A Corner of Mathematical Olympiad and Competition Book II: Phnom Penh, October 24, 2014 Prepare By: Keo Sodara

A Corner of Mathematical
Olympiad and Competition

Book II
1. Estonian Mathematical Olympiad and Competition
2. Singapore Mathematical Olympiad Team Selection
Test and Trainings
3. IMO Winter and Summer Camp Trainings Test
Problems with Solutions
Phnom Penh, October 24, 2014

Prepare by: Keo Sodara
Contact: [email protected]
[email protected]
www.facebook.com/aprissigi
(+855) 0967308278
Estonian Mathematical
Olympiad and Competition
1993/1994-1995/1996, 1998/1999
And
2000/2001-2012/2013
Estonian Mathematical Olympiad ’93
(Selected Problems)
The mathematical olympiad in Estonia is held annually in three stages:

at school, city/district, and all-Estonian levels. At each level, separate
problems are usually given to students of each grade from 9 to 12, in
compliance with the school curriculum. Younger students also take
part at the first two levels.
Below, after each problem number, the numbers in brackets indicate

the grade(s) and stage (II or III) where it was used.
Problem 6 originates from Iceland ’92; problem 4 is due to an oral

contribution from Sweden; problems 2, 3 and 5 are taken from various
Russian sources; the rest seem to be original.
Problems
1. (9, II) Find all triples of pairwise different real numbers (a, b, c) ,
such that the system of equations
ax = b



bx = c

cx = a

has a solution.
2. (10, II) The point of osculation of the hypotenuse of a right triangle

with its inscribed circle divides the hypotenuse into intervals of
lenghts a and b . Find the area of the triangle.
3. (10, III) Natural numbers m , n and k have the properties: mn

divides nm , and nk divides kn . Prove that mk divides km .
4. (10-11, III) For which natural numbers n is it possible to cover

a rectangle of size 3 × n unit squares with shapes shown on the
figure, without overlapping? (Each shape is available in unlimited
quantity.)
5. (10, III) Let r be the radius of the inscribed circle of a right trian-
gle, and let h be the height of the triangle, drawn to its hypotenuse.
This height divides the original triangle into two smaller triangles,
let r1 and r2 be the radii of the circles inscribed in these triangles.
Prove the following equalities:
a) r1 + r2 + r = h ;
b) r12 + r22 = r2 .
6. (11, II) In a right triangle ABC , the medians drawn to its side AC
and to the hypotenuse are perpendicular to each other. Find the
area of the triangle if the side AB has length 1 .
7. (11, III) Find all functions f (n) having the following three prop-
erties:
a) for any natural number n , f (n) is also a natural number;
b) f (n + m) = f (n)f (m) holds for any natural numbers n and m ;
c) the equation f (f (n)) = (f (n))2 has some natural number n0 as
its solution.
8. (12, II) Prove that, for any natural number k > 2 , there exist k
pairwise different natural numbers n1 , n2 , . . . , nk such that
1 1 1 3
+ + ··· + = .
n1 n2 nk 17
9. (12, III) A rectangle is cut into five rectangular pieces having equal
areas. Prove that at least two of the pieces are identical.
10. (12, III) Let us call a natural number “beautiful” if it is a perfect

square itself, and its presentation in the decimal system is obtained
by writing side by side two or more squares of natural numbers.
(For example, 169 is a “beautiful” number, as 169 = 132 , 16 = 42
and 9 = 32 .) How many “beautiful” numbers divisible by 1993 are
there?
Hints and Solutions
1. There are no such triples. It is easy to see that abc 6= 0 , and

a + b + c = 0 if a solution x 6= 1 exists. Now, one of the num-
bers a, b, c must have a sign opposite to the two others, and thus
x should be both positive and negative.
2. From the Pythagorean Theorem a
we find
a
(r + a)2 + (r + b)2 = (a + b)2 ,
b
and √ r r
−a − b + a2 + 6ab + b2
r= .
2 r b
(r + a)(r + b)
So, the area is S= = ab .
2
3. By the assumptions, (mn )k = (mk )n divides (nk )m = (nm )k which di-

vides (km )n = (kn )m . Thus, (mk )n divides (km )n , and this implies mk
dividing km .
4. Assign numbers to the unit squares as shown on the figure:

···
1 1 1 1 1
···
−1 −1 −1 −1 −1
···
1 1 1 1 1
···
(A) (B)
The sum of numbers covered by shape (B) is always equal to zero;

for shape (A) this sum is either 1 or −1 . As the sum of all numbers
assigned to the squares is n , we have to use at least n shapes, that
is, we cannot use shape (B) at all. Now, it is easy to see that
covering is possible if and only if n is an even number.
5. Calculating the area of the C
triangle ABC in two different r r
L
ways, we obtain the equal- M
ity h(x + y) = 2(r2 + xr + yr) ; x
2r(x + y + r) y
thus h = . As O
(r1 ) h
x+y (r2 )
the triangles AP C, CP B and y
A x B
ACB are all similar, we get K P
r1 r2 r r
= = , and r1 + r2 + r = (2x + 2y + 2r) = h . Now, we
x+r y+r x+y x+y
r2
also have r12 + r22 = ((x + r)2 + (y + r)2 ) = r2 ,
(x + y)2
as (x + y)2 = (x + r)2 + (y + r)2 by the Pythagorean Theorem.
C
6. Denote 6 CED = 6 AED = 6 EAB = α
and CE = EB = EA = x . Then we have
FE FA FE 1 x
= cos α = , implying = and
DE AB FA 2
x DE FE
FE = . As = cos α = , we get D α
E
3 CE DE α
3 √ √
x2 = and AC = 4x2 − 1 = 2 . The area F ·
4 x
√
AB · AC 2
is equal to = .
2 2
α α
A 1 B
7. Condition b) implies f (n + 1) = f (n)f (1) for any natural number n ,

and thus f (n) = (f (1))n . Denote f (1) = k , then by condition c) there
is a natural number n0 such that kk 0 = (kn0 )2 . This implies either
n
k = 1 or k n0 = 2n0 ; in the latter case we have k = 2 and n0 = 1 or

n0 = 2 . Therefore, the only functions having the required properties
are f (n) ≡ 1 and f (n) = 2n .
8. For k = 2 , take n1 = 6 , n2 = 102 . Now, use the identity

1 1 1
= +
m m + 1 m(m + 1)
taking the greatest of the denominators as m , at each step.

9. Quite obviously, the assertion holds
when we cut a rectangle in two, three, 2
or four pieces. The only way to cut
1
a rectangle in five rectangular pieces
5
so that no two, three or four of them 3
make up a rectangle is shown on the
figure. By comparing the areas, it is 4
easy to show that pieces 1 , 3 and
2 , 4 must be identical.
10. Denote x = 1993 and y = 2 · 1993 , then the number

(108 · x + y)2 = x2 · 1016 + 2xy · 108 + y 2
is “beautiful” (it is obtained by writing side by side the perfect

squares x2 , 2xy ja y 2 ), and it is divisible by 1993 . Now, for any
“beautiful” number n divisible by 1993 , check that 100n has the
same properties.
Estonian Mathematical Competitions ’94
(Selected Problems)
The Estonian Mathematical Olympiad is held annually in three stages:

at school, city/district, and national levels. At each level, separate
problems are usually given to students of each grade from 9 to 12, in
compliance with the school curriculum.
Since 1993, we also have two “open” mathematical competitions each

year. At these, any student can participate, and everyone is given the
same set of problems.
Here we present a variety of problems from both of these competitions.

For each problem, we indicate in brackets the competition, stage (II
or III) and grade(s) where it was used. Many of these problems are,
in fact, not original but taken from various sources and more or less
modified.
1. (II, 9) The sides of a convex quadrangle serve as diameters for four

circles. Prove that these circles fully cover the quadrangle.
2. (II, 10) Find the largest natural number n such that 999
| {z. . . 99} is
999 digits
divisible by 9n .
3. (II, 11) Prove that for any real numbers x and y , we have
| sin x| > | cos y| if and only if | sin y| > | cos x| .
4. (II, 12) For a point C on the diameter AB of a semicircle, let P

and Q be points on the semicircle such that 6 ACP = 6 BCQ = β .
Prove that, for any given value of β , the length of segment P Q
does not depend on the choice of point C .
5. (II, 11–12) The “Top Twenty” of popular songs is selected weekly.

It is known that:
1) the same 20 songs are never selected in the same order on any
two consequtive weeks;
2) once a song has moved backward in the ordering, it cannot
improve its position in the future.
Determine the maximum number of consequtive weeks the same 20
songs can be selected.
6. (III, 9) Label the sides and diagonals of a regular pentagon by
natural numbers 1 to 10 and consider all triangles with vertices
at the vertices of the pentagon. Is it possible that for each of these
triangles the sum of labels at its sides is the same?
7. (III, 10) Find the least natural number having exactly 100 different
natural divisors (including 1 and the number itself).
sin β
8. (III, 10) Let α, β, γ, δ be such that 0◦ < α, β, γ, δ < 90◦ , sin α =
sin γ
tan β tan δ
and sin δ = . Prove that tan α = .
tan γ cos γ
9. (III, 10) Three men decided to have a picnic together with their
wives. They arrived at the railway-station one by one, each of them
greeting by handshake all those already present, except his/her wife
or husband. Later one of the men asked all the others how many
companions each had shaken hands with upon arrival, and he got
five different answers. What would have been his own answer?
10. (III, 11) Compute the sum of all the five-digit natural numbers
that do not change when the order of their digits is reversed (we do
not accept 0 as the first digit of a natural number).
11. (III, 11) Prove that for any triangle the equality
cot α β
R 2 + cot 2
=
r 2 sin γ
holds, where α, β, γ are the angles of the triangle and R , r are the
radii of its circumcircle and inscribed circle, respectively.
12. (III, 11) For any sequence A = { a1 , a2 , . . . } of real numbers denote

∆A = {b1 , b2 , . . .} where bi = ai+1 − ai . Determine the value of term a1
of sequence A , knowing that a19 = a94 = 0 and the sequence ∆(∆A)
is constant with all its terms equal to 1 .
13. (III, 12) For a triangle ABC , let A1 , B1 ja C1 be points on its

sides BC , AC and AB respectively, such that the three segments
AA1 , BB1 ja CC1 intersect in one point O and
AO BO CO
+ + = 1994 .
OA1 OB1 OC1
AO BO CO
Find the value of · · .
OA1 OB1 OC1
14. (III, 12) Prove that, for any natural number n , the number
n
n! = 1 · 2 · . . . · n has less than trailing zeros.
4
15. (III, 12) All three vertices of a regular triangle lie on the sides of
a unit square (some of them may coincide with the vertices of the
square). Determine the minimum and maximum values for the area
of such a triangle.
16. (OPEN) The two sides BC and CD of an inscribed quadrangle

ABCD are of equal length. Prove that the area of this quadrangle
1
is equal to S = · AC 2 · sin 6 A .
2
17. (OPEN) Consider the sequence of real numbers a0 , a1 , a2 , . . . :

a0 = x , a1 = y , ai = ai−2 + ai−1 (i > 2) .
Determine all values of x and y for which this sequence is bounded

(i.e. there exists a positive number M such that |ai | < M for all
i = 0, 1, 2, . . . ).
18. (OPEN) A number of balls of radius R are

m
laid on the table in the form of a regular tri- mm
angle with each side containing N balls (see mmm
mmmm
figure for N = 5 ). A frame is put around this mmmmm
triangle and further layers of similar balls are
laid on it, each successive layer having one
ball less on each of its sides (so the last, N -th
layer consists of a single ball). Determine:
a) the total height of the pyramid (for arbitrary N );
b) the total number of balls for N = 10 ;
c) the total number of balls for arbitrary N .
19. (OPEN) Consider a function f (x) such that

f (f (x)) = f (x) + 1994 x
for any real value of x .

a) Prove that f (x) = 0 if and only if x = 0 ;
b) Find one such function.
AC AB + BC
20. (OPEN) Prove that if = in a triangle ABC , then
BC AC
6 B = 2·6 A.
Selected problems from Estonian mathematical
contests 1995-96
2nd round
1. Find all positive integers n such that 6n is divisible by 6 + n .

2. In a town there are three streets – T , L and H . All the citizens
living in the street T tell only the truth, all the citizens from the
street L only lie and in the speech of the citizens from the street
H true and false sentences alternate.
Once the fire-brigade noticed a cloud of smoke above the town and
right away the phone rang.
Caller:“There is a fire in our street!”
Fireman:“What street is it?”
Caller:“The street H .”
What was the street the fire-brigade had to go to?
3. Four students A , B , C and D took part in a contest of solving
logical problems. They all gathered different number of points.
Before the contest the students told the following.
A: “I shall win the contest”
B: “I am a boy and shall get the first place”
C: “The boys are wrong, D will get the place right after me and
there will be no boys after A”
D: “C is right and there will be no girls after A”
It is known that two contestants were boys and two girls, two of
the sentences turned out to be right and two wrong. Find the final
places and the genders of all the contestants.
4. An n × n × n cube is coloured blue and cut into n3 unit cubes.
Prove that for no natural number n the number of uncoloured unit
cubes and the number of unit cubes with at least one coloured face
can be equal.
5. 50 different numbers are chosen from the numbers 1, 2, . . . , 100 .
Prove that from these 50 numbers we can choose two whose sum
is a perfect square.
3rd round
1. Does there exist a positive integer such that its last digit is non-
zero and that it becomes exactly two times bigger when the order
of its digits is reversed?
2. Three children wanted to make a table-game. For that purpose
they wished to enumerate the mn squares of an m×n game-board
by the numbers 1, . . . , mn in such way that the numbers 1 and mn
lie in the corners of the board and the squares with successive
numbers have a common edge. The children agreed to place the
initial square (with number 1) in one of the corners but each child
wanted to have the final square (with number mn ) in different
corner. For which numbers m and n is it possible to satisfy the
wish of any of the children?
3. John and Mary play the following game. First they choose integers
n > m > 0 and put n sweets on an empty table. Then they start
to make moves alternately. A move consists of choosing a non-
negative integer k 6 m and taking k sweets away from the table
(if k = 0 , nothing happens in fact). In doing so no value for k
can be chosen more than once (by none of the players) or can be
greater than the number of sweets at the table at the moment of
choice. The game is over when one of the players can make no
more moves.
John and Mary decided that at the beginning Mary chooses the
numbers m and n and then John determines whether the per-
former of the last move wins or looses. Can Mary choose m and
n in such way that independently of John’s decision she will be
able to win?
4. Prove that 1n + 2n + . . . + 15n is divisible by 480 for any odd n > 5 .
5. Let p be a fixed prime number. Find all pairs (x, y) of positive
integers satisfying the equation p(x − y) = xy .
6. In a plane there are n triangles such that any three of them have
a common vertex and four of them have vertex in common. Find
the greatest possible value for n .
7. In the spase there are n tetrahedra such that any two of them have
two common vertices and no three of them have three vertices in
common. Find the greatest possible value for n .
Final (selectional) round
x2 + y 2 + 6
1. The numbers x , y and are positive integers. Prove
xy
x2 + y 2 + 6
that is a perfect cube.
xy
2. Let a , b , c be the sides of a triangle and α , β , γ the opposite
angles of the sides respectively. Prove that if the inradius of the
triangle is r then a sin α + b sin β + c sin γ > 9r .
3. Find all functions f : R → R satisfying the following conditions for
all x ∈ R :
(a) f (x) = −f (−x) ;

(b) f (x + 1) = f (x) + 1 ;
1 1

(c) f = 2 f (x) , if x 6= 0 .
x x
x2 x3 xn
4. Prove that the polynomial Pn (x) = 1 + x + + + . . . + has no
2 6 n!
zeros if n is even and has exactly one zero if n is odd.
5. Let H be the ortocenter of an obtuse triangle ABC and A1 , B1 ,
C1 arbitrary points taken on the sides BC , AC , AB , respectively.
Prove that the tangents drawn from the point H to the circles with
diameters AA1 , BB1 , CC1 are equal.
6. Each face of a cube is divided into n×n equal squares. The vertices
of the squares are called nodes, so each face of the cube has (n + 1)2
nodes.
(a) For n = 2 , does there exist a closed broken line whose links
are the edges of the squares which contains each node exactly
once?
(b) for n arbitrary, prove that each such broken line divides the
surface area of the cube into two equal parts.
Open contests
1. Exactly one of the following statements is known to be true.

(a) All the following statements are true;
(b) None of the following statements is true;
(c) At least one of the following statements is true;
(d) All the previous statements are true;
(e) None of the previous statements is true.
Find the true statement.
2. A 4 × 4 square is divided into unit squares, some of which are then
coloured and into every uncoloured square we write the number of
its coloured neighbours (the squares are said to be neighbours if
they have a vetrex or an edge in common). Is it possible to colour
the squares in such way that in every uncoloured square there will
be
(a) the number 2;
(b) the number 3?
3. Prove that 3n +n3 is divisible by 7 if and only if 3n ·n3 +1 is divisible
by 7.
4. Is it possible to draw 19 lines in a plane in such way that they
have exactly 95 points of intersection and no three lines have a
point in common?
5. In a company mn soldiers were drawn up into m rows and n
columns. First they got a command to rearrange themselves inside
the rows into increasing order of tallness and after that they got
a second command to perform similar rearrangement inside the
columns. Prove that after the final rearrangement the soldiers in
the rows still stand in increasing order of tallness.
6. In a pentagon (not necessarily convex) all the sides are of length
1 and the product of cosines of any four angles is zero. Find all
possible values for the area of the pentagon.
7. The game-board has a form of rectangle with width 2 and height
n . On each but one square of the board one of the 2n − 1 bricks
is placed, one square is left empty. The bricks have numbers
1, 2, . . . , 2n − 1 and the bricks number 1 and 2 are in the top row. A
move consists of moving a brick from a square having a common
edge with the empty square onto the empty square. Is it possible
to interchange the places of the bricks number 1 and 2 using the
finite number of described moves, if
(a) n=2;
(b) n=3?
8. Which of the numbers 21996! and 21996 ! is greater?
9. There are n petrol stations on a circular motorway. Each of the
stations has a certain amount of fuel, while all the stations together
have exactly the amount of fuel which is necessary to run through
the motorway. Prove that it is always possible to choose the first
station such that starting from that station with an empty cistern
a car can run through the motorway and return to the initial point.
Problems from Estonian mathematical contests 1998/1999
Problems from the final round
10th grade
1. Find all pairs of integers (a, b) such that a2 + b = b1999 .
Answer: (0; −1), (0; 0) and (0; 1).
Solution. The given condition is equivalent to a2 = b(b1998 −1). If b > 2 then b and b1998 −1
are both positive and relatively prime, hence both perfect squares. But since b1998 is also
a perfect square, we get a contradiction. So b 6 1. The cases b = 1, b = 0 and b = −1
all give a = 0. At last note that b 6 −2 is impossible because it implies a2 < 0.
2. Find all values of a such that absolute value of one of the roots of the equation
x2 + (a − 2)x − 2a2 + 5a − 3 = 0
is twice of absolute value of the other root.

5 7 5
Answer: , and .
4 5 3
Solution. Solving the equation, one gets
(a − 2)2 − 4 · (−2a2 + 5a − 3)
p
−(a − 2) ±
x= =
2
√
2−a± a2 − 4a + 4 + 8a2 − 20a + 12
= =
2
√
9a2 − 24a + 16 2 − a ± (3a − 4)2
p
2−a±
= = ,
2 2
so
2 − a − (3a − 4) 2 − a + (3a − 4)
x1 = = −2a + 3 , x2 = =a−1.
2 2
Assume |x1 | = 2|x2 |, i.e. | − 2a + 3| = 2 · |a − 1|. The numbers a − 1 and −2a + 3 are
3
negative iff a < 1 and a > respectively. Thus in the case a < 1 the condition reduces
2
3
to −2a + 3 = 2(1 − a), and in the case a > the condition reduces to 2a − 3 = 2(a − 1).
2
3
Hence both cases are impossible. So 1 6 a 6 , which gives −2a + 3 = 2(a − 1) and
2
5
a= .
4
Now assume |x2 | = 2|x1 |, i.e. |a − 1| = 2 · | − 2a + 3|. In the case a < 1 this condition
5
reduces to 1 − a = 2(−2a + 3), which gives a = , but this is not sound with the case
3
3
assumption a < 1. In the case 1 6 a 6 the condition reduces to a − 1 = 2(−2a + 3)
2
7 3
which gives a = , and in the case a > the condition reduces to a − 1 = 2(2a − 3)
5 2
5
which gives a = .
3
1
3. The incircle of the triangle ABC , with the center I , touches the sides AB , AC and BC
in the points K , L and M respectively. Points P and Q are taken on the sides AC and
BC respectively, such that |AP | = |CL| and |BQ| = |CM |. Prove that the difference of
areas of the figures AP IQB and CP IQ is equal to the area of the quadrangle CLIM .
Solution. We have SAP I = SCLI since |AP | = |CL| and both triangles have altitude LI .
Analogously we have SALI = SCP I since |AL| = |CP |. But △ALI ∼ = △AKI because
the sides are respectively equal. Hence
SAP IK − SCP I = SAKI + SAP I − SCP I = SALI + SCLI − SALI = SCLI .
Analoguously we find that
SBM IK − SCQI = SCM I .
Thus we have
SAP IQB − SCP IQ = SAP IK + SBM IK − SCP I − SCQI = SCLI + SCM I = SCLIM .
4. 32 stones, with pairwise different weights, and lever scales without weights are given. How
to determine by 35 scaling, which stone is the heaviest and which is the second by weight?
Solution. At first we determine the heaviest stone by 31 scalings and find at the same
time 5 stones, one of which must be the second by weight. This can be achieved by pairing
all the 32 stones and comparing the stones in each pair; after that pairing the heavier
stones and comparing again the stones in each pair, etc. This takes 16 + 8 + 4 + 2 + 1 = 31
scalings. By the last scaling we find the heaviest stone and the second by weight may be
only one of these five which have been in pair with the heaviest.
To complete the solution we must determine by 4 scaling, which of these 5 stones is the
heaviest. It can be done linearly.
5. Let C be an interior point of line segment AB . Equilateral triangles ADC and CEB
are constructed to the same side from AB . Find all points which can be the midpoint of
the segment DE .
Answer: the midline of the equilateral triangle constructed on the segment AB which is
parallel to AB (endpoints excluded).
E
Fq
D
A C B
Figure 1
Solution. Let F be the midpoint of the segment DE . Lengthen the segments AD and
BE till intersecting in point G (see figure 1). The location of the point G does not depend
on the point P and the quadrangle ECDG is a parallelogram with the point F as the
midpoint of its diagonals. While the point C moves along AB from A to B , point F
moves from the midpoint of AG to the midpoint of BG along the segment connecting
these points. According to the condition of the problem, the endpoints must be excluded.
2
11th grade
1. Find all pairs of integers (m, n) such that
4mn
(m − n)2 = .
m+n−1
k(k + 1) k(k − 1)

Answer: the pairs (k, −k) and , , (where k is an arbitrary integer),
2 2
(1, 0) and (0, 1) excluded.
Solution. By multiplying both sides of the equation with the denominator of the right
side, we get (m − n)2 (m + n − 1) = 4mn, which gives (m + n)2 = (m − n)2 (m + n). Hence
m + n = 0 or m + n = (m − n)2 . The first case gives n = −m, i.e. all the pairs (k, −k),
where k is integer, are suitable. In the second case take m − n = k , then m + n = k2
k2 + k k2 − k
and m = , n= . To ensure the denominator of the fraction in the problem
2 2
is not zero the condition m + n 6= 1 must be added, so k 6= 1 and k 6= −1.
2. Find the value of the expression
1 2 1999 2000 2000 2000

f +f +. . .+f +f +f +. . .+f
2000 2000 2000 2000 1999 1
x2
assuming f (x) = .
1 + x2
1
Answer: 1999 .
2
Solution. One gets the answer directly using the fact that for any non-zero real number
x,
2
1
1 x2 x x2 1 1
f (x) + f = + 2 = + 2· =
x 1 + x2
1+ 1 1+x 2 x 1+ 1
2
x x
x 2 1
= + =1.
1 + x2 1 + x2
3. For the given triangle ABC , prove that a point X on the side AB satisfies the condition
−−→ −−→ −−→ −−→ −→ −−→
XA · XB + XC · XC = CA · CB iff X is the basepoint of the altitude or median of the
triangle ABC (~v · ~u denotes the scalar product of vectors ~v and ~u ).
C
O
-
~
A X B
Figure 2
3
−→ −−→ −−→ −−→ −−→ −−→
Solution. Since CA = XA − XC and CB = XB − XC (see figure 2), the condition in the
−−→ −−→ −−→ −−→ −−→ −−→ −−→ −−→
problem is equivalent to the condition XA· XB + XC · XC = = (XA− XC)·(XB − XC).
−−→ −−→ −−→
Transforming this, we get the equation XC · (XA + XB) = 0.
−−→ −−→ −−→ −−→ −−→
This condition holds iff XA + XB = 0 or XC ⊥ XA + XB . The first case holds iff X is
the midpoint of the side AB , the second case holds iff X is on the altitude.
4. For which values of n it is possible to cover the side wall of staircase of

n steps (for n = 6 in the figure) with plates of shown shape? The width
and height of each step is 1 dm, the dimensions of plate are 2 × 2 dm
and from the corner there is cut out a piece with dimensions 1 × 1 dm.
Answer: n = 3k or n = 3k + 2, except n = 3 and n = 5.
n(n + 1)
Solution. The area of the side wall of the staircase is 1 + 2 + . . . + n = dm2 , the
2
area of each plate is 3 dm2 . Thus one of the numbers n and n + 1 must be divisible by
3, i.e. it isn’t possible to plate the side wall of the staircase if n ≡ 1 (mod 3).
Consider the cases n = 2, n = 3, n = 5. Clearly, if n = 2 then the plating is possible.

If n = 3 then one of the plates must be in the vertex of the wall and it is not possible to
cover the remaining part with plates with required shape (see figure 3). If n = 5 then two
plates must be in vertices of the wall. The remaining part has shape 3 × 3 and thus can’t
be covered with 3 plates, because otherwise one of these plates must cover two different
vertices of the 3 × 3 square.
n=2 n=3 n=5
Figure 3
In the cases n = 6 and n = 9 the wall can be covered as shown in figure 4.
n=6 n=9
Figure 4
Now we give a method how to construct plating in case n = 3k + 6 from plating in case
n = 3k . Thus the wall can be plated for any n = 3k , k > 1. For this leave unplated the
lower ribbon of the wall with height 6 dm and plate the upper part as in case n = 3k .
The ribbon can be plated as shown in figure 5.
4
...
| {z }
k rectangles
Figure 5
At last we show how to get the plating in the case n = 3k + 2 from the plating in the
case n = 3k . This implies the wall can be plated for any n = 3k + 2, k > 1. For this leave
unplated the lower ribbon with height 2 dm and plate the upper part as in case n = 3k .
The ribbon can be covered as shown in figure 6.
...
| {z }
k rectangles
Figure 6
5. On the squares a1, a2, . . . , a8 of a chessboard there are respectively 20 , 21 , . . . , 27 grains

of oat, on the squares b8, b7, . . . , b1 respectively 28 , 29 , . . . , 215 grains of oat, on the
squares c1, c2, . . . , c8 respectively 216 , 217 , . . . , 223 grains of oat etc. (so there are 263
grains of oat on the square h1). A knight starts moving from some square and eats after
each move all the grains of oat on the square to which it had jumped, but immediately
after the knight leaves the square the same number of grains of oat reappear. With the
last move the knight arrives to the same square from which it started moving. Prove that
the number of grains of oat eaten by the knight is divisible by 3.
Solution: The arrangement of grains of oat described in the problem implies that on each
white square of the chessboard there are 22n grains, while on each black square there
are 22n+1 grains, where n = 0, 1, 2, . . . , 31. Therefore the number of grains on any black
square is congruent to 1 modulo 3 and the number of grains on any white square is
congruent to 2 modulo 3. Since the knight moves always from a square of one colour to a
square of the other colour the number of grains it eats with each two consecutive move is
divisible by 3. But knight makes an even number of moves, because the initial and final
square are of the same colour. Hence the assertion of the problem holds.
12th grade
1. Let a, b, c and d be non-negative integers. Prove that the numbers 2a 7b and 2c 7d give
the same remainder when divided by 15 iff the numbers 3a 5b and 3c 5d give the same
remainder when divided by 16.
′ ′
Solution. First we show that if |a′ − a| = |b′ − b| = 2 then 2a 7b ≡ 2a 7b (mod 15) and
′ ′
3a 5b ≡ 3a 5b (mod 16). Indeed we can assume that a′ = a + 2. If b′ = b + 2, we obtain
′ ′
2a 7b = 2a 7b · 22 72 = 2a 7b · (2 · 7)2 ≡ 2a 7b · (−1)2 = 2a 7b (mod 15)
5
and
′ ′
3a 5b = 3a 5b · 32 52 = 3a 5b · (3 · 5)2 ≡ 3a 5b · (−1)2 = 3a 5b (mod 16) .
If b′ = b − 2, we can use the same relations noting that 74 ≡ 1 (mod 15) and
54 ≡ 1 (mod 16).
Now we prove that for every pair of non-negative integers (a, b) there exists a pair (a′ , b′ )
′ ′ ′ ′
such that 2a 7b ≡ 2a 7b (mod 15), 3a 5b ≡ 3a 5b (mod 16), a′ ∈ {0, 1, 2, 3} and b′ ∈ {0, 1}.
We conclude that both of the exponents can be changed by a number divisible by 4 without
changing the remainder of dividing by the required number. Thus we can consider only
the case where a, b ∈ {0, 1, 2, 3}. If b 6 1, take a′ = a and b′ = b; if b > 1 then b′ = b − 2
and a′ can be chosen from the set {0, 1, 2, 3} so that it differs from the number a exactly
by 2.
′ ′
It remains to prove that the remainders of the numbers 2a 7b when divided by 15 and the
′ ′
remainders of the numbers 3a 5b when divided by 16 are pairwise different if the numbers
a′ and b′ come from the abovementioned sets. This can be seen from the following tables.
The remainder of The remainder of

Number dividing by 15 Number dividing by 16
0 0 0 0
2 7 1 3 5 1
21 70 2 31 50 3
2 0 2 0
2 7 4 3 5 9
3 0 3 0
2 7 8 3 5 11
20 71 7 30 51 5
1 1 1 1
2 7 14 3 5 15
22 71 13 32 51 13
3 1 3 1
2 7 11 3 5 7
2. Find the value of the integral
Z1 p
ln x + 1 + x2 dx.
−1
Answer: 0.
Solution. Let f (x) = ln(x + 1 + x2 ). We always have 1 + x2 > x2 , so 1 + x2 > |x|,

p p
thus f (x) is defined for every x. We will show that the function f (x) is odd, i.e.
f (−x) = −f (x). Indeed,
√ √ !
q ( 1+x 2 − x)( 1+x2 + x)
f (−x) = ln(−x + 1+(−x)2 ) = ln √ =
1+x2 + x
! !
1 + x2 − x2 1
= ln √ = ln √ =
1 + x2 + x x + 1 + x2
p
= − ln(x + 1 + x2 ) = −f (x) .
6
Now let
Z1 Z0 Z1
I= f (x) dx, I1 = f (x) dx, I2 = f (x) dx .
−1 −1 0
Making the substitution x = −t in the integral I1 , we obtain
Z0 Z1 Z1
I1 = − f (−t) dt = f (−t) dt = − f (t) dt = −I2 .
1 0 0
Consequently
I = I1 + I2 = −I2 + I2 = 0 .
3. Prove that the line segment, joining the orthocenter and the intersection point of the me-
dians of the acute-angled triangle ABC is parallel to the side AB iff tan 6 A · tan 6 B = 3.
Solution: Let the basepoints of the altitudes drawn from the vertices A and C be D and
E respectively and let the orthocenter be H (see figure 7). Note first that H lies on the
line parallel to AB and passing through the intersection point of the medians if and only if
|CE| |CE|
= 3, thus it is enough to prove that = tan 6 A · tan 6 B . As 6 AHE = 6 CHD ,
|EH| |EH|
we also have 6 EAH = 6 BCE , i.e. the right triangles CEB and AEH are similar and
|CE| |AE| |CE|
tan 6 B = = . Noting that tan 6 A = , we obtain the necessary relation.
|EB| |EH| |AE|
C
D
·
H
·
A E B
Figure 7
4. Let us put pieces on some squares of 2n × 2n chessboard in such a way that on every
horizontal and vertical line there is an odd number of pieces. Prove that the whole number
of pieces on the black squares is even.
Solution: Enumerate all the horizontal and vertical lines by numbers 1, . . . , 2n and assume
that the square (1, 1) is black (this does not restrict generality). Let A be the number
of pieces on the squares with even horizontal and vertical line number, B the number of
pieces on the squares with odd horizontal and vertical line number and C the number of
pieces on the squares with even horizontal and odd vertical line number. Then A + C is
the number of pieces with even horizontal line number and B + C the number of pieces
with odd vertical line number. As the number of such horizontal and vertival lines is the
same, these numbers have the same parity and thus the number of pieces on the black
squares A + B = (A + C) + (B + C) − 2C is even.
5. The numbers 0, 1, 2, . . . , 9 are written (in some order) on the circumference. Prove that
a) there are three consecutive numbers with the sum being at least 15;
b) it is not necessarily the case that there exist three consecutive numbers with the sum
more than 15.
7
Solution: a) The sum of numbers on the circumference is 45 and thus adding the ten
triple sums gives 3 · 45 = 135. Note that in every two neighbouring triples there are two
common numbers and the third one is different, so their sums must be different. Hence,
if the largest sum would be 14, we should have the sums 13 and 14 altering. But then
the sum of every six-tuple should be 27 which is not possible, because the sums of two
neighbouring six-tuples must also be different.
b) Write on the circumference the numbers in the following order: 3, 8, 1, 5, 9, 0, 6, 7, 2,
4. Now it is elementary to check that no triple gives the sum greater than 15.
Problems from the open contests
Younger group (up to 10th grade)
1. John knows n > 3 positive real numbers and he writes them all on the blackboard (every
number may occur more than once). Mary writes under every number the arithmetic mean
of the other n − 1 numbers and then deletes the initial numbers. She repeats the process
for 1998 times. After that Mary notices that there are exactly the initial numbers on the
blackboard. How many different real numbers does John know?
Answer: John knows one positive real number.
Solution. Let M be the greatest of the initial numbers. For arbitrary b1 , . . . , bn−1
n−1
z }| {
b1 + . . . + bn−1 M + ... + M
6 =M
n−1 n−1
and the equality holds if and only if b1 = . . . = bn−1 = M . If there were at least two
numbers less than M on the blackboard then all the next numbers would also be less
than M . And then of course on the 1998th time there cannot be M on the blackboard.
If only one of the numbers on the blackboard is less than M then the next time there
would be already two numbers less than M . Therefore all the initial numbers were equal
and John knows only one positive real number.
2. Two different points X and Y are chosen in the plane. Find all the points Z in this plane
for which the triangle XY Z is isosceles.
s
c1 c2
Xq qY
Answer: All the points on the circumferences c1 and c2 except the five points lying on
the line XY (see the figure).
8
Solution. If the points X , Y and Z are the vertices of an isosceles triangle then
|XZ| = |XY |, |Y Z| = |XY | or |XZ| = |Y Z|. The points Z such that |XZ| = |XY |
lie on the circumference c1 . The points Z such that |Y Z| = |XY | are situated on the
circumference c2 . The points Z such that |XZ| = |Y Z| are situated on the midperpen-
dicular s of the segment XY . It is clear that the points determine a triangle if and only
if they do not lie on the same line.
3. Prove that for every integer k the following assertions are equivalent (if one is true then
the other is also true and vice versa):
a) exist nonnegative integers a and b so that k = a2 + b2 + ab,
b) exist nonnegative integers c and d so that k = c2 + d2 − cd.
Solution. We shall use the relations
n2 + m2 + nm = (n + m)2 + m2 − (n + m)m, (1)

2 2 2 2
n + m − nm = (n − m) + m + (n − m)m. (2)
Let k be expressed in the form k = a2 + b2 + ab. Denoting a = n, b = m on the left side

and c = n + m, d = m on the right side of the relation (1) we obtain
a2 + b2 + ab = k = c2 + d2 − cd.
Let now k be expressed in the form k = c2 + d2 − cd. Without loss of generality we can
assume that c > d. Denoting c = n, d = m on the left side and a = n − m, b = m on
the right side of the relation (2) we obtain
c2 + d2 − cd = k = a2 + b2 + ab.
9
4. Find all the four-digit numbers n such that multiplying n by we obtain the number
2
which consists of the same digits as n but in the opposite order.
Answer: 1818 and 1998.
Solution. Let n = abcd. Since 4,5 · abcd < 10000 then a = 1 or a = 2. The digit d has
9
to be even. At the same time multiplying it by the last digit of the product must be
2
1 or 2. Consequently a = 2 and d = 6 or a = 1 and d = 8. Let us examine these cases
separately.
9
1) If a = 2 and d = 6 then · abcd > 9000 which contradicts d = 6.
2
2) If a = 1 and d = 8 then we obtain
9
(1000 + 100b + 10c + 8) · = 8000 + 100c + 10b + 1
2
3465 + 55c = 440b
63 + c = 8b.
This equation has two solutions: c = 1, b = 8 and c = 9, b = 9. These values will give us
the numbers 1818 and 1998.
1. The teacher wrote an addition exercise on the blackboard

A C E
+ + = ,
B D F
9
where A, B , C , D , E , F are positive integers, all three fractions are reduced and their
denominators are pairwise relatively prime. The pupil adds the fractions writing the least
common multiple of the denominators of the summands as the denominator of the result.
Prove that the fraction that the pupil writes is reduced.
Solution. As the denominators of the fractions are pairwise relatively prime, their least
ADF + CBF + EBD
common multiple is BDF and so the resulting fraction is in form .
BDF
Let’s suppose antithetically that this fraction is reducable by some number greater than
one. Then there exists a prime number p which divides both the numerator and the
denominator of this fraction. As the product BDF is divisible by the prime number p
so one of terms if divisible by it: without loss of generality we can assume that this term
is B . But then two summand in sum ADF + CBF + EBD are divisible by p and as
the sum is divisible by p by assumption, the summand ADF is divisible by p as well.
Therefore either A, D or F is divisible by p. In the first case we get a contradiction with
A
the fraction being reduced, in other cases with assumption that the denominators of
B
the initial fractions were pairwise relatively prime.
Comment. Analogical assertion can be proved in case where there are more than three
fractions as summands.
2. Prove that the value of the expression
1
1+
1
1+
1
1+
1+n
is not an integer for any integer n.
Solution. The expression given in the problem is not defined when n = −1 or n = −2.
n+2 n+2
The value of this expression is 1 + what is an integer iff is an integer i.e.
2n + 3 2n + 3
the number 2n + 3 divides the number n + 2. We will show that no such integer exists.
As n 6= −1 and n 6= −2 then |3 + 2n| 6= 1. If there exits an integer n for which the
n+2
fraction could be reduced with some number d > 1 then both 2n + 3 and n + 2
2n + 3
would be divisible by d, so 2 · (n + 2) = 2n + 4 would be divisible by d as well and
(2n + 4) − (2n + 3) = 1 would be divisible by d—a contradiction. Therefore the fraction
n+2
is reduced for all integers n 6= −1, −2 and the value of given expression can’t be
2n + 3
an integer for any integer n.
3. On the plane there are two non-intersecting circles with equal radii and with centres O1
and O2 , line s going through these centres, and their common tangent t. The third circle
osculates these two circles in points K and L respectively, line s in point M and line t
in point P . The point of tangency of line t and the first circle is N .
a) Find the length of the segment O1 O2 .
b) Prove that the points M , K and N lie on the same line.
√
Answer: a) 2 2R.
10
R
Solution. a) The radius of the third circle is obviously ; let its centre be O3 (see
2
figure 8) and |O1 O2 | = c. Considering that the triangle O1 M O3 is a right triangle with
R 3R R c R 2 c 2 3R 2
the hypotenuse R + = and legs and , we get + = , from
2 2 2 √ 2 2 2 2
where R2 + c2 = (3R)2 , c2 = 8R2 and c = 2 2R.
A1
Np Pp t A9 A2
pO3
K p pL A8 A3
p p p s
O1 M O2
A7 A4
A6 A5
Figure 8 Figure 9
b) Let’s denote 6 P M K = α. From the isosceles triangle O3 M K we get 6 M KO3 = α and

6 KO3 M = π−2α. As the segments P M and N O1 are parallel the angle 6 KO1 N = π−2α
and from the isosceles triangle KO1 N we get 6 O1 N K = α. Therefore the segments M K
and N K parallel, that means the points N , K and M lie on the same line.
4. For which values of n (n > 3) is it possible to draw on a plane such a closed broken line
consisting of n links that every link has exactly one point in common with every other
link so that this point is an end point or an inner point for both links, and no point on
the plane is an end point for more than two links?
Answer: it is possible iff n is odd.
Solution. When n = 2k + 1 is odd we can get the necessary construction in the following
way. Let A1 , A2 , . . . , A2k+1 be the vertices of a regular (2k+1)-gon. We draw a segment
from every vertex Ai to vertices Ai+k (or A(i+k)−(2k+1) , if i + k > 2k + 1) and Ai−k (or
A(i−k)+(2k+1) , if i − k < 1). We can see that a closed broken line is obtained. Indeed, from
one side every vertex is connected to exactly two other vertices, from the other side it is
possible to reach any vertex from any other vertex, moving by links. To be convinced in
the latter assertion it is enough to notice that starting from the vertex Ai we can reach
the vertex Ai+1 using two links and hence continuing in the same way any other vertex.
In addition we have to prove that every link has exactly one common point with every
other link (either end point or inner point). Without loss of generality let’s consider the
link A1 Ak+1 . All the other links Ai Aj (i < j ) can be divided into three classes:
1) i = 1, j = k + 2; in this case these links have a common end point A1 ;
2) i = k , j = 2k + 1; in this case these links have a common end point Ak ;
3) 1 < i < k + 1, k + 1 < j < 2k + 1; in this case the vertices Ai and Aj lie on different
sides of the line A1 Ak and because of convexity of initial regular (2k+1)-gon the
links A1 Ak and Ai Aj have a common inner point.
An example for the case n = 9 is given on the figure 9.
Now let’s show that no broken line having an even number of links does not satis-
fy the requested conditions. Let’s denote the vertices of the broken line in the or-
der of passing by B1 , B2 , . . . , Bn . Let’s consider the line B1 B2 . As every segment
B3 B4 , B4 B5 , . . . , Bn−1 Bn has to intersect it, the vertices B3 , B5 , . . . , Bn−1 lie on one side
and the vertices B4 , B6 , . . . , Bn on the other side of that line. Therefore the segments
B2 B3 and Bn B1 can’t have common points.
11
5. Two palmists were asked several questions about the life of mr. X, each of which had to
22
be answered with yes“ or no“. The palmist A answered correctly to of questions
” ” 43 4
which were answered uncorrectly by the palmist B , palmist B answered correctly to
7
of questions which were answered uncorrectly by the palmist A. To how many questions
did the palmists A and B give the same answer, when the palmist A answered correctly
51% of all questions?
Answer: 50%.
Solution. As the palmist A answered uncorrectly to 49% of all questions there were
4
·49 = 28 per cent questions that were answered uncorrectly by A but correctly by B and
7
21% questions that both answered uncorrectly. On the other side, as A answered correctly
22
to of questions answered uncorrectly by B , then the part of questions answered
43
21
uncorrectly by both palmists was among questions answered uncorreclty by palmist
43
B . As both palmists answered uncorrectly 21% of all questions, the palmist B answered
uncorrectly 43% of all questions and there were 22% of questions that were answered
uncorrectly by B but correctly by A. So there were 28 + 22 = 50 pro cent of questions
that were answered differently by A and B and so their opinion coincided in another half
of questions.
Older group (11th–12th grade)
2. Let a be an integer. Find the all real solutions of the equation
[x] = ax + 1
where [x] denotes the integer part of x.
1
Answer: If a < −1 then x = − ; if a = 0 then every such real number x is suitable that
a
3 2
1 6 x < 2; if a = 2 then x = − or x = −1; if a > 2 then x = − ; if a = −1 or a = 1
2 a
then the equation has no solutions.
Solution. We will use the inequalities x − 1 < [x] 6 x holding for all real numbers x.
1) If a = 0 then [x] = 1 that is 1 6 x < 2.
2) Since [x] 6 x < x + 1, the equation has no solutions if a = 1.
3) Let a > 2. We will obtain the inequalities (a − 1) x > −2 and (a − 1) x 6 −1 that is
2 1
− <x6− .
a−1 a−1
2 1
Since − > −2 and − < 0, we get two possibilities if a = 2: [x] = −2 or
a−1 a−1
3
[x] = −1, the corresponding values of x are x = − and x = −1 (these are really
2
2
solutions). If a > 2 then the only possibility is [x] = −1 and x = − which satisfies the
a
equation.
12
4) Let a 6 −1. Then a − 1 < 0 and we obtain the inequalities
2 1
>x> .
1−a 1−a
1 2
Since > 0 and 6 1 then consequently [x] = 0. The possibility a = −1 gives
1−a 1−a
1
us a contradiction, if a < −1 then the solution is x = − .
a
3. A n × m-table filled with positive integers was written on the paper. John wrote after
each row the greatest common divisor of the numbers of this row and below each column
the least common multiple of the numbers of this column. Let a be the least common
multiple of the column of the greatest common divisors and let b be the greatest common
divisor of the row of the least common multiples. Prove that b is divisible by a.
Solution. Let cij denote the element of the ith row and the jth column. We fix the row
i and the column j arbitrarily. Then the least common multiplier of the numbers of the
jth column is divisible by cij and cij is divisible by the greatest common divisor of the
numbers of the ith row. Thus the number written below the jth column is divisible by the
number written after the ith row. Since j was chosen arbitrarily, the number written after
the ith row is the common factor of all the numbers written below the columns. Thus the
greatest common factor b of all the numbers written below the columns is divisible by
the number written after the ith row. Since i was chosen arbitrarily, the number b is the
common multiple for all the numbers written after the rows. Thus b is divisible by the
least common multiple a of all the numbers written after the rows.
4. On the conference of linguists there were n > 3 participants who could speak altogether
14 different languages. It is known that for every three linguists existed a language that
was spoken by all three. But every language was spoken by no more than a half of linguists.
Find the minimal possible value of n.
Answer: The minimal possible value of n is 8.
Solution. If n 6 5 then we choose an arbitrary triple and we get a language that is spoken
by at least 3 that is more than a half of the participants, contradicting the conditions of
the problem. If n = 6 or n = 7 then we choose arbitrary 6v linguists. It is possible to
form 20 different triples of the linguists. By pigeon-hole principle there exists a language
that is the same for at least two triples and therefore is spoken by at least 4 linguists.
This is again a contradiction because 4 is more than a half of 6 or 7. If n = 8 then it is
possible to find the construction satisfying the conditions of the problem:
1 A C E G I K M
2 A C E H J L N
3 A D F G I L N
4 A D F H J K M
5 B C F G J K N
6 B C F H I L M
7 B D E G J L M
8 B D E H I K N
(here the letters A, . . . , N denote the languages and the numbers 1, . . . , 8 denote the
participants).
13
Comment: Constructing the example for the case n = 8 we can use the following condi-
tions that have to be satisfied (why?):
a) Every language is spoken by exactly 4 linguists;
b) Every linguist can speak exactly 7 languages;
c) For every three linguists there is one and only one language spoken by all of them.
5. On the side BC of the triangle ABC a point D different from B and C is chosen so
that the bisectors of the angles ACB and ADB intersect on the side AB . Let D ′ be the
symmetrical point to D with respect to the line AB . Prove that the points C , A and D ′
are on the same line.
Solution. Let E be the intersection point of the bisectors of the angles ACB and ADB
and let 6 ACB = δ1 , 6 ADB = δ2 and 6 CBA = α.
D′
δ1 δ2
α
C B
D
By the property of the bisector for the triangles ABC and ABD we get
|AC| |AE| |AD|

= = ,
|BC| |BE| |BD|
and by the law of sines for the same triangles
|AC| sin α sin α

= ◦
= ,
|BC| sin(180 − α − δ1 ) sin(α + δ1 )
|AD| sin α sin α
= ◦
= .
|BD| sin(180 − α − δ2 ) sin(α + δ2 )
Consequently
sin α sin α
= ,
sin(α + δ1 ) sin(α + δ2 )
hence
sin(α + δ1 ) = sin(α + δ2 ).
There are two possibilities for this equation.
14
1) α + δ1 = α + δ2 , that is δ1 = δ2 . This is not possible because the point D is in the
interior of the segment BC .
2) α + δ1 + α + δ2 = π , that is δ1 + δ2 + 2α = π . Since the triangles ADB and AD ′ B
are congruent then 6 AD ′ B = 6 ADB = δ2 and 6 ABD ′ = 6 ABD = α and the point A
is on the side CD ′ of the triangle BCD ′ .
1. Let a be an integer, which square divided by n gives the remainder 1. Which remainder
can be obtained dividing the number a by n, if
a) n = 16;
b) n = 3k , where k is a positive integer?
Answer: a) 1, 7, 9 or 15; b) 1 or 3k − 1.
Solution. a) As the square of an even number is an even number and gives an even remain-
der dividing by 16 the number a must be an odd number. Examining three possible cases
we find that only 1, 7, 9 and 15 fit. It is possible to reduce the number of examinations
needed by noticing that the squares of the numbers i and 16 − i give the same remainder
by dividing by 16.
b) It’s clear that the squares of the remainders 1 and n − 1 give the remainder 1 modulo
n. Let’s assume that the number a2 gives the remainder 1 by dividing by n = 3k and
let’s prove that number a gives the remainder 1 or 3k − 1 by dividing by 3k . From the
assumption we get that 3k divides the number a2 − 1 = (a − 1)(a + 1), hence there exist
natural numbers i and j that i + j = k where 3i divides the number a − 1 and 3j divides
the number a + 1. If both i and j were positive then both numbers a − 1 and a + 1 would
be divisible by 3, but that’s impossible. Therefore one of the nest cases must occur: i = 0
and j = k , or j = 0 and i = k . Hence one of the numbers a − 1 and a + 1 is divisible by
3k , or in another way, a gives the remainder 1 or 3k − 1 by dividing by 3k .
Comment. The arguments in the part b) will still hold if number 3 is replaced by any
other odd prime number.
2. A polynomial an xn + an−1 xn−1 + . . . + a1 x + a0 is called alternating, when n > 1 and

for all i = 1, 2, . . . , n the coefficients ai and ai−1 are nonzero real numbes with different
signs. Let P (x) and Q(x) be arbitrary alternating polynomials. Prove that the polynomial
R(x) = P (x)Q(x) is alternating.
Solution. Let’s define α(x) = P (−x) and β(x) = Q(−x). Then α and β are polynomials
which coefficients all have same sign. Hence in the product γ of the polynomials α and
β all coefficients have the same sign as well (all coefficients are positive if the signs of
coefficients in α and β are same and negative if the signs of coefficients in α and β are
different). Therefore
R(x) = P (x)Q(x) = α(−x)β(−x) = γ(−x)
and the polynomial R is alternating.
3. Two right triangles are given, of which the incircle of the first triangle is the circumcircle
of the second triangle. Let the areas of the triangles be S and S ′ respectively. Prove that
S √
> 3 + 2 2.
S′
15
Solution. Let the lenghts of the legs of the outer triangle be a and b, the lenght of the
hypotenuse be c and the radius of its incircle be r (see figure 10). Then the length of
the hypotenuse of the inner triangle is 2r and the altitude h drawn to hypotenuse is not
1 ab a+b+c
greater than r . Hence S ′ = · 2rh = rh 6 r 2 . We know that S = = r , from
2 2 2
what using the inequality between arithmetical mean and geometrical mean we get
p √ √
ab = (a + b + c) · r = ((a + b) + a2 + b2 ) · r > (2 ab + 2ab) · r =
√ √
= (2 + 2) ab · r ,
√ √
e.g. ab > (2 + 2) · r . Squaring both sides of the equation and dividing by 2 we get
√ 1 1 S √
S > (3 + 2 2) · r 2 . Bearing in mind that ′ > 2 we see that ′ > 3 + 2 2.
S r S
c
· h
a r r
r
b
Figure 10
4. On n cells of an infinite squared board there is one piece on each cell. If one of the four
neighbours of the cell containing the piece A contains the piece B and the cell behind it is
empty, the piece A can be moved over the piece B to the empty cell behind it. Does there
exist such a combination of pieces from which it is possible in finite number of moves to
obtain the situation where the final combination of pieces is the same as in the beginning
but moved by one cell in any direction, if
a) n = 1999;
b) n = 2000;
c) n = 1998?
Answer: a) no; b) yes; c) no.

Solution. a) Let’s colour the cells black and white as in the chessboard. As the number
1999 is odd, in the initial combination there are on cells of one colour (for example white)
more pieces than on cells of another colour (for example black). As from one side moving
the pieces as required in the problem we can move each piece only on cells coloured with
same colour but from another side in the final combination we should have more pieces
on black cells as on white cells we get a contradiction.
b) It’s easy to see that we can move a block consisting of 2 × 2 pieces by one cell in any
direction. We can get a suitable initial combination placing 2000 pieces on the squared
board in 500 blocks as shown in the figure 11.
c) Let’s colour the cells with four colours as shown in the figure 12. As the number 1998
is not divisible by four there will be on cells of some colour (for example red) more pieces
than on cells of some other colour (for example blue). As we can move each piece only on
cells of same colour we get a contradiction as the problem requires that we should be able
to move the pieces so that in final configuration there are more pieces on blue cless than
on red cells.
16
1 2 ... 500 1 2 1 2 1 ···
3 4 3 4 3 ···
w w w w ... w w ···
1 2 1 2 1
w w w w ... w w .. .. .. .. .. ..
. . . . . .
Figure 11 Figure 12
5. Inside the square ABCD there is the square A′ B ′ C ′ D ′ so that the segments AA′ , BB ′ ,
CC ′ and DD ′ do not intersect each other neither the sides of the smaller square (the
sides of the larger and the smaller square do not need to be parallel). Prove that the sum
of areas of the quadrangles AA′ B ′ B and CC ′ D ′ D is equal to the sum of areas of the
quadrangles BB ′ C ′ C and DD ′ A′ A.
Solution. When the centres of smaller and larger square coincide then the quadrangles
AA′ B ′ B , BB ′ C ′ C , CC ′ D ′ D and DD ′ A′ A are congruent and the assertion of the problem
holds. Therefore it is enough to show that the sum of areas of the quadrangles AA′ B ′ B
and CC ′ D ′ D does not change by any parallel displacement of smaller square (with this
displacement we alwyas can make the centres of the squares coincide).
A B
A′
D′
B′
C′
D C
Figure 13
Without loss of generality we can assume that the vertex A′ of the smaller square is not
further off the side AB than the vertex B ′ (otherwise we can change the labels of the
vertices A and B ; C and D ; A′ and B ′ ; C ′ and D ′ ). Then the diagonal A′ B lies inside
the quadrangle AA′ B ′ B and divides it into two triangles ABA′ and A′ B ′ B (see figure 13).
Also the vertex C ′ lies not further off the side CD than the vertex D ′ , because of what the
diagonal C ′ D lies inside the quardangle CC ′ D ′ D and divides it into two triangles CDC ′
and C ′ D ′ D . The areas of triangles ABA′ and CDC ′ does not change by any parallel
displacement of smaller square parallel to the side AB of the larger square (because their
basis AB and CD and the altitudes drawn on them don’t change) and the sum of the
areas of these triangles doesn’t change by any parallel displacement of smaller square
parallel to the side BC of the larger square (because their basis and the sum of their
altitudes don’t change). As every parallel displacement can be done by two displacement
perpendicular to each other so the sum of the areas of the triangles ABA′ and CDC ′
doesn’t change by any parallel displacement of smaller square inside the larger square.
Analogically we can see that the sum of areas of triangles A′ B ′ B and C ′ D ′ D doesn’t
change by parallel displacement parallel to any side of smaller square (because their basis
A′ B ′ and C ′ D ′ and the sum of their altitudes don’t change). So the sum of the areas of
the quardangles AA′ B ′ B and CC ′ D ′ D does not change by any parallel displacement of
the smaller square inside the larger square.
17
Estonian math ompetitions 2000/2001 2. Find the largest real number K having the following property: for any positive
real numbers a; b; satisfying the inequality a b 6 K , the inequality ab 6 K
also holds.
+ +
We thank the IMO ommunity for many of these problems whi h have been taken
from various materials distributed at the re ent IMO-s. 3. Prove that, for any integer n > , the number | {z: : : } is divisible by , but is
0 11 1 3
n
3 digits n
not divisible by +1 . 3
n
Autumn Open Contest: O tober 2000
4. The terms of the sequen e a1; a2; a3; : : : satisfy the ondition a a 1 a 2
for any n > . Find the sum of the rst terms of this sequen e, if the sum
n = n n
of the rst terms is and the sum of the rst terms is .
3 2000
Juniors (up to 10th grade)
1997 2002 2002 1997
1. How many positive integers less than and not ontaining digits other 5. On a plane n points are given, no three of them ollinear. At most how many line
than and are there? segments it is possible to draw between these points in su h a way that the line
20002001
segments form no triangle with verti es at the given points?
0 2
2. Find the two last digits of the number :::
1! + 2! + 3! + .
+ 2000!
3. Consider points C1 , C2 on the side AB of a triangle ABC , points A1 , A2 on Solutions of Autumn Open Contest
the side BC and points B1 , B2 on the side CA su h that these points divide the
orresponding sides to three equal parts. It is known that all the points A1 , A2 , J1. .
B1 , B2 , C1 and C2 are on y li . Prove that triangle ABC is equilateral. Answer: 136
The set of integers under onsideration onsists of all integers with up to digits
4. Real numbers x and y satisfy the system of equations ontaining only digits and , all -digit integers of the form and the
7
integer . There are 1 integers with exa tly k digits and , and 3
0 2 8 20000
k
integers of the form . So the required number of integers is
8 20002000 2 0 2 2
> x
<x+y + y
>
> = 10
20000
: (2
0 + 21 + : : : + 26) + 8 + 1 = (27 1) + 9 = 136 :
>
> x(x + y)
>
:
y
= 20
J2. Answer: 13 .
Find the sum of all possible values of the expression x y . The produ t : : : has , and as fa tors, therefore being divisible by
. Hen e the last two digits of n are zeros for any n > and it su es to
1 2 10 2 5 10
+
nd two last digits of . The two last digits of the summands are
100 ! 10
5. Let m m for any odd integer m and m m for any even integer m .
= +3 =
01 02 06 24 20
:::
, , , , , , , and , yielding as the answer.
20
1! + 2! +
40 20
+ 9!
80 13
a) Find all integers k su h that k .
2
b) Prove that, for every odd integer K , there exist pre isely three dierent
= 1
C
integers k su h that k K .
) How many dierent integers k with the property k K exist for an even
=
B1 A2
integer K ?
= r r
B2 r
A1
r
r r
Seniors (grades 11 and 12) A C1 C2 B
Figure 1
1. Points A , B , C , D , E and F are given on a ir le in su h a way that the three
hords AB , CD and EF interse t in one point. Express angle EF A in terms of J3. Label the points on the sides of the triangle so that jAC1j jC1C2j jC2Bj ,
angles ABC and CDE (nd all possibilities). jBA1 j jA1 A2j jA2 C j and jCB1 j jB1 B2 j jB2 Aj (see Fig. 1). Then we have
= =
= = = =
1 2
6 BA C 6 BA C 6 BCA and 6 BC A 6 BC A 6 BAC . Sin e points
1 2 2 1 2 1 1 2 In ases (a) and (b) similar arguments give 6 EF A 6 CDE 6 ABC and
A1 , A2 , C1 and C2 are on y li , we get 6 BA2 C1 Æ 6 AC2A1 6 BC2A1 , 6 CDE , respe tively.
= = = = =
6 EF A = 6 ABC
whi h gives 6 BCA 6 BAC . The equality 6 BAC 6 CBA follows by symme-
= 180 =
try.
= =
E F
J4. Answer: 10 . A C A C A C A C
By Viete's theorem, the possible values of x y are in luded in the set of roots
of the quadrati equation E F
+
F E
a2 10a + 20 = 0 : D B D F
B D B D E
B
This equation has two dierent roots be ause D 2 > . Viete's (a) (b) ( ) (d)
formulae give to be the sum of these roots. It remains to he k that is not
= 10 4 20 0
10 11
among the roots (as y x x y y from the rst equation, x y 6 enables
=
+
+ = 11 Figure 2
us to nd the orresponding values for x and y ).
11 ( + )
Consider ase (d). Sin e EF A and ADE are opposite angles of a y li quadri-
J5. Answer: a) , and ; ) .
1 2 8 5
lateral ADEF and 6 ADE 6 ADC 6 CDE 6 ABC 6 CDE , we have
= + = +
a), b) Observe that if m is odd, then both m and m are even. Hen e if 6 EF A = 180Æ 6 ADE = 180Æ 6 ABC 6 CDE :
K k is odd, then k K , and k and k are not both odd. This gives the p
following three possibilities.
= = 2
S2. Answer: 3 3 .
1) If both k and k are even, then k k k K . Let a b+ 6 K. By the AM-GM inequality we have
2) If k is odd and k is even, then k k K .
= 2 = 4 = 8 +
k
3) If k is even and k is odd, then k k k K .
a + b + 3 3 K2
= 3 = (2 3) = 4 3
K = 2 = 2 ( 3) = 2 (2 3) = 4 6
ab 6 6 K K :
The numbers K , K and K are pairwise distin t sin e modulo they
=
3 3 27
are ongruent to , and , respe tively. For a), K gives k 2 f ; ; g .
8 4 3 4 6 4
Hen e if K 6 , or equivalently K 6 p , the required ondition is satis-
0 1 2 = 1 1 2 8 2
) Let now K be even. If k is even, then we get the same three possibilities for 1 3 3
k as above. If k is odd, then k is even and k an be either even or odd.
ed. However, if K > p and a b K , then a b K and
27
4) If k is even, then k k k K K .
5) If k is odd, then k k K .
= 2 = 4 = 4( 3) = 4 12 3 3 = = = + + =
k K
3
K2
> K , so the ondition is not satised.
= 3 = 2 3 = 2( 3) 3 = 2 9
Sin e K is even, the numbers K , K , K and K are ongruent to ab K
=
, , and , respe tively, modulo . Moreover, K is ongruent to either
8 4 3 4 6 4 12
27
or modulo . Hen e these ve numbers are pairwise distin t. S3. We use indu tion on n .
0 5 2 4 8 2 9
3 7 8
S1. Angle EF A is equal to either 6 ABC 6 CDE , or 6 ABC 6 CDE , or The proposition holds for n sin e is divisible by 0 and is not
divisible by 1 .
Base: = 0 1 3 = 1
6 CDE 6 ABC , or Æ 6 ABC 6 CDE .
Answer: +
3 = 3
180
Given the hords AB and CD , the hord EF an be drawn in four essentially Step:Observe the equality
dierent ways point E an lie on the ir le between points D and A , between 11 : : : 1 = 11 : : : 1 1 00 : : : 0 1 00 : : : 0 1 :
points A and C , between points C and B or between points B and D (see | {z } | {z } | {z } | {z }
Fig. 2). 3n+1 3n 3n 1 3n 1
Let us nd 6 EF A for ase ( ). Sin e EF C and CDE are angles subtended The rst fa tor here is divisibleby but not by +1 by the indu tion hypothesis,
n n
by the same hord EC , we have 6 EF C 6 CDE ; similarly 6 CF A 6 ABC . and the se ond fa tor is divisible by but not by . Sin e is prime, this implies
3 3
Hen e that the produ t is divisible by +1 , but not by +2 .
= = 3 9 3
n n
3 3
6 EF A = 6 CF A + 6 EF C = 6 ABC + 6 CDE : S4. Answer : .
2012
3 4
Denote a1 p and a2 q . It is easy to see that
= = Spring Open Contest: Mar h 2001
a p if k ; ; ; : : :;
a q if k ; ; ; : : :;
k = = 1 7 13
a q p if k ; ; ; : : :;
k = = 2 8 14 Juniors (up to 10th grade)
p if k ; ; ; : : :;
k = = 3 9 15
a k =
q if k ; ; ; : : :;
= 4 10 16
1. Eight students, Anne, Mary, Cathy and Tina, Anthony, Mark, Carl and Tom have
a
a p q if k ; ; ; : : :.
k = = 5 11 17
to work in four pairs, one boy and one girl in ea h pair. They know ea h other,
k = = 6 12 18
with only these ex eptions: Anthony knows neither Anne nor Mary; Mark doesn't
Observe that the sum of any six onse utive members of the sequen e is equal to know Mary and both Carl and Tom know neither Cathy nor Tina. How many
zero. Denoting S a1 : : : a , we get ways are there to divide the students into pairs, so that ea h boy ould work with
a girl he knows?
k = + +
S p if k ; ; ; : : :;
k
S p q if k ; ; ; : : :;
k = = 1 7 13
S q if k ; ; ; : : :;
k = + = 2 8 14
2. In a triangle ABC , the lengths of the sides are onse utive integers and median
q p if k ; ; ; : : :;
k = 2 = 3 9 15
drawn from A is perpendi ular to the bise tor drawn from B . Find the lengths
S
S q p if k ; ; ; : : :;
k = 2 = 4 10 16
of the sides of triangle ABC .
if k ; ; ; : : :.
k = = 5 11 17
S k = 0 = 6 12 18
3. In a s hool lo ker room there are lo kers
Hen e q p S1997 and q p S2002 , whi h give q and in three rows. The lo kers in ea h row are la-
60
K
with S2000 p q . belled from left to right with numbers to
= = 2002 2 = = 1997 = 5
p L
in the top row, to in the middle row and M
= 2007 = + = 2012 1 20
to in the bottom row.
21 40
n2 2
S5. for even n and n for odd n . 1 41 60
Kate's, Lisa's and Mary's lo kers are lo ated as shown in the gure. Ea h of the
Answer:
Divide the points into two subsets with ardinalities as lose to ea h other as
4 4
possible, and draw a line segment between any two points from dierent subsets. three lo ker numbers is divisible by the number of Mary's house, whi h is not . 1
Then ea h losed line formed by these line segments ontains an even number a) What is the number of Mary's house?
of links and hen e the line segments do not form any triangles2 with verti es at b) What ould be the numbers on the girls' lo kers?
the given points. The number of line segments is n 2 n for even n and

4. Integers a , b , and d satisfy ja bdj jad b j . Prove that either
jaj jbj or j j jdj .
= + = + = 1
2 4
n2
n
n for odd n .
= = 1 = = 1
1 + 1 1
5. A onvex hexagon is onstru ted from n pie es, ea h of whi h
=
Now prove that there annot be more line segments. Consider any olle tion of line
2 2 4
segments satisfying the onditions of the problem. Let m be the maximal number is an equilateral triangle (one example is given in the gure).
of line segments in ident to one point, and let X be any point in ident to m line a) Prove that the hexagon is equiangular.
segments. Let A be the set of the other endpoints of these m line segments, and b) Find all possible values of n .
B be the set of the other n m points (in luding X ). Ea h point of A an be
joined only to points of B be ause any two joined points from A together with
X would form a triangle. Hen e ea h of the m points of A o urs as an endpoint Seniors (11th and 12th grade)
for at most n m line segments. On the other hand, ea h of the n m points
from B o urs as an endpoint for at most m line segments by the hoi e of m . 1. The serial numbers of lottery ti kets are -digit integers. It is known that the
So there is at most m n m n m m m n m segment-endpoint pairs, serial number of a winning ti ket has seven distin t digits and is divisible by ea h
7
and sin e every line segment has two endpoints, we have at most m n m line
( )+( ) = 2 ( )
segments. It remains to noti e that this expression a hieves its maximum when
( )
of its digits.
the dieren e of m and n m is as small as possible, i.e. if m n m for even a) Prove that the serial numbers of all winning ti kets onsist of the same digits.
n and if jm n m j for odd n . b) Find the largest possible serial number of a winning ti ket.
=
( ) = 1
5 6
2. Let us all a onvex hexagon ABCDEF if 6 A 6 C 6 E 6 B 6 D 6 F . From the gure we obtain L K and M L K . Sin e
K , L and M are all divisible by the number of Mary's house n , the dieren es
boring + + = + + = + 21 = + 14 = + 35
a) Is every y li hexagon boring? and M L are also divisible by n . It follows that the only
b) Is every boring hexagon y li ? L K
possible value of n is . Now, sin e 6 K 6 , we obtain K or K ,
= 21 = 14
3. Find all real-valued fun tions f x dened for all real numbers whi h satisfy the and hen e L and M or L and M .
7 1 20 = 7 = 14
ondition f x f x2 for ea h real x .
( ) = 28 = 42 = 35 = 49
2001 + (0) = 2001
J4. If the numbers a bd and ad b have the same sign, then a bd ad b
and a bd ad b a b d . Hen e a b or d . If the numbers
+ + + = +
4. For some < x; y < , two of the three expressions 2 x 2 y , 2 x y a bd and ad b have distin t signs, then
0 = + = ( )( ) = =
and have equal values and the third one is dierent.
0 sin + sin sin ( + )
a bd ad b a b d
and a b or d . In both ases jaj jbj or j j jdj . If jaj jbj , then
+ + 0 = + + + = ( + )( + )
1
a) Whi h of the three expressions has a dierent value? ja bdj is divisible by jaj , therefore jaj jbj . In ase j j jdj , we
= = = = =
b) Give an example of x and y for whi h su h a situation o urs. similarly obtain j j jdj .
1 = +
= = 1
= = 1 =
5. There are small boxes numbered from to , and one large box. John puts J5. b) all integers n > .
some balls in some (or all) of the small boxes, and starts relo ating them by the
10 1 10
Answer: 6
following rules: a) Let a vertex of the hexagon be the vertex of k triangles. Then the interior
during ea h move, John removes all balls from any small box numbered n
angle at this vertex must be k Æ . Sin e the interior angles of a onvex hexagon
are less than Æ , none of them an be larger than Æ . Sin e the sum of the
60
where the number of the balls equals n ; angles is Æ Æ , it follows that all the angles are equal to Æ .
180 120
he adds these balls into boxes to n (one ball into ea h box) and puts
720 = 6 120 120
the remaining ball into the large box.
1 1
b) There must be at least pie es, be ause there is at least one triangle on ea h
side of the hexagon, and sin e the hexagon has no angles equal to Æ , ea h
6
He ontinues this way until he annot make another move a ording to these rules. triangle an lie on only one side of the hexagon.
60
Find the largest possible total number of balls in the small boxes at the beginning
of the game, for whi h it is possible to put all balls in the large box by the end of
the game.
Solutions of Spring Open Contest
n=6 n=7 n=8
J1. Answer : . 4 Figure 3
It is lear that Carl and Tom an only work with Anne and Mary: we obtain two The onstru tions for n , n ja n are given in Figure 3. We an
ways to form two pairs. Now, Anthony and Mark have to work with Cathy and ontinue the same way, adding more large triangles in the middle.
= 6 = 7 = 8
Tina, whom they both know: there are also two ways to form the two remaining
pairs. Hen e, altogether there are four ways to form the pairs. S1. Answer: b) .
9867312
J2. , and . a) Obviously, the serial number annot ontain and must be even. Now, we
annot have among the digits, be ause every even number divisible by ends
Answer: 2 3 4 0
Let D be the midpointof BC , then the medianis AD . Sin e the bise tor of 6 B is with a . If the serial number didn't have as one of its digits, it would ontain
5 5
also an altitude in the triangle ABD , that triangle is equilateral, i.e. jBDj jBAj and should be divisible by , but the sum of the remaining seven digits is ,
0 9
and in the original triangle ABC we have jBC j jABj . Sin e the lengths of
=
ontradi tion. Hen e is one of the digits. Now, the serial number is divisible by
3 3 31
the sides of triangle ABC are onse utive integers, the dieren e jBC j jABj is
= 2
and the sum of its digits is between and . The only possible sum is
9
either or . In the rst ase jABj , jBC j and the length of the side and the serial number onsists of the digits ; ; ; ; ; ; .
9 32 39 36
AC must be either or , whi h is impossible. In the se ond ase we obtain
1 2 = 1 = 2
1 2 3 6 7 8 9
jAB j , jBC j and jAC j .
0 3
b) Any -digit number onsisting of these digits is divisible by , , and .
Now we must nd the largest possible even number among these whi h is divisible
= 2 = 4 = 3 7 1 3 6 9
J3. Answer: a) ; b) ; ; or ; ; .
7 7 28 42 14 35 49 by 7 . This number is
8 = 56 . 9867312
7 8
S2. Answer: a) yes, b) no. Sin e x sin

x and

x = os

x sin

x , then
3

= sin

+

= os
in both ases 2 x 2 y 2 x 2 x .
2 2 2
B sin + sin = sin + os = 1
A ? Let 2 x 2 y 2 x y . We shall prove that both of these are equal to .
Applying the formula for x y and squaring, we get
sin + sin = sin ( + ) 1
sin( + )
C sin
2 x + sin2 y = sin2 x os2 y + sin2 y os2 x + 2 sin x sin y os x os y ;
D sin
2 x (1 os
2y) + sin2 y (1 os
2x) 2 sin x sin y os x os y = 0 ;
2 x sin2 y x sin y os x os y = 0 :
E F 2 sin 2 sin
Figure 4 Figure 5 Sin e < x; y < , neither x nor y equals . Hen e we must have
, i.e. x y and x y equals to either
0 sin sin 0
x y x y
or , when e 2 x y . So the only expression that an have a dierent
sin sin os os = 0 os( + ) = 0 sin( + ) 1
a) Let ABCDEF be a y li hexagon. Sin e the quadrilaterals ABDF , CDF B
and EF BD are also y li (see Fig. 4), we obtain value is 2 x y .
1 sin ( + ) = 1
sin ( + )
6 BDF = 6 A; 6 DF B = 6 C; 6 F BD = 6 E: b) Taking < x < and y x , we get
0 = +
Now, 6 A 6 C 6 E , and hen e 6 A 6 C 6 E .
2 2
Thus 6 B 6 D 6 F 6 A 6 C 6 E , whi h proves that the 2 x + sin2 y = sin2 x + sin2 x +

( ) + ( ) + ( ) = + + = 2
2 x + os2x = 1 :
hexagon ABCDEF is boring.
+ + = 4 2 = 2 = + + sin = sin
2
b) Let us ompress a regular hexagon along its two opposite sides (see Fig. 5). Sin e < y < x y x < , we have 2 x y 6 .
The new hexagon is boring sin e all its angles are equal, but it is not y li sin e + = +2
3
sin ( + ) = 1
three of its verti es lie on one ir le and the rest on another ir le. 2 2 2
x2 2 S5. .
S3. and f x x .
Answer: 41
fx
( 2001)
We shall rst show that the total number of balls in the small boxes annot ex eed
. John an empty box number only on e, sin e no balls are put into it during
Answer. ( ) = ( ) =
2001 2001
the relo ations. He an also empty box only on e, sin e at most one ball is put
41 10
Sin e for any real number y there is an x y f su h that y = x f ,
2001
(0)
= 2001 + (0)
into it (from the tenth box). Also, boxes , and an be emptied only on e.
9
2
Box number an be emptied at most twi e (at most balls will be added to it
8 7 6

the equality f y y f
( ) = 2001 holds for any real y . Taking y we
2001
(0)
= 0
from boxes with bigger numbers). Box an also be emptied at most twi e, box
5 5
at most times, box at most times and box at most times. John an
4
f 2

get f (0) = and hen e f
(0)
or f . Therefore the fun tion
(0) = 0 (0) = 2001
therefore put no more than balls in the large box.
3 4 2
41
7 1 21
We shall now nd a way to pla e balls in the smallboxes, so that all boxes ould
2001
2 2
must be either f y y or f y y
( ) = . It is easy to he k that both
( ) =
( 2001)
be emptied. No balls are added to box , therefore it must ontain balls. One
41
of these satisfy the given onditions. ball will be added to box , therefore it must ontain balls in the beginning.
2001 2001 10 10
Similarly there must be , and balls in boxes , and respe tively. Sin e
9 8
S4. a) 2 x y an be the only expression with a dierent value; b) balls are added to box , it must ontain balls in the beginning in order to
6 4 2 8 7 6
be emptied twi e. Box must ontain one ball, box three balls, boxes and
Answer: sin ( + ) 5 5 5
hoose < x < arbitrarily and take y x .
0
2
= +
2 must ontain ball. The number of balls in the small boxes is now . It is easy
4 3 2 1
a) Let 2 x y . We shall prove that in this ase also 2 x 2 y . to he k that if John always empties the box with the smallest possible number,
1 41
all balls will nally be in the large box.
sin ( + ) = 1 sin + sin = 1
From the equation 2 x y we have either x y or x y .
sin ( + ) = 1 + =
2
+ =
3
2
9 10
Final Round of National Olympiad: Mar h 2001 (1) A is a word;
(2) if w is a word, then ww and ww are also words, where w is obtained from
w by repla ing all letters A with B and all letters B with A ( xy denotes the
9th grade
on atenation of x and y );
1. John had to solve a math problem in the lass. While leaning the bla kboard, he (3) all words are reated by rules (1) and (2).
a identally erased a part of his problem as well: the text that remained on board Prove that any two words with the same number of letters dier exa tly in half of
was x , where marks an erased digit. Show that John an their letters.
still solve his problem, knowing that x is an integer.
37 (72 + 3 ) = 14 45
2. Dividing a three-digit number by the number obtained from it by swapping its 11th grade
rst and last digit we get as the quotient and the sum of digits of the original
number as the remainder. Find all three-digit numbers with this property.
3
1. The angles of a onvex n -gon are ; ; : : : ; n . Find all possible values of n
and the orresponding values of .
2
3. A ir le of radius is tangent to two adja ent sides of a square and interse ts
its two remaining sides at the endpoints of a diameter of the ir le. Find the side
10
length of the square. 2. A student wrote a orre t addition operation BA DC FE to the bla kboard, su h
+ =
4. It is known that the equation jx j jx j : : : jx j a has exa tly that both summands are irredu ible fra tions and F is the least ommon multiple
one solution. Find a . of B and D . After that, the student redu ed the obtained sum EF orre tly by
1 + 2 + + 2001 =
5. A table onsisting of rows and olumnsis lled with integers ; ; : : :; an integer d . Prove that d is a ommon divisor of B and D .
in su h a way that ea h of these integers o urs in the table exa tly times and
9 2001 1 2 2001
the integers in any olumn dier by no more than . Find the maximum possible
9
3. Points D , E and F are taken on the sides BC , CA , AB of a triangle ABC ,
value of the minimal olumn sum (sum of the numbers in one olumn).
3
respe tively, so that the segments AD , BE and CF have a ommon point O .
Prove that jjOD
AOj jAE j jAF j
j jEC j jF B j .
= +
10th grade 4. Let x and y be non-negative real numbers su h that x y . Prove that
x2y2 x2 y2 6 .
+ = 2
1. A onvex n -gon has exa tly three obtuse interior angles. Find all possible values
( + ) 2
of n . 5. ConsiderÆall trapezoids in a oordinate plane with interior angles of Æ , Æ , Æ
and , su h that their bases are parallel to one of the oordinate axes and
90 90 45
2. Find the minimum value of n su h that, among any n integers, there are three all verti es have integer oordinates. Dene the of su h a trapezoid as the
135
whose sum is divisible by . total number of points with integer oordinates inside and on the boundary of the
size
trapezoid.
3
3. There are three squares in the pi ture. Find the sum of D
a) How many pairwise non- ongruent su h trapezoids of size are there?
angles ADC and BDC . b) Find all positive integers not greater than that do not appear as sizes of
2001
any su h trapezoid.
50
A B C
4. We all a triple of positive integers a; b;
( ) harmoni if a b . Prove that,
1
+
1
=
1
12th grade
for any given positive integer , the number of harmoni triples a; b; is equal
to the number of positive divisors of 2 .
( )
1. Solve the system of equations
5. A tribe alled Ababab uses only letters A and B, and they reate words a ording
x y
to the following rules: x :
sin =
sin y =
11 12
2. Find the maximum value of k for whi h one an hoose k integers out of 3) If , then 6 a b 6 363 whi h implies a > , a ontradi -
; ; : : :; n so that none of the hosen integers is divisible by any other hosen tion.
= 3 300 32 = 200 + 7 10
integer.
1 2 2
p
9-3. .
3. Let I and r be the midpoint and radius of the in ir le of a right-angled triangle
Answer: 10 + 5 2
ABC with the right angle at C . Rays AI and BI interse t the sides BC and
AC at points D and E , respe tively. Prove that 6B C
jAE j jBDj r .
1 1 1
+ =
4. Prove that, for any integer a > , there is a prime p su h that a a2 : : : a 1 p
O
q
is omposite.
1 1+ + + +
5. Consider a table, lled with real numbers in su h a way that ea h number -
A
in the table is equal to the absolute value of the dieren e of the sum of numbers
3 3
in its row and the sum of numbers in its olumn. Figure 6
a) Prove that any number in this table an be expressed as a sum or a dieren e Introdu e a oordinate system where the sides of the square tangent to the ir le
of some two numbers in the table. are on the oordinate axesthen the entre of the ir le is O ; (see Fig. 6).
b) Show that there exists su h a table with numbers in it not all equal to 0. Let the side of the square be a (evidently a > ) and the interse tion points of
(10 10)
the ir le with its two other sides be A and B . As AB is the diameter of the
10
ir le, the ommon point C a; a of these two sides lies on the ir le. Sin e CO
p
is a radius,pwe obtain p a 2 a 2 , giving a
( )
Solutions of Final Round and
.
( 10) + ( 10) = 10 10 = 5 2
a
9-1. x . = 10 + 5 2
9-4. .
Answer: = 1271
From the given equality we obtain x) = . To nd the number
y = 24 + x , note that
Answer: 1001000
Note that if x is a solution of the equation, x is also a solution. For
111(24 + 14 45
uniqueness we have x x , or x . In this ase
2002
< 1445 < 222000 = 111 2000 ;
= 2002 = 1001
111 1000 = 111000
a 1000 + 999 + : : : + 2 + 1 + 0 + 1 + 2 + : : : + 999 + 1000 =
therefore y is a 4-digit number, with as its rst digit. Evidently y must end =
with . Let y ab , where 6 a; b 6 . Writing out the multipli ation we
1
= (1000 + 1) + (999 + 2) + : : : + (2 + 999) + (1 + 1000) =
see that b ends with , hen e b and there is a arry of at least from
5 = 1 5 0 9
= 1000 1001 = 1001000 :
the third position. Sin e there is no arry to the rst position, we have a 6 .
+ 5 4 = 9 1
If the arry from the third position were more than , we would have a > , a
2
ontradi tion. Hen e a and x .
1 8
Although this is not required in the problem, it an be veried that
is indeed the only solution of the equation for a .
Remark.
= 2 = 1295 24 = 1271
x = 1001 = 1001000
9-2. Answer: 441 and . 882
9-5. 24.
We look for a number ab su h that ab ba a b , or a b . Answer:
The numbers 1 an be in the same olumn only with numbers 2, 3 and 4. As there
Hen e 6 6 , and we have 3 ases.
= 3 +( + + ) 32 = 100 + 7
1
1) If , then 6 a
3
whi h implies 6 a 6 . If a , are altogether of these, the 1-s an be at most in four olumns. If all 1-s are
b6 in the same olumn, the minimal olumn sum is 9. If the 1-s are in two olumns,
4 9
then b and b . If a , then b and b is not an
= 1 100 32 = 100 + 7 163 4 5 = 4
integer.
128 = 100 + 7 = 4 = 5 160 = 100 + 7
one of these must ontain at least 5 of them and the sum of this olumn is at most
2) If , then 6 a whi h implies 6 a 6 . If a , . If the 1-s are in four olumns, then the sum of all numbers in
these olumns is , hen e the minimal olumn sum is at most
5 1 + 4 4 = 21
b6
then b and b is not an integer. If a , then b , giving
= 2 200 32 = 200 + 7 263 7 8 = 7 9 (1 + 2 + 3 + 4) = 90
h 90 i
b = 8.
224 = 200 + 7 = 8 256 = 200 + 7
4
. If the 1-s are in three olumns, we should have 3-s and 4-s in these
= 22
13 14
olumns to obtain the largest olumn sum. In this ase the sum of numbers in the right-angled triangles, we have
three olumns is and the minimal olumn sum is at most 24.
From the table below we see that this value is indeed attainable.
9 (1 + 3 + 4) = 72
6 ADF = 6 ADC 6 F DG = 6 ADC 6 DF G =
1 1 1 2 2 6 7 ... 2001
2

6 ADC 6 BDC ;
1 1 1 2 2 6 7 ... 2001 =
1 1 1 2 2 6 7 ... 2001
2
3 3 3 2 2 6 7 ... 2001 that implies 6 ADC 6 BDC 6 ADF . The segments AF and DF are
3 3 3 2 5 6 7 ... 2001
+ = +
transformed into ea h other by a Æ rotation around F . Hen e AF D is an
2
3 3 3 5 5 6 7 ... 2001 90
4 4 4 5 5 6 7 ... 2001 isos eles right-angled triangle with 6 ADF , yielding 6 ADC 6 BDC . 3
4 4 4 5 5 6 7 ... 2001
= + =
4 4
4 4 4 5 5 6 7 ... 2001 Remark: There are also solutions using the osine theorem or the identity
10-1. Answer: The possible values of n are , and . 4 5 6

tan( +

) =

.
tan + tan
The sum of the angles of a n -gon is n . Sin e three of these angles are 1 tan tan
10-4. As a and b are non-zero integers, we have
( 2)
greater than and less than , and the remaining n angles are greater than
2
3
0 and less or equal to , we obtain n < n < n .
2
( 3) 0 + 3
2
( 2) ( 3)
2
+3
1
a b
+ () a ab b
1
=
1 +
=
1

() ab a b ()
= ( + )
Dividing by and transforming yields < n < . As n is an integer, we have
7
7 () ab a b = 0 () ab a b 2 2 () + =
6 n 6 , and it is easy to he k that all these three values are indeed possible. () a b 2:
2
( )( ) =
4 6
10-2. Answer: n .= 5 Now let a b . If a and b are positive, then a > and b > . On
1
+
1
=
1
0 0
The sum of any three integers ongruent to , and modulo is divisible the other hand, if a > and b > , then a and b are positive. Hen e the
by . Also, the sum of any three integers ongruent to ea h other modulo is
0 1 2 3
harmoni triples a; b; are in one-to-one orresponden e with pairs of positive
0 0
divisible by . Consequently, among any ve numbers there are three whose sum
3 3
integers r; s , where rs 2 , and there are as many su h harmoni triples as
( )
is divisible by . On the other hand, among the numbers , , and there are
3
there are positive divisors of 2 .
( ) =
no three with a sum divisible by .
3 1 3 4 6
3
10-5. We use indu tion on the length of a word. Let u1 and u2 be any dierent words
D of the same length, and suppose the laim holds for all shorter words. As there is
only one word of length 1, u1 and u2 are onstru ted by rule (2). This implies
that there exist words v1 and v2 so that u1 v1v1 or u1 v1v1 and u2 v2v2
B or u2 v2v2 . Note that v1 and v2 are of the same length. If v1 v2 v , then
= = =
A C
one of the words u1 and u2 is vv and the other vv , diering exa tly in half of
= = =
their letters. If v1 6 v2 , then v1 and v2 dier exa tly in half of their letters by
the indu tion hypothesis, and it remains to show that the latter halves of u1 and
=
F G
u2 also dier exa tly in half of their letters. If these halves are v1 and v2 or v1
Figure 7 and v2 , this is obviously true. The words v1 and v2 , as well as v2 and v1 , dier
exa tly in the letters where v1 and v2 oin idediering therefore also exa tly
10-3. Answer:

3
. in half of their letters. Hen e in any ase u1 and u2 dier in half of their letters.
Consider points F and G as shown on Fig. 7. As BCD and DGF are ongruent
4
11-1. Answer: The only possibilities are n , and n , . = 3 =
6
= 4 =
5
15 16
Obviously n > . As the sum of angles of the n -gon is n n n ,
3
( + )
= ( 2)
11-4. Denote = 1 x , then x = 1 and from x + y=2 we get y = 1+ . Now
x2 y2 (x2 + y2 ) )2 (1 + )2 ((1 )2 + (1 + )2 ) =
2
we have n nn . Be ause of onvexity, we have n nn <
=
2 (
( + 1)
2)
=
2 (
+1
2) =
=
(1
((1 )(1 + ))2 (2 + 22) =
yielding n < . If n , we obtain ; if n , then .
5 = 3 =
6
= 4 =
5
= 2(1 2)2 (1 + 2) = 2(1 4)(1 2) :
11-2. Let D0 and B0 0be the0 multipliers of the0 rst and the se ond fra tion, respe tively. Sin e x; y > ,4we have2 jj 6 that implies 6 2 6 and 6 4 6 .
Hen e 6 .
0 1 0 1 1 0 1 1
Then E AD B C and F BD DB0 , with B0 and D0 oprime sin e F
2(1 )(1 ) 2
is the least ommon multiple of the denominators. If, for a prime p , p divides d
= + = =
with k > , then p divides both E and F . Suppose p does not divide B . From
k
11-5. Answer: a) ; b) , , , , , , , , ja .
7 1 2 3 4 6 8 10 16 28 32
k
F BD0 we obtain that p divides D0 , hen e p also divides B 0 C E AD0 .
0
k
Consider a trapezoid of height h and the length of its shorter base a (see Fig. 9).
Therefore, p divides either B0 or C , and as B0 and D0 are oprime, p divides
= =
The longer base of the trapezoid is of length a h and thus there is a total of +
C . From F DB 0 we get that D is divisible by p , hen e p is a ommon fa tor
=
k
N (a; h) = (a +1) + (a +2) + : : : + (a + h +1) =
(2 + a h +2)(h +1)
of C and D , ontradi ting the irredu ibility of DC . We on lude that p divides k 2
B , and similarly also D . Sin e this is true for any prime divisor p of d , then B points with integral oordinates inside and on the border of this trapezoid.
and D are both divisible by d .
11-3. Draw a line parallel to BC through A and denote its interse tion points with 6
rays BE and CF by L and M , respe tively (see Fig. 8). From similar triangles q q q
a q q q q q q
jAE j jALj jAF j jAM j
q q q q q q q q q q
AEL and CEB we have
jEC j jBC j . Analogously jF B j jBC j . Moreover,
q q q q q q q q q q
= = q
h q q q q q q q q
from similar triangles AOL and DOB we get jjOD AOj jALj
q q q q q q q q q q
j jBDj , and analogously -
q q q q q q q q q q
= q q q q q q q q q q
jAOj jAM j Figure 9
jODj jDC j . Hen e
=
jAOj jALj jAM j jALj jAM j jALj jAM j jAE j jAF j a) We have to nd the number of distin t pairs a; h for whi h N a; h .
Taking into a ount that , we onsider two ases:
( ) ( ) = 2001
jBC j jBC j jEC j jF B j :
+ +
jODj jBDj jDC j jBC j
2001 = 3 23 29
1) If h k is even, then N a; h a k k where k > and
= = = + = +
+
= 2 ( ) = ( + + 1) (2 + 1) 2 +1 3
C
a k
+ >k
+ 1 >
k
+ 1. The fa tor k an be , or , yielding the
2 + 1
2 + 1 3 23 29
pairs ; , ; and ; .
2
(665 2) (75 22) (54 28)
D 2) If h k is odd, then N a; h a k k , where h N a; h
k > and a k > k . The fa tor k an be , ,
= 2 1 ( ) = (2 +2 + 1) ( )
L a
or , yielding the pairs ; , ; , ; and ; .
1 2 +2 +1 2 +3 1 3 23 1 2 + 3
O 29 (999 1)
a (330 5) (20 45) (5 57)
E 2 3 + 6
F b) For h ; ; ; : : :; we express the size of a trapezoid in a
terms of a (see the table); if h > , then N a; h > for any
= 1 2 3 7 3 4 + 10
a
a > . It is easy to he k that numbers , , , , , , ,
A B 7 ( ) 50 4 5 + 15
a
, and are the only ones that annot be expressed by
1 1 2 3 4 6 8 10
5 6 + 21
M
any of the formulae in the table. a
16 28 32
6 7 + 28
Figure 8 12-1. Answer: The only solution is x y . a
= = 0
7 8 + 36
17 18
Clearly x y is a solution. We know that j xj 6 jxj , where equality holds 12-4. If a , then p gives the desired result:
i x (this an be easily proved using derivatives). Now
= = 0 sin = 2 = 11
: : : + 210 = 211
= 0
1+ 2+4+ 1 = 2047 = 23 :
89
jxj > j xj = jyj > j sin yj = jxj ;
sin
If a > , then a > and there exists a prime2 p that divides a . Hen e a
and at least one of the inequalities is stri t if x 6 or y 6 . is ongruent to modulo p and M a a : : : a 1 is divisible by p .
2 1 1 1
p
We also have M > a > p , implying that M is omposite.
= 0 = 0 1 p = 1+ + + +
12-2. Answer: n.
p 1+ p
Let the hosen integers be a1; : : : ; a and, for ea h i ; : : :; k , let n be the 12-5. a) Let r1 , r2 , r3 be the sums of numbers in the rst, se ond and third row, and
exponent of in the prime fa torization of a , i.e. a b with b odd. Sin e 1 , 2 , 3 be the sums of numbers in the rst, se ond and third olumn. Denote by
k = 1 i
ni
6 b 6 n , there are n possibilities for the numbers b . If k > n , then a the element in the i -th row and j -th olumn, and noti e that all the elements
2 i i = 2 i i
there exist indi es i and j su h that b b b and n > n . Then a of the table are non-negative.
1 i 2 1 i + 1 ij
b ni
is divisible by a b. Sin e r1 r2 r3 1 2 3 , we have
i = j = i j i = 2
nj
j = 2
If k 6 n , then hoose any k numbers in the set f n ; : : :; n g . None of them
+ + = + +
is divisible by another sin e n < n . jr1 1j j r2 r3 2 3 j j r2 2 r3 3)j =
+ 1 2
2 2 ( + 1)
a11 = = ( + ) ( + ) = ( ) +(
jr2 2j jr3 3 j a22 a33 :
12-3. Let 6 IAE 6 BAI and 6 DBI 6 IBA , then 6 EIA 6 BID = =
(see Fig. 10). Applying the sine rule for triangle AEI and the equality As all the elements are non-negative, a22 and a33 annot both have minus sign
= = = = = = +
r jAI j , we obtain here and, onsequently, a11 is equal to the sum or dieren e of two numbers in
the table. The proof for all other elements of the table is similar.
= sin
jAE j jAI j r
sin( + )
=
sin 6 AEI =
sin sin 6 AEI
: b) The tables below satisfy the required ondition for any real x > : 0
B x x x x
;
0 0
x 0 x x x x :
0 x 0 x 2x 2x
2
D I IMO Team Sele tion Test: April 2001
r First Day

C E A 1. Consider on the oordinate plane all re tangles whose
Figure 10 (i) verti es have integer oordinates;
(ii) edges are parallel to oordinate axes;
From triangle BDI , we similarly get (iii) area is , where k ; ; : : :.
2
k
= 0 1 2
Is it possible to olor all points with integer oordinates in two olors so that no
jBDj = jBI j r
: su h re tangle has all its verti es of the same olor?
sin( + ) sin 6 IDB sin sin 6 IDB
=
2. Point X is taken inside a regular n -gon of side length a . Let h1; h2; : : : ; h be
Sin e 6 AEI = os
sin and sin 6 IDB = os , we have the distan es from X to the lines dened by the sides of the n -gon. Prove that
n
sin os sin os sin( + ) 2
: :::+ > :
1 1 1 1 1 1
jAE j jBDj + = +
r sin( + ) r sin( + )
=
r sin( + )
=
r h1
+
h2
+
hn a
19 20
3. Let k be a xed real number. Find all fun tions f : R ! R su h that 3. If k then f x x or f x ; if k 6 then f x k or
.
Answer: = 1 ( ) = ( ) = 0 = 1 ( ) = 1
fx
f (x) + f (y) 2 = kf (x + y2 )
( ) = 0
Substituting y in the original equation we get
= 0
for all real numbers x and y . k( f x f 2:
1) ( ) = (0) (1)
If k 6 , then substituting x in (1) we get f or f k .
So the solutions in the ase k 6 are tthe onstant fun tions f x and
= 1 = 0 (0) = 0 (0) = 1
Se ond Day
fx k .
= 1 ( ) = 0
4. Consider all produ ts by 2, 4, 6, ..., 2000 of the elements of the set
( ) = 1
If k , then from (1) we get f
= 1 . Substituting x in the original
(0) = 0 = 0
A
n 1 1 1
; ; ; : : :; ;
o
. Find the sum of all these produ ts.
1 1
equation we get f y 2 f y2 , and furhter substituting y we nd that
or f .
= ( ) = ( ) = 1
2 3 4 2000 2001
f(1) = 1 (1) = 0
5. Find the exponent of in the representation of the number : :{z:: : : } as For any non-negative real number z there is a real number y su h that y2 z , =
therefore from f y 2 f y2 we get that f z > for any z > . Also,
37 111 11
|
3372000 digits
produ t of prime powers.
( ) = ( ) ( ) 0 0
substituting x y2 in the original equation we get f y2 f y 2 , so
= ( ) = ( )
f z 6 for any z 6 . Sin e f y 2 f y2 f y 2 f y 2 , then we

6. Let C1 and C2 be the in ir le and the ir um ir le of the triangle ABC , respe - must have f y f y , i.e. f is an odd fun tion.
( ) 0 0 ( ) = ( ) = (( ) ) = ( )
tively. Prove that, for any point A0 on C2 , there exist points B0 and C 0 su h that
Now let x be any real number and z > , then denoting pz y we get
( ) = ( )
C1 and C2 are the in ir le and the ir um ir le of triangle A0 B 0 C 0 , respe tively. 0 =
f x z f x y2 f x f y 2 f x f y2 f x f z :
( + ) = ( +

) = ( )+ ( ) = (2) ( )+ ( ) = ( )+ ( )
Solutions of Sele tion Test
Hen e if a 6 b , then f b f a b a f a f b a > f a , i.e. f is
non-de reasing.
( ) = ( + ( )) = ( ) + ( ) ( )
1. Answer: Yes. Sin e f is an odd fun tion, (2) holds also when x and z are both negative. Now
Color the points with integer oordinates in three olors so that on ea h diagonal we show, using indu tion on n , that f nx nf x for any real x and integer n .
y x k all points are of the same olor and the olors hange in a y li manner
Indeed, this holds for n and if f nx nf x then
( ) = ( )
when k in reases. Sin e for even m and for odd
= +
m m = 0 ( ) = ( )
m , it is easy to understand that all three olors are present in verti es of ea h
2 1 (mod 3) 2 2 (mod 3)
f ((n +1)x) = f (nx + x) = f (nx)+ f (x) = nf (x)+ f (x) = (n +1)f (x) :
re tangle under onsideration. Now re oloring the verti es of some olor with one
of the remaining two olors, we obtain a oloring with the required properties. Sin e f is odd, we also have f ( nx) = f (nx) = nf (x) , i.e. f (nx) = nf (x)
holds for all integers n .
2. Let S be the area of the n -gon and r its inradius, then S n ar . On the other =
Earlier we proved that f (1) = 1 or f (1) = 0 . If f (1) = 0 then substituting
x = 1 in f (nx) = nf (x) we get that f (n) = 0 for all integers n , and sin e f is
2
hand, S a h1 h2 : : : h . Using the AM-HM inequality, we get
=
1
( + + + n) non-de reasing, we have f (x) = 0 for all real x . We show now that if f (1) = 1
2
then f (x) = x for all x . For integers we get it from f (nx) = nf (x) , substituting
x = 1 . For a rational number we have
n a
6 h1 h2 n : : : h
+ + + n
=
2S
na
= r: b
:::+
1 1 1
h1 h2 hn
+ +
a
a
a = f (a) = f b = b f ;
Comparing the lengths of ir umferen es of the n -gon and its in ir le, we get b b
na > r . Hen e
2
so f ab = ab . Assume that for some real number x we have f (x) 6= x , then
1
h1
+
1
h2
+ :::+
1
hn
> nr > a :2
f (x) = x + " where " 6= 0 . If " > 0 , then let r be a rational number su h
21 22
that x < r < x " , and if " < , then let r be a rational number su h that and note that
x > r > x " . In the rst ase we get r < x " f x 6 f r r , in the se ond
+ 0
37k+1 37k 37
ase r > x " f x > f r r , a ontradi tion.
+ + = ( ) ( ) =
1000 1 = 1000 1 =
+ = ( ) ( ) =

37k 37k + 37k 2 + : : : + 37k 36 :
4. Answer: 499
1001
. = 1000 1 1 + 1000 1000 1000
The value of The exponent of in the representation of number 37 is k by the
2001
k
indu tion hypothesis. Hen e it su es to show that the exponent of in the
37 1000 1 + 1

representation of
37
1+
1
2
1+
1
3
::: 1+
2001
1
1
37k + 37k 2 + : : : + 37k 36
is equal to the sum of all produ ts of the elements of set A by 1, 2, 3, ..., 2000, 1 + 1000 1000 1000
and the value of is . Sin e then 37k 1 (mod 37) . Let 37k = 37q + 1 ,
then
1 1000 1 (mod 37) 1000 1000

1
1
1
1
::: 1
1
1
2 3 2001
37 + 100037 2 + : : : + 100037 36=
k k k
is equal to a similar sum where the produ ts by 2, 4, 6, ..., 2000 are taken with
1 + 1000
2 36
a plus sign but the produ ts by 1, 3, 5, ..., 1999 are taken with a minus sign. = 1 + (37q + 1) + (37q + 1) : : : + (37q + 1)
Denote the required sum by S , then q + 1) + (2 37q + 1) + : : : + (36 37q + 1) =
1 + (37

=
37 36
q
37 + 37 = 37
2 18 q + 37 37 (mod 372) :
2 S = 1+
1
2
1+
1
3
::: 1+
2001
1
+
2
1

1

:::
1

So 1 + 1000
37 k
+
37 2 : : :
1000
37 36 is divisible by
k
+ + 1000 but not by 2 ,
k
37 37
+ 1
2
1
3
1
2001
2 =
and the exponent of in the representation of 37 +1 is k .
37 1000
k
1 +2
=
3
:::
4 2002
+
1
:::
2 2000
2 = Hen e the exponent of in the representation of 372000 is .
37 1000 1 2001
6. Let I and O be the in enter and the ir um enter of the triangle ABC , respe -
2 3 2001 2 3 2001
; tively. We know by Euler's formula that jOI j2 R2 Rr , where r and R are
2002 1 1
= + 2 = 999
the radii of the in ir le and the ir um ir le, respe tively.
2 2001 2001 = 2
and S . 1001
Assume now that there exists a point A0 on the ir le C2 su h that it is impossible
= 499
2001 to onstru t the points B0 and C 0 as required in the problem. Let the tangents
5. . drawn from A0 to the ir le C1 tou h C1 in B0 and C 0 , hen e B0C 0 is not tan-
Answer: 2001
As and are relatively prime it is su ient to nd the exponent of in the gent to the ir le C1 . Suppose the line B0C 0 and the ir le C1 have no points in
representation of the number
37 9 37
ommon0 0(the ase of two ommon points is similar). Let the distan e between the
line B C and the ir le C1 be Æ > . 0
3372000 372000 Now start moving the points B0 and C 0 along the ir le C2 towards A0 in su h
|999 : :{z
:: : : 99} = 10 1 = 1000 1 :
a way that the distan es from the ir le C1 to the straight lines A0B0 and A0C 0
3372000 numbers remain equal (note that they are both equal to at the beginning) denote
this distan e by . The distan e Æ obviously de reases, whereas the distan e
0
We show by indu tion on k that the exponent of 37 in the representation of in reases, hen e at some moment they must be ome equal. Now we an in rease
37
1000 is k . In the ase k we have
k
1 +1 = 0
the radius r by Æ > to make it the in ir le of the triangle A0B0C 0 . Hen e
the triangle A0B0C 0 has ir umradius R and inradius r Æ , but the distan e jOI j
= 0
370 3 37 ;
is the same as for the triangle ABC , hen e Euler's formula for triangle A0B0 C 0
1000 1 = 999 = 3 +
i.e. the exponent of is . Suppose now that for some k our assertion holds,
37 1 be omes violated.
23 24
Estonian math ompetitions 2001/2002 3. For any positive integer n , denote by S (n) the sum of its positive divisors (in-
luding 1 and n ).
a) Prove that S (6n) 6 12S (n) for any n .
n does the equality S (6n) = 12S (n) hold?
b) For whi h
from various materials distributed at the re ent IMO-s.
4. In a triangle ABC we have 6 B = 2 6 C and the angle bise tor drawn from A
interse ts BC in a point D su h that jAB j = jCDj . Find 6 A .
Autumn Open Contest: O tober 2001
5. Let b1 ; b2 ; : : :; b n be a rearrangement of positive real numbers a1 ; a2 ; : : :; a n.
Juniors (up to 10th grade) Prove that
a)

a1 +
1 a + 1 : : : a + 1 > 2 n
;
b1 2
b2 n
b
1
n
1. A gure onsisting of ve equal-sized squares is pla ed as 7
shown in a re tangle of size 78 units. Find the side length
b) if equality holds here for an odd n then at least one of the numbers a i is .
of the squares.
2. Find the remainder modulo 13 of the sum
8 Solutions of Autumn Open Contest
1 + 2 + 3 + : : : + 2000
2001 2001 2001 2001
+ 2001 2001
: p
J1. Answer: 5 .
3. Find all triples (x; y; z) of real numbers satisfying the system of equations (where Let a be the required side length, then the proje tions of ea h side of any square
[r℄ frg
and denote the integer and fra tional part of r , respe tively): to the sides of the re tangle are x and y where x2 + y2 = a2 . We have
8
< x + [y℄ + fz g = 200;2
> 8 = 2x + y + x + y = 3x + 2y
>
fxg + y + [z ℄ = 200;1 : and
:
[x℄ + fyg + z = 200;0
7 = 3x + y ;
4. Consider a point M inside triangle ABC su h that triangles ABM , BCM and p p
CAM have equal areas. Prove that M is the interse tion point of the medians of yielding y=1 x=2 a= x +y = 5
, and 2 2 .
triangle ABC .
J2. Answer: 0 .
5. For any integer n > 1 onsider all squares with verti es in points having non- 1001
Arrange all terms of the sum ex ept 13 2001
(whi h is divisible by ) in pairs
negative integer oordinates not greater than n .
(k ; (2002 k) )
2001
2002 2001
13 k + (2002 k)
. Sin e is divisible by ,
2001 2001
is
a) How many su h squares are there for n = 4? ongruent to k +( k) = 0 2001
13 2001
modulo , i.e. the sum of ea h pair is divisible
b) Find a general formula for the number R n of su h squares for any n. by 13 0
and hen e the required remainder is .
J3. Answer: x = 100;15 y = 100;95 z = 99;05
the only su h triple is ; ; .
Seniors (grades 11 and 12) [r℄+frg = r
Adding all three equations and using x + y + z = 300;15 we have . Sub-
(y [y℄)+(z fzg) = 99;95
tra ting from here the rst given equation, we obtain
1. The sum of two distin t positive integers, obtainable from ea h other by rear- or fyg + [z ℄ = 99;95 fyg = 0;95 [z℄ = 99
, yielding and . Similarly we get
rangement of digits, onsists of 2001 equal digits. Find all possible values of the [x℄+ fzg = 100;05 fxg +[y℄ = 100;15 [x℄ = 100 fzg = 0;05 fxg = 0;15
and , i.e. , ,
digits of the sum. and [y℄ = 100 .
2. The side lengths of a triangle and the diameter of its in ir le, taken in some order, J4. It su es to prove that if triangles ABM and BCM have equal areas then M
form an arithmeti progression. Prove that the triangle is right-angled. lies on the median drawn from B. Let K and H be the perpendi ular proje tions
1 2
of A and C to BM , and let BM interse t AC in a point P (see Figure 1). Then Note. R R = n + (n 1) + : : : + 1
Using the identities n n 1
2 2 2
and
jBM j jAK j = S =S = jBM j2 jCH j 1 + 2 + : : : + n = n(n + 1)(2
2 2
6
n + 1)2
2 ABM BC M
and hen e jAK j = jCH j . If AC is perpendi ular to BM , then K = H = P ,
we an show by indu tion that
jAP j = jP C j and BP is a median. If AC is not perpendi ular to BM , then (n + 1) (n + 1) 1 : 2 2
R =
learly one of K and H lies inside triangle ABC and the other one outside of it. n
12
Hen e 6 AKP = 90Æ = 6 6 AP K = 6 CP H , i.e. triangles AKP and
CHP and
CHP are ongruent, whi h again yields jAP j = jP C j and BP being a median. S1. Answer: 1 2 3 4 5 6 7 8
, , , , , , , .
Let A 0 1 2 3
be any of the digits , , , . Taking
B n = A (A+5) : : : (A+5)(A+5)A(A+6); m = A (A+5) : : : (A+5)(A+6)A(A+5);
| {z } | {z }
1997 digits 1997 digits
H r r r r
M r r r r r r r or
r r r r r r r r r
n =(A+1) : : : (A+1) A(A+2); m =(A+1) : : : (A+1)(A+2)A;
A P C r r r r r r r r r | {z } | {z }
K n=1 n=2 n=3
1999 digits 1999 digits
we have n and m obtainable from ea h other by rearrangement of digits and
Figure 1 Figure 2
n + m =(2A+1) : : : (2A+1); n + m =(2A+2) : : : (2A+2) ;
| {z } | {z }
Answer: a) 50 ; b) 1 n + 2 (n 1)2 + : : : + (n 1) 22 + n 12 .
2
J5. 2001 digits 2001 digits
We rst study the possible squares for n = 1; 2; 3 .
1 8
For n = 1 we have a single square of side length 1 .
respe tively. Hen e to are possible digits.
m+n = 99 9
:::}
n = 2 we have 2 2 possible lo ations
p for the square of side length 1
Suppose now that . Moving from right to left it is easy to see
For and new
| {z
2001 digits
possible squares of side lengths 2 and 2 , one of ea h. that there an be no arries during the addition. Hen e any digit A o urs in
For n = 3 , we have 3 3 possible lo ations for the square of side length
p 1 22
, n in these and only these positions where the digit 9 A o urs in m . Sin e n
possible lo ations for ea h of the squares of side lengths 2 and 2 and three new and m are obtainable from ea h other by rearrangement of digits then any digit
types of squares, one of ea h (see Figure 2). A o urs in n the same number of times as 9 A 6= A , and the number of digits
in n, m and also n+m must be even a ontradi tion.
We see that for ea h n we have 1=1 2
possible lo ation for ea h of the new
S2. Drawing a line through the in enter of a triangle parallel to ea h of its sides it is
squares (i.e. squares having all their verti es at the edges of the grid) and for the
next values of n we have 2 ; 3 ; 4 ; :::
2 2 2
possible lo ations for these squares. It
easy to see that the diameter of the in ir le is shorter than any of the sides. Let
x d>0
n and be the diameter of the in ir le and the dieren e of the arithmeti
x + d , x + 2d and x +3d .
remains to noti e that the number of the new squares is sin e we an pla e
one of its verti es either in a orner of the grid or in one of the n 1 points on
progression, then the side lengths are
S of the triangle in two ways we get
Finding the area
the side of the grid, thereby determining the lo ations of the other three verti es.
q
n x
p = S = p p (x + d) p (x + 2d) p (x + 3d) :
Hen e for any we have
2
R n = 1 n + 2 (n 1) + 3 (n 2) + : : : + (n 1) 2 + n 1
2 2 2 2 2
; s
R4 = 50 . p=
3(x + 2d) 3(x + 2d)x = 3(x + 2d)(x + 4d)(x + 2d)x
yielding Sin e
2 4 we have
16 , yield-
ing 3x = x + 4d x = 2d and x + d = 3d x + 2d = 4d
. Hen e the side lengths are ,
3 4
and x + 3d = 5d , i.e. the triangle is right-angled. Sin e i1 b =a i2 and b2
i =a i1 , we an omit a1
i and a2
i and use the indu tion
S3. Answer: b) for n not divisible by 2 or 3 .
hypothesis.
1 1 1
n 6n If i3 6= i1 , we nd a4 =b3 = = , a = b 4 = a = a 1 et . Sooner or
Considering the representations of and as produ ts of primes we see that i i
a3 a1 5 i i i
any positive divisor of 6n is the produ t of a positive divisor of 6 and a positive i +1 = i1 for some even k (sin e a 1 6= 1 ). Similarly to the
i i 4 i
6n are numbers of the form d , 2d , 3d
later we must have
n.
k i
divisor of Hen e the positive divisors of a 1 ; a 2 ; : : :; a k and use the indu tion hypothesis.
6d
previous ase we an now omit
d n , and
i i i
and where is a positive divisor of
S (6n) 6 S (n) + 2S (n) + 3S (n) + 6S (n) = 12S (n) :
Spring Open Contest: February 2002
Here equality holds if and only if the abovementioned four series of divisors do
not interse t, i.e. no divisor d of n an be represented as 2d or 3d where d is
0 0 0
another divisor of n , or equivalently n is not divisible by 2 or 3 .
Juniors (up to 10th grade)
Answer: 72 .
S4.
Æ 1. Is it possible to arrange the integers 1 to 16
Let jAB j = jCD j = a , 6 C = and 6 A = 2 , then 6 CAD = 6 BAD = , a) on a straight line;
6 B = 2 and 6 BDA = + . Applying the sine rule in triangles ACD and ABD b) on a ir le
we have
so that the sum of any two adja ent numbers is the square of an integer?
sin jADj sin 2
sin = a = sin( + ) ; 2. Does there exist a re tangle with integer side lengths with the square of its diagonal
equal to 2002 ?
2 sin os =Æ sin( + ) tan = tan 0 < ; < 90Æ
In a triangle ABC we have jAB j = jAC j and 6 BAC = . Let P 6= B be a point
yielding , or . Sin e we have
= . 180 = 2 +2 + = 5
Now from = 36Æ 6 A = 2 = 72Æ we have and . 3.
on AB and Q a point on the altitude drawn from A su h that jP Qj = jQC j .
S5. a) Using the AM-GM inequality we have: Find 6 QP C .
1
a1 + : : : a +
1 > 2
r
a1
:::2
r
a
n
=2 n
r
a1 : : : a n
=2 n
: 4. Dene a1 ; a2 ; : : : ; a ; : : : as follows:
b1 : : : b
n
b1 b1 n
b b
a1 = 0; a2 = 1; a = 5a n > 2.
n n n
a ;
b) If n = 1 then a1 +
1
= 2 and hen e a1 = 1 . Suppose now that the laim is true
n n 1 n 2 for
a1 For whi h n is a divisible by: a) 5 ; b) 15 ?
1 = 2r a
n
for any odd integers less than n . The equality holds if and only if a +
i
b b i 5. For whi h positive integers n is it possible to write n real numbers, not all equal
1 . If a = b for some i then a = b = 1 and we are done.
i i
0
for ea h i , i.e. a =
to , on a ir le so that ea h of these numbers is equal to the absolute value of
b i i i i i the dieren e of its two neighbouring numbers?
a 6= b for all i then onsider some i1 su h that a 1 6= 1 . Then b 1 equals to
i
If
1
i i i i
some a2 i2 6= i1 , and 2
where a =b =
. Also, b 2 equals to some a 3 where Seniors (11th and 12th grade)
i 1 i
a i i i
1 =a
i1
i3 6= i2 . a3 =b2=
a , b and of a right triangle form a geometri progression, and ab = 1 .
Hen e .
i i
a2 i1
1. The sides
=i
i
If i3 1 then Find a, b, .
+1 a + 1 = a + 1 a + 1 = a + a1 a + a1 =

a 2. Let a; b be any real numbers su h that jaj 6= jbj . Prove that
i1
b1 i
i2
b i2 a a
i1
i2
i2
i3
i1
i2
i2
i1
ab
= (a + a ) a1 + a1 = 2a a2 = 2 : + b > 1 :

a

2
i1 i1 i1 a b
i1 i1 i1
5 6
Let ABCD be a rhombus with 6 DAB = 60 . Let K , L be points on its sides AD Sin e jAB j = jAC j then the altitude drawn from A is also an angle bise tor.
3.
Æ
and DC and M a point on the diagonal AC su h that KDLM is a parallelogram. Note that jQB j = jQC j = jP Qj (see Figure 3), i.e. the triangles BQC , BQP
BKL is equilateral.
Prove that triangle and P QC are isos eles. Denote 6 QBC = 6 QCB = , 6 QBP = 6 QP B =
and 6 QP C = 6 QCP = Æ , then 6 QCA = . From triangle ABC we now have
4. Call a 10 -digit natural number magi if it onsists of 10 distin t digits and is + 2 + 2 = 180Æ and from triangle P BC we have 2Æ +2 +2 = 180Æ . Hen e
divisible by 99999 . How many su h magi numbers are there (not starting with
6 QP C = Æ = .
digit 0 )? 2
5. Find the maximum number of distin t four-digit positive integers onsisting only J4. Answer: a) for any odd n ; b) for n = 6k + 1 .
of digits 1 2
, and 3 su h that any two of these numbers have equal digits in at a) From the equality a = 5a 1 a 2 we see that a is divisible by 5 if and
5 a1 = 0 5 a2 = 1
n n n n
most one position? only if a 2 is divisible by . Sin e is divisible by but is not
5 5
n
divisible by , then a n is divisible by if and only if n is odd.
b) Taking n+1 instead of n in the given equality we have
Solutions of Spring Open Contest
an+1 = 5a n a n 1 = 5 (5a n 1 a n 2 ) an 1 = 24a n 1 5an 2 :
J1. Answer: a) yes; b) no.
From here we see that a is divisible by 3 if and only if a 2 is divisible by 3 .
4 = 16 < 16 + 1 6 = 36 > 16 + 15 9 16
n+1
a1 = 0 3 a2 = 1 and a3 = 5 1 0=5
n
2 2
Sin e and then only an be adja ent to Sin e is divisible by but are not divisible
(giving 16 + 9 = 25 = 5 2
). Hen e it is impossible to arrange the numbers 1 to 16 by 3 , then a is divisible by 3 if and only if n = 3m + 1 for some m . Hen e a
15 n = 3m + 1 , i.e.
n n
on a ir le in the required manner. A suitable arrangement on a straight line is: is divisible by if and only if n is both odd and of the form
n = 6k + 1 .
16; 9; 7; 2; 14; 11; 5; 4; 12; 13; 3; 6; 10; 15; 1; 8 :
J5. Answer: if and only if n is divisible by 3 .
J2. Answer: no. Sin e ea h number written on the ir le is equal to the absolute value of the dif-
We need to determine whether there exist positive integers a and b su h that feren e of two others, then all these numbers are non-negative. Let a be maximal
a2 + b = 2002
2
2002 7 among these numbers (then a > 0 ) and let b and be the numbers adja ent to
b > > 0. b 6 a then also b 6 a , and the equality a = b
. Note that is divisible by and the square of any integer
0 1 2 4 7 a b it, with Sin e
b=a = 0 . Hen e any number a on the ir le must have a
is ongruent to , , or modulo . Hen e and must both be divisible
7 a +b 2 2
49 2002 49 holds only if and
0 0
by , but then is divisible by , and is not divisible by .
and as its neighbours, and any number must have its both neighbours equal.
We see that the numbers on the ir le must be a , a , 0 , a , a , 0 , . . . (see Figure 4)
A and hen e n must be a multiple of 3 .
P
22 ar
r
Æ
ar
Q 0r r
Æ r r
a a
r
0
B C Figure 4
On the other hand it is easy to he k that for any n = 3k and a > 0 the numbers
Figure 3
a; a; 0; a; a; 0; : : :; a; a; 0} satisfy the required onditions.
6 QP C = . | {z
J3. Answer:
2 k triples a; a; 0
7 8
s p s p 3456
5 1 5+1 S4. Answer: .
S1. Answer:
2 1 , and
2 .
We an write any ten-digit number ab defghij as
b ab defghij = 100000 ab de + fghij =
Assume w.l.o.g. that a 1. Hen e
q
from ab = 1 b =1
we have
3
and b=1 . From the Pythagorean Theorem we
= 99999 ab de + ab de + fghij :
1
+1=q q 1=0 x x 1
2
now have
q
2
, or
4
q2 . Sin e the equation
2
has Hen e ab defghij 99999
is divisible by if and only if the sum ab de + fghij is
p s p sp divisible by 99999 . Sin e ea h summand here is positive and less than 99999 , we
5+1 q=
5+1 1= 5 1 must have ab de + fghij = 99999 , or equivalently
2 as its only positive solution, then
2 and
q 2 .
a+f = b+g = +h= d+i = e+j = 9
S2. For any real y and x > 0 we have:
a) x > 1 , if x > 1 and y > 0 or x < 1 and y < 0 ;
y (sin e the sum ontains only digits 9 , no arries an o ur on addition). We see
b) x < 1 , if x > 1 and y < 0 or x < 1 and y > 0 ;
y that magi numbers are in one-to-one orresponden e with numbers of the form
) x = 1 , if x = 1 or y = 0 .
ab de where a; b; ; d; e are ve distin t digits su h that a 6= 0 and the sum of
9 9 8 6 4 2 = 3456
y
no two of them is . There are su h numbers ab de .
We shall now onsider the ases where ab is positive, negative or equal to zero.
If ab > 0 , then a and b have the same sign and ja + bj > ja bj > 0 , hen e S5. Answer: 9 .

a + b > 1 and a + b > 1 .

ab
Note that we annot have more than 33 = 9 integers with the required property
a b a b sin e the pairs of rst two digits of any two of them must be distin t. A suitable
If ab < 0 , then a and b have opposite signs and ja bj > ja + bj > 0 , hen e set of 9 integers is 1111 1222 1333 2123 2231 2312 3132 3213 3321
, , , , , , , , .

0 < aa + bb < 1 and aa + bb > 1 .
ab
Final Round of National Olympiad: Mar h 2002
If ab = 0 , then a
+ b = 1 sin e a + b 6= 0 .
ab
a b a b
9th grade
S3. The rhombus ABCD onsists of two equilateral triangles ABD and BCD . We
shall prove that jKD j = jLC j (see Figure 5) then triangles KBD and LBC are
K and L are taken on the sides BC CD ABCD
jKB j = jLB j , 6 KBD = 6 LBC . Hen e 6 KBL = 6 DBC = 60Æ ,
1. Points and of a square so that
ongruent and 6 AKB = 6 AKL . Find 6 KAL .
i.e. the triangle BKL is equilateral.
2. Do there exist distin t non-zero digits a, b and su h that the two-digit number
D L C ab is divisible by , the number b is divisible by a and a is divisible by b?
K 3. Let a1 ; a2 ; : : : ; a be pairwise distin t real numbers and m be the number of
a + a (where i 6= j ). Find the least possible value of m .
n
M distin t sums i j
4. Mary writes 5 numbers on the bla kboard. On ea h step John repla es one of the
A B numbers on the bla kboard by the number x + y z , where x , y and z are three
Figure 5 of the four other numbers on the bla kboard. Can John make all ve numbers on
the bla kboard equal, regardless of the numbers initially written by Mary?
To prove the equality jKD j = jLC j note that LM is parallel to AD and
6 LMC = 6 DAC = 6 DCA = 6 LCM . Hen e the triangle MLC is isos eles, 5. There were n> 1 aborigines living on an island, ea h of them telling only the
jLC j = jLM j = jKDj .
i.e. truth or only lying, and ea h having at least one friend among the others. The
9 10
new governor asked ea h aborigine whether there are more truthful aborigines or 3. The tea her writes a 2002 -digit number onsisting only of digits 9 on the bla k-
liars among his friends, or an equal number of both. Ea h aborigine answered that board. The rst student fa tors this number as ab with a>1 and b>1 and
there are more liars than truthful aborigines among his friends. The governor then
0
repla es it on the bla kboard by two numbers a and b su h that ja a j = 2 and
0 0
ordered one of the aborigines to be exe uted for being a liar and asked ea h of jb b0 j = 2 . The se ond student hooses one of the numbers on the bla kboard,
the remaining n 1 aborigines the same question again. This time ea h aborigine fa tors it as d with >1 and d>1 and repla es the hosen number by two
answered that there are more truthful aborigines than liars among his friends. numbers 0 and d0 su h that j 0j = 2 and jd d0j = 2 . The third student again
Determine whether the exe uted aborigine was truthful or a liar, and whether hooses one of the numbers on the bla kboard and repla es it by two numbers
there are more truthful aborigines or liars remaining on the island. following a similar pro edure, et . Is it possible that after a ertain number of
students have been to the bla kboard all numbers written there are equal to 9 ?
10th grade
4. Let a1; a2; a3; a4 ; a5 be real numbers su h that at least N of the sums a + a , i j
where i < j , are integers. Find the greatest value of N for whi h it is possible
that not all of the sums a + a are integers.
i j
1. The greatest ommon divisor d and the least ommon multiple v of positive
integers m and n satisfy the equality 3m + n = 3v + d . Prove that m is divisible 5. John built a robot that moves along the border of a regular o tagon, passing ea h
by n. side of the o tagon in exa tly 1 minute. The robot begins its movement in some
vertex A of the o tagon, and further on rea hing ea h vertex an either ontinue
2. Let ABC be a non-right triangle with its altitudes interse ting in point H . Prove movement in the same dire tion, or turn around and ontinue in the opposite
that ABH is an a ute triangle if and only if 6 ACB is obtuse. dire tion. In how many dierent ways an the robot move so that after n minutes
it will be in the vertex B opposite to A?
3. John takes seven positive integers a1 ; a2 ; : : : ; a7 and writes the numbers aa ,
+a i 6= j
i j
a i j and ja i aj
j for all on the bla kboard. Find the greatest possible
number of distin t odd integers on the bla kboard.
12th grade
4. Find the maximum length of a broken line on the surfa e of a unit ube, su h that
its links are the ube's edges and diagonals of fa es, the line does not interse t itself 1. Peter, John, Kate and Mary are standing at the entran e of a dark tunnel. They
and passes no more than on e through any vertex of the ube, and its endpoints have one tor h and none of them dares to be in the tunnel without it, but the
are in two opposite verti es of the ube. tunnel is so narrow that at most two people an move together. It takes 1 minute
for Peter, 2 minutes for John, 5 minutes for Kate and 10 minutes for Mary to
5. The tea her writes numbers 1 at both ends of the bla kboard. The rst stu- pass the tunnel. Find the minimum time in whi h they an all get through the
dent adds a 2 in the middle between them; ea h next student adds the sum tunnel.
of ea h two adja ent numbers already on the bla kboard between them (hen e
there are numbers 1; 3; 2; 3; 1 on the bla kboard after the se ond student; 2. Does there exist an integer ontaining only digits 2 and 0 , whi h is a k -th power
1; 4; 3; 5; 2; 5; 3; 4; 1 after the third student et .) Find the sum of all numbers of a positive integer with k > 2?
on the bla kboard after the n -th student.
3. Prove that for positive real numbers a, b and the inequality
11th grade
2(a + b + ) < (a + b + )
4 4 4 2 2 2 2
holds if and only if there exists a triangle with side lengths a, b and .
1. Determine all real numbers a su h that the equation x8 + ax4 + 1 = 0 has four
real roots forming an arithmeti progression. 4. All verti es of a onvex quadrilateral ABCD lie on a ir le ! . The rays AD ,
BC interse t in point K and the rays AB , DC interse t in point L . Prove that
2. Inside an equilateral triangle there is a point su h that the distan es from it to the ir um ir le of triangle AKL ! if and only if the ir um ir le
is tangent to
the sides of the triangle are 3 4 , and 5 . Find the area of the triangle. of triangle CKL is tangent to !.
11 12
5. There is a lottery at John's birthday party with a ertain number of identi al i.e. there are at least 2n 3 distin t sums. Taking a =i we have 1+2 = 3
(n 1) + n = 2n 1
i
prizes, whereas ea h of the guests an win at most one prize. It is known that as the minimal sum and as the maximal sum, so there are
if there was one prize less than there a tually is, then the number of possible exa tly 2n 3 distin t sums in this ase.
distributions of the prizes among the guests would be 50% less than it a tually
9-4. Answer: yes.
is, while if there was one prize more than there a tually is, then the number of
50% Denote the numbers written by Mary by a , b , , d and e (not ne essarily distin t).
b x = +d e
possible distributions of the prizes would be more than it a tually is. Find
First John an repla e ea h of a and by . Then he an repla e ea h
d e+x x = e e + e e = e:
the number of possible distributions of the prizes.
of and by and nally repla e both numbers x by
Solutions of Final Round (a; b; ; d; e) ! (x; x; ; d; e) ! (x; x; e; e; e) ! (e; e; e; e; e) :
9-5. Answer: the exe uted aborigine was truthful and after the exe ution only liars
Answer: 45 .
9-1.
Æ remained on the island.
Let AM be the perpendi ular drawn from A to KL (see Figure 6). Sin e ABK First note that there was a truthful aborigine on the island before the exe ution,
and AMK are ongruent right triangles then jAM j = jAB j = jAD j , i.e. AML sin e otherwise all friends of ea h aborigine would have been liars, and hen e their
and ADL are also ongruent right triangles, and answers would have been true a ontradi tion.
6 KAL = 6 KAM + 6 LAM = 6 KAB + 6 LAD ; Suppose now there was a truthful aborigine on the island after the exe ution.
Then both his answers must have been true but this is impossible sin e the
when e exe ution of one aborigine ould not hange the dieren e of the numbers of liars
26 KAL = 6 KAM + 6 LAM + 6 KAB + 6 LAD = 90Æ
and truthful aborigines among his friends from positive to negative.
Let m = dm and n = dn where g d (m ; n ) = 1 . Then v = m n d and we have
0 0 0 0 0 0
6 KAL = 45Æ .
10-1.
and 3m0d + n0d0 = 3m0n0d + d , yielding 3m0 + n0 = 3m0n0 +1 and (3m0 1)(n0 1) = 0 .
Sin e 3m 1 6= 0 , then n0 1 = 0 and hen e n = d divides m .
D L C 10-2. If H is the ortho enter of triangle ABC , then C is the ortho enter of triangle
M ABH . We'll onsider three possible ases.
(1) If ABC is an a ute triangle (see Figure 7), then H lies inside triangle ABC
K
and C lies outside triangle ABH , when e ABH is an obtuse triangle.
(2) If ABC is an obtuse triangle with 6 ACB a ute, then assume w.l.o.g. that
A B 6 B is obtuse (see Figure 8). Then C and H lie on opposite sides of AB . Hen e
Figure 6 C is outside triangle ABH , and ABH is an obtuse triangle.
(3) If 6 ACBis obtuse (see Figure 9), then C lies inside triangle ABH , when e
9-2. Answer: no.
ABH is an a ute triangle.
Note that if a, b and satisfy the required onditions and one of them is even,
a b
then all three are even. Then
2 2
, and
2 also satisfy the required onditions.
C H H
Hen e we an assume w.l.o.g. that a, b and are all odd. Also note that none

of these numbers an be 5 , sin e then the other two should also be 5 . Hen e it
H B C
su es to onsider 1 , 3 , 7 and 9 and one of a , b and must be 3 or 9 let
this be a . Then b is divisible by 3 , whi h gives fb; g = f3; 9g , a ontradi tion.

9-3. Answer: 2n 3. A B A C A B
We an assume w.l.o.g. that a1 < a2 < : : : < a . Then n Figure 7 Figure 8 Figure 9
a1 + a2 < a1 + a3 < : : : < a1 + a < a2 + a < : : : < a 1 + a ;
n n n n 10-3. Answer: 30 .
13 14
If there are m odd integers among a i, then the maximum number of odd integers Answer: a =
82
written on the bla kboard is
11-1.
9 .
t=x 4
t x =t 4
m(m 1)
+ 2 m(7 m) = 1 (m m + 28m 4m ) =
Substituting we get a quadrati equation in , and the equation 0
f (m) = 2 2
2 2
has at most two roots whi h in this ase have equal absolute values. Hen e the
four roots of the equation x + ax + 1 = 0 8 4
x have to be of the form 0 and

= 32 m(9 m) = 32 814 29 m x
2
1 . Assuming w.l.o.g. that x >x 1 0 we see that these roots form an arithmeti
progression if and only if x = 3x t + at +1
0 . Sin e the roots of the equation
2
are
x 4
81x 4
81x = 1
1
x =
1 8
a = 82x =
82 4 4
and the maximum value of f (m) is f (4) = f (5) = 30 . 9 9
then 0 and 0 , we have 0 , yielding 0 and 0 .
a1 = 2 , a2 = 4 ,
It remains to he k that these numbers an all be distin t, e.g. for
a3 = 6 a4 = 25 = 52 , a5 = 125 = 53 , a6 = 15625 = 56 , a7 = 9765625 = 510 . 36 + 25p3
and 11-2. Answer:
4 .
Consider a triangle ABC P with a point jP Aj = 3 jP B j = 4 inside su h that ,
and jP C j = 5 60Æ
. Rotating the triangle by C A B B around , we map to and
s
to some B0 (see Figure 12). P P0 Then jP 0B j = jP Aj = 3
maps to , where ,
s
jP 0B 0 j = jP B j = 4 jP 0C j = jP C j = 5
and CP P 0 . Sin e is an equilateral
triangle then jP P 0j = 5 jP B j + jP 0B j = 4 + 3 = 5 = jP P 0j
. Hen e
2 2 2 2 2 2
,
s and 6 P BP 0 = 90Æ AP B
. From triangles BP 0 B 0 and we have
s 6 ABP + 6 BAP = 6 ABP + 6 B 0 BP 0 = 120Æ 90Æ = 30Æ
Figure 10 Figure 11 and 6 AP B = 180Æ 30Æ = 150Æ AP B . The osine law in now gives
p
p jAB j = jAP j + jBP j 2 jAP j jBP j os 6 AP B = 25 + 12 3
2 2 2
3+4 2 p
36 + 25p3
10-4. Answer: .
1 S=
3
p 4 jABj = 4
The links of su h a broken line are the edges of the ube (of length ) and the 2
2
and the area is .
diagonals of its fa es (of length ). Sin e the line passes ea h vertex at most
on e, it an have at most 7 links. Coloring the verti es as shown on Figure 10 we
see that opposite verti es are of dierent olour and the endpoints of the diagonal B0 C
of ea h fa e are of the same olour hen e an odd number of links have to be
4 5
P0
edges of the ube. Also, it is lear from this olouring that no more than three
onse utive links an be diagonals (sin e the ube has only 4 verti es of ea h
5
olour). It is now easy to he k that a broken line with
p 1 edge and 6 diagonals is
impossible, hen e its length annot ex eed 3+4 2 ; a suitable broken line of this 3 4 3
length is shown on Figure 11. P
10-5. Answer: 3 +1
n
.
B A
Let S be the sum of the numbers on the bla kboard after the n -th student. We Figure 12
= 3 +1
. Indeed, S0 = 2 = 3 + 1 and ea h
n
shall prove by indu tion that Sn
n0
k -th student is ounted in exa tly two 11-3. Answer: no.
2002 999 : : : 9 3 4 = ab
number that is on the bla kboard after the
(k+1) 1 The initial -digit number is ongruent to modulo . If N
3 4 3
of the sums written by the -th student, with the ex eption of the two -s
written by the tea her whi h are ounted only on e hen e and N is ongruent to modulo then one of a and b is ongruent to and
the other is ongruent to 1 modulo 4 and the same is true for a0 and b0 . Hen e
S = S + 2S 2 = 3(3 + 1) 2 = 3
k k +1
+1: at all times there is a number on the bla kboard whi h is ongruent to 3 modulo
4 9 1 4
k +1 k k
, while is ongruent to modulo .
15 16
11-4. Answer: 6 . Note that it is possible to get everyone through the tunnel in 17 minutes:
If there are four integers and one non-integer among a i then N =6 . To prove 1) Peter and John go through the tunnel (2 minutes);
the maximality we denote the fra tional part of x by fxg and note that: 2) Peter brings ba k the tor h (1 minute);
(a) iffag =6 fbg and is any real number then at most one of + a and + b 3) Kate and Mary go through the tunnel (10 minutes);
4) John brings ba k the tor h (2 minutes);
is an integer;
if a = b then a + b is an integer if and only if fag = 0 or fag = 0;5 ;
5) Peter and John go through the tunnel (2 minutes).
(b)
if fag 6= fbg and a + b is an integer then neither fag nor fbg is 0 or 0;5 .
It remains to show the minimality of this total. Clearly they have to go through
( )
the tunnel an odd number of times and bring ba k the tor h at least twi e, hen e
Considering now the possible partitions of the set fa1; a2 ; a3 ; a4 ; a5 g to subsets they have to go through the tunnel at least 3 times in one dire tion and 2 times
(of elements with equal fra tional parts) and keeping in mind the above remarks 7
6
in another dire tion. If they go through the tunnel or more times then the total
(a), (b) and ( ) we see that the only ase when there an be more than integer 10+2+5 1 = 17
+a fa1 g = fa2g = fa3g = fa4 g = fa5g , and in this ase all
time annot be less than minutes. If they go through the tunnel
sums a i j is when 5 times then ea h pass in the initial dire tion takes at least 2 minutes and one
these sums are integers. of these (with Mary) takes 10 minutes. If Peter brings ba k the tor h both times
11-5. Answer: 2 (2k 1 k 1
1) for n = 2k , and 0 for n = 2k + 1 .
then Kate and Mary annot go through the tunnel together and they need at least
10 + 5 + 2 + 2 1 = 17 minutes. If someone else brings ba k the tor h at one time
Colour the verti es of the o tagon alternately bla k and white. Sin e A and B then they need at least 10 + 2 + 2 + 2 + 1 = 17 minutes.
are of the same olour and in ea h minute the robot moves from a vertex of one
olour to a vertex of the opposite olour, then n must be even. 12-2. Answer: no.
We now label the verti es by 1 8 A = 1 and B = 5 , and denote to so that by Consider an integer N ontaining only digits 2 and 0 and ending in t zeroes
a = (a ; a ; : : :; a )
(k ) (k ) (k ) (k )
the numbers of possibilities, for the robot to rea h ( t>0 ), then
1; 2; : : : ; 8 A=1 k
1 2 8
verti es from in minutes. Using indu tion by m we shall
N = 2 : : : 2 10 = 1 : : : 1 2 5 ;
m>1
t t+1 t
prove that for any
a = (2 + 2 ; 0; 2 ; 0; 2 2 ; 0; 22 ; 0) :
where the dotted part in 2 : : :2 an ontain both 2 -s and 0 -s (in 1 : : :1 both 1 -s
0 1 : : :1 2 5
(2m) 2m 2 m 1 2m 2 2m 2 m 1 m 2
and -s respe tively). Sin e is not divisible by or then in the ase
Obviously we have a = (2; 0; 1; 0; 0; 0; 1; 0) (2)
. Suppose now that the laim is true
when N =n k
both t+1 and t have to be multiples of k , yielding k = 1 .
form=k 2 =s and denote
k 1
, then 12-3. The given inequality is equivalent to
= (s + s; 0; s ; 0; s s; 0; s ; 0) :
a(2 k) 2 2 2 2 a4 + b4 + 4 2a b 2b 2 a
2 2 2 2 2 2
<0:
Sin e the robot an move to any vertex M from either of its neighbouring verti es, Transforming the left side of this inequality we have
we have
a4 + b + 2a b 2b 2 a = (a + b )
4 4 2 2 2 2 2 2 2 2 2 2
4a b =
2 2
a(2 k +1)
= (0; 2s + s; 0; 2s 2 2
s; 0; 2s2 s; 0; 2s2 + s) = (a + b 2ab)(a + b + 2ab) =
2 2 2 2 2 2
= (a b) ) (a + b) =
2 2 2 2
and similarly
= (a b + )(a b )(a + b + )(a + b ) :
a(2 k +2)
= (4s + 2s; 0; 4s ; 0; 4s 2 2 2
2s; 0; 4s ; 0) :
2
Hen e the given inequality is equivalent to
Sin e 4s = 2 2
2s = 2 2k
and
k
, we have proved the laim for m = k + 1 . Hen e (a + b + )(a + b )(b + a)( + a b) > 0 : (1)
the number of possibilities to rea h B =5 from A =1 in n = 2k minutes is
22k 2
2
= 2 (2k 1
1) k 1 k 1
. Here the rst term is positive and at most one of the other three an be negative
(e.g. a + b < 0 and b + a < 0 would give 2b < 0 a ontradi tion). Hen e
12-1. Answer: 17 minutes. (1) holds if and only if a , b and satisfy the triangle inequalities.
17 18

!1 !2 be the ir um ir les of triangles AKL CKL (see Figure 13). n
12-4. Let and and
We have possible distributions of k prizes among n guests, and
Suppose that ! and ! 2 are tangent to ea h other in point C , and let l 2 be their k
ommon tangent. Then
n
6 KLC = 6 KCl2 = 6 BCl2 = 6 BDC : k+1
= nk+1k n
k
: (2)
Hen e KL k BD and 6 ADB = 6 AKL therefore then angle between AB and
! in A is equal to the angle between AL and the tangent to ! Let n be the number of guests and m the a tual number of prizes, then we
the tangent to 1
n
= 2

n

n

= 3 n
2 m
A. A, B , L ! !1
m 1 m+1
in Sin e the points are ollinear then the tangents to and in have and . Substituting from (2) we
A oin ide, i.e. these ir les are tangent to ea h other. m

n m+1
This argument an be reversed to show that
in A implies ! and !2
! and !1
being tangent to ea h other in
being tangent to ea h other
C.
have
m
mn 1 = 2 mn 1 n m n

m+1 m
= 23 mn
and . Hen e
n m + 1 = 2m n = 3m 1
, yielding 2(n m) = 3(m + 1)
, and . Plugging in
n = 3m 1 4m 2 = 3m + 3
here we have m=5 n = 14 , when e and . It
14
remains to al ulate
5 = 2002 .
l2
Kq
IMO Team Sele tion Test: May 2002
Dq
First Day
qC
q
!2 1. The prin ess wishes to have a bra elet with r rubies and s emeralds arranged
q in su h order that there exist two jewels on the bra elet su h that starting with
q B L
A these and enumerating the jewels in the same dire tion she would obtain identi al
! sequen es of jewels. Prove that it is possible to fulll the prin ess's wish if and
l1 !1 only if r and s have a ommon divisor.
2. Consider an isos eles triangle KL1 L2 with jKL1j = jKL2 j , and let KA , L1 B1 ,
Figure 13
L2 B2 os 6 B1 AB2 <
3
Alternative solution. Let !1 and !2 be the ir um ir les of triangles AKL
be its angle bise tors. Prove that

5 .
and CKL . If ! and !1 are tangent to ea h other in A then some homoth- 3. In a ertain ountry there are 10 ities onne ted by a network of one-way nonstop
ety relative to A maps ! to !1 . Sin e K is the interse tion point of AD with ights so that it is possible to y (using one or more ights) from any ity to any
!1 and L is the interse tion point of AB with !1 , and points B and D lie on ! , other. Let n be the least number of ights needed to omplete a trip starting
then this homothety takes D to K and B to L , when e KL k BD . Sin e BK
from one of the ities, visiting all others and returning to the starting point. Find
and DL interse t in C C maps B to K and
then some homothety relative to the greatest possible value of n.
D to L. This homothety then maps the ir um ir le CDB to the ! of triangle
ir um ir le !2 of triangle CKL . Hen e ! and !2 are tangent to ea h other
in C. Se ond Day
Similarly we an show that ! and !2 being tangent to ea h other in C implies
! !1 A.
and being tangent to ea h other in
4. Let ABCD be a y li quadrilateral su h that 6 ACB = 26 CAD and
12-5. Answer: 2002 . 6 ACD = 26 BAC . Prove that jCAj = jCB j + jCDj .
19 20

5. Let 0<< 2 x ; x ; : : :; x
and 1 2 n be real numbers su h that The sine rule in triangle AL1 B2 yields
s sin 6 B AL
sin x + sin x + : : : + sin x > n sin :
1 2 n
t
= sin = os 2=2
os =2
1
Prove that s
= 1 + os
2
AB B
sin(x1 ) + sin(x2 ) + : : : + sin(x n ) > 0 :
and
t 1 + os . The osine rule in triangle 1 2 yields
6. Pla e a pebble at ea h non-positive integer point on the real line, and let n be a
s = t + t 2t os = 2t (1 os );
2 2 2 2 2
xed positive integer. At ea h step we hoose some n onse utive integer points, s
= 2(1 os ) 1 + os
1 + os = 2(1 os )
2
and . Hen e and
remove one of the pebbles lo ated at these points and rearrange all others arbi-
t
trarily within these points (pla ing at most one pebble at ea h point).
Determine whether there exists a positive integer
we an pla e a pebble at a point with oordinate greater than
n su h that for any given N > 0
N in a nite number
os = 31 ++ 2 os
2 os

= 1 2 2 3
3 + 2 os < 1 5 = 5 :
of steps des ribed above.
K
Solutions of Sele tion Test
B1 s D B2
1.
0 r
Note that if g d (r; s) = d > 1 then the prin ess's wish an be fullled. Let r =
d
s s t t s
and s =
0 0 0
we pla e on the bra elet r rubies and s emeralds, then again r
0
d
rubies and s0 emeralds, et . ( d r0 + s0
times) now any two jewels at distan e
22
have the required property.
L2 A L1
Suppose now that the required ordering exists. Label the positions on the bra elet
by 0; : : :; n 1 where n = r + s (thinking of them as modulo n ) and denote by P (i)
the jewel at position i for an ordering P . It su es to show that g d (r; n) > 1 .
Figure 14
Let P be the required ordering, and let a , a + i (where 0 0 and hen e P (b) = P (b + i) = P (b + 2i) = : : : for any position b . Let
Let 1 10 be the ities and denote by ij the minimum number of ights
required to rea h L from L . Let
k be the smallest positive integer su h that n divides ki , and let R be the set of
j i
all positions with rubies. For any b in R we have k distin t rubies at positions m = max x ;
b; b + i; : : :; b + (k 1)i : denote the set of these rubies O(b) and all the orbit 6= ij
i j
of b . A standard argument shows that the set of all rubies on the bra elet is we an assume w.l.o.g. that i = 1, j = m + 1 and the shortest path from L1 to
the disjoint union of some number of orbits, with ea h orbit ontaining k rubies.
L
Hen e k divides r and sin e n divides ki with 06i<n then g d (r; n) > 1 .
m+1 is
L1 ; L2 ; : : :; L ; L :
Denote = 6 L1 KL2 , = 6 KL1 L2 = 6 KL2 L1 and = 6 B1 AB2 (see Fig-
m m+1
2.
ure 14). Sin e B1 and B2 are symmetri relative to KA , we have B1 B2 ? KA We ontinue this path, ying from L to L , then to L et . and nally
and B1 B2 k L1 L2 , i.e. triangle AB1 B2 is isos eles and AK is its angle bise tor.
m+1 m+2 m+3
from L ba k to L1 :
Sin e 6 B2 B1 L1 = 6 L2 L1 B1 = 6 B2 L1 B1 , then B2 L1 B1 is also isos eles. Denote
n
s = jB1 B2 j = jB2 L1 j and t = jAB1 j = jAB2 j . L1 ; L2 ; : : :; L ; L m m+1 !L m+2 ! : : : ! L ! L1 :
n
21 22
Here ea h of the 10 (m +1) + 1 = 10 m se tions denoted by arrows ontains On the other hand,
at most m ights, hen e the entire round-trip ontains at most
X
n
X
n
m + 11 (sin x + : : : + sin x ) + ( os x + : : : + os x ) = os(x x ) 6 n2 ;
m 2
= 121
2 2
m + m (10 m) = m (11 m) 6
1 n 1 n i j
2 4 i=1 j =1
ights, i.e. no more than 30 ights.
a ontradi tion.
An example of a network requiring exa tly 30 ights is shown on Figure 15. 6. Answer: there is no su h n.
For n =1 and n =2 we annot pla e a pebble in any point with a positive
r
D oordinate hen e let n > 3. Consider the innite sum
C
r 2 S =a x1
+a + a + :::
x2 x3
2 where x1; x2; x3 ; : : : are the oordinates of the points with pebbles at some
r r r r r r
given moment and a a positive real number to be determined later. We
a n
r B
show that it is possible to hoose a ording to so that the initial sum
S0 = a + a + a + a + :::
0 1 2 3
onverges (for this it su es to have a > 1 )
r A and at ea h step the sum S an only de rease. Hen e we always have S 6 S0
and it is impossible to pla e a pebble at a point with an arbitrarily large positive
Figure 15 Figure 16 oordinate N , sin e for su iently large N we have a > S0 .
N
To show this onsider for n = 2k 1 the equation
Denote 6 CAD = and 6 BAC = , then 6 ACB = 2 and 6 ACD = 2
4.
Figure 16). Sin e ABCD is y li then 3 + 3 = 6 BCD + 6 BAD = 180
Æ
(see
and
1+ x+x +:::+x 2 k 1
= x +:::+x
k 2k 2
(3)
+ = 60Æ . ABC and ACD gives
Applying the sine rule to triangles
and for n = 2k the equation
jCB j = 2R sin ; jCDj = 2R sin ; jCAj = 2R sin( + 2 ) ; 1+ x+x +:::+x 2 k 1
=x k +1
+:::+x 2k 1
: (4)
where R is the ir umradius of ABCD . Hen e it is su ient to show that
For 0 6 x 6 1 the left side ex eeds the right side but for su iently large positive
sin + sin = sin( + 2) if + = 60Æ . Indeed:
x the right side ex eeds the left side. Hen e the equation has a rooth a > i1 . It
n+1
sin + sin = 2 sin +2 os 2 = 2 12 os +2 = remains to show that for any integers m and t su h that 1 6 t 6

2 the
= os(30Æ ) = sin(60Æ + ) = sin( + 2) : sum of any t 1 elements of A = fa ; a +1; : : :; a + 1g does not ex eed the
m m m n
sum of any t elements of A (here m; : : :; m + n 1 are the hosen n onse utive
integer points and t is the number of points having a pebble before this step and
5. Suppose the laim does not hold, i.e. no pebble after this step hen e t 1 points have no pebble before this step and
a pebble after this step). Note that it su es to have m = 0 and prove that the
sin(x 1 ) + sin(x 2 ) + : : : + sin(x n ) < 0 sum of t 1 A does not ex eed the sum of t smallest elements,
largest elements of
i.e.
whi h gives
1+ a+a + :::+ a > a + :::+a
os x + : : : + os x > os
2 t 1 n t+1 n 1
;
sin (sin x + : : : + sin x ) > n os
1 6 t 6 n+1 = k
1 1
n n
h i
x=a
where
2 . This dire tly follows from (3) or (4) for and
a>1
and hen e
from the fa t that sin e then deleting an equal number of terms from ea h
(sin x + : : : + sin x ) + ( os x + : : : + os x )
1 n
2
1 n
2
> n2 : side makes the left side larger than the right side.
23 24
Estonian math ompetitions 2002/2003 Seniors (grades 11 and 12)
1. Find all quadruples of integer numbers ( a; b; ; d) for whi h
52
a 77b 88 91d = 2002 :
from various materials distributed at the re ent IMO-s.
2. Four rays spread out from point O in a 3-dimensional spa e in a way that the
angle between every two rays is . Find os .
3. Three onse utive positive integers ea h happen to be a power of some prime
Autumn Open Contest: O tober 2002 number. Find all triples of integers with this property.
4. n points of integral oordinates in the plane have been painted white. If some
Juniors (up to 10th grade) points A and B are white, we may paint white the point symmetri al to A in
respe t to B (at every step we paint only one new point). For whi h smallest
natural number n an the initial n points be hosen in a way that any point of
All inner angles of a 7-gon are obtuse, their sizes in degrees being pairwise dierent integral oordinates ould be painted white in a nite number of steps?
integer numbers divisible by 9. Find the sum of the two biggest angles of this 7-
gon. 5. During an ele tion ampaign K tabloid newspapers published ompromising ma-
terial about P politi ians; every politi ian being written about in an odd number
2. Cir les with entres O1 and O2 interse t in two points, let one of whi h be A. of newspapers and every newspaper writing about an odd number of politi ians.
The ommon tangent of these ir les tou hes them respe tively in points P and a) Prove that the number of newspapers K and the number of politi ians P are
Q. It is known that points O1 , A and Q are on a ommon straight line and either both odd or both even numbers.
points O2 , A and P are on a ommon straight line. Prove that the radii of the b) Find the total number of all possibilities whi h newspaper wrote about whi h
ir les are equal. politi ians.
3. A 4-digit number ab d is divisible by the produ t of 2-digit numbers ab and d .
Find all 4-digit numbers with su h property. Solutions of Autumn Open Contest
4. Mari and Jüri ordered a round pizza. Jüri ut the pizza into four pie es by two J1. Answer: 315 .
Æ
straight uts, none of whi h passed through the entre point of the pizza. Mari
Let the interior angles of the 7-gon in degrees be a ; a ; : : :; 9a7 ,
9 1 9 2 where
an hoose two pie es not aside of these four, and Jüri gets the rest two pie es.
a ; a ; : : :; a
1 2 7 are integers and a1 < a2 < : : : < a7 . As all the interior angles
Prove that if Mari hooses the pie e that overs the entre point of the pizza, she
are obtuse, we have a1 > 11 and
will get more pizza than Jüri.
a1 + a2 + : : : + a7 > 11 + 12 + 13 + 14 + 15 + 16 + 17 = 98 :
5. The lo k on the wall shows four numbers (possible times from 00:00 to 23:59),
On the other hand a1 + a2 + : : : + a7 = 100 , be ause the sum of the interior angles
the shapes of the numbers being as shown in the pi ture.
of a 7-gon is (7
Æ Æ
2) 180 = 900 . Bearing in mind that a1 < a2 < : : : < a7
we nd that the only possibilities are (a1 ; a2 ; : : :; a7 ) = (11; 12; 13; 14; 15; 17; 18)
a) How many times a day does the mirror image of the lo k on the wall or-
and (a1 ; a2 ; : : :; a7 ) = (11; 12; 13; 14; 15; 16; 19) . In either ase the sum of the two
respond to some time (the mirror image of number 1 is equal to number
1)?
biggest angles is 9 (a6 + a7 ) = 9 35 = 315 .
Æ
b) How many times a day does the mirror image of the lo k on the wall orre-

spond to the same time as the lo k? J2. Solution 1. By onstru tion 6 O1 P Q = = 6 P QO2 . We also have
2
jO1P j = jO1Aj and jO2Qj = jO2Aj that implies
6 O1P A = 6 P AO1 = 6 O2AQ = 6 AQO2:
1 2
Consequently 6 AO1P = 6 QO2A and is that 100 + n is divisible by d . We have obtained 100 + n = k d = k n ab
(where k > 1 , be ause 100 + n is a 3-digit and d is a 2-digit number). Therefore
6 AP Q = 6 O1P A = 6 AQO2 = 6 P QA : 100 must be divisible by a one-digit number n , implying n equals 1, 2, 4 or 5.
2 2
Consider these four ases separately.
Thus the triangles O1 P Q and O2 QP are similar be ause the respe tive angles
1) Case n=1 gives the equation 101 = k ab . As 101 is prime and k > 1 , this
are equal. As PQ is their ommon side, these triangles are ongruent, giving
jO1P j = jO2Qj .
ase has no solutions.
2) Case n = 2 gives 102 = k 2 ab that implies 51 = k ab . As 51 = 3 17 , the
Æ only possibility is k = 3 , ab = 17 , and we obtain one 4-digit answer 1734 .
Solution 2. Let 6 O1 QP = 6 AQP = . As 6 O1 P Q = 90 , we have
6 P O1A = 6 P O1Q = 90Æ . Be ause the triangle P O1A is isos eles, we obtain 3) Case n=4 gives 104 = k 4 ab that implies 26 = k ab . As 26 = 2 13 , the
only possibility is k = 2 , ab = 13 , and we obtain one 4-digit answer 1352 .
6 O1P A = 6 O1 AP =
1
180
Æ (90
Æ ) = 45Æ + : 4) Case n = 5 gives 105 = k 5 ab that implies 21 = k ab . As 21 = 3 7,
2 2 this ase has no solutions (the number 21 annot be expressed as a produ t of a
Æ
On the other hand 6 P QO2 = 90 implies 6 AQO2 = 90
Æ . The triangle AQO 2-digit number and a number ex eeding 1).
is isos eles, onsequently 6 AQO2 = 90
Æ . The onditions of the problem imply2 J4. Constru t a straight line parallel to one of the uts passing through the entre
6 O AQ = 6 O AP , therefore
2 1 point of the pizza O this line passes through two pie es of all four, dividing
both into two parts. Let a, b, and d be areas as shown in the gure 2. If Mari
45 +
Æ = 90
Æ ; hooses the pie e having the entre point of the pizza, the total area of her share
2
is (a + b) + ( d) and the total share of Jüri is ( a b) + ( + d) . To omplete
Æ Æ Æ
giving = 30 . Thus 6 P O1 A = 60 and 6 O1 AP = 6 O2 AQ = 60 , i.e. the
the solution, one has to prove that a + b + d > a b + + d , or b d > d b .
Æ
triangles O1 AP and O2 AQ are equilateral. As 6 O1 P Q = 6 P QO2 = 90 , the
This is similar to b>d that evidently holds.
A to the straight line P Q is equal to jO1P j . On the other
1
distan e of the point
2
hand the distan e equals
1
2
jO2Qj , giving us jO1P j = jO2Qj .
PSfrag repla ements
O a
0
1
1
0
d b
PSfrag repla ements O1 O2
Figure 2
A
J5. Answer: a) 121 ; b) 11 .
P Q a) The numbers having a number as its mirror image are 0, 1, 2, 5 and 8. The
number 8 an only be the last digit of hours or minutes, the mirror image of whi h
Figure 1
is respe tively the rst digit of minuts or hours, implying that 8 annot be in any
time onsidered in the problem. Thus the rst digit of hours an be 0 , 1 or 2 and
J3. Answer: 1352 , 1734 .
the last digit of hours an be 0 , 1 , 2 or 5 , totally giving 3 4 1 = 11 numbers
We need the integer ab d = 100 ab + d be divisible by ab , therefore d must ( 25 is not an hour) as hour time. Sin e the mirror image of minutes is hours, the
be divisible by ab . Let d = n ab , then n is a one-digit number (sin e a 6= 0 ). minutes an be any of these 11 numbers. Consequently the mirror image of the
The ne essary ondition that ab d = (100 + n) ab would be divisible by ab d lo k orresponds to some time 11 11 = 121 times a day.
3 4
b) For having the mirror image time equal to the real image time, the minutes Hen e the entre point K of the segment A3 A4 is also on the plane ! , having
must equal the mirror of the hours, i.e. any of the hours possible found in a) gives jKA1j = jKA2 j . As A1 M1 ? KA2 and A2 M2 ? KA1 , we have that KM1 OM2
exa tly one su h time. Therefore the mirror image time equals the real image time is a y li quadrangle that implies
11 times a day.
= 6 A1OA2 = 6 M1 OM2 = 180Æ 6 M1 KM2 :
S1. Answer: the only su h quadruple is (2 2 ; ; 1; 1) .
After fa torisation of the bases of powers and 2002 , we may rewrite the given
As KA2 is a median of A2 A3A4 , the ratio jKM1j : jKA2j = 1 : 3,
equation as 6 M1 KM2 jKM1j jKM1 j 1
therefore os =
jKA1j =
jKA2j =
3
. Finally we obtain
2a
(2 13a) (7b 11b) (23 11 ) (7d 13d ) = 2 7 11 13 os = os 6 M1 KM2 = 1
.
3
or in the form
A3
2
a 7b+d 11b+ 13a+d = 2 7 11 13:
2 +3
For the last equation to hold, the powers of respe tive primes on the left and right
side must be equal. This implies a system of equations
8 0000000000
1111111111 K 11111111
00000000
> 2a + 3 =
PSfrag repla ements
> 1 000000000011111111
1111111111 00000000
< 000000000011111111
1111111111 00000000
b +d = 1
:
000000000011111111
1111111111
M
0000000000
1111111111
00000000
00000000
11111111
>
> b+ = 1 0000000000
1111111111
000000000000
111111111111
1 00000000
11111111
00000000000
11111111111
0000000000
1111111111 O 11111111
000000000000
111111111111 00000000
:
a+d = 1 0000000000
1111111111
000000000000
111111111111 M2
00000000000
11111111111
00000000
11111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
000000000000
111111111111 A4
00000000000
11111111111
0000000000
1111111111 00000000
11111111
000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
The se ond and the third equation give = d , and the se ond and the fourth
000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
a = b . Multiplying the third equation by 2 and subtra ting from the rst gives 000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
000000000000
111111111111
00000000000
11111111111
0000000000
1111111111 00000000
11111111
= 1 and b = 2 . The rest of the unknown variables must then be a = 2 000000000000
111111111111
00000000000
11111111111
1111111111
0000000000
000000000000
111111111111 00000000
11111111
00000000000
11111111111
d 0000000000
1111111111
A2111111111100000000A
11111111
000000000000
111111111111
11111111111
00000000000
000000000000000000
and = 1. We have obtained the only quadruple satisfying the onditions 11111111 1
; ;
(2 2 1 ; 1) .
Figure 3
1
S2. Answer:
3
.
; ; ; ; ; ;
S3. Answer: (1 2 3) , (2 3 4) , (3 4 5) and (7 8 9) . ; ;
At least one of the three onse utive integers is divisible by 2 and hen e must be
Consider the points A1 , A2 , A3 and A4 on the respe tive rays equidistant from 1 2 3
O . Consider any pair of points Ai1 ; Aj1 and Ai2 ; Aj2 , where i1 6= j1 and i2 6= j2 . a power of 2 (with positive exponent). Considering 2 = 2 , 2 = 4 and 2 = 8 ,
k
As jOAi1 j = jOAi2 j and jOAj1 j = jOAj2 j , also 6 Ai1 OAj1 = = 6 Ai2 OAj2 ,
we nd the solutions. We prove that no more solutions exist. Say 2 belongs to
we obtain that the triangles OAi1 Aj1 and OAi2 Aj2 are ongruent, implying
a triple in question, where k > 4. Sin e the only onse utive even numbers that
k k k
jAi1 Aj1 j = jAi2 Aj2 j . Therefore all segments Ai Aj , where i 6= j are of equal are powers of 2 are 2 and 4 , the triple must be in the form 2 1, 2 and 2 + 1 .
lengths. Consequently all the fa es of Ai Aj Ak , having i; j; k pairwise dierent,
As one of three onse utive integers is always divisible by 3, one of the numbers
k k
are equilateral triangles, and A1 A2 A3 A4 is a regular tetrahedron.
2 1 and 2 + 1 must be a power of 3 . Consider these two ases separately.
n k 1 ; onsider the sides of the equation modulo 8 : 2k 1 7 , but
Cut this tetrahedron with a plane ! ontaining the points O , A1 and A2 . The
1) Say 3 = 2
2x 2x+1
straight line A1 O passes through the triangle A2 A3 A4 in its entre M1 and the
3 1 and 3 3 , a ontradi tion.
straight line A2 O passes through the triangle A1 A3 A4 in its entre M2 (see gure
n k
2) Say 3 = 2 + 1 ; onsider the sides of the equation modulo 4 : 2 + 1 1
k
2x 2x+1
3). By onstru tion M1 and M2 are on the plane ! . The straight line A2 M1 and 3 1, 3 3 implying that n is even. Now onsider the sides of the
3y 3y +1 3y +2 6z
as a median of the triangle A2 A3A4 bise ts the segment A3A4 and the straight equation modulo 7 : sin e 2 + 1 2, 2 + 1 3, 2 + 1 5 and 3 1 ,
line A1 M2 A1 A3A4 bise ts the segment A3 A4 as well.
as a median of the triangle 3
z 2 , 36z+4 4 at the same time, we must have 2k + 1 = 3n 2 (mod 7) .
6 +2
5 6
Consequently the rst number of the triple 2
k 1 is divisible by 7 , therefore it Remark. The onstru tion des ribed above an be presented more onveniently if
is a power of 7 : 2
k 1 = 7
m. Consider the sides of the equation modulo 16 : one performs su h ane transdu tion that takes the three verti es of the parallel-
2
k 1 15 , but 72w 1 , 72w+1 7 a ontradi tion. ; ; ;
ogram to points (0 0) , (0 1) and (1 0) then the points to be hosen into the
subset are exa tly the points at least one of the oordinates of whi h is an even
0
1
0
1 number.
1 0
0 1
00
11
00
11
00 0
11 1 0
1 1
0 Solution 2. Similarly to the previous solution we noti e that it su es to nd su h
1 0
0 1 0
1 1 0
0 1 a set of points of integral oordinates on the plane that:
0
1 1 0
0 1 0
1 1 0
0 1 0 0
1
1 1 (i) is symmetri al in respe t of every its point (i.e. any point that is symmetri al
1
0 0
1 0
0 1 1 0
0 1 0
1 1 0
0 1 to any point in the set in respe t to any other point in the set also belongs to
1 0
0 1 1 1
0
0 1 1 0
0 1 0
1 the set);
0
1 0
0 1 1 0
0 1 0
1 1 0
0 1 (ii) ontains all three initial white points;
0
1 1 0
0 1 0
1 1 0
0 1 0 0
1
1 1 (iii) does not ontain all points of integral oordinates on the plane.
1
0 0
1 0
0 1 1 0
0 1 0
1 1 0
0 1
In order to onstru t su h a set we onsider the parities of the three initial points
1 0
0 1 1 1
0
0 1
0 1 0
0 1 0
1 and see that one an (if ne essary, by shifting the oordinate system by one unit)
1 0
0 1 1 0
0 1 0
1 always obtain the situation where all of these three points have at least one of
1
0
0 1 0
0 1 0
1 the oordinates an even number. Now the set in question an be expressed by all
1
1 0
0 1 of these points of integral oordinates on the plane that have at least one of the
oordinates an even number (if two points of integral oordinates are symmetri al
in respe t to some third point of integral oordinates, then the respe tive oor-
Figure 4 dinates of the two points are of the same parity; the onditions (ii) and (iii) are
evidently satised).
S4. Answer: n = 4.
Solution 1. It is evident that one annot nd suitable olouring in the ase n=2 S5. Answer: b) 2
( K 1)( P 1)
.
(one an only olour the points on the straight line determined by the two initial Solution 1. a) Consider the situation des ribed in the problem as a table onsisting
points) and that suitable olouring exists in the ase n= 4 (e.g. olour points of K rows and P olumns, orresponding respe tively to the newspapers and the
; ; ; ;
(0 0) , (0 1) , (1 0) and (1 1) ). politi ians, and olour the square in the i th row and the j th olumn bla k in
It remains to prove that no su h olouring exists in the ase n = 3. If one the ase the i th newspaper wrote about the j th politi ian. Let ai and bj be
hooses three points ollinear, then no points outside the ommon straight line respe tively the number of oloured squares in the i th row and in the j th olumn.
an be oloured white. Now we prove that if one hooses three points at three For a1 + : : :+ aK = b1 + : : :+ bP (either sum represents the total number of oloured
verti es of a parallelogram (su h parallelogram always exists if the points are not squares) and a ording to the onditions of the problem all the addends are odd,
hosen ollinear), one annot olour the fourth vertex (having evidently integral the ounts of the addends K and P must be of the same parity.
oordinates) of this parallelogram. For this it su es to show su h subset of all b) The number of the squares not being in the last row and in the last olumn
points of integral oordinates that is symmetri in respe t of any of its point and is ( K 1)(P 1) , and as we an either olour or leave un oloured any of these
ontains the three verti es of the parallelogram, but does not in lude the fourth squares, the total number of possibilities to olour these squares is 2
(K P
1)( 1)
.
vertex. We obtain su h set of points in the following way: Show that any su h olouring an be ompleted exa tly uniquely to the olouring
(1) Consider a latti e on the plane, the axes of the latti e being determined by of the whole table satisfying the onditions of the problem (i.e. the number of
the edges of the parallelogram, one latti e-point in the opposite vertex of the oloured squares in every row and in every olumn is odd). Indeed: the square
un- oloured vertex and the distan e between latti e points being equal to in the last olumn must be oloured exa tly in these rows (from the 1 st until the
twi e the respe tive side-length of the parallelogram. ( K 1 )th) that have an even number of squares oloured. After that the square
(2) Choose the latti e-points and the midpoints of segments that onne t the in the last row must be oloured exa tly in these olumns (from the 1 st until the
neighbouring latti e points and are parallel to the axes of the latti e (these P th) that have an even number of squares oloured. It remains to he k that the
points are shown white in gure 4). last row has an odd number of squares oloured as well this is proved by the
7 8
fa t that the numbers of rows and olumns are by a) odd, all the rows ontain
altogether as many oloured squares as do all the olumns, and all the rows ex ept
2. The shape of a dog kennel from above is an equilat-

C
q
eral triangle with side length 1 m and its orners in
points A , B and C , as shown in the pi ture. The
q q q
the last and all the olumns ontain an odd number of oloured squares.
hain of the dog is of length 6 m and its end is xed
K A B
Solution 2. a) Consider the situation des ribed in the problem as a graph of K to the orner in point A . The dog himself is in point K in a way that the hain
yellow and P brown verti es, where one has edges only between verti es of dier- is tight and points K , A and B are on the same straight line. The dog starts to
ent olours; an edge between the i th yellow and the j th brown vertex shows that move lo kwise around the kennel, holding the hain tight all the time. How long
the i th newspaper wrote about the j th politi ian. A ording to the onditions of is the walk of the dog until the moment when the hain is tied round the kennel
the problem an odd number of edges emerges from every vertex of the graph, the at full?
sum of all these numbers equals twi e the number of edges (as every edge ounts
3. A ar, a motor y le, a moped and a bi y le were driving at unvaried speeds in a
at its both vertex) and is thus an even number. Hen e the number of verti es
K + P is an even number, implying that K and P are of the same parity. straight road. At 12:00 the ar passed the moped driving in the same dire tion and
A1 ; : : :; AK and the brown
b) Let the yellow verti es be verti es B1 ; : : : ; BP .
met the bi y le and the motor y le driving in the opposite dire tion respe tively
One has 2
( K 1)(P 1)
possibilities for joining the verti es A1 ; : : :; AK 1 and
at 14:00 and 16:00. The motor y le met the moped at 17:00 and passed the bi y le
at 18:00. At what time did the moped meet the bi y le?
B1 ; : : :; BP 1 (one may or may not have an edge between the verti es in any
of the pairs ( Ai ; Bj ) ). We shall prove that every su h graph an be ompleted 4. Consider the points A1 and A2 on the side AB of the square ABCD taken in
to a graph satisfying the onditions of the problem (i.e. an odd number of edges su h a way that jAB j = 3 jAA1 j and jAB j = 4 jA2 B j , similarly onsider points B1
emerging from every vertex) by adding edges that emerge from AK and BK in an and B2 , C1 and C2 , D1 and D2 respe tively on the sides BC , CD and DA .
exa tly unique way. Indeed, one must draw edges from the vertex AK exa tly into The interse tion point of straight lines D2 A1 and A2 B1 is E , the interse tion
the verti es Bj ( j = 1; : : :; P 1 ) that had an even number of edges emerging. point of straight lines A2 B1 and B2 C1 is F , the interse tion point of straight
After that one must draw edges from the vertex BP exa tly into the verti es Ai lines B2 C1 and C2 D1 is G and the interse tion point of straight lines C2 D1 and
(i = 1; : : :; K ) that had an even number of edges emerging. It remains to he k D2 A1 is H . Find the area of the square EF GH , knowing that the area of ABCD
that the vertex BP now also has an odd number of edges this is proved by the is 1 .
fa t that a ording to a) the total number of verti es is even, all the rest of the
verti es has an odd number of edges emerging and the total sum of edges emerging 5. Is it possible to write one of the letters A, B , C or D in every square of an
from any vertex is twi e the number of edges, thus an even number. innite he ked paper in su h a way that every 2 2 square ontains all four
letters?
Seniors (11th and 12th grade)
1. Find the values of sharp angles and that satisfy
8
> 3
> os2 + os2 =
<
Spring Open Contest: February 2003 2
:
>
sin
> 1
: sin =
4
Juniors (up to 10th grade) 2. Juku has 2003 wooden sti ks (nothing is known about their lengths). Juku on-
stru ts a re tangle of all these sti ks, breaking some sti ks into two parts. Find
the least number of breakings that would be su ient for any lengths of the sti ks.
1. A four-digit number a not ontaining digit 9 is a square of an integer. If we
3. The sequen e fFng is dened as follows:
in rease every digit of a by 1 , we obtain a square of an integer again. Find all
4-digit numbers with su h property. F1 = 1 ; F2 = a; Fn = Fn 1 + Fn 2 ; if n > 3:
9 10
a) Do there exist integers a and N> 1 that no member of the sequen e fFng
would be divisible by N?
b) Do there exist integers a and N >1 that some two onse utive members of
the sequen e fFng would be divisible by N ?
4. Consider the points D , E and F on the respe tive sides BC , CA and AB of the
triangle ABC in a way that the segments AD , BE
CF have a ommon point
and C
jAP j jBP j jCP j
q
P. B
jP Dj = x , jP E j = y jP F j = z . Prove that xyz x + y + z) = 2 . q q q
Let and (
K A
5. Consider nn squares painted bla k and white as a hessboard in a omputer
monitor. One an mark any re tangle onsisting of whole squares by mouse-
dragging and invert the olours of the squares in a marked re tangle by mouse-
li king. Find the smallest number of mouse- li ks by what one an onvert all
the squares mono hrome.
Figure 5
J3. Answer: At 15.20.
Solution 1. Constru t a diagram with the graphs of movement of the ar, the
motor y le, the moped and the bi y le, and have one axis for time in hours and
the other for distan e in kilometers. Let A, B , C , D, E and F be points in this
diagram that respe tively orrespond to the meeting of the ar and the moped, the
ar and the bi y le, the ar and the motor y le, the motor y le and the moped,
the motor y le and the bi y le, and the moped and the bi y le (see gure 6). In
Solutions of Spring Open Contest this ase B is the midpoint of segment AC and D is the midpoint of segment CE .
Hen e F is the interse tion point of the medians AD and EB in triangle ACE ,
implying that jAF j = 23 jADj and the dieren e in time oordinates between
J1. Answer: 2025 is the only su h number.
Solution. Let a = x2 and a + 1111 = y2 . Then
points A and F is
2
3
(17 12) =
10
3
hours. Therefore the moped met the bi y le
10
1111 = 101 11 = y2 x2 = (y + x)(y x) 3
hours, or 3 hours and 20 minutes later than it met the ar, i.e. at 15.20.
that implies y + x = 101 , y x = 11 or y + x = 1111 , y x = 1 . The rst ase
leads to x = 45 , y = 56 and x = 2025 . The se ond ase leads to x = 555 and
2
y = 556 , but the squares of these numbers are not 4-digit numbers.
J2. Answer: 14 m.
Solution: The dog's path onsists of six 120
Æ ar s of a ir le, the radii of whi h
are 6 , 5 , 4 , 3 , 2 ja 1 metres (see gure 5). Hen e the total length of the dog's
path is
2
1
3
2 (6 + 5 + 4 + 3 + 2 + 1) =
3
21 = 14 (m):
11 12
x the ar
G
6
the motor y le
C D C2 C1 C
the moped
D B2
the bi y le
B D1
F E F
H
A D2 B1
-
12.00 14.00 16.00 17.00 18.00 t A A1 A2 B
Figure 6
E
Solution 2. Say zero time and zero point the time and pla e of the meeting of Figure 7
the ar and the moped, and let the positive dire tion of the axis be the dire tion
where the ar and the moped move. Let the speeds of the ar, the motor y le, J4. Right triangles AA D
1 2 and EA1 A2 are similar, sin e 6 AA1 D2 = 6 EA1 A2 (see
the moped and the bi y le be respe tively a , b , and d , then at time t their gure 7). We shall show that jD2A1 j = jA1A2 j , i.e. these triangles are ongruent.
respe tive lo ations are at , 4a b(t 4) (sin e the ar and the motor y le met at Indeed,
time t = 4 ), t and 2a d(t 2) (sin e the ar and the bi y le met at time t = 2 ). s
t = 5 , we obtain an equation p
As the motor y le and the moped met at time
jD2 A1j = jD2Aj2 + jAA1 j2 = 1
16
jAB j2 + 19 jAB j2 = 125 jAB j
4a b = 5 ; (1)
and
and from the meeting of the motor y le and the bi y le at time t=6 we get

4 a b
2 =2 a 4 d: (2)
jA1A2 j = jAB j jAA1j jA2 B j = 1
1
3
1
4
jAB j = 125 jAB j :
What should be omputed is the meeting time of the moped and the bi y le, Therefore jD2 A1 j = jA1 A2 j , jAA1 j = jA1E j , jA2 B j = jD2 Aj = jEA2j = jHD2 j and
a+d
i.e. the time x that satises x = 2a d(x 2) , or x = 2
+d
. Having
jHE j = jHD2j + jD2 A1j + jA1E j = jA2B j + jA1A2 j + jAA1 j = jAB j :
expressed b from the equation (1) and substituted it into the equation (2), we
obtain 4a 2(4a 5 ) = 2a 4d , or 6a = 10 + 4d . Adding to both sides 6d gives Consequently the squares ABCD and EF GH are of the same side length, i.e.
a+d EF GH is 1 as well.
= 2
10 10 the area of the square
a d
6( + ) = 10( + d) , implying x = 2 = . Therefore the moped
+d 6 3
10 J5. Answer: no.
met the bi y le hours or 3 hours 20 minutes later than it met the ar, i.e. at
3 Solution. Assume the required onguration of the letters exists. Then no row an
15.20. ontain two similar letters onse utively, therefore one must have three dierent
letters onse utively without loss of generality assume they be A , B and C .
The next row must ontain aligned to them the letters C, D and A , the row
after the next again has A, B and C , et . We see that any of the three olumns
ontains only two letters, a ontradi tion with the assumption made above.
S1. Answer: = = 30Æ .
13 14
Substitutions os
2
=1 sin
2
and os2 = 1 sin
2
give a system of equation S3. Answer: a) yes; b) no.
similar to the given system a) Having a = 3 , the residues of Fn modulo N = 5 are the following
8
> sin2 + sin2 = 1
> ; ; ; ; ; ; ::: :
1 3 4 2 1 3
<
2
:
>
sin
1
>
: sin =
Sin e every term in the sequen e is determined by two previous terms, the residues
4 modulo 5 are also determined by two previous residues, hen e the quadruple
; ; ;
of residues 1 3 4 2 will repeat innitely and onsequently no member of the
By letting x = sin and y = sin , one has a system of equations in x and y sequen e is divisible by 5 .
8 Fn = Fn Fn Fn Fn Fn
b) Having written 1+ 2 in the form = 1 , we see that
> x2 + y 2 = 1
>
2
< every term of this sequen e equals the dieren e of two next onse utive terms.
2
>
: Therefore if some two onse utive terms of the sequen e were divisible by N > 1,
xy
1
>
: = all the terms pre eeding them would also be divisible by N, in luding the rst
4
term F1 = 1 a ontradi tion. Consequently for any hoi e of a two onse utive
Having added two times the se ond equation to the rst, we obtain terms of the sequen e are pairwise primes.
(x + y)2 =
1
2
+2 14 = 1 . Sin e and are a ute angles, x and y are pos- S4. Let the base points of heights drawn to the segment AB in triangles ABC and
itive numbers, therefore x + y = 1 , or y = 1 x . Substituting it into the se ond ABP be respe tively K and L (one or both of these points may lie on the
extension of side AB ) and let Q be the interse tion point of the segment CK
x (1 x) = or
1
with straight line s , passing through point P and being parallel to side AB (see
equation in the system, we obtain
4

gure 8). Let SXY Z be the area of triangle XY Z , then
1 1 2
0= x2 x + = x ; jP F j jQK j jP Lj SABP
4 2
jCF j = jCK j = jCK j = SABC
= 30Æ .
1 1
x= x+y = 1 y=
from whi h
2
and The equation now gives
2
and
jP Dj SBCP jP E j SACP
= 30Æ . and similarly
jADj = SABC and jBE j = SABC . Sin e
S2. Answer: 2 . jP Dj 1 1
Solution. Juku may ut one sti k into two halves; onsider these halves one pair jADj = 1 + jAP j = 1 + x
of sides of a re tangle and divide the rest of the sti ks into two sets A and B. If
jP Dj
the total lengths of the sti ks in sets A and B are equal, no more sti ks need to
be broken but if, for instan e, the total length of the sti ks in A is bigger, we jP E j 1 jP F j 1
put one by one sti ks from A to B until the inequality of total lengths reverses. and similarly
jBE j = 1 + y and jCF j = 1 + z , we have
Now it su es to break the last sti k that was put from A to B (in the ase the
total lengths equalized no breaking is ne essary). 1 1 1 SBCP + SACP + SABP
+ + = =1 ;
One does not nd su ient to break only 1 sti k e.g. in the ase where the 1+ x 1+ y 1+ z SABC
2002
lengths are 1 , 2 , . . . , 2 . If it would be enough to break only one sti k in
order to onstru t a re tangle, one had its two sides of integer and two sides of y z x z x y x
or (1+ )(1+ )+(1+ )(1+ )+(1+ )(1+ ) = (1+ )(1+ )(1+ ) , giving after mul- y z
fra tional lengths (if all four sides were of integral length, their sum would be an tipli ation and olle ting similar terms the required equation xyz x y z
( + + ) = 2.
even number, but 1 + 2 + : : : + 22002 is odd). Hen e one pair of opposite sides
should be onstru ted only of initial sti ks, but it is not possible, sin e one an
nd a sti k longer than the total sum of the others in every subset of the initial
sti ks.
15 16
C 4. Ella the Wit h was mixing a magi elixir whi h onsisted of three omponents:
140 ml of reindeer moss tea, 160 ml of y agari extra t, and 50 ml of moonshine.
She took an empty 350 ml bottle, poured 140 ml of reindeer moss tea into it and
E D
started adding y agari extra t when she was disturbed by its bla k at Mesto.
Q P s So she mistakenly poured too mu h y agari extra t into the bottle and noti ed
her fault only later when the bottle lled before all 50 ml of moonshine was added.
Ella made qui k al ulations, arefully shaked up the ontents of the bottle, poured
out some part of liquid and added some amount of mixture of reindeer moss tea
and y agari extra t taken in a ertain proportion until the bottle was full again
and the elixir had exa tly the right ompositsion. Whi h was the proportion of
A K L F B reindeer moss tea and y agari extra t in the mixture that Ella added into the
Figure 8 bottle?
S5. Answer: n , if n is even; n 1 , if n is odd.
5. Is it possible to over an nn hessboard whi h has its enter square
ut out with tiles shown in the pi ture (ea h tile overs exa tly 4
Solution. Every mouse- li k an onvert mono hrome at most 4 pairs of squares
of dierent olours on the outer edge of the re tangle. As one has 4( n 1) of squares; tiles an be rotated and turned around) if a) n = 5 ; b)
su h pairs initially, one requires at least n 1 mouse- li ks. For even n one needs n = 2003 ?
at least n mouse- li ks, sin e the orner squares are not of same olour and by
marking any re tangle ontaining a orner square one onverts mono hrome only 10th grade
2 pairs of squares of dierent olours on the outer edge of the re tangle.
These numbers prove to be su ient. We may hoose subsequently the 2nd, 4th, 1. The pi ture shows 10 equal regular pentagons where ea h
6th, . . . row and then the 2nd, 4th, 6th, . . . olumn for odd n we have altogether
two neighbouring pentagons have a ommon side. The
2 n 2 1 = n 1 mouse- li ks and for even
n
n we have altogether 2 = n mouse-
2
smaller ir le is tangent to one side of ea h pentagon and
the larger ir le passes through the opposite verti es of
li ks.
these sides. Find the area of the larger ir le if the area of
the smaller ir le is 1 .
Final Round of National Olympiad: Mar h 2003
m2 + n2
9th grade 2. Find all possible integer values of where m and n are integers.
mn
1. Let A1 , A2 , . . . , Am and B2 , B3 , . . . , Bn be the points on a ir le su h that 3.
Æ
In the a ute-angled triangle ABC all angles are greater than 45 . Let AM and
A1A2 : : :An is a regular m -gon and A1 B2 : : : Bn is a regular n -gon whereby BN be the heights of this triangle and let X and Y be the points on MA and
n > m and the point B2 lies between A1 and A2 . Find 6 B2 A1 A2 . NB , respe ively, su h that jMX j = jMB j and jNY j = jNAj . Prove that MN
2. Find all positive integers n su h that
and XY are parallel.
hni h i h i a , b , and
6= n + 2n :
4. Let be positive real numbers not greater than 2 . Prove the inequality
n+
6 2 3
ab
a+b+
6 4
:
x
Here [ ℄ denotes the largest integer not greater than x. 3
3. In the re tangle ABCD with jAB j < 2jADj , let E be the midpoint of AB and 5. The game Clobber is played by two on a strip of 2 k squares. At the beginning
F a point on the hord CE su h that 6 CF D = 90Æ . Prove that F AD is an there is a pie e on ea h square, the pie es of both players stand alternatingly. At
isos eles triangle. ea h move the player shifts one of his pie es to the neighbouring square that holds
17 18
a pie e of his opponent and removes his opponent's pie e from the table. The 12th grade
moves are made in turn, the player whose opponent annot move anymore is the
winner.
1. Jüri and Mari both wish to tile an nn hessboard with ards shown
in the pi ture (ea h ard overs exa tly one square). Jüri wants that for
Prove that if for some k the player who does not start the game has the winning
ea h two ards that have a ommon edge, the neighbouring parts are of
strategy, then for k + 1 and k + 2 the player who makes the rst move has the dierent olor, and Mari wants that the neighbouring parts are always of the same
winning strategy.
olor. How many possibilities does Jüri have to tile the hessboard and how many
possibilities does Mari have?
2. Solve the equation
px = log x .
2
11th grade
3.
Æ
Let ABC be a triangle with 6 C = 90 and D a point on the ray CB su h that
jAC j jCDj = jBC j2 . A parallel line to AB through D interse ts the ray CA at
1. E . Find 6 BEC .
Juhan is touring in Europe. He stands on a highway and wat hes ars. There are 4. Call a positive integer lonely if the sum of re ipro als of its divisors (in luding
three ars driving along the highway at onstant speeds: an Opel and a Trabant in 1 and the integer itself ) is not equal to the sum of re ipro als of divisors of any
one dire tion and a Mer edes in the opposite dire tion. At the moment when the other positive integer. Prove that a) all primes are lonely; b) there exist innitely
Trabant passes Juhan, the Opel and the Mer edes lie at equal distan es from him many non-lonely positive integers.
in opposite dire tions. At the moment when the Mer edes passes Juhan, the Opel
and the Trabant lie at equal distan es from him in opposite dire tions. Prove that 5. On a lottery ti ket a player has to mark 6 numbers from 36. Then 6 numbers from
at the moment when the Opel passes Juhan, also the Mer edes and the Trabant these 36 are drawn randomly and the ti ket wins if none of the numbers that ame
lie at equal distan es from him in opposite dire tions. out is marked on the ti ket. Prove that a) it is possible to mark the numbers on
9 ti kets so that one of these ti kets always wins; b) it is not possible to mark the
2. Prove that for all positive real numbers a , b , and numbers on 8 ti kets so that one of ti kets always wins.
p p
3
ab +
1
a
+
1
b
+
1

>2 3:
Solutions of Final Round
When does the equality o ur?

9-1. Answer: .
3. Let ABC be a triangle and A1 , B1 , C1 points on BC , CA , AB , respe tively,
m n
su h that the lines AA1 , BB1 , CC1 meet at a single point. It is known that A ,
B1 , A1 , B are on y li and B , C1 , B1 , C are on y li . Prove that a) C , A1 , Let O be the enter of the ir le. Then
6 B2 A1 A2 = 6 B2 OA2=2 = (6 A1OA2 6 A1 OB2 )=2 = =m =n .
C A are on y li ; b) AA BB CC
1, 1, 1, ABC .
1 are the heights of
Prove that there exist innitely many positive integers n su h that
p n is not 9-2. Answer: n = 6k + 1 where k = 0, 1, . . .
4.
and n
pn℄ . (Here [x℄ denotes the largest integer
an integer is divisible by [ not
If the equality holds (does not hold) for some n, then it holds (does
greater than x .) +6
hn i hni
not hold) for n
+ 6. Indeed, ( + 6) + n = + n
+ 7 and
6 6
h + 6 i h 2( + 6) i h i h 2 i
n n n n
5. For whi h positive integers n is it possible to over a (2n+1) (2n+1) + = + + 7 . Therefore it su es to he k the values
hessboard whi h has one of its orner squares ut out with tiles shown 2 3 2 3
in the gure (ea h tile overs exa tly 4 squares; tiles an be rotated n = 0, 1 , 2 , 3 , 4 , 5 . Doing so, we nd that n
= 1 is the only ase when the
and turned around)? equality is violated.
19 20
9-3. Sin e E is the midpoint of AB , the right-angled triangles EAD and EBC Let O be the ommon enter of the ir les and ABCDE one of the pentagons
Æ
are ongruent. Also, sin e 6 DAE = 6 DF E = 90 , the quadrilateral AEF D whereby the inner ir le tou hes the side AB and the outer ir le passes through
is y li . Therefore 6 AF D = 6 AED = 6 BEC . On the other hand, D . Let Q be the enter of ABCDE . Then 6 AOB = 360Æ=10 = 36Æ and
6 ADF = 180
Æ 6 AEF 6 BEC . Hen e 6 AF D = 6 ADF .
= 6 ADB = 6 AQB=2 = 360Æ=10 = 36Æ . So the isos eles triangles AOB and ADB
are ongruent and the radius of the inner ir le is exa tly half of the radius of the
Solution 2. Let K be the interse tion of CE and DA . Then EA is the midline
KCD , therefore AK = AD . Sin e 6 KF D = 90Æ , the points K , F , and D
outer ir le.
of
A . This means AF = AD .
lie on the ir le entered at 10-2. Answer: 2 and 2.
Using the fa t that mn divides m + n , we on lude that m divides n and n
2 2 2
9-4. Answer: 2 : 3 .
a 0
divides m . Therefore m and n have the same prime divisors. Now let m = p m
2
When the bottle got lled rst time, the reindeer moss tea made up 140 ml of b 0 0 0
and n = p n where the prime p is not a divisor of m and n . If for example
its 350 ml volume. Therefore the mixture that was poured out onsisted of 2 5 = a > b , then mn and m2 are both divisible by pa+b but n2 is not (it is only
=
of reindeer moss tea and 3 5 of other substan es. Sin e the amount of liquid
divisible by p2b where 2b < a + b ). That is, (m2 + n2)=(mn) is not an inte-
So it must be a = b and onsequently m = n . Then
that was poured out of the bottle and the amount that was added afterwards are
ger, a ontradi tion.
equal and the bottle nally ontained 140 ml of reindeer moss tea again, also the
m2 + n2 n
2 2
=
mixture that was added was made up of exa tly 2 5 of reindeer moss tea. So the
amount of moonshine in that mixture was 3 5 . = mn
=
n2 = 2 .
10-3. The quadilateral AXY B is y li sin e 6 AXB = 180
Æ 6 MXB = 135Æ and
9-5. Answer: a) yes; b) yes.
analogously 6 AY B = 135Æ . Also, the quadrilateral ANMB is y li . Therefore
6 MXY Æ 6 AXY = 6 ABN = 6 AMN whi h means that MN and XY
= 180
a) See gure 9.
are parallel.
b) Figure 10 shows how to tile the border of width 2 around the (4k 1) (4k 1)
4 re tangle an be overed with two tiles). Starting
square (we use the fa t that 2 p
10-4. First, from the inequality between arithmeti mean and geometri mean we get
from the tiling for n = 3 (its existen e an be seen from the gure 9) and applying a + b + > 3 3 ab . Se ond, sin e a 6 2 , b 6 2 , and 6 2 , we have ab 6 8 . Now
this onstru tion 500 times, we get the tiling for n = 2003 . p p
ab ab )2
6 pab =
3 3
a + b + 3 3 ab
(
3
6 64
3
=
4
3
:
2 4k 10-5. If the length of the strip is 2( k + 1) , then at his rst move, the rst player beats
his opponent's pie e that stands at the end of the strip. This divides the pie es
into two se tions: one of length 2 k and the other of length 1 . Sin e the se ond
se tion annot hange, the situation is equivalent to playing the game on a strip
of length 2 k where the se ond player makes the rst move. So the rst player an

2 4k - follow the winning strategy.
k
If the length of the strip is 2( +2) , then at his rst move, the rst player beats his
opponent's pie e that stands on the third square from the end with his pie e that
stands on the fourth square from the end. After that the board again ontains two
se tions: one of length 2 k and the other of length 3 . If the se ond player makes
2 4k a move in the rst se tion, the rst player responds a ording to the winning
strategy. If the se ond player makes a move in the se ond se tion, the rst player
also makes his move in the se ond se tion whi h thereafter has only one pie e left.
Figure 9 Figure 10
11-1. Let t1 and t2 be the time instants when the Trabant and the Mer edes passed
10-1. Answer: 4. Juhan, respe tively. The Mer edes was at t1 some distan e d away from Juhan
21 22
and by t2 it rea hed him. The Opel also was at t1 the distan e d away from length 4 m+2 whi h ontains 2 m+1 bla k squares and 2 m+1 white squares,
Juhan (on the other side) and at t2 its distan e to Juhan was the same as the the board altogether has an odd number of bla k squares and an odd number of
distan e the Trabant overed between t1 and t2 . This means that the velo ity of white squares. Ea h tile, not depending of its position, always overs 3 squares of
the Mer edes equals the sum of velo ities of the Opel and the Trabant. If t3 is the one olor and 1 square of other olor, hen e the total number of tiles must be odd.
time instant when the Opel rea hes Juhan, having overed the distan e d sin e On the other hand, the board has (4 m + 3)2 1 = 16 m2 + 24m + 8 squares. This
t1 , the Mer edes has overed the distan e d , passing Juhan at t 2 in the opposite number is divisible by 8, so the number of tiles must be even, a ontradi tion.
dire tion, plus the distan e that the Trabant has overed between t1 and t3 .
p
11-2. Answer: equality o urs when a=b= = 3.
2 (4m 4)
Using twi e the inequality between arithmeti and geometri means, we get
s v s
u
p3 p up
ab +
1
a
+
1
b
+
1

> 3 ab + 3 3 1
ab
> 2t 3 ab 3 3 1
ab
=
p
2 4m -
= 2 3 :
s
p 1
The equality holds if and only if a = b = and 3 ab = 3 3 . These onditions
ab
s
p 1 3 p 2 (4m 4)
give
3
a3 = 3 3 or a= . Hen e a=b= = 3.
a3 a
11-3. a) Let AA1 , BB1 , and CC1 interse t at P. From the property of interse ting Figure 11 Figure 12
hords we have jP AjjP A1 j = jP B j jP B1 j and also jP C j jP C1j = jP B j jP B1 j .
n
So jP Aj jP A1j = jP C j jP C1j from whi h it follows that the points C , A1 , C1 , 12-1. Answer: both Jüri and Mari have 4 possibilities.
and A are on y li . Let us pla e n ards on the n squares of main diagonal. This an be done in 4
n
b) We know that AB1 A1 B , BC1 B1 C , and CA1 C1 A are y li quadrilaterals.
ways, sin e ea h ard an be in one of 4 positions. After that, the position of
Therefore 6 AB1 B = 6 AC1 C be ause 6 CB1 B and 6 CC1 B are equal. Now
other ards is uniquely determined for both Jüri and Mari: if we know the olors
6 AA1 B = 6 AB1 B = 6 AC1C = 6 AA1 C . Hen e 6 AA1 B = 90Æ . Analogously of two neighbouring edges of a square, the ard an be pla ed on that square only
6 BB1 C = 90Æ and 6 CC1A = 90Æ . in one way. So we an ll the whole board diagonal by diagonal.
11-4. The numbers m + m and m + 2m all have the desired property where m is
2 2 12-2. Answer: x=4 and x = 16 .
p p
an arbitrary positive integer. Indeed, the numbers m2 + m and m2 + 2m p
First note that 4 and 16 satisfy the equation. To show that there are no more
are not integers, sin e m < m + m < m + 2m < (m + 1) . In addition, solutions, onsider the fun tion f (x) = x log2 x . Finding the derivative
p
2 2 2 2
p p

x ln 2 2
m2 + m m2 + 2m = m whi h is a divisor of both m2 + m and m2 +2m . f 0 (x) =
2 2
, we see that f (x) has only one extremum point x =
= .
2x ln 2 ln 2
11-5. Answer: all even numbers. Be ause f (x) is ontinuously dierentiable in its domain (0; 1) , it must have a
lo al extremum between ea h two zero points. Therefore f (x) annot have more
Figure 11 gives the tiling for n = 2 , gure 12 shows how to extend the tiling for
n = 2(m + 1) (we use the fa t that 2 4 re tangle an be
than 2 zero points.
n = 2m to the ase
overed with two tiles). 12-3. Answer: 45 .
Æ
Let now n = 2m + 1 . Let us olor the rows alternatingly bla k and white. Sin e Triangles EDC and ABC are similar with the similarity ratio
there are an even number of olumns of full length 4 m+3 and one olumn of k = jDC j=jBC j . From the given fa ts we get jDC j=jBC j = jBC j=jAC j .
23 24
Now jEC j = jAC j k = jAC j jBC j=jAC j = jBC j . Triangle BCE is right-angled of remaining six ti kets, say on C and D. Now the drawn set an ontain 1 , 12
and isos eles, so 6 CEB = 45Æ . and one number from ea h of 4 remaining ti kets (not A , B , C , D ), and again,
none of the 8 ti kets wins.
12-4. a) For a prime p , the sum of re ipro als of its divisors is (p + 1)=p . If there exists
another positive integer a 6= p whose sum of re ipro als of divisors is (p + 1)=p ,
then p a 's divisors. Then a is divisible at least by 1 , p , and a ,
must be one of IMO Team Sele tion Test: May 2003
and the sum of their re ipro als is greater than 1 + 1 =p .
b) Suppose that there exist an integer a with divisors d1 , . . . , dk and a se ond First Day
integer b with divisors e1 , . . . , el su h that 1=d1 + : : : + 1=dk = 1=e1 + : : : + 1=el .
Let p be a prime that is not a divisor of neither a nor b . Then ap has divisors d1 , 1. Two treasure-hunters found a treasure ontaining oins of value
. . . , dk , d1 p , . . . , dk p , and bp
has divisors e1 , . . . , el , e1 p , . . . , el p . Computing a1 < a2 < : : : < a2003 (the quantity of oins of ea h value is unlimited).
their re ipro al sum, we get for ap
bp , respe tively:
and The rst treasure-hunter forms all the possible sets of dierent oins ontaining
odd number of elements, and takes the most valuable oin of ea h su h set. The
1 1 1 1 1 1 1
+ :::+ + + :::+ = 1+ + :::+ se ond treasure-hunter forms all the possible sets of dierent oins ontaining
d1 dk d1p dk p p d1 dk even number of elements, and takes the most valuable oin of ea h su h set.

1 1 1 1 1 1 1 Whi h one of them is going to have more money and how mu h more? (H.
+ :::+ + + :::+ = 1+ + :::+
e1 el e1 p el p p e1 el Nestra )
The results are equal. So if there exist non-lonely integers a and b , then all 2. Let n be a positive integer. Prove that if the number 99 : : : 9}
| {z is divisible by n,
integers ap and bp where p is a prime not dividing a and b are also non-lonely. n
Now we an take a = 6 and b = 28 (in general, any two perfe t numbers), then then the number 11 : : : 1}
| {z is also divisible by n. (H. Nestra )
= = = = = =
1 + 1 2 + 1 3 + 1 6 = 2 and 1 + 1 2 + 1 4 + 1 7 + 1 14 + 1 28 = 2 .= = n
3.
0
Let N be the set of all non-negative integers and for ea h n 2 N denote n = n +1 .
12-5. a) Mark the numbers on 9 ti kets as follows:
The fun tion A : N ! N is dened as follows:
3
(1; 2; 3; 4; 5; 6) (10; 11; 12; 13; 14; 15) (19; 20; 21; 22; 23; 24)
0
(i) A(0; m; n) = m for all m; n 2 N ;
; ; ; ; ; ; ; ; ; ; ; ; ; ; ; (
(1 2 3 7 8 9) (10 11 12 16 17 18) (25 26 27 28 29 30) n; if k = 0,
; ; ; ; ;
(4 5 6 7 8 9) ; ; ; ; ;
(13 14 15 16 17 18) ; ; ; ; ;
(31 32 33 34 35 36) 0
(ii) A(k ; 0; n) = 0; if k = 1, for all k; n 2 N ;
1; if k > 1
If none of the ti kets in the rst olumn wins, two of the six drawn numbers must
A(k0 ; m0; n) = A(k; A(k0 ; m; n); n) for all k; m; n 2 N .
belong to f1; 2; : : :; 9g . If none of the ti kets in the se ond olumn wins, two of
(iii)
the six drawn numbers must belong to f10; 11; : : :; 18g . If none of the ti kets in Compute A(5; 3; 2) . (H. Nestra )
the last olumn wins, three of the six drawn numbers must belong to f19; : : :; 36g .
Sin e only 6 numbers are drawn, all these onditions annot be satised at the
same time. Se ond Day
b) If some number is marked on 3 ti kets form 8, then the drawn set of numbers
4. A de k onsists of 2
n ards. The de k is shued using the following operation: if
an ontain this number and one number from ea h of 5 remaining ti kets, so none
the ards are initially in the order
of 8 ti kets wins. Now suppose that ea h number is marked on at most 2 ti kets.
All 8 ti kets together have 48 numbers marked but sin e there are only 36 dierent
a1 ; a2 ; a3 ; a4 ; : : :; a2n 1 ; a2n ;
numbers, at least 12 numbers must o ur twi e. Without loss of generality, let
they be 1 , 2 , . . . , 12 . Let us take two ti kets, say A and B, whi h both have then after shuing the order be omes
1 marked. They ontain 10 more numbers, so one of the numbers 2 , . . . , 12 is
not marked on either of them. Let this number be 12 . Then 12 is marked on two a2n 1 +1 ; a1 ; a2n 1 +2 ; a2 ; : : :; a2n ; a2n 1 :
25 26
Find the smallest number of su h operations after whi h the original order of the b 1 mod p and 1 + b + : : : + bp 1
1+1+
| {z
: : : + 1} p
0 mod . Hen e,
ards is restored. (R. Palm ) p
bp
( m = (b 1)m (1 + b + : : : + bp
1)
1
) is divisible by ( b 1)mp = (b 1) n
b 1)n j bn 1 as required.
1 1 1
5. Let a; b; be positive real numbers satisfying the ondition + + = 1. and thus (
ab a b
Prove the inequality 3. Answer: 65536 .
p k k>1 n
p a p b p
Solution. First we show by indu tion on that for all integers and any
+ + 6 323 : the equality A(k; 1; n) = n holds. Indeed, if k = 2 , we have
1+ 2a 1+ 2 b 1+ 2
A(2; 1; n) = A(10 ; 00; n) = A(1; A(10; 0; n); n) =
A(1; 0; n) = A(00; 0; n) = n :
When does the equality hold? (L. Parts )
=
6. Let ABC be an a ute-angled triangle, O its ir um enter and H its ortho enter.
The orthogonal proje tion of the vertex A to the line BC lies on the perpendi ular Assuming now that A(k; 1; n) = n holds for some k > 1. Then
AC .
jCH j . (J. Willemson ) A(k0 ; 1; n) = A(k0 ; 00; n) = A(k; A(k0; 0; n); n) = A(k; 1; n) = n ;
bise tor of the segment Compute
jBOj
0
hen e we have the required equality for k = k + 1 and the indu tion is omplete.
In a similar way we an use indu tion on m to prove that for any natural numbers
Solutions of Sele tion Test m and n the equalities A(1; m; n) = m+n , A(2; m; n) = mn and A(3; m; n) = nm
hold.
1. Answer: The rst treasure-hunter gets one heapest oin more than the se ond Next we use indu tion on k to show that for any integer k > 0 the equality
one. A(k; 2; 2) = 4 holds. Indeed: for k = 1 we have A(1; 2; 2) = 2 + 2 = 4 . If
Solution. Take all the odd oin sets of the rst treasure-hunter and form an even A(k; 2; 2) = 4 for some k > 0 we get k0 > 1 and
A(k0 ; 2; 2) = A(k0 ; 10; 2) = A(k; A(k0 ; 1; 2); 2) = A(k; 2; 2) = 4 :
oin set from ea h one of them by the following rule. If the odd set ontains the
heapest oin then leave it out, otherwise add the heapest oin to the set. This
way we obtain exa tly all the sets of the se ond treasure-hunter plus one empty
oin set. It is lear that the des ribed operation does not hange the most valuable
We also note that for any natural k
oins of the sets ex ept for the set having originally only one heapest oin. Hen e,
A(k0 ; 3; 2) = A(k0 ; 20; 2) = A(k; A(k0 ; 2; 2); 2) = A(k; 4; 2) :
the rst treasure-hunter gets one heapest oin more than the se ond one.
n j |11 {z n j 99
Finally we ompute
2. The ondition of : : : 1} is equivalent to 9 : : : 9} .
| {z We will prove a more
n n A(5; 3; 2) = A(40 ; 3; 2) = A(4; 4; 2) = A(30; 30; 2) = A(3; A(30; 3; 2); 2) =
general result for any positional number system. Namely, we will show by indu - 4
tion on n that for any positive integers n and b the ondition n j b
n 1 implies = A(3; A(3; 4; 2); 2) = 22 = 65536 :
(b 1)n j b
n 1 . Taking b = 10 gives the desired result.
For n = 1 the laim is true as b
n 1 = (b 1)(bn 1 + : : : + b + 1) . Now assume 4. Answer: At least 2n shuings are needed.
that for all numbers less than n and any b the laim holds. Take any positive n
Solution. Assume the ards are labeled by numbers 1; 2; : : : ; 2 . We note that
integer b and assume n j b
n 1 . Consider two possible ases. n
the position of the ard number x after the rst shuing is f (x) = 2x mod 2 + 1 .
a) If n and b 1 are oprime, the onditions n j b
n 1 and b 1 j bn 1 imply After k th shuing its position be omes f k (x) = 2k x mod 2n + 1 . Our task is to
(b 1)n j b
n 1. nd the least number k su h that for ea h x the equality f k (x) = x holds, or
b) If n and b 1 have a ommon prime fa tor p , let n = mp ; then equivalently, 2
k 1 mod 2n + 1 .
bn 1 = bmp 1 = (bp )m 1 . As m < n , we an use the indu tion hy- Let k = 2n . Sin e
pothesis for m: bp )m 1 = bn 1 is divisible by n = mp and hen e
as (
also by m, bp 1)m j (bp)m 1 = bn 1 . Sin e p j b 1 we get
we have ( 2
2 n (2n + 1)2 n
2(2 + 1) + 1 1 mod 2n + 1 ;
27 28
then after 2n th shuing the order of the ards is restored. Assume that for some ongruent. Indeed, 6 AB 0 B = 90Æ = 6 HB 0 C , jBB 0 j = jB 0 C j and jAB 0j = jB 0H j
m < 2n the order is restored after m th shuing. In this ase we must have (sin e AB 0 H is also a right isos eles triangle). This ompletes the proof.
m > n , sin e the ard number 1 only rea hes the position 2n after n th shuf-
ing. Subtra ting now the ongruen e 2
m 1 mod 2n + 1 from the ongruen e
2
2 n 1 mod 2n + 1 , we obtain
m
2 (2
2n m 1) 0 mod 2n + 1 :
This is the ontradi tion as 2
m n
and the modulus 2 + 1 are oprime and the value
of 2
2n m 1 is less than the same modulus be ause of m > n. Hen e, k = 2n is
the least number of shuings restoring the original order of ards
p
5. Answer: the equality holds i a = b = = 3.
Solution. The given equality is equivalent to ab = a + b + . Sin e a; b; > 0 we

an nd su h 0 < ; ; < that a = tan , b = tan , = tan . Using the
2
tan x + tan y
equality tan( x + y) = repeatedly, we obtain tan( + + ) = 0
x tan y 1 tan
tan2 x
or + + = . Using now the equality = sin
2
x and taking into
1 + tan2 x
a ount that the angles , and are a ute, we an transform the required
inequality to the form
p
+ sin + sin 6
3 3
sin :
2
Sin e + + = and the sine fun tion is on ave in the segment [0 ; ℄ , we an
use Jensen's inequality to obtain
p

6 sin ( + + ) = sin
1 1 1 1 3
sin + sin + sin = ;
3 3 3 3 3 2
whi h implies the required
p
inequality. We also see that the equality holds i
== = , or a=b= = 3.
3
p
6. Answer: 2.
0 0
Solution. Let A and B be the base points of the heights drawn from the verti es
A and B , respe tively. From the onditions of the problem we on lude that
triangle AA0 C is an isos eles right triangle with the right angle at the vertex
A0 . 6 A0 AC = 6 A0 CA = 45Æ . Hen e the triangle BB 0 C is also right and
Thus,
Æ Æ
isos eles. We have 6 AOB = 26 ACB = 2 45 = 90 and the triangle AOB is
jCH j = p2 it is enough to
right and isos eles as well. In order to prove that
jBOj
prove that jAB j = jCH j . We will show that the triangles ABB 0 and HCB 0 are
29 30
Estonian Math Competitions
2003/2004
The Gifted and Talented Development Centre
Tartu 2004
second day, the first Liarian answered that he lied on exactly 2 days a year, the second,
that he lied on at least 2 days a year, and the third, that he lied on at most 2 days a year,
Selected Problems from Open Contests and so on, until the last, 2004th day of the year, when the first answered that he lied
on exactly 2004 days, the second said that he lied on at least 2004 days, and the third
replied that he lied on at most 2004 days a year. How many days a year does each of the
1. Diameter AB is drawn to a circle with radius 1. Two straight lines s and t touch the three Liarians actually lie? (Juniors.)
circle at points A and B , respectively. Points P and Q are chosen on the lines s and t,
Answer: the first Liarian lies on days, the second on days and the third on
respectively, so that the line P Q touches the circle. Find the smallest possible area of the
2003 1002 0
days a year.
quadrangle AP QB . (Juniors.)
Solution. Since the answers of the first Liarian are all mutually exclusive, he could tell
Answer: 2. the truth on at most one day. If he had lied on all days, the answer he gave on the 2004th
Solution. As ABCD is a trapezium, its area can be found as follows: day would have been true — a contradiction. Thus, the first Liarian tells the truth on
one day (the 2003th day of the year) and lies on the remaining 2003 days of the year.
SAP QB =
jAP j jBQj jAB j jAP j jBQj :
+
= + Suppose that the second Liarian lies on n days of a year, then his answers are true on the
2
first n days and false on the remaining 2004 n days. Hence n = 2004 n, and n = 1002
On the other hand, jAP j = jP K j and jBQj = jQK j, thus SAP QB = jP K j jQK j jP Qj.
+ = (the second Liarian lies on the last 1002 days of the year).
The smallest possible length of the segment P Q is 2. Suppose now that the third Liarian lies on n days of a year, where n 1. Then the
answers he gave on the first n 1 days are false and the remaining answers (starting
2. Find all pairs of positive integers (m; n), m n, such that a rectangle of size m n from the nth day) are true. Thus he lies on n 1 days a year — a contradiction. However,
can not be divided into rectangles of sizes 2 5 and 1 3. (Juniors.) if he lies on 0 days a year, all his answers are true and there is no contradiction. Hence
Answer: (2; 2), (2; 4), (4; 4), (2; 7) and (1; 3k ,
1) (1 3 ; k 2) , where k is any positive integer. the third Liarian lies on 0 days a year.
Solution. The answer is obvious for rectangles with one side of length 1. Hence, we need 4. Circles 1 and 2 with centres O1 and O2 , respectively, intersect at points A and B
to consider the rectangles with both of the sides longer than 1. so that the centre of each circle lies outside the other circle. Line O1 A intersects circle 2
All the rectangles 3 m can be divided into 1 3 pieces. All the rectangles 5 (2k ) can be again at point P2 and line O2 A intersects circle 1 again at point P1 . Prove that the points
divided into 2 5 pieces. All the rectangles 5 (2k + 1) can be divided into k 1 pieces O1 , O2 , P1 , P2 and B are concyclic. (Juniors.)
of 2 5 and 5 pieces of 1 3. Thus, all the rectangles with one side having the form
3s + 5t, where s; t 0, can be divided in the required way. Since 6 = 3 2, 8 = 3 1 + 5 1,
Solution 1. Triangles AO1 P1 and AO2 P2 are isosceles and have equal base angles, thus
9 = 3 3 and 10 = 5 2, every positive integer n > 7 can be represented as 8 + 3k , 9 + 3k
also their vertex angles are equal. Now
or 10 + 3k for some integer k 0. Hence we are done, if one of the sides of the original \P O P
1 1 2 = \P O A \AO P
1 1 = 2 2 = \P O P 1 2 2
rectangle is 3, 5, 6 or n > 7 and the length of the other side is at least 2.
(See figures 2 and 3 for the possible positions of points P1 and P2 , according to the angle
We still need to consider the rectangles with both sides from \O P O 2 = \O P O
being smaller or larger than a right angle). Also,
the set f2; 4; 7g. The rectangles 2 2, 2 4 and 4 4 cannot
1 1 1 2 2
be divided, as they cannot contain a 2 5 piece, but their \P BP \P BA \ABP \P O A \AO P \P O P :
1 1
1 2 = 1 + 2 = 1 1 + 2 2 = 1 2 2
area is not divisible by 3. A similar argument holds for 2 7 2 2
rectangle — cutting away either one or zero 25 pieces leaves Hence the angles \P1 O1 P2 , \P1 O2 P2 and \P1 BP2 lying on segment P1 P2 are equal, and
a part with area not divisible by 3. the points O1 , O2 , P1 , P2 and B are concyclic.
Figure 1 It is easy to construct a required division for 4 7 rectangle
P1 P2
(see Figure 1), hence 7 7 rectangle can be divided as well. A
1 A 2 1 2
3. On Liarians planet, a year consists of 2004 days and for each inhabitant on the P2 P1
planet, each day of the year is either a truth-day or a lie-day: on truth-days a Liarian
speaks only the truth and on lie-days the Liarian always lies (the number of a Liarian’s O1 O2 O1 O2
truth-days or lie-days can also be 0). On each day of a given year, three Liarians were
B
asked, “How many days a year do you lie?” On the first day of the year, the first Liarian B
answered that he lied on exactly 1 day a year, the second Liarian said that he lied on
Figure 2 Figure 3
at least 1 day a year, and the third replied that he lied on at most 1 day a year. On the
1 2
Solution 2. Since (1) ABCD is not cyclic;
(2) the sides AB , BC , CD and DA have pairwise different lengths;
\O P A \O AP
1 1 = 1 1 = \O AP2 2 = \O P A ;2 2
(3) the circumradii of the triangles ABC , ADC , BAD and BCD are equal?
points O1 , P1 , P2 and O2 lie on the same circle. On the other hand, since
b) Does there exist such a non-convex quadrangle? (Seniors.)
\O P O \O BO \O AP \O AO Æ
1 1 2 + 1 2 = 1 1 + 1 2 = 180
Answer: a) no; b) yes.
(for the case shown on figure 2) or Solution. a) Assume the quadrangle ABCD satisfies the conditions of the problem. Let
\O P O \O BO \ Æ \O P A \O AO R be the common circumradius of triangles ABC , ADC , BAD and BCD. The Sine law
2 + = 180 + 2 =
for triangles ABC and BAD gives
1 1 1 2 1 1 1
= \ 180
Æ \O AO \O AO Æ
1 2 + 1 2 = 180
jAB j R=
jAB j
(for the case shown on figure 3), also points O1 , P1 , O2 and B lie on the same circle.
These two circles coincide, since both of them are the circumcircle of triangle O1 P1 O2 .
sin \ACB = 2
sin \ADB ;
or sin \ACB = sin \ADB . If \ACB = \ADB , the quadrangle ABCD would be cyclic,
Solution 3. We note that jO1 B j = jO1 P1 j, jO2 B j = jO2 P2 j, \AO1 O2 = \BO1 O2 ja \AO2 O1 =
hence \ACB + \ADB = (see Figure 4). Similarly we find \DAC + \DBC = , but
\BO2O1. Letting R0 , R1 and R2 be the radii of the circumcircles of triangles O1O2B , then the sum of interior angles of ABCD would be greater than 2 , a contradiction.
O1 O2 P1 and O1 O2 P2 , and applying the Sine law, we get
b) Let ABD be an isosceles triangle with jAB j = a and jBD j = jDAj > 2a. Let C be
R1
jO P j
1 1 jO P j
1 1 jO B j
1
R0 = the orthocenter of the triangle ABD and let B1 and D1 be the feet of the perpendiculars
2 =
sin \P O O 1 2 1
=
sin \AO O 2 1
=
sin \BO O 2 1
= 2
drawn from the vertices B and D to the opposing sides (see Figure 5).
jO B j
2 jO P j
2 2 jO P j
2 2
R2 : Denote \ADC = . As the triangles ADD1 and ABB1 are right with a common acute
=
sin \BO O 1 2
=
sin \AO O 1 2
=
sin \P O O 2 1 2
= 2
angle, they are similar and \ABC = \ABB1 = \ADD1 = \ADC = . Due to symme-
try we also have \BAC = \BDC = . As the triangles ADC and BDC are congruent,
Hence triangles O1 O2 B , O1 O2 P1 and O1 O2 P2 share a common side O1 O2 and have equal
their circumradii are equal — let it be R. The Sine law for triangles ABC and BDC gives
radii of circumcircles.
If \O1 P1 O2 is larger than a right angle, then all the midperpendiculars of the triangles jCB j jCB j jCB j
intersect at the same side of line O1 O2 as point B . If \O1 P1 O2 is smaller than a right RABC = R;
2 sin \CAB =
2 sin
=
2 sin \CDB =
angle, then all the centres of the circumcircles lie on the other side of line O1 O2 from
point B . Thus, in both cases the triangles share a common circumcircle. and the Sine law for triangles BAD and ABC gives
Remark. The claim of the problem is actually true without assuming that the centres of
RBAD =
jAB j jAB j jAB j jAB j R:
the circles lie outside the other circle. 2 sin \ADB =
2 sin 2
=
2 sin( )
2
=
2 sin \ACB =
5. Let R + be the set of all positive real numbers. Find all functions f : R + !R +
, such D
B
that for every x; y 2 R + the equality

x C
y 2 f (x) = f C
y D0
D
holds. (Seniors.)
a B0
Answer: All functions of the form f (x) = 2 , where a 2 R + . A
x B A
1
Solution. Taking x = 1 y 2 f (1) = f
, we obtain . Substitution z =
1
gives f (z ) = Figure 4 Figure 5
y y
f (1) Thus we have shown that the circumradii of the triangles ABC , ADC , BAD and BCD
. Substituting this to the original equation we see that f (1) can be any positive real
z2 are equal. We still have to prove that the lengths of the sides of ABCD are all different.
number.
Indeed, as AB is the hypotenuse of ABB1 , we have jAB j > jBB1 j > jBC j. Similarly,
6. a) Does there exist a convex quadrangle ABCD satisfying the following conditions
3 4
in triangle ADD we have jAD j > jDD j > jCD j.
1 1 Since jBDj > a, jBC j < a and
2 hence y < . On the other hand, if ; x; y are real numbers satisfying x; y 0 and y < ,
jBDj < jBC j jCDj, we get
+ there exists a triangle with sides + x + y , + x and (since all the triangle inequalities
are satisfied). We will show that numbers less than 63 can not be represented in two
jCDj > jBDj jBC j > a a a jAB j : 2 = =
different ways in the form s( +x+y; +x; ), where x; y; are integers and x; y 0,
y < (disregarding the order of the numbers). Note that when x increases, the value
Thus jDAj > jCD j > jAB j > jBC j as required. As a non-convex quadrangle can not be
of the expression s( +x+y; +x; ) increases as well. Taking this into account, we will
cyclic, all the conditions for the quadrangle ABCD are satisfied.
consider all the possibilities for .
7. Find all positive integers n such that the number (1) If = 1, we have y = 0 and for x = 0; 1; 2; 3, we have 3; 15; 35; 63 as the values for
the expression s( +x+y; +x; ), respectively.
n1 n2 nn 1
nn (2) Let = 2. If y = 0, then for x = 0; 1 the expression s( +x+y; +x; ) takes the values
+ + :::+ +
1! 2! (n 1)! n! 48 and 128, respectively. If y = 1, then for x = 0 we have s( +x+y; +x; ) = 63.
is an integer. (Seniors.) (3) Let 3 . As x; y
< , we obtain s( +x+y; +x; ) 3 1 = 3 3
0 and y
81 > 63.
Answer: n = 1; 2; 3.
Hence we have covered all the cases when s( +x+y; +x; ) can be less than 63. We have
Solution. Suitability of the solutions 1, 2 and 3 can be verified directly. We will prove also seen that
that there are no other solutions.
Let n 2, then the given expression can be written as: s(4; 4; 1) = s(3; 2; 2) = 63
p p
n1 n2 nn 2 nn 1 nn 63 3 7
+ + ::: + + + = and the respective area is = .
1! 2! (n 2)! (n 1)! n! 4 4
n (n 1)! n2 3 4 : : : (n 1)
= + + ::: + 9. Find all functions f that are defined on the set of positive real numbers, have real
(n (n
1)! 1)!
values and satisfy for all positive real numbers x and y the equation
nn 2 (n 1) nn 1 nn 1
+ + + =
(n (n (n
1 1
1)! 1)! 1)!
f (x)f (y ) = f (xy ) + + :
n (nn2 3 4 : : : (n 1) + : : : + nn
1)! +
2
n
( 1) + 2 nn :
1 x y
=
(n 1)! (Seniors.)
As the first n 1 terms in the numerator and the denominator are divisible by n 1, the
Answer: f (x) = 1 + .
1
last summand 2 nn 1 has to be divisible by n 1 as well. As the numbers n 1 and x
nn 1 are coprime, n 1 must divide 2, hence n = 2 or n = 3. Solution. Taking y = 1 , we get for each x
8. Find the smallest real number x for which there exist two non-congruent triangles f (x)f (1) = f (x) +
1
+ 1 : (1)
with integral side lengths having area x. (Seniors.) x
p Taking also x = 1, we obtain a quadratic equation f (1)2 f (1) 2 = 0, that has solutions
f (1) = 1 and f (1) = 2. Thus f (1) 1 6= 0 and from the equation (1) we get
3 7
Answer: .
4
Solution. Denote
f (x) =
1
f (1)
1 +
1
x
:
s(a; b; ) = (a + b + )(b + a)( + a b)(a + b ) : 1

a triangle with sides a, b and can be found from the Heron’s formula and
The area ofp Now either f (x) =
1
2
1+
1
x
1
or f (x) = 1 + . The second solution satisfies the initial
x
s(a; b; ) 1
is equal to . W.L.O.G. we may assume that b = + x and a = b + y = + x + y, condition, whereas the first one does not for e.g. x = y = .

4 2
where x; y 0 . Then
10. There are N lightbulbs on a circle, labelled clockwise with numbers 1 to N . Ini-
s(a; b; ) = s( + x + y; + x; ) = tially, none of the lightbulbs are lit. Then the following operation is performed for each
= (3 + 2x + y )( y )( + y )( + 2x + y ) ; positive divisor d of the number N (1 and N included): starting from bulb number 1
5 6
and moving clockwise, the state of each d-th bulb is changed i.e. it is lit, if it is off, and Selected Problems from the Final Round of National
switched off, if it is lit, and this is repeated exactly N times. (E.g. for N = 6 and d = 3,
the bulbs are lit or switched off in the following order: 3; 6; 3; 6; 3; 6.) Olympiad
For which values of N will all bulbs be lit, after the procedure has been completed for
all divisors of N ? (Seniors.)
1. Find all pairs of real numbers (x; y ) that satisfy the equation
Answer: N = 2
k
, where k is any non-negative integer.
x+6 13 4 y
Solution. We note that for a given divisor d of the number N , all states of bulbs having + = :
N y xy x
number divisible by d are changed, and only these. Since there are such bulbs and
d (Grade 9.)
exactly N changes are made, the state of each bulb is changed exactly d times. In partic-
ular, for d = 1 the state of each of the N bulbs is changed once. Answer: x= 3 ,y = 2 .
To have all bulbs lit by the end of the procedure, the state of each bulb must be changed Solution. Multiply the equation by xy to obtain
an odd number of times. As we showed, the state of bulb m is changed once for divisor
1, and additionally, for each such divisor d > 1, that is also a divisor of m, the state ( x + 6) x + 13 = (4 y ) y
is changed d times. Obviously, the condition is satisfied when all divisors d > 1 of N
are even, i.e. N = 2k , where k 0. Suppose now that the number N has odd divisors or equivalently
greater than 1, and let d be the smallest of them. Then the state of bulb number d is x2 + 6x + y 2 4 y + 13 = 0 :
changed once for divisor 1, d times for divisor d, and 0 times for all other divisors of N ,
since d is not divisible by them. Hence the state of bulb d is changed an even number of Notice that the left side can be expressed as sum of squares:
times and the bulb is not lit by the end of the procedure.
( x + 3)2 + (y 2)
2
= 0 ;
11. On the circumcircle of triangle ABC , point P is chosen, such that the perpendic-
ular drawn from point P to line AC intersects the circle again at a point Q, the perpen- giving x + 3 = 0 and y 2 = 0 , or x = 3 and y = 2 .
dicular drawn from point Q to line AB intersects the circle again at a point R and the
perpendicular drawn from point R to line BC intersects the circle again at the initial 2. The positive differences ai aj of five different positive integers a1 , a2 , a3 , a4 , a5 are
point P . Let O be the centre of this circle. Prove that \P OC = 90Æ . (Seniors.) all different (there are altogether 10 such differences). Find the least possible value of
the largest number among the ai . (Grade 9.)
Solution 1. Rotate triangle P QR clockwise 90Æ and denote
the triangle obtained P 0 Q0 R0 . Since P Q ? AC , QR ? AB A Answer: 12 .
q
P
and RP ? BC (see figure 6), P 0 Q0 k AC , Q0 R0 k AB and Solution 1. Say a1 < a2 < a3 < a4 < a5 . Since the 10 positive differences ai aj are all
R0 P 0 k BC . Hence triangle Q0 R0 P 0 is similar to triangle different, the largest a5 a1 must be greater than or equal to 10, giving a5 . In the
11
ABC . Since the triangles ABC and Q0 R0 P 0 share a com- case a5 = 11 the differences must be exactly 1; 2; : : : ; 10, hence
mon circumcircle, they are actually congruent and either Q O : : : + 10 =
coincide or are rotated 180Æ from the midpoint of the cir-
55 = 1 + 2 +
B C a a (a5 a3 ) + (a5 a2 ) + (a5 a1 ) +
cle. In the first case, C = P 0 , and in the second case, points
= ( 5 4) +
R (a4 a3 ) + (a4 a2 ) + (a4 a1 ) +
C and P 0 are endpoints of the same diameter of the tri-
+
(a3 a2 ) + (a3 a1 ) +
angle. Rotating triangle P 0 Q0 R0 back to triangle P QR, we
+
a a1 ) =
obtain in both cases that \P OC = 90Æ .
_ _
Figure 6 + ( 2
= a
4 5 + 2 4 a a
2 2 a :
4 1
P C AQ Æ . Analo-
_ _ _ P Q_
+
Solution 2. Since chords AC and intersect at a right angle, This is impossible, because 4a5 + 2a4 2a2 4a1 is even.
_ AQ
_ RB
_
= 90
2
We have obtained that the largest of the numbers must be larger than or equal to 12. For
AQ + RB Æ and RB + P C Æ . These equations give P C instance, we may choose the numbers 1; 3; 8; 11; 12 (the differences ai aj are therefore
gously, = 90 = 90 = = =
90
Æ. 2 2 1 = 12 , 1, 3 = 11
11 2 = 3 8, 4 = 12 8, 5 = 8 3, 7 = 8 1, 8 = 11 3, 9 = 12 3,
10 = 11 and 11 = 12 1).
1
Solution 2. Construct the example for the largest number 12 similarly to the previous
solution and notice that if the largest of numbers ai equals 11, the differences in question
7 8
equal 1; 2; : : : ; 10. Therefore exactly 5 of these differences are odd. Denote by x and y Therefore x0 y 0 z 0 = 20 and, since x0 , y 0 and z 0 are pairwise coprime, they are equal to (in
the number of odd, resp. even, numbers among the ai . The difference of two integers some order) 1; 1; 20 or 1; 4; 5. Requiring x < y < z for the numbers x = 24x0 , y = 30y 0
is odd iff the numbers are of different parity. There are xy possibilities to choose a pair and z = 40z 0 , the first case implies x0 = 1, y 0 = 1 and z 0 = 20, while the second case gives
of integers of different parity, hence there are xy such odd differences. So x and y must two possibilities: x0 = 1, y 0 = 4, z 0 = 5, or x0 = 1, y 0 = 5, z 0 = 4. The triples (x; y; z )
satisfy the system of equations are (24; 30; 800), (24; 120; 200) and (24; 150; 160), respectively. The first and the third
triple satisfy the conditions of the problem, but the second triple does not, because the
x+y =5 greatest common divisor of y = 120 and z = 200 is 40 rather than 10.
;
xy = 5 Solution 2. Let p . a denote the exponent of prime p in the canonical form of number a.
which has no positive integral solutions. First, consider the exponents of 2: the conditions of the problem imply
min(2 . x; 2 . y ) 2 . 6 = 1 ;
3. Three different circles of equal radii intersect in point Q. The circle C touches all of = (2)
them. Prove that Q is the center of C . (Grade 9.) min(2 . x; 2 . z ) = 2 .8=3; (3)
q
. x; 2 . y; 2 . z ) . 2400 = 5 : (4)
Solution. Consider an arbitrary point P inside the circle C
max(2 = 2
C
qq
different from the center point of C . Fix r > 0. We show that The equation (3) gives us 2 . z 3 and 2 . x 3. The equation (2) now implies that

there exist at most two circles of radius r that touch C from 2. y = 1, and from (3) and (4) we obtain that one of the exponents 2 . x and 2 . z equals
inside and pass through P . For this roll the circle of radius 3 and the other is 5. Altogether we now have two possibilities:
r inside C ; evidently the circle being rolled passes through ( (2 . x = 5
r 2 . x = 3
point P in at most two positions (or equivalently: any fixed P
circle that touches C from inside has at most two common 2.y =1 ; .y =1
2 : (5)
points with such a circle that passes through P and has the 2.z =5 .z =3
2
centre point in the centre point of C — see figure 7). Hence no Consider now the exponents of 3:
such point P can be the intersection point of three different Figure 7
circles of different radii, all touching the circle C . min(3 . x; 3 . y ) = 3 . 6 = 1 ; (6)
min(3 . x; 3 . z ) = 3 . 8 = 0 ; (7)
4. Find all triples of positive integers (x; y; z ) satisfying x < y < z , g d(x; y ) = 6,
g d(y; z ) = 10, g d(z; x) = 8 and l m(x; y; z ) = 2400. (Grade 10.)
max(3 . x; 3 . y; 3 . z ) = 3 . 2400 = 1 : (8)
Answer: (24; 30; 800) and (24; 150; 160). The equations (6) and (8) imply that 3 .x = 3 .y = 1 , and the equation (7) gives that
3 . z = 0. Therefore
Solution 1. As 6 and 8 both divide x, l m(6; 8) = 24 divides x. Similarly l m(6; 10) = 30 (
divides y and l m(10; 8) = 40 divides z . Hence there exist positive integers x0 , y 0 , z 0 , that 3 . x = 1
x = 24x0 , y = 30y 0, z = 40z 0 , and 3.y =1 : (9)
3.z =0
6 =
0 0 0 0
g d(x; y ) = g d(24x ; 30y ) = 6 g d(4x ; 5y ) ;
Finally, consider the exponents of 5:
g d(y; z ) = g d(30y ; 40z ) = 10 g d(3y ; 4z ) ;
10 =
0 0 0 0
8 = g d(z; x) = g d(40z 0 ; 24x0) = 8 g d(5z 0 ; 3x0 ) : min(5 . x; 5 . y ) = 5 . 6 = 0 ; (10)
. y; 5 . z ) . 10 = 1 ; (11)
These equations imply g d(4x0 ; 5y 0 ) = 1, g d(3y 0 ; 4z 0 ) = 1 and g d(5z 0 ; 3x0 ) = 1. Hence
min(5 = 5
x0 ; y 0; z 0 are pairwise coprime and g d(x0 ; 5) = 1, g d(y 0 ; 4) = 1 and g d(z 0 ; 3) = 1. max(5 . x; 5 . y; 5 . z ) = 5 . 2400 = 2 : (12)
We show now that l m(24x0 ; 30y 0 ; 40z 0 ) = l m(120x0 ; 120y 0 ; 120z 0 ). For this it suffices to The equation (11) gives us 5 .y 1 and 5 .z 1. The equation (10) now implies 5 .x = 0
prove that either side of the equation divides the other. Evidently lhs divides rhs. For and from (11) and (12)we obtain that one of the exponents 5 . y and 5 . z equals 1 and
the other direction, prove that 120x0 , 120y 0 , 120z 0 all divide lhs. Indeed, 24x0 and 5 divide the other is 2. Altogether there are again two possibilities:
lhs and, since g d(x0 ; 5) = 1, we have g d(24x0 ; 5) = 1 giving that 24x0 5 = 120x0 divides ( (5 . x = 0
lhs. Similarly prove that 120y 0 and 120z 0 divide lhs as well. Now 5 . x = 0
5.y =1 ; .y =2
5 : (13)
2400 =
0 0
l m(x; y; z ) = l m(24x ; 30y ; 40z ) =
0 5.z =2 .z =1
5
= l m(120 x0 ; 120y 0; 120z 0 ) = 120 l m(x0 ; y 0; z 0 ) = 120 x0 y 0z 0 :
9 10
Since no other prime divides l m(x; y; z ) = 2400, no other prime divides x, y , z either. (1) invert the order of the letters in the word;
The conditions (5), (9) ja (13) now give altogether four possibilities: (2) replace two consecutive letters: BA ! UU, AU ! BB, UB ! AA, UU ! BA, BB !
( ( AU or AA ! UB.
x = 23 31 50 = 24 x = 23 31 50 = 24
y = 21 31 51 = 30 ; y = 21 31 52 = 150 ;
( z = 2 5 3 1 5 0 = 800 ( z = 2 5 3 1 5 0 = 160
5 0 2 5 0 1 Given that BBAUABAUUABAUUUABAUUUUABB is a BAU word, does BAU have
x = 2 3 5 = 96 x = 2 3 5 = 96 a) the word BUABUABUABUABAUBAUBAUBAUB?
y = 21 31 51 = 30 ; y = 21 31 52 = 150 :
z = 23 30 52 = 200 z = 23 30 51 = 40 b) the word ABUABUABUABUAUBAUBAUBAUBA?
We may check that all the conditions of the problem, except x < y < z , are satisfied, (Grade 11.)
which leaves us the two former possibilities.
5. In the beginning, number 1 has been written to point (0; 0) and 0 has been written Solution 1. a) Define the value of a word w as the number b a where b and a are the
to any other point of integral coordinates. After every second, all numbers are replaced numbers of B s and As, respectively, in w . Note that any allowed operation preserves
with the sum of the numbers in four neighbouring points at the previous second. Find the value of word modulo 3. The value of the known BAU word is 7 8 = 1 and the
the sum of numbers in all points of integral coordinates after n seconds. (Grade 10.) value of the word under consideration is 9 8 = 1, so the latter cannot belong to BAU.
Answer: 4
n
. b) At first, note that the allowed operations enable to interchange any two consecutive
letters of a word. Indeed, if these letters are BA, AU or UB then replace them with UU,
Solution. At the initial moment, the sum of numbers in points of integral coordinates is BB or AA, respectively, then invert the order of letters, then replace the letters UU, BB
1 = 4 . After every second, any number x on the plane contributes to the sum of the
0
or AA back, and finally invert the order once more. If the letters to be interchanged are
next second as 4x, hence the total sum of numbers increases 4 times at every step, giving AB, UA or BU then perform all these steps in reverse order.
4 after n seconds.
n
Iteration of the interchanging operation enables to interchange any two letters (move
the second letter side by side with the first, interchange them, and finally move the first
6. Real numbers a, b and satisfy a2 + b2 + 2 = 1 and a3 + b3 + 3 . Find a + b + .
= 1
letter to the original place of the second). Note that the known BAU word contains 7
(Grade 10.)
letters B, 8 letters A and 10 letters U, the word under consideration has 8 letters B, 9
Answer: 1. letters A and 8 letters U. Thus we can replace some consecutive letters UU with BA and
Solution. As a2 + b2 + 2 = 1, we have 1 a; b; 1, implying a2 a3 , b2 b3 and then reorder the letters in the word, obtaining the required word.
2 3 . Since a2 + b2 + 2 = a3 + b3 + 3 , all three inequalities must have equality: a2 = a3 , Solution 2. a) Part b) of the previous solution implies that a word w belongs to BAU if
b2 = b3 and 2 = 3 . Therefore a; b; can only be equal to 0 or 1 and exactly two of them and only if there is a word w 0 in BAU containing any letter the same number of times as
must be 0 and the third one 1 because a2 + b2 + 2 = 1. All cases imply a + b + = 1. w. Suppose the right numbers of letters can be obtained from the known BAU word by a
sequence of the allowed operations. Let x be the difference of the number of performed
7. Find all functions f which are defined on all non-negative real numbers, take non- replacements BA ! UU and the number of the performed replacements UU ! BA;
negative real values only, and satisfy the condition analogously define y and z for the rules AU ! BB and UB ! AA, respectively. The
known word contains 7 letters B, 8 letters A ja 10 letters U, the word under consideration
x f (y ) + y f (x) = f (x) f (y ) (f (x) + f (y )) has 9 letters B, 8 letters A and 8 letters U. This leads to the system of equations
for all non-negative real numbers x, y . (Grade 11.) 8
p < 7 x + 2y z = 9
Answer: f (x) 0 and f (x) x. :
8 x y + 2z = 8
Solution. Taking x = y = 1, one gets 2f (1) = 2f (1)3 , so either f (1) = 1 or f (1) = 0. If
10 + 2x y z = 8
f (1) = 1 then, taking y = 1 in the
p original equality, one gets x + f (x) = f (x)(f (x) + 1) for (each equation describes the number of one letter). This system has no integral solu-
all x. From this we get f (x) x as one solution. If f (1) = 0 then the same substitution tions, so the answer must be “no”.
leads to f (x) 0. b) Here the word under consideration has 8 letters B, 9 letters A and 9 letters U. The
It remains to check that both functions satisfy the conditions of the problem. corresponding system of equations is
8
8. The alphabet of language BAU consists of letters B, A, and U. Independently of the < 7 x + 2y z = 8
choice of the BAU word of length n from which to start, one can construct all the BAU 8 x y + 2z = 9 :
words with length n using iteratively the following rules: : 10 + 2x y z = 8
11 12
r
One of the solutions of it is x = 0, y = z = 1. Hence one can construct the word under 2S
Analogously we find b = = , i.e. triangle ABC is equilateral.
consideration from the known word, performing both the replacements AU ! BB and k
UB ! AA once and then reordering the letters.
11. Let a, b, be positive real numbers such that a2 + b2 + 2 = 3 . Prove that
9. Inside a circle, point K is taken such that the ray drawn from K through the centre
O of the circle and the chord perpendicular to this ray passing through K divide the 1
ab
+
1
b
+
1
a
: 1
circle into three pieces with equal area. Let L be one of the endpoints of the chord 1 + 2 1 + 2 1+ 2
mentioned. Does the inequality \KOL < 75Æ hold? (Grade 12.) A (Grade 12.)
Answer: yes.
L K M
Solution 1. Applying the AM-GM inequality to each denominator, one obtains
Solution. Let AB be the diameter containing K (see figure 8).
Moving point K together with chord LM along this diame-
O
1+2
1
ab
+
1+2
1
b
+
1+2
1
a
1+
1
a2 + b2
+
1+
1
b2 + 2
+
1+
1
2 + a2
:
ter away from O , the angle KOL decreases, so the segment
bounded by chord LM and arc LAM decreases, too. Hence it Applying now the AM-HM inequality to the whole expression, one obtains
suffices to show that, if \KOL = 75Æ , the area of this segment is
B
more than one third of the area of the circle. Figure 8 1
a 2 + b2
+
1
b 2 + 2
+
1
2 + a2

Æ ; then the area of the sector r
1 + 1+ 1 +
r be the radius and \KOL
150
Let LOM is 2

= 75 = 3
360 3 =
5
r 2
and the area of triangle LOM is
(1 + a2 + b2 ) + (1 + b2 + 2 ) + (1 + 2 + a2 )
12 9 9
:
SLOM = 2 SLOK = jLK j jOK j = r2 sin \KOL os \KOL =
=
3 + 2( a 2 + b2 + 2 )
=
3 + 2 3
= 1
r2 sin 2\KOL = r2 :
1 1
=
2 4
Solution 2. Applying the AM-HM inequality, one obtains
Hence the area of the segment bounded by chord LM and arc LAM is 1
+
1
+
1
9
=
1 + 2 ab 1 + 2 b 1+ 2 a 3 + 2 ab + 2b + 2 a
5
r 2 1
r2 =
5 1

r2 : =
9
=
9
:
12 4 12 4 a2 + b2 + 2 + 2ab + 2b + 2 a ( a + b + )2
1 1 5 1 4 1
As > 3 implies < , we have > = . Consequently, the area of the The Jensen inequality for the square function establishes
4 12 12 4 12 3
segment under consideration is indeed greater than one third of the area of the circle.
a + b + )2
a b 2 + 2
2
( + 3
= = 1 :
10. Let K , L, M be the basepoints of the altitudes drawn from the vertices A, B , C 9 3 3
! ! ! !
of triangle ABC , respectively. Prove that AK + BL + CM = 0 if and only if ABC is Thus
equilateral. (Grade 12.)
! ! !
Solution. If triangle ABC is equilateral then vectors AK , BL and CM have equal lengths
1
ab
+
1
b
+
1
a
a b : 9
2
1
! ! ! !
1 + 2 1 + 2 1+ 2 ( + + )
and the sizes of the angles between them are equal to 120Æ . So AK + BL + CM = 0 .
! ! ! !
Assume now that AK +BL+CM = 0 . Let a, b, be the lengths of the sides BC , CA, AB , Solution 3. Multiplying both sides of the desired inequality by (1+2ab)(1+2b )(1+2 a),
1
! 2S , jBL
respectively, and let S be the area of triangle ABC . Then jAK j =
!j = 2S and one gets the equivalent inequality
8
a b
! S ! ! !
jCM j = . Rotating vectors AK , BL and CM counterclockwise by 90Æ, they become
2 1 + ab + b + a
a b :2 2 2
4
parallel to the corresponding sides of the triangle and, if we draw them one by one onto
the plane, every vector starting from the endpoint of the previous, we obtain a triangle Note that
similar to ABC because the sum of the vectors is zero. Let k be the similarity coefficient;
r a2 + b2 + 2 p
! ! ! S !j ka implying a S
3
ab ;3 2 2 2
then jAK j = ka, jBLj = kb and jCM j = k . Hence jAK
2 2 1 = =
= = = . 3 3
a k
13 14
giving ab 1; thus Solution 2. b) Note at first that the size of the angle between two arbitrary line segments
p
connecting two vertices is an integral multiple of , and so the angle can occur in case
a2 b2 2 = (ab )2 ab : n n
n
2
of even n only. So must be odd since and n are coprime, which gives (mod n).
2 2
So indeed one has n n n
Thus (i + ) = i + i + n), i.e. any diameter transforms to a diameter.
ab + b + a p p (mod
ab ab a2 b2 2 :
1 + 4 2 2 2
q q
2 2 2
= Now consider two cases.
4
If line segment Ai Aj is a diameter then j Ak Ak
n Aj
12. Let n and be coprime positive integers. For any integer i, denote by i0 the re-
i+ n) and Ai Aj ? Ak Al if and only Aj

(mod
if j k l j (mod n) (see figure 11).
2
mainder of division of product i by n. Let A0 A1 : : : An 1 be a regular n-gon. Prove Ai Am
that Then j k = (j k ) (l j ) = Ai
l j (mod n), i.e. j 0 k0 l0 j 0 (mod n) Al Al
a) if Ai Aj k Ak Al then Ai Aj k Ak Al ;
0 0 0 0 and hence A A ? A A .
i j 0
k0
l 0 0
b) if Ai Aj ? Ak Al then Ai Aj ? Ak Al .
0 0 0 0
If line segment Ai Aj is not a diameter then Figure 11 Figure 12
consider the vertex Am such that Ai Am is
(Grade 12.) a diameter (see figure 12). Then Ai Aj ? Aj Am because the angle Ai Aj Am is supported
by the diameter Ai Am . As diameters transform to diameters, also angle Ai Aj Am is
Solution 1. a) Clearly Ai Aj k Ak Al iff i k l j (mod n) (see figure 9). Consequently, supported by a diameter giving Ai Aj ? Aj Am . Since Ai Aj ? Aj Am and Ai Aj ? Ak Al ,
0 0 0 0
0 0 0
Ai Aj k A k Al ) i k l j (mod n) ) we get Aj Am k Ak Al and, by a), also Aj Am k Ak Al . Hence Ai Aj ? Ak Al .
0 0 0 0 0 0 0 0
) (i k) (l j ) (mod n) )
) i k l j (mod n) ) IMO Team Selection Contest
) i0 k0 l0 j 0 (mod n) )
) Ai Aj k Ak Al :
0 0 0 0 First Day
Ak
1. Let k > 1 be a fixed natural number. Find all polynomials P (x) satisfying the
Aj Aj
condition P (xk ) = (P (x))k for all real numbers x.
Ai Ai A k+ n4 Answer: P (x) = 0 and P (x) = xn , where n is an arbitrary non-negative integer; in the
Al case of odd k also P (x) = xn .
Ak Al+ n4 Al
Solution. Let the degree of a polynomial P (x) be n > 0, then
Figure 9 Figure 10
P (x) = an xn + an 1 xn 1
+ : : : + a 1 x + a0 ;
_ _ _ _
b) Using a well-known theorem of geometry, we obtain
where an 6= 0.
Ai Aj ? Ak Al () Ai Ak Aj Al Ak Aj Al Ai () + = +
Let i be the largest index smaller than n for which ai 6
= 0 (suppose that such an index i
exists), then
() k i l j j k i l ( ) + ( ) ( ) + ( ) (mod n) ()
() k l i j n 2( + ) 0 (mod ) P (xk ) = an xkn + ai xki + ai 1 xki k
+ : : : + a1 xk + a0
where the arcs considered are taken on the circumcircle of the polygon. Thus and
AA ?A A ) k l i j
i j k l 2( + ) 0 (mod n) ) (P (x))k = (an xn + ai xi + ai 1 xi 1
+ : : : + a 1 x + a 0 )k :
) k l i j 2 ( + ) n) )
0 (mod
Find next the coefficient of the term x(k 1)n+i in both polynomials. As i < n, we get
) k 0 l 0 i0 j 0
2( + ) n) )
0 (mod
kn > (k 1)n + i > ki, and the coefficient of this term in the polynomial P (x ) is
k
) A i Aj ? Ak Al : 0 0 0 0 therefore 0. On the other hand, we get the term an xn in (P (x))k iff we take the term
15 16
an xn from k 1 factors and ai xi from one factor; therefore in the polynomial (P (x))k , Construction for n = 4k :
the coefficient of this term is kank 1 ai 6= 0. This contradiction shows that there is no such
index i and the polynomial P (x) has the form P (x) = an xn . Also notice that if the degree
; k
(0 4 ) 4k 1
n of the polynomial P (x) is 0, then also P (x) = a0 = an xn . 9 9
k ; k >
> (6 k ; k + 1)
1 6 2 >
>
From the equality P (xk ) = (P (x))k , we now get an xnk = ank xnk for all real x, i.e. an = ank
(2 1 2 ) 1
= k ; k + 3) =
(2 k ; k + 2)
3 2
k 5 (6 3 6
k 6
or an (akn 1 1) = 0. Hence an 2 f 1; 0; 1g for odd k (an = 0 is possible only if n = 0) and >
> ;
>
>
an 2 f0; 1g for even k. (1 4 ; k 2)
; 4 k 3 (4 k + 1; 8k 1) k
4 2
9 9
2. Let O be the circumcentre of the acute triangle ABC and let lines AO and BC inter- (4k 1; 4k + 2) 3 > (2; 8k 4 >
>
> 2)
=
sect at point K . On sides AB and AC , points L and M are chosen such that jKLj = jKB j k ; k + 4) = ; k
and jKM j = jKC j. Prove that segments LM and BC are parallel.
(4 3 4
k 7 (4 8 4)
; k 8
;
>
> >
>
1
(2 k + 1; 6k) 4 k 1 (2k ; k + 2) 4k
2 6 4
Solution. Draw heights for triangles KBL and KCM from
A
q
the vertex K and let their bases be S and T , respectively.
Also lengthen the segment AK until it intersects the cir- Construction for n = 4k + 1:
cumcircle of ABC at point P (see Figure 13). As segment
L M (0 4 ; k + 1) 4k + 1 (5 k + 1; 7k + 1) 2k
AP is a diameter of the circumcircle of ABP C , the trian-
1 1
S O T 9 9
gles ABP and ACP are right-angled. Triangle ASK is k + 1; k) > k + 1; 8k + 1) 2k + 2 >
>
similar to triangle ABP and triangle AT K is similar to B
C
(
k + 2; k
1
>
=
(2
k + 2; 8k) 2k + 4
=
triangle ACP (their corresponding sides are parallel), so
K ( 1)
k 3 (2
k
>
> >
>
jAS j jAK j jAT j P k; 1) ; k k; 7k + 2) k;
jAB j jAP j jAC j . Hence triangle AST is similar to
= = (2 2 1 (3 4
9 9
triangle ABC and therefore ST k BC . As jLS j jSB j and Figure 13 (5k; 5k + 2) 2 > k; 6k + 1) 2k + 1 >
=
> (4
>
jMT j jT C j, we obtain LM k BC .
= k 1; 5k + 3) = k 1; 6k + 2) 2k + 3 =
(5
; k 4
1
(4
; k
>
> >
>
3. For which natural number n is it possible to draw n line segments between vertices (4 k + 2; 6k) 2 k 2 (3 k + 1; 7k) k
4 1
of a regular 2n-gon so that every vertex is an endpoint for exactly one segment and these
segments have pairwise different lengths? (There are actually many other constructions for both cases.)
Answer: n = 4k and n = 4k + 1, where k is an arbitrary positive integer.
Solution. Colour the vertices of the 2n-gon alternately black and white. Call the smallest Second Day
number of sides needed to pass when moving from one vertex to another the weight of
the segment with endpoints at these vertices. We see that segments with odd weigths 4. Denote
connect vertices with different colours, but segments with even weights connect vertices
with the same colour. X
m
k
Suppose that the required construction exists for a given n. As there is an equal number
f (m) = ( 1)
k
os :
k=1
m+1
2
of vertices of both colours and all segments with odd weights occupy an equal number
of vertices of both colours, also all segments with even weights must take occupy an For which positive integers m is f (m) rational?
equal number of vertices of both colours. Therefore the number of segments connecting
two white vertices equals the number of segments connecting two black vertices, and Answer: for all positive integers m.
the number of segments with even weights is an even number. Therefore there must be Solution Fix a positive integer m arbitrarily and take
an even number of even numbers among numbers 1; 2; : : : ; n, which is possible only if

n 0 (mod 4) or n 1 (mod 4). a = os
2(2 m + 1)
6
= 0 :
We show next that such sets exist for n = 4k and n = 4k + 1. In the following tables,
the segments are grouped into blocks of parallel segments. In each row of a block, (x; y ) Using the formula
shows that a segment is drawn between vertices x and y , next comes the weight of this
1
segment and, after the end of a block, the number of segments in this block is shown. os x os y = ( os( x y ) + os(x + y )) ;
2
17 18
we get the chain of equations Let now be k > l. Then blog2 n + 1 2(blog3 n + 1) or
X
m
k b n 2blog3 n + 1 ; (15)
f (m) a a=
log2
k
= ( 1) os
k=1
m+1
2
this holds for n = 2 and n = 8. With immediate check we see that for 4 n 7 and
X 9 n < 27 the inequality (15) does not hold. If n 27 then
m
k k
= ( 1) os os =
k=1
2m + 1 2(2m + 1) log2 n log2 27 = log2 3(log3 n log3 27) < 2(log3 n 3) = 2 log3 n 6 ;
Xm k
1
k hence 2 log3 n log2 n > 6 log2 27 > 6 5 = 1. Therefore blog2 n < b2 log3 n
2blog3 n + 1 for all n 27 . Hence (15) holds iff n = 2 or n = 8.
= ( 1) os +
2 2m + 1 2(2m + 1)
k=1
k Therefore the only possible values for n are 1, 2, 3 and 8. For them, we get A(1) = 1 = 20 ,
+ os
2m + 1
+
2(2m + 1)
=
A(2) = 2 = 21 , A(3) = 4 = 22 and A(8) = 32 = 25 . Hence all these values suit.
1
Xm (2k 1) (2k + 1)

k 6. Call a convex polyhedron a footballoid if it has the following properties.
= ( 1) os + os =
2 2(2m + 1) 2(2m + 1)
k=1
1
(2m + 1)
(1) Any face is either a regular pentagon or a regular hexagon.
m
= os + ( 1) os =
2 m + 1) 2(2 2(2m + 1) (2) All neighbours of a pentagonal face are hexagonal (a neighbour of a face is a face that
=
1
2
( a+( 1)
m
0) =
1
2
a:
has a common edge with it).
Find all possibilities for the number of pentagonal and hexagonal faces of a footballoid.
1
Hence f (m) = for all m. Answer: there are 12 pentagonal and 20 hexagonal faces.
2
Solution. We show first that there exists a footballoid with 12 pentagonal and 20 hexag-
5. Find all natural numbers n for which the number of all positive divisors of the onal faces. Start with a regular icosahedron and abstract from every vertex a regular
number lcm (1; 2; : : : ; n) is equal to 2k for some non-negative integer k . 1
pyramid with lateral edge of the edge of the icosahedron. In such a way, we get 12
3
Answer: 1, 2, 3 and 8.
regular pentagons instead of 12 vertices of icosahedron and we get 20 regular hexagons
Solution. Let P be the set of all prime numbers. Let Æ (m) denote the number of positive instead of 20 faces of icosahedron. All neighbours of any pentagonal face are hexagonal.
divisors of natural number m and let A(n) = Æ (lcm (1; : : : ; n)). Denote by p . m the Now show that it is the only possibility. Let B be a footballoid. Consider an arbitrary
exponent of prime number p in the canonical representation of m. Notice that vertex of B ; let it belong to x pentagonal and y hexagonal faces. Then x + y 3 as every
Y vertex of a polyhedron belongs to at least 3 faces. As the sizes of the interior angles of
A(n) = Æ (lcm (1; : : : ; n)) = Æ pmax(p.1;:::;p.n) = pentagonal and hexagonal faces are 108Æ and 120Æ , respectively, we get x 108Æ + y 120Æ <
360 . Hence x + y 3 and x > 0. Therefore x + y = 3, which means that every vertex of
p2P Æ
Y Y
= Æ pblogp n = b
( logp n + 1) : a footballoid belongs to exactly 3 faces, at least one of which is pentagonal. As these 3
p2P p2P faces are pairwise neighbours and pentagonal faces cannot be neighbours, every vertex
must belong to exactly one pentagonal and two hexagonal faces.
Therefore A(n) is a power of 2 iff all numbers in the form blogp n + 1, where p 2 P, are Consider an arbitrary hexagonal face. All its vertices belong to one pentagonal and one
powers of 2. Let blog2 n + 1 = 2k and blog3 n + 1 = 2l . As log2 n log3 n, we get k l. hexagonal face. Therefore the neighbours of a hexagonal face are alternately pentagonal
Consider two cases. and hexagonal, so there are exactly three of both kinds.
If k = l then Now cover every pentagonal face with a regular pentagonal pyramid, whose lateral
blog2 n = blog3 n ; (14) edges are continuations of the (hexagonal) neighbours of this face. In this way, hexag-
onal faces become equilateral triangles and pentagonal faces are replaced with vertices
this holds for n = 1 and n = 3. With immediate check we see that, for n = 2 and in which five edges meet. As any two neighbours of a pentagonal face which are neigh-
4 n < 8, the equation (14) does not hold. If n 8 then bouring themselves meet under the same angle (two regular hexagons and one regular
pentagon can meet in one vertex in principle in only one way), also the triangles meet-
log2 n 3 = log2 n log2 8 = log2 3(log3 n log3 8) > log3 n log3 8 ; ing in a vertex of our new polyhedron meet under the same angle. Therefore the new
hence log2 n log3 n > 3 log3 8 > 3 2 = 1 . Therefore blog2 n > blog3 n for all n 8. polyhedron is a regular icosahedron. It has 12 vertices and 20 faces, so the footballoid
Hence (14) holds iff n = 1 or n = 3. B had to have 12 pentagonal and 20 hexagonal faces.
19 20
2004/2005
Tartu 2005
As rays BK and CL are opposite-directed, we have
|AB| + |AC|
|BC| = ⇐⇒ |BC| = |BK| + |CL| ⇐⇒ K = L .
2
3. On some square of an infinite squared plane, there is a cube which covers the square
exactly. The top face of the cube is white, the other faces are black. With one step, one
can turn the cube over any edge so that it starts covering a neighbouring square. Is it
possible to achieve a situation where the cube lies on the initial square with the white
Selected Problems from Open Contests face at the bottom, making exactly
a) 2004 steps; b) 2005 steps? (Juniors.)
1. In the final tournament of football championship, the teams are divided into groups
of four. Each team plays one game with every other team in its group. A win gives 3
points, a draw 1 point and a loss 0 points. From each group, two teams advance so that
each advancing team gets at least as many points as each non-advancing team. A
a) What is the smallest possible score of an advancing team?
b) What is the biggest possible score of a non-advancing team? (Juniors.) D E
Answer: a) 2; b) 6.
Solution. a) If one team in a group wins all matches and the other three draw all matches,
B KL C
then the second advancing team collects just 2 points. Let us prove that a team cannot
advance with less than 2 points. Indeed, if team A gets at most 1 point, it must lose at Figure 1 Figure 2 Figure 3
least two games. Therefore two teams get at least 3 points and outperform A, so that A
cannot be in top two. Answer: a) yes; b) no.
b) If one team in a group loses all matches and the other three win cyclically (X wins Solution 1. a) Turn the cube two steps forward, one step to the right, two steps backward,
Y , Y wins Z and Z wins X ), then three teams collect 6 points and one of them does one step to the left (see Figure 2). After these six steps, the cube gets back onto the initial
not advance. Let us prove that a team with more than 6 points must advance. If team square but the white face is now at the bottom. The rest 1998 steps are made in pairs:
C gets at least 7 points, it must win at least two games. As the losing teams can get at turn the cube onto arbitrary neighbouring square and then turn it back onto the initial.
most 6 points, team C must be in top two. b) We colour squares dark and light by diagonals so that the cube lies on a light-coloured
square at the beginning (see Figure 3). Since, from any square, the cube can move only
2. In triangle ABC let D, E be the midpoints of AB and AC , respectively. Prove that to squares of the opposite colour, the cube lies on a light-coloured square after any even
the intersection point of the bisectors of angles BDE and CED lies on AB if and only if number of steps and on a dark-coloured square after any odd number of steps. Thus
the length of BC is equal to the arithmetic mean of the lengths of AB and AC . (Juniors.) after 2005 steps, the cube lies on a square different from the initial.
Solution 2. a) Turn the cube two steps forward, one step to the left, two steps forward,
Solution. Let K and L be the intersection points of BC with the bisectors of two steps to the right, four steps backward, one step to the left. After these 12 steps, the
angles BDE and CED, respectively (see Figure 1). As DE k BC , we have cube is back on the initial square but the white face is now at the bottom. Repeating this
|AB| |AC|
∠BDK = ∠EDK = ∠BKD, and |BK| = |BD| = . Similarily |CL| = . cycle, we see that, after any odd number of repetitions, the white face of the cube is at
2 2 the bottom and, after any even number of repetitions, the white face is at the top. Since
Therefore
2004 = 12 · 167, we obtain the desired result by repeating this cycle for 167 times.
|AB| + |AC| b) Assume that the cube makes a circuit and gets back onto the initial square. Let a,
|BK| + |CL| = .
2 b, c, and d be the numbers of steps made during this circuit respectively to the right,
to the left, up, and down. Then a = b and c = d. Thus the cube makes altogether
1 2
K
a + b + c + d = 2(a + c) steps which is even number. Hence the cube cannot get back to Solution. Let ∠O1AG1 = α and ∠O1 G1 A = β H1 H2
the initial square after 2005 steps. (see Figure 4). If line G1 K touches circle C1 , then β P
∠AH1 G1 = 90◦ and ∠H1 G1 O1 = ∠AG1 O1 = β. β
G1 O1 O2 G2
Also, ∠G1 P A = 90◦ and ∠P AO1 = G1 AO1 = α.
a+b
4. Relatively prime positive integers a and b are chosen in such a way that is From the right-angled triangles AH1 G1 and P1 αα P2
a−b G1 P A, we get
also a positive integer. Prove that at least one of the numbers ab + 1 and 4ab + 1 is a A
perfect square. (Juniors.) α + 2β = 90◦ , Figure 4
a+b a m+1 β + 2α = 90◦ .
Solution 1. Let = m. Then a + b = ma − mb implying = . As a and b
a−b b m−1
are relatively prime, there exists an integer k such that m + 1 = ka and m − 1 = kb. By Solving the system gives α = β = 30◦ . Therefore ∠P AG1 = 60◦ and
multiplying these equalities, we get m2 − 1 = k 2 ab implying k 2 ab + 1 = m2 . Number k ∠P AG2 = 120◦ − 60◦ = 60◦ . Thus we have ∠P AG1 = ∠P AG2 . As circles C1 and
as a common divisor of numbers m − 1 and m + 1 must be a divisor of their difference C2 lie in equal angles and touch at P , their radii must be equal. Therefore AK is the
2. Hence k can only be 1 or 2 and we are done. symmetry axis. By symmetry, line G2 K is a tangent to C2 and touches the circle at H2 .
Solution 2. As
a+b a−b+b+b 2b 7. A king wants to connect n towns of his kingdom with one-directional airways so
= =1+ ,
a−b a−b a−b that, from each town, exactly two airlines depart. From each town, it must be possible
2b to fly to every other town with at most one change. Find the biggest n for which this
we see that must be an integer. Numbers b and a − b are relatively prime because plan is feasible. (Seniors.)
a−b
a and b are relatively prime. Hence 2 must be divisible by a − b. Therefore a − b = 1 or
Answer: 6.
a − b = 2. The former case implies 4ab + 1 = 4(b + 1)b + 1 = (2b + 1)2 , the latter case
implies ab + 1 = (b + 2)b + 1 = (b + 1)2 . Solution. From a fixed town, one can get directly to two towns and further to at most
four more towns. Thus the number of towns cannot exceed 1 + 2 + 4 = 7.
a √ Let us assume that a suitable airway plan for 7 towns exists. From each town, there
5. The teacher has chosen positive integers a and b such that · a2 + b2 is an integer. must be a unique way to every other town (either direct or with one change), otherwise
b
a) Silly-Sam claims that a is divisible by every prime factor of b. Prove that he is right. some town would have less than 6 possible final destinations. Without loss of generality
assume that there is a direct flight from town L1 to towns L2 and L3 , from town L2 to
b) Silly-Sam claims that actually b 6 a. Is he right this time? (Seniors.)
towns L4 and L5 , and from town L3 to towns L6 and L7 (see Figure 5).
Answer: b) no.
Solution. a) Let p be √ an arbitrary prime factor of b. If the observed expression is an L4
integer,√the number a a2 + b2 must be divisible by p. As p is prime, either a is divisible L5
L2
by p or a2 + b2 is divisible by p. In the latter case, squaring gives that a2 +b2 is divisible
by p2 . By the initial assumption, b2 is divisible by p2 , hence also a2 is divisible by p2 . L1 L6
Therefore a is divisible by p in both cases. L3
a √ L7
b) The teacher may choose a = 12 and b = 16. In this case, · a2 + b2 = 15 is an
b
integer. Therefore the inequality b 6 a might be wrong. Figure 5 Figure 6 Figure 7
6. Two circles C1 and C2 with centres O1 and O2 , respectively, are touching exter-
From L2 , there must be a way to all towns in list L1 , L3 , L6 and L7 . As the direct flights
nally at P . On their common tangent at P , point A is chosen, rays drawn from which
from L2 take to towns L4 and L5 , there must be a direct flight from L4 to two towns in
touch the circles C1 and C2 at points P1 and P2 both different from P . It is known that
the list and from L5 to the other two. Without loss of generality assume that there is a
∠P1 AP2 = 120◦ and angles P1 AP and P2 AP are both acute. Rays AP1 and AP2 intersect
direct flight from L4 to L3 . But now there can be a flight from L4 to none of L1 , L6 , L7
line O1 O2 at points G1 and G2 , respectively. The second intersection between ray AO1
because otherwise there would be two ways to get from town L4 to town L3 , L6 , L7 ,
and C1 is H1 , the second intersection between ray AO2 and C2 is H2 . Lines G1 H1 and
respectively.
AP intersect at K . Prove that if G1 K is a tangent to circle C1 , then line G2 K is tangent
to circle C2 with tangency point H2 . (Seniors.) A suitable airway plan for 6 towns exists, as shown on Figure 6 or Figure 7.
3 4
8. For which integers a does there exist two different finite sequences of positive inte- 2004 2005
gers i1 < i2 < · · · < ik and j1 < j2 < · · · < jl such that
(ai1 + 1)(ai2 + 1) · · · (aik + 1) = (aj1 + 1)(aj2 + 1) · · · (ajl + 1) ?
2004
2004
(Seniors.)
Answer: −1, 0, and 1.
Solution. In the case a = −1, both sides of the equation equal to zero whenever both Figure 8 Figure 9 Figure 10
sequences contain an odd number. In the case a = 0, both sides equal to one irrespective
of the sequences. In the case a = 1, all factors of the products are equal to 2, so the Answer: a) Kalle; b) Kalle; c) Juku.
products are equal whenever the sequences have the same length. Therefore suitable
Solution. a) To Juku’s first move, Kalle can reply with a move after which a part of shape
sequences exist in these three cases.
2 × 2 has been eaten up (see Figure 8). Irrespectively of Juku’s second move, Kalle can
Let us prove that if |a| > 1, then such sequences do not exist. Suppose the contrary, make one more move. After that, Juku has no move.
i.e. i1 , . . . , ik and j1 , . . . , jl are different sequences that lead to equal products. We
b) When Juku eats up some two pieces, Kalle replies by eating two pieces which lie
may assume that no integer is in both sequences or else the respective terms can be
symmetrically with respect to the midpoint of the cake (see Figure 9). This guarantees
cancelled in the products. After deletions, both sequences are still nonempty or else we
that, after every move by Kalle, pieces not yet eaten up are situated symmetrically to
get an equation between 1 and the product of integers not being equal to 1. We can also
the midpoint of the cake and, as it is not possible to eat two symmetric pieces at one
assume that i1 < j1 .
move by the same player, Kalle can always follow the strategy described. Hence Juku’s
Multiplying and then removing the parentheses on both sides gives an equation be- moves come first to the end.
tween sums of powers of a. Both sides contain term 1 and we can reduce that. This
c) On his first move, Juku can eat two pieces between which the midpoint of the cake
ends up in the equation of form
lies and later use the strategy of Kalle from part b) (see Figure 10).
ai1 +···+ik + · · · + ai1 = aj1 +···+jl + · · · + aj1 . Remark. This game is called Cram, its full analysis for the case where length and width
are both odd numbers seems to be quite complicated and is not completed yet.
The smallest exponent is i1 on the left-hand side and j1 on the right-hand side. As
j1 > i1 + 1, the right-hand side is divisible by ai1 +1 . On the left-hand side, all terms
10. Find all functions f : R → R satisfying
except the last one are divisible by ai1 +1 . Hence the left-hand side is not divisible by
ai1 +1 , a contradiction. f (x + f (y)) = x + f (f (y))
Comment. From the solution, we get that for each a (|a| > 1), any finite subset I of
for all real numbers x and y whereby f (2004) = 2005. (Seniors.)
natural numbers can be assigned a unique number
Y Answer: f (x) = x + 1 is the only such function.
f (I) = (ai + 1). Solution 1. Taking y = 0, we get the equality f (x + f (0)) = x + f (f (0)). Making the
i∈I substitution x + f (0) = z, we obtain f (z) = z − f (0) + f (f (0)) for every real number z.
The number determines the subset uniquely. This can be used in proving that the num- Hence f is a linear function f (x) = x + a. To find a, take x = 2004 in the last expression
ber of finite subsets of natural numbers is countable. and, by using the known value of the function, obtain a = 1. A quick checking shows
that the function f (x) = x + 1 satisfies the conditions of the problem.
Solution 2. Taking x = −f (f (y)), we see that f (y) = 0 for some y. Then f (x) = x + f (0).
9. Mother has baked a platecake and cut it into m×n square pieces of equal size. Kalle From the condition f (2004) = 2005, we find f (0) = 1. Thus f (x) = x + 1.
and Juku play the following game. Each player at his move eats two pieces having a
common side. Moves are made by turns, Juku starts. A player who cannot move loses.
Who wins the game if 11. Three rays are going out from point O in space, forming pairwise angles α, β and
γ with 0◦ < α 6 β 6 γ 6 180◦ . Prove that
a) m = 3, n = 3;
α β γ
b) m = 2004, n = 2004; sin + sin > sin .
2 2 2
c) m = 2004, n = 2005? (Seniors.) (Seniors.)
5 6
O Solution 1. Choose points A, B, and C on the three Solution. Let a1 , a2 , a3 , a4 be the smallest numbers
10 120 5 15
α α rays, respectively, so that |OA| = |OB| = |OC| = d and b1 , b2 , b3 , b4 be the biggest numbers of the first,
d d 12 24 3
2 2 and ∠BOC = α, ∠COA = β, ∠AOB = γ . These second, third and fourth rows, respectively. Obvi-
2 8
three points must be different and do not lie on the 4 ously b4 = a4 and b3 = a3 b4 . In the second row,
same line. From isosceles triangle BOC with vertex there exists a number which is the product of the
B C Figure 12
angle α and side length d (see Figure 11), we obtain biggest element in the third row and some other
α element in the second. Thus b2 > a2 b3 = a2 a3 a4 .
Figure 11 |BC| = 2d sin . Analogously from triangles COA
2 Finally for the first row, we obtain similarily b1 > a1 b2 > a1 a2 a3 a4 . All numbers in the
β γ
and AOB, find |CA| = 2d sin and |AB| = 2d sin . As triangle are greater than 1, otherwise we could find two equal numbers in it. Since all
2 2 numbers are different, we have a1 a2 a3 a4 > 2 · 3 · 4 · 5 = 120. Hence b1 > 120. Number
|BC| + |CA| > |AB|, 120 is achievable as follows from Figure 12.
we obtain
α β γ
Selected Problems from the Final Round of National
2d sin + 2d sin > 2d sin ,
2 2 2 Olympiad
giving the desired inequality.
Solution 2. At first, we show that γ 6 α + β. Consider the two of the given three rays
which form angle of size γ , and build two cones by moving the third ray around both 1. Rein solved a test on mathematics that consisted of questions on algebra, geometry
rays. On the plane defined by the axes, the first cone cuts angle α off from angle γ and and logic. After checking the results, it occurred that Rein had answered correctly 50%
the second cone cuts angle β. Assume γ > α+β, then the last two angles do not overlap, of questions on algebra, 70% of questions on geometry and 80% of questions on logic.
therefore the cones have no common points except the vertex O, a contradiction. Thus Thereby, Rein had answered correctly altogether 62% of questions on algebra and logic,
γ 6 α + β. On the other hand, α + β + γ 6 360◦ , giving and altogether 74% of questions on geometry and logic. What was the percentage of
correctly answered questions throughout all the test by Rein? (Grade 9.)
γ α β γ
6 + 6 180◦ − . Answer: 65%.
2 2 2 2
Solution: Let a, g, and l be the numbers of correctly answered questions on algebra,
Therefore geometry and logic, and A, G, and L be the total number of questions on algebra, ge-
γ

α β

α β β α α β ometry and logic, respectively. The conditions of the problem imply a = 0.5A, g = 0.7G,
sin 6 sin + = sin cos + sin cos < sin + sin l = 0.8L, a + l = 0.62(A + L), g + l = 0.74(G + L). After substituting to the fourth and
2 2 2 2 2 2 2 2 2
fifth equation, we obtain 0.5A + 0.8L = 0.62A + 0.62L, or equivalently 0.12A = 0.18L,
α β γ β giving
because , and are grater than 0◦ but do not exceed 90◦ , and also cos < 1 and
2 2 2 2 A = 1.5L,
α
cos < 1.
2 and 0.7G + 0.8L = 0.74G + 0.74L, or equivalently 0.04G = 0.06L, giving
G = 1.5L.
12. We call a number triangle amazing if all its elements are 21 84 7
different positive integers and, under every two neighbouring 4 12 Now
numbers, the quotient by division of the greater of two by the 3
smaller is written. In the figure, one amazing triangle with side a + g + l = 0.5A + 0.7G + 0.8L = 0.75L + 1.05L + 0.8L = 2.6L
length 3 is shown. Find the smallest number which can occur as the greatest element in and
an amazing triangle with side length 4. (Seniors.)
A + G + L = 1.5L + 1.5L + L = 4L.
Thus the percentage of correct answers was
a+g+l 2.6
= = 65%.
A+G+L 4
7 8
2. Represent the number
√
q
3
1342 167 + 2005
in the form where it contains only addition, subtraction, multiplication, division and
square roots. (Grade 10.)
√ Figure 13 Figure 14
Answer: 2 167 + 1.
Solution 1. First, isolate the terms divisible by 167:
√ √ √ Solution. Suppose all dark squares are covered. Since one hook cannot cover more than
1342 167 + 2005 = 1336 167 + 2004 + 6 167 + 1 = one dark square, in total at least nine hooks are needed. As only eight of these are avail-
√ √
= 8 · 167 167 + 12 · 167 + 6 167 + 1. able, one of the dark squares must remain uncovered. There are three fundamentally
different possibilities: the free square lies in the corner, in the middle of a side or in the
Second, represent the result in the form centre of the board. The corresponding tilings are shown in the Figure 14.
√ √ √ √
1342 167 + 2005 = (2 167)3 + 3 · (2 167)2 + 3 · (2 167) + 1 =
√ 4. Real numbers x and y satisfy the system of equalities
= (2 167 + 1)3 .
(
sin x + cos y = 1
Therefore .
cos x + sin y = −1
√ √
q
3
1342 167 + 2005 = 2 167 + 1.
Prove that cos 2x = cos 2y. (Grade 11.)
√
Solution 2. Search the answer in the form a 167 + b, where a and b are integers. Then Solution 1. After squaring both sides, we obtain
we must have
sin2 x + 2 sin x cos y + cos2 y = 1,
√ √
(a 167 + b)3 = 1342 167 + 2005, cos2 x + 2 cos x sin y + sin2 y = 1.
or equivalently After adding the equations and dividing by 2, we obtain
√ √ √
167a3 167 + 3 · 167a2 b + 3ab2 167 + b3 = 1342 167 + 2005. sin x cos y + sin y cos x = 0,
Thus a and b must satisfy the system or equivalently
(
167a3 + 3ab2 = 1342 sin(x + y) = 0.
.
501a2 b + b3 = 2005 Hence x + y = kπ, where k is integral. Therefore 2x = 2kπ − 2y, giving cos 2x = cos 2y.
Solution 2. After adding the equations, we obtain
The second equation can be rewritten in the form (501a2 +b2 )b = 2005. Since a and b dif-
fer from 0 and are integers, 501a2 + b2 must be a divisor of 2005 that is greater than 501. sin x + sin y + cos x + cos y = 0,
The only possibility is now 501a2 + b2 = 2005, giving b = √ 1, a = 2. Simple check
√ shows
that a = 2, b = 1 satisfy the first equation as well. Hence (2 167+1)3 = 1342 167+2005. which is equivalent to
x+y x−y x+y x−y
2 sin cos + 2 cos cos = 0,
3. A 5×5 board is covered by eight hooks (a three unit square figure, shown 2 2 2 2
in the picture) so that one unit square remains free. Determine all squares of implying
the board that can remain free after such covering. (Grade 10.)
x−y x+y x+y
Answer: All the squares that are colored dark in the Figure 13. cos sin + cos = 0.
2 2 2
9 10
x−y x−y π assumption that C is more important than A. Analogously, if C comes up first, then
If cos = 0, then = (2k − 1) · , giving 2x − 2y = (2k − 1) · 2π.
2 2 2 we get a contradiction with the assumption that B is more important than A. Thus our
x+y x+y x+y
Hence cos 2x = cos 2y. If sin + cos = 0, then tan = −1, original assumption was false.
2 2 2
x+y π π Remark. In graph theory, the relation “is more important than” of this problem is called
implying = kπ − and x + y = 2kπ − . Therefore
2 π 4
π
2 postdominance.
cos y = cos − − x = cos + x = − sin x, leading to sin x + cos y = 0 that
2 2
contradicts the first equation of the initial system. 7. In a fixed plane, consider a convex quadrilateral ABCD. Choose a point O in the
plane and let K , L, M , and N be the circumcentres of triangles AOB, BOC , COD, and
5. Let a, b, and n be integers such that a + b is divisible by n and a2 + b2 is divisible DOA, respectively. Prove that there exists exactly one point O in the plane such that
by n2 . Prove that am + bm is divisible by nm for all positive integers m. (Grade 11.) KLMN is a parallelogram. (Grade 11.)
Solution 1. We prove that a and b are divisible by n, then the claim immediately follows. Solution. If O is the point described in the problem, then D
As M
we must have KL ⊥ BO because K and L both lie on
the perpendicular bisector of BO. Similarly LM ⊥ CO, C
2ab = (a + b)2 − (a2 + b2 ), MN ⊥ DO, and NK ⊥ AO. Let O be the intersec-
tion point of the diagonals of ABCD (see Figure 15). O
2ab is divisible by n2 . Let p be any prime in the prime decomposition of n and let α be N
Then both KL and MN are perpendicular to BD, giving
its exponent. Then the exponent of p is at least 2α in the prime decomposition of 2ab, L
KL k MN . Similarly LM k NK . Therefore the oppo-
and at least 2α − 1 in the prime decomposition of ab. Therefore at least one of numbers
site sides of KLMN are parallel, meaning that KLMN
a and b must be divisible by pα . As a + b is divisible by n and hence by pα , also the other
is a parallelogram. On the other hand, if O is a point for A B
of the numbers a and b must be divisible by pα . Altogether, this means that both a and K
which KLMN is a parallelogram, we have KL k MN .
b are divisible by n.
Then also BO k DO, giving that O lies on the line BD. Figure 15
We can show similarily that O lies also on the line AC .
6. A post service of some country uses carriers to transport the mail; each carrier’s Therefore O is the intersection point of the diagonals.
task is to bring the mail from one city to a neighbouring city. It is known that it is
possible to send mail from any city to the capital P . For any two cities A and B, call
8. Does there exist an integer n > 1 such that
B more important than A, if every possible route of mail from A to the capital P goes
through B. n −1
22 −7
a) Prove that, for any three different cities A, B, and C , if B is more important than A
is not a perfect square? (Grade 11.)
and C is more important than B, then C is more important than A.
b) Prove that, for any three different cities A, B, and C , if both B and C are more Answer: Yes.
n
important than A, then either C is more important than B or B is more important Solution 1. Let us show that if n = 5, then the number 22 −1 − 7 is not a perfect square.
than C . (Grade 11.) Note that 210 = 1024 ≡ 1 (mod 11), giving 231 = 2 · (210 )3 ≡ 2 (mod 11). Hence the
remainder of division of 231 − 7 by 11 is 6. On the other hand, squares of integers can
Solution. a) Let t be any possible mail route from A to P . Since B is more important have remainders 0, 1, 4, 9, 5, and 3 in division by 11.
than A, the route t goes through B. The end part of t from B to P is a mail route from Solution 2. Computation gives 231 − 7 = 32768 · 65536 − 7 = 2147483641 but
B to P . Since C is more important than B, this route goes through C . Therefore, t goes 463402 = 2147395600 and 463412 = 463402 + 2 · 46340 + 1 = 2147488281. So
through C . 463402 < 231 − 7 < 463412 .
5
b) Assume that the claim doesn’t hold, that is, C is not more important than B and B Remark. The number 11 is the least modulus with respect to which 22 −1 − 7 is not a
is not more important than C . Then there exist a route from B to P not going through quadratic residue. There exist greater such numbers, e.g. 31.
C and a route from C to P not going through B. Consider any route from A to P .
Since B and C are more important than A, this route goes through both B and C . Start
moving from A along this route and find out which of the cities B and C comes up
first. If it is B, then continue along the route to P that doesn’t pass through C . So we
have found a route from A to P that doesn’t go through C , a contradiction with the
11 12
9. Punches in the buses of a certain bus company always cut exactly six
holes into the ticket. The possible locations of the holes form a 3 × 3 table Ai+1
as shown in the figure. Mr. Freerider wants to put together a collection
of tickets such that, for any combination of punch holes, he would have a Q
ticket with the same combination in his collection. The ticket can be viewed P
both from the front and from the back. Find the smallest number of tickets in such a Ai
collection. (Grade 12.)
Figure 16 Figure 17
Answer: 47.
Solution. Instead of holes, we can deal with non-holes — the locations that are not cut
through during punching. The number of possibilities to choose 3 locations for non- As P and Q are different, there exist such vertices of the polygon Ai and Ai+1 that the
holes from 9 locations is point Q is inside or on the side of Ai P Ai+1 , not coinciding with the vertex P (see Figure
17). But this implies ∠Ai QAi+1 > ∠Ai P Ai+1 , a contradiction. Hence there exists only
9
= 84. one point satisfying the conditions of the problem.
3
Remark. For n = 4, 6 one may find several different points satisfying the conditions of
One ticket can represent either one punch combination that is symmetric with respect the problem.
to the central axis parallel to the longer sides of the ticket or two different combinations
that are mirror images of each other with respect to this axis. In the case of symmetric
combinations, either all three non-holes lie in the second column (1 possibility) or one 11. A string having a small loop in one end is set over a horizontal pipe so
non-hole lies in the second column and other two lie in the same rows, one in the first that the ends hang loosely. After that, the other end is put through the loop,
column and the other in the second (3 · 3 = 9 possibilities). So there are 1 + 9 = 10 pulled as far as possible from the pipe and fixed in that position whereby α
symmetric combinations and 84 − 10 = 74 non-symmetric combinations. The number this end of the string is farther from the pipe than the loop. Let α be the
of tickets needed to cover these combinations is angle by which the string turns at the point where it passes through the loop
(see picture). Find α. (Grade 12.)
74
10 + = 47. π
2 Answer: .
3
Solution. Let O and r be the centre point and the radius of the pipe, respectively. Let
l be the length of the string, A and B the loose end and the end with the loop, re-
10. Consider a convex n-gon in the plane with n being odd. Prove that if one may find spectively, and let C and D be the first and the last tangent point with the surface of
a point in the plane from which all the sides of the n-gon are viewed at equal angles, the pipe (see Figure 18). At first, find the length of AO, denote by d(α). Obviously
then this point is unique. (We say that segment AB is viewed at angle γ from point O iff |BC| = |BD| = r cot α. As ∠COD = π − 2α, the string touches the pipe along the arc
∠AOB = γ .) (Grade 12.) CD at angle π + 2α and thus the length of the string along the arc CD is r(π + 2α). After
Solution. Draw the rays from the point described in the problem through the vertices of subtracting the lengths of segments BC and BD and the arc CD from the total length
of the string, we obtain
the polygon. The point can lie either inside or outside the polygon, therefore there are
two possibilities for the rays: they divide either all the plane or only an angle into equal |AB| = l − 2r cot α − r(π + 2α).
angles (see Figure 16). The latter case would imply that the outermost rays were both
We also have
incident to one vertex and all the others were incident to two vertices of the polygon,
r
giving that n is even. This contradiction shows that the point satisfying the conditions |BO| = .
of the problem lies inside the polygon. sin α
Assume now that there are two different points P and Q inside A1 A2 . . . An , from Altogether d(α) = |AB| + |BO|, or equivalently
which all the sides are viewed at equal angles. Then for every i = 1, 2, . . . , n (tak- r
d(α) = l − 2r cot α − r(π + 2α) + .
ing An+1 = A1 ), sin α
Now we find α for which the value of d(α) is the greatest. The derivative equals
2π 2π
∠Ai P Ai+1 = , ∠Ai QAi+1 = . 2r r cos α
n n d0 (α) = − 2r − ,
sin2 α sin2 α
13 14
be periodic modulo 2N . On the other hand, if ai and aj are congruent modulo 2N ,
they are also congruent modulo N . Therefore, the sequence is periodic modulo N , a
contradiction.
O b) We are going to prove that the sequence ai = 2i−1 is periodic modulo no even num-
bers and all odd numbers. Let first the modulus N be even. If the sequence were
periodic modulo N , then, using a similar argument as in a), we get that the sequence
r r
would be periodic modulo any factor of N . But the sequence 1, 0, 0, 0, . . . of remainders
modulo 2 is not periodic, so our sequence cannot be periodic modulo N . Now let the
α α modulus N be odd. As the set of possible remainders in division by N is finite, there
D C
exist two indices i and j with i < j such that ai = 2i−1 and aj = 2j−1 are congruent
B
modulo N . Then the difference 2j−1 − 2i−1 = 2i−1 (2j−i − 1) is divisible by N . Since N
A is odd, 2j−i − 1 must be divisible by N . Hence a1 = 1 and aj−i+1 = 2j−i are congruent
modulo N , implying also that a2 = 2a1 and aj−i+2 = 2aj−i+1 are congruent modulo N ,
Figure 18 that a3 = 2a2 and aj−i+3 = 2aj−i+2 are congruent modulo N etc, that is, the sequence is
periodic modulo N .
or equivalently,
13. A crymble is a solid consisting of four white and one black unit cubes
2 − 2 sin2 α − cos α 2 cos2 α − cos α

d0 (α) = r =r . as shown in the picture. Find the side length of the smallest cube that can
sin2 α sin2 α
be exactly filled up with crymbles. (Grade 12.)
The condition d0 (α) = 0 gives the equation 2 cos2 α − cos α = 0, implying cos α = 0 or Answer: 10.
1 π π π
cos α = . Thus α = or α = , bearing in mind that 0 < α 6 . In order to find Solution. Since a crymble consists of 5 unit cubes, the volume of the cube made up from
2 2 3 2
the maximal value, we consider d0 (α) in the neighbourhoods of α found out before. The crymbles and hence also the length of its side must be divisible by 5. A cube with the
denominator sin2 α is positive in these neighbourhoods, the numerator 2 cos2 α − cos α side length 5 cannot be filled up with crymbles. To prove this, colour 27 unit cubes as
1 shown in the Figure 19. One crymble cannot fill more than one coloured cube, therefore
is negative iff 0 < cos α < . Therefore d0 (α) is positive and d(α) is increasing in the at least 27 crymbles are needed. But their volume 27 · 5 = 135 is larger than the volume
2
π π π 53 = 125 of the cube.
interval 0 < α < , d0 (α) is negative and d(α) is decreasing in the interval < α < .
3 3 2 A cube with side length 10 can be filled up with crymbles. By putting together two
π
Hence the function d(α) obtains its maximal value at α = . crymbles, construct the solid that is in the Figure 20. From two such solids, make a
3
Remark. The value for α found out in the solution is attainable iff the ratio between the
√ 2 × 2 × 5 cuboid. From such cuboids, it is possible to put together a 10 × 10 × 10 cube.
5π + 2 3
length of the string and the diameter of the pipe is at least .
6
12. A sequence of natural numbers a1 , a2 , a3 , . . . is called periodic modulo n if there
exists a positive integer k such that, for any positive integer i, the terms ai and ai+k are
equal modulo n. Does there exist a strictly increasing sequence of natural numbers that
a) is not periodic modulo finitely many positive integers and is periodic modulo all the
other positive integers;
b) is not periodic modulo infinitely many positive integers and is periodic modulo in- Figure 19 Figure 20
finitely many positive integers? (Grade 12.)
Solution. a) Suppose that the sequence a1 , a2 , a3 , . . . is not periodic modulo finitely
many positive integers, let N be the largest of these. Since 2N > N , this sequence must
15 16
IMO Team Selection Contest c1
c1
First Day Q1
c2 O1
l
Q1
O2 Q2 P M
1. On a plane, a line l and two circles c1 and c2 of different radii are given such that l
touches both circles at point P . Point M 6= P on l is chosen so that the angle Q1 MQ2 is l
as large as possible where Q1 and Q2 are the tangency points of the tangent lines drawn Q2
c2
P M
from M to and
c1 c2 ,
respectively, differing from l. Find ∠P MQ1 + ∠P MQ2 .
Figure 21 Figure 22
Answer: π.
Solution. Consider first the case where c1 and c2 are on the same side from l (see Figure
21). Let O1 and O2 be the circumcentres and r1 and r2 the radii of c1 and c2 , respec- • each Automorian is loved by some Automorian;
tively. Without loss of generality, assume r1 > r2 . Denote ∠P MQ1 = α1 , ∠P MQ2 = α2 • if Automorian A loves Automorian B, then also all Automorians honouring A
1 1 love B;
and |P M| = d. As ∠P MO1 = ∠P MQ1 and ∠P MO2 = ∠P MQ2 , we see that
2 2 • if Automorian A honours Automorian B, then also all Automorians loving A hon-
α1 α2
∠Q1 MQ2 = α1 − α2 is maximal if and only if ∠O1 MO2 = − is maximal; the our B.
2 2
α1 α2

latter holds if and only if tan − is maximal because the angle is in the first Is it correct to claim that every Automorian honours and loves the same Automorian?
2 2
quadrant. The formula of tangent of difference gives
Answer: Yes.
α1 α2 r1 r2
tan − tan − Solution. Denote by f (A) the Automorian loved by A and by g(A) the Automorian
α1 α2

2 2 = d d .
tan − = α α r 1 r2
honoured by A. The conditions of the problem imply the following:
2 2 1 + tan
1
· tan
2
1+ ·
2 2 d d • for every Automorian A, there exists an Automorian C such that f (C) = A;
Representing the result in the form • for every Automorian C , f (g(C)) = f (C);
• for every Automorian C , g(f (C)) = g(C).
α α r 1 − r2 r − r

1 2
tan − = rr = 1 2
√ ,
2 2 d+
1 2 √ d r 1 r2 We will show that f (A) = g(A) for every A. Applying f to both sides of the third
d r1 r2 √ +
r1 r2 d condition, we get
we obtain that the value of the denominator of the last expression is minimal in the case f (g(f (C))) = f (g(C)).
√
d = r1 r2 . Now
Using the second condition in both sides, we get
√ √
α1 r1 r1 α2 r2 r2
tan = =√ , tan = =√ , f (f (C)) = f (C).
2 d r2 2 d r1
α1 α2 α1 α2 π Using the first condition, this implies
i.e. tan and tan are reciprocals of each other. Therefore + = and
2 2 2 2 2 f (A) = A
∠P MQ1 + ∠P MQ2 = α1 + α2 = π.
In the other case when c1 and c2 are on the different sides from l (see Figure 22), the for all A. Using the second condition again, we obtain the desired result:
maximal size of the angle Q1 MQ2 is π which is the greatest size an angle can have. In
this case, the point M lies on the common tangent to c1 and c2 intersecting l. Then f (A) = f (g(A)) = g(A).
∠P MQ1 + ∠P MQ2 = ∠Q1 MQ2 = π. Thus, every Automorian loves and honours himself.
2. On the planet Automory, there are infinitely many inhabitants. Every Automorian
3. Find all pairs (x, y) of positive integers satisfying the equation
loves exactly one Automorian and honours exactly one Automorian. Additionally, the
following can be noticed: (x + y)x = xy .
17 18
Answer: (2, 6) and (3, 6). We see that in both inequalities, equality actually holds. Consequently, x1 = x2 ,
Solution. We have xy = (x + y)x > xx implying y > x. Let y = nx where n > 1 is a 2
x10 = x20 = x30 , and − a = 4. From the latter, we find a = −6. Thus x10 x20 x30 = 8
rational number. From the equality given in the problem, we get 3
leading to x10 = x20 = x30 = 2 which gives b = 12.
(x + nx)x = xnx . On the other hand, taking a = −6, b = 12 gives 6x2 − 24x − 4a = 6(x − 2)2 and
x3 + ax2 + bx − 8 = (x − 2)3 satisfying the conditions of the problem.
1
Raise both sides to the power of and then divide them by x; we obtain
x 5. On a horizontal line, 2005 points are marked, each of which is either white or black.
1 + n = xn−1 . (1) For every point, one finds the sum of the number of white points on the right of it and
the number of black points on the left of it. Among the 2005 sums, exactly one number
p occurs an odd number of times. Find all possible values of this number.
On the right hand side, the exponent n − 1 can be represented as a reduced fraction ,
q
therefore the number Answer: 1002.
p√ Solution. It is easy to see that the sums computed for a white point V and a black point
xn−1 = x q =q
xp M immediately following V on its right are equal. Note also that the sums are equal
also if the two points of different colour lie in the opposite order. If one interchanges
is either natural or irrational. The left hand side of (1) cannot be irrational, thus it is
two consecutive points of different colour, only the two equal sums corresponding to
natural. Hence n is natural. By choice, n > 1.
these two points change giving rise to two new equal sums. Hence, for any k, such
If n = 2, (1) gives x = 3 implying y = 2x = 6. If n = 3, (1) gives x = 2 implying transitions preserve parity of the number of occurrences of k among the 2005 sums.
y = 3x = 6. Note that (1) implies x > 2, thus 1 + n > 2n−1 . The latter inequality does
Assume there are n white and 2005 − n black points on the line. By a sequence of
not hold in the case n > 4. Hence no more solutions exist.
transitions described, collect all white points to the left. Then, going from left to right,
the corresponding sums are
Second Day
n − 1, n − 2, . . . , 1, 0, 0, 1, . . . , 2003 − n, 2004 − n.
4. Find all pairs (a, b) of real numbers such that the roots of polynomials 6x2 −24x−4a According to the invariant discovered in the first paragraph, exactly one number must
and x3 + ax2 + bx − 8 are all non-negative real numbers. occur an odd number of times also in this sequence. As the middle numbers occur in
pairs, the single number occurring an odd number of times is either the leftmost n − 1
Answer: (−6, 12). or the rightmost 2004 − n. If the former case takes place, then n − 2 = 2004 − n leading
Solution. Let x1 , x2 be the roots of the first polynomial and x10 , x20 , x30 be the roots of the to n = 1003 and n − 1 = 1002. If the other case takes place, then n − 1 = 2003 − n, giving
2 n = 1002 and 2004 − n = 1002 just like in the first case.
other polynomial. Division of the first polynomial by 6 gives x2 − 4x − a whose roots
3
are x1 and x2 , too. By Viète’s formulae,
6. On a plane, line l and a circle having no common points are given. Let AB be the
2 diameter of the circle being perpendicular to l whereby B is nearer to l than A. Let C
x1 + x2 = 4, x1 x2 = − a be a point on the circle different from both A and B. Line AC intersects l at point D.
3
Points B and E, the latter obtained as the tangency point of a line drawn from D to the
and circle, lie on the same side from AC . Line EB intersects l at point F ; line F A intersects
the circle second time at point G. Prove that the point symmetric to G with respect to
x10 + x20 + x30 = −a, x10 x20 + x20 x30 + x30 x10 = b, x10 x20 x30 = 8. AB lies on F C .
Now Solution. See IMO-2004 Shortlist.
2 2
4 x1 + x2 2
4= = > x1 x2 = − a
2 2 3
and
2 2 p
− a = (x10 + x20 + x30 ) > 2 3 x10 x20 x30 = 4.
3 3
19 20
WE THANK:
2005/2006
Tallinn City Government
Tartu City Government
Narva City Government
Tartu 2006
D N C
Selected Problems from Open Contests
A M B
1. A farmer noticed that, during the last year, there were exactly as many calves born
as during the two preceding years together. Even better, the number of pigs born during Figure 1
the last year was one larger than the number of pigs born during the two preceding
years together. The farmer promised that if such a trend will continue then, after some
years, at least twice as many pigs as calves will be born in his cattle, even though this far divisible by 3, a2004 is divisible by 3 and so is a. But then a2004 is also divisible by 9
this target has not yet ever been reached. Will the farmer be able to keep his promise? making the sum of its digits divisible by 9. Since 2004 is not divisible by 9, the sum of
(Juniors.) the digits of a2004 can not be equal to 2004, a contradiction.
b) Assume there exists such a number a. We will use the fact that a natural number and
Answer: no.
its sum of digits give the same remainder when divided by 3. It follows that a2006 ≡
Solution. Let Fn and Gn be the numbers of calves and pigs born during the nth year,
2006 ≡ 2 mod 3. On the other hand, the number a2006 = (a1003 )2 is a perfect square that
respectively. We will prove that if the farmer’s promise has been false during the pre-
can not give the remainder 2 when divided by 3, a contradiction.
vious years, it will remain so after the nth year as well. From the problem statement,
we have Fn = Fn−1 + Fn−2 and Gn = Gn−1 + Gn−2 + 1. If the number of born pigs
was less than twice the number of born calves during the previous years, we must have 4. A 9 × 9 square is divided into unit squares. Is it possible to fill
Gn−1 6 2Fn−1 − 1 and Gn−2 6 2Fn−2 − 1. Consequently, each unit square with a number 1, 2, . . . , 9 in such a way that, when-
ever one places the tile so that it fully covers nine unit squares, the
Gn = Gn−1 + Gn−2 + 1 6 2Fn−1 − 1 + 2Fn−2 − 1 + 1 = tile will cover nine different numbers? (Juniors.)
= 2( Fn−1 + Fn−2 ) − 1 = 2Fn − 1
Answer: no.
or Gn 6 2Fn − 1, i.e. the promise will not be true during the nth year. Since the farmer Solution 1. Assume that the numbers can be written in the required way. Put the tile over
has kept the cattle for at least two years, we can claim by induction that the promise will the central square; w.l.o.g. we can assume that the numbers are placed like in Figure 2,
never come true. left. Next move the tile like in Figure 2, middle. Two upper left vacant squares can have
neither 6 nor 7. Thus we must have 8 and 9 there, in some order. Now place the tile like
2. Let ABCD be a parallelogram, M the midpoint of AB and N the intersection of CD in Figure 2, right. We can see that either way we must cover number 8 twice, hence the
and the angle bisector of ABC. Prove that CM and BN are perpendicular iff AN is the required configuration of numbers does not exist.
angle bisector of DAB. (Juniors.)
8, 9
Solution. The triangle CNB is isosceles since ∠CNB = ∠ MBN = ∠CBN (see Figure 1). 8, 9
Thus we have | NC| = | BC|. 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
Assume first CM ⊥ BN. Since BN is both bisector and altitude for triangle BMC, we 6 8 6 8 6 8
7 9 7 9 7 9
have | BM| = | BC|. Consequently | BM| = |CN |, implying that N is the midpoint of CD
and MN is a segment joining the midpoints of the sides of the parallelogram. Then we
must have | AM| = | DN | = | NC| and | AD | = | MN | = | BC|. Thus the sides of AMND Figure 2
are equal and we have a rhombus. Its diagonal AN bisects DAM.
Assume now that AN bisects DAB. Then ∠ DN A = ∠ BAN = ∠ DAN, which implies Solution 2. Put the tile over the central square; w.l.o.g. we can assume that the numbers
| DN | = | DA|. On the other hand, | DA| = |CB| = | NC|. Thus N is the midpoint of CD. are written like in Figure 3, left. We will analyse which number can be written into the
Since M is the midpoint of AB, we have that MBCN is a rhombus with the diagonals gray central square. Moving the tile one position left, we see that the central square can
CM and BN being perpendicular. not contain 1, 2, 3 or 4. Moving the tile one position right, we see that the central square
can not contain 5. Moving the tile one position down, we see that the central square can
3. Does there exist a natural number with the sum of digits of its kth power being not contain 6, 7, 8 or 9. Thus the required numbering is not possible.
equal to k, if a) k = 2004; b) k = 2006? (Juniors.)
5. Find all real numbers with the following property: the difference of its cube and
Answer: a) no; b) no. its square is equal to the square of the difference of its square and the number itself.
Solution. a) Assume there exists such a number a. Since the sum of digits of a2004 is (Juniors.)
1 2
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 A′ D′
6 8 6 8 6 8 6 8
7 9 7 9 7 9 7 9
A D
Figure 3 P
C
B
C′
B′
Figure 5
Figure 4 Solution 2. Independently of the location of P, the equalities
| PA| · | PA′ | = | PC| · | PC′ |, | PB| · | PB′ | = | PD | · | PD ′ |
Answer: 0, 1 and 2.
Solution 1. Let x be a number with this property. Then x3 − x2 = ( x2 − x )2 , which are valid. Multiplying these, we get
leads to the equation x4 − 3x3 + 2x2 = 0 or x2 ( x2 − 3x + 2) = 0. Hence x2 = 0 or
x2 − 3x + 2 = 0. Solving these quadratic equations, we get x0 = 0, x1 = 1, x2 = 2. | PA| · | PB| · | PA′ | · | PB′ | = | PC| · | PD | · | PC′ | · | PD ′ |.
Solution 2. Transform the equation x3 − x2 = ( x2 − x )2 to obtain x2 ( x − 1) = x2 ( x − 1)2 . Points A, B, C, D are concyclic if and only if
Therefore x2 ( x − 1)2 − x2 ( x − 1) = 0 or x2 ( x − 1)( x − 2) = 0. Clearly, the solutions of
the last equation are 0, 1, 2. | PA| · | PB| = | PC| · | PD |
6. A solid figure consisting of unit cubes is shown in the picture. Is it or, taking into account the previous equality, if and only if
possible to exactly fill a cube with these figures if the side length of the
| PA′ | · | PB′ | = | PC′ | · | PD ′ |,
cube is a) 15; b) 30? (Juniors.)
Answer: a) no; b) yes. which holds if and only if A′ , B′ , C′ , D ′ are concyclic.
Solution. a) Since the figure consists of four unit cubes, the number of unit cubes in every Note. As can be seen from solution 2, the assertion of the problem holds regardless of
solid composable from these figures is divisible by 4. Since the cube with side length 15 positions of the circles.
contains an odd number of unit cubes, this cube is not among these solids.
8. A computer outputs the values of the expression (n + 1) · 2n for n = 1, n = 2, n = 3,
b) From two figures, it is possible to assemble a cube with side length 2 (see Figure 4).
etc. What is the largest number of consecutive values that are perfect squares? (Juniors.)
From these cubes, it is possible to assemble a cube with side length 30.
Answer: 2.
7. Two non-intersecting circles, not lying inside each other, are drawn in the plane. Solution. Two consecutive values can be perfect squares, for example, for n = 7 and
Two lines pass through a point P which lies outside each circle. The first line intersects n = 8 we get 8 · 27 = (25 )2 and 9 · 28 = (3 · 24 )2 .
the first circle at A and A′ and the second circle at B and B′ ; here A and B are closer
Now prove that three consecutive values cannot be perfect squares. Assume that (n +
to P than A′ and B′ , respectively, and P lies on segment AB. Analogously, the second
1) · 2n and (n + 3) · 2n+2 are both perfect squares. If n is even then both 2n and 2n+2
line intersects the first circle at C and C′ and the second circle at D and D ′ . Prove that
are perfect squares. Therefore also n + 1 and n + 3 must be perfect squares, which is
the points A, B, C, D are concyclic if and only if the points A′ , B′ , C′ , D ′ are concyclic.
impossible. If n is odd, i.e. n = 2k + 1 for some k > 0, then (n + 1) · 2n = (2k + 2) ·
(Juniors.)
22k+1 = (k + 1) · 22k+2 and (n + 3) · 2n+2 = (2k + 4) · 22k+3 = (k + 2) · 22k+4 . Here, the
Solution 1. Since A, A′ , C′ , C are concyclic (see Figure 5), we have ∠ AA′ C′ + ∠ ACC′ = factors 22k+2 and 22k+4 are perfect squares, therefore also k + 1 and k + 2 must be perfect
180◦ , hence ∠ B′ A′ C′ = ∠ ACD. Analogously ∠C′ D ′ B′ = ∠ DBA. Points A, B, C, D are squares, which is impossible for non-negative k.
concyclic if and only if ∠ ACD = ∠ DBA, which is equivalent to ∠ B′ A′ C′ = ∠C′ D ′ B′ ,
the last equality holds if and only if points A′ , B′ , C′ , D ′ are concyclic. 9. Let a, b, c be positive integers. Prove that the inequality
( x − y ) a ( x − z)b (y − z)c > 0
3 4
starts all over. Find all possible values for p1 and q1 such that the teacher can hold Juku
at school forever. (Seniors.)
Answer: either p1 is any integer and q1 = 0 or p1 = 1 and q1 = −2.
Solution. If q1 = 0 then we have the equation x2 + p1 x = 0 with solutions − p1 and 0.
The teacher can write another equation x2 − p1 x = 0 with solutions p1 ja 0, then again
Figure 6 Figure 7 x2 + p1 x = 0, etc. Thus all pairs ( p1 , 0) satisfy the conditions of the problem.
If q1 = −1 then the product of the solutions must be −1 and the solutions −1 and 1
in some order. Since the equations x2 − x + 1 = 0 and x2 + x − 1 = 0 have no integer
holds for all reals x, y, z if and only if a, b, c are even. (Juniors.) solutions, no pairs of the form ( p1 , −1) satisfy the conditions of the problem.
Solution. If a, b, c are positive then the inequality holds. On the other hand, assume that If q1 = −2 then the product of the solutions must be −2 and the solutions are either
the inequality holds for all reals x, y, z. Choosing z < x < y and dividing the given 2 and −1 or 1 and −2. In the first case, the teacher can choose between the equations
inequality by a positive number ( x − z)b (y − z)c , we get the inequality ( x − y)a > 0, x2 + 2x − 1 = 0 and x2 − x + 2 = 0, none of them having integer solutions. In the second
from which we conclude due to x − y < 0 that a is even. Analogously, choosing y < case, the teacher can write the equation x2 + x − 2 = 0 with solutions 1 and −2. Thus
z < x, we conclude that c is even. Finally, choosing x < y < z, we get after dividing we see that the pair (1, −2) staisfies the conditions of the problem.
the inequality by a positive number ( x − y)a (y − z)c that ( x − z)b > 0 from which we Now let q1 be an integer not equal to 0, −1 nor −2. If x1 and x2 are the solutions of
conclude due to x − z < 0 that b is even. x2 + p1 x + q1 = 0, Viète formulae imply that x1 + x2 = − p1 , x1 x2 = q1 and
10. All the streets in a city run in one of two perpendicular directions, forming unit x12 + x22 = ( x1 + x2 )2 − 2x1 x2 = p12 − 2q1 < p12 + q12 .
squares. Organizers of a car race want to mark down a closed race track in the city in Thus the sum of squares of the coefficients of the equation strictly decreases for q1 6∈
such a way that it would not go through any of the crossings twice and that the track [−2; 0]. Since sum of squares is non-negative, we must sooner or later reach one of the
would turn 90◦ right or left at every crossing. Find all possible values of the length of two situations: the solutions are not integers or the constant term belongs to the interval
the track. (Seniors.) [−2; 0]. The latter case is impossible, since every equation x2 + px + q = 0 uniquely
Answer: all positive integers divisible by 4, except for 8. determines its predecessor x2 − ( p + q) x + pq = 0, hence the pairs ( p1 , 0) ja (1, −2) can
Solution. Define natural coordinates with the origin at some crossing and consider two only arise from equations with constant terms 0 or −2, respectively. Thus there are no
consecutive track fragments of length 1. One of them is parallel to x-axis and the other other pairs of numbers satisfying the conditions of the problem.
one to y-axis; moving along the first one, parity of the x-coordinate changes, and mov-
ing along the second one, parity of the y-coordinate changes. Moving along the track, 12. Let ABC be an acute triangle and choose points A1 , B1 and C1 on sides BC, CA
parities of x- and y-coordinates change alternately, but when returning to the initial and AB, respectively. Prove that if the quadrilaterals ABA1 B1 , BCB1 C1 and CAC1 A1 are
point, both parities must be the same as in the beginning. Since the pairs of parities cyclic then their circumcentres lie on the sides of ABC. (Seniors.)
repeat after every four track fragments, the length of the track must be divisible by 4. Solution. Since BCB1 C1 is cyclic (see Figure 8), we have ∠ BB1 C = ∠ BC1 C = α. Similar-
There exists a suitable track of length 4 going around one block. There is no track of ily, let ∠CC1 A = ∠CA1 A = β and ∠ AA1 B = ∠ AB1 B = γ. Considering the angles with
length 8. If there were one, it would have four fragments parallel to x-axis and four vertices at points A1 , B1 and C1 , we get the following system of equations:
fragments parallel to y-axis. Hence, we could not deviate more than 2 units in either 
direction and the whole track should fit into 2 × 2 square. It is easy to see that the track  β + γ = 180◦
can not contain three corners of the square, but then there will be less that 8 possible γ + α = 180◦

turning points left. α + β = 180◦ .
All the other positive integers divisible by 4 are attainable. Figure 6 shows a track of Adding the equalities and dividing by 2 gives α + β + γ = 270◦ , implying α = β = γ =
length 12 and we can increase this length repeatedly by 4 units using the operation in 90◦ . Thus the segments BC, CA and AB are the diameters of the circles and contain their
Figure 7. circumcentres.
11. After the schoolday is over, Juku must attend an extra math class. The teacher s1
13. Martin invented the following algorithm. Let two irreducible fractions and
writes a quadratic equation x2 + p1 x + q1 = 0 with integer coefficients on the blackboard t1
and Juku has to find its solutions. If they are not both integers, Juku may go home. If the s2
be given as inputs, with the numerators and denominators being positive integers.
solutions are integers, then the teacher writes a new equation x2 + p2 x + q2 = 0, where t2
Divide s1 and s2 by their greatest common divisor c and obtain a1 and a2 , respectively.
p2 and q2 are the solutions of the previous equation taken in some order, and everything
Similarily, divide t1 and t2 by their greatest common divisor d and obtain b1 and b2 ,
5 6
A smallest on the board and strictly less than the other number in the pair. The situation
with all numbers being equal to zero can this way only occur after B’s move.
B1
Now let m be odd and n even. If none of the numbers is yet zero, B can ensure that after
C1
his move the following conditions hold: the smallest number on the board is even, the
quantity of the smallest numbers is odd and the number of occurrencies of every other
number is even. Indeed, if A changes the smallest number then B can change it again,
but if A changes some other number then B can change another number equal to the
B A1 C
one A changed. It goes on until some number becomes zero, afterwards B can divide
all the remaining numbers into pairs and use the strategy described above.
Finally, let m and n be odd. Then after A’s first move, there is a position described in
the previous paragraph on the board. Thus A can use the strategy of B and win.
Figure 8 15. Kati cut two equal regular n-gons out of paper. To the vertices of both n-gons, she
wrote the numbers 1 to n in some order. Then she stabbed a needle through the centres
a1 b2 + a2 b1 of these n-gons so that they could be rotated with respect to each other. Kati noticed
respectively. After that, form a new fraction , reduce it, and multiply the that there is a position where the numbers at each pair of aligned vertices are different.
t1 b2
numerator of the result by c. Martin claims that this algorithm always finds the sum of Prove that the n-gons can be rotated to a position where at least two pairs of aligned
the original fractions as an irreducible fraction. Is his claim correct? (Seniors.) vertices contain equal numbers. (Seniors.)
Answer: yes. Solution. Assume that the lower n-gon is fixed and move the upper n-gon. Let the initial
Solution. Since position of n-gons be the one found by Kati. For each vertex of the upper n-gon, there
is an angle by which rotating clockwise the upper n-gon brings this vertex atop of the
s 1 t2 s 2 t1 vertex of the lower n-gon with the same number. There are n different vertices, but only
c· · + ·
c · (a1 b2 + a2 b1 ) c d c d s 1 t2 + s 2 t1 s s n − 1 different rotation angles since the angle 0◦ is excluded by conditions. Hence for
= = = 1 + 2,
t1 b2 t2 t1 t2 t1 t2 two vertices of the upper n-gon, the rotation angles are equal.
t1 ·
d
16. A real-valued function f satisfies for all reals x and y the equality
the resulting fraction has correct value. We still need to prove that it is irreducible. For
that, it is enough to show that the numbers c and t1 b2 are relatively prime. f ( xy) = f ( x )y + x f (y).
Suppose there exists a prime p dividing both c and t1 b2 . Since c = gcd(s1 , s2 ), we have
Prove that this function satisfies for all reals x and y 6= 0 the equality
that p also divides both s1 and s2 . Consequently, t1 and t2 can not be divisible by p,
s1 s2
because the fractions and are irreducible. Thus p does not divide t1 t2 , implying x f ( x )y − x f (y )
t1 t2 f = .
y y2
t2
that p can not divide t1 b2 = t1 · either, a contradiction.
d (Seniors.)
f ( xy) − x f (y)
14. Two players A and B play the following game. Initially, there are m equal positive Solution. From the given expression we obtain f ( x ) = , this equality
integers n written on a blackboard. A begins and the players move alternately. The y
x
player to move chooses one of the non-zero numbers on the board. If this number k is holds for any reals x and y 6= 0. Taking at place of x, we get
y
the smallest among all positive integers on the board, the player replaces it with k − 1;
if not, the player replaces it with the smallest positive number on the board. The player x x
f · y − f (y )
who first turns all the numbers into zeroes, wins. Who wins if both players use their x y y f ( x )y − x f (y )
best strategies? (Seniors.) f = = .
y y y2
Answer: A wins if mn is odd; B wins if mn is even. Note. It is possible to prove (for example, using reduction to the Cauchy equation), that
Solution. If the quantity m of numbers is even then B has the following winning strategy. the only continuous functions satisfying the given conditions are
B divides all the numbers into pairs and if A makes a move and changes some number, (
B changes the other number in the pair, ensuring that after his move all pairs contain ax ln | x |, if x 6= 0,
f (x) =
equal numbers. This is possible, since after A’s move the number A wrote must be the 0, if x = 0.
7 8
F
is 7 − 1 · 3 = 4. This means that the term a361 is divisible by both 17 and 118, i.e. it is
divisible by 2006.
Note 1. Since 17 and 118 are relatively prime, the existence of the suitable n follows from
O
the Chinese Remainder Theorem: the remainders of n upon division by 17 and 118 must
be 4 and 7, respectively.
E D Note 2. The least number satisfying the given conditions corresponds to n = 87, in this
A
case a87 = 627878 = 313 · 2006.
B C 19. Let n > 2 be a fixed integer and let ai,j (1 6 i < j 6 n) be some positive integers.
For a sequence x1 , . . . , xn of reals, let K ( x1 , . . . , xn ) be the product of all expressions
G ( xi − x j )ai,j where 1 6 i < j 6 n. Prove that if the inequality K ( x1 , . . . , xn ) > 0 holds
independently of the choice of the sequence x1 , . . . , xn then all integers ai,j are even.
(Seniors.)
Figure 9 Solution 1. Suppose the contrary: some of the numbers ai,j are odd. Let l be the smallest
index for which there are odd numbers among the numbers ai,l (1 6 i < l); also let k
17. Four points A, B, C, D are chosen on a circle in such a way that arcs AB, BC, and be the largest index for which ak,l is odd. Then ak,l is the only odd number among the
CD are of the same length and the arc DA is longer than these three. Line AD and the numbers ai,j where k 6 i < j 6 l. Now choose x1 , . . . , xn as follows:
line tangent to the circle at B intersect at E. Let F be the other endpoint of the diameter x1 > x2 > . . . > x k −1 > x l > x k +1 > . . . > x l −1 > x k > x l +1 > . . . > x n ;
starting at C of the circle. Prove that triangle DEF is equilateral. (Seniors.)
i.e., choose some n numbers in decreasing order and swap the positions of the k-th and
Solution 1. Let O be the centre of the circle and let G be the point where BE intersects
the l-th number. Then in the given expression, the factors ( xi − x j )ai,j , where i < k or
the tangent drawn from D to the circle. Since the total length of arcs DA and AB is
j > l, are positive, since the bases of the power are positive. All the remaining factors
larger than the total length of arcs BC and CD, the points G and E lie on different sides
( xi − x j )ai,j where k 6 i < j 6 l have even exponents with the exception of ( xk − xl )ak,l ,
from B (see Figure 9). Points B and D are symmetric with respect to the line CF, hence
1 which has negative base and odd exponent. So the whole product is negative.
G lies on the line CF and ∠ BGF = ∠ DGF. Now ∠ BED = (∠ BOD − ∠ AOB) = Solution 2. Assume that ak,l is odd for some indices k and l. Fix x1 , . . . , xk−1 , xk+1 , . . . , xn
2
1 in such a way that they are pairwise different and consider the product K ( x1 , . . . , xn ) as
∠COD = ∠CFD, therefore ∠GED = ∠GFD. Consequently G, E, F, D are concyclic. a polynomial of one variable K ( xk ). The root xl of this polynomial has odd multiplicity.
2
Since ∠EGF = ∠ DGF, chords EF and DF of the circumcircle of quadrangle GEFD are Therefore the graph of K ( xk ) intersects the x-axis at xl and we can choose xk such that
equal. K ( xk ) is negative, a contradiction.
Solution 2. Since the lengths of the arcs BA and BC are equal, the lines EB and AC
are parallel. Therefore ∠ BED = ∠CAD = ∠ BDA. Thus triangle BED is equilateral.
Further, ∠CBF = 90◦ . By symmetry, the lines BC and AD are parallel, hence BF is Selected Problems from the Final Round of National
perpendicular to AD. Segment BF is the altitude of the equilateral triangle BED, it Olympiad
bisects its base ED. This segment is also the altitude of the triangle DEF and it bisects
its base. This is possible only when the triangle DEF is equilateral.
Note. The assertion of the problem holds also in the case when the arc DA is shorter 1. Find all pairs of positive integers (a, b) such that
than the other three, the point G then lies on the other side.
ab = gcd(a, b) + lcm(a, b).
18. In the sequence (an ) with general term an = n3 − (2n + 1)2 , does there exist a term
that is divisible by 2006? (Seniors.) (Grade 9.)
Answer: yes. Answer: (2, 2).
Solution. First, a4 = 43 − 92 = −17 and a7 = 73 − 152 = 118. Since n3 − (2n + 1)2 Solution. As the left-hand side and summand lcm(a, b) on the right-hand side are both
is a polynomial, a4+17k is divisible by 17 and a7+118l is divisible by 118 for all natural divisible by a, also gcd(a, b) has to be divisible by a. On the other hand, gcd(a, b) 6 a as
numbers k and l. Since 119 is divisible by 17, adding 118 decreases the remainder by 1 a is positive. Thus gcd(a, b) = a. Analogously we obtain that gcd(a, b) = b. Therefore
on division by 17. Therefore, when 361 = 7 + 118 · 3 is divided by 17, the remainder a = b and the equation has the form a2 = a + a or a(a − 2) = 0. The only positive
solution of the equation is a = 2 and thus also b = 2.
9 10
C
√ 2 2
3c 1 3c c 2 3c c
h: = tan 30◦ = √ , therefore h = . Hence b2 = h2 + = + = c2 ,
2 3 2 2 4 4
a
b b
a giving b = c. We now have that CDE is an equilateral triangle and its angles are 60◦ .
M h |CH | 1 | DH | 1
Solution 3. As = sin 30◦ = and = , segment CD satisfies the Angle
30 ◦ 30 ◦ |CA| 2 | DA| 2
A c D c H c E c B Bisector Property. Hence CD bisects angle ACH whose size is 60◦ and ∠ DCH = 30◦ .
2 2 Then the vertex angle of the isosceles triangle CDE is 60◦ and the base angles are also
60◦ .
Solution 4. Let the medians of triangle CDE meet at M. By the Ray Property, | DM| =
C′ 1 1 1
| AC| = | BC| = | EM|. We obtain |CH | = | AC| as in the previous solution and thus
3 3 2
Figure 10 1
|CM| = | AC|. So |CM| = | DM| = | EM|, i.e., the intersection point M of medians of
3
CDE is also the circumcentre of CDE. It follows that CDE is an equilateral triangle and
Solution 2. It is known that ab = gcd(a, b) lcm (a, b) for positive integers a and b. Thus all its angles are of size 60◦ .
we obtain the equality gcd(a, b) lcm (a, b) = gcd(a, b) + lcm(a, b) which is equivalent to Solution 5. Assume that ∠ DCE > 60◦ . Then ∠CED < 60◦ . Hence | DE| > |CD | giving
(gcd(a, b) − 1) (lcm(a, b) − 1) = 1. The last equality expresses 1 as a product of two non- also | AD | > |CD |. Then in triangle ACD, we have ∠ ACD > 30◦ , and in triangle
negative integers; this is only possible if both of them are equal to 1. Thus gcd(a, b) − BCE by symmetry, ∠ BCE > 30◦ . Adding the inequalities, we get ∠ ACB > 120◦ —
1 = 1 and lcm(a, b) − 1 = 1 or gcd(a, b) = 2 and lcm(a, b) = 2. Hence a = b = 2. a contradiction. Analogously we obtain a contradiction by starting with assumption
Solution 3. Let gcd(a, b) = d. We represent integers a and b as a = da′ and b = db′ ∠DCE < 60◦ . Hence ∠DCE = 60◦ and thus ∠CDE = ∠CED = 60◦ .
where a′ and b′ are relatively prime. Then lcm(a, b) = da′ b′ . The equation gets the
1 4. Consider a rectangular grid of 10 × 10 unit squares. We call a ship a figure made
form d2 a′ b′ = d + da′ b′ , or a′ b′ = . As a′ b′ is an integer, we must have d − 1 = 1
d−1 up of unit squares connected by common edges. We call a fleet a set of ships where no
and a′ b′ = 1. Hence d = 2 and a′ = b′ = 1. Therefore the solution of the equation is two ships contain squares that share a common vertex (i.e. all ships are vertex-disjoint).
a = da′ = 2 and b = db′ = 2. Find the least number of squares in a fleet to which no new ship can be added. (Grade
9.)
2. Let there be n > 2 real numbers such that none of them is greater than the arith-
metic mean of the other numbers. Prove that all the numbers are equal. (Grade 9.) Answer: 16.
Solution. Call a fleet full if no new ships can be added. We have to find the least number
Solution. Let a be the greatest among the numbers. Suppose, by contradiction, that not
of squares in a full fleet.
all the numbers are equal. Then there must be some numbers less than a. Considering
First we show that a full fleet covering 16 unit squares exists. Put on the grid 16 one-
numbers other than a, we therefore know that their arithmetic mean is less than a. But
square ships as shown in Figure 11. Note that then each square of the grid has a common
this contradicts the conditions of the problem. Hence all the numbers are equal.
vertex with one of those ships and thus no ship can be added.
3. Triangle ABC is isosceles with AC = BC and ∠C = 120◦ . Points D and E are Second we prove that there can not be fewer than 16 unit squares in a full fleet. Suppose
chosen on segment AB so that | AD | = | DE| = | EB|. Find the sizes of the angles of a full fleet is fixed. Consider the set of 16 unit squares painted gray in Figure 11. For
triangle CDE. (Grade 9.) each of these 16 squares, there is a square of the full fleet that shares (at least) a common
Answer: all the angles are 60◦ .
Solution 1. The base angle of the isosceles triangle ABC is (180◦ − 120◦ ) : 2 = 30◦ . Let
H be the foot of the altitude drawn from vertex C (see Figure 10). Reflect the triangle
ABC with respect to side AB, the point C going to C′ . As ∠CAC′ = 60◦ and | AC| =
| AC′ |, triangle ACC′ is equilateral and AH is its median. Moreover, point D divides the
median with ratio 2 : 1. Thus the medians of ACC′ meet at D, and CD is both a median
and an angle bisector to ACC′ . We obtain ∠ DCH = 30◦ and ∠ DCE = 60◦ . As CDE is
isosceles, ∠CDE = ∠CED = 60◦ .
Solution 2. Denote |CA| = |CB| = a, |CD | = |CE| = b and | AD | = | DE| = | EB| = c for
brevity. Let h be the altitude drawn from vertex C to side AB. In triangle CH A, we have Figure 11
11 12
diameter is also an integer. Find all possible side lengths of the triangle. (Grade 11.)
Answer: 3, 4 and 5.
Solution. Let p and m be the lengths of the sides of the triangle and let n be the length of
the hypotenuse, where p is a prime (Figure 14). Then p2 + m2 = n2 , implying p2 = (n −
m)(n + m). Since p is prime, the terms of the right-hand side must satisfy n − m = 1,
n + m = p2 .
The perimeter of the triangle is p + m + n. In order to find the incircle diameter d, we
note that the total length of the segments tangent to the incircle, originating from the
Figure 12 Figure 13 vertex of the right angle, equals the length of the incircle diameter, whereas the total
length of the four tangents originating from the other two vertices is 2n. Thus, d + 2n =
p + m + n. According to the assumption, p + m + n is divisible by d = p + m − n.
vertex with it. All these squares of the fleet must be different. Hence there are at least Substituting m + n and m − n from above, we get that p + p2 = p( p + 1) is divisible by
16 squares in the fleet. p − 1. Since p and p − 1 are coprime, it must be that p + 1 is divisible by p − 1. Hence,
p − 1 = 1 or p − 1 = 2. We see that p = 2 is impossible, since n − m and n + m cannot
5. Consider a rectangular grid of 10 × 10 unit squares. We call a ship a figure made
be of different parity. Thus, p = 3, m = 4 and n = 5. A triangle with side lengths 3, 4, 5
up of unit squares connected by common edges. We call a fleet a set of ships where no
is clearly a right triangle.
two ships contain squares that share a common vertex (i.e. all ships are vertex-disjoint).
Find the greatest natural number that, for each its representation as a sum of positive
integers, there exists a fleet such that the summands are exactly the numbers of squares
contained in individual ships. (Grade 10.) p m
Answer: 25.
Solution. First we prove that, for all n > 25, we can divide n into summands so that a
n
fleet with respective ship sizes can not be put on the grid. In particular, we prove that
one can not put more than 25 ships of size 1 on the grid. Let us divide the grid into Figure 14
squares of 2 × 2 (as in Figure 12), there are 25 of them. As each 2 × 2 square can contain
at most one ship of size 1, then the total number of such ships is at most 25.
Second we show that, for any representation of 25 as a sum of positive integers, there is a 8. The sequence ( Fn ) of Fibonacci numbers satisfies F1 = 1, F2 = 1 and Fn = Fn−1 +
fleet with respective ship sizes. Let us initially put 25 ships of size 1 on the grid as shown Fn−2 for all n > 3. Find all pairs of positive integers (m, n), such that Fm · Fn = mn.
in Figure 13. Then, starting from the upper left corner, shift along the line together (Grade 11.)
as many ships as the first summand of the representation tells; then shift together as
Answer: (1, 1), (1, 5), (4, 6), (5, 1), (5, 5) and (6, 4).
many ships as the second summand tells etc. The fleet obtained this way satisfies the
conditions of the problem. Solution. By induction on n, it is easy to prove that Fn > n for all n > 6 and Fn > 2n for
all n > 8. Thus, if m > 6 and n > 6, then Fm · Fn > m · n. W.l.o.g., we may now assume
6. Find the greatest possible value of sin(cos x ) + cos(sin x ) and determine all real m 6 5 (the remaining solutions can be obtained by exchanging m and n).
numbers x, for which this value is achieved. (Grade 11.)
• If m = 1, then 1 · Fn = 1 · n, or Fn = n. From above, the only solutions are n = 1
Answer: the greatest possible value is sin 1 + 1, which is achieved iff x = 2kπ, where k and n = 5 and the suitable pairs are (1, 1), (1, 5), (5, 1) and (5, 5).
is an arbitrary integer.
Solution. Since the value of cos x is in the interval [−1; 1] and since sin x is increasing • If m = 2, we get 1 · Fn = 2 · n, or Fn = 2n. Since there are no solutions for n < 8,
in this interval, the greatest possible value of the first addend is sin 1, and the value is there are no solutions at all.
achieved iff cos x = 1, or x = 2kπ, where k is an arbitrary integer. The greatest possible 3 3
• If m = 3, we get 2 · Fn = 3 · n, or Fn = n. Since n < 2n, there are no solutions, as
value of the second addend is 1, which is achieved iff sin x = 0, or x = lπ, where l is 2 2
an arbitrary integer. Both terms achieve the maximal value simultaneously iff x = 2kπ, in the previous case.
where k is an arbitrary integer; the value of the expression is then sin 1 + 1. 4
• If m = 4, then 3 · Fn = 4 · n, or Fn = n. Here, the only solution is n = 6, giving
7. In a right triangle, the length of one side is a prime and the lengths of the other 3
pairs (4, 6) and (6, 4).
side and the hypotenuse are integral. The ratio of the triangle perimeter and the incircle
13 14
• If m = 5, then 5 · Fn = 5 · n, or Fn = n; this case is analysed above. Solution. First, we prove that a suitable sequence of moves exists for even n. Divide the
board into blocks of 2 × 2 squares (see Figure 16) and place the pawn on the upper left
corner square. To move through the first block, take the following moves: down-right,
9. In a triangle ABC with circumcentre O and centroid M, lines OM and AM are
up, down-left, down. Repeat the same combination of moves until the pawn reaches the
perpendicular. Let AM intersect the circumcircle of ABC again at A′ . Let lines BA′ and
bottom-most block in a column of blocks. In the bottom-most block, move down-right,
AC intersect at D and let lines CA′ and AB intersect at E. Prove that the circumcentre
left, up-right, and right; the first column of blocks is passed. In the bottom-most block
of triangle ADE lies on the circumcircle of ABC. (Grade 11.)
of the second column, move down-right, left, up-right, up, and continue by moving
upwards block by block. By passing the columns alternatingly up and down, the pawn
A
visits each square exactly once.
H G
O
M
B C
F
A′
Figure 16 Figure 17
D
E
Next, we prove that a sequence does not exist for odd n. Colour the second, fourth,
Figure 15 n2 − n
sixth, etc. row dark (see Figure 17), then there are dark squares. Note that every
2
Solution. Let F, G and H be the base points of the medians drawn from vertices A, B diagonal move starts from a dark square or ends on a dark square. Since a suitable
and C, respectively (see Figure 15). Then, triangle A′ OA is isosceles with height OM sequence does not visit a square twice and does not contain two consecutive diagonal
and | A′ M| = | MA|. Since the centroid divides a median in ratio 2 : 1, we get | FM| = moves, each diagonal move corresponds to a different dark square. Thus, we can make
1 n2 − n n2 − n
| MA|, and | A′ F| = | FM|. On the other hand, | BF| = | FC|. Hence, A′ BMC is a at most diagonal moves and, consequently, at most + 1 non-diagonal
2 2 2
parallelogram. Parallel sides then imply that triangles ABD and AHC are similar with moves, or n2 − n + 1 moves altogether. For n > 3, this number is smaller than n2 − 1,
similarity ratio 2 — the ratio of the lengths of AB and AH. Analogously, triangles ACE the number of moves required to visit all squares.
and AGB are similar with the same ratio. Homothecy with centre A and ratio 2 brings
triangle ABC to triangle AED, while the circumcentre O of triangle ABC is transformed 11. We call a ship a figure made up of unit squares connected by common edges.
to the second intersection point of AO and the said circumcircle. Prove that if there is an odd number of possible different ships consisting of n unit
Note. The use of homothecy can be avoided by finding the second intersection point squares on a 10 × 10 board, then n is divisible by 4. (Grade 12.)
P of line AO and the circle and by proving that | AP| = 2| AO|, | DP| = 2|CO| and Solution. Let n be such that the number of possible different ships of n squares is odd.
| EP| = 2| BO|. Divide all ships in classes, such that all ships in the same class are precisely those that
can be obtained from one another by shifts, vertical and horizontal reflections. Then
10. A pawn is placed on a square of a n × n board. There are two types of legal
there must exist a class with an odd number of ships.
moves: (a) the pawn can be moved to a neighbouring square, which shares a common
side with the current square; or (b) the pawn can be moved to a neighbouring square, Let L be a ship in such a class. Assume that L is not symmetrical w.r.t. either the vertical
which shares a common vertex, but not a common side with the current square. Any or horizontal axis of symmetry of its tight bounding box. Then no ship in this class is
two consecutive moves must be of different type. Find all integers n > 2, for which it is symmetrical w.r.t. this axis. Thus, we can divide all ships in this class into pairs: a ship
possible to choose an initial square and a sequence of moves such that the pawn visits and its reflection from this axis; a contradiction with the odd cardinality of this class.
each square exactly once (it is not required that the pawn returns to the initial square). Therefore, L must be symmetrical w.r.t. both the vertical and the horizontal axis.
(Grade 11.) The side lengths of the rectangle bounding L must be even, for there is an even number
of ways to place a rectangle with an odd side on a board with an even side length
Answer: n = 2k, where k is an arbitrary positive integer. 10; again a contradiction in parity. Thus, the vertical and horizontal axes of symmetry
15 16
divide the squares of L into four disjoint reflection symmetrical sets. Since there is an b) Assume that for some n, there exist n suitable integers a1 < a2 < . . . < an . Then, on
equal number of squares in each set, the total number of squares of the ship is divisible the one hand,
by 4.
a1 + a2 + . . . + a n − 2 < a n − 2 + a n − 2 + . . . + a n − 2 = ( n − 2 ) a n − 2
12. Find the smallest possible distance of points P and Q on a xy-plane, if P lies on
but, on the other hand,
the line y = x and Q lies on the curve y = 2x . (Grade 12.)
1 + ln ln 2 a n −1 a n > ( n − 1) a n > ( n − 2) a n −2 .
Answer: √ .
2 ln 2 Thus, a1 + a2 + . . . + an−2 < an−1 an , and the sum of a1 , a2 , . . . , an−2 can not be divisible
Solution. We find the minimum point of h( x ) = 2x − x. Since h′ ( x ) = 2x ln 2 − 1, we get by the product of an−1 and an .
2x ln 2 − 1 = 0, and Solution 2. a) Choose arbitrary pairwise distinct numbers b1 , b2 , . . . , bn and let m be
1 ln ln 2 the least common multiple of all sums of (n − 2) terms. For each i = 1, 2, . . . , n, take
2x = and x=− . ai = mbi . Then the product of any two numbers ak al is ak al = (mbk ) · (mbl ), which is
ln 2 ln 2
divisible by the sum of the remaining numbers, since m2 is divisible by this sum.
Since h′ ( x ) is increasing, this is indeed a minimum. The value of h at this point is
14. Let O be the circumcentre of an acute triangle ABC and let A′ , B′ and C′ be the
1 ln ln 2 1 + ln ln 2 circumcentres of triangles BCO, CAO and ABO, respectively. Prove that the area of
h( x ) = + = .
ln 2 ln 2 ln 2 triangle ABC does not exceed the area of triangle A′ B′ C′ . (Grade 12.)
Here 1 + ln ln 2 = ln(e ln 2). Since 2 < e < 4, we get ln(e ln 2) > ln(2 ln 2) = ln ln 4 > Solution 1. First, we prove that for a fixed circumcircle, a triangle with maximal area
> ln ln e = 0. Thus, the value of h at the minimum is positive, so the graph of g( x ) = 2x is equilateral. Assume that a triangle KLM with maximal area has two sides of un-
is always higher than the graph of f ( x ) = x. Consider points A( x, f ( x )) and B( x, g( x )) equal lengths, say, KM and LM. Take a point M ′ on the circumcircle of KLM such that
and let C be the projection of B to the graph of f . Then, triangle ABC is a right isosceles |KM′ | = | LM′ | (Figure 18). Triangles KLM and KLM′ have a common base but the
| AB| h( x ) altitude of the first triangle is smaller, a contradiction.
triangle, since ∠ BAC = 45◦ . Consequently, | BC| = √ = √ and the distance | BC|
2 2
is minimal iff h( x ) is minimal. The sought distance is thus M
M′
M
Q
h( x ) 1 + ln ln 2
√ = √ . M′
2 2 ln 2
log2 e − log2 log2 e
Note. The answer can be expressed in many different ways, e.g., by √ .
2 X
P
13. Prove or disprove the following statements. K L K L L′
a) For every integer n > 3, there exist n pairwise distinct positive integers such that
the product of any two of them is divisible by the sum of the remaining n − 2 num- Figure 18 Figure 19
bers.
Next, we prove that for a fixed incircle, a triangle with minimal area is equilateral.
b) For some integer n > 3, there exist n pairwise distinct positive integers, such that Assume w.l.o.g. that in a triangle KLM with minimal area, ∠KLM > ∠KML. Con-
the sum of any n − 2 of them is divisible by the product of the remaining two num- sider a triangle KL′ M′ , where points L′ and M′ lie on lines KL and KM such that
bers. ∠KL′ M′ = ∠KM′ L′ and L′ M′ is tangent to the incircle (Figure 19). Let X be the in-
tersection point of lines LM and L′ M′ . Draw perpendicular segments LP and MQ from
(Grade 12.)
points L and M to line L′ M′ . Then | XM′ | > | XL′ | and | MQ| > | LP|, since | XQ| > | XP|
Answer: a) true; b) false. and right triangles MQX and LPX are similar. Thus, the area of triangle MM ′ X is
Solution 1. a) Take n numbers (n2 )!, 2(n2 )!, 3(n2 )!, . . . , n(n2 )!. The product of any two greater than the area of triangle LL′ X and consequently, the area of KLM is greater than
of these numbers is divisible by (n2 )!(n2 )!, whereas the sum of the remaining numbers the area of KL′ M′ , a contradiction.
is k(n2 )!, where k is some positive integer smaller than 1 + 2 + . . . + n, which in turn is Now, let R be the circumradius of ABC. Since the sides of triangle A′ B′ C′ are perpen-
smaller than n2 . dicular to OA, OB and OC (see Figure 20) and bisect these segments, point O is the
17 18
A
15. The Ababi alphabet consists of letters A and B, and the words in the Ababi lan-
C′
guage are precisely those that can be formed by the following two rules:
B′
1) A is a word.
2) If s is a word, then s ⊕ s and s ⊕ s̄ are words, where s̄ denotes a word that is obtained
β
O β by replacing all letters A in s with letters B, and vice versa; and x ⊕ y denotes the
αα concatenation of x and y.
The Ululu alphabet consists also of letters A and B and the words in the Ululu language
B C are precisely those that can be formed by the following two rules:
A′
1) A is a word.
Figure 20 2) If s is a word, then s ⊗ s and s ⊗ s̄ are words, where s̄ is defined as above and x ⊗ y
is a word obtained from words x and y of equal length by writing the letters of x
R and y alternatingly, starting from the first letter of x.
incentre of A′ B′ C′ and the inradius is . An equilateral triangle with circumradius R
2 R
and an equilateral triangle with inradius have equal area S by the property of the Prove that the two languages consist of the same words. (Grade 12.)
2
centroid. From above, we now get S ABC 6 S 6 S A′ B′ C ′ . Solution. Since each step doubles the length of a word, both languages contain only
Solution 2. Let R be the circumradius of ABC and let α, β and γ be the angles at the words of length 2n , where each such word has been obtained in exactly n steps.
vertices of the triangle. Then ∠ BOC = 2α, ∠COA = 2β, and ∠ AOB = 2γ. Thus, First, we show that each language contains 2n words that can be obtained in exactly n
steps. Indeed, in 0 steps, we obtain only the word A in both languages. Every k-step
R2 word gives two different (k + 1)-step words and any two different k-step words give
S ABC = SBOC + SCOA + S AOB = (sin 2α + sin 2β + sin 2γ).
2 different (k + 1)-step words, since the initial word is always a part of the new word.
Since triangle BOC is isosceles, the midperpendicular OA′ of side BC is also an an- Thus, the number of k + 1-step words is twice the number of k-step words. Now, it
gle bisector, and ∠ BOA′ = ∠COA′ = α. Similarly, ∠COB′ = ∠ AOB′ = β and suffices to prove that every Ababi word is an Ululu word.
R Any 0-step Ababi word is clearly an Ululu word. Assume that the claim holds for all
∠ AOC′ = ∠BOC′ = γ. Consider triangle B′ OC′ . The height drawn to side B′ C′ is , k-step Ababi words and consider a k + 1-step word t. Then, for some word s in the
2
R R2 Ababi language, t = s ⊕ s or t = s ⊕ s̄. By the induction hypothesis, there must exist a
so | B′ C′ | = (tan β + tan γ) and the area of this triangle is SB′ OC ′ = (tan β + tan γ). sequence of k operations by the Ululu rules that allows to construct the word s from the
2 8
R2 R2 word A.
Analogously, SC ′ OA′ = (tan γ + tan α) and S A′ OB′ = (tan α + tan β). Thus,
8 8
• If the word t is obtained in the Ababi language by the rule t = s ⊕ s, apply the
R2 aforementioned sequence of Ululu rules to the word AA. It is easy to see that after
S A′ B′ C ′ = SB′ OC ′ + SC ′ OA′ + S A′ OC ′ = (tan α + tan β + tan γ).
4 each step, the new word is of the form a ⊕ a, where a is the corresponding inter-
h πi mediate word in the construction process of s, since each Ululu operation has the
Function f ( x ) = tan x − 2 sin 2x is concave in the interval 0; , since the second same effect on the two equal halves of a word.
2
derivative
• If the word t is obtained by the rule t = s ⊕ s̄, apply the aforementioned sequence of
2 sin x
f ′′ ( x ) = + 8 sin 2x Ululu rules to the Ululu word AB. In this case, after each step, the obtained word is
cos3 x a ⊕ ā, where a is the corresponding intermediate word in the construction process
is non-negative in this interval. Jensen’s inequality now gives of s, since if the two halves of a word are “negations” of each other, any Ululu
operation preserves this property.
α+β+γ π 2π
f (α ) + f ( β) + f (γ ) > 3 f = 3 tan − 2 sin = 0, Consequently, t is a word in the Ululu language.
3 3 3
so tan α + tan β + tan γ > 2 sin 2α + 2 sin 2β + 2 sin 2γ. The final equality yields S A′ B′ C ′ >
S ABC .
19 20
WE THANK:
2006/2007
Tartu 2007
with side lengths a′ , b′ , c′ where a′ = b and b′ = c. Then
b a′ b′ c
Selected Problems from Open Contests = = = ,
a a b b
c2
1. Find all positive integers N with at most 4 digits such that the number obtained by hence b2 = ac. Second let K satisfy b2 = ac. Taking a′ = b, b′ = c and c′ = , we see
b
reversing the order of digits of N is divisible by N and differs from N. (Juniors.) that
Answer: 1089 and 2178. a′ b′ c′
Solution. Let N be any number satisfying the conditions. Let N ′ be the number with = = ,
a b c
digits in reversed order. Then N ′ = k · N for some integer k ≥ 2. Clearly, N must have
at least 2 digits. It cannot end with 0 as otherwise N ′ < N, which is impossible. Let i.e., triangle K ′ with side lengths a′ , b′ , c′ is just like required by the definition of disguis-
N = a . . . b and N ′ = b . . . a where a and b are digits. Then b ≥ ka ≥ 2a. On the other ability.
hand, a multiple of b ends with a. The pairs (a, b) that meet all conditions imposed by If a, b, c are integers then obviously b ≥ 2. A case study of possible values of b shows
the observations made so far are the following: that, for b = 2, . . . , 5, there exist no integers a and c such that a < b < c and ac = b2 and
c < a + b. For b = 6, we can take a = 4 and c = 9, giving perimeter 19. Thus we have
(1, 3), (1, 7), (1, 9), (2, 4), (2, 6), (2, 7), (2, 8), (2, 9), (3, 7), (3, 9), (4, 8), (4, 9). one more condition: a + b + c ≤ 19. As a ≥ 1 and c ≥ b + 1, this implies 2b + 2 ≤ 19
and b ≤ 8. So it suffices to check that for b = 7 and b = 8, no integers a and c such that
For pairs (1, 3), (2, 4), (2, 7), (2, 9), (3, 7), (3, 9), (4, 8), (4, 9), the conditions ka ≤ b and a < b < c and ac = b2 and c < a + b exist.
kb ≡ a (mod 10) are contradictory. Among the remaining pairs, (1, 7) gives k = 3, (b) Let K be a triangle satisfying the conditions of the problem. Then gcd(a, b, c) = 1
while the first digit of 3 · 1 . . . 7 cannot be 7. Analogously, (2, 6) gives k = 2, while the and b2 = ac. This implies that gcd(a, c) = 1 (as each common prime divisor of a and c
first digit of 2 · 2 . . . 6 cannot be 6. Hence only the pairs (1, 9) and (2, 8) are possible. would also divide b). Thus, both a and c are perfect squares.
As 91 is not divisible by 19 and 82 is not divisible by 28, no two-digit numbers satisfy
the conditions. For three-digit numbers, there must be either N = 1c9 and N ′ = 9c1 3. In a school tennis tournament with m ≥ 2 participants, each match consists of 4
with k = 9 or N = 2c8 and N ′ = 8c2 with k = 4. The first case leads to equality sets. A player who wins more than half of all sets during a match gets 2 points for
900 + 10c + 1 = 9 · (100 + 10c + 9) which gives 80c + 80 = 0, so no suitable c exists. The this match. A player who wins exactly half of all sets during the match gets 1 point,
second case fails analogously. and a player who wins less than half of all sets gets 0 points. During the tournament,
In the case of four-digit numbers, we have either N = 1cd9 and N ′ = 9dc1 with k = 9 each participant plays exactly one match against each remaining player. Find the least
or N = 2cd8 and N ′ = 8dc2 with k = 4. In the first case, the equality 9000 + 100d + number of participants m for which it is possible that some participant wins more sets
10c + 1 = 9 · (1000 + 100c + 10d + 9) leads to d = 89c + 8. The only solution in digits is than any other participant but obtains less points than any other participant. (Juniors.)
c = 0, d = 8. In the second case, the equality 8000 + 100d + 10c + 2 = 4 · (2000 + 100c + Answer: 6.
10d + 8) leads to 2d = 13c + 1. The only solution in digits is c = 1, d = 7.
Solution. Let m = 5. A participant who wins more sets than any other during the
2. Call a scalene triangle K disguisable if there exists a triangle K ′ similar to K with tournament must win more than half of all sets he plays. This implies that he must win
two shorter sides precisely as long as the two longer sides of K, respectively. Call a more sets than his opponent in at least one match, i.e., he must win at least one match.
disguisable triangle integral if the lengths of all its sides are integers. But in order to obtain less points than anyone else, he must lose more matches than
win. As each participant plays 4 matches, this special participant must win exactly one
(a) Find the side lengths of the integral disguisable triangle with the smallest possible match and lose at least two. Under such conditions, he can win at most 8 sets during
perimeter. the tournament but this is not more than half of the number 16 of all sets.
(b) Let K be an arbitrary integral disguisable triangle for which no smaller integral Thus, for m = 5, the described situation is impossible. If it were possible for some m
disguisable triangle similar to it exists. Prove that at least two side lengths of K are such that m < 5, we could obtain a suitable tournament table also for m = 5 by adding
perfect squares. an appropriate number of players whose matches all end in draw.
The following table shows a situation for m = 6 where all conditions are fulfilled:
(Juniors.)
Answer: (a) 4, 6, 9. Player Marks Sets won
Solution. (a) Let K be a triangle with side lengths a, b, c, where a < b < c. We show that 1. 4:0 4:0 1:3 1:3 1:3 4 11
K is disguisable iff b2 = ac. First let K be disguisable and let K ′ be a similar triangle to K 2. 0:4 2:2 3:1 2:2 3:1 6 10
1 2
n
3. 0:4 2:2 2:2 3:1 2:2 5 9 by exactly students. Thus k and n must be even.
4. 3:1 1:3 2:2 2:2 2:2 5 10 2
It remains to show that the situation is possible for arbitrary even k and n. For this,
5. 3:1 2:2 1:3 2:2 2:2 5 10 enumerate the questions by numbers from 1 to k and the students by numbers 1 to
6. 3:1 1:3 2:2 2:2 2:2 5 10 n. Let every student with an odd number answer correctly exactly the questions with
an odd number and every student with an even number answer correctly exactly the
4. Call a k-digit positive integer a hyperprime if all its segments consisting of 1, 2, . . . , questions with an even number. The requirements are fulfilled.
k consecutive digits are prime. Find all hyperprimes. (Juniors.)
6. Let an = 1 + 2 + . . . + n for every n ≥ 1; the numbers an are called triangular. Prove
Answer: 2, 3, 5, 7, 23, 37, 53, 73, 373. that if 2am = an then a2m−n is a perfect square. (Seniors.)
Solution. One-digit hyperprimes are precisely the one-digit primes 2, 3, 5, 7.
In a larger hyperprime, all digits must be prime. The last digit can be neither 2 nor Solution. We depict an as a set of points orga- (n − m) 2
5, and no two consecutive digits can be equal (this would form a composite segment). nized triangularly as shown in Fig. 1. From
Adding one digit to all one-digit primes, while following these requirements, we obtain two ends of the base, separate two triangles
|
numbers 23, 27, 37, 53, 57, 73. Among these, only 23, 37, 53, 73 are primes. both containing am points. For counting 2am
Note that all segments of any hyperprime are hyperprime. Thus all three-digit hyper- points, we count the points in the intersec-
primes can be obtained from two-digit hyperprimes by adding one digit to the end. tion of the two triangles twice, while leav-
Following the requirements above, we get hyperprime candidates 237, 373, 537, 737, ing the points in the upper rhomboid un- am
{z
among which only 373 is really a prime and a hyperprime. counted; for counting an points, every point
is taken into account once. Thus if 2am = an
Hyperprimes with more than 3 digits are impossible since their every segment of 3
then the intersection contains as many points
digits should be 373.
as the rhomboid. The former contains a2m−n
points while the latter contains (n − m)2 .
}
5. In an exam with k questions, n students are taking part. A student fails the exam
Remark. One can also prove the claim al- | {z }
if he answers correctly less than half of all questions. Call a question easy if more than a 2m −n
half of all students answer it correctly. For which pairs (k, n) of positive integers is it gebraically using the formula of the sum of
possible that arithmetic progression. Figure 1
(a) all students fail the exam although all questions are easy; 7. Three circles with centres A, B, C touch each other pairwise externally, and touch
(b) no student fails the exam although no question is easy? circle c from inside. Prove that if the centre of c coincides with the orthocentre of triangle
ABC, then ABC is equilateral. (Seniors.)
(Juniors.)
Solution 1. Let the tangent point of circles with centres A and B be C′ , the tangent point
Answer: (a) there are no such pairs; (b) all pairs (k, n) with both k and n even.
of circles with centres B and C be A′ and the tangent point of circles with centres C and
Solution. Let v be the total number of correct answers given by all students.
A be B′ (see Fig. ??). Let A′′ , B′′ , C′′ be the tangent points of circle c with the circles with
k
(a) If all students fail then each of them gives less than correct answers, i.e., v < centre A, B, C, respectively. Let H and I be the orthocentre and the incentre of ABC,
2 respectively. Assume that H is the centre of c.
k nk n
n· =
2 2
. If all questions are easy then, for each of them, more than correct answers
2 We prove first that triangle ABC is acute. Line H A′′ passes through A and contains both
n nk a radius of c and the altitude of ABC drawn from A. If angle BAC were not acute then
are given, i.e., v > k · = , contradiction. the orthocentre of ABC would be on ray AA′′ while the centre of c would be outside
2 2
k k this ray since | AA′′ | < | H A′′ |. Analogously, the other angles of ABC must be acute.
(b) If no student fails then each of them gives at least correct answers, i.e., v ≥ n · =
2 2 Draw tangents to c from points A′′ , B′′ , C′′ . As ABC is acute, point H lies inside it. Hence
nk k each of the three arcs of c with endpoints A′′ , B′′ , C′′ is less than 180◦ . Consequently,
; the equality holds iff each student gives exactly correct answers. On the other
2 n 2 these tangents intersect each other pairwise, forming a triangle DEF whose incircle is c.
hand, if no question is easy then no more than correct answers are given to each of
n nk 2 n As both BC and EF are perpendicular to H A′′ , they are parallel. Analogously, CA and
them, i.e., v ≤ k · = , whereby equality holds iff each question gets exactly correct FD are parallel, and AB and DE, too. Thus triangles ABC and DEF are similar.
2 2 2
nk Prove now that the orthocentre of DEF is I. Points A′ , B′ , C′ lie on the sides of ABC;
answers. These two inequalities can both hold only if v = . Consequently, each it is known that they are also the points where the incircle of ABC touches the sides.
2
k Thus I A′ , C′′ D and B′′ D are the radical axes of c and two circles touching it and each
student answers exactly questions correctly and each question is answered correctly
2
3 4
other. The radical axes meet at D. Thus ID, IE and IF are perpendicular to BC, CA, AB, these into (??), we obtain
respectively, and consequently also to the corresponding sides of DEF.
From all this, we get that the distance between the orthocentre and the incentre is the r cos ∠CAB + | AC| = r cos ∠CBA + | BC|. (2)
same in triangles ABC and DEF. As these triangles are similar but not equal, this can
happen only if the distance is zero, i.e., the orthocentre and incentre coincide. This As the opposite angle of a bigger side is bigger in every triangle, | AC| < | BC| would
implies that ABC is equilateral. imply ∠CBA < ∠CAB and cos ∠CBA > cos ∠CAB, leading to r cos ∠CAB + | AC| <
F r cos ∠CBA + | BC| which contradicts (2). Analogously, also | AC| > | BC| cannot be.
Consequently, | AC| = | BC|. In the same way, we get | AB| = | AC|, i.e., triangle ABC is
equilateral.
M Remark. The claim of the problem holds also without the assumption that the tangency
of c with the three circles is inner.
B ′′ A′′ B ′′ A′′
B A B A
8. Let b be an even positive integer for which there exists a natural number n such
C′ bn − 1
H H that n > 1 and is a perfect square. Prove that b is divisible by 8. (Seniors.)
A′ B′ b−1
C K C L bn − 1
Solution. As b is even, the perfect square is odd. Hence it is congruent to 8
b−1
D C ′′ E C ′′ modulo 1, i.e., the number
bn − 1
Figure 2 Figure 3 − 1 = b + b2 + . . . + b n − 1 = b 1 + b + . . . + b n − 2
b−1
Solution 2. Let r be the radius of c and let r A , r B , rC be the radii of circles with centre A,
B, C, respectively. Let h A be the length of the altitude of ABC drawn from A. Let A′ , B′ , is divisible by 8. As the factor 1 + b + . . . + bn−2 is odd, it is relatively prime to 8 and
C′ be the feet of altitudes of triangle ABC drawn from vertices A, B, C, respectively. hence b is divisible by 8.
From triangle H A′ C, we get | H A′ |2 = |CH |2 − |CA′ |2 ; from triangle AA′ C, we get Remark. One can also prove the claim by considering b modulo 8.
|CA′ |2 = | AC|2 − | AA′ |2 . Here, |CH | = r − rC , | AC| = r A + rC and | AA′ | = h A . Thus
9. The Fibonacci sequence is determined by conditions F0 = 0, F1 = 1, and Fk =
| H A′ |2 = (r − rC )2 − (r A + rC )2 + h2A = r2 − 2(r + r A )rC − r2A + h2A .
Fk−1 + Fk−2 for all k ≥ 2. Let n be a positive integer and let P( x ) = am x m + . . . + a1 x + a0
Analogously, we obtain be a polynomial that satisfies the following two conditions:
| H A′ |2 = r2 − 2(r + r A )r B − r2A + h2A . (1) P( Fn ) = Fn2 ; (2) P( Fk ) = P( Fk−1 ) + P( Fk−2 ) for all k ≥ 2.
These two equalities together give r B = rC . Analogously, r B = r A . Thus the radii of the Find the sum of the coefficients of P. (Seniors.)
circles drawn around A, B, C touching pairwise each other are equal. This can be only
Answer. Fn .
if the sides of ABC are all equal.
Solution. We are asked to find P(1). If n = 1 then P(1) = P( F1 ) = F12 = 1, giving
Solution 3. Let r A , r B , rC be defined as in Solution 2. Like in Solution 1, note that H lies
P(1) = F1 . If n ≥ 2 then using condition P( Fk ) = P( Fk−1 ) + P( Fk−2 ), 2 ≤ k ≤ n,
inside triangle ABC. As H is the centre of c, radii H A′′ and HB′′ are equal (see Fig. ??)
repeatedly, we get
which means that | H A| + r A = | HB| + r B . Adding rC to both sides of this equality, we
get P( Fn ) = P( Fn−1 ) + P( Fn−2 ) = 2P( Fn−2 ) + P( Fn−3 ) = 3P( Fn−3 ) + 2P( Fn−4 ) = . . .
| H A| + | AC| = | HB| + | BC|. (1) = Fn P( F1 ) + Fn−1 P( F0 ) = Fn P(1) + Fn−1 P(0).
Take triangle KLM whose midlines are the sides of ABC; then triangles KLM and ABC Using the given condition again for k = 2, we obtain
are similar. Thereby, AH is the perpendicular bisector of side LM as LM k BC and
| LA| = | AM|. Also, BH is the perpendicular bisector of side MK. Thus H is the circum- P(1) = P( F2 ) = P( F1 ) + P( F0 ) = P(1) + P(0),
1
centre of triangle KLM and ∠ MH A = ∠ MHL = ∠ MKL = ∠CAB. which gives P(0) = 0. Altogether, Fn2 = P( Fn ) = Fn P(1), showing that P(1) = Fn .
2
From the right triangle MAH, we get | H A| = r cos ∠ MH A = r cos ∠CAB where r is Remark. We could ask whether there exist polynomials for every n satisfying the con-
the radius of the circumcircle of KLM. Analogously, | HB| = r cos ∠CBA. Substituting ditions of the problem. Using the condition P( Fk ) = P( Fk−1 ) + P( Fk−2 ), 2 ≤ k ≤ n,
5 6
T
for finding the other values P( Fk ), 2 ≤ k ≤ n, we get P( Fk ) = Fn Fk , for all k such that
0 ≤ k ≤ n. Elsewhere, the values of the polynomial are not determined. One suitable
polynomial is P( x ) = Fn x.
10. Does there exist a natural number n such that n > 2 and the sum of squares of
some n consecutive integers is a perfect square? (Seniors.)
Answer: yes.
Solution 1. For n = 11, we can construct the following example: D1
A2
(−4)2 + (−3)2 + (−2)2 + (−1)2 + 02 + 12 + 22 + 32 + 42 + 52 + 62 = 112 . C1
B2
n
Solution 2. It is easy to prove by induction that 12 + 22 + . . . + n2 = (n + 1)(2n + 1). P
6
If n = 24 then all three factors in the last product are perfect squares (4, 25 and 49, B1
C2
respectively). Thus the product is also a perfect square.
l1
Remark 1. For n = 3, . . . , 10, no suitable examples exist because there is a number mod- A1
ulo which no sum of squares of n consecutive integers is a quadratic residue. l2
D2
n sum expression bad modulus
3 ( a − 1 )2 + . . . + ( a + 1 )2 3a2 + 2 3 Figure 4
4 ( a − 1 )2 + . . . + ( a + 2 )2 4a2 + 4a + 6 4
5 ( a − 2 )2 + . . . + ( a + 2 )2 5a2 + 10 25 Solution 1. Let r1 and r2 be the radii of c1 and c2 , respectively. Since quadrangle
6 ( a − 2 )2 + . . . + ( a + 3 )2 6a2 + 6a + 19 4 A1 A2 D1 D2 is cyclic, ∠ A2 A1 D1 = ∠ D1 D2 A2 . In addition, ∠ A1 PA2 = ∠ D2 PD1 (see
7 ( a − 3 )2 + . . . + ( a + 3 )2 7a2 + 28 49 Fig. 4). Thus triangles A1 A2 P and D2 D1 P are similar. Their incircles are c1 and c2 , hence
8 ( a − 3 )2 + . . . + ( a + 4 )2 8a2 + 8a + 44 16
| A P|
1 r
9 ( a − 4 )2 + . . . + ( a + 4 )2 9a2 + 60 9 = 1.
| D2 P | r2
10 ( a − 4 )2 + . . . + ( a + 5 )2 10a2 + 10a + 85 25
Since quadrangle B1 B2 C1 C2 is cyclic, ∠ B2 B1 C1 = ∠C1 C2 B2 , whence ∠TB1 A1 =
Remark 2. Using diophantine equation theory, it has been proven that the sum of squares ∠TC2 D2 . As also ∠TA1 B1 = ∠TD2 C2 , triangles A1 B1 T and D2 C2 T are similar. Their
of numbers from 1 to n is a perfect square only for n = 0, n = 1 and n = 24. Thus the incircles are c1 and c2 again, hence
choice n = 24 in Solution 2 is the only possibility to succeed.
Remark 3. Sloane’s Encyclopedia of Integer Sequences contains the sequence A001032 with | A1 T | r
= 1.
description “Numbers n such that the sum of squares of n consecutive positive integers | D2 T | r2
can be a perfect square” (i.e., in addition to the conditions of our problem, the numbers | A1 P | | A T|
whose squares are considered must be positive). The sequence starts with numbers 1, Now consider triangles A1 PT and D2 PT. We have = 1 and ∠TA1 P =
| D2 P | | D2 T |
2, 11, 23, 24, 26, 33, 47, 49, 50, 59, 73, 74, 88, 96, 97, 107, 121, 122, 146, 169, 177, 184, 191, ∠TD2 P. Thus these triangles are similar and
193, 194, 218, 239, 241, 242, 249, 289, 297, 299, 311, 312, 313, 337, 338, 347, 352, 361, 362,
376, 383, 393, 407, 409, 431, 443, 457, 458, 479, 481, 491. r 1 | A P| | PT |
= 1 = = 1.
r2 | D2 P | | PT |
11. Tangents l1 and l2 common to circles c1 and c2 intersect at point P, whereby tangent
points remain to different sides from P on both tangent lines. Through some point Solution 2. Let r1 and r2 be the radii of c1 and c2 , respectively. Let s be the line that passes
T, tangents p1 and p2 to circle c1 and tangents p3 and p4 to circle c2 are drawn. The through P and is perpendicular to the line joining the centres of c1 and c2 . Consider the
intersection points of l1 with lines p1 , p2 , p3 , p4 are A1 , B1 , C1 , D1 , respectively, whereby composition of two plane transformations: reflection w.r.t. s and homothety w.r.t. P
the order of points on l1 is: A1 , B1 , P, C1 , D1 . Analogously, the intersection points of l2 r2
with factor . This composite transformation takes c1 to c2 and P to P.
with lines p1 , p2 , p3 , p4 are A2 , B2 , C2 , D2 , respectively. Prove that if both quadrangles r1
A1 A2 D1 D2 and B1 B2 C1 C2 are cyclic then radii of c1 and c2 are equal. (Seniors.) Denote the image of any point X under this transformation by X ′ . As the transformation
respects all angles between lines, the equality of angles in cyclic quadrilateral B1 B2 C1 C2
7 8
√ √
implies ∠ B2′ B1′ P = ∠ B2 B1 P = ∠C1 C2 P. As lines B1′ B2′ and C2 C1 both touch circle c2 and 2+ 2 1 2+ 2 1
inradius and the perimeter gives r. Thus = r, whence r = √ =
intersect PC2 under the same angle, line B1′ B2′ coincides with line C2 C1 . Analogously, 2 2 2 2+ 2
√ √
A1′ A2′ coincides with line D2 D1 . 2− 2 2 1 √ 3 √
Hence the transformation takes the intersection point of lines A1 A2 and B1 B2 to the = 1− . Therefore r2 = 1 + − 2 = − 2, so r2 is not rational.
2 2 2 2
intersection point of lines C2 C1 and D2 D1 , i.e., T is taken to T. But if r1 6= r2 then the abc a2 b2 c 2
transformation obviously can have only one fixpoint. Consequently, r1 = r2 . Solution 2. (a) By the formula S = , we get R2 = . Thus it suffices to prove
4R 16S2
that S2 is rational. By Heron’s formula,
12. Find all positive integers n such that one can write an integer 1 to n2 into each unit
square of a n2 × n2 table in such a way that, in each row, each column and each n × n 1 2 2
S2 = p( p − a)( p − b)( p − c) = 2a (b + c2 ) − (b2 − c2 )2 − (a2 )2 .
block of unit squares, each number 1 to n2 occurs exactly once. (Seniors.) 16
The last expression clearly evaluates to a rational number.
Answer: 1.
(b) In part (a), it was proven that S2 is rational. For the triangle with side lengths 1, 1
Solution. In the case n = 1, the conditions can be fulfilled trivially. Assume n ≥ 2. √ !2 √
√ 2+ 2 6+4 2 3 √
Denote the unit square in the ith row and jth column by 1 2 ... n n +1 and 2, we get p2 = = = + 2, hence p2 is not rational and
(i, j). Let A be the n × n block containing both (1, 1) and 1
2 4 2
(n, n). Let B and C be the n × n blocks obtained from A by S2
2 neither is r2 = .
shifting it by one unit down and right, respectively (see . p2
Fig. 5). ..
The numbers in the bottommost row of B must be the n
same as the numbers in the topmost row of A in some or- Selected Problems from the Final Round of National
n +1
der since both blocks must contain each number 1, . . . , n2 Olympiad
exactly once. Analogously, the rightmost column of C
must contain the same numbers as the leftmost column Figure 5
of A in some order. Now, the number in (n + 1, n + 1) 1. Two medians drawn from vertices A and B of triangle ABC are perpendicular.
cannot occur in the topmost row of A since all these occur already in the row number Prove that side AB is the shortest side of ABC. (Grade 9.)
n + 1 left from the square under consideration. Analogously, this number cannot occur
in the leftmost column of A. As A contains all numbers 1 to n2 , this number must occur Solution. Let the medians intersect in point M and let the median drawn from vertex
elsewhere in A. But then it occurs twice in the n × n block that contains squares (2, 2) C intersect AB in point F (see Fig. 6). Then, F is the midpoint of the hypotenuse of
and (n + 1, n + 1) which is prohibited. the right triangle ABM and thus the midpoint of the circumcircle of ABM, so we get
| AB| = 2| FM|. Since M divides median CF in ratio 2 : 1, we have | AB| = |CM|. The
13. Consider triangles whose each side length squared is a rational number. Is it true largest angle of triangle AMC is the obtuse angle AMC, therefore AC is the longest side
that of this triangle. We get | AC| > | MC| = | AB|. The proof of | BC| > | AB| is analogous.
(a) the square of the circumradius of every such triangle is rational; Remark. One can also solve the problem using the Pythagorean theorem and the fact
(b) the square of the inradius of every such triangle is rational? that the centroid divides the medians in ratio 2 : 1.
(Seniors.) 2. Juhan wants to order by weight five balls of pairwise different weight, using only
Answer: (a) yes; (b) no. a balance scale. First, he labels the balls with numbers 1 to 5 and creates a list of weigh-
ings, such that each element in the list is a pair of two balls. Then, for every pair in
Solution 1. (a) Fix a triangle from the family under consideration. Let a, b, c be its side
the list, he weighs the two balls against each other. Can Juhan sort the balls by weight,
lengths, γ the size of the angle opposite to the last side and R the circumradius. The
using a list with less than 10 pairs? (Grade 9.)
a2 + b2 − c 2 ( a2 + b 2 − c 2 )2
cosine law gives cos γ = , whence cos2 γ = . As a2 , b2 , c2 are Answer: no.
2ab 4a2 b2
rational, also cos2 γ is rational. Therefore 1 − cos2 γ, i.e., sin2 γ, is rational. The sine law Solution. There are 10 possible pairs of 5 balls. Suppose w.l.o.g. that Juhan does not
c c2 weigh the pair (1, 2). Then, it is not possible to distinguish orderings 1, 2, 3, 4, 5 and
gives R = , hence R2 = . Thus R2 is rational.
2 sin γ 4 sin2 γ 2, 1, 3, 4, 5, since the remaining 9 weighings give the same result. Thus, Juhan’s list must
√
(b) Consider the right isosceles triangle with side lengths 1, 1, 2. Let r be its inradius. contain all 10 pairs.
1
The area of this triangle, computed via leg lengths, is ; computing the area via the 3. Two radii OA and OB of a circle c with midpoint O are perpendicular. Another
2
9 10
circle touches c in point Q and the radii in points C and D, respectively. Determine
∠ AQC. (Grade 10.)
Answer: 45◦ .
C 3 vertices 4 vertices 5 vertices 6 vertices 7 vertices
b a
2 2 B
8 vertices 9 vertices 10 vertices 12 vertices
D Q Figure 8
E
y x
M D union is an n-gon (not necessarily convex). Find all possible values of the number of
b a vertices n. (Grade 10.)
O′
2 2x 2y 2
Answer: all integers from 3 to 12, except 11.
A′ A
Solution. The n-gon must have at least 3 vertices. We show first that the number of
A B
O C vertices is at most 12. Indeed, each vertex of the n-gon is either a vertex of one of the
F
triangles or an intersection point of some two sides of the two triangles. There are 6
Figure 6 Figure 7 triangle vertices and 6 possible intersection points, since every side of the first triangle
can intersect at most two sides of the second triangle. Thus, n ≤ 12.
Solution. By symmetry, ∠ AQC = ∠ BQD (see Fig. 7). Since AQB is an internal angle
of a regular octagon, we have ∠ AQB = 135◦ . Let now O′ be the midpoint of the circle Next, suppose that the n-gon has 11 vertices. If 6 of those vertices are vertices of the two
through C, D and Q. The quadrilateral OCO′ D has three right angles: ∠COD = 90◦ by triangles, then neither triangle can contain a vertex of the other triangle. Thus, each side
assumption, while angles OCO′ and ODO′ are angles between a radius and a tangent. of each triangle intersects the other triangle either never or twice, so we cannot have an
1 odd number of intersection points. On the contrary, if the n-gon has 6 intersection points
Thus, CO′ D is also a right angle, so ∠CQD = ∠CO′ D = 45◦ and as vertices, every side of each triangle must intersect the second triangle twice, and thus
2
all vertices of one triangle must be outside the other triangle. Thus, all 6 triangle vertices
1 1 are also vertices of the n-gon, and n = 12.
∠ AQC = (∠ AQB − ∠CQD ) = (135◦ − 45◦ ) = 45◦ .
2 2 The remaining configurations from 3 to 12 are all possible (see Fig. 8).
Remark. One can also make use of the similarity of triangles QO′ D and QOA′ (where A′
6. The identifier of a book is an n-tuple of numbers 0, 1, . . . , 9, followed by a check-
is the other endpoint of the diameter of c drawn through A), yielding that angle AQA′
sum. The checksum is computed by a fixed rule that satisfies the following property:
subtends the diameter. Another approach is to apply the tangent chord property on
whenever one increases a single number in the n-tuple (without modifying the other
tangent AC and secant CD to obtain that triangles AQC and CQD are similar.
numbers), the checksum also increases. Find the smallest possible number of required
4. Prove that the sum of the squares of any three pairwise different positive odd in- checksums if all possible n-tuples are in use. (Grade 10.)
tegers can be represented as the sum of the squares of six (not necessarily different) Answer: 9n + 1.
positive integers. (Grade 10.) Solution. Consider the checksum of (0, 0, . . . , 0). Increasing the first number, we get n-
Solution. Let a > b > c be positive integers. Then tuples (1, 0, . . . , 0), . . . , (9, 0, . . . , 0) with 9 new checksums. Increasing the second num-
ber in the last n-tuple, we get (9, 1, . . . , 0), . . . , (9, 9, . . . , 0), and again obtain 9 new val-
a2 b2 b2 c 2 c 2 a2 ues. Continuing like this, we see that the number of different check values is at least
a2 + b2 + c 2 = + + + + + = 9n + 1.
2 2 2 2 2 2
2 2 2 2 2 It is easy to see that the sum of all numbers in the n-tuple is a valid checksum. On the
aa+b
− b b + c b − c a + c a − c 2
= + + + + + . other hand, the sum of n numbers 0, 1, . . . , 9 is at least 0 and at most 9n, so with this
2
2 2 2 2 2
rule, we have exactly 9n + 1 different checksums.
Since a, b and c are all odd, the latter is a sum of squares of six positive integers.
7. Find all real numbers a such that all solutions to the quadratic equation x2 − ax +
5. Two triangles are drawn on a plane in such a way that the area covered by their a = 0 are integers. (Grade 11.)
11 12
Answer: a = 0 and a = 4. ∠CAR = ∠CBQ = ∠PAB, so triangles ACR and ABP are similar. If AQ ⊥ BC, then AR
Solution. Let x and y be the solutions of the quadratic equation. Viète formulae give is the height of ABC and ∠ ARC = ∠ APB = 90◦ , so P = R and thus P lies on AQ. If P
x + y = xy = a. The case y = 1 gives a contradiction 1 + x = x, while y 6= 1 gives lies on AQ, then again P = R and ∠CRA = ∠CPA = 180◦ − ∠ BPA = 180◦ − ∠CRA, or
x=
y
. Thus, x is an integer if and only if y = 2 or y = 0. Now, y = 2 gives x = 2 ∠CRA = 90◦ , so AQ ⊥ BC.
y−1 A
Solution 2. First consider the case where AB
and a = x + y = 4; y = 0 gives x = 0 and a = x + y = 0.
is the diameter of the circle through A, B and
Remark. Oneqcan also proceed from the fact that the sum a as well as the difference
p P. Then ∠ APB = 90◦ and ∠QBA = 90◦ giv-
a2 − 4a = (a − 2)2 − 4 of the solutions is integral. ing that AQ is the diameter of circumcircle
of ABC. Thus AQ ⊥ BC and P lies on line
8. A 3-dimensional chess board consists of 4 × 4 × 4 unit cubes. A rook can step from AQ.
any unit cube K to any other unit cube that has a common face with K. A bishop can On the other hand, when the centre of
step from any unit cube K to any other unit cube that has a common edge with K, but the circle passing through points A and B
does not have a common face. One move of both a rook and a bishop consists of an moves away from the leg AB on the half-
arbitrary positive number of consecutive steps in the same direction. Find the average plane containing point C, points Q and P R
number of possible moves for either piece, where the average is taken over all possible will move toward points B and C, respec- B P C
starting cubes K. (Grade 11.) tively. When the centre of the circle moves
away from AB on the other halfplane, points
Answer: the rook has on average 9 moves, the bishop has 10.5. Q
Q and P will move toward points C and B,
Solution. The rook has always 3 pos- respectively. Therefore in either case point P Figure 10
R R sible moves in the direction of one neither lies on line AQ nor AQ is perpendic-
axis, regardless of the choice of the ular to BC.
starting cube K (see Fig. 9), and thus
9 possible moves on average. 10. Find all pairs (m, n) of positive integers such that mn − nm = 3. (Grade 11.)
In any of the 8 “middle” cubes of the Answer: (4, 1).
B chess board, the bishop has 5 possi-
Solution. First, m and n must have different parity, for otherwise the lhs is even.
B B ble moves in the direction of every
axis, and thus 15 moves in total. In • If m is odd and n even, then mn ≡ 1 (mod 4), so nm ≡ 2 (mod 4), which is possible
Figure 9 any of the 24 cubes that have a com- only for m = 1. But then mn = 1 and the lhs is smaller than 3. Thus, in this case
mon face with a middle cube, the there are no solutions.
bishop has 5 moves on the plane parallel to the common face, and 3 moves on each • If now m is even and n odd, then nm ≡ 1 (mod 8), so mn ≡ 4 (mod 8). Thus,
of the remaining two planes, or 11 moves in total. For the remaining 32 cubes on the n ≤ 2, and n odd gives n = 1, m = 4 as the only solution.
edge of the board, the bishop has 3 moves in the direction of every axis, and 9 moves in
total. Averaging over all cubes, we get the result.
11. Some circles of radius 2 are drawn on the plane. Prove that the numerical value of
Remark. The number of possible moves shows to some extent the strength of every piece. the total area covered by these circles is at least as big as the total length of arcs bounding
The solution to this problem implies that in 3-dimensional chess, a bishop is perhaps the area. (Grade 11.)
stronger than a rook (as opposed to regular chess). On the other hand, the bishop is
weakened by the fact that it can always reach only half of the cubes or squares. Solution 1. The boundary line of the area consists of circular arcs, each corresponding to
While a 3-dimensional 4 × 4 × 4 board and a regular 8 × 8 board have the same number rl
a circular sector. According to the formula S = , where r is the radius and l the length
of cells, a regular rook and bishop have 14 and 8,75 moves on average, respectively. 2
of arc, we obtain that in the case r = 2 the area of circular sector equals numerically the
length of the circular arc on its boundary.
9. A circle passing through the endpoints of the leg AB of an isosceles triangle ABC
intersects the base BC in point P. A line tangent to the circle in point B intersects the We prove that no two such sectors have common interior points. First, the sectors from
circumcircle of ABC in point Q. Prove that P lies on line AQ if and only if AQ and BC the same circle share only the centre of the circle. Now let one sector be part of circle w1
are perpendicular. (Grade 11.) with centre O1 and the other be part of circle w2 with centre O2 . Assuming that the sec-
tors have a common interior point C, we draw the radii O1 A and O2 B of circles w1 and
Solution 1. Let line AQ intersect base BC in point R (see Fig. 10). On one hand, ABC w2 through C, respectively. Obviously A and B lie on the boundary of the area covered
is isosceles, so ∠ ACR = ∠ ABP. On the other hand, the tangent chord property gives by circles, meaning that B and A do not lie in the interior of w1 and w2 , respectively.
13 14

Therefore |O1 B| ≥ |O1 A| and |O2 A| ≥ |O2 B|. We draw the mediator of segment AB x x n z
Since < 1, + 1 < 2 holds for all n. On the other hand, since > 1, there exists
(see Fig. 11). The inequalities imply that O1 and A lie on the one side of the mediator y y y
N
and O2 and B lie on the other side. Hence segments O1 A and O2 B cannot have common z
points, a contradiction. A
an integer N such that > 2, contradiction.
y
We have obtained that the total length of the bound- B
ary line of the area equals the sum of the lengths of 14. Does there exist an equilateral triangle
circular arcs. But the total area is greater or equal (a) on a plane; (b) in a 3-dimensional space;
than the sum of the areas of the sectors.
Solution 2. We shall use induction on the number of such that all its three vertices have integral coordinates? (Grade 12.)
O1
circles. For one circle, the circumference and the area Answer: (a) no; (b) yes.
O2
are equal (4π). We prove that whenever we add a −→ −→
circle, the area a covered by the intersection of the Solution 1. (a) Suppose that such a triangle ABC exists. Then vectors AB and AC have
−→
new circle with the old area is at most as big as the Figure 11 integral coordinates. Denote AB = ( x, y), then the vector of the height drawn from
√
perimeter p of that intersection. 3 −→
vertex C is (y, − x ). The coordinates of AC are then either
First, notice that a ≤ 4π. Assume now that also p ≤ 4π. We write p = 2πr, so r ≤ 2
2. Since a circle maximizes the area for a fixed perimeter, we see that the area of the √ √ √ !
−→ 1 3 x + 3y y − 3x
intersection is a ≤ πr2 ≤ 2πr = p as desired. AC = ( x, y) + (y, − x ) = ,
2 2 2 2
12. Consider a cylinder and a cone with a common base such that the volume of the or
part of the cylinder enclosed in the cone equals the volume of the part of the cylinder √ √ √ !
outside the cone. Find the ratio of the height of the cone to the height of the cylinder. −→ 1 3 x− 3y y + 3x
AC = ( x, y) − (y, − x ) = , .
(Grade 12.) 2 2 2 2
1
Answer: 1 + √ . In either case, the coordinates are integral only for x = y = 0.
3 (b) Triangle ABC with A = (1, 0, 0), B = (0, 1, 0) and C = (0, 0, 1) is equilateral.
Solution. Denote by v and V, h and H the volume and the height of the cylinder and the
Solution 2. (a) W.l.o.g. assume that one vertex of the triangle is A(0, 0). We also assume
cone, respectively, and denote by S the area of the common base. The vertex of the cone
w.l.o.g. that the ordinates of B and C are non-negative and these vertices do not lie on
must lie outside the cylinder, for otherwise the volume of the intersection would be at
1 the y-axis.
most of the total volume of the cylinder. Since all the coordinates are integral, the slopes of AB and AC are rational. Denote the
3 h
Denote = x. The part of the cone that lies outside the cylinder is a cone similar to slope angles of lines AB and AC by β and γ, respectively, then γ = β ± 60◦ . Denote
H H −h tan β = k. We have
the original cone with scale factor = 1 − x and volume (1 − x )3 V. The volume √ √ √
H
v tan β ± tan 60◦ k± 3 (k ± 3)(1 ± k 3)
of the part of the cylinder inside the cone is thus V − (1 − x )3 V = . From v = Sh and tan γ = = √ = .
√ 2 1 ∓ tan β tan 60◦ 1∓k 3 1 − 3k2
1 3 3± 3 √ √ √ √ √
V = SH we get ( x3 − 3x2 + 3x ) = x, so x = . Since h < H implies x < 1, the But (k ± 3)(1 ± k 3) = k ± k2 3 ± 3 + 3k = 4k ± (k2 + 1) 3. As k2 + 1 > 0, it is
3 √ 2 2
3− 3 impossible that tan γ and k would be simultaneously rational.
only possible solution is x = and we get the desired ratio.
2 Remark. There are several other solutions to this problem. One may w.l.o.g. denote
the vertices A(0, 0), B( x1 , y1 ) and C( x1 , y1 ), where x1 , y1 , x2 and y2 are relatively prime,
13. Let x, y, z be positive real numbers such that x n , yn and zn are side lengths of some derive the equalities x12 + y12 = x22 + y22 = 2( x1 x2 + y1 y2 ) and consider all cases of re-
triangle for all positive integers n. Prove that at least two of x, y and z are equal. (Grade mainders of x1 , y1 , x2 and y2 modulo 2.
12.) Another approach would be to notice that ( x1 + x2 )2 + (y1 + y2 )2 is divisible by 3 and
Solution. Assume that x, y, z are all different and assume w.l.o.g. x < y < z. For any n, use the fact that perfect squares have only remainders 0 and 1 modulo 3.
the triangle inequality implies x n + yn > zn , or The most straightforward way is to compute the area S of the triangle with vertices

n n x1 x2 x3
x z A( x1 , y1 ), B( x2 , y2 ) and C( x3 , y3 ) in two different ways: 2S = y1 y2 y3 and 2S =
+1 > . 1 1 1
y y
15 16
√
a2 3
(where a is the side of the triangle). Since all the coordinates as well as the side
2
length are integral, such a triangle cannot exist.
15. Let a, b, c be positive integers such that gcd(a, b, c) = 1 and the product of every
two of these integers is divisible by the third one.
(a) Prove that every one of these integers equals the least common multiple of the re-
maining two integers divided by the greatest common divisor of these two integers.
n = 4q n = 4q + 1 n = 4q + 2 n = 4q + 3
(b) Give an example of such integers a > 1, b > 1 and c > 1.
Figure 12
(Grade 12.)
Answer: (b) For example, a = 6, b = 10, c = 15. r 4
Solution 1. (a) Let d = gcd(a, b), a = a′ d and b = b′ d, where gcd(a′ , b′ ) = 1. Then
4
lcm(a, b) .
lcm(a, b) = a′ b′ d and = a′ b′ . We prove that a′ b′ = c. ..
gcd(a, b) .
..
On the one hand, since ab = a′ b′ d2 is divisible by c and gcd(d, c) = 1 because of 4
gcd(a, b, c) = 1, a′ b′ must be divisible by c. On the other hand, the conditions of the 4
problem imply that ca = ca′ d is divisible by b = b′ d, i.e. ca′ is divisible by b′ . As 4
gcd(a′ , b′ ) = 1, c is divisible by b′ . Analogously, c is divisible by a′ . Since a′ and b′ are 4 r
coprime, c is divisibly by a′ b′ . Altogether we have c = a′ b′ as desired. The claims about
4 4 ··· 4 r r 4 4 ··· 4
a and b are proven analogously.
(b) Let x, y and z be different numbers that are pairwise coprime, e.g. different primes. Figure 13
Having a = xy, b = yz and c = xz, the numbers a, b and c satisfy the conditions of the
problem. Indeed,
• If r = 1, we mark groups of 8 squares. Since there are q2 such groups, we again get
lcm(a, b) xyz A(4q + 1) ≤ 8q2 .
= = xz = c,
gcd(a, b) y • If r = 2, we mark as in the first case, but leave the last two rows empty. This way,
we mark 2q rows with n squares, or 2q(4q + 2) squares in total. Thus, A(4q + 2) ≤
analogously for other cases. For an example we may take x = 2, y = 3, z = 5 that yields 8q2 + 4q.
6, 15 and 10.
• If r = 3 and q > 0, we mark every fourth row and column and the middle squares
Solution 2. (a) Let p be some prime factor of at least one of a, b and c. Since gcd(a, b, c) =
of the remaining 3 × 3 groups. There are (q + 1)2 marked middle squares and q
1, we may assume w.l.o.g. that p does not divide c. At the same time, since ca is divisible
marked rows and columns, giving (q + 1)2 + 2(4q + 3)q − q2 marked squares in
by b and cb is divisible by a, the factor p must have the same exponent α in both a and
total. Thus, A(4q + 3) ≤ 8q2 + 8q + 1. For q = 0, we get n = 3 and A(3) = 0.
b. Similarly, the exponents of any prime factor p′ in the prime factorization of a, b and c
are α′ , α′ and 0 in some order. Since every exponent in such a triple equals the difference We now show that these bounds are tight.
of the maximum and the minimum of the remaining two, the result follows.
• For r = 0 or r = 1, we can divide a 4q × 4q grid into q2 squares of size 4 × 4. In
16. Some squares of an n × n grid are marked in such a way that in every 4 × 4 square, every such square, at least 8 squares must be marked, so A(n) ≥ 8q2 .
at least half of the squares are marked. Find the smallest possible number of marked • For r = 2 or r = 3, cut an r × r square from the lower left corner and divide the
squares in the grid. (Grade 12.) remaining squares into q L-shaped strips of width 4 (see Fig. 13, right). The ith strip
then contains 2i 4 × 4 squares. The last two squares intersect in the corner of the L
Answer: 8q2 for n = 4q or n = 4q + 1; 8q2 + 4q for n = 4q + 2; 8q2 + 8q + 1 for n = 4q + 3
and the intersection is a square of size (4 − r ) × (4 − r ). Every strip also contains
and q ≥ 1; 0 for n = 3.
an r × r square in the corner.
Solution. Let A(n) be the smallest possible number of marked squares. Write n = 4q + r,
where 0 ≤ r < 4. First, we bound A(n) from above. In order to bound A(n) from below, we add the smallest possible number of
marked squares in the 4 × 4 squares and the r × r squares and subtract the largest
• If r = 0, we mark half of the squares in groups of two rows as shown in Fig. 12, so possible number of marked squares in the (4 − r ) × (4 − r ) squares.
A(4q) ≤ 8q2 .
17 18
The case r = 2 gives Solution 2. Let ai be the number of times the ith switch changes its position during the
whole process. According to the conditions of the problem, each switch moves either
(2q + 2)q when it is moved directly by the operator or its left neighbour
A(n) ≥ 8 · (2 + 4 + . . . + 2q) + 0 · (q + 1) − 4 · q = 8 · − 4q = 8q2 + 4q. j moves
a k down. As all
2 switches are down at the beginning, the ith switch moves down i times. Thus a1 = 1
j a k 2
If r = 3 every 3 × 3 square contains at least one marked square, so i −1
and ai = i + for all i ≥ 2.
2
(2q + 2)q j k
a 1
A(n) ≥ 8 · (2 + 4 + . . . + 2q) + 1 · (q + 1) − 1 · q = 8 · + 1 = 8q2 + 8q + 1. We prove by induction that ai = 2(i − 1) for all i ≥ 2. As a2 = 2 +
1
= 2+ = 2,
2 2 2
this claim holds for i = 2. Assuming that it holds for i, we obtain
ja k
IMO team selection contest ai + 1 = i + 1 + i
= i+1+
2 (i − 1 )
= 2i,
2 2
First day i.e., the claim holds also for i + 1.
Altogehter, this shows that a1 is odd and ai is even for all i ≥ 2. Hence, after the process,
1. On the control board of a nuclear station, there are n electric switches (n > 0), all the first switch is up and all the others are down.
in one row. Each switch has two possible positions: up and down. The switches are Solution 3. Interpret the position of switches on the board as binary numbers so that the
connected to each other in such a way that, whenever a switch moves down from its ith switch from the left corresponds to the ith lowest binary digit: being down encodes
upper position, its right neighbour (if it exists) automatically changes position. At the 0 and being up encodes 1. Changing the ith switch then works like addition of 2i −1
beginning, all switches are down. The operator of the board first changes the position modulo 2n . The initial position encodes number 0 and the final position encodes 1 · 20 +
of the leftmost switch once, then the position of the second leftmost switch twice etc., 2 · 21 + . . . + n · 2n−1 modulo 2n .
until eventually he changes the position of the rightmost switch n times. How many
We prove by induction that 1 · 20 + 2 · 21 + . . . + n · 2n−1 ≡ 1 (mod 2n ). If n = 1 then
switches are up after all these operations?
this holds. Assume that the claim holds for n = k. Multiplying this congruence by 2
Answer: 1. gives
Solution 1. Enumerate the switches with numbers 1 to n from left to right. We prove first
that the result of two consecutive changes does not depend on the order of the changes. 1 · 21 + 2 · 22 + . . . + k · 2 k ≡ 2 (mod 2k+1 ).
Let x and y be the numbers of the switches changed, x < y.
Adding 20 + 21 + . . . + 2k to both sides gives
• If there exists a number z such that x ≤ z < y and switch number z is down then
changing the position of x can influence only switches from x to z, changing the 1 · 20 + 2 · 21 + 3 · 22 + . . . + ( k + 1 ) · 2 k ≡ 2 + 2 k + 1 − 1 ≡ 1 (mod 2k+1 ),
position of switch y can influence only this switch and switches right from y. Thus
i.e., the claim holds for n = k + 1.
the results of the changes are independent of each other.
Remark. In Solution 3, one could prove by induction a stronger claim: 1 · 20 + 2 · 21 +
• If no such z exists then changing switch number x causes a change of switch number
. . . + n · 2n−1 = (n − 1) · 2n + 1.
y. After that, switches x to y − 1 are all down while all switches in the right from
them are in the same position as if switch number y were changed. Thus after 2. Let D be the foot of the altitude of triangle ABC drawn from vertex A. Let E and
moving both x and y in either order, switches from x to y − 1 are down and the F be points symmetric to D w.r.t. lines AB and AC, respectively. Let R1 and R2 be the
switches with larger number are in the position as when switch y were moved circumradii of triangles BDE and CDF, respectively, and let r1 and r2 be the inradii of
twice. the same triangles. Prove that
We prove now that, after all operations, precisely the leftmost switch is up. This claim |S ABD − S ACD | ≥ | R1 r1 − R2 r2 |
holds trivially for n = 1. Assume the claim holding for n switches and consider a board
with n + 1 switches. According to what was proven above, the moves can be performed where SK denotes the area of figure K.
in arbitrary order. Therefore, first change switch number 2 once, then switch number 3
Solution 1. Consider first the case where D lies between points B and C (see Fig. 14). As
twice etc., until the last switch n times. By the induction hypothesis, switch number 2 1 1
is up and all the others are down. Each switch has to be moved once more; if we do it S ABD = · | AD | · | BD | and S ACD = · | AD | · |CD |, we have
2 2
from right to left then switches n + 1 to 3 go up, then moving switch 2 down brings all
them down and finally switch 1 is moved up. Thus 1 is the only switch remaining up. 1
S ABD − S ACD = · | AD | · (| BD | − |CD |).
2
19 20
Let G be the incentre of triangle BDE and let G ′ be the projection of G to line BD. Then Analogously we obtain
| GG′ | = r1 . By symmetry, ∠BEA = ∠BDA = 90◦ , hence quadrangle BEAD is cyclic
and line segment AB is its circumdiameter. Thus | AB| = 2R1 . As triangles ADB and 1 1 γ
S ACD = | AD |2 tan γ, R2 r 2 = | AD |2 tan γ tan .
| AB| | GB| 2 2 2
GG ′ B are similar, we have = , implying 2R1 r1 = | AD | · | GB|. Let H be the
| AD | | GG′ | From these equalities, we can conclude that S ABD − S ACD and R1 r1 − R2 r2 have the
incentre of triangle CDF; then analogously 2R2 r2 = | AD | · | HC|. Hence same sign since β and γ belong to the first quarter where tan is increasing. W.l.o.g.,
1 assume that both are non-negative (otherwise interchange B and C). Then β ≥ γ and
R1 r 1 − R2 r 2 = · | AD | · (| GB| − | HC|). the desired inequality is equivalent to S ABD − R1 r1 ≥ S ACD − R2 r2 . Now
2
1 1 1 β
Triangle ADG is isosceles because ∠ ADG = 90◦ − ∠ BDE = 90◦ − ∠ DAG. Thus S ABD − R1 r1 = | AD |2 tan β 1 − tan =
2 2 2 2
| AD | = | AG|. Analogously, | AD | = | AH |. Thus | AG| = | AH |. β β
2 tan tan
Subtracting equality | AD |2 + |CD |2 = | AC|2 from | AD |2 + |BD |2 = | AB|2 gives 2
| BD | −
1 2 β 2
= | AD |2 1 − tan = | AD |2 ,
|CD |2 = | AB|2 − | AC|2 which is equivalent to | BD | − |CD | · | BD | + |CD | = | AB| − 2 β 2 β
1 − tan2 1 + tan
| AC| · | AB| + | AC| . Consequently, 2 2

| BD | − |CD | · | BC| = | GB| − | HC| · | AB| + | AC| . whence
 

As | BC| < | AB| + | AC|, we must have | BD | − |CD | ≥ | GB| − | HC|, which gives the  1 
desired inequality. S ABD − R1 r1 = | AD |2 1 − 
β
E 1 + tan
E 2
A A
and, analogously,
F
F  
K 1
S ACD − R2 r2 = | AD |2 1 − γ
.
G L
H G
1 + tan
2
H H′
By β ≥ γ and tan being increasing, the inequality S ABD − R1 r1 ≥ S ACD − R2 r2 follows.
B G′ D C B C D C′
bn − 1
Figure 14 Figure 15 3. Let n be a natural number, n ≥ 2. Prove that if is a prime power for some
b−1
If D does not lie between B and C (see Fig. 15) then assume w.l.o.g. that it is on ray BC. positive integer b then n is prime.
Reflect line segment AC w.r.t. line AD; points C and H transform to some points C′ and bn − 1
H ′ , respectively. Now apply the solution above for triangle ABC′ . The desired claim Solution. Clearly b ≥ 2. Assume that = pl where p is prime, then n ≥ 2 implies
b−1
follows then by using |C′ D | = |CD | and | H ′ C′ | = | HC|. l ≥ 1. If n = xy where both x and y are greater than 1 then consider the representation
Solution 2. Denote ∠ BAD = β and ∠CAD = γ. Then
bxy − 1 bxy − 1 by − 1 by − 1
1 1 = y · = (1 + b y + . . . + b y ( x − 1) ) · .
S ABD = · | AD | · | BD | = | AD |2 tan β. b−1 b −1 b−1 b−1
2 2
As the product is a power of p, both factors must be powers of p. As x > 1 and y > 1,
As in Solution 1, show that quadrangle BEAD is cyclic. Let K be the point of intersec- both factors are multiples of p. Then by − 1 is a multiple of p. Thus all addends in the
| AB| | AD | first factor are congruent to 1 modulo p which implies that the first factor is congruent
tion of its diagonals. As R1 = , we get R1 = . Furthermore, r1 = | GK | and
2 2 cos β to x modulo p. Hence x is divisible by p. As x was an arbitrary non-trivial factor of n,
∠BDE ∠BAD β β β this shows that n = pm for a positive integer m.
∠GDK = = = . Thus r1 = | DK | tan = | AD | sin β tan . Conse-
2 2 2 2 2 Now consider the representation
quently,
m m 2
| AD | β 1 β bp − 1 bp − 1 bp − 1 bp − 1
R1 r 1 = · | AD | sin β tan = | AD |2 tan β tan . = p m −1 ·...· p · .
2 cos β 2 2 2 b−1 b −1 b −1 b−1
21 22
bp − 1 Taking y = a in the initial relation, we get f ( x ) = a + f ( x + 1), and so g( x − 1) − g( x ) =
Each factor is both greater than 1 and a power of p. As is a positive integral
b−1 a − 1 for all x. Since a 6= 1, g is unbounded and by continuity, takes all real values, so
power of p, the numerator is divisible by p, i.e., b p ≡ 1 (mod p). By Fermat’s little f (z) = 1 + z for all z.
theorem, b p ≡ b (mod p). Thus b ≡ 1 (mod p) and b − 1 is divisible by p. But then Let now a = 1, i.e., f (1) = 0. Then x = 0 yields f ( f (y)) = y for all reals y. Taking now
the numerator b p − 1 must be divisible by p2 , i.e., b p ≡ 1 (mod p2 ). If m ≥ 2 then the
2 y = f (1 − x ) in the initial relation, we get f ( x + f ( f (1 − x ))) = f (1 − x ) + f ( x + 1),
bp − 1 or 0 = f (1 − x ) + f ( x + 1). Finally, taking y = 1 − x yields f ( x + f (1 − x )) = 1 −
representation above contains factor p = 1 + b p + . . . + b p( p−1) . On one hand,
b −1 x + f ( x + 1), so f ( x + f (1 − x )) = 1 − x − f (1 − x ). Let h( x ) = x + f (1 − x ), then
this is congruent to p modulo p2 as all addends are congruent to 1. On the other hand, f (h( x )) = 1 − h( x ) holds for all x. Replacing x with − x and taking y = 1 in the initial
this factor is a power of p while being greater than p, hence it is a multiple of p2 . This relation, we get f (− x ) = 1 + f (1 − x ), so h( x + 1) − h( x ) = 2. Again, h is continuous
contradiction shows that m = 1, qed. and must take all real values, so f (z) = 1 − z for all z.
Remark 1. Fermat’s little theorem can easily be avoided in the solution. Cutting this It is straightforward to verify that both solutions indeed satisfy the initial relation.
out from the solution above, it still shows that if m ≥ 2 then b p − 1 is not divisible by
bp − 1 6. Consider a 10 × 10 grid. On every move, we colour 4 unit squares that lie in the
p2 . Continuing from this, we see that is divisible by p but not by p2 . Hence this
b−1 intersection of some two rows and two columns. A move is allowed if at least one of the
factor must be p. Now 4 squares is previously uncoloured. What is the largest possible number of moves that
bp − 1 can be taken to colour the whole grid?
= 1 + b + . . . + b p−1 > b p−1 ≥ 2 p−1 ≥ p
b−1 Answer: 81.
gives a contradiction. Solution. By always choosing the first line, the first column and a square of the remaining
Remark 2. In the special case b = 2, l = 1, the problem reduces to the well-known fact 9 × 9 grid as the lower right square, the whole grid can be coloured in 81 moves.
that a Mersenne’s number Mn can be prime only if n is prime. We now prove that it is not possible to make more than 81 moves. Consider a sequence
of moves. Select for each move one square that is chosen for the first time during this
Second day move and colour the remaining squares already before starting the sequence. Then,
take all squares that were not selected and colour them in advance, i.e., already before
starting the sequence of moves. Since all selected squares must be different, every move
4. In square ABCD, points E and F are chosen in the in- D F C in the sequence now colours exactly one square.
terior of sides BC and CD, respectively. The line drawn Next, consider a bipartite graph with the 10 rows and 10 columns as vertices. Every time
from F perpendicular to AE passes through the intersec- a square is coloured, draw an edge between the row and the column corresponding to
tion point G of AE and diagonal BD. A point K is chosen this square. We claim that the graph is connected before we start the sequence of moves.
on FG such that | AK | = | EF|. Find ∠EKF. Indeed, suppose that during some move, we pick rows (a, b) and columns (c, d), such
that only the square (b, d) is coloured for the first time, i.e., we add the edge (b, d). But
Answer: 135◦ . K
then b is already connected with d through b − c − a − d, so the number of connected
Solution. Since AGFD is a cyclic quadrilateral (see Fig. 16), E components does not decrease. Since the graph of a fully coloured grid is connected, it
∠GAF = ∠GDF = 45◦ and ∠GFA = ∠GDA = 45◦ , so G
must also be connected in the beginning. But a connected graph with 20 vertices must
triangle AGF is isosceles and | GA| = | GF|. Now, right A B have at least 19 edges, so we can add only 100 − 19 = 81 new edges, and hence any
triangles AGK and FGE are congruent, and | GK | = | GE|,
sequence can have at most 81 moves.
so triangle GKE is also isosceles. Finally, ∠GKE = 45◦ and Figure 16
∠EKF = 180◦ − ∠GKE = 135◦ .
5. Find all continuous functions f : R → R such that for all reals x and y
f ( x + f (y)) = y + f ( x + 1).
Answer: f ( x ) = 1 + x and f ( x ) = 1 − x.
Solution. Taking y = − f ( x + 1), we see that there is a value a such that f (a) = 0. We
consider two cases.
Let first a 6= 1. Taking y = x + 1, we get f ( x + f ( x + 1)) = x + 1 + f ( x + 1). Let
g( x ) = x + f ( x + 1), then f ( g( x )) = 1 + g( x ) for all x. Since f is continuous, so is g.
23 24
2007/2008
Tartu 2008
WE THANK:
Problem authors: Juhan Aru, Oleg Koshik, Toomas Krips, Kaie Kubjas, Härmel Nestra,
Hendrik Nigul, Uve Nummert, Reimo Palm, Oleksandr Rybak (Ukraine),
Laur Tooming, Jan Willemson, Indrek Zolk
Translation: Emilia Käsper, Ago-Erik Riet
Edition: Reimo Palm, Indrek Zolk
Estonian Mathematical Olympiad
Mathematics Contests in Estonia

The Estonian Mathematical Olympiad is held annually in three rounds – at the school,
town/regional and national levels. The best students of each round (except the final)
are invited to participate in the next round. Every year, about 110 students altogether
reach the final round.
In each round of the Olympiad, separate problem sets are given to the students of each
grade. Students of grade 9 to 12 compete in all rounds, students of grade 7 to 8 partici-
pate at school and regional levels only. Some towns, regions and schools also organise
olympiads for even younger students. The school round usually takes place in Decem-
ber, the regional round in January or February and the final round in March or April in
Tartu. The problems for every grade are usually in compliance with the school curricu-
lum of that grade but, in the final round, also problems requiring additional knowledge
may be given.
The first problem solving contest in Estonia took place already in 1950. The next one,
which was held in 1954, is considered as the first Estonian Mathematical Olympiad.
Apart from the Olympiad, open contests are held twice a year, usually in October and
in December. In these contests, anybody who has never been enrolled in a university
or other higher education institution is allowed to participate. The contestants compete
in two separate categories: the Juniors and the Seniors. In the first category, students
up to the 10th grade are allowed to participate; the other category has no restriction.
Being successful in the open contests generally assumes knowledge outside the school
curriculum.
According to the results of all competitions during the year, about 20 IMO team candi-
dates are selected. IMO team selection contest for them is held in April or May. This
contest lasts two days; each day, the contestants have 4.5 hours to solve 3 problems,
similar to the IMO. All participants are given the same problems. Some problems in our
selection contest are at the level of difficulty of the IMO but somewhat easier problems
are usually also included.
The problems of previous competitions can be downloaded from
http://www.math.olympiaadid.ut.ee/eng.
Besides the above-mentioned contests and the quiz “Kangaroo” some other regional
competitions and matches between schools are held as well.
*
This booklet contains problems that occurred in the open contests, the final round of
national olympiad and the team selection contest. For the open contests and the final
round, selection has been made to include only original and interesting problems. The
team selection contest, containing only original problems, is presented entirely.
1
OC-1. An n-boomerang consists of 2n − 1 unit squares arranged in an

L-shape with both legs of length n (n = 4 in the figure). Find all integers
n > 2 for which there exists a rectangle with integer side lengths that can
be partitioned into n-boomerangs. (Juniors.)
Answer: the only suitable integer is 2.
Solution. If n = 2, the rectangle of size 2 × 3 can be partitioned into two 2-boomerangs
(Fig. 1).
Let us prove that if n > 3 then there are no rectangles that can be partitioned into
n-boomerangs. Let ( x, y) denote the unit square located in row x and column y, where
x and y are positive integers. Denote by ( a, b) − (c, d) − (e, f ) the boomerang with end-
squares in unit squares ( a, b) and (e, f ) and corner-square in unit square (c, d). Clearly
(1, 1) has to be covered by an end-square or a corner-square of some boomerang.
If a boomerang covers (1, 1) with its corner-square then (2, 2) can be covered by another
boomerang again with a corner-square or an end-square.
Figure 1 Figure 2 Figure 3 Figure 4
1. If another boomerang covers (2, 2) with a corner-square (Fig. 2) then the

boomerangs covering (n + 1, 1) and (1, n + 1) leave the squares (n + 2, 2) and
(2, n + 2) empty. For any position of the boomerang covering (3, 3), one of (n + 2, 2)
and (2, n + 2) can not be covered anymore.
2. If another boomerang covers (2, 2) with its end-square (Fig. 3) then let it be without
loss of generality positioned as (2, 2) − (2, n + 1) − (n + 1, n + 1). The boomerang
covering (3, 2) is then positioned as (3, 2) − (n + 2, 2) − (n + 2, n + 1). The second
and third boomerang now have an isolated empty rectangle with side lengths less
than n between them.
If the first boomerang covers (1, 1) with its end-square (Fig. 4) then we can assume
without loss of generality that it is positioned as (1, 1) − (1, n) − (n, n). The boomerang
covering (2, 1) must now be positioned as (2, 1) − (n + 1, 1) − (n + 1, n). Those two
have an empty rectangle in between that is too small to fit any boomerangs.
OC-2. Do there exist four different integers a, b, c, d, all greater than one, satisfying
gcd( a, b) = gcd(c, d) and a) ab = cd; b) ac = bd? (Juniors.)
2
Answer: a) yes; b) no.
Solution 1. a) Let x, y, z and w be arbitrary different pairwise co-prime positive integers.
Let a = xy, b = zw, c = xz and d = yw. All these numbers are greater than 1. Then
gcd( a, b) = gcd( xy, zw) = 1 and also gcd(c, d) = 1, whereas ab = cd = xyzw.
b) Assume for contradiction that such a, b, c, d exist. Let s = gcd( a, b) = gcd(c, d). Write
a = a′ s, b = b′ s, c = c′ s, d = d′ s, then gcd( a′ , b′ ) = 1 and gcd(c′ , d′ ) = 1. The equation
ac = bd becomes a′ s · c′ s = b′ s · d′ s, equivalently a′ c′ = b′ d′ . Thus d′ divides a′ c′ and
hence, since c′ and d′ are co-prime, d′ divides a′ . Analogously, since a′ divides b′ d′ and
b′ and a′ are co-prime, a′ divides d′ . It follows that a′ = d′ and a = d. This contradicts
the assumption that all of a, b, c, d are different.
Solution 2. b) Assume for contradiction that such a, b, c, d exist. Write the equation as
a d
= .
b c
When we put equal fractions into lowest terms, we get equal fractions (in lowest terms)
with equal numerators and also equal denominators. The number we divide by is the
greatest common divisor of the numerator and the denominator. Since we are given
gcd( a, b) = gcd(c, d) the denominators and numerators will be divided through by the
same number, that is, the numerators and denominators must be equal to begin with.
Thus a = d and b = c, contradicting the assumption that a, b, c, d are different.
OC-3. How many 5-digit natural numbers are there such that after deleting any one
digit, the remaining 4-digit number is divisible by 7? (Juniors.)
Answer: 8.
Solution. Let M = abcde be a number with the required property. By deleting a and b
we get A = bcde and B = acde, respectively. Since they are divisible by 7, so is their
difference B − A = 1000( a − b), hence a − b is divisible by 7, hence a and b are congruent
modulo 7. Analogously, we have that b and c, that c and d, and finally that d and e are
congruent modulo 7. Thus all the digits are congruent modulo 7.
If M has digits that are at least 7, we can subtract 7 from each such digit to obtain a new
number M′ . It is easy to see that M satisfies the condition in the problem if and only if
M′ does. Since all digits give the same remainder, we are left to consider xxxxx where
0 6 x 6 6. By deleting a digit we get xxxx = x · 1111 that is divisible by 7 only if x = 0.
Indeed, 1111 and 7 are co-prime. Thus every digit of M is either 0 or 7. The first two
digits must be 7 (since the number has 5 digits and any number we get by deleting a
digit has 4 digits), the last three digits can be any of 0 or 7 independently. Thus there
are 2 · 2 · 2 = 8 suitable numbers.
OC-4. A magician wants to do the following trick, using an n-year-old volunteer

from the audience. On a board, the magician writes n different positive integers in a
row. Now, between every two consecutive integers, the volunteer writes the difference
of the inverses of the left-hand and right-hand numbers. He finds that all the differences
are equal. Show that the magician can do the trick with every volunteer who is at least
2 years old. (Juniors.)
Solution. If the volunteer is n years old then the magician can pick a number N that is
divisible by all the integers from 1 to n, e.g. their least common denominator or product,
3
and write on the board the numbers
N N N
, , ..., .
1 2 n
1 2 n
These are different positive integers whose inverses are , , ..., respectively. We
N N N
1
see that the differences of consecutive numbers are all equal to − .
N
OC-5. A squaric is a square that has been divided into 8 equal tri-
angles by perpendicular bisectors of its sides and its diagonals. Each
of those lines divides the squaric into two parts; we can take one of
the parts and reflect it over a second dividing line that is perpendicu-
lar to the original line (equivalently, we rotate along the line by 180◦
in space). Every triangle has been coloured by one of four colours and there are two
triangles of each colour. Show that regardless of the initial colouring, the squaric can be
taken to an end position where at every side of the square both triangles have the same
colour. (Juniors.)
Solution. Denote the positions of triangles by numbers 1 to 8 and axes of reflection by
letters x, u, y, v as seen in Fig. 5. Additionally, let A, B, C, D be the colours used.
y D C C D B C C B D D D D
v u C B B C B D D B C B B C
⇒ ⇒ ⇒ ⇒ ⇒
D B D B D C D C B C B C
6 5 A A A A A A A A A A A A
7 4
x D C B B B B
8 3 D B D C C D
1 2 ⇒ ⇒
C B C D C D
A A A A A A
Figure 5 Figure 6
Assume without loss of generality that the triangle in position 1 has colour A. If the
other triangle of colour A is not in position 2, we take it there by reflections that leave
the triangle in position 1 fixed.
Position of colour A triangle Axes of reflection

3 u, y, x
4 u, x
5 x
6 y, x
7 v, y, x
8 v, x
Now the triangles at the bottom side of the squaric have the same colour.
Assume without loss of generality that the triangle at position 3 has colour B. If the
other triangle of colour B is not located at position 4, we take it there as follows, always
leaving the triangle in position 1 unmoved:
4
Position of colour B triangle Axes of reflection
5 y, v, y
6 v, y
7 y
8 v, y, v, y
Now the bottom and right sides of the squaric have triangles of suitable colours.
Finally assume without loss of generality that the triangle in position 5 has colour C.
If the other triangle coloured C lies in position 6, we are done, since then the triangles
at positions 7 and 8 must have colour D. If the other triangle of colour C does not lie
in position 6, we use the following reflections to solve the squaric, always leaving the
triangle in position 1 unmoved (cf. Fig. 6):
Position of colour C triangle Axes of reflection

7 y, u, y, v, y
8 u, y
OC-6. Is it true that every polynomial P( x ) = am x m + . . . + a1 x + a0 with integer

coefficients whose value P(z) for every integer z is a composite number can be written
as P( x ) = Q( x ) · R( x ), where Q and R are polynomials with integer coefficients, neither
of which is constantly 1 or −1? (Seniors.)
Answer: no.
Solution. Let P( x ) = x2 + x + 4. If a is any integer then P( a) = a2 + a + 4 = a( a + 1) + 4.
One of a and a + 1 has to be even, thus P( a) is even. Since a and a + 1 cannot have
opposite signs, a( a + 1) is non-negative, thus P( a) > 4. Thus P( a) is composite.
Write P( x ) = Q( x ) · R( x ), where Q and R are polynomials with integer coefficients.
Since the leading coefficient of P is 1, the leading coefficients of Q and R are both 1 or
both −1. So if Q or R is constant it has to be 1 or −1. If neither is a constant, since P is
a square polynomial, Q and R have to be linear polynomials. But P has no roots, hence
it is not a product of linear polynomials. Therefore P( x ) = x2 + x + 4 cannot be written
as a product of polynomials both different from −1 and 1.
A
OC-7. Let O be the circumcentre of triangle ABC.
Lines AO and BC intersect at point D. Let S be a point S
on line BO such that DS k AB and lines AS and BC inter-
sect at point T. Prove that if O, D, S and T lie on the same O
circle, then ABC is an isosceles triangle. (Seniors.)
Solution. We have that OAB is an isosceles triangle; so is
T
OSD since DS and AB are parallel (Fig. 7). It follows that B D C
the triangles OAS and OBD are equal, using that |OA| =
|OB|, |OS| = |OD | and ∠SOA = ∠ DOB. Thus ∠OSA = Figure 7
∠ODB, from which it follows that ∠OST = ∠ODT. The
points O, D, S, T are located on a circle, so that the points D and S that are symmetric
with respect to the perpendicular bisector of the segment AB are located on different
sides of the line OT. It follows that OST and ODT are opposite angles of the inscribed
quadrilateral ODTS and their sum is 180◦ . Thus ∠OST = ∠ODT = 90◦ . The altitude
5
from vertex A of the triangle ABC goes through the circumcentre of ABC, so it is also
the perpendicular bisector of BC. It is possible only if | AB| = | AC |.
Remark. If ABC is a right triangle, the validity of the claim depends on definitions used.
By the usual high-school definition a right triangle does not satisfy the assumptions of
the problem: if the right angle is at A, then S = D and we cannot talk about line DS; if
the right angle is at B there is no intersection point T; if the right angle is at C, the lines
DS and AB would coincide but coinciding lines are not considered parallel at school.
OC-8. Wolf and Fox play the following game on a board with a finite number of unit
squares. In the beginning, all squares are white and empty. First, Wolf picks up a game
piece from a pile, and either places it on a white square, paints this square gray and
removes all the other pieces from the board, or places it on an empty gray unit square.
Then, Fox makes a move by the same rules, only her colour is red and not gray. The
players continue taking turns and the last player to make a move wins (assume there is
an infinite supply of game pieces). Who wins if both play optimally? (Seniors.)
Answer: Wolf.
Solution 1. Let Wolf have the following strategy. If there are white squares on the board
he will place a piece on one (and colour it gray), otherwise on an empty gray square.
Let us prove this is a winning strategy. Since Wolf starts occupying white squares and
colours them at every move, then after each of his moves there are more gray than red
squares. After every move of Fox there are at least as many gray squares than red. This
is true until there are no more white squares. The square coloured last contains a piece,
all the other squares are empty and the game continues so that each player places pieces
on empty squares.
If the last white square is coloured by Wolf there will be more gray squares than red.
Although the square coloured last contains a piece and he cannot move there anymore,
there are at least as many gray squares left as red. This means Wolf can move after every
move of Fox. If the last square is coloured by Fox, it will contain a piece and she cannot
move there afterwards. Therefore Wolf will have more empty gray squares than Fox has
empty red ones. In both cases Wolf gets to make the last move.
Solution 2. Wolf can, after the first move of each player, colour a white square on his
second move, and after that copy moves of Fox.
Solution 3. It is clear that the game always ends, so somebody has a winning strategy.
Suppose Fox has a winning strategy. At the first move no player has a choice (they
colour a white square). At the second move Wolf has two choices.
• He colours a new white square. Then there are two gray squares, one containing a
piece, and one empty red square on the board. By assumption, Fox has a winning
strategy.
• He places a piece on the gray square. Fox can only colour a new white square. After
Fox moves there are two red squares, one containing a piece and one empty gray
square on the board. By symmetry now Wolf has a winning strategy.
Contradiction.
OC-9. The teacher gives every student a triple of positive integers. First, every stu-
dent has to reduce the second and third number by dividing them by their greatest com-
6
mon divisor, then reduce the first and third number of the resulting triple by dividing
them by their greatest common divisor, and finally, reduce the first and second number
of the new triple by dividing them by their greatest common divisor. Then, everybody
has to multiply the numbers in the final triple and tell the result to the teacher. It is
known that the initial triples only differ by the order of numbers. Find the greatest
possible number of different correct answers that the students could get. (Seniors.)
Answer: 3.
Solution. Let ( a, b, c) be the initial triple and

c b
d1 = gcd(b, c), d2 = gcd a, , d3 = gcd a, .
d1 d1

b c a b c
After the first and second division, we get triples a, , and , , , re-
d1 d1 d2 d1 d1 d2
spectively.

a b b c
Let us prove that gcd , = d3 . Since and are co-prime, their divisors d2
d2 d1 d1 d1
a b
and d3 are co-prime. Since d3 divides a, it hence divides . Since d3 divides , it
d2 d1
a b a a b
also divides gcd , . On the other hand, since divides a, clearly gcd ,
d2 d1 d2 d2 d1
divides
d3 whichproves the claim. Therefore the triple after the third division is
a b c abc
, , and the correct answer is 2 2 2 .
d2 d3 d1 d3 d1 d2 d1 d2 d3
If we swap b and c in the initial triple, d1 is left unchanged and d2 and d3 are swapped
which leaves the final answer unchanged. Therefore the answer depends only on what
we choose as the first component of the triple. Thus there can not be more than 3 differ-
ent answers.
We get three different correct answers if we pick a triple ( p2 qr, pq2 r, pqr2 ) where p, q
and r are pairwise different primes. Indeed this triple changes as
( p2 qr, pq2 r, pqr2 ) 7→ ( p2 qr, q, r ) 7→ ( p2 q, q, 1) 7→ ( p2 , 1, 1),
giving the answer p2 . By changing cyclically the order of the components in the triple,
answers q2 and r2 are obtained.
OC-10. In a square grid of dimension m × n where m, n > 5, every square has been
coloured black or white. At each step, we can pick some horizontal or vertical strip of
width 1 and odd length that contains squares of both colours, and colour all squares in
this strip by the colour occurring less in the strip. Show that by these steps we can give
all squares the same colour. (Seniors.)
Solution 1. Let the grid have m rows and n columns. At first we shall show that the
squares in each row can be given the same colour. If n is odd we cover a row by one
strip and colour all the squares by the colour that occurs less in that row. If n is even we
cover all squares in the row except for the first one by a strip (of odd length) and give
them one colour. If the first square is coloured differently from the rest of the row we
7
cover the squares 1, 2 and 3 by a strip and give them the same colour, then we cover the
squares 1, 2, 3, 4 and 5 by a strip and colour them by the colour of squares 4 and 5. Now
all the squares in that row have the same colour.
Let us now do the same construction for columns. After that the squares in each column
have the same colour, but also all the colours are equal since after the first stage all
columns looked identical.
Solution 2. At first, prove that a rectangle with dimensions 1 × 5 that contains squares of
both colours can be coloured by each colour. Now we can make the first row monochro-
matic by sequentially choosing the 1 × 5 blocks (or not choosing if a block is already
monochromatic). Then we can similarly give every column the colour of its first square.
Remark. One can find other solutions, precisely by dividing the grid into at most four
sub-grids, with each dimension odd and at least 3, and making the sub-grids and even-
tually the whole grid monochromatic.
OC-11. We are given different positive integers a1 , a2 , . . . , an where n > 3 and every
integer except the first and last one is the harmonic mean of its neighbours. Show that
none of the given integers is less than n − 1. (Seniors.)
1 1 1
Solution. As given, the numbers , , ..., form an arithmetic progression. By
a1 a2 an
1 1 1
symmetry we may assume that > > . . . > , equivalently a1 < a2 < . . . < an .
a1 a2 an
Since a1 , a2 , . . . , an are positive integers,
a − a1 n−2

1 1 1 1 1
> − = ( n − 2) − = ( n − 2) · 2 > .
a2 a2 a n a1 a2 a1 a2 a1 a2
1 n−2
Hence > , and multiplying both sides by a1 a2 gives a1 > n − 2. Since a1 is an
a2 a1 a2
integer, a1 > n − 1. The numbers a2 , . . . , an are greater than a1 and thus greater than
n − 1.
OC-12. Two circles are drawn inside a parallelogram ABCD so that one circle is tan-
gent to sides AB and AD and the other is tangent to sides CB and CD. The circles touch
each other externally at point K. Prove that K lies on the diagonal AC. (Seniors.)
D C
Solution 1. Let O1 and O2 be the centres of
the first and the second circle, respectively
(Fig. 8). Consider the triangles O1 AK and
O1
K O2 O2 CK. Their angles AO1 K and CO2 K are
equal since their sides O1 K and O2 K lie on the
same line and the sides O1 A and O2 C are par-
A B 1 1
allel since ∠O1 AB = ∠ DAB = ∠ BCD =
2 2
Figure 8 ∠O2 CD. As ∠ DAB = ∠ BCD, we have
|O1 A| |O2 C |
= . Therefore the triangles O1 AK and O2 CK are similar. Thus ∠O1 KA =
|O1 K | |O2 K |
∠O2 KC, from which it follows that the points A, K and C are collinear.
Solution 2. Consider the homothety with centre K that takes one of the circles onto the
8
other one. Then the line AB is taken to the line CD and the line AD is taken to the line
CB. Thus the intersection point A of the lines AB and AD goes to the intersection point
C of the lines CD and CB. It follows that the points A, K and C are collinear.
OC-13. Let x and y be arbitrary real numbers.

a) If x + y and x + y2 are rational numbers, can we deduce that x and y are rational
numbers?
b) If x + y, x + y2 and x + y3 are rational numbers, can we deduce that x and y are
rational numbers?
(Seniors.)
√ √
1− 2 1+ 2 5
Solution. a) Pick x = and y = . Then x + y = 1 and x + y2 = . We see
2 2 4
that x + y and x + y2 are rational but x and y are not.
b) Let x + y, x + y2 and x + y3 be rational. If y = 0 or y = 1 then both x and y are
rational. If y 6= 0 and y 6= 1 then the following is also a rational:
( x + y3 ) − ( x + y2 ) y3 − y2 y ( y2 − y )
= = = y.
( x + y2 ) − ( x + y ) y2 − y y2 − y
Thus ( x + y) − y = x is also rational.
OC-14. A sequence ( an ) of natural numbers is given by the following rule:

lcm( an−1 , an−2 )
an = for all n > 2.
gcd( an−1 , an−2 )
It is known that a560 = 560 and a1600 = 1600. Find all possible values of a2007 . (Seniors.)
Answer: a2007 = 140 is the only possible value.
Solution. Explore the behaviour of the sequence in general. Note that it is sufficient to
consider the behaviour of the sequence for each prime factor separately. We have
lcm( p a , pb ) pmax(a,b)
a b
= min ( a,b )
= p| a−b| .
gcd( p , p ) p
Therefore for each prime factor one may consider the behaviour of the sequence of ex-
ponents. Thus explore the properties of the sequence (bn ) defined by bn = |bn−1 − bn−2 |
for all n > 2. It is easy to see that either all terms of the sequence are even or there
is a cycle (even, odd, odd). Thus bn+3 and bn have the same parity. Also observe that
bn+3 6 bn for all n.
Consider now the prime factors appearing in given terms. For prime factor 7 one has
b560 = 1 and b1600 = 0. Since 3 | 1601 − 560, previous observations imply that b1601 = 1.
Then b1602 = |1 − 0| = 1. Now the fact that 3 | 2007 − 1602 leads to b2007 = 1. Hence the
exponent of 7 in a2007 is equal to 1.
For the prime factor 5, we have b560 = 1 and b1600 = 2. Analogously with previous case
we obtain b1601 = 1 and also b1602 = 1. Hence as before b2007 = 1, thus the exponent of
5 in a2007 is 1.
9
The prime factor 2 remains. For this b560 = 4 and b1600 = 6. Examine the possible
values of b1601 . Since it must have the same parity as b560 and may not be greater than
it, the only candidates are 0, 2 and 4. Suppose b1601 = 0. Then both b1601 and b1600 are
divisible by 6. Taking into account the definition of the sequence (bn ) implies that 6
divides also all previous terms, including b560 . This leads to contradiction, thus b1601 is
not 0. Suppose that b1601 = 2. Since consequent terms are even, all terms of the sequence
must be even. Dividing all terms by 2 leads to sequence, that still satisfies the definition,
thus all previously considered observations must be valid. The term b560 transforms to
2 and the term b1601 to 1, that means they have different parity. This is a contradiction
analogously to the cases of previous primes.
The last remaining possibility is b1601 = 4 (it is easy to see that a corresponding sequence
exists). Now performing calculations we obtain b1602 = 6 − 4 = 2, b1603 = 4 − 2 = 2,
b1604 = 2 − 2 = 0, b1605 = 2 − 0 = 2 and further the cycle (2, 0, 2) repeats, therewith the
value of terms with the number divisible by 3 is 2. Thus b2007 = 2, hence the exponent
of 2 in a2007 is 2.
Since the terms a560 and a1600 have no other prime factors, taking preceding into account
implies that the term a2007 neither has other prime factors. Hence the only solution is
a2007 = 7 · 5 · 22 = 140.

Olympiad
FR-1. On a railway connecting cities A and B, trains run at full speed except for two
railway segments, where poor track conditions force them to slow down. If any one of
those two segments were repaired, the average speed of a train between A and B would
increase by a third. How much would the average speed between A and B increase if
both segments were repaired? (Grade 9.)
Answer: 2 times.
Solution. Let the train journey between A and B take time t when neither segment is
repaired. If the first segment was repaired, the average speed would increase by a third,
4 3
in other words, times, so the journey would take time t. Thus, repairing the first
3 4
1 1
segment would save t time. Similarly, repairing the second segment would save t.
4 4
1
Repairing both segments would save t and the average speed would increase 2 times.
2
FR-2. Find all possible values of abc · ( a + b + c), given that bca = ( a + b + c)3 and
b 6= 0. (Grade 9.)
Answer: 2008.
Solution. There exist five three-digit cubes: 125 = 53 , 216 = 63 , 343 = 73 , 512 = 83 and
729 = 93 . Of these, only 512 satisfies bca = ( a + b + c)3 . Thus, a = 2, b = 5, c = 1 and
abc · ( a + b + c) = 251 · (2 + 5 + 1) = 2008.
10
O3
O2
C
B A
O1 B
O3 O1
A
O2 C D
O4
Figure 9 Figure 10
FR-3. a) Circles c1 and c2 touch externally at point A, circles c2 and c3 touch externally
at point B, and circles c3 and c1 touch externally at point C. Suppose that triangle ABC
is equilateral. Are the radii of c1 , c2 and c3 necessarily equal?
b) Circles c1 and c2 touch externally at point A, circles c2 and c3 touch externally at point
B, circles c3 and c4 touch externally at point C, and circles c4 and c1 touch externally at
point D. Suppose that ABCD is a square. Are the radii of c1 , c2 , c3 and c4 necessarily
equal? (Grade 9.)
Solution. a) Let O1 , O2 and O3 be the midpoints of c1 , c2 and c3 , respectively (Fig. 9).
Triangles O1 CA, O2 AB and O3 BC are isosceles, as each triangle has two radii of the
same circle as its two sides. Let ∠O1 CA = ∠O1 AC = α, ∠O2 AB = ∠O2 BA = β and
∠O3 BC = ∠O3 CB = γ. Suppose that triangle ABC is equilateral, so ∠ ABC = ∠ BCA =
∠CAB = 60◦ . As ∠O1 AC + ∠CAB + ∠O2 AB = 180◦ , we have α + β = 120◦ . Similarly,
β + γ = 120◦ and γ + α = 120◦ . The last three equations together give α = β = γ = 60◦ .
Thus, triangles O1 CA, O2 AB, O3 BC are equilateral and as ABC is also equilateral, they
are in fact equal.
b) Choose the midpoints of the three circles as O1 (6; 0), O2 (0; 3), O3 (−6; 0), O4 (0; −3)
(Fig. 10). Then O1O2O3O4 is a rhombus and points A(2; 2), B(−2; 2), C (−2; −2),
D (2; −2) on the sides of the rhombus form a square. Take each vertex of the rhom-
bus to be the midpoint of a circle drawn through the two closest vertices of the square.
Then these four circles touch externally at A, B, C, D, yet they do not all have equal
radii (e.g., |O1 A| 6= |O2 A|).
FR-4. Let n be a positive integer. Rays originating from the mid-

point X of a revolving stage divide the stage into 2n + 2 equal sec-
tors, coloured alternatingly black and white (n = 3 in the figure).
Similarly, equally spaced rays originating from X divide the fixed
floor area outside the revolving stage into 2n alternatingly black-
and-white sectors. Prove that regardless of the position of the re-
volving stage, there exists a sector of the stage that is completely
embraced by a single fixed floor sector of the same colour. (Grade 9.)
Solution. Consider the rays dividing the revolving stage into 2n + 2 sectors. Since the
rest of the floor is divided into 2n sectors, there exist two neighbouring rays that pass
through the same floor sector. If the revolving stage sector between those two rays has
11
the same colour as the floor sector, we are done. If, on the other hand, the two sectors
are of different colour, then the stage sector symmetrically opposite to the original sector
satisfies our conditions. This stage sector is completely embraced by the floor sector
symmetrically opposite the original floor sector, however, when turning 180◦ , the stage
sectors change colour n + 1 times, while the floor sectors change colour only n times, so
the two sectors symmetrically opposite to the original sectors are of the same colour.
FR-5. Circles c1 and c2 with midpoints O1 and O2 intersect at point P. Circle c2 inter-
sects O1O2 at point A. Prove that there exists a circle touching c1 at P and O1O2 at A iff
∠O1 PO2 = 90◦ . (Grade 10.)
Solution. Assume first there exists a circle c touching c1 at P and O1O2 at A (Fig. 11). Let
O be the midpoint of c, then line O1O passes through P. Consider triangles OPO2 and
OAO2 . Clearly |OP| = |OA| and |O2 P| = |O2 A| as the radii of circles c and c2 . Also,
the triangles share a third side OO2 , so they are equal. As ∠OAO1 = 90◦ , we must also
have ∠OPO2 = 90◦ .
Assume now ∠O1 PO2 = 90◦ . Then line O2 P is perpendicular to radius O1 P and thus
touches c1 at P. As |O2 P| = |O2 A|, the line drawn through P perpendicular to O2 P
and the line drawn through A perpendicular to O2 A intersect at a point O such that
|OP| = |OA|. A circle with midpoint O and radius OP then touches c1 at P and O1O2
at A.
P
O C D
O1 A O2 A B
D′
Figure 11 Figure 12
FR-6. Do there exist 5 different points in the plane such that all triangles with vertices
at these points are right triangles and
a) no four of the chosen points lie on the same line;
b) no three of the chosen points lie on the same line?
(Grade 10.)
Solution 1. a) Choose four vertices of a square and the intersection point of its diagonals.
b) Consider a set of points in the plane such that all triangles with vertices in those
points are right triangles and no three points lie on the same line. Choose some two
points A and B; all the remaining points then lie either on the circle with diameter AB,
or on either line perpendicular with AB drawn through endpoint A or B (Fig. 12). At
most four points (including A and B) can lie on the circle, since any two of such three
12
points must be the two endpoints of some diameter. Also, in addition to A and B, there
can be at most one point on either perpendicular.
Suppose now that C and D are two points satisfying our conditions such that C lies on
the circle and D lies on one of the two lines, say, on the perpendicular drawn through
B. If C and D lie on the same side of AB, then ∠ ACD > ∠ ACB = 90◦ , and ACD is
not a right triangle. If, on the other hand, C and D lie on opposite sides of AB, then
∠ DBC > ∠ DBA = 90◦ , so DBC is not a right triangle. Thus, either all points lie on the
circle, or they all lie on the two perpendiculars. In either case, there can be at most 4
such points.
Solution 2. b) Assume by contradiction that it is possible to choose 5 points satisfying
the conditions. Since each three points form the vertices of a right triangle, there are 10
right triangles with vertices in these 5 points. Thus, there exists a point O that is the
vertex of at least two right angles. Let OAB and OXY be the two triangles with right
angles at O.
Now, if either X or Y was lying on line OA, the other point would have to lie on AB.
But then we would have three points on the same line, since at most one of X and Y
can coincide with A or B. Analogously, neither X nor Y can lie on OB. Now if X (resp.
Y) and B lie on opposite sides of line OA, then XOB (resp. YOB) is an obtuse triangle.
Similarly, X and A (or Y and A) cannot lie on opposite sides of OB. Thus, both X and Y
must lie within the right angle AOB, but then XOY is not a right triangle.
FR-7. Call a rectangle splittable if it can be divided into two or more square parts such
that the side of each square is of integral length and there is a unique square with small-
est side length. Find the dimensions of the splittable rectangle with the least possible
area. (Grade 11.)
Answer: 5 × 7.
Solution. The unique smallest square of the partition cannot lie
on the side of the rectangle, for it would have a larger square
on either side and the area between the two squares could only
be filled by squares no larger than the smallest square. Anal-
ogously, the smallest square cannot lie in the corner. Now, the
distance between the smallest square and any side of the triangle
must be at least one unit longer than the side length of the small- Figure 13
est square, for otherwise the area between the smallest square
and the side could not be filled. Thus, the length of each side of the rectangle is at least
1 + 2 + 2 = 5 and each square on a side must have side length at least 2. Thus, if the
rectangle has a side of length 5, on this side we must have a square with side length at
least 3. But then the distance between the smallest square and this side is at least 3. The
same holds for the opposite side of length 5, so the length of the longer side must be at
least 1 + 3 + 3 = 7. It is possible to partition a 5 × 7 rectangle in the desired way (see
Fig. 13). The area of this rectangle is 35, which is indeed the smallest possible area, since
any rectangle with shorter side length greater than 5 has area at least 6 · 6 = 36.
FR-8. Circles c1 and c2 with respective diameters AB and CD of different length touch
externally at point K. An external tangent common to both circles touches c1 at A and
c2 at C. Line BD intersects c1 again at point L and c2 at point M. Prove that triangles
13
A C
C A
O1 K O1 K O2
O2
M L
D B
L M
B D
Figure 14 Figure 15
AKL and BKM are similar. (Grade 11.)

Solution. Let O1 and O2 be the midpoints of circles c1 and c2 , respectively (Fig.
14 and 15). The isosceles triangles BO1 K and CO2 K are similar, since their correspond-
ing legs are parallel: AB k CD and point K lies on O1O2 . Thus, the bases are also
parallel, so K lies on BC. Now on one hand, ∠KAL = ∠KBM, while on the other hand
∠ ALK = ∠ ABK = ∠KCD = ∠KMB. We see that triangles AKL and BKM have two
pairs of equal angles and hence are indeed similar.
FR-9. Let a, b, c be real numbers. Prove that a2 + 4b2 + 8c2 > 3ab + 4bc + 2ca. When
does equality hold? (Grade 11.)
Answer: Equality holds iff a = 2b = 4c.
Solution. Bringing all terms to the lhs, we get
a2 + 4b2+ 8c2 − 3ab − 4bc− 2ca =

3 2 2 2 2
2 1 2
= a − 3ab + 3b + b − 4bc + 4c + 4c − 2ca + a =
4 4
√ !2 !2
3 √ 1
= a − 3b + (b − 2c)2 + 2c − a > 0.
2 2
√
3 √ 1
Equality holds iff equations a = 3b, b = 2c, 2c = a hold simultaneously, in other
2 2
words, iff a = 2b = 4c.
Remark. One may find other solutions, precisely using AM-GM on (1.5a2 , 6b2 ),
!2 √ !2
3 7 √
(0.5a2 , 8c2 ), (2b2 , 8c2 ), grouping the lhs as a − b − c + b − 7c , or con-
2 2
sidering the lhs as a quadratic trinomial in a, b and c and investigating the respective
discriminants.
FR-10. Does there exist a convex hexagon ABCDEF such that the circumcircles of
triangles ABC, CDE and EFA intersect at a common point inside the hexagon? (Grade
11.)
Answer: no.
14
B
Solution. Suppose that such a hexagon exists and let O be

the common intersection point of the three circumcircles A
(Fig. 16). Then quadrilaterals ABCO, CDEO and EFAO are
O
all inscribed, so ∠ BAO + ∠ BCO = 180◦ , ∠ DCO + ∠ DEO = C
180◦ and ∠ FEO + ∠ FAO = 180◦ . Adding the three equa- F
tions, we get ∠ BCD + ∠ DEF + ∠ FAB = 3 · 180◦ . On the
other hand, all internal angles of a convex hexagon are D
less than 180◦ , so the sum of the three angles cannot reach
E
3 · 180◦ , contradiction.
Figure 16
FR-11. Find the least possible value of (1 + u2 )(1 + v2 ),
where u and v are real numbers satisfying u + v = 1. (Grade
12.)
25
Answer: .
16
1 1
Solution 1. Write u = + x and v = − x. Then
2 2
1 2 1 2 1

2 2 2
(1 + u )(1 + v ) = 1 + +x 1+ −x = 1+ +x +x ·
2 2 4
2

1 5 25 5 2 25 3 2
· 1 + + x2 − x = + x2 − x2 = + x + x4 − x2 = + x + x4 .
4 4 16 2 16 2
3
Since x2 and x4 are both non-negative, the obtained sum is minimal when x = 0. The
2
25
latter gives (1 + u2 )(1 + v2 ) = .
16
Solution 2. As u + v = 1, we get
(1 + u2 )(1 + v2 ) = 1 + u2 + v2 + u2 v2 =
= 1 + (u + v)2 − 2uv + u2 v2 = 2 − 2(uv) + (uv)2 .
Let s = uv. For a fixed sum u + v = 1, the product s = uv is maximal when u = v. Thus,
1 2 1
we can bound s 6 = . Now, we need to minimize 2 − 2s + s2 = (s − 1)2 + 1,
2 4i
1 1
which is decreasing in −∞; and obtains the minimum at s = .
4 4
Solution 3. Notice that u = 1, v = 0 gives (1 + u2 )(1 + v2 ) = 2, while for any u > 1
or v > 1 (or equivalently, v < 0 or u < 0), (1 + u2 )(1 + v2 ) > 2. Thus, we may
restrict to the case u, v ∈ [0, 1]. Now consider a triangle ABC such that its side BC and
altitude AH (Fig. 17) have unit length and H divides BC to parts of length u and v. Then
1 1 1
u + v = 1 and the law of sines gives · | AB| · | AC | · sin ∠ BAC = · | BC | · | AH | = ,
2 2 2
1
so (1 + u2 )(1 + v2 ) = | AB|2 | AC |2 = . The value sin ∠ BAC is maximal when
sin2 ∠ BAC
H is the midpoint of BC. Indeed, let c be the circumcircle of ABC in the case when
H is the midpoint. For any other configuration, A lies outside this circle c and thus
the angle BAC is smaller (note that the angle BAC is always acute as BC cannot be the
5
longest side of ABC). Now if H is the midpoint of BC, we get | AB|2 = | BC |2 = , and
4
15
25
| AB|2 | BC |2 = .
16
Remark. One may find other solutions, precisely determining the minima of g(u) =
(1 + u2 )(1 + (1 − u)2 ) using derivatives, or writing out Jensen’s inequality for a convex
function l ( x ) = ln(1 + x2 ).
A
A A′ D
O1 O
O2
B
E
B u H v C C
Figure 17 Figure 18
FR-12. In a convex quadrilateral ABCD, | AB| = | BC | = |CD |. Diagonals AC and

BD intersect at point O. Prove that the circumcircles of triangles AOB and COD are
mutually tangent iff AC is perpendicular to BD. (Grade 12.)
Solution. Assume first that the circumcircles of AOB and COD are mutually tangent
(Fig. 18). Draw a tangent common to both circles from O, and let the tangent line inter-
sect BC at E. Then | AB| = | BC | implies ∠EOB = ∠OAB = ∠ BCO. Similarly, ∠EOC =
∠ODC = ∠CBO. Triangle BOC now gives ∠EOB + ∠EOC + ∠OBC + ∠OCB = 180◦ or
2∠EOB + 2∠EOC = 180◦ , so finally ∠ BOC = ∠EOB + ∠EOC = 90◦ .
Assume now AC is perpendicular to BD. The circumcentres O1 and O2 of right triangles
AOB and COD lie on the respective hypotenuses AB and CD. We have ∠O1OA =
∠O1 AO = ∠ BCO and ∠O2OD = ∠O2 DO = ∠CBO. As BOC is also a right triangle,
∠ BCO + ∠CBO = 90◦ . Finally, ∠O1OA + ∠ AOD + ∠O2OD = ∠ BCO + 90◦ + ∠COB =
180◦ , so the circumcircles of AOB and COD touch at O.
FR-13. All natural numbers that are less than a fixed positive integer n and relatively
prime to it are added one-by-one in increasing order. How many intermediate sums
(starting from the lonely first addend and including the final sum) are divisible by n, if
a) n is an odd prime number?
b) n is the square of an odd prime number?
(Grade 12.)
Answer: a) 1; b) 1.
Solution. a) Let n = p where p is an odd prime. The addends are 1, 2, . . . , p − 1, thus the
intermediate sums have form 1 + . . . + k where 1 6 k 6 p − 1. Suppose p | 1 + . . . + k.
k ( k + 1)
Then p | k (k + 1) as 1 + . . . + k = . Thus either p | k or p | k + 1. This is possible
2
only for k = p − 1.
b) Let n = p2 where p is an odd prime. Any number less than p2 is added if and only
if it is not divisible by p. Divide the addends into p groups, each consisting of p − 1
16
members:
1, 2, . . . . . . , p − 1,
p + 1, p + 2, . . . . . . , p + p − 1,
.......................................................
( p − 1) p + 1, ( p − 1) p + 2, . . . . . . , ( p − 1) p + p − 1.
Let an intermediate sum be divisible by p2 ; then it is divisible by p, too. As the rows
are equivalent modulo p, we can use part a) of the problem to deduce that the last
intermediate sum of every row is divisible by p and the others are not. Hence, in the
whole intermediate sum under consideration, the last row cannot occur partially, i.e.,
our intermediate sum consists of whole rows of addends.
p ( p − 1)
The sum of the elements of the first row is . The sum of the numbers of each
2
following row is by p( p − 1) larger than that of the row preceding it. Thus the row sums
p ( p − 1) p ( p − 1) p ( p − 1)
are 1 · ,3· ,5· etc.. The sum of the numbers of the first i rows
2 2 2
p ( p − 1) p ( p − 1)
is (1 + 3 + . . . + (2i − 1)) · = i2 · .
2 2
p ( p − 1)
If p2 | i2 · then p | i2 · ( p − 1), implying p | i. Hence i = p, i.e., the sum is the
2
final sum.
Remark. It is easy to show that the entire sum of ϕ(n) addends is divisible by n for all
integers n > 2. If n is neither a prime nor the square of a prime then there can be more
intermediate sums divisible by n. For example, if n = 16 then the intermediate sum
1 + 3 + 5 + 7 containing only half of the addends is divisible by 16. If n = 27 or n = 39
then two intermediate sums in addition to the final sum are divisible by n, etc.
FR-14. Consider a point X on line l and a point A outside the line. Prove that if there
exists a point Z1 on l such that the three side lengths of triangle AXZ1 are all rational,
then there exist two other points Z2 and Z3 on l such that the side lengths of triangles
AXZ2 and AXZ3 are also all rational. (Grade 12.)
Solution. We consider three separate cases.
• If AXZ1 is equilateral, i.e., | AX | = | AZ1 | and ∠XAZ1 = 60◦ , then take Z2 on
the extension of Z1 X across X such that | XZ2 | = 0.6 | AX |, and take Z3 to be the
reflection of Z2 across the perpendicular bisector of XZ1 (Fig. 19). The law of
cosines implies | AZ2 | = | AX |2 + | XZ2 |2 − 2 · | AX | · | XZ2 | · cos 120◦ = 1.96 | AX |2 ,
so | AZ2 | = 1.4 | AX | and the side lengths of AXZ2 as well as AXZ3 are rational.
Let now AXZ1 be not equilateral.
A A
Z2 X Z1 Z3 Z3 X Z1 Z2 X Z3 Z1 Z2
Figure 19 Figure 20 Figure 21
17
• Assume | XZ1 | 6= | AX | and | XZ1 | 6= | AZ1 |. Choose Z2 on ray XZ1 such that
∠Z2 AX = ∠ AZ1 X, and choose Z3 on ray Z1 X such that ∠Z3 AZ1 = ∠ AXZ1
(Fig. 20). Points Z2 and Z1 differ since ∠ AZ1 X 6= ∠Z1 AX, and points Z3 and X
differ since ∠ AXZ1 6= ∠XAZ1 . Triangles Z2 XA and Z3 AZ1 are similar to triangle
| AX | | Z A|
AXZ1 with similarity ratios and 1 , so their side lengths are rational, and
| Z1 X | | Z1 X |
| XZ3 | is rational, too.
• Assume w.l.o.g. | XZ1 | = | AZ1 | (Fig. 21). Choose Z2 as before, then Z2 differs from
X and Z1 and the side lengths of AXZ2 are rational. Take Z3 to be the reflection of
Z1 across the perpendicular bisector of XZ2 . Then Z3 differs from Z1 , as AXZ1 is
not an isosceles right triangle. Triangle AXZ3 is equal to triangle AZ2 Z1 , and the
latter has rational side lengths.
FR-15. A finite number of thin straight pins are attached to a vertical wall such that
no two pins touch each other. If a pin is detached, it slides straight down the wall,
keeping its original angle to the floor. Prove that there exists a pin that can slide freely
down to the floor without being stopped by any of the other pins. (Grade 12.)
Solution 1. If there exists a vertical pin that can slide freely, we are done. Assume now
that no vertical pin can slide down freely. Draw a horizontal line l where the wall meets
the floor and project the endpoints of each pin onto l. If there are no pin points between
some left endpoint and l, colour the projection point on l blue. Similarly, if some right
endpoint is the lowermost pin point on its projection line, colour the corresponding
projection point yellow. Clearly, the leftmost projection point on l is coloured blue, while
the rightmost point is yellow. Thus, moving on l left-to-right, some two consecutive
coloured points must be blue and yellow, respectively. We claim that these points are
the two endpoints of the same pin, and thus this pin can slide down. Indeed, on the
segment between the blue and the yellow point, any lowest pin point above l must
belong to the same pin as the left (blue) and the right (yellow) endpoint.
Solution 2. We prove by induction on the number of pins. The claim clearly holds for
one pin. Assume there is more than one pin, and consider three cases.
1. There exists a pin p such that below every point of p, there is a point of some other
pin. Remove p, then by the induction assumption, some pin v can slide down freely.
Now, put p back. Then p cannot be the only pin stopping v, since below every point
of p, there is a point of another pin, and at least one of those pin points should also
be stopping v.
2. There exists a pin p which cannot slide down freely such that all pins stopping p
lie entirely below p. Remove p and all the remaining pins that do not lie below
p. Then, there must exist a pin v that can slide down freely, but then v can also
slide down in the original configuration, since the only pins possibly stopping it
are those below p.
3. If the previous two cases do not hold, then each pin has some points that have no
other pin points below them, and either the pin can slide down or one of the pins
stopping it does not lie entirely below this pin. Let p be the pin with the rightmost
point amongst all pins (if there is more than one such pin, choose the one with the
topmost such point). Remove p. By the induction assumption, there now exists a
18
pin v that can slide down. Put p back. If v can still slide down, we are done. In the
other case, the only pin stopping it is p. Since v must have some free points with
no other pin points below, the left endpoint of v must reach further left than the left
endpoint of p. We claim that now p can slide down. Indeed, any pin points below
p that also lie below v cannot be stopping p, as they would also be stopping v. But
any other pin below p can also not be stopping p, as p has the rightmost endpoint,
so any pins stopping p should lie completely below p.
Remark 1. One can find other solutions, precisely using a directed graph with pins as
its vertices and an edge from vertex a to vertex b if pin b is stopping pin a: it suffices to
prove that this graph does not contain any directed cycles.
Remark 2. The claim does not always hold when the pins are not straight. For example,
two half-circle pins can be placed to mutually stop each other.
IMO team selection contest

First day
TS-1. There are 2008 participants in a programming competition. In every round,

all programmers are divided into two equal-sized teams. Find the minimal number of
rounds after which there can be a situation in which every two programmers have been
in different teams at least once.
Answer: 11.
Solution 1. After every round consider the biggest set of programmers where the pro-
grammers have been in the same team in all rounds so far. Before the first round it
consists of 2008 programmers. With every round its size can decrease by at most twice,
since the programmers belonging to it are divided among two teams in the new round
and at least half of them will again be in the same team. Thus the number of rounds is
at least log2 2008, i.e. at least 11.
We shall show that 11 rounds suffice. Order the 2008 programmers in some way and
add both at the end and at the beginning 20 imaginary programmers. Number the pro-
grammers by 11-digit binary numbers from 0 to 2047, adding leading zeros if necessary.
In round i the programmers are divided into teams according to the ith digit of their
number. In every round the kth imaginary programmer from the beginning and the kth
imaginary programmer from the end are in different teams since their corresponding bi-
nary numbers have all digits different. Hence both teams have in every round an equal
number of programmers. Also, every pair of programmers belong to different teams in
at least one round since their numbers differ in at least one binary digit.
Solution 2. Let us prove by induction on k that if the number of programmers 2n sat-
isfies the inequalities 2k−1 < 2n 6 2k then k rounds suffice. If k = 1 then we have 2
programmers and clearly one round is enough. Assume the claim is true for some k.
Assume there are 2n programmers where 2k < 2n 6 2k+1 . Divide them into two groups
of s = 2k and t = 2n − s programmers, and number them by 1, . . . , s and s + 1, . . . , s + t
respectively. By the induction hypothesis the programmers in both the first and the sec-
ond group can be divided into equal-sized teams so that after k rounds every two (in
19
each group) have competed against each other at least once. For rounds 1, . . . , k make
up two new equal teams by taking one team corresponding to each group and putting
them together. Assume without loss of generality that in round k one team consists of
s
the programmers of the first group with numbers 1, . . . , and of the second group with
2
t
numbers s + 1, . . . , s + . In the new round swap and make up a team of programmers
2
s t
with numbers 1, . . . , and s + + 1, . . . , s + t. We can check that every two program-
2 2
mers (also from different groups) have now been in different teams at least once.
The other part can be done like in Solution 1.
TS-2. Let ABCD be a cyclic quadrangle whose midpoints of diagonals AC and BD

are F and G, respectively.
a) Prove the following implication: if the bisectors of angles at B and D of the quad-
1
q
rangle intersect at diagonal AC then · | AC | · | BD | = | AG | · | BF | · |CG | · | DF |.
4
b) Does the converse implication also always hold?
Answer: b) No.
Solution 1. a) Let E be the intersection point of the bisectors from B and D. By the
bisector property,
| AB| | AE| | AD |
= = . (1)
| BC | | EC | | DC |
By Ptolemy’s theorem, | AB| · |CD | + | AD | · | BC | = | AC | · | BD |. Using this in (1), we
obtain
2 · | BC | · | AD | = | AC | · | BD |, (2)
2 · | AB| · |CD | = | AC | · | BD |. (3)

| FA| | AC |
Let F be the midpoint of AC. Then ∠ FAD = ∠CAD = ∠CBD. By (2), = =
| AD | 2| AD |
| BC |
. Hence triangles FAD and CBD are similar. Analogously by (3), triangles FAB and
| BD |
| FA| | FB|
CDB are similar. Consequently, triangles FAD and FBA are similar. Thus =
| FD | | FA|
which implies
1
| AC |2 = | FB| · | FD |. (4)
4
| DA| | DC |
By (1), = . Thus bisector property implies that the bisectors of angles at A
| AB| |CB|
and C intersect at diagonal BD. Let G be the midpoint of BD. Analogously to what we
did before, we obtain
1
| BD |2 = | AG | · |CG |. (5)
4
20
The desired claim follows now by multiplying the corresponding sides of (4) and (5)
and finding the square root.
b) Let ABCD be a rectangle where | AB| > | BC |. Clearly | AG | = | BF | = |CG | = | DF | =
1 1
| AC | = | BD |, implying the rhs of the implication of part a) (lhs of the converse). But
2 2
| AB| | AD |
> 1 > shows that the bisectors of angles at B and D do not intersect on
| BC | | DC |
diagonal AC. Hence the converse implication is false.
Solution 2. a) Denote the interior angles of ABCD by ∠ A, ∠ B, ∠C, ∠ D. In triangle DAB,
cosine law gives
| BD |2 = | AB|2 + | AD |2 − 2 · | AB| · | AD | · cos ∠ A.
In triangle BCD, taking into account that ∠C = 180◦ − ∠ A, cosine law gives
| BD |2 = |CB|2 + |CD |2 + 2 · |CB| · |CD | · cos ∠ A.
Multiplying these two equalities leads to
| BD |4 = (| AB|2 + | AD |2 )(|CB|2 + |CD |2 ) − 4 · | AB| · | AD | · |CB| · |CD | · cos2 ∠ A +

+ 2 (| AB|2 + | AD |2 ) · |CB| · |CD | − (|CB|2 + |CD |2 ) · | AB| · | AD | · cos ∠ A.
−→ −→ −→
On the other hand, 2 AG = ( AB + AD ) implies
4 · | AG |2 = | AB|2 + | AD |2 + 2 · | AB| · | AD | · cos ∠ A
and, analogously (using ∠C = 180◦ − ∠ A),
4 · |CG |2 = |CB|2 + |CD |2 − 2 · |CB| · |CD | · cos ∠ A.
Multiplying these equalities leads to
16 · | AG |2 · |CG |2 =
= (| AB|2 + | AD |2 )(|CB|2 + |CD |2 ) − 4 · | AB| · | AD | · |CB| · |CD | · cos2 ∠ A −

− 2 (| AB|2 + | AD |2 ) · |CB| · |CD | − (|CB|2 + |CD |2 ) · | AB| · | AD | · cos ∠ A.
If the bisectors of angles by B and D intersect on diagonal AC, the bisector property
| AB| | AD |
gives = or | AB| · |CD | = | AD | · |CB|. Thus
|CB| |CD |
(| AB|2 + | AD |2 ) · |CB| · |CD | − (|CB|2 + |CD |2 ) · | AB| · | AD | =

= | AB| · | AD | · |CB|2 + | AD | · | AB| · |CD |2 − (|CB|2 + |CD |2 ) · | AB| · | AD | = 0.
Consequently, | BD |4 = 16 · | AG |2 · |CG |2 . Considering triangles ABC and CDA, we

obtain in a similar way that | AC |4 = 16 · | BF |2 · | DF |2 . Multiplying the last equalities
and taking the 4th root from both, we obtain the desired result.
21
TS-3. Let n be a positive integer and x, y positive real numbers such that x n + yn = 1.
Prove the inequality
! !
n n
1 + x2k 1 + y2k 1
∑ 1 + x4k ∑ 1 + y4k < (1 − x)(1 − y) .
k =1 k =1
1 + x2k 1
Solution. Note first that < . Indeed,
1 + x4k xk
1 + x2k 1 x k + x3k − 1 − x4k ( x3k − 1)(1 − x )

− = = < 0,
1 + x4k xk (1 + x4k ) x k (1 + x4k ) x k
since the conditions of the problem imply 0 < x < 1. Now we estimate
n n
1 + x2k 1 xn − 1
∑ 4k
< ∑ k = n
x ( x − 1)
.
k =1 1 + x k =1 x
A similar inequality can be proven for y, so we obtain

! ! ! !
n n n n
1 + x2k 1 + y2k 1 1
∑ 1 + x4k ∑ 1 + y4k < ∑ xk ∑ yk =
k =1 k =1 k =1 k =1
xn − 1 yn − 1 1
= · = .
x n ( x − 1) y n ( y − 1) ( x − 1)(y − 1)
Remark. This problem, proposed by Estonia, appeared in the IMO-2007 Shortlist.
Second day
TS-4. Sequence ( Gn ) is defined by G0 = 0, G1 = 1 and Gn = Gn−1 + Gn−2 + 1 for

every n > 2. Prove that for every positive integer m there exist two consecutive terms
in the sequence that are both divisible by m.
Solution. Define G−1 = 0, then Gn = Gn−1 + Gn−2 + 1 holds also when n = 1. Consider
the pairs ( Gn , Gn+1 ) of consecutive members of the sequence. There are only m2 pairs
modulo m, hence there are pairs ( Gk , Gk+1 ) and ( Gl , Gl +1 ) with k < l that are compo-
nentwise congruent modulo m. Since Gn−2 = Gn − Gn−1 − 1, two consecutive terms
in the sequence determine the previous term uniquely. The same is true modulo m.
Therefore also pairs ( Gk−1 , Gk ) and ( Gl −1 , Gl ) are componentwise congruent modulo
m. Continuing, we see that ( G−1 , G0 ) and ( Gl −k−1 , Gl −k ) are componentwise congru-
ent modulo m. Since G−1 = G0 = 0, the terms Gl −k−1 and Gl −k are divisible by m as
required.
TS-5. Points A and B are fixed on a circle c1 . Circle c2 , whose centre lies on c1 ,
touches line AB at B. Another line through A intersects c2 at points D and E, where D
lies between A and E. Line BD intersects c1 again at F. Prove that line EB is tangent to
c1 if and only if D is the midpoint of the segment BF.
22
Solution 1. Let K be the second intersection point of the line AD and the circle c1 (Fig. 22).
The triangles KFD and BAD are similar since the corresponding angles are equal. The
triangle BAD is similar to the triangle EAB since, by tangent-secant theorem, ∠ ABD =
∠ BED and they have a common angle at the vertex A. Let O be the centre of the circle
c2 . Since AB is the tangent to the circle c2 at point B, AB ⊥ BO. It follows that AO is a
diameter of the circle c1 since O is on the circle c1 by assumption. Hence also OK ⊥ AK
from which it follows that OK is an altitude of the isosceles triangle ODE. Thus | DK | =
|KE|.
The line EB is tangent to the circle c1 at B if and only if ∠EBK = A
∠ BAD. Since ∠ ABD = ∠ BED, the last equality is equivalent to F
the triangles EKB and BAD being similar. By the same equality
| AB| D
of angles, the two triangles are similar if and only if =
| BE| B
| DB|
. Since EAB and KFD are similar triangles, the last equality
|KE| K
| FD | | DB| O
is equivalent to = . This is equivalent to | FD | =
| DK | |KE|
| DB| since the denominators are equal.
Solution 2. As in the first solution we show that | DK | = |KE|. E
Let | AD | = x, | DK | = |KE| = y, | BE| = z, | DB| = u, | FD | = v,
| AB| = w. By the property of intersecting chords, uv = xy. Figure 22
2
Since AB is a tangent, w = x ( x + 2y). The triangles ABD and
AEB are similar since ∠ ABD = ∠ BED and at vertex A they
u z uw
have a common angle. Thus = and hence z = .
x w x
The condition that the line EB is tangent to the circle c1 is equivalent to
z2 = y( x + 2y). We shall show that the last condition is equivalent to u = v:
u2 w2
z2 = y( x + 2y) ⇔ = y( x + 2y) ⇔ u2 x ( x + 2y) = x2 y( x + 2y) ⇔
x2
⇔ u2 = xy ⇔ u2 = uv ⇔ u = v.
TS-6. A string of parentheses is any word that can be composed by the following rules.
1) () is a string of parentheses.
2) If s is a string of parentheses then (s) is a string of parentheses.
3) If s and t are strings of parentheses then st is a string of parentheses.
The midcode of a string of parentheses is the tuple of natural numbers obtained by find-
ing, for all pairs of opening and its corresponding closing parenthesis, the number of
characters remaining to the left from the medium position between these parentheses,
and writing all these numbers in non-decreasing order. For example, the midcode of
(()) is (2, 2) and the midcode of ()() is (1, 3). Prove that midcodes of arbitrary two
different strings of parentheses are different.
Solution. We can assume that the two strings have equal lengths because otherwise their
midcodes differ by length. We prove the desired claim by induction on the length. In the
23
case of length 2, the claim holds trivially. Let s and t be two longer strings of parentheses.
Consider, for both of them, the longest prefix that forms a string of parentheses itself.
The first and the last character of such prefix form a pair of opening and corresponding
closing parenthesis.
If the prefixes of s and t under consideration have different lengths 2k and 2l, respec-
tively, where assume w.l.o.g. that k < l, then consider the first k numbers in the mid-
codes of both strings. Let the opening parentheses occur at positions a1 , . . . , ak and the
corresponding closing parentheses occur at positions b1 , . . . , bk in word s. The number
a + bi − 1
in midcode that corresponds to the ith pair of parentheses is i . As the first 2k
2
characters of s form a string of parentheses, numbers a1 , . . . , ak , b1 , . . . , bk are precisely
1, . . . , 2k in some order. Thus the sum of k smallest members of the midcode of s is
k
a i + bi − 1 1 + 2 + . . . + 2k k
∑ 2
=
2
− .
2
i =1
In the midcode of t, the sum of k smallest members is larger since, otherwise, the sum
of position indices of some k pairs of parentheses would be 1 + 2 + . . . + 2k. This would
imply that the corresponding closing parenthesis for each opening parenthesis among
those at positions 1, 2, . . . , 2k occurs within the same positions, leading to l 6 k, a con-
tradiction.
If both prefixes under consideration have length 2k then, for both cases, the part of the
word between the first and the last character of the prefix forms a string of parentheses,
as does the part of the word remaining after the prefix (provided they are non-empty).
As s and t differ, either the first mentioned parts of the words or the second mentioned
parts differ.
In the former case, the induction hypotheses implies that their midcodes also differ. In
the midcodes of s and t, these midcodes are represented by numbers that are by 1 larger,
whereby all these numbers are less than 2k. In addition, both midcodes contain k (from
the pair of parentheses embracing the prefix) and the remaining numbers are larger than
2k. Thus the midcodes of s and t differ.
In the latter case, the induction hypothesis again implies that the midcodes of the parts
after the prefix are different. In the midcodes of s and t, these midcodes are represented
by numbers that are by 2k larger. All other numbers in the midcode are less than 2k.
Hence the midcodes differ.
Remark. A tuple of positive integers x1 , . . . , xn is a midcode of some string of parentheses
k n
iff it is monotone, ∑ xi > k 2
for every k = 1, . . . , n, and ∑ xi = n2 .
i =1 i =1
Problems listed by topics

Number theory: OC-2, OC-3, OC-4, OC-9, OC-14, FR-13, TS-4
Algebra: OC-6, OC-11, OC-13, FR-1, FR-2, FR-9, FR-11, TS-3
Geometry: OC-7, OC-12, FR-3, FR-5, FR-8, FR-10, FR-12, FR-14, TS-2, TS-5
Discrete mathematics: OC-1, OC-5, OC-8, OC-10, FR-4, FR-6, FR-7, FR-15, TS-1, TS-6
24
2008/2009
Tartu 2009
WE THANK:
Estonian Ministry of Education and Research
University of Tartu
Problem authors: Juhan Aru, Maksim Ivanov, Oleg Koshik, Toomas Krips,
Kaie Kubjas, Härmel Nestra, Uve Nummert, Oleksandr Rybak (Ukraine),
Translators: Emilia Käsper, Härmel Nestra, Ago-Erik Riet
Editor: Indrek Zolk
Estonian Mathemati al Olympiad

may be given.
curriculum.
contest lasts two days; each day, the contestants have 4.5 hours to solve 3 problems, sim-
ilarly to the IMO. All participants are given the same problems. Some problems in our
*
round, selection has been made to include only problems that have not been taken from
other competitions or problem sources and seem to be interesting enough. The team
selection contest is presented entirely.
1
OC-1. The feet of the altitudes drawn from vertices A and B of an acute triangle ABC
are K and L, respectively. Prove that if | BK | = |KL| then the triangle ABC is isosceles.
(Juniors.) C
L
Solution 1. From | BK | = |KL| we have that BKL is an isosceles
π K
triangle and ∠KBL = ∠KLB (Fig. 1). Now ∠KLC = − ∠KLB,
2
π
∠KCL = ∠BCL = − ∠KBL. Thus ∠KLC = ∠KCL. So |KC| = A B
2
|KL| = | BK |. As the altitude AK is now also a median, the
triangle ABC is isosceles and | AB| = | AC|. Figure 1
π
Solution 2. Similarly to Solution 1 ∠KBL = ∠KLB. As ∠ AKB = ∠ ALB = (Fig. 1), the
2
points A, B, K and L are concyclic. Now ∠KAB = ∠KLB = ∠KBL = ∠KAL whence the
altitude AK is also the bisector of angle CAB. Thus ABC is isosceles.
Remark. The claim can be proven for any triangle.
1 1 1
OC-2. A computer program adds numbers , , , etc., and represents every inter-
1! 2! 3!
mediate sum as a fraction in lowest terms. Prove that for every positive integer k there
is a fraction among the results whose denominator is divisible by at least k different
primes. (Seniors.)
n! n! n!
Solution 1. For every positive integer n, denote an = + + . . . + , then the n-th
1! 2! n!
an
intermediate sum is .
n!
Let k be a fixed positive integer. Let t be such that there are at least 2k − 1 primes
less than or equal to t; then t! is divisible by at least 2k − 1 primes. If there are at least k
at
different prime divisors of the denominator of the fraction we get by writing in lowest
t!
terms, we have what we were looking for. Otherwise, at is divisible by all but at most
k − 1 prime divisors of t! – there are at least k of those. Note that at+1 = (t + 1) · at + 1.
Thus at and at+1 are coprime. So at+1 is not divisible by the prime divisors of t! we
at +1
considered. Thus, writing in lowest terms, the denominator will have at least k
( t + 1) !
prime divisors, so in this case we also have the fraction we were looking for.
Solution 2. Define an as in Solution 1; note that, for every n > 1, an = n · an−1 + 1. Let
n be a number divisible by k different primes. Then the n-th intermediate sum is of
n · an −1 + 1
the form . The numerator is obviously coprime with n whence we cannot
n!
reduce the fraction by dividing numerator and denominator by any prime divisor of n
– there are k of those by construction. The number n! in the denominator is divisible
by all of those primes, thus the denominator is divisible by at least k primes even after
writing the fraction in lowest terms.
2
OC-3. Three circles in a plane have the sides of a triangle as their diameters. Prove
that there is a point that is in the interior of all three circles. (Seniors.)
C
Solution 1. Let the triangle ABC be given. Assume w.l.o.g. that
the largest angle is at C (Fig. 2). As the angles at A and B are
acute, the foot of the altitude drawn from C – let this be F – lies
A
between A and B. Since CFA is a right angle, the circle with
F B
diameter CA goes through point F, thus the altitude CF is a
Figure 2 chord to the circle. Analogously, CF is a chord to the circle with
diameter CB. Hence all the points on the altitude CF except C
and F lie inside both of the circles. Since point F is in the interior of the triangle with
diameter AB, there are interior points of the segment FC that are in the interior of that
circle.
Solution 2. Given the triangle ABC, let the angles at A, B, C have C
respectively sizes α, β, γ (Fig. 3). Let I be the intersection of the

α β α+β α+β+γ I
angle bisectors of ABC. Since + = < = β
2 2 2 2 α
2 α β
2
π α β π π
, we have ∠ AIB = π −
2 2
+ > π− = , i.e. AIB A B
2 2 2 2 2
is an obtuse-angled triangle with the obtuse angle at I. This
Figure 3
shows that I is inside the circle with diameter AB. Analogously
we show that I is inside both of the other circles.
OC-4. There are three ants at vertex A1 of the regular n-gon A1 A2 . . . An initially. Each
minute some two of them move simultaneously to a neighbouring vertex in different
directions (one clockwise and the other counter-clockwise), while the third ant stands
still. For which n can it happen that after some time all the ants meet at a vertex different
from A1 ? (Seniors.)
Answer: if n is divisible by 3.
Solution. Note that the sum of numbers of the vertices on which the ants sit is an invari-
ant modulo n. As initially this sum is 3, it will always be 3 modulo n.
We now show that if 3 does not divide n then it is not possible for the ants to meet at a
different vertex. Suppose that the ants are all at k after a move. Then 3k is congruent to
3 modulo n, or 3k − 3 = 3(k − 1) is divisible by n. This is possible only if k = 1.
Finally show that if n is divisible by 3 then it is possible to meet at another vertex. Let
n
the first and second ants move times, ending up in A n3 +1 , resp. A 2n +1 . Then the third
3 3
n
and second ants move times, taking the third ant to A n +1 and the second ant to A n +1 .
3 3 3
After that all the ants sit at A n3 +1 .
OC-5. A unit square is removed from the corner of the n × n grid where
n > 2. Prove that the remainder can be covered by copies of the figures
consisting of 3 or 5 unit squares depicted in the drawing. Every square
must be covered once and the figures must not go over the bounds of the grid. (Seniors.)
Solution. Assume w.l.o.g. that the unit square removed is the one in the bottom right
3
corner. Further let us write “n × n grid” for the grid with the bottom right corner square
removed.
The cases n = 2, 3, 4 can be done by trial (Figures 4, 5 and 6).
We show how to extend the construction for n to a construction for n + 3; by repeating
this procedure, we get a solution for any positive integer n.
Let n be odd; consider an (n + 3) × (n + 3) grid. We cover the n × n grid in its bottom
right corner, then cover the (n − 1) × 3 band on the left and the 3 × (n − 1) band above
by 2 × 3 rectangles formed by two 3-square figures (Fig. 7). We are left with a 4 × 4 grid
in the top left corner which we can already cover.
Let n be even; consider an (n + 3) × (n + 3) grid. Cover the n × n grid in its lower right
corner, then cover the (n − 2) × 3 band on the left and the 3 × (n − 2) band above by
2 × 3 rectangles. The remaining part in the top left corner can be covered like shown in
Fig. 8.
Figure 4 Figure 5 Figure 6 Figure 7 Figure 8

Remark. It is also possible to extend the construction for n to a construction for n + 2,
dividing into cases according to the remainder of n modulo 3.
OC-6. Find all posivite integers n for which there are exactly 2n pairs of integers (a, b)
where 1 6 a 1. The pairs whose second component is at most n − 1 have been
counted for n − 1. We have to add pairs with the second component n and the first
component a proper divisor of n. Denoting the number of proper divisors of n by d(n),
we have
g ( n ) = g ( n − 1 ) + d ( n ). (1)
We tabulate according to (1).

n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
d (n) 0 1 1 2 1 3 1 3 2 3 1 5 1 3 3 4 1
g(n) 0 1 2 4 5 8 9 12 14 17 18 23 24 27 30 34 35
We see that only n = 15 satisfies the conditions among the numbers considered.
Note that two consecutive numbers greater than 4 have at least 4 proper divisors in
total. Indeed, one of them is an even number greater than 4 and has at least 3 proper
divisors (1,2 and half of itself), the other one has at least one proper divisor (number 1).
So for every n > 4 we have g(n + 2) > g(n) + 4.
It follows that if for some n we have g(n) > 2n then g(n + 2) > g(n) + 4 > 2n + 4 =
2(n + 2). Since the inequality holds for n = 16 and n = 17, we obtain by induction that
g(n) > 2n for any n > 16. Thus there are no more numbers satisfying the condition in
the problem.
4
D C c D
γ C
A α β B A B
c
U V U
Figure 9 Figure 10
OC-7.
a) An altitude of a triangle is also a tangent to its circumcircle. Prove that some angle
of the triangle is larger than 90◦ but smaller than 135◦ .
b) Some two altitudes of the triangle are both tangents to its circumcircle. Find the
angles of the triangle. (Seniors.)
Answer: b) 120◦ , 30◦ , 30◦ .

Solution 1. Let our triangle be ABC and the sizes of angles at A, B, C be α, β, γ, respec-
tively (Fig. 9). Assume the altitude of ABC drawn from A is a tangent to the circumcir-
cle; the point of tangency is A since it lies on the circle. Let D be the foot of the altitude.
W.l.o.g. assume C is between B and D.
By the tangent-secant theorem, ∠ DAC = ∠ DBA = β. The angle ACB is an exterior
angle of the triangle ACD, hence γ = ∠ ACB = ∠ ADC + ∠ DAC = 90◦ + β. Thus
γ > 90◦ and also β = γ − 90◦ .
a) In the triangle ABC we have
180◦ = α + β + γ = α + 2γ − 90◦ ,
whence 2γ + α = 270◦ . Thus 2γ < 270◦ which gives γ < 135◦ .

b) Obviously the altitude drawn from the vertex of the obtuse angle goes inside the
circumcircle and thus cannot be a tangent. So we have to let the altitude drawn from B
be a tangent to the circumcircle (Fig. 10). Then in addition to the identity β = γ − 90◦ we
also have α = γ − 90◦ . Thus 180◦ = 3γ − 180◦ whence γ = 120◦ and also α = β = 30◦ .
Solution 2. We use notations from the previous solution.
a) Since DA is a tangent and DC is a secant to the same circle drawn from the same
point we have | DA| > | DC|. Thus ∠ ACD > ∠CAD. Since ∠ ACD + ∠CAD = 180◦ −
1
∠ ADC = 180◦ − 90◦ = 90◦ , it follows that 90◦ > ∠ ACD > · 90◦ = 45◦ . As γ =
2
180◦ − ∠ ACD, we obtain 90◦ < γ < 135◦ .
b) Let O be the centre of the circumcircle. As AO ⊥ AD and BC ⊥ AD we have AO k
BC. Analogously BO k AC. Thus ACBO is a parallelogram, but since |OA| = |OB| it is
a rhombus. Since the diagonal OC has the same length as the sides, ACO and BCO are
equilateral triangles. Thus γ = ∠ ACO + ∠ BCO = 60◦ + 60◦ = 120◦ . The diagonal of a
1
rhombus bisects its angle, thus α = β = · 60◦ = 30◦ .
2
OC-8. Find all functions f from positive real numbers to positive real numbers such
that the curve y = c · f ( x ) is symmetric with respect to the line y = x for every positive
real number c. (Seniors.)
5
a
Answer: f ( x ) = where a is an arbitrary positive real number.
x
Solution. The symmetry of the curve y = g( x ) with respect to the line y = x is equivalent
to the condition that g( g( x )) = x for each positive real number x. Indeed, the point
( x, g( x )) lies on the curve y = g( x ). By symmetry, the point ( g( x ), x ) is also on the
curve – but this is equivalent to g( g( x )) = x.
Using this for the curve y = c · f ( x ), we have for every c > 0 and x > 0 that
c · f (c · f ( x )) = x. (2)
1 1 f (1 )
Taking c = in the identity (2) we have · f (1) = x, i.e. f ( x ) = for every
f (x) f (x) x
positive real x.
a
It remains to check that every function of form f ( x ) = where a > 0 satisfies the
x
ca
conditions of the problem. Consider the function g where g( x ) = c · f ( x ) = and
x
c > 0. Now
ca ca
g( g( x )) = g = ca = x,
x x
which means that the curve y = c · f ( x ) is symmetric with respect to the line y = x.
Remark. There are other possibilities to determine all possible functions. E.g. taking
x = c = 1 in the identity (2), we have f ( f (1)) = 1, yielding by the substitution x = f (1)
in (2) that for every positive real number c it holds that c · f (c) = f (1).

Olympiad
FR-1. In triangle ABC, points F and E are chosen on sides AC and BC, respectively,
such that 2|CF| = | FA| and 2|CE| = | EB|. Outside triangle ABC, points K and L are
chosen on rays AE and BF, respectively, such that 2|KE| = | EA| and 2| LF| = | FB|.
Prove that ABKL is a parallelogram. (Grade 9.)
Solution 1. Note that △ ABF ∼ △CLF and △ ABE ∼ △KCE with similarity ratio 2
(Fig. 11). Indeed, ∠ BFA = ∠ LFC and ∠ BEA = ∠CEK (opposite angles) and we know
that | AF| = 2|CF|, | BE| = 2|CE|, | BF| = 2| LF| and | AE| = 2|KE|. Thus, LC and CK
are parallel to AB and 2| LC| = 2|CK | = | AB|. Consequently, points L, C and K are
collinear, | LK | = | AB| and LK k AB, implying that ABKL is a parallelogram.
C K
Solution 2. The assumptions of the problem directly yield: L
• △ ABC ∼ △ FEC with similarity ratio 3; F E

3
• △ AKC ∼ △ AEF with similarity ratio ;
2
3 A B
• △ BCL ∼ △ BEF with similarity ratio .
2 Figure 11
6
3 3
It follows that 3| EF| = | AB|, | EF| = |KC| and | EF| = |CL|, where EF (and thus
2 2
also KC and CL) are parallel to AB. Consequently, K, C and L are collinear, KL k AB,
and |KL| = |KC| + |CK | = 3| EF| = | AB|, implying that ABKL is a parallelogram.
Solution 3. Let lines BC and AL intersect at N, and let lines AC and BK intersect at M.
BF AF
Since = = 2, we have that BL and AC are medians of ABN and LC is the mid-
FL FC
segment of side AB: LC k AB and | AB| = 2| LC|. Similarly, BC and AK are medians of
ABM and CK is the mid-segment of side AB: CK k AB and | AB| = 2|CK |. Hence, L, C
and K are collinear, LK k AB and | LK | = | AB|, implying that ABKL is a parallelogram.
FR-2. Call a positive integer m magic if the sum of its digits equals the product of its
digits.
a) Prove that for all n = 1, 2, . . . , 10, there exists a magic number consisting of precisely
n digits.
b) Prove that there exist infinitely many magic numbers. (Grade 9.)
Solution 1. a) The numbers 1, 22, 123, 1124, 11125, 111126, 1111127, 11111128, 111111129
and 1111111144 are magic.
b) Given any positive integer for which the product of its digits is larger than the sum,
we can construct a magic number by appending a suitable number of 1-digits: each
appended 1 increases the sum by 1 without changing the product. Now, for any n > 0,
the product of the digits of 22 . . . 2} equals 2n > 2n, the sum of the digits, so for any n,
| {z
n
we can construct a magic number with at least n digits.
Solution 2. b) Let m > 1 be a magic number. We can always construct a larger magic
number m′ by appending 2 to m, followed by a suitable number of 1-digits, as append-
ing a 2 increases the sum of the digits by 2 and the product of the digits by at least 2,
while appending a 1 increases the sum by 1 without changing the product.
FR-3. Juku and Miku play a game on a rhombus of side length n consisting
of two equilateral triangles divided into equilateral triangular tiles with side
length 1 (n = 3 in the figure). Each player has one token. At the beginning
of the game, the tokens lie on the topmost and bottommost tile, respectively.
Players alternate moves by sliding their token one step to an adjacent tile
(tiles are adjacent if they share a side). A player wins the game by capturing
his opponent’s token (moving his own token to the same tile where the opponent’s
token lies); or by reaching his opponent’s starting tile. Suppose Juku makes the first
move. Does either of the players (who?) have a winning strategy? (Grade 9.)
Answer: Juku has a winning strategy.
Solution. We show that Miku cannot capture Juku’s token. Colour the tiles with the
triangle “pointing upwards” black, and the remaining tiles white, then any two adjacent
tiles are of different colour and every move takes the token from a black tile to a white
tile or vice versa. As the starting tiles are of different colour, after every two moves, the
tokens again lie on tiles of different colour and no Miku’s move can result in capturing
Juku’s token.
7
Consequently, Juku can safely take the shortest route to his opponent’s starting tile.
Since the shortest route has equal length for Juku and Miku, Juku is bound to reach his
goal first.
FR-4. Find all pairs of positive integers (m, n) such that in an m × n rectangular grid,
the number of unit squares touching at least one side of the rectangle equals the number
of remaining unit squares. (Grade 9.)
Answer: (5, 12), (6, 8), (8, 6), (12, 5).
Solution. We can exclude m = 1 or n = 1, as then all unit squares touch a side.
Assume now m > 2, n > 2. There are 2m + 2n − 4 unit squares touching a side, so the
number of remaining squares is mn − 2m − 2n + 4. We require 2m + 2n − 4 = mn −
2m − 2n + 4, which is equivalent to (m − 4)(n − 4) = 8. As m − 4 > −2 and n − 4 > −2,
we can eliminate {−1, −8} and {−2, −4} as possible factorings of 8. The remaining
possibilities {1, 8} and {2, 4} yield four solutions for (m, n).
FR-5. Let the angles of a triangle measure x, y, z in degrees.

x y z
a) Prove that if , , are all rational, then x, y, z are also all rational.
y z x
x y z
b) Prove that if exactly one of , , is rational, then x, y, z are all irrational.
y z x
(Grade 10.)
Solution. Note that
180 x+y+z x y z y z
= = + + = 1+ + . (3)
x x x x x x x
y 1 z 180
a) Assume = x and are rational. By (3), is a sum of three rationals and thus
x y x x
itself rational. Hence, x is rational. The proof for y and z is analogous.
x y y z z x
b) Assume w.l.o.g. (and thus also ) is rational and , (and thus also , ) are
y x z x y z
180
irrational. The equation (3) then represents as a sum of one irrational and two
x
180 x
rational numbers. Hence, is irrational, and so is x. As is rational, y must also be
x y
irrational.
x+y
Assume now z is rational, then x + y = 180 − z is also rational. Now as a ratio
y
x+y x
of a rational and an irrational number is irrational, yet = + 1 as a sum of two
y y
rationals must be rational, contradiction.
FR-6. Find all triples of positive integers ( x, y, z) satisfying 99x + 100y + 101z = 2009.
(Grade 10.)
Answer: (1, 9, 10), (2, 7, 11), (3, 5, 12), (4, 3, 13), (5, 1, 14).
8
Solution. We bound the lhs from both sides:
2009 = 99x + 100y + 101z 6 101( x + y + z),

2009 = 99x + 100y + 101z > 99( x + y + z).
2009 2009
Rearranging, we get 19 < 6 x+y+z 6 < 21, implying x + y + z = 20.
101 99
Now, rewriting the original equation as 100( x + y + z) + z − x = 2009 gives z = x + 9.
Substituting z in x + y + z = 20 then gives y = 11 − 2x. Since x and y are positive
integers, we must have 0 < x 6 5. The values 1, 2, 3, 4, 5 of x lead to solutions (1, 9, 10),
(2, 7, 11), (3, 5, 12), (4, 3, 13), (5, 1, 14) of the equation, respectively.
FR-7. In an acute triangle ABC, draw a perpendicular y to AB through B, and a

perpendicular z to AC through C. Prove that the intersection point of y and z lies on the
perpendicular drawn to BC through A iff | AB| = | AC|. (Grade 10.)
Solution 1. Let the perpendicular drawn to BC through A be x, and let x intersect BC at
E (Fig. 12).
q Assume x, y and z intersect at D, then from right triangles ABD and ACD,
| AB| = | AD | · | AE| = | AC|.
C Conversely, assume | AB| = | AC|, then triangle ABC is

isosceles with vertex angle A. Altitude AE bisects the ver-
z tex angle, so segments AB and AC are symmetric wrt x. But
x E D then also perpendiculars y and z drawn through B and C are
A symmetric wrt x, and their intersection points with x coin-
cide.
y
Solution 2. Let the notations be as above. Since AB is per-
B pendicular to BD and AC to CD, points A, B, C, D lie on the
same circle with diameter AD. As chord BC is perpendicular
Figure 12 to AD, the intersection point E bisects BC. Hence, AE is both
an altitude and a median in triangle ABC, so | AB| = | AC|.
Conversely, assume | AB| = | AC|, then ∠ ACB = ∠ ABC and in triangle ABC, the alti-
tude drawn from vertex A coincides with the angle bisector. Let y and z intersect at F.
π π
Since ∠ BAF = ∠ FCB = − ∠ ACB = − ∠ ABC = ∠CBF = ∠CAF, AF is also an
2 2
angle bisector. Thus, line AF coincides with x, so x, y and z intersect at F.
FR-8. Mari and Jüri play a game on an 2 × n rectangular grid (n > 1) whose sides of
length 2 are glued together to form a cylinder. Alternating moves, each player cuts out
a unit square of the grid. A player loses if his/her move causes the grid to lose circular
connection (two unit squares that only touch at a corner are considered to be discon-
nected). Suppose Mari makes the first move. Which player has a winning strategy?
(Grade 10.)
Answer: Mari, if n is odd; Jüri, if n is even.
Solution 1. Consider the grid immediately before one of the players is forced to make
a losing move. Divide missing squares into horizontal blocks of consecutive squares
in one row of the grid. Clearly, no upper and lower block can overlap, as this would
disconnect the grid. If the whole upper row is missing, then the lower row has no
9
missing squares, and vice versa. In this case, there have been n moves, so Mari wins if
n is odd and Jüri wins if n is even.
Assume now no horizontal missing block covers the whole cylinder. Notice that if a
block is missing in the upper row, then its neighbouring column cannot have a missing
square in the lower row, and vice versa. Thus, there is always a full column between
two missing blocks. On the other hand, if there was more than one column between
some two missing blocks, it would be possible to make another move without losing
by extending one of the missing blocks. Finally, if two consecutive missing blocks were
both in the same row, it would be possible to make a move by removing the square
between them.
We conclude that before the last move, the missing blocks are in the upper and lower
row, alternatingly, implying that the number of missing blocks is even. As there is
exactly one full column between two neighbouring missing blocks, the number of full
columns f is also even. The remaining columns have exactly one missing square, so the
number of missing squares is m = n − f . If n is odd, m is odd and Mari wins; if n is
even, m is even and Jüri wins.
Solution 2. Consider the planes of symmetry of the cylinder that intersect the midpoints
of its bases. Choose a plane p such that at least one of its intersections with the cylinder
surface coincides with a grid line.
If n is even, both intersections coincide with the grid and reflecting over p divides the
unit squares into symmetric pairs. Jüri’s strategy is to cut out a square symmetric to
the square Mari chose. We claim that if Mari’s move does not disconnect the grid, then
neither does Jüri’s follow-up move. Indeed, if Mari removes a square not adjacent to
p, then Mari’s move does not affect the grid around Jüri’s square. Thus, the moves
are mirror images of each other and have identical impact. If Mari removes a square
adjacent to p, then before Jüri’s move, the grid around his square has one additional
missing adjacent square compared to the grid around Mari’s square before her move.
Thus, Jüri’s move can only disconnect the grid along the line between his square and
Mari’s. However, since the two removed squares are in the same row, and the two other
squares in the same column must be present (for otherwise Mari’s move would have
disconnected the grid), the disconnect does not happen. It follows that the grid can
only disconnect after Mari’s move, so Jüri has a winning strategy.
If n is odd, reflecting over p divides all squares into symmetric pairs save for two
squares that p intersects in the middle. Let Mari first choose one of those two squares.
Then Jüri cannot choose the other square without losing, as this would disconnect the
square. Thus, Mari can now use the strategy of choosing symmetric squares to win the
game.
FR-9. The teacher asks Arno to choose some of the positive factors of 200910 such that
no chosen factor divides another chosen factor. At most how many factors can Arno
choose? (Grade 10.)
Answer: 11.
Solution. Since 2009 = 72 · 41, where 7 and 41 are primes, we can represent all factors
of 200910 in the form 7n · 41m , where 0 6 n 6 20 and 0 6 m 6 10. Since there are 11
possible choices for m, Arno can choose at most 11 factors. Otherwise, by the Pigeonhole
principle, two of the chosen factors would have the same exponent m, and the one with
10
a smaller exponent n would divide the other.
Now, we show that none of the 11 factors of the form 720−m · 41m , m = 0, 1, . . . , 10,
is divisible by another. Assume by contradiction that for some two m1 and m2 , 720−m1 ·
41m1 divides 720−m2 · 41m2 . But then 20 − m1 6 20 − m2 and m1 6 m2 , implying m1 = m2 ,
contradiction.
FR-10. Let n > 18 be a positive integer such that n − 1 and n + 1 are both primes.
Prove that n has at least 8 different positive factors. (Grade 11.)
Solution 1. First, since of the three consecutive integers n − 1, n, n + 1, two are primes,
the third, n, must be divisible by both 2 and 3. Thus, 1, 2, 3 and 6 are factors of n.
Next, n > 18 = 3 · 6. If n = 5 · 6 = 30, we get four additional factors 5, 10, 15 and 30.
We can excldue n = 4 · 6 = 24 and n = 6 · 6 = 36, as 24 + 1 and 36 − 1 are not prime.
n n n n n
Finally, if n > 6 · 6, then 6 < , so n has four additional factors , , , larger than 6.
6 1 2 3 6
Solution 2. As above, note that n is divisible by 2 and 3. We can express the number
of factors via the exponents in the canonical representation of n. If n has a third prime
factor, then its canonical representation contains at least 3 primes with exponents at
least 1, so the number of factors is at least (1 + 1) · (1 + 1) · (1 + 1) = 8. Assume now
n = 2a 3b . If a = 1, then b > 2, and vice versa, so the number of different factors in this
case is at least (1 + 1) · (3 + 1) = 8. Finally, if a, b > 1, the number of different factors is
at least (2 + 1) · (2 + 1) = 9.
FR-11. Find all real numbers k that satisfy 0 6 a + b − kab 6 1 for all real numbers a
and b such that 0 6 a 6 1 and 0 6 b 6 1. (Grade 11.)
Answer: 1 6 k 6 2.
Solution. First, we show that if 1 6 k 6 2, then k satisfies the condition. Let 0 6 a, b 6 1,
then a2 6 a and b2 6 b. We get
0 6 (a − b)2 = a2 + b2 − 2ab 6 a + b − kab 6 a + b − ab = 1 − (1 − a)(1 − b) 6 1.
Next, we show that the condition does not hold for k < 1 and k > 2. Take a = b = 1,
then a + b − kab = 2 − k. Now k < 1 yields a − b − kab = 2 − k > 1 and k > 2 yields
a − b − kab = 2 − k < 0.
FR-12. Numbers 1 to n2 are written in some order in the unit squares of an n × n

square such that in any rectangle consisting of some of those unit squares, the sum of
the numbers in two opposite corner squares equals the sum of the numbers in the other
two corner squares. Find all possible values for the sum of all numbers on a diagonal of
the n × n square. (Grade 11.)
n ( n2 + 1 )
Answer: .
2
Solution. Denote the unit square in row i, column j by (i, j) and the number written
in that square by tij . Let D = t11 + t22 + . . . + tnn be the sum of the numbers on the
main diagonal. For all i, j = 1, 2, . . . , n, we have tii + t jj = tij + t ji . By summing n2 such
n2 ( n2 + 1 )
equations over all values of i and j, we get 2nD = 2 · (1 + 2 + . . . + n2 ) = 2 · ,
2
11
as each of the n numbers on the diagonal occurs an equal 2n times on the lhs, and each
n ( n2 + 1 )
of the n2 numbers occurs twice on the rhs. Solving for D gives D = .
2
FR-13. Call a point in a plane rational (resp. irrational) if both its coordinates are ratio-
nal (resp. irrational).
a) Does every point on the plane lie on some line defined by two rational points?
b) Does every point on the plane lie on some line defined by two irrational points?
(Grade 11.)
Solution. a) Two different rational points A( x1 , y1 ) and B( x2 , y2 ) define a line given by
( x − x1 )(y2 − y1 ) = (y − y1 )( x2 − x1 ),
or ax + by + c = 0, where a = y2 − y1 , b = x1 − x2 and c = y1 ( x2 − x1 ) − x1 (y2 − y1 ) are

all rational. √ √
We prove that point ( x, y) = 2, 3 does not satisfy ax + by + c = 0 for any triple of
√ √ √ √
rational numbers (a, b, c). Indeed,
√ assume a 2 + b 3 +√ c = 0, then a 2 = − b 3 − c,
2 2 2 2 2 2
which gives 2a = 3b + 2bc 3 + c , implying that 2bc 3 = 2a − 3b − c is rational.
The latter is only possible if b = 0 or c = 0.
√
If b = 0, then a 2 = −c is rational. This is only possible when a = c = 0, i.e., points A
and B coincide.√ √
If c = 0, then a 2 = −b 3. Now a = 0 implies rb = 0 √ and vice versa; in this case again,
a 3 6 a
the points coincide. Thus, a, b 6= 0 and − = = . However, is rational while
√ b 2 2 b
6 is irrational, contradiction. √ √
Thus, no line defined by two rational points contains point 2, 3 .
b) Take an arbitrary point ( x, y). Consider points A ( x + α1 , y + β 1 ) and
B ( x + α2 , y + β 2 ), where
( (
(1, 2), x ∈ I, (1, 2), y ∈ I,
(α1 , α2 ) = √ √ ( β1 , β2 ) = √ √
2, 2 2 , x ∈ Q, 2, 2 2 , y ∈ Q.
Points A and B are irrational, since each of their coordinates is a sum of a rational and
an irrational number. It is also easy to verify that point ( x, y) lies on line AB:
x − ( x + α1 ) − α1 − β1 y − (y + β1 )
= = −1 = = .
( x + α2 ) − ( x + α1 ) α2 − α1 β2 − β1 (y + β 2 ) − ( y + β 1 )
FR-14. Four equally sized spheres are placed in a larger sphere such that each of the
small spheres touches the remaining three small spheres and the large sphere. Is the
total volume of the four small spheres equal to, larger or smaller than the remaining
volume of the large sphere? (Grade 12.)
12
Answer: smaller.
Solution 1. Assume the total volume of the small spheres is at least one half of the volume
of the large sphere. Let the radii of the small and big sphere be r and R, respectively.
Since the ratio of the volumes equals the ratio of the radii in cubic power, we have
R R
2 · 4 · r3 > R3 , or r > . The strict inequality r > is clearly impossible, as the centre
2 2
of the large sphere would have to lie in each of the smaller spheres simultaneously. If
R
r = , any two small spheres should touch at the centre of the large sphere (i.e., the 6
2
points at which the small spheres touch should all coincide), which is also impossible.
We conclude that the total volume of the small spheres is less than one half of the volume
of the large sphere.
Solution 2. As above, we show that the total volume of the small spheres is less than
half of the volume of the large sphere. Let the radius of the small spheres be 1, then
the radius of the large sphere can be written as 1 + a, where a is the distance from the
centre to the vertex in a regular tetrahedron of side length 2. Consider a triangle with
vertices in the centre of this tetrahedron, and at two of its vertices (i.e., at the centres
of two small spheres). The triangle has side lengths 2, a and a. The triangle inequality
gives 2a = a + a > 2, so a > 1.
4
The volume of half the large sphere is V = · π · (1 + a)3 , while the total volume of the
3
4
small spheres is v = 4 · · π · 1. As a > 1, (1 + a)3 > 8, and V > 2v.
3 √
6
Remark. The exact value of a in Solution 2 is .
2
FR-15. Let n be a non-negative integer such that 3n + 3n+1 + . . . + 32n is a perfect

square. Prove that n is divisible by 4. (Grade 12.)
Solution. Adding as a sum of geometric progression, we get
n n +1 2n n n n 3n +1 − 1
3 +3 +...+3 = 3 · (1 + 3 + . . . + 3 ) = 3 · .
2
3n +1 − 1
The factors 3n and are relatively prime: the former can have only 3 as its prime
2
factor while the latter is not divisible by 3. As the product is a perfect square, both
factors are perfect squares. The factor 3n gives now that n is even.
Suppose n is not divisible by 4. As n is even, n ≡ 2 (mod 4) whence n + 1 ≡ 3 (mod 4).
Note that 34 = 81; from 16 | 80 we get 34 ≡ 1 (mod 16) whence 3n+1 ≡ 33 ≡ 11
3n +1 − 1
(mod 16) and 3n+1 − 1 ≡ 10 (mod 16), so ≡ 5 (mod 8). But a perfect square
2
is not congruent to 5 modulo 8.
Remark. From the assumption n ≡ 2 (mod 4), an analogous contradiction can also be
derived using module 5.
The solutions can also be written down in easier terms, omitting the application of the
geometric progression sum formula and just studying the behaviour of the sums of
powers of 3 modulo 8 (or modulo 5).
13
FR-16. Find all real numbers a such that polynomial x3 + ax − 2(a + 4) has exactly
two distinct real roots. (Grade 12.)
Answer: −12, −3.

Solution 1. Note that x3 + ax − 2(a + 4) = ( x − 2) · x2 + 2x + (a + 4) . Hence 2 is a
root of our polynomial irrespective of a. Consider two cases.
If x = 2 is a single root then the quadratic polynomial x2 + 2x + (a + 4) must have
exactly one real root, i.e., its discriminant equals zero. Thus 4 − 4(a + 4) = 0, giving
a = −3.
If x = 2 is a double root then x = 2 must be a root of x2 + 2x + (a + 4), hence 22 + 2 ·
2 + (a + 4) = 0, giving a = −12. As x2 + 2x − 8 6= ( x − 2)2 , the other root is different
from 2 indeed.
Hence the only possibilities are a = −3 and a = −12.
Solution 2. As a polynomial with real coefficients cannot have one imaginary root and
other roots real, the given cubic polynomial has exactly two distinct roots, i.e., one of
its roots must be a double root. This root is also a root of its derivative 3x2 + a. Letting
a = −3x2 in the original polynomial, we get the equation −2x3 + 6x2 − 8 = 0 for roots
common to the polynomial and the derivative. As −2x3 + 6x2 − 8 = −2( x − 2)2 ( x + 1),
the possible common roots are x = 2 and x = −1. From 3x2 + a = 0, we now establish
the corresponding possibilities a = −12 and a = −3.
FR-17. Prove that the ratio of the lengths of the two diagonals of a parallelogram
equals the ratio of its side lengths iff the angles at the intersection of the diagonals are
equal to the interior angles of the parallelogram. (Grade 12.)
| BC|
Solution 1. Since the diagonals of a parallelogram bisect each other, condition =
|CD |
|CA| | BC| |CP|
is equivalent to = (Fig. 13). The Sine Law in triangles BCD and CPD
| BD | |CD | | PD |
| BC| sin ∠ BDC |CP| sin ∠ PDC sin ∠ BDC | BC| |CP|
give = and = = . Thus, =
|CD | sin ∠CBD | PD | sin ∠ PCD sin ∠ PCD |CD | | PD |
iff sin ∠CBD = sin ∠ PCD. From triangle BCD, we observe ∠CBD + ∠ PCD < 180◦ ,
hence sin ∠CBD = sin ∠ PCD iff ∠CBD = ∠ PCD. Since triangles CPD and BCD share
a common angle ∠ PDC = ∠ BDC, the latter condition translates to ∠CPD = ∠ BCD.
| BC| |CA|
To conclude, we have that condition = is equivalent to ∠CPD = ∠ BCD.
|CD | | BD |
| DC| |CA|
Analogously, = is equivalent to ∠CPB = ∠ BCD.
|CB| | BD |
D C
Solution 2. Let the diagonals AC and BD of parallelo-
P gram ABCD intersect at P and let a = | AB| = |CD |,
b = | BC| = | AD | and d = | BD |, e = | AC|. We show
a d
A B that = iff ∠ BCD = ∠CPD (analogously, we can
b e
a e
Figure 13 then show that = iff ∠ BCD = ∠CPB).
b d
14
b2 e2
In parallelogram ABCD, we have 2(a2 + b2 ) = d2 + e2 , or 2a2 1 + 2 = d2 1 + 2 (by
a d
a d
the Cosine Law in triangles ABC and BCD). Thus, condition = is equivalent to
b e
a d
condition 2a2 = d2 . The two equations = and 2a2 = d2 are in turn equivalent to
b e
2b 2a d
= = , i.e., triangle BCD with side lengths b, a and d is similar to triangle CPD
e d a
e d
with side lengths , and a, or ∠ BCD = ∠CPD.
2 2
Remark. To construct a parallelogram where the ratio of diagonal lengths equals the
pone side a and an adjacent angle α. Then the other side of the
ratio of side lengths, fix
parallelogram is b = a 1 + cos2 α − a cos α.
FR-18. Some unit squares are removed from a rectangular grid in such a way that
whenever a unit square is removed, all unit squares in the rectangular area obtained by
extending the left side of the unit square to the top of the grid and the bottom side of the
square to the right side of the grid are also removed. Finally, a count is written in each of
the remaining unit squares indicating the total number of remaining squares above that
square in the same column and to the right of that square in the same row. Prove that
there are at least as many even counts as odd counts written in the remaining squares.
(Grade 12.)
Solution. Consider a configuration of the rectangular grid after removing some squares.
We prove that if we can legally remove exactly two more adjacent squares, then after
updating the counts, the number of even and odd counts both decrease by 1.
Indeed, if we can remove two squares in the same row, the columns of these squares
must be empty above that row, and full below that row. In this row, each count decreases
by 2, so the parities do not change. In the two columns below those two squares, each
row contains one odd and an one even number (the numbers of squares remaining
above are equal and the numbers of squares remaining to the right differ by one). After
removing the two squares, both counts change by one, so the total number of even and
odd counts remains the same. Finally, the counts in the two removed squares are 0 and
1, so removing those squares decreases both the number of even and odd counts by 1.
Analogously, removing two squares in the same column decreases the number of even
and odd counts by 1.
Now, starting from the given configuration, keep removing pairs of adjacent unit
squares until it is no longer possible. In the end, we must have one of two configu-
rations:
• An empty grid. This grid contains no even or odd counts.

• A grid with k > 1 rows where the topmost row has 1 square, the next row 2 squares,
etc., until the bottommost row has k squares. Then, the square in the top row has
count 0, the squares in the second row have counts 2 and 0, the third row contains
4, 2 and 0, etc.
In both configurations, there are at least as many even counts as odd counts. Compared
to the original configuration, we decreased the even and odd counts by an equal num-
15
ber, so also in the original configuration, there had to be at least as many even counts as
odd counts.
IMO Team Selection Contest

First day
TS-1. For arbitrary pairwise distinct positive real numbers a, b, c, prove the inequality
( a2 − b 2 )3 + ( b 2 − c 2 )3 + ( c 2 − a2 )3
> 8abc.
( a − b )3 + ( b − c )3 + ( c − a )3
Solution. Denote a − b = x and b − c = y; then c − a = −( x + y). For the denominator

of the lhs,
(a − b)3 + (b − c)3 + (c − a)3 = x3 + y3 − ( x + y)3 = −3x2 y − 3xy2 =

= −3xy( x + y) = 3(a − b)(b − c)(c − a).
For the numerator, analogously,
(a2 − b2 )3 + (b2 − c2 )3 + (c2 − a2 )3 = 3(a2 − b2 )(b2 − c2 )(c2 − a2 ).
( a2 − b 2 )3 + ( b 2 − c 2 )3 + ( c 2 − a2 )3
Thus = (a + b)(b + c)(c + a). Hence the given in-
( a − b )3 + ( b − c )3 + ( c − a )3
equality is equivalent to √ (a + b)(b + c)(c + a) > 8abc. This inequality holds due to
AM-GM (use a + b > 2 ab and note that a 6= b; similarly for other pairs).
TS-2. Call a finite set of positive integers independent if its elements are pairwise co-
prime, and nice if the arithmetic mean of the elements of every non-empty subset of it is
an integer.
a) Prove that for any positive integer n there is an n-element set of positive integers
which is both independent and nice.
b) Is there an infinite set of positive integers whose every independent subset is nice
and which has an n-element independent subset for every positive integer n?
Answer. b) No.
Solution 1. a) Let A = {n! + 1, 2 · n! + 1, . . . , n · n! + 1}. For any two elements k · n! + 1
and l · n! + 1 where 1 6 k < l 6 n,
gcd(k · n! + 1, l · n! + 1) = gcd(k · n! + 1, (l · n! + 1) − (k · n! + 1))

= gcd(k · n! + 1, (l − k)n!).
The number k · n! + 1 is not divisible by any of 2, . . . , n, thus by none of their prime

divisors; the only prime divisors of (l − k)n! are those of 2, . . . , n. So gcd(k · n! + 1, l ·
n! + 1) = 1, thus A is independent.
16
From arbitrary m elements k1 · n! + 1, k2 · n! + 1, . . . , km · n! + 1 of A where 1 6 m 6 n,
we get
(k1 · n! + 1) + (k2 · n! + 1) + . . . + (km · n! + 1) = (k1 + k2 + . . . + km )n! + m

n!
= m · (k1 + k2 + . . . + k m ) · + 1 .
m
n!
Since is an integer, the sum of the m selected elements is divisible by m, so their
m
arithmetic mean is an integer. Hence A is nice.
b) Note that the difference of any two elements of an n-element nice set is divisible by
every positive integer less than n. Indeed, let X be an n-element nice set. Fix an integer i,
2 6 i 6 n − 1, and arbitrary a, b ∈ X. If s is the sum of some i − 1 elements of X different
from a and b then both a + s and b + s are divisible by i, thus so is their difference a − b.
Assume there is an infinite set B of positive integers such that for every n there is n-
element independent subset and that every independent subset of B is nice. We show
that every finite independent subset A can be extended to larger independent subsets
by adding new numbers. Indeed, there is only a finite number, say k, of primes which
divide some element of A. By Pigeonhole principle, a k + 1-element independent subset
of B has an element that is coprime with all elements of A, and we can add this element
to A to form a larger independent set.
Let a and b be different coprime elements of B (they exist since there is an 2-element
independent subset of B) and consider an infinite subset of B containing both a and
b and having pairwise coprime elements (it can be constructed by infinitely repeating
the steps described above starting from A = { a, b}). Since any of its finite subsets
containing a and b is nice, a − b is divisible by every positive integer. So a = b and we
obtained a contradiction showing there is no such set B.
Solution 2. a) Use Dirichlet’s Theorem: for any two coprime numbers a and d there are
infinitely many primes of the form a + kd where k > 0. Choosing a = 1 and d = n! we
have infinitely many primes of the form kn! + 1 where k > 0. Choosing some n of them
we have an independent set A. Similarly to Solution 1 we establish the niceness of A.
b) Let B be an infinite set of positive integers such that for every n there is n-element
independent subset and that every independent subset of B is nice. Let a ∈ B. Let
C be any independent subset of B of size greater than a. As established in Solution
1, the difference of any two elements of B is divisible by a. So all elements of B are
congruent modulo a and thus have the same greatest common divisor with a. Since B is
independent, it has to be 1.
We have that all elements of each independent subset of B of size greater than a are
coprime with a. So we can add a to any such set to form an independent set with more
than a elements containing a. So also a has to be coprime with a, thus a = 1. Since the
choice of a was arbitrary, B = {1} which is impossible.
TS-3. Find all natural numbers n for which there exists a convex polyhedron satisfying
the following conditions:
(i) Each face is a regular polygon.
(ii) Among the faces, there are polygons with at most two different numbers of edges.
17
(iii) There are two faces with common edge that are both n-gons.
Answer: 3, 4, 5, 6, 8, 10.
Solution. For n = 3, 4, 5, the well-known regular polyhedrons satisfy the conditions.
If we remove a regular pyramid around each vertex of a regular tetrahedron, cube,
or dodekahedron, the faces of these polyhedrons become 6-gons, 8-gons and 10-gons,
respectively. Clearly one can make the cuts so that all these faces are regular polygons.
At place of vertices that are cut off, equilateral triangles appear. Hence also 6, 8, 10
satisfy the conditions of the problem.
M
C
Show now that there are no other answers. Suppose we
have a polyhedron satisfying the conditions for some n.
Consider two neighbouring n-gons; let AB be the edge sep-
B
arating them (Fig. 14). Suppose A belongs to exactly k other
N ( n − 2) π π
faces. Then 2π > 2 · +k· or, equivalently,
n 3
12
A n < . Thus k > 2 implies n < 6 and k = 1 implies
k
n < 12.
All we have to show is that n = 7, 9, 11 are impossible; so
Figure 14
assume n is odd and 6 < n < 12. Then vertex A, as well as
vertex B, belongs to exactly one face other than the two n-gons. Let m be the number of
vertices of these faces.
Let N be one of the two n-gons under consideration and let M be the face meeting
them at B. Let C be the other endpoint of the edge separating M and N . Let γ be
the sum of all interior angles of the faces other than M and N that meet at C. Then
( n − 2) π
clearly γ > (if only one face meets M and N at C then it must be an n-gon,
n
and if there are more faces then the sum of their angles at C is greater). We obtain the
inequalities
20 π 5π ( m − 2) π ( n − 2) π ( n − 2) π 5π
π> π = 2π − − > 2π − − >γ> > .
21 3 7 m n n 7
From π > γ we see that there are at most 2 other faces meeting M and N at C. If
there were 2 of them, they both would have no more than 5 vertices; therefore they
would not be n-gons and thus they both would be polygons of the second kind. But for
this situation, π > γ gives that they both must be triangles which is contradictory to
5π
γ> . Consequently, there is only one face meeting M and N at C and this must be
7
an n-gon.
Continuing the same way, we find that the faces surrounding N are alternately n-gons
and m-gons. As n is odd by assumption, we get n = m. Thus three n-gons meet at A,
implying n < 6. This completes the solution.
Remark. The condition (iii) of the problem is necessary. Without it, a required poly-
hedron would exist for arbitrary n: a right prism whose bases are regular n-gons and
lateral faces are squares would do.
18
Second day
TS-4. Points A′ , B′ , C′ are chosen on the sides BC, CA, AB of triangle ABC, respec-
| BA′ | |CB′ | | AC′ |
tively, so that = = . The line which is parallel to line B′ C′ and goes
| A′ C| | B′ A| |C′ B|
through point A′ intersects the lines AC and AB at P and Q, respectively. Prove that
| PQ|
> 2.
| B′ C′ |
| BA′ | |CB′ | | AC′ |
Solution 1. Denote = = = k.
| A′ C| | B′ A| |C′ B|
If k = 1 then B′ C′ is the segment connecting midpoints of sides AB and AC of the tri-
angle ABC, so it is parallel to the side BC. Thus the lines PQ and BC coincide, meaning
| PQ| | BC|
P = C and Q = B. Thus ′ ′ = ′ ′ = 2.
|B C | |B C |
| AB′′ |
Now assume that k 6= 1 (Fig. 15). Let B′′ be the point on the side AC such that ′′ =k
| B C|
(in other words, B′′ is symmetric to B′ w.r.t. the perpendicular bisector of AC).
By the intercept theorem, B′′ C′ k BC and A′ B′′ k AB. Thus BC′ B′′ A′ is a parallelogram;
also △ PA′ C ∼ △ B′ C′ B′′ since the corresponding sides are collinear (the triangles exist
| A′ P| | A′ C| | A′ C| 1
since k 6= 1). Hence ′ ′ = ′ ′′ = ′
= . Let C′′ be a point on the side AB
|B C | |C B | | BA | k
| BC′′ | | A′ Q| | A′ B|
such that ′′ = k. By analogy, we have that ′ ′ = = k. It follows that
|C A| |B C | |CA′ |
| PQ| | A′ P| + | A′ Q| 1
′ ′
= ′ ′
= + k > 2.
|B C | |B C | k
C C
P P
A′ A′
B ′′
B′
A′′ B′
Q Q
A C ′′ C′ B A C′ B
Figure 15 Figure 16
Solution 2. From the intercept theorem,
| PQ| | PA| | PB′ | + | AB′ | | PB′ |

= = = + 1.
| B′ C′ | | AB′ | | AB′ | | AB′ |
Let d(X, l ) be the distance of the point X from the line l and S∆ the area of the triangle
∆. Considering lines perpendicular to B′ C′ through points P, resp. A (Fig. 16), we have
| PB′ | d( P, B′ C′ ) d( A′ , B′ C′ ) S△ A ′ B ′ C ′
′
= ′ ′
= ′ ′
= .
| AB | d( A, B C ) d( A, B C ) S△ AB′ C ′
19
Thus
| PQ| | PB′ | S△ A ′ B ′ C ′
′ ′
> 2 ⇐⇒ ′
> 1 ⇐⇒ > 1 ⇐⇒ S△ A′ B′ C ′ > S△ AB′ C ′ .
|B C | | AB | S△ AB′ C ′
| BA′ | |CB′ | | AC′ |

We show now that indeed S△ A′ B′ C ′ > S△ AB′ C ′ . As = = , we have
| A′ C| | B′ A| |C′ B|
| BA′ | |CB′ | | AC′ |
= = = x where 0 < x < 1. We obtain
| BC| |CA| | AB|
1 1
S△ AB′ C ′ = · | AC′ | · d( B′ , AB) = · ( x · | AB|) · ((1 − x ) · d(C, AB))
2 2
1
= x (1 − x ) · · | AB| · d(C, AB) = x (1 − x ) · S△ ABC .
2
x + (1 − x ) 2

1 1
Since from AM-GM, x (1 − x ) 6 = , we have S△ AB′ C ′ 6 · S△ ABC .
2 4 4
1 1
Analogously S△ A′ BC ′ 6 · S△ ABC and S△ A′ B′ C 6 · S△ ABC . To summarize,
4 4
S△ A′ B′ C ′ = S△ ABC − (S△ AB′ C ′ + S△ A′ BC ′ + S△ A′ B′ C )
3 1
> S△ ABC − · S△ ABC = · S△ ABC > S△ AB′ C ′ .
4 4
| BA′ | |CB′ | | AC′ | | BA′ | |CB′ | | AC′ |
Solution 3. As = ′ = , we also have = = , let this
| A′ C| | B A| |C′ B| | BC| |CA| | AB|
ratio be x. Then
−−→′ −→ −→′ −→
AC = x · AB, AB = (1 − x ) · AC,
−−→′ −→ −→ −→ −→
AA = AB + BA′ = AB + x · BC =
−→ −→ −→ −→ −→
= AB + x · ( AC − AB) = (1 − x ) · AB + x · AC.
| PQ| | AP|
Let ′ ′
= v; we need to show that v > 2. Since PQ k B′ C′ , we obtain =
|B C | | AB′ |
| AQ| −→ −→ −→ −→ −−→ −→
′
= v. Hence AP = v · AB′ = v · (1 − x ) · AC and AQ = v · AC′ = v · x · AB. As P,
| AC |
−→ −→ −−→ −→ −→ −→
A′ , Q are collinear, we have PA′ = z · PQ or AA′ − AP = z · ( AQ − AP ). Substituting
−−→ −→ −→
the formulae for AA′ , AP, AQ, we have
−→ −→ −→ −→ −→
(1 − x ) · AB + x · AC − v · (1 − x ) · AC = z · v · x · AB − z · v · (1 − x ) · AC.
−→ −→
Since AB and AC are not collinear, the identity can only hold if their coefficients on both
sides are the same. We have the system of equations

1 − x = zvx,
x − v(1 − x ) = −zv(1 − x ).
20
1−x
From the first equation zv = , and substituting this into the second we get x −
x
(1 − x )2
1−x +x (1 − x )2 + x 2
v (1 − x ) = − · (1 − x ). Thus v = x = . Since (1 − x )2 +
x 1−x (1 − x ) x
x > 2(1 − x ) x, the last equation gives v > 2, just as required.
2
| PQ| | AP|
Solution 4. Let ′ ′ = v; we have to show v > 2. As PQ k B′ C′ , we have =
|B C | | AB′ |
| AQ|
= v. W.l.o.g., assume | AP| > | AC| and | AQ| 6 | AB|. Then
| AC′ |
| AQ| v · | AC′ | v · | AC′ | v · | AC′ | v v

= = = = =
| QB| | QB| | AB| − | AQ| | AB| − v · | AC′ | | AB |
−v
1+ k
k −v
| AC ′ |
and, analogously,
|CP| |CP| | AP| − | AC| v · | AB′ | − | AC| | AC| 1+k

= ′
= ′
= ′
= 1− ′
= 1− .
| PA| v · | AB | v · | AB | v · | AB | v · | AB | v
| AQ| | BA′ | |CP| v

From Menelaus’ theorem, · ′ · = 1. Now substituting gives 1+k ·
| QB| | A C| | PA| k − v
v − (1 + k ) 1+k
k· = 1 which is equivalent to kv − k(1 + k) = − v. Hence v(1 + k) =
v k
1 1
(1 + k) + k(1 + k), leading to v = + k. Therefore v > 2.
k k
TS-5. A strip consists of n squares which are numerated in their order by integers
1, 2, 3, . . . , n. In the beginning, one square is empty while each remaining square con-
tains one piece. Whenever a square contains a piece and its some neighbouring square
contains another piece while the square immediately following the neighbouring square
is empty, one may raise the first piece over the second one to the empty square, remov-
ing the second piece from the strip.
Find all possibilites which square can be initially empty, if it is possible to reach a state
where the strip contains only one piece and a) n = 2008; b) n = 2009.
Answer: a) 2, 5, 2004, 2007; b) none.
Solution. Interpret the move so that the piece at distance 2 from the empty square is
removed from the strip and the piece from the intermediate square moves to the empty
square. In the following, assume n > 5.
First show that, under this interpretation, the location of any piece, as long as it is on the
board, does not change more than by one square. For that, we show that after a move in
some direction, no piece can make next move in the same direction. Perform induction
on the distance between the piece and the end of the strip from which the piece went
farther on the move – assume that the claim holds for smaller distances. Clearly there
are two possibilities for the state after the move: (1) there are no pieces between the
piece that moved and the end of the strip under consideration; (2) there are other pieces
in that area but the nearest among them is separated from the piece that moved by at
21
least two empty squares. But for moving in the same direction, there must be a piece on
the square where the piece came from. By the induction hypothesis, the pieces between
the piece that moved and the end of the strip under consideration do not get farther by
more than one square. Hence the piece that moved cannot move further in the same
direction.
Note that whenever we divide the squares into two non-empty groups then, in order
to achieve the desired final state, one must first obtain a state where some pieces from
different groups are on neighbouring squares – otherwise one could not do a move that
takes off the last piece of one group. Thus if there are three consecutive empty squares,
in both sides of which there are pieces, then the required final state is impossible to
achieve because the pieces from the two groups do not move nearer to each other by
more than one square and there always remains an empty square between them.
Analogously, if there is a lonely piece separated by at least two empty squares from the
nearest piece then the required final state is impossible to achieve since the other pieces
do not reach its neighbouring square and it cannot move itself.
Independently on the location of the empty square, the state after the first move of the
game has two consecutive empty squares and other squares containing a piece. Call
these two squares central (Fig. 17). We show that, from this state, it is possible to reach
the desired final state if and only if there are 2 pieces at one side and an even number of
pieces at the other side of the central squares.
Suppose that this condition is fulfilled; w.l.o.g. there are 2 pieces on the left and 2k pieces
on the right. Show by induction on k that the desired final state can be achieved. If k = 0
then there are only 2 pieces on the board and they are on consecutive squares, so one
move reaches the goal. If k > 0 then first move with the piece on square 2 to the right
and with the piece on square 5 to the left. In the resulting state, pieces are on squares
3 and 4, the following two squares are empty and 2(k − 1) consecutive squares contain
pieces. By the induction hypothesis (it holds for the substrip starting from square 3) it
is possible to achieve the desired final state from this state.
Suppose now there being more than 2 pieces on both sides of the central squares. In
order to achieve the desired final state, the pieces from both sides must get together. It
is possible only if the rightmost piece of the left-hand side moves to the right and the
leftmost piece of the right-hand side moves to the left. The pieces immediately following
these pieces are removed and the remaining pieces do not move nearer by more than
one square. Thus an empty square remains on both sides of the central squares where
no remaining pieces can ever come. Hence when one of the pieces on the central squares
makes a move, the other one is removed and three consecutive squares become empty,
whereby there are still pieces on both sides from them. Thus the desired final state
cannot be achieved.
initial state
after the initial move
central squares
Figure 17
22
Also if there is 1 piece on one side from the central squares then the desired final state
is unreachable. If there is 0 pieces on one side then there will be one piece after one
move, consequently the final state is unreachable again. It remains to study the case
with 2 pieces on one side – w.l.o.g., on the left – and an odd number 2k + 1 of pieces
on the other side. Show by induction on k that the required final state is unreachable
even if the strip had more empty squares to the left from square 1. If k = 0 then both
possible moves (from 2 to the right and from 1 to the left) leads to a stub configuration
with 2 lonely pieces. Let k > 0. The move from 1 to the left gives a state where three
consecutive squares are empty, thus uninteresting. The moves from 2 to the right and
from 5 to the left remain, which both are certainly performed during the play. Suppose
w.l.o.g. that the very first move is from 2 to the right because the possible moves before
it concern the pieces on the right only and they can be performed also if the move from 2
to the right is already done. Hence in the state after two moves, squares 3 and 4 contain
pieces, the next two squares are empty and then 2(k − 1) + 1 consecutive squares with
pieces follow. By the induction hypothesis, the desired state is unreachable.
We have got that the desired state is possible only for even n since only then there can
be 2 pieces on one side and an even number of pieces on the other side from two central
squares. In the case of even n, the state where squares 3 and 4 are empty can arise
if, initially, 2 or 5 is empty. Analogously, the symmetric case is handled. Hence, for
n = 2008, the squares 2, 5, 2004, 2007 can be initially empty.
TS-6. For any positive integer n, let c(n) be the largest divisor of n not greater than
√
n and let s(n) be the least integer x such that n < x and the product nx
is divisibleby
n
an integer y where n < y < x. Prove that, for every n, s(n) = (c(n) + 1) · +1 .
c (n)

n
Solution. Take y = c(n) · + 1 . Then
c (n)

n n
n = c (n) · < y < ( c ( n ) + 1) · +1
c (n) c (n)
while

n n n
y | c (n) · · ( c ( n ) + 1) · + 1 = n · ( c ( n ) + 1) · +1 .
c (n) c (n) c (n)

n
This implies s(n) 6 (c(n) + 1) +1 .
c (n)
n
For establishing minimality, choose x such that n < x < (c(n) + 1)( + 1). It has to
c (n)
be proven that no integer strictly between n and x divides nx.
y
Let y be an arbitrary integer strictly between n and x. Let d = gcd(y, n) and y′ = ,
d
n
n′ = .
d
23
n
As c(n) and are the central divisors of n, we have
c (n)
n n
c (n) + 6 d+ . (4)
c (n) d
1
(Indeed, consider function f (z) = z + ; it is decreasing between 0 and 1. Then c(n) +
z
n √ c (n) n √ min(d, n ) min(d, nd ) c (n)
= n · f ( √ ) and d + = n · f ( √ d ) as well as √ 6 √ 6 1.)
c (n) n d n n n
Adding n + 1 to both sides of (4) and factorizing, we get

n n
( c ( n ) + 1) · + 1 6 ( d + 1) · +1 . (5)
c (n) d
But d | y implies d | y − n and furthermore d 6 y − n. Thus n + d 6 y, implying

n y
+ 1 6 = y′ . (6)
d d
Inequalities (5) and (6) together give

′ n n
dy = y < x < (c(n) + 1) · + 1 6 ( d + 1) · + 1 6 ( d + 1) y ′ .
c (n) d
Hence x is strictly between two consecutive multiples dy′ and (d + 1)y′ of y′ , therefore
y′ ∤ x.
If y | nx were the case then reducing by d would lead to y′ | n′ x. As y′ is relatively prime
to n′ , this would imply y′ | x. This contradiction completes the solution.
Remark. The table for small values of functions c and s is
n 1 2 3 4 5 6 7 8 9 10 11 12
c (n) 1 1 1 2 1 2 1 2 3 2 1 3
s (n) 4 6 8 9 12 12 16 15 16 18 24 20
n 13 14 15 16 17 18 19 20 21 22 23 24
c (n) 1 2 3 4 1 3 1 4 3 2 1 4
s (n) 28 24 24 25 36 28 40 30 32 36 48 35
Problems Listed by Topic

Number theory: OC-2, OC-6, FR-2, FR-4, FR-9, FR-10, FR-15, TS-2, TS-6
Algebra: OC-8, FR-5, FR-6, FR-11, FR-13, FR-16, TS-1
Geometry: OC-1, OC-3, OC-7, FR-1, FR-7, FR-14, FR-17, TS-3, TS-4
Discrete mathematics: OC-4, OC-5, FR-3, FR-8, FR-12, FR-18, TS-5
24
2009/2010

Tartu 2010
WE THANK:
University of Tartu
Problem authors: Juhan Aru, Maksim Ivanov, Urve Kangro, Oleg Koshik,
Mihkel Kree, Toomas Krips, Härmel Nestra, Hendrik Nigul, Uve Nummert,
Translators: Urve Kangro, Härmel Nestra
Editors: Uve Nummert, Reimo Palm

The Estonian Mathematical Olympiad is held annually in three rounds – at
the school, town/regional and national levels. The best students of each
round (except the final) are invited to participate in the next round. Every
year, about 110 students altogether reach the final round.
In each round of the Olympiad, separate problem sets are given to the stu-
dents of each grade. Students of grade 9 to 12 compete in all rounds, students
of grade 7 to 8 participate at school and regional levels only. Some towns,
regions and schools also organise olympiads for even younger students. The
school round usually takes place in December, the regional round in January
or February and the final round in March or April in Tartu. The problems
for every grade are usually in compliance with the school curriculum of that
grade but, in the final round, also problems requiring additional knowledge
may be given.
The first problem solving contest in Estonia took place already in 1950.
The next one, which was held in 1954, is considered as the first Estonian
Mathematical Olympiad.
Apart from the Olympiad, open contests are held twice a year, usually in
October and in December. In these contests, anybody who has never been
enrolled in a university or other higher education institution is allowed to
participate. The contestants compete in two separate categories: the Juniors
and the Seniors. In the first category, students up to the 10th grade can par-
ticipate; the other category has no restriction. Being successful in the open
contests generally assumes knowledge outside the school curriculum.
According to the results of all competitions during the year, about 20 IMO
team candidates are selected. IMO team selection contest for them is held in
April or May. This contest lasts two days; each day, the contestants have 4.5
hours to solve 3 problems, similarly to the IMO. All participants are given
the same problems. Some problems in our selection contest are at the level of
difficulty of the IMO but somewhat easier problems are usually also included.
The problems of previous competitions are available at the Estonian Math-
ematical Olympiad’s website http://www.math.olympiaadid.ut.ee/eng.
Besides the above-mentioned contests and the quiz “Kangaroo” other re-
gional competitions and matches between schools are held as well.
*
This booklet contains problems that occurred in the open contests, the fi-
nal round of national olympiad and the team selection contest. For the open
contests and the final round, selection has been made to include only prob-
lems that have not been taken from other competitions or problem sources
and seem interesting enough. The team selection contest is presented entirely.
3
OC-1. Find all positive integers n such that 1 + 22 + 33 + 4n is a perfect
square. (Juniors.)
Answer: n = 1.
Solution 1. Let 1 + 22 + 33 + 4n = x2 . This implies 32 = x2 − 4n or, equiv-
alently, 25 = ( x − 2n )( x + 2n ). As the l.h.s. is a power of 2, the factors in the
r.h.s. are of the form x − 2n = 2a and x + 2n = 25− a where a is 0, 1 or 2. Sub-
tracting the first of the two equalities from the second gives 2n+1 = 25− a − 2a .
This leads to an integral n only if a = 2; then n = 1. A check shows that
1 + 22 + 33 + 41 = 62 indeed.
Solution 2. Observe that 1 + 22 + 33 + 4n = 32 + 4n = 25 + 22n = 25 ·
(1 + 22n−5 ). If n > 3, then 2n − 5 > 1; hence 22n−5 is an even integer and
1 + 22n−5 is therefore odd. Thus in the prime factorization of the number
given in the problem, the exponent of 2 is 5. As this is odd, the number cannot
be a perfect square. If n = 2 or n = 1, then 32 + 4n = 48 or 32 + 4n = 36,
respectively, where only the latter is a perfect square. Consequently, only
n = 1 is possible.
Solution 3. If n = 1, 2, 3, then the given number is 36, 48, 96, respectively,
where only the first is a perfect square. If n > 4, then 2 · 2n + 1 > 2 · 16 + 1 >
32, implying (2n )2 < (2n )2 + 32 < (2n )2 + 2 · 2n + 1 = (2n + 1)2 . As the
number under question is equal to (2n )2 + 32, it falls between two consecutive
perfect squares, hence cannot be a perfect square itself.
OC-2. Given a convex quadrangle ABCD with | AD | = | BD | = |CD | and

∠ ADB = ∠DCA, ∠CBD = ∠BAC, find the sizes of the angles of the quad-
rangle. (Juniors.)
Answer: 75◦ , 120◦ , 45◦ , and 120◦ .
A
Solution 1. Denote ∠ ADB = ∠ DCA = α and
∠CBD = ∠BAC = β (Fig. 1). In triangle DAC α β
we have | DA| = | DC | and therefore ∠ DAC =
∠DCA = α; analogously in triangles DAB and D α α+β
DBC, we have ∠ DBA = ∠ DAB = α + β and β B
∠DCB = ∠DBC = β, respectively. So ∠BCA =
β − α. From triangle ABC now β + α + β + β +
β − α = 180◦ or, equivalently, 4β = 180◦ , giving
β = 45◦ . From triangle ADB we get α + β + α +
β + α = 180◦ or, equivalently, 3α = 180◦ − 2β =
α
90◦ and α = 30◦ . Therefore, the sizes of the an-
β−α
gles of quadrangle ABCD are ∠ DAB = α + β =
C
75◦ , ∠ ABC = α + 2β = 120◦ , ∠ BCD = β = 45◦ ,
and ∠CDA = 360◦ − 75◦ − 120◦ − 45◦ = 120◦ . Fig. 1
4
Solution 2. We use the same notation as in the Solution 1. Triangle BCD
is isosceles, hence ∠ DCB = ∠ DBC = β. As D is the circumcenter of ABC,
we have ∠ BDC = 2∠ BAC = 2β. The sizes of the angles of triangle BCD are
therefore β, β, and 2β; thus β + β + 2β = 180◦ , whence β = 45◦ . As ∠ BCA =
∠BDA , we have ∠ BCD = α + α = β, whence α = 2 β = 30◦ . Consequently,
2 2 3
◦
the sizes of the angles of quadrangle ABCD are ∠ DAB = ∠ ABD = 1802 −α =
75◦ , ∠ ABC = ∠ ABD + ∠CBD = 75◦ + 45◦ = 120◦ , ∠ BCD = β = 45◦ , and
∠CDA = ∠CDB + ∠BDA = 90◦ + 30◦ = 120◦ .
Remark. The convexity of the quadrangle actually follows from the other
constraints of the problem. Namely, consider the circle with center D, passing
through points A and B, and a point C on it. If A and B were on different sides
from line CD, we would have ∠ ADB > ∠ BCA > ∠ DCA, hence ∠ ADB and
∠DCA could not be equal.
OC-3. In the buffet of the kitchen, there are three candy boxes, each con-
taining the same number of candies. Every time when Juku goes into the
kitchen, he takes either three candies from one box or one candy from ev-
ery box. Prove that irrespectively of how Juku takes the candies, he always
retains the possibility to completely clean out all candy boxes. (Juniors.)
Solution. The difference of the numbers of candies in any two boxes can
only be a multiple of 3 because it is 0 in the beginning and, with every move,
it changes by either 0 or 3. Hence, starting from an arbitrary intermediate
state, Juku can clean out the boxes as follows: he takes one candy from each
box as many times as possible, after which one box is empty and the number
of candies in each of the other two is divisible by 3, and then empties the
other boxes by taking three candies from one box every time.
OC-4. Four musketeers together bought a plot of rectangular shape and

paid for it equally. They divided the plot by two cuts into four pieces of
rectangular shape, from which every musketeer got one. It turned out that
one musketeer obtained as much land as the other three in total. Prove that
the price per acre of one musketeer’s piece turned out as large as the sum of
the prices per acre of the other three musketeers’ pieces. (Juniors.)
Solution. Let a and b be the side lengths of the plot. Assume that the
cuts divided the side of length a to parts of length x and a − x where x be-
ing the greater part, and the side of length b to parts of length y and b − y
where y being the greater part. Then the area of the largest piece was xy.
The condition that this area equals the sum of the areas of the other three
pieces can be written as follows:
xy = ( a − x )y + x (b − y) + ( a − x )(b − y).
5
Dividing both sides by x ( a − x )y(b − y), one obtains
1 1 1 1
= + + .
( a − x )(b − y) x (b − y) ( a − x )y xy
If the price that every musketeer paid for the plot was 1, then the l.h.s. of the
last equality is precisely the price per area unit of the piece with area ( a − x ) ·
(b − y). Analogously, the r.h.s. equals the sum of the prices per area unit of
the other three pieces. Hence multiplying the sides of this equality by the
number of area units per acre, the claim of the problem follows.
OC-5. Let a be a fixed real number. Find all real numbers b such that, for
every real number x, at least one of the numbers x2 + ax + b and x2 − ax + b
is non-negative. (Juniors.)
Answer: b > 0.
Solution. Note that x2 + ax + b and x2 − ax + b sum up to 2x2 + 2b. If
b > 0, then it is non-negative for arbitrary real number x, implying that at
least one of the numbers added was non-negative. If b < 0, then taking x = 0
turns both summands negative.
Remark. This problem can be solved also in technical ways, by calculating
the negative and non-negative domains of the quadratic polynomials.
OC-6. Call a positive integer n prime-prone if there exist at least three prime
numbers from which we can get n by removing the last digit. Prove that
every two prime-prone positive integers differ from each other by at least 3.
(Juniors.)
Solution. As the prime numbers under consideration have at least two
digits, the last digit can be only 1, 3, 7, or 9. Thus n is prime-prone if and only
if, among numbers 10n + 1, 10n + 3, 10n + 7, and 10n + 9, at least three are
primes.
If n = 3k, then 10n + 3 = 30k + 3 and 10n + 9 = 30k + 9 are divisi-
ble by 3 and hence composite. If n = 3k + 2, then 10n + 1 = 30k + 21 and
10n + 7 = 30k + 27 are divisible by 3 and hence composite again. Conse-
quently, all prime-prone integers are congruent to 1, and hence to each other,
modulo 3. Thus they differ by a multiple of 3, i.e., by at least 3.
OC-7. Does there exist a prime number p such that both p3 + 2008 and p3 +
2010 are primes as well? (Seniors.)
Answer: no.
Solution. Let p be any prime number. If p is not divisible by 7, then p3
is congruent to either 1 or −1 modulo 7. Since 2008 ≡ −1 (mod 7) and
2010 ≡ 1 (mod 7), either of the numbers p3 + 2008 and p3 + 2010 is divisible
by 7 and hence composite. If p is divisible by 7, then p = 7 and p3 + 2010 =
73 + 2010 = 2353 = 13 · 181 is composite, too.
6
OC-8. In a regular n-gon, either 0 or 1 is written at each vertex. Using
non-intersecting diagonals, Juku divides this polygon into triangles. Then
he writes into each triangle the sum of the numbers at its vertices. Prove that
Juku can choose the diagonals in such a way that the maximal and minimal
number written into the triangles differ by at most 1. (Seniors.)
Solution. If all numbers written at the vertices of the polygon are equal,
then the claim holds trivially. Hence assume that there are both zeros and
ones among the numbers at the vertices. We prove by induction that, for
every convex polygon, the partition into triangles can be chosen in such a
way that Juku writes either 1 or 2 to each triangle.
If n = 3, then this claim holds since the sum of the numbers at the vertices
of a triangle can be neither 0 nor 3. If n = 4 (Fig. 2), then draw the diagonal
that connects the vertices where 0 and 1 are written, respectively, or, if such
a diagonal does not exist, then an arbitrary diagonal. In both cases, only
sums 1 and 2 can arise. If n > 5, then choose two consecutive vertices with
different labels and a third vertex P that is not neighbour to either of them
(Fig. 3). Irrespective of whether the label of P is 0 or 1, we can draw the
diagonal from it to one of the two consecutive vertices chosen before so that
the labels of its endpoints are different. Now the polygon is divided into two
convex polygons with smaller number of vertices so that both 0 and 1 occur
among their vertex labels. By the induction hypothesis, both polygons can be
partitioned into triangles with sum of labels of vertices either 1 or 2.
P
⋆ 1 1 0
0 ⋆ 0 1 0 1
Fig. 2 Fig. 3
OC-9. Circle c passes through vertices A and B of an isosceles triangle ABC,

whereby line AC is tangent to it. Prove that circle c passes through the cir-
cumcenter or the incenter or the orthocenter of triangle ABC. (Seniors.)
Solution. Consider three cases | AB| = | AC |, | BC | = | BA|, and |CA| = |CB|.
1. We show that if | AB| = | AC | (Fig. 4), then circle c passes through the
circumcenter of ABC. Let O be the point at the same side from AB
as C that is the intersection of the perpendicular bisector of side AB
and circle c. Then ∠OAB = ∠OBA and, by inscribed angles theorem,
7
C
C
C
O E
H A I B
A A B
B
c c c
Fig. 4 Fig. 5 Fig. 6
∠OBA = ∠OAC. Hence O lies on the bisector of angle CAB. Since

| AB| = | AC |, this angle bisector is also the perpendicular bisector of
side BC. Consequently, O is the intersection point of the perpendicular
bisectors of the sides of triangle ABC.
2. For the case | BC | = | BA| (Fig. 5), we show that circle c passes through
the orthocenter of triangle ABC. Let E be the foot of the altitude of
triangle ABC drawn from B and let H be the second intersection point
of this altitude with circle c (in the special case with tangency and no
intersection, take H = B). By the inscribed angles theorem, ∠EBA =
∠EAH. Thus ∠ ACB + ∠CAH = ∠CAB + ∠EBA = 90◦ whence AH ⊥
BC. Consequently, H is the orthocenter.
3. Finally, we show that if |CA| = |CB| (Fig. 6), then circle c passes through
the incenter of triangle ABC. Let I be the intersection point of the bi-
sector of angle CAB with circle c. By the inscribed angles theorem,
∠CAI = ∠ IBA. Hence ∠BAI = ∠ IBA whence I lies on the perpen-
dicular bisector of side AB. As |CA| = |CB|, this perpendicular bisector
is also the bisector of angle ACB. Consequently, I is the intersection
point of the angle bisectors.
OC-10. Let n > 2. Positive integers a1 , a2 , . . . , an whose sum is even and

which satisfy ai 6 i for every i = 1, 2, . . . , n, are given. Prove that it is possible
to choose signs in the expression a1 ± a2 ± . . . ± an in such a way that its value
becomes 0. (Seniors.)
Solution 1. Prove the claim by induction on n. If n = 2, then the only way
to choose integers that satisfy the conditions of the problem is a1 = 1 and
a2 = 1. In this case, a1 − a2 = 0. Assume now that the claim holds whenever
2 6 n 6 k and show that it holds also for n = k + 1. Consider two cases.
1. If ak+1 = ak , then a1 + a2 + . . . + ak−1 is even. As this case is possible
only for k > 2, the induction hypothesis is applicable for n = k − 1.
Thus it is possible to choose signs in the expression a1 ± a2 ± . . . ± ak−1
8
in such a way that it evaluates to 0. Adding ak − ak+1 to it, the desired
expression for n = k + 1 is obtained.
2. If ak 6= ak+1 , then consider integers a1 , . . . , ak−1 , | ak − ak+1 |. As | ak −
ak+1 | and ak + ak+1 have the same parity, the sum of these k numbers is
even. Also note that 1 6 | ak − ak+1 | 6 k. Thus these numbers satisfy the
conditions of the problem, so it is possible to choose signs in the expres-
sion a1 ± a2 ± . . . ± ak−1 ± | ak − ak+1 | in such a way that it evaluates to
0. As either | ak − ak+1 | = ak − ak+1 or | ak − ak+1 | = ak+1 − ak , this also
leads to a corresponding expression for numbers a1 , a2 , . . . , ak , ak+1 .
Solution 2. Prove by induction on i that, for each i and s such that 1 ≤ i ≤ n
and 1 ≤ s ≤ a1 + . . . + ai , it is possible to choose some of the numbers a1 , . . . ,
ai that sum up to s. If i = 1, then this claim holds since a1 = 1. Assume that
the claim holds for i = k − 1 and consider the case i = k. Let S = a1 + . . . + ak
and S′ = a1 + . . . + ak−1 . If 1 ≤ s ≤ S′ , then the desired statement holds by the
induction hypothesis. If S′ < s ≤ S, then 0 ≤ s − ak ≤ S′ (the first inequality
holds because s − ak ≥ s − S′ − 1 > 0, implied by ak ≤ k and S′ ≥ k − 1;
the second inequality follows from S = S′ + ak ). Therefore, to get the sum s,
we can choose the number ak , and if s − ak > 0, then add to it those numbers
among a1 , . . . , ak−1 whose sum is s − ak , using the induction hypothesis.
Let now a1 + a2 + . . . + an = 2T. Choose the numbers among a1 , a2 , . . . ,
an that sum up to T. This divides all the numbers into two groups with equal
sum. It remains to write minuses in front of every term of the group that does
not contain a1 .
Solution 3. Start choosing signs from right to left. Denote S1 = an and
define Sk+1 , k = 1, . . . , n − 1, as follows: if Sk > 0, then Sk+1 = Sk − an−k ,
otherwise Sk+1 = Sk + an−k . We show that then always |Sk | 6 n − k + 1. This
holds if k = 1. Assume therefore that it holds for k = m and prove it for
k = m + 1. If Sm > 0, then Sm+1 = Sm − an−m 6 (n − m + 1) − 1 = n − m
and Sm+1 = Sm − an−m > 0 − (n − m), hence |Sm+1 | 6 n − m. If Sm < 0, then
Sm+1 = Sm + an−m < 0 + n − m and Sm+1 = Sm + an−m > −(n − m + 1) +
1 = −(n − m), hence |Sm+1 | 6 n − m again.
Now |Sn | 6 1 since |Sk | 6 n − k + 1 for every k = 1, . . . , n. Thus Sn = 0
as the sum of all terms is even. If in this formal sum, the term a1 has minus
sign, turn all signs to the opposite one.
Remark. Solution 2 shows that the assumption ak 6 k for all k = 1, . . . , n
could be replaced with the more general assumption ak 6 1 + a1 + . . . + ak−1
for all k = 1, . . . , n.
One can also note that the lemma proved at the beginning of Solution 2
does not need the assumption that the sum of all numbers is even.
9
OC-11. The diagonals of trapezoid ABCD with bases AB and CD meet at
P. Prove the inequality SPAB + SPCD > SPBC + SPDA , where SXYZ denotes
the area of triangle XYZ. (Seniors.)
Solution 1. Let a = | AB| and b = |CD | and let h a and hb be the altitudes of
triangles PAB and PCD drawn from P (Fig. 7). Denote S1 = SPAB + SPCD and
S2 = SPBC + SPDA . Then S1 = 21 ( ah a + bhb ) and S1 + S2 = 12 ( a + b)(h a + hb ),
whence S2 = 12 ( ahb + bh a ). Since triangles PAB and PCD are similar, a > b
implies h a > hb and also a 0,
2 2
i.e., S1 > S2 .
Solution 2. Let M and N be the intersection A a B
points of the arms BC and DA of the trape- ha
zoid with the line being parallel to the bases
of the trapezoid and passing through point P. N M
Let l be the length of MN, let d be the length P
hb
of the midline of the trapezoid, and let h and
S be the height and the area of the trapezoid,
respectively. Let S′ = SPBC + SPDA . Then D b C
′ 1
S = dh and S = 2 lh whence it suffices to Fig. 7
show that l < d.
W.l.o.g., assume | AB| < |CD |. Comparing the heights of similar triangles
PAB and PCD shows that MN is closer to base AB than to base CD. Thus
MN is situated between the midline and the shorter base AB. Consequently,
MN is shorter than the midline.
Solution 3. Let a = | AB| and b = |CD |. Let h and S be the height and
the area of the trapezoid, respectively, and let h a and hb be the heights corre-
sponding to vertex P of triangles PAB and PCD, respectively. Similar trian-
gles PAB and PCD imply h a : hb = a : b. As h a + hb = h we get
a b
ha = · h, hb = · h.
a+b a+b
Now
1 1 a 2 + b2
SPAB + SPCD = ( ah a + bhb ) = · · h.
2 2 a+b
It suffices to show that SPAB + SPCD > S2 or, equivalently,
1 a 2 + b2 1 a+b
· ·h > · · h,
2 a+b 2 2
or, equivalently, 2( a2 + b2 ) > ( a + b)2 . But the last inequality is equivalent to
( a − b)2 > 0 (a 6= b since a and b are the lengths of the bases of the trapezoid).
10
OC-12. Call pure any positive integer n that does not occur in any integer
sequence c0 , c1 , c2 , . . . , where 0 < c0 < n and
( 1
2 c i −1 if ci−1 is even,
ci =
3ci−1 − 1 if ci−1 is odd,
for every i > 1. (For instance, 10 is not pure since it occurs in the sequence 5,
14, 7, 20, 10, . . . )
a) Is every positive multiple of 3 pure?
b) Prove that if an integer n > 1 is pure but not divisible by 3, then n + 1
is divisible by 6.
(Seniors.)
Answer: a) yes.
Solution. a) Note that 3ci−1 − 1 is never divisible by 3 and if 12 ci−1 is di-
visible by 3, then also ci−1 is divisible by 3. Thus, if some term ck = n is
divisible by 3, then, up to it, only dividing by 2 is used to build the terms
(i.e., ci = 21 ci−1 for every i such that 1 6 i 6 k) and, consequently, c0 > c1 >
. . . > ck = n. But this contradicts the condition c0 < n. Hence every positive
multiple of 3 is pure.
b) If n is not divisible by 3, then n = 3k + 1 or n = 6k + 2 or n = 6k + 5.
If n = 3k + 1, then taking c0 = 2k + 1 gives c1 = 6k + 2 and c2 = 3k + 1 = n.
Thereby k > 0 since n > 1, therefore c0 < n. Hence none of such numbers n
is pure. If n = 6k + 2, then taking c0 = 2k + 1 gives c1 = 6k + 2 = n, whereby
c0 < n. Hence also none of such numbers n is pure. Hence, among the
positive integers n > 1 not divisible by 3, only those of the form n = 6k + 5
can be pure.
Remark. Not every integer of the form n = 6k + 5 is pure. For example,
23 = 6 · 3 + 5 occurs in the sequence 21, 62, 31, 92, 46, 23, . . .
OC-13. Let a and b the lengths of the legs of a given right triangle. Prove
that angle ϕ, where 0 < ϕ < 90◦ , is an acute angle of this triangle if and only
if ( a cos ϕ + b sin ϕ)( a sin ϕ + b cos ϕ) = 2ab. (Seniors.)
Solution 1. The equality given in the problem is equivalent to
( a2 + b2 ) sin ϕ cos ϕ + ab(sin2 ϕ + cos2 ϕ) = 2ab
and hence also to
( a2 + b2 ) sin ϕ cos ϕ = ab. (1)
Let α and β be the
√ angles opposite to legs with
√ length a and b, respectively.
Then sin α = a/ a2 + b2 , sin β = cos α = b/ a2 + b2 , implying
( a2 + b2 ) sin α cos α = ab.
Comparing this to (1) shows the equivalence of the equality of the problem
and the equality sin ϕ cos ϕ = sin α cos α, i.e., equality sin 2ϕ = sin 2α. As
11
0 < α, β < 90◦ , this implies 2ϕ = 2α or 2ϕ = 180◦ − 2α, whence ϕ = α or
ϕ = 90◦ − α = β. Hence, ϕ satisfies the equality if and only if it equals one of
the acute angles of the right triangle.
Solution 2. Let ABC be the given trian- C
gle with right angle at vertex C. Let c be the
length of its hypothenuse and h be the height
corresponding to the hypothenuse. Let C ′ be
b h a
a point on the circumcircle of ABC such that
′
one acute angle of triangle ABC is ϕ (Fig. 8). A c
ϕ B
Let a′ and b′ be the lengths of the legs of tri-
angle ABC ′ and h′ be the height of the trian- b′ a′
′ h′
gle ABC corresponding to its hypothenuse.
Then ( a2 + b2 ) sin ϕ cos ϕ = c2 sin ϕ cos ϕ =
(c sin ϕ)(c cos ϕ) = a′ b′ = ch′ . Since the equal- C′
ity in the problem is equivalent to the equal-
ity (1) from the Solution 1, it is also equiva- Fig. 8
lent to ch′ = ab. But ab = ch, hence it is also
equivalent to h = h′ . This condition holds if and only if ABC ∼ ABC ′ or
ABC ∼ BAC ′ , i.e., ϕ equals one of the acute angles of triangle ABC.
Selected Problems from the Final Round

of National Olympiad
FR-1. Let a, b and c be positive integers such that ab is divisible by 2c, bc is
divisible by 3a and ca is divisible by 5b. Find the least possible value of abc.
(Grade 9.)
Answer: 900.
Solution. Since ab is divisible by 2c and ca is divisible by 5b, ab · ca must be
divisible by 2c · 5b, hence a2 is divisible by 2 · 5. Therefore a2 is divisible by 2
and 5, hence a is divisible by 2 and 5. Similarly b is divisible by 2 and 3, and c
is divisible by 3 and 5. Consequently abc is divisible by 2 · 5 · 2 · 3 · 3 · 5 = 900.
On the other hand, a = 10, b = 6 and c = 15 satisfy the conditions and
abc = 900.
FR-2. Prove the inequality

22 + 1 32 + 1 20102 + 1 1
2010 < 2 + 2 +...+ < 2010 .
2 −1 3 −1 20102 − 1 2
(Grade 9.)
n2 + 1 1 1
Solution. Since = 1+ − , the given sum can
(n − 1)(n + 1) n−1 n+1
1 1 1 1 1 1
be rewritten in the form 1 + − + 1 + − + . . . + 1 + − =
1 3 2 4 2009 2011
12
1 1 1 1 1 1 1
2010 + − − . Because 0 < − − < , the inequality
2 2010 2011 2 2010 2011 2
is proved.
FR-3. Juku drew a regular hexagon and chose three triangles with different
areas whose vertices were among the vertices of the hexagon. Prove that the
sum of the areas of the triangles is equal to the area of the hexagon. (Grade 9.)
Solution. Any triangle whose vertices are among the vertices of a regular
hexagon is one of the following:
• a triangle ∆1 whose vertices are three consecutive vertices of the hexa-
gon;
• a triangle ∆2 whose two vertices are adjacent vertices of the hexagon
and the third one is adjacent to none of the first two;
• a triangle ∆3 where any two vertices are not adjacent vertices of the
hexagon.
Since the areas of the chosen triangles are different, the triangles must be
equal to the triangles ∆1 , ∆2 , ∆3 . The hexagon can be divided into four parts
(Fig. 9): the triangle ∆3 surrounded by three triangles ∆1 . The area of the tri-
angle ∆2 (marked by a dotted line in Fig. 9) is twice the area of the triangle ∆1
beacuse they have the same base but the height of ∆2 is twice the height of ∆1 .
FR-4. Points A′ , B′ and C ′ are chosen correspondingly on the sides AB, BC,
| A′ B| | B′ C| |C′ A |
and CA of an equilateral triangle ABC so that | AB| = | BC| = |CA| = k. Find
all positive real numbers k for which the area of the triangle A′ B′ C ′ is exactly
half of the area of the√triangle ABC. (Grade 10.)
Answer: k = 12 ± 63 .
Solution. Let α be the angle at the vertex A (Fig. 10). The area of the triangle
AA′ C ′ is S AA′ C′ = 12 · | AA′ | · | AC ′ | · sin α = 12 · (1 − k)| AB| · k| AC | · sin α =
(1 − k)k S ABC . Similarly SBB′ A′ = (1 − k)k S ABC and SCC′ B′ = (1 − k)k S ABC .
Hence the triangles AA′ C ′ , BB′ A′ and CC ′ B′ are of equal area. Therefore
A
α
C′
A′
B B′ C
Fig. 9 Fig. 10
13
the area of the triangle A′ B′ C ′ is half of the area of the triangle ABC iff the
area of the triangle AA′ C ′ is one sixth of the area of the triangle
√
ABC, i.e.
(1 − k)k = 16 . The solutions of k2 − k + 16 = 0 are k1,2 = 12 ± 63 , both of them
are positive.
Remark. As seen from the solution, the result actually holds for an arbi-
trary triangle.
FR-5. Three players A, B and C play the following game. At the beginning
of the game, each player has a sheet of paper with the name of the player
written on it. Player A chooses one of the other players and replaces the name
on this player’s sheet with the name on his own sheet. Then player B makes
a similar move, then player C and after that the turn to move goes to player
A again. The game ends when all the sheets have the same name written on
them and the winner is the player whose name it is. Does any of the players
have a winning strategy (i.e., a strategy that allows a player to win no matter
what his opponents play)? (Grade 10.)
Answer: no.
Solution 1. Player B does not have a winning strategy, since on the first
move player A can write the name A on his sheet, after that the name B is not
on any of the sheets. Similarly player C does not have a winning strategy.
To prove that even player A does not have a winning strategy, we show
that players B and C have a joint strategy which guarantees that among the
names written on the sheets there are at least two different names. Namely,
if player A on his move writes a name on the sheet of player B, then B writes
a name on the sheet of player A, otherwise on the sheet of player C. Player C
always writes a name on the sheet of player B.
In the beginning both players B and C have names different from the name
on the sheet of player A. Hence A cannot win in one move. Independent of
which name A changes on his move, after B moves, the name on the sheet of
C differs from the name on the sheet of A, and after C moves, both B and C
have names on their sheets different from the one on the sheet of A, as in the
beginning. So the cycle repeats.
Solution 2. Denote the players starting from any player in the order of their
turns by X, Y, and Z. Show that the players Y and Z can together always keep
X from winning. Indeed, X can win only on his move because Y and Z can
always play so that their move does not result immediately in X winning. X
can win on his turn only if before his move he and somebody else have his
name on their sheets. The player Z cannot prevent this situation only if the
same situation occurred already before his move and his sheet has the name
of X on it. But after Y moves, then either X or Z has the same name on their
sheets as Y has, and so Y can always prevent both X and Z having the same
name on their sheets. Thus none of the three players has a winning strategy.
14
FR-6. A regular 2010-gon is divided into pieces of triangular shape. Find
the least possible number of pieces. (Grade 10.)
Answer: 2008.
Solution. All the interior angles of the 2010-gon can be built from the in-
ner angles of the triangular pieces. As the sum of the inner angles of the
2010-gon is 2008 · 180◦ and that of every triangle is 180◦ , there must be at
least 2008 triangles. On the other hand, each convex 2010-gon can be divided
into exactly 2008 triangles by choosing one vertex and cutting the figure into
pieces along the diagonals that start from this vertex.
FR-7. Let x, y and z be positive integers satisfying gcd( x, y, z) = 1. Prove

that if (y2 − x2 ) − (z2 − y2 ) = ((y − x ) − (z − y))2 , then x and z are perfect
squares. (Grade 11.)
Solution. Remove the parentheses, collect the terms and divide both sides
by 2 to get x2 + y2 + z2 − 2xy − 2yz + xz = 0. This equality can be written
as ( x − y + z)2 = xz. Hence xz is a square of an integer. If x and z have a
common divisor d, then xz is divisible by d2 , and by the previous equality
( x − y + z)2 is divisible by d2 , therefore x − y + z is divisible by d. Since x and
z are divisible by d, y must be divisible by d, hence d = 1, i.e. x and z do not
have common divisors. Since xz is a square of an integer, it follows that both
x and z are squares of integers.
FR-8. Find all pairs of integers (m, n) such that for all positive real numbers
x and y the inequality x m + yn > x n ym holds. (Grade 11.)
Answer: (0, 0).
Solution. If m = 0, then the inequality is 1 + yn > x n . This holds for all
positive real numbers x and y iff n = 0. Hence (0, 0) is a solution. Let now
both m and n be different from zero. If the pair (m, n) satisfies the condi-
tion, then substituting x and y by 1x and 1y we see that the pair (−m, −n) also
satisfies the condition. Hence we can assume without loss of generality that
m > n and m > 0. If m > n, then by taking x = 1 we get 1 + yn > ym , which
does not hold for y large enough. Hence m = n. By taking x = y = 4 we get
2 · 4m > 42m which does not hold for any positive integer m. Therefore there
are no more suitable pairs.
FR-9. Let D be the midpoint of side BC of triangle ABC. Prove that the
intersection point of medians of triangle ABD and that of triangle ACD are
equidistant from line AD. (Grade 11.)
Solution. Triangles ABD and ACD have equal area since | BD | = |CD | and
the altitudes drawn from A coincide (Fig. 11). As these triangles have a com-
mon side AD, also the altitudes drawn from vertices B and C, respectively,
must be equal. Thus B and C are equidistant from line AD. Since the point
15
A
M
N
D
B C
Fig. 11
of intersection of medians cuts 13 part of every median, the distance of the

point of intersection of medians of ABD from line AD is thrice shorter than
the distance of point B from line AD. An analogous relation holds also for
triangle ACD. Hence the claim follows.
FR-10. A unit L-shape consists of three unit squares as shown in

the picture. Prove that for any positive integer k it is possible to cut
a similar L-shape with k times larger side lengths into unit L-shapes.
(Grade 11.)
Solution. Let the L-shape be placed so that the two longer sides meet at
the top left corner. Starting from the top left we place on it k unit L-shapes
diagonally with the same orientation as the large L-shape (Fig. 12). The rest
consists of two equal staircase-like parts; it is enough to show that one of
them, e.g. the lower part, can be covered. The staircase has k stairs, the lowest
one at the height k − 1 and the highest at the height 2k − 2.
In case k = 1 the staircase is empty, in case k = 2 it can be covered with
one unit L-shape. Assume that the claim holds for the staircase with k stairs
and consider the staircase with k + 2 stairs. Separate a strip of width 2 from
the left and bottom. The rest can be covered by the induction assumption.
The topmost part of the strip is covered with one unit L-shape. Now we have
Fig. 12 Fig. 13 Fig. 14 Fig. 15
16
to cover the rest of the strip whose lower and left sides have correspondingly
the lengths k + 2 and 2k.
• If k is divisible by 3, then cut the figure into two strips of sizes 2 × 2k
and k × 2 and cover both of them with 2 × 3 rectangles consisting of two
unit L-shapes (Fig. 13) and we are done.
• If k ≡ 1 (mod 3), then cut the figure into two strips of sizes 2 × (2k − 2)
and (k + 2) × 2 and cover both of them with 2 × 3 rectangles (Fig. 14).
This is possible because 2k − 2 and k + 2 are divisible by 3.
• If k ≡ 2 (mod 3), then cut the figure into two strips of sizes 2 × (2k − 4)
and (k − 2) × 2, and a corner part, which is a L-shape with k = 2. Both
strips can be covered by 2 × 3 rectangles since 2k − 4 and k − 2 are di-
visible by 3; the corner part can be covered by induction basis (Fig. 15).
FR-11. A ball bearing consists of two cylinders with the same axis and n
equal balls between them. The centers of all the balls are on the same plane
perpendicular to the axis of the cylinders and each ball touches both cylinders
and two adjacent balls. Let r be the radius of the balls and let R be the radius
of the outer cylinder. Prove that Rr < n+ π
π . (Grade 11.)
Solution 1. Consider the regular n-gon with vertices at the centers of the
balls (Fig. 16). Its edges are of length 2r and its perimeter is n · 2r. The radius
of the circumcircle of the n-gon is R − r and the length of the circumcircle is
2π ( R − r ). Since a chord of a circle is always shorter than the corresponding
arc of the circle, we have n · 2r < 2π ( R − r ) or nr + πr < πR, which implies
r π
R < n+π .
Solution 2. Consider the isosceles triangle with vertices at the centers of
two adjacent balls and at the closest point to them on the common axis of the
cylinders (Fig. 17). The two equal sides of the triangle are of length R − r, the
base is of length 2r and the vertex angle is 2π n . The altitude drawn onto the
base divides the triangle into two equal right triangles with the hypotenuse
r
R − r, one of the legs r and the opposite angle πn . Hence R− π π
r = sin n < n ,
r π
whence nr < πR − πr, which implies R < n+π .
Fig. 16 Fig. 17
17
FR-12. The sequence ( an ) is defined by a1 = 1 and an = n · ( a1 + . . . + an−1 )
for all n > 1. Find all indices n for which an is divisible by 1 · 2 · . . . · n.
(Grade 12.)
Answer: 1 and all positive even numbers.
Solution. For each n > 2 denote Sn = a1 + . . . + an−1 . Then an = Sn · n and
for all n > 2 we have Sn = Sn−1 + an−1 = Sn−1 + Sn−1 · (n − 1) = Sn−1 · n.
Hence Sn = Sn−1 · n = Sn−2 · (n − 1)n = . . . = S2 · 3 · . . . · n = n! 2 because
S2 = 1 = 12·2 . Consequently an = Sn · n = n! · n2 for all n > 2. Therefore, for
n > 2, an is divisible by n! iff n is even, and n = 1 also satisfies the condition.
FR-13. The lengths of the sides of a quadrilateral are a, b, c, d and its area
is S. Prove that a2 + b2 + c2 + d2 > 4S. For which quadrilaterals does the
equality hold? (Grade 12.)
Answer. The equality holds only for squares.
Solution. Without loss of generality we can assume that a, b, c and d are
the lengths of consecutive sides of the quadrilateral. A diagonal divides the
quadrilateral into two triangles. From one partition we get the inequality
2 + 2 > S, whence ab + cd > 2S, and from the other partition 2 + 2 > S,
ab cd bc da
whence bc + da > 2S. Therefore ab + bc + cd + da > 4S.

On the other hand, by adding the inequalities a2 + b2 > 2ab, b2 + c2 > 2bc,
c2 + d2 > 2cd, and d2 + a2 > 2da, and dividing by 2 we get a2 + b2 + c2 + d2 >
ab + bc + cd + da, which implies the required inequality.
The equality holds iff all the inequalities used are, in fact, equalities. In
the first inequality the equality holds iff all the angles are right angles. In the
second step the equalities hold iff all sides are of equal length.
FR-14. In a coordinate city there are n > 3 tramlines parallel to the x-axis
such that each line begins from x-coordinate 0 and ends at x-coordinate n.
Exactly one tram of length 1 is moving on each line: on the first line with
speed 1, on the second line with speed 2 etc, until on the last line with the
speed n. When a tram reaches the end of its line it instantly starts moving
back without turning around. In the morning all trams start moving at the
same time from the starting position where the x-coordinate of the back end
of the tram is 0. Prove that the trams’ projections onto the x-axis never cover
the whole interval from 0 to n. (Grade 12.)
Solution. The projections of the trams can cover the whole interval only
when one projection covers [0, 1], another [1, 2] etc. until [n − 1, n]. Consider
the moments when the projection of the slowest tram covers one of these
intervals. When the slowest tram moves by 1 unit, then the fastest and the
third fastest trams move correspondingly by n and n − 2 units. Together these
two trams move by 2n − 2 units which is exactly one to and fro cycle.
Denote the integer positions of the trams on the round trip by numbers 0
to 2n − 3, i.e the starting position is 0 and each next one until returning to the
18
starting point is greater by one. Call these numbers the position characteris-
tics. If the sum of the position characteristics of two trams is 2n − 2, then their
projections cover the same interval because one of them has moved the same
amount from the starting point as the other one still has to go to reach it. If
the sum of the position characteristics is 0, then they both are in the starting
positions, so they again cover the same interval. Hence, when the sum of the
position characteristics is divisible by 2n − 2, the projections cover the same
interval.
At the beginning the sum of the position characteristics of the fastest and
the third fastest tram is 0 and each time they together move by 2n − 2 units the
sum of their position characteristics stays divisible by 2n − 2. Consequently,
when the projection of the slowest tram covers an interval with integer end-
points, the projections of these two trams cover the same interval, hence at
least one of the intervals is not covered.
FR-15. Find the minimal distance between two points, one of which is on
the graph of function y = e x and the other on the graph of function y = ln x.
(Grade 12.) √
Answer: 2.
Solution. The graphs of functions y = e x and y = ln x are symmetrical
w.r.t. line y = x (Fig. 18). Hence the distance between points on these graphs
is minimal iff both points are at minimal distance from line y = x. The dis-
tance between the graph of the function y = e x and the graph of the function
y = x is minimal at the point where the tangent is parallel to y = x. Then,
y′ = 1 for the function y = e x , whence e x = 1, giving x = 0 and y = √1. The
minimal distance is therefore between points (0, 1) and (1, 0) and it is 2.
y = ex
y = ln x
1
1
y=x
Fig. 18
19
First day
a−b
TS-1. For arbitrary positive integers a, b, denote a ⊖ b = .
gcd( a, b)
Let n be a positive integer. Prove that the following conditions are equiv-
alent:
(i) gcd(n, n ⊖ m) = 1 for every positive integer m < n;
(ii) n = pk where p is a prime number and k is a non-negative integer.
Solution. Note at first that da ⊖ db = a ⊖ b for all positive integers a, b, and
d. Indeed,
da − db d · ( a − b) a−b
da ⊖ db = = = = a ⊖ b.
gcd(da, db) d · gcd( a, b) gcd( a, b)
Show now that if n is a prime power and m < n, then n ⊖ m is relatively
prime to n. Indeed, let n = pk where p is a prime number, and let m = pi s
where gcd( p, s) = 1. Then m < n implies i < k. Now
p k−i − s
n ⊖ m = p k−i ⊖ s = = p k−i − s
gcd( pk−i, s)
because gcd( pk−i , s) = 1 by the choice of s. Also, for the same reason,
gcd( p, pk−i − s) = 1, hence gcd(n, n ⊖ m) = gcd( pk , pk−i − s) = 1.
It remains to show that if n is not a prime power, then there exists a posi-
tive integer m such that m < n and the integers n ⊖ m and n share a common
prime factor. Since n is not a prime power, it has at least two different prime
factors. Let p and q be some prime factors of n, whereby p < q. Let n = pk t
where gcd( p, t) = 1. Take m = n − pk+1. As n is divisible by both pk and
q which are relatively prime, it is also divisible by their product pk q. Conse-
quently, pk+1 < pk q 6 n, i.e., 0 < m < n. Now
t − (t − p) p
n ⊖ m = n ⊖ ( n − p k +1 ) = t ⊖ ( t − p ) = = =p
gcd(t, t − p) gcd(t, p)
since gcd(t, p) = 1. We see that n ⊖ m and n have a common prime factor p.
TS-2. Let n be a positive integer. Find the largest integer N for which there
exists a set of n weights such that it is possible to determine the mass of all
bodies with masses of 1, 2, . . . , N using a balance scale (i.e. to determine
whether a body with unknown mass has a mass 1, 2, . . . , N, and which
namely).
n
Answer: N = 3 2−1 .
Solution. The possibility to determine mass m means the possibility to
place the weights on the two scalepans so that the difference of total masses
on the two scalepans is exactly m.
20
Every weight can be placed on either of the two pans or on neither of
the pans. For n weights this makes 3n different placements. Note that the
placement where none of the weights is on the scales does not determine
any mass. Also, for each placement there is a symmetric placement with
all the weights on the two pans swapped, which determines the same mass.
n
Therefore with n weights it is possible to determine at most 3 2−1 different
masses.
We show by induction that it is possible to determine all masses from 1
3 n −1
to 2 using n weights with masses 1, 3, . . . , 3n−1 . For n = 1 it is obvious.
Assume that the claim holds for n = k. By the induction assumption we
k
can determine all masses from 1 to 3 2−1 by weights 1, 3, . . . , 3k−1 . Using the
weight with mass 3k , we can determine the mass 3k , and using it together
k
with the other weights also the masses 3k + 1, . . . , 3k + 3 2−1 and 3k − 1, . . . ,
k k 3 k −1 3 k −1 3 k +1 − 1
3k − 3 2−1 . Since 3k − 3 2−1 = 2 + 1 and 3k + 2 = 2 , the claim is
also true for n = k + 1.
TS-3. Let the angles of a triangle be α, β, and γ, the perimeter 2p and the
radius of the circumcircle R. Prove the inequality
2
9R
cot α + cot β + cot γ > 3
2 2 2
−1 .
p2
When is the equality achieved?
Answer: the equality holds for equilateral triangles.
Solution. Let the opposite sides of the angles α, β, and γ be correspond-
ingly a, b, and c. Since cot2 α = 1/ sin2 α − 1 and from the law of sines
1/ sin α = 2R/a, we have cot2 α = 4R2 /a2 − 1; similarly cot2 β = 4R2 /b2 − 1
and cot2 γ = 4R2 /c2 − 1. The inequality can therefore be written as
4 · 9R2

1 1 1
2
4R · + 2 + 2 −3 > 3· −1 ,
a2 b c ( a + b + c )2
or
1 1 1 27
+ 2+ 2 > .
a2 b c ( a + b + c )2
Dividing both sides by 3 and taking the square root gives
s
1 1 1 1 3
· 2
+ 2
+ 2
> .
3 a b c a+b+c
The left side is the quadratic mean of 1a , 1b , 1c and the right side is the harmonic
mean of the same numbers, hence the inequality holds.
The equality holds iff a = b = c.
21
Second day
TS-4. In an acute triangle ABC the angle C is greater than the angle A. Let
AE be a diameter of the circumcircle of the triangle. Let the intersection point
of the ray AC and the tangent of the circumcircle through the vertex B be K.
The perpendicular to AE through K intersects the circumcircle of the triangle
BCK for the second time at point D. Prove that CE bisects the angle BCD.
Solution. Since AE is a diam- D
eter of the circumcircle of the tri-
angle ABC, ∠ ACE = ∠ECK = B
90◦ . So it suffices to show that
∠ ACB = ∠DCK (Fig. 19). Let
L be the point of intersection of E
lines AE and DK. Then ∠ BAC = A L
∠CBK = ∠CDK by the inscribed
angles theorem. Also ∠ ABC = C
∠ AEC = ∠CKD where the latter K
equality follows from the simi- Fig. 19
larity of the right triangles ACE
and ALK. Hence the two triangles ABC and DKC are similar, and therefore
∠ ACB = ∠DCK.
Remark. This problem was proposed to the Baltic Way competition in 2008
(not by Estonia) but was not selected.
TS-5. Let P( x, y) be a non-constant homogeneous polynomial with real co-

efficients such that P(sin t, cos t) = 1 for every real number t. Prove that there
exists a positive integer k such that P( x, y) ≡ ( x2 + y2 )k .
Solution 1. Let n be the degree of the polynomial P, i.e.,
P( x, y) = an x n + an−1 x n−1 y + . . . + a1 xyn−1 + a0 yn ,
where n > 0. Note that n must be even because otherwise the condition
P(sin t, cos t) = 1 for t = 0 would imply a0 = 1 while the same condition for
t = π would imply a0 = −1.
Since P has no constant
p term, P(0, 0) = 0. Now assume that x 6= 0 or
y 6= 0 and take c = x2 + y2 . Since
2 2
x y
p + p = 1,
x 2 + y2 x 2 + y2
p
there
p exists some real number t such that sin t = x/ x2 + y2 and cos t =
2 2
y/ x + y and therefore P(sin t, cos t) = 1. By homogenicity, P( x, y) =
y
cn · P xc , c , hence
q n q n
x y
P( x, y) = x 2 + y2 · P p ,p = x 2 + y2 .
x 2 + y2 x 2 + y2
22
for all x, y. The case n = 2k implies P( x, y) = ( x2 + y2 )k which satisfies also
the condition P(0, 0) = 0.
Solution 2. Like in Solution 1, express the polynomial as a sum of n + 1
monomials with coefficients a0 , . . . , an and show that n = 2k.
We prove the claim of the problem by induction on k. In case k = 0
(omitting the extra assumption that P is non-constant) the claim holds ob-
viously. Assume now that k > 0 and the claim holds for k − 1. Substituting
t = 0 and t = π2 into P(sin t, cos t) = 1 gives a0 = 1 and an = 1, respec-
tively. Hence the polynomial P( x, y) − ( x2 + y2 )k does not have terms with
x n and yn . Let Q( x, y) be such that P( x, y) − ( x2 + y2 )k ≡ xy · Q( x, y). Then
sin t cos t · Q(sin t, cos t) = 0 for every real number t, hence Q(sin t, cos t) = 0
for every t such that sin 2t 6= 0. By continuity of Q(sin t, cos t) as a function
of t, it follows that Q(sin t, cos t) ≡ 0. Now define R( x, y) = Q( x, y) + ( x2 +
y2 )k−1 . As both Q and R are homogeneous polynomials of degree 2(k − 1),
the assumptions of the problem hold for polynomial R. By the induction hy-
pothesis, R( x, y) ≡ ( x2 + y2 )k−1 . Hence Q( x, y) ≡ 0 and P( x, y) ≡ ( x2 + y2 )k .
TS-6. Every unit square of a n × n board is colored either red or blue so that
among all 2 × 2 squares on this board all possible colorings of 2 × 2 squares
with these two colors are represented (colorings obtained from each other by
rotation and reflection are considered different).
a) Find the least possible value of n.
b) For the least possible value of n find the least possible number of red
unit squares.
Answer: a) 5; b) 10.
Solution. a) Since there are 24 = 16 = 42 possibilities
to color a 2 × 2 square in two colors and a n × n square
contains (n − 1)2 such subsquares, we must have n − 1 >
4, or n > 5. For n = 5 a suitable coloring is given in Fig. 20.
b) Fig. 20 presents a coloring with 10 red squares. We
will show that this is the least possible.
Note that in the 5 × 5 square there are 4 unit squares
in the corners, 12 squares on the sides (not in the corners), Fig. 20
and 9 inner squares. Each corner square is contained in
exactly one, side square in two and inner square in four 2 × 2 squares. All 16
colourings of 2 × 2 squares contain a total of 64 unit squares of which 32 are
red by symmetry. Therefore, if the 5 × 5 square contains k red squares, among
them a corner squares, b side squares and c inner squares, then a + b + c = k
and a + 2b + 4c = 32. The equation a + 2b + 4c = 32 implies c 6 8. If c = 8,
then a = b = 0. If c = 7, then the only possibility to have k < 10 is b = 2 and
a = 0. If c 6 6, then always k = a + b + c > 10.
23
Thus it is enough to show that there are no colorings with required prop-
erties with a = 0 and b 6 2. Indeed, in this case the 5 × 5 square has at least
two sides not containing any red squares. Without loss of generality, let one
of them be the upper side. We saw in part a) that for n = 5 each coloring of
2 × 2 squares must occur exactly once. Since among all 16 colorings of 2 × 2
squares there are 4 such where both upper unit squares are blue, and two
upper rows of the 5 × 5 square contain exactly 4 2 × 2 squares, all four such
colorings must be located in the two upper rows, among these the completely
blue coloring. Since the same is true for the other side which does not contain
any red squares, the two sides must meet and a completely blue 2 × 2 square
must be in the corner where the two sides meet. With-
out loss of generality, let it be the left side. Then the two
squares on Fig. 21 must be red, because otherwise there
would be more than one completely blue 2 × 2 square. But
now there are two 2 × 2 squares with red square in the
lower right corner and the rest of them blue. Therefore
there is no coloring satisfying the conditions with a = 0
Fig. 21 and b 6 2 and the least number of red squares is k = 10.

Number theory: OC-1, OC-6, OC-7, OC-12, FR-1, FR-7, FR-12, TS-1
Algebra: OC-4, OC-5, OC-13, FR-2, FR-8, FR-15, TS-5
Geometry: OC-2, OC-9, OC-11, FR-4, FR-6, FR-9, FR-11, FR-13, TS-3, TS-4
Discrete mathematics: OC-3, OC-8, OC-10, FR-3, FR-5, FR-10, FR-14, TS-2, TS-6
24
2010/2011

Tartu 2011
WE THANK:
University of Tartu
Problem authors: Maksim Ivanov, Urve Kangro, Oleg Koshik,

Toomas Krips, Tuan Le (USA), Härmel Nestra, Uve Nummert,
Laur Tooming, Jan Willemson
Translators: Härmel Nestra, Kadi Liis Saar
Editor: Reimo Palm

The Estonian Mathematical Olympiad is held annually in three rounds: at
In each round of the Olympiad, separate problem sets are given to the
students of each grade. Students of grade 9 to 12 compete in all rounds,
students of grade 7 to 8 participate at school and regional levels only. Some
towns, regions and schools also organize olympiads for even younger stu-
dents. The school round usually takes place in December, the regional
round in January or February and the final round in March or April in
Tartu. The problems for every grade are usually in compliance with the
school curriculum of that grade but, in the final round, also problems re-
quiring additional knowledge may be given.
and the Seniors. In the first category, students up to the 10th grade can
participate; the other category has no restriction. Being successful in the
open contests generally assumes knowledge outside the school curriculum.
Based on the results of all competitions during the year, about 20 IMO
April or May, lasting two days; each day, the contestants have 4.5 hours to
solve 3 problems, similarly to the IMO. All participants are given the same
problems. Some problems in our selection contest are at the level of diffi-
culty of the IMO but somewhat easier problems are usually also included.
The problems of previous olympiads are available at the Estonian Math-
Besides the above-mentioned contests and the quiz “Kangaroo” other
regional and international competitions and matches between schools are
held as well.
*
This booklet presents the problems of the open contests, the final round
of national olympiad and the team selection contest. For the open contests
and the final round, selection has been made to include only problems that
have not been taken from other competitions or problem sources and seem
interesting enough. The team selection contest is presented entirely.
1
OC-1. (Juniors.) Find all pairs ( a, b) of real numbers with a + b = 1, which
satisfy ( a2 + b2 )( a3 + b3 ) = a4 + b4 .
Answer: (0, 1), (1, 0), and ( 12 , 21 ).
Solution 1. As a3 + b3 = ( a + b)( a2 − ab + b2 ), the given equation can be
expressed as ( a2 + b2 )( a2 − ab + b2 ) = a4 + b4 . Expanding brackets gives
− a3 b + 2a2 b2 − ab3 = 0, which factorizes to − ab( a − b)2 = 0. Hence a = 0
or b = 0 or a − b = 0. Together with the condition a + b = 1, we get the
following solutions: (0, 1), (1, 0), and ( 21 , 12 ).
Solution 2. Denote ab = c. Then a2 + b2 = ( a + b)2 − 2ab = 1 − 2c,
a + b3 = ( a + b)( a2 − ab + b2 ) = 1 · (1 − 2c − c) = 1 − 3c, and a4 + b4 =
3
( a2 + b2 )2 − 2a2 b2 = (1 − 2c)2 − 2c2 = 2c2 − 4c + 1. The given equation

( a2 + b2 )( a3 + b3 ) = a4 + b4 can now be expressed as (1 − 2c)(1 − 3c) =
2c2 − 4c + 1, or equivalently, c(4c − 1) = 0. Hence, c = 0 or c = 14 . Now the
two simultaneous equations a + b = 1 and ab = c give the solutions a = 0,
b = 1 and a = 1, b = 0 for c = 0, and a = 12 , b = 12 for c = 14 .
OC-2. (Juniors.) Consider a parallelogram ABCD.

a) Prove that if the incenter of the triangle ABC is located on the diagonal
BD, then the parallelogram ABCD is a rhombus.
b) Is the parallelogram ABCD a rhombus whenever the circumcenter of
the triangle ABC is located on the diagonal BD?
Answer: b) no.
D C
Solution. a) As the incenter of the triangle
ABC is located on diagonal BD (Fig. 1), we
can conclude that BD is the bisector of ∠ ABC.
Therefore ∠ ABD = ∠CBD. However, since I
ABCD is a parallelogram, ∠ ABD = ∠CDB.
Hence the triangle BCD is isosceles, i.e. | BC | =
|CD |. Thus, ABCD is a rhombus. A B
b) Let ABCD be a rectangle with different
side lengths. The circumcenter of triangle ABC Fig. 1
is located on the intersection of the diagonals of the rectangle. We see that
all the required conditions are satisfied, however ABCD is not a rhombus.
OC-3. (Juniors.) The numbers 0, 1, and 2 are written in the vertices of a

triangle. One step involves increasing two of the three numbers by m or
decreasing one of the three numbers by n. Is it possible that after some steps
there are numbers 1, 2, and 3 (in an arbitrary order) written in the vertices if
a) m = 3, n = 6;
b) m = 4 12 , n = 6?
2
Solution 1. a) Both the step that involves increasing two of the numbers
by 3 and the step that involves decreasing one of the numbers by 6 result
in the sum of all three numbers being changed by 6. Thus the remainder
when the sum of the three numbers is divided by 6 will always be the same
regardless of the number of steps taken. But as the sums 0 + 1 + 2 and
1 + 2 + 3 give different remainders when divided by 6, it is impossible to
reach the required end situation from the given initial situation.
b) First increase the second and the third numbers three times by 4 12 ;
we end up with 0, 14 21 , 15 12 in the vertices. Now increase the first and the
second numbers by 4 12 and also increase the first and the third numbers by
4 21 ; so we end up with 9, 19 and 20 written in the three vertices, respectively.
Finally decrease the first number once by 6 and the other two three times
by 6, achieving the situation in question.
Solution 2. a) Consider one of the numbers. The remainder when this
number is divided by 3 is the same regardless of the number of steps taken.
Therefore, if we want to achieve the situation where 1, 2, 3 are located in the
three vertices, the numbers 1 and 2 should stay in the same vertices where
they were at the beginning and 3 has to be in the vertex where 0 was. Notice
that two increasings are exactly cancelled out by one decreasing. Thus, the
vertices where the numbers remain the same should have undergone an
even number of increasings and the vertex where 0 is replaced by 3 should
have been exposed to an odd number of increasings. Hence there should
have been an odd number of increasings in total which is impossible since
each increasing step influences the numbers in two vertices.
b) As in Solution 1.
OC-4. (Juniors.) Find all pairs (n, k) of positive integers that satisfy the
equality n! + (n + 1)! = k! + 120.
Answer: (4, 4), (5, 6).
Solution 1. Note that for every n, n! + (n + 1)! = n! + n! · (n + 1) =
n! · (n + 2) 6 (n + 2)!. Thus if n! + (n + 1)! = k! + 120, then due to 120 = 5!
we have k! + 5! 6 (n + 2)!. This inequality in turn implies k < n + 2 and 5 <
n + 2. Hence 5 6 n + 1, leading to 0 6 (n + 1)! − 5! = k! − n!. Consequently,
k > n, i.e., the cases to be considered are k = n and k = n + 1. If k = n,
then the initial equation leads to (n + 1)! = 120, giving n = 4, k = 4. If
k = n + 1, then analogously n = 5, k = 6.
Solution 2. If k < n, then n! > k!. If, additionally, n > 4, then (n + 1)! >
120, giving n! + (n + 1)! > k! + 120. But if n < 4, then n! + (n + 1)! 6 30 <
k! + 120. The desired equality can hold in neither of the cases.
If k > n + 1, then k! − ( n + 1)! is positive and is also divisible by (n + 1)!,
hence k! − (n + 1)! > (n + 1)!. On the other hand, n! − 120 < (n + 1)!,
giving k! − (n + 1)! > n! − 120. Thus, there is no solution in this case either.
Hence n = k or n + 1 = k, leading to two solutions (n, k) = (4, 4), (5, 6).
3
Solution 3. The equation implies that 120 is divisible by the minimum of
n! and k!. As 120 = 5!, either n 6 5 or k 6 5. Consider both cases.
If n ∈ {1, 2, 3}, then n! + (n + 1)! − k! < 120, whence the equation has
no solution. If n = 4, then 120 = 144 − k!, whence k! = 24 and k = 4. If
n = 5, then 120 = 840 − k!, whence k! = 720 and k = 6.
If k ∈ {1, 2, 3, 4, 5}, then the initial equation implies that n! + (n + 1)!
lies between 121 and 240. This is possible only if n = 4, since if n = 3, then
n! + (n + 1)! = 30, and if n = 5, then n! + (n + 1)! = 840. If n = 4, then
n! + (n + 1)! = 144, which corresponds to k = 4.
OC-5. (Juniors.) Consider the diagonals A1 A3 , A2 A4 , A3 A5 , A4 A6 , A5 A1

and A6 A2 of a convex hexagon A1 A2 A3 A4 A5 A6 . The hexagon whose ver-
tices are the points of intersection of the diagonals is regular. Can we con-
clude that the hexagon A1 A2 A3 A4 A5 A6 is also regular?
Answer: yes.
Solution 1. We show that the hexagon A1 A2 A3 A4 A5 A6
has all its side lengths equal and all its angles equal. As the
internal hexagon is regular, the grey triangles in Fig. 2 all
have two angles of equal size and so they are isosceles. Ad-
ditionally, all these six isosceles triangles have their bases
of equal lengths, thus they are all congruent. The black
triangles on Fig. 2 are isosceles because the grey triangles
are isosceles. Additionally, their vertex angles are equal, as Fig. 2
they all are equal to the angles of a regular hexagon. Therefore the black
triangles are all congruent and thus their bases are of equal length. Now
the angles of the external hexagon are all formed of the angles of two grey
triangles and one black triangle. As both of the latter are congruent, the
external hexagon has its angles of equal size.
Solution 2. Lengthen the sides of the internal regular hexagon until in-
tersection. The points of intersection are exactly the vertices of the initial
external hexagon. Because of the symmetry of the internal hexagon the
points of intersection are symmetrically located about the midpoint of the
internal hexagon. Thus, the external (initial) hexagon is also regular.
OC-6. (Juniors.) A cashier has a stack of n notes lying on top of each other.
He has to turn all notes front side up, however the order of the notes is not
important. Every step consists of taking a block of consecutive notes and
turning them around in the stack. Find the smallest number of steps that
will suffice him to turn all notes in the stack front side up, irrespective of
the initial position of notes.
Answer: n2 for even n, and n+ 1
2 for odd n.
Solution 1. Let the cashier always choose the part of the stack which
starts from the top most note facing the wrong way and ends with the bot-
tom most note facing the wrong way. Then after k steps there are at least
4
k notes facing the right way on top of the stack and so are k notes at the
bottom of the stack. Therefore for even n, after at least n2 steps all the notes
are facing upwards. For odd n it is guaranteed that after n− 1
2 steps there are
n − 1 notes facing the right way, and one more step may be needed to turn
the middle note the right way. So, overall at least n+ 1
2 steps are sufficient.
To show the necessity, assume that the stack itself is located in between
a bigger stack where all the notes are already facing the right way, and call
the number of pairs of neighboring notes where one of the notes is the right
way around and the other is the wrong way around degree of disarrange-
ment. If the degree of disarrangement of the stack is 0, we have achieved
the required situation (we introduce the concept of the bigger stack to en-
sure that if all the notes in our stack are the wrong way around the degree
of disarrangement of the stack is non-zero). At every step, the degree of
disarrangement of the stack cannot decrease by more than 2, since the rel-
ative order of consecutive notes can only change at the boundaries of the
block to be turned around. Assume the initial stack consists of notes facing
alternately the right way and the wrong way, whereby for odd n assume the
top and the bottom notes are the wrong way around. Then for odd n the
degree of disarrangement to start with is n + 1 and so we need at least n+ 2
1
steps. For even n the initial degree of disarrangement is n because one of the
endmost notes is the right way around and thus at least n2 steps are needed.
Remark. In the first part of the solution we cannot claim that at every
step the number of notes facing the wrong way decreases by 2. In addition
to the two endmost notes facing the wrong way, all the notes between them
are also turned around and as a result the number of notes facing the wrong
way may even increase. For that reason it is necessary to monitor the pro-
gression of success from the stack’s ends onwards.
Solution 2. Let us show that every possible stack can be put in order as
required by no more than n2 (for even n) or n+ 1
2 (for odd n) steps as follows.
For n = 2 the statement holds. Indeed, if both notes are facing the wrong
way, we simply turn them around together, and if only one of the notes is
facing the wrong way we only turn this one around. Now partition the
stack into smaller stacks, each containing two consecutive notes and solve
the problem for each stack separately. For even n no more than n2 steps are
required. For odd n the last note has no counterpart, however it still may
require one step, so overall no more than n+ 1
2 steps are needed.
The necessity is shown as in Solution 1.
Solution 3. Let us present one more strategy to achieve the required situ-
ation by n2 (for even n) or n+ 1
2 (for odd n) steps for any stack of notes. If no
more than half of the notes are facing the wrong way we simply turn each
of them around individually. This requires as many steps as is the number
of such notes, i.e. for even n no more than n2 steps and for odd n no more
than n− 1
2 steps. If more than half of the notes are facing the wrong way, first
5
turn the whole stack around and then individually turn around all those
notes facing the wrong way. For even n no more than 1 + ( n2 − 1) = n2 steps
are required. For odd n no more than 1 + n− 1 n +1
2 = 2 steps are required.
The necessity is again shown as in Solution 1.
OC-7. (Juniors.) Consider a positive integer N with exactly 6 positive divi-

sors d1 , . . . , d6 such that 1 = d1 < d2 < d3 < d4 < d5 < d6 = N. Call such
an integer N good if the sum d4 + d5 is divisible by the sum d2 + d3 .
a) Find the smallest positive integer N which has exactly 6 positive di-
visors and which is not good.
b) Prove that there are infinitely many positive integers N all with ex-
actly 6 positive divisors and all not good.
Answer: a) 20.
Solution 1. a) Considering the numbers from 1 to 20 we see that exactly
three of them have 6 divisors: 12 (the divisors are 1, 2, 3, 4, 6, 12), 18 (1, 2, 3,
6, 9, 18), and 20 (1, 2, 4, 5, 10, 20). For 12 the sum d4 + d5 = 4 + 6 is divisible
by the sum d2 + d3 = 2 + 3 and similarly, for 18 the sum d4 + d5 = 6 + 9
is divisible by d2 + d3 = 2 + 3. However, for 20 the sum d4 + d5 = 5 + 10
is not divisible by the sum d2 + d3 = 2 + 4. Thus, the smallest non-good
number with exactly 6 factors is 20.
b) Take N = 4p, where p is an arbitrary prime number larger than 4.
Then N has exactly 6 different divisors: 1, 2, 4, p, 2p, 4p, in increasing order.
Indeed, as N = 22 p1 , where 2 and p are two different prime numbers, all of
its divisors can be expressed as 2i p j where i 6 2 and j 6 1. From here we
obtain exactly 3 · 2 = 6 choices: i can be either 0, 1 or 2 and for every i we
have two choices for j: 0 or 1.
Here, d4 + d5 = p + 2p = 3p is odd because p > 2 and thus is not
divisible by an even number d2 + d3 = 2 + 4 = 6. Therefore, none of the
numbers expressed as N = 4p where p > 4 is a prime is good. As there
are infinitely many prime numbers, there must also be infinitely many such
numbers N.
Solution 2. Let us find all N > 1 which have exactly 6 divisors.
1) If N = pk , where p is a prime, then it has the divisors 1, p, . . . , pk , i.e.
k + 1 divisors overall. Thus, all N = p5 satisfy this requirement.
2) Let N have two different prime divisors, i.e. N = pk ql . For k > 2 and
l > 2 we see that N has at least 9 different divisors: 1, p, p2, q, q2 , pq, p2 q,
pq2 , and p2 q2 . For l = 1, N has the divisors 1, p, . . . , pk , and q, pq, . . . , pk q,
i.e. 2(k + 1) divisors in total. Thus, all N = p2 q satisfy the requirement.
3) Let N have at least three prime divisors p, q, r. Then N has at least 8
different divisors: 1, p, q, r, pq, pr, qr, and pqr, and we get no more numbers.
Let us now consider N = p5 and N = p2 q in more detail.
i) If N = p5, then di = pi−1 and d4 + d5 = p3 + p4 = p3 (1 + p) is
divisible by d2 + d3 = p + p2 = p(1 + p). Thus they are all good.
6
ii) If N = p2 q, where q < p, then N has the divisors 1, q, p, pq, p2 , and
p2 q, in increasing order, and d4 + d5 = pq + p2 = p(q + p) is divisible by
d2 + d3 = q + p and thus, they are all good, too.
iii) If N = p2 q, where p < q < p2 , then N has the divisors 1, p, q, p2 , pq,
and p2 q, in increasing order, and d4 + d5 = p2 + pq = p( p + q) is divisible
by d2 + d3 = p + q. Thus, they are all good, too.
iv) Finally, let N = p2 q, where q > p2 . Then N has the divisors 1, p, p2 ,
q, pq, and p2 q, in increasing order and d4 + d5 = q + pq = q(1 + p) is not
divisible by d2 + d3 = p + p2 = p(1 + p) because the prime number q can-
not be divisible by another prime number p. Thus all these numbers have
exactly 6 different divisors and they all are non-good. To get the smallest
of these numbers, we have to take p and q as small as possible, i.e. p = 2
and q = 5 (to achieve q > p2 = 4). Then N = 22 · 5 = 20. Finally, there
are infinitely many of these numbers N because we have infinitely many
choices for prime numbers p and q such that q > p2 . For example, we can
take p = 2 and q an arbitrary prime number bigger than 5. As there are
infinitely many prime numbers, we have proven the statement.
OC-8. (Seniors.) Prove that none of the integers that contain one 2, one 1,
and all the rest zeros, can be expressed as a sum of two perfect squares or
as a sum of two perfect cubes.
Solution. All the numbers described in the problem are divisible by 3
(as their sum of digits is divisible by 3). Note that all perfect squares leave
the remainder 0 or 1 when divided by 3, and therefore, for the sum of the
two perfect squares to be divisible by 3, they both have to be divisible by
3. Now, as the numbers are both divisible by 3, their squares are divisible
by 9 and thus the sum of the squares is divisible 9. However, the number
described in the problem is not divisible by 9, a contradiction.
Note that when a cube number is divided by 3, it will leave the re-
mainder of either 0, 1 or −1. Indeed, (3k)3 = 9(3k3 ) and (3k ± 1)3 =
27k3 ± 27k2 + 9k ± 1 = 9(3k3 ± 3k2 + k) ± 1. The numbers described in
the problem give the remainder 3 when divided by 9 (as their sum of digits
gives the remainder 3 when divided by 9), therefore we conclude that it is
impossible to express them as sums of two cubes.
OC-9. (Seniors.) Consider an acute-angled triangle ABC and its circumcir-

cle. Let D be a point on the arc AB which does not include point C and
let A1 and B1 be points on the lines DA and DB, respectively, such that
CA1 ⊥ DA and CB1 ⊥ DB. Prove that | AB| > | A1 B1 |.
Solution 1. If CD is the diameter of the circumcircle of triangle ABC, then
A1 = A and B1 = B and the statement holds. Assume that CD is not the di-
ameter (Fig. 3). Then A1 6= A and B1 6= B. The point A1 lies on the ray AD
if and only if the point B1 does not lie on the ray BD (depending on which
side of the diameter through point C point D is located). Thus, ∠CAA1 =
7
∠CBB1 (because the sum of opposite C
angles of a cyclic quadrilateral ADBC
is 180◦ ). Thus, the right-angled tri-
angles AA1 C and BB1 C are similar.
From ∠ ACA1 = ∠ BCB1 we see that
∠ ACB = ∠ A1 CB1 . This together with
| AC|
| BC|
= ||AB 1CC|| gives that ACB and B1
1 A B
A1 CB1 are similar. As | AC | > | A1 C |,
we conclude that | AB| > | A1 B1 |. A1
Solution 2. Since the angles CA1 D
and CB1 D are right angles, the points D
C, A1 , D, and B1 form a cyclic quadri- Fig. 3
lateral and thus ∠CAB = ∠CDB =
∠CA1 B1 . Similarly, ∠CBA = ∠CB1 A1 . Therefore the triangles ABC and
A1 B1 C are similar. As |CA| > |CA1 |, we deduce that | AB| > | A1 B1 |.
Solution 3. The radius R of the circumcircle of the quadrilateral CADB
is at least as large as the radius R1 of the circumcircle of the quadrilateral
CA1 DB1 because CD is a chord in the first one and a diameter in the second
one. The sine law in triangles ADB and A1 DB1 gives | AB| = 2R sin ∠ D and
| A1 B1 | = 2R1 sin ∠D. As R > R1 , we deduce | AB| > | A1 B1 |.
Remark. The statement holds for all triangles ABC and all points D on
the circumcircle, given D is not one of the vertices of the triangle.
OC-10. (Seniors.) Find all pairs (m, n) of positive integers for which the
m × n grid contains exactly 225 rectangles whose side lengths are odd and
whose edges lie on the lines of the grid.
Answer: (1, 29), (5, 9), (9, 5), and (29, 1).
Solution. The m × n grid is formed by m + 1 horizontal and n + 1 ver-
tical lines. Number the horizontal lines with numbers from 1 to m + 1 and
the vertical lines with numbers from 1 to n + 1. Rectangles with odd side
lengths arise if and only if two horizontal lines with different parity and
two vertical lines with different parity intersect.
Assume that at least one of the numbers m and n is even. We can assume
without loss of generality that m = 2k. Then there are exactly k + 1 odd-
numbered and k even-numbered horizontal lines and thus there are k(k + 1)
pairs of lines of different parity. But this means that overall the number of
rectangles with odd side lengths is even and cannot be 225. Therefore m
and n are both odd numbers. Let now m = 2k − 1 and n = 2l − 1. Then we
have exactly k even-numbered and k odd-numbered horizontal lines and l
even-numbered and l odd-numbered vertical lines. Overall it is possible to
form k · k · l · l = (kl )2 rectangles with odd side lengths. From (kl )2 = 225
we get kl = 15. The solutions are k = 1, l = 15 or k = 3, l = 5 (or vice
versa). So m = 1, n = 29 or m = 5, n = 9 (or vice versa).
8
OC-11. (Seniors.) Given a triangle ABC where | BC | = a, |CA| = b and
| AB| = c, prove that the equality
1 1 3
+ =
a+b b+c a+b+c
holds if and only if ∠ ABC = 60◦ .
Solution. By finding the common denominator on the left hand side,
transform the equation to ( a + 2b + c)( a + b + c) = 3( a + b)(b + c). Ex-
panding the brackets and simplifying gives b2 = a2 + c2 − ac. Comparing
the latter with the cosine law b2 = a2 + c2 − 2ac cos β, we see that the equal-
ity holds if and only if cos β = 12 , i.e., β = 60◦ .
OC-12. (Seniors.) A square ABCD lies in the coordinate plane with its ver-
tices A and C lying on different coordinate axes. Prove that one of the ver-
tices B or D lies on the line y = x and the other one on y = − x.
Solution 1. Assume without loss of generality that A is located on the
x-axis and C is located on the y-axis, let these points have coordinates of
A ( a, 0) and C (0, c). As the diagonals of a square bisect each other, we know
that the intersection point P of diagonal is also the mid-point of AC, i.e.
−
→
P ( 2a , 2c ) and PC = (− 2a , 2c ).
As the diagonals of a square are perpendicular to each other and of the
−
→ −→
same length, the vectors PB and PD have the same length as the vector
−→
PC and are perpendicular to it. But for a given vector ~u = (s, t), there
are exactly two vectors perpendicular to and having the same length as it:
−→
~v = (−t, s) and −~v = (t, −s). For the vector ~u = PC = (− 2a , 2c ) we get
−
→ −→
~v = ( 2c , 2a ) and w.l.o.g. we can assume that PB = ~v and PD = −~v. Now
from here B( a+ c a+c a−c c− a
2 , 2 ) and D ( 2 , 2 ). Thus, we see that the point B is
located on the line y = x and point D is located on the line y = − x.
Solution 2. W.l.o.g., assume that the vertices of the square are labelled
counter-clockwise with A ( a, 0), C (0, c), where a, c > 0 (other cases are
similar). Let O be the origin, then ∠ AOC = 90◦ , i.e. the circumcircle (with
y
C y B
B C
D
O A x
O A x
D
Fig. 4 Fig. 5
9
diameter AC) of the square ABCD passes through the origin O. Based on
the assumptions made, B definitely lies in the first quadrant and D has to lie
in the second quadrant (Fig. 4) or in the fourth quadrant (Fig. 5), otherwise
the circle with the diameter BD cannot pass the origin. Now note that the
vertices of the square divide its circumcircle into four equal arcs of 90◦ , each
having an inscribed angle of 45◦ subtending on it. Thus, ∠ AOB = ∠ BOC =
45◦ , i.e., B lies on the line with equation y = x (if A = O or C = O, then one
of those angles will lose its meaning, however, the other one is still 45◦ and
that is sufficient). Similarly, ∠COD = 45◦ , if D lies in the second quadrant,
or ∠ AOD = 45◦ , if D lies in the fourth quadrant. In both cases, D lies on
the line y = − x; this condition is also met in the special case D = O.
OC-13. (Seniors.) Let a, b, c be fixed real numbers, where 0 6 a, b, c 6 4.

Prove that the system of equations
 2
 p − aq = −3
q2 − br = −4
 2
r − cp = −5
has no real solutions ( p, q, r ).
Solution. Adding up all equations gives p2 − cp + q2 − aq + r2 − br =
2
−12. From the inequality ( p − 2c )2 > 0 we have p2 − cp > − c4 > −4 and
similarly, q2 − aq > −4 and r2 − br > −4. Adding up these inequalities,
we see that to avoid a contradiction with the equality derived first, all three
inequalities must actually be equalities, i.e. a = b = c = 4 and p = q = r =
2. But this does not satisfy the initial equations.
OC-14. (Seniors.) Let ABC be a triangle with integral side lengths. The
angle bisector drawn from B and the altitude drawn from C meet at point P
inside the triangle. Prove that the ratio of areas of triangles APB and APC
is a rational number.
Solution 1. Let H be the foot of the alti- C
tude drawn from C. First prove that | AH |
and | BH | are rational numbers. For that, L
use the Pythagorean theorem for triangles
ACH and BCH to obtain | AH |2 + |CH |2 = K
2 2 2 2
| AC | and | BH | + |CH | = | BC | . There-
P
fore | AC |2 − | BC |2 = | AH |2 − | BH |2 =
(| AH | − | BH |) · (| AH | + | BH |) = (| AH | − A H B
| BH |) · | AB|. We see that | AH | − | BH | =
| AC|2 −| BC|2 Fig. 6
| AB |
is rational and so are | AH | =
| AB |+(| AH |−| BH |)
2 and | BH | = | AH | − (| AH | − | BH |). Let now K be the
projection of P to BC (see Fig. 6). As P lies on the angle bisector of B,
it is equidistant from both AB and BC, i.e., | PH | = | PK |. Consequently,
10
S APB | AB| · | PH | | AB| S |CP| · | BH | | BH |
= = . As CH ⊥ AB, also BPC = = .
SBPC | BC | · | PK | | BC | S APC |CP| · | AH | | AH |
S | AB| | BH |
Thus, APB = · is rational as a product of two rational numbers.
S APC | BC | | AH |
Solution 2. Let H be the foot of the altitude drawn from C and let L be
the projection of P to AC (see Fig. 6). Now ∠CPL = 90◦ − ∠ ACH = 90◦ −
(90◦ − ∠CAB) = ∠CAB, giving ||PH |
PL |
| PH |
= | PC|·cos ∠CAB
. The angle bisector
| PH | | BH |
theorem gives | PC|
= | BC|
= cos ∠ ABC. Consequently,
S APB | AB| · | PH | | AB| cos ∠ ABC
= = · .
S APC | AC | · | PL| | AC | cos ∠CAB
| AB|
As the side lengths of the triangle ABC are integers, is rational. By the
| AC |
cosine law, the cosines of the angles of triangle ABC are rational, whence
cos ∠ ABC S
is rational. Altogether, APB is rational.
cos ∠CAB S APC
OC-15. (Seniors.) Prove that the set of integers {0, 1, 2, . . . , 2n − 1} can be

partitioned into n + 1 disjoint subsets A0 , A1 , . . . , An such that both of the
following hold:
a) If k + l = n, then the subsets Ak and Al have the same number of
elements.
b) If s and t are non-negative integers and s + t 6 n, then for an arbitrary
element z in the set As+t , there exist elements x and y from the sets As
and At , respectively, such that x + y = z.
Solution 1. Divide the set A into subsets such that the subset Ak consists
of only those numbers which have exactly k ones in their binary represen-
tation. Then A0 = {0}, A1 = {1, 2, 4, . . . , 2n−1 }, . . . , An = {2n −1}. Let us
show that both conditions are met. The first condition is met because the
numbers with k ones are in one-to-one correspondence with the numbers
with n − k ones: given a number, simply replace all ones in its binary repre-
sentation by zeros and vice versa. To show that the second condition is met,
choose an arbitrary number z from the set As+t . Its binary representation
contains exactly s + t ones. Construct a binary number x by choosing s ones
from the binary representation of z and filling all other binary places by ze-
ros, analogously construct a second number y based on remaining ones in
z. Then x ∈ As , y ∈ At and x + y = z.
Solution 2. Let us prove the statement by induction. If n = 1, then
A = {0, 1}, and taking A0 = {0} and A1 = {1} we get a partition that
satisfies both of the requirements.
Assume now that we have a partition C0 , C1 , . . . , Cn for the set C =
{0, 1, . . . , 2n − 1}. Construct a partition of the set A = {0, 1, . . . , 2n+1 − 1}
based on that. First, generate the sets B0 , B1 , . . . , Bn as follows: the elements
11
of the subset Bi are derived from the elements of the subset Ci by adding
2n to them. The subsets B0 , B1 , . . . , Bn form a partition of the set A \ C =
{2n , 2n + 1, . . . , 2n+1 − 1} and from the construction for all i = 0, 1, . . . , n
the corresponding subsets Bi and Ci have the same number of elements.
Now, let Ai = Ci ∪ Bi−1 for all i = 1, 2, . . . , n and in addition to that,
A0 = C0 and An+1 = Bn . Then, | A0 | = | An+1 | = 1, and if k + l = n + 1, then
also k, l 6= 0, | Ak | = | Bk−1 | + |Ck | = |Ck−1 | + |Ck | and | Al | = | Bl −1 | + |Cl | =
|Cl −1 | + |Cl |. As (k − 1) + l = k + (l − 1) = n, we see that |Ck−1 | = |Cl | and
|Ck | = |Cl −1 |; thus | Ak | = | Al |.
To verify that the second condition is met, let z be an arbitrary element
of As+t . If t = 0, then As = As+t and At = A0 = {0}, so we can take
x = z and y = 0. Now assume t > 1. If z < 2n , then z is an element of
Cs+t and thus there exist elements x and y in the sets Cs ⊂ As and Ct ⊂ At ,
respectively, such that x + y = z. If z > 2n , then z is an element of Bs+t−1,
i.e. z − 2n is an element of the set Cs+t−1 and thus the sets Cs and Ct−1
contain elements x and y, respectively, such that x + y = z − 2n . But now
x + (y + 2n ) = z, where y and z + 2n are elements of the sets Cs ⊂ As and
Bt−1 ⊂ At , respectively. So the statement holds for all positive n.
Remark. The subsets A0 , A1 , . . . , An formed in Solution 2 are actually the
same as those formed in Solution 1.
OC-16. (Seniors.) How many positive integers are there that are divisi-
ble by 2010 and that have exactly 2010 divisors (1 and the integer itself
included)?
Answer: 24.
Solution. Let N be a positive integer that is divisible by 2010 and that
has exactly 2010 positive divisors. Since 2010 = 2 · 3 · 5 · 67, also N should
be divisible by these four primes. Thus, N = 2a · 3b · 5c · 67d · s, where a, b, c,
d > 0 and s is not divisible by any of the primes 2, 3, 5, 67. All the factors of
N can be expressed as 2i · 3 j · 5k · 67l · t, where 0 6 i 6 a, 0 6 j 6 b, 0 6 k 6 c,
0 6 l 6 d, and t is a factor of s. There are a + 1 choices for i (from 0 to a) and
similarly, there are b + 1, c + 1 and d + 1 choices for j, k and l, respectively.
Therefore, N has δ( N ) = ( a + 1)(b + 1)(c + 1)(d + 1)δ(s) different factors,
where δ( x ) stands for the number of factors of x. We require δ( N ) = 2010.
As a + 1 > 1, b + 1 > 1, c + 1 > 1, and d + 1 > 1, we see that each of
these numbers is divisible by some prime numbers and the number δ( N ) =
( a + 1)(b + 1)(c + 1)(d + 1)δ(s) can thus be expressed as a product of at
least four prime numbers. But as 2010 itself is a product of exactly four
prime numbers, we conclude that a + 1, b + 1, c + 1, and d + 1 are exactly
those primes 2, 3, 5, and 67, in some order, and δ(s) = 1. From the latter
condition we see that s = 1 because any numbers bigger than 1 has more
than one factor. So for N to satisfy the conditions, N must be expressible as
2a · 3b · 5c · 67d , where a, b, c, d are the numbers 1, 2, 4, and 66 in some order.
Thus there are 4! = 24 numbers satisfying the conditions.
12
FR-1. (Grade 9.) Juku discovered that of the things in his satchel, 60 per-
cent were ugly and 76 percent were useless. He scrapped all things that
were both ugly and useless, and added things that were both beautiful and
useful. After this, of the things in Juku’s satchel, 25 percent are ugly and 45
percent are useless. How many percent of the things in Juku’s satchel were
both beautiful and useful initially?
Answer: 4 percent.
Solution. Observe that the amount of things that are beautiful but use-
less and things that are useful but ugly remained unchanged. The differ-
ence between the percentages of these things was 76% − 60% = 16% before
displacement but is 45% − 25% = 20% after that. Hence the overall num-
ber of things in the satchel decreased 20 : 16 = 1.25 times. Things that are
beautiful but useless form 45% of all things after the displacement, hence
they formed 45% : 1.25 = 36% before it. As there were 100% − 60% = 40%
of beautiful things in total, the things that were both beautiful and useful
constituted 40% − 36% = 4% of the content of the satchel.
Remark. This problem can of course be solved in completely standard
ways via linear equations.
FR-2. (Grade 9.) There are 8 identical dice. The numbers 4, 5, 6 6

are written on three faces of the dice, as shown in the figure, and
the remaining faces carry the numbers 1, 2, 3 so that the sum of 4 5
the numbers written on each pair of opposite faces is 7.
a) Show that using these dice, it is possible to form a 2 × 2 × 2 cube so
that every two faces that touch each other carry the same number.
b) Is it possible to do this in such a way that only numbers 4, 5, 6 occur
on the outer surface of the resulting cube?
Answer: b) no.
Solution. a) Put together four dice as depicted in Fig. 7. These dice form
the lower layer of the cube. On top of this, place another similar layer
turned upside down. By the con- 3 4 4
struction of the layer, the num-
2 6 5 5 6 2 1 5 6
bers on the faces touching each
other within one layer coincide 4 3 3
everywhere. As the second layer 4 3 1 3
is turned upside down, the num-
5 6 2 2 6 5 5 4 2 2 6 5
bers on faces of cubes of different
layers that touch each other also 3 4 6 4
coincide.
Fig. 7 Fig. 8
13
b) Suppose it is possible to form a 2 × 2 × 2 cube so that its surface con-
tains only numbers 4, 5, 6. As exactly 3 faces of each unit cube are visible, all
three numbers must occur on those. Place the cube in such a way that the
upper layer has 6 in its southeastern corner (see Fig. 8). Then, as the only
possibility, the upper layer must have 4 in its southwestern corner and 5 in
its northeastern corner. Now it is impossible to place a dice in the north-
western corner since it should touch both of its neighbors with number 1.
FR-3. (Grade 10.) Prove that

a2 + bc b2 + ca c2 + ab
+ + > a+b+c
b+c c+a a+b
for all positive real numbers a, b, c.
Solution 1. W.l.o.g., assume a > b > c. Then
a2 + bc b2 + ca c2 + ab
+ + =
b+c c+a a+b
a2 + ( b + c ) c − c2 b2 + ( c + a ) a − a 2 c 2 + ( a + b ) b − b2
= + + =
b+c c+a a+b
a2 − c2 b2 − a 2 c 2 − b2
= +c+ +a+ + b > a + b + c,
b+c c+a a+b
since
a2 − c2 b2 − a 2 c 2 − b2 a 2 − b2 + b2 − c 2 b2 − a 2 c 2 − b2
+ + = + + =
b+c c+a a+b b+c c+a a+b
! !
1 1 1 1
= ( a 2 − b2 ) − + ( b2 − c 2 ) − =
b+c c+a b+c a+b
( a2 − b2 )( a − b) (b2 − c2 )( a − c)
= + > 0.
(b + c)(c + a) (b + c)( a + b)
Solution 2. Rearranging and transforming the expression gives
a2 + bc b2 + ca c2 + ab
−a+ −b+ −c =
b+c c+a a+b
a2 + bc − ab − ac b2 + ca − bc − ba c2 + ab − ca − cb
= + + =
b+c c+a a+b
( a − b)( a − c) (b − c)(b − a) (c − a)(c − b)
= + + =
b+c c+a a+b
( a2 − b2 )( a2 − c2 ) + (b2 − c2 )(b2 − a2 ) + (c2 − a2 )(c2 − b2 )
= =
(b + c)(c + a)( a + b)
a 4 + b4 + c 4 − a 2 b2 − a 2 c 2 − b2 c 2
= =
(b + c)(c + a)( a + b)
( a 2 − b2 ) 2 + ( c 2 − a 2 ) 2 + ( b2 − c 2 ) 2
= > 0.
2(b + c)(c + a)( a + b)
14
FR-4. (Grade 10.) Find all quadruples (w, x, y, z) of positive integers such
that w x + wy = wz .
Answer: (w, x, y, z) = (2, n, n, n + 1), where n is any positive integer.
Solution 1. Consider the following cases.
If w = 1, then no solution can exist, since the l.h.s. of the equality equals
2 while the r.h.s. equals 1.
If w > 2, then x < z and y < z, i.e., x 6 z − 1 and y 6 z − 1. Thus
w x + wy 6 wz−1 + wz−1 = 2 · wz−1 6 w · wz−1 = wz . To satisfy the equation,
equalities must hold in both inequalities and thus x = y = z − 1 and w = 2.
This gives the solutions (w, x, y, z) = (2, n, n, n + 1), where n is an arbitrary
positive integer.
Solution 2. In the case w = 1 there are no solutions because 1 + 1 =
2 is not a power of 1. Assume in the rest that w > 1. W.l.o.g., assume
x 6 y < z. Then the equation takes the form w x (1 + wy− x ) = wz , whence
1 + wy− x = wz− x . Consequently, 1 + wy− x is a positive power of w and is
divisible by w. If y − x were positive, then wy− x would also be divisible by
w, whence 1 should be divisible by w, which is impossible. The remaining
case y − x = 0 leads to 2 = wz− x that gives w = 2 and z − x = 1 as the only
possibility. Hence the solutions of the equation are of the form (w, x, y, z) =
(2, n, n, n + 1), where n is any positive integer. Checking shows that all
these quadruples satisfy the equation.
FR-5. (Grade 10.) Each side of a convex quadrangle ABCD is a diameter of

a circle. All four circles pass through the same point O, different from the
vertices of the quadrangle, and no two circles have common points other
than those mentioned. Prove that ABCD is a rhombus.
Solution. Since AB, BC, CD, and DA
are diameters (Fig. 9), AOB, BOC, COD,
and DOA are right angles. Hence AOC Q
and BOD are straight angles, i.e., the di- D C
agonals of the quadrangle meet at O. The
circles drawn on the opposite sides AB
and CD cannot have common points be- O
sides O, since otherwise one circle would
pass through three collinear points (two
A P B
vertices of the quadrangle and the point
O). Consequently, the circles drawn on
AB and CD must touch at O. Let P and
Q be the centers of these circles, respec- Fig. 9
tively; then O lies on the segment PQ. As
| AP| = | PO| and |CQ| = | QO|, isosceles triangles give ∠BAC = ∠PAO =
∠POA = ∠QOC = ∠QCO = ∠DCA. Thus AB and CD are parallel. Simi-
larly, the remaining sides are parallel. Thus ABCD is a rectangle; the diag-
onals of a rectangle are perpendicular only if the rectangle is a rhombus.
15
− + −
+ 0 + 0
− + − −
+ − + 0 +
− 0 0
0 + + − 0 − + 0
+ − 0 − C
0 + − + 0
− + − C
A B
A B D C
D C − + −
0 − + 0 +
+ 0 − + −
A B A B
Fig. 10 Fig. 11
FR-6. (Grade 10.) Every face of a unit cube has one of numbers −1, 0,
1 written on it in such a way that every two faces with a common edge
contain different numbers. Is it possible to form
a) a cube of size 2 × 2 × 2;
b) a cube of size 3 × 3 × 3
so that in the grids that come up on the faces, every two squares with a
common side contain different numbers and the sum of all numbers on
each face equals 0?
Answer: a) yes; b) yes.
Solution. First note that the placement of the numbers on the faces of the
unit cube is unique. Indeed, let a number x be written on some face; then
the neighboring faces contain alternately the other numbers y and z, while
the opposite face again contains x. This means that each of the numbers −1,
0, 1 occurs in one pair of opposite faces. Figures 10 and 11 show suitable
constructions (where − and + denote −1 and 1, respectively).
Remark. These constructions can be easily generalized to arbitrary (even
or odd, respectively) integral size of the cube.
FR-7. (Grade 10.) The size of the angle ABC, expressed in degrees, in a
right triangle ABC is an integer. It is known that for some positive integer
n, one can choose points K0 = A, K2 , . . . , K2n on the hypotenuse AB and
points K1 = C, K3 , . . . , K2n+1 = B on the leg CB in such a way that each
triangle Ki−1 Ki Ki+1 with i = 1, . . . , 2n is isosceles with base Ki−1 Ki+1 . Find
all possible values of the size of angle ABC.
Answer: 2◦ , 6◦ , 10◦ , 18◦ , and 30◦ .
Solution. Let ∠ ABC = α (Fig. 12). Then the base angle of the last
isosceles triangle K2n−1 K2n K2n+1 is α. The base angle of the second last
isosceles triangle K2n−2 K2n−1 K2n has the size 180◦ − (180◦ − 2α) = 2α.
The base angle of the next triangle before it, K2n−3 K2n−2 K2n−1 , has the size
16
180◦ − (180◦ − 4α) − α = 3α. Generally, the size of the base angle of triangle
K2n−i K2n−i+1 K2n−i+2 is 180◦ − (180◦ − 2 · (i − 1)α) − (i − 2)α = iα (i = 3,
. . . , 2n). Thus the base angle of triangle ACK2 = K0 K1 K2 has the size 2nα.
Now in the triangle ABC we get 90◦ = ∠ BAC + ∠ ABC = 2nα + α, whence
◦
α = 2n90+1 . By the conditions of the problem, α must be an integer, hence
2n + 1 is an odd divisor C = K1
of 90 and is greater than
1 (as a triangle cannot 3α K3
have two angles of the 3α
size 90◦ ). Such divisors α
are 3, 5, 9, 15, and 45 4α 4α 2α 2α α
that give the solutions A = K K2 K4 B = K5
0
30◦ , 18◦ , 10◦ , 6◦ , and 2◦ ,
respectively. Fig. 12
FR-8. (Grade 11.) Find all integers that cannot be expressed as a sum of at
least three consecutive terms of some non-constant arithmetic sequence of
integers.
Answer: 1 and −1.
Solution 1. First prove that 1 and −1 are not expressible as the sum of at
least three consecutive terms of an arithmetic sequence of integers. Let a1 ,
a2 , . . . , ak be k consecutive terms of an arithmetic sequence, where k > 3.
They sum up to s = a1 + 2
ak
· k. If k is odd, then s is divisible by k. If k is even,
k
then s is divisible by 2 > 1. In both cases, s differs from 1 and −1.
Now prove that every integer s other than 1 or −1 is expressible as the
sum of at least three consecutive terms of an arithmetic sequence of inte-
gers. If s = 0, then s = −1 + 0 + 1. If s is different from zero and is even,
i.e., s = 2t, where t 6= 0, then −t, 0, t, 2t sum up to 2t = s. If s is odd, i.e.,
s = 2t + 1, then −t + 1, . . . , 0, 1, . . . , t − 1, t, t + 1 are consecutive terms of
an arithmetic sequence; they sum up to t + (t + 1) = 2t + 1 = s, since the
terms −t + 1 through t − 1 mutually cancel.
Solution 2. Let a1 be the first of the consecutive terms and d be the com-
mon difference of consecutive terms. The sum of n consecutive terms is
2a1 + d(n − 1)
s = · n. Thus 2s = (2a1 + d(n − 1))n. If s = 1 or s = −1,
2
then this equality cannot hold because n > 3 divides neither 2 nor −2. If
s = 0, then choose the portion of the arithmetic progression to be −1, 0,
1. If s differs from these numbers, then let n = 2|s|, d be an arbitrary odd
1 − ( n − 1) d −1 − ( n − 1 ) d
number, and a1 = if s > 0, and a1 = if s < 0.
2 2
FR-9. (Grade 11.) Find all integral solutions of the equation x3 − y3 =

3xy + 1.
Answer: ( x, y) = (n + 1, n), where n is any integer, or ( x, y) = (−1, 1).
17
Solution. First assume x > y. Then 3xy + 1 = x3 − y3 = ( x − y)( x2 +
xy + y2 ) > ( x − y) · 3xy. Thus 3xy = x3 − y3 − 1 > 1 − 1 = 0 because
x > y. If 3xy > 0, then 3xy > 3, hence the inequality 3xy + 1 > ( x − y) · 3xy
derived above implies x − y = 1. If 3xy = 0, then either x = 0 or y = 0
and in both cases the only possibility is x − y = 1 again. An elementary
check shows that all pairs ( x, y) = (n + 1, n), where n is an integer, satisfy
the initial equation.
Now assume x = y. Then the equation has no solutions, since the l.h.s.
is 0 while the r.h.s. is positive.
Finally assume x < y. Then the l.h.s. of the equation is negative, show-
ing that xy is negative. Hence x < 0 and y > 0. Denoting − x = z and multi-
plying the equation by (−1) leads to new equation z3 + y3 = 3zy − 1. Then
3zy − 1 = z3 + y3 = (z + y)(z2 − zy + y2 ) > (z + y)zy. Hence z + y < 3,
giving z = y = 1, i.e., x = −1, y = 1, as the only possibility. It is easy to
check that this satisfies the equation.
FR-10. (Grade 11.) Let CM be the median of a triangle ABC. Prove that
the product of the circumradius of ACM and the altitude drawn from M
in ACM equals the product of the circumradius of BCM and the altitude
drawn from M in BCM.
Solution 1. Let ∠CAB = α and ∠CBA = C
β (see Fig. 13), and let r and s be the cir-
cumradii of the triangles ACM and BCM,
respectively. By the sine law in the triangle U
|CM | V
ACM we obtain sin α = 2r, reducing to r =
|CM | |CM | α β
2 sin α . Analogously, s = 2 sin β . Let U and V
A M B
be the feet of altitudes drawn from the point
M in triangles ACM and BCM, respectively. Fig. 13
| AB |
Then | MU | = | AM | sin α = 2 sin α. Anal-
ogously, | MV | = | AB |
2 sin β. Thus r · | MU | =
|CM |·| AB |
4 = s · | MV |.
Solution 2. Let x and y be the altitudes drawn from M in the triangles
ACM and BCM, respectively. Let r and s be the circumradii of these tri-
angles, respectively. The areas of triangles ACM and BCM are equal be-
cause of | AM | = | BM | and the common altitude drawn from C. Therefore
| AC | · x = | BC | · y. Denote ∠ AMC = γ. The sine law gives | AC | = 2r sin γ
and | BC | = 2s sin(180◦ − γ) = 2s sin γ. Hence 2r sin γ · x = 2s sin γ · y. As
γ 6= 0, this implies rx = sy.
Solution 3. The formulas S = abc ah
4R and S = 2 , where a, b, c are the
side lengths, R is the circumradius and h is the altitude corresponding to a,
together give Rh = bc 2 . The product of the circumradius of ACM and the
| AM | · |CM | | BM | · |CM |
altitude drawn from M is thus . Analogously, for
2 2
triangle BCM. These two products are equal, since | AM | = | BM |.
18
FR-11. (Grade 11.) The inhabitants of a city of naturals are natural num-
bers. Every two different inhabitants may either be or not be friends. Call
a city neighborly if every two inhabitants share a common friend if and only
if one of the numbers is divisible by the other. Can a city whose inhabitants
are precisely 1, 2, . . . , 2011 be neighborly?
Answer: no.
Solution 1. Let p be a prime inhabitant of a neighborly city 1
(see Fig. 14). As p is divisible by 1 and p 6= 1, the inhabitants
1 and p share a common friend k. As k is divisible by 1 and i
k 6= 1, the inhabitants 1 and k share a common friend i.
If i 6= p, then k is a common friend of i and p. This means p
that i is divisible by p because the primality of p does not per-
mit the divisibility the other way round. If i = p, then 1 is a k
common friend of k and p. This analogously means that k is
divisible by p. In both cases, the city contains a multiple of p Fig. 14
that is greater than p.
In the city with inhabitants 1, . . . , 2011, the prime inhabitant 2011 has no
larger multiples. Hence this city cannot be neighborly.
Solution 2. Suppose that this city is neighborly. Choose arbitrary 12
prime inhabitants (for instance, the first 12 primes). Each of them shares a
common friend with 1; let these friends be a1 , a2 , . . . , a12 . These numbers are
all different since otherwise one of them would be a common friend to two
prime numbers. W.l.o.g., assume a1 < a2 < . . . < a12 . As 1 is a common
friend of all them, each of a2 , . . . , a12 must be divisible by the previous term.
This implies that a12 > 212 = 2048, a contradiction.
Remark. The use of primality of 2011 in Solution 1 can be replaced with
the application of Chebyshev’s theorem, choosing p arbitrarily in such a
way that 1005 < p < 2010.
FR-12. (Grade 12.) Find the last digit of the number 11 + 22 + 33 + . . . +

20112011.
Answer: 8.
Solution. Consider the sum modulo 2 and modulo 5. As powers of odd
numbers are odd and powers of even numbers are even, the number of
odd summands equals the number of odd elements in set {1, . . . , 2011}.
As there are an even number of odd elements in this set, the sum given in
the problem is even. Concerning modulo 5, note that 0 to every power is
congruent to 0 and a4 is congruent to 1, whenever 1 6 a 6 4. Thus for all a
such that 1 6 a 6 20, ( a + 20) a+20 ≡ a a+20 = a a · a20 = a a · ( a4 )5 ≡ a a · 1 =
a a (mod 5). Hence the remainders modulo 5 repeat periodically with the
period 20. As there are 100 full periods and 100 is divisible by 5, the sum of
the last 2000 summands is congruent to 0 modulo 5. It remains to compute
the remainders of the first 11 summands: 11 = 1, 22 = 4, 33 = 27 ≡ 2,
44 ≡ 40 = 1, 55 ≡ 01 = 0, 66 ≡ 12 = 1, 77 ≡ 23 ≡ 3, 88 ≡ 30 = 1,
19
99 ≡ 41 = 4, 1010 ≡ 02 = 0, 1111 ≡ 13 = 1. Hence the overall sum is
congruent to 3 modulo 5. Consequently, the last digit of this sum is 8.
FR-13. (Grade 12.) Does there exist a positive real number C such that the
inequality
x1 x2 + x1 x3 + x1 x4 + x2 x3 + x2 x4 + x3 x4 6 C ( x1 x2 + x2 x3 + x3 x4 + x4 x1 )
holds for arbitrary positive real numbers x1 , x2 , x3 , x4 ?
Answer: no.
Solution. For simplicity, denote A = x1 x2 + x1 x3 + x1 x4 + x2 x3 + x2 x4 +
x3 x4 and B = x1 x2 + x2 x3 + x3 x4 + x4 x1 . Choose x1 = x3 = u and x2 =
x4 = 1. Then A > x1 x3 = u2 and B = 4u. Hence A B > 4 . As u can be
u
arbitrarily large, no constant C such that A 6 CB exists.
FR-14. (Grade 12.) In a rectangle ABCD we have | AB| = a and | BC | = b,

where a > b. Let E be a point in the interior of side AB such that there is
exactly one possibility to choose points F, G, H on the sides BC, CD, DA,
respectively, in such a way that EFGH is a rectangle, too. Find the ratio of
the areas of rectangles EFGH and ABCD.
Answer: 12 .
Solution. The rectangles ABCD and EFGH D G C
have a common center O (see Fig. 15). As rect-
angles are cyclic quadrangles, the point F lies
on the circle with center O and radius |OE|. O
This circle intersects the side BC at two points H F
symmetric w.r.t. the midpoint of the side. To
have exactly one point common to the circle
and the side, the side must be tangent to the
circle and F must be the midpoint of BC. Anal- A E B
ogously, H must be the midpoint of DA. In tri- Fig. 15
angle EFH, the side HF has length a and the
corresponding altitude is 2b , giving 12 · a · 2b = ab
4 as the area of the triangle.
The triangle GFH has the same area. Hence the area of rectangle EFGH is
2 · ab ab
4 = 2 that makes up a half of the area ab of the rectangle ABCD.
FR-15. (Grade 12.) Ants has three pencils, each of a different color. In how
many ways can he paint the faces of a regular octahedron in such a way that
faces with a common edge always have different colors? Colorings that can
be obtained from each other via rotations of the octahedron are considered
the same.
Answer: 15.
Solution. One color can occur at most 4 times (at most twice among
the faces adjacent to either one of some two opposite vertices). Thus the
possible numbers of colors are 4, 4, 0 or 4, 3, 1 or 4, 2, 2 or 3, 3, 2.
20
i) Case 4, 4, 0. There are 3 possibilities to choose two colors from the
three. After that, there is only one possibility to paint the octahedron. Thus
there are 3 possibilities to paint.
ii) Case 4, 3, 1. Ordering the 3 colors can be done in 6 ways. After that,
there is only one possibility to paint the octahedron, since the color used 4
times must occur twice among the faces adjacent to one vertex and twice
among the faces adjacent to the opposite vertex. The remaining two colors
can be deployed in principle in only one way. Thus there are 6 possibilities
to paint.
iii) Case 4, 2, 2. Choosing the color that is used 4 times can be done in
3 different ways. After that, the octahedron can be painted in only 1 way,
since after 4 faces have been painted with the same color, faces with either of
the other colors must meet at the same vertex. Thus there are 3 possibilities
to paint.
iv) Case 3, 3, 2. Choosing the color that is used only twice can be done
in 3 ways. If the faces painted with this color met at a common vertex V,
the faces adjacent to the opposite vertex would be painted alternately with
the other two colors. But then the remaining two faces adjacent to V would
have to be painted with the same color, that contradicts the case assump-
tion. Hence the color that occurs twice is used on a pair of opposite sides.
The other colors occur alternately on the surface formed by the remaining
six faces. Thus there are 3 possibilities to paint.
Consequently, the number of all colorings is 3 + 6 + 3 + 3 = 15.
FR-16. (Grade 12.) Inside a regular 2n-gon, an arbitrary point is chosen

and connected to every vertex of the 2n-gon. The triangles obtained are
colored alternately black and white so that triangles with a common side
are of different color. Prove that the sum of the areas of all white triangles
equals the sum of the areas of all black triangles.
Solution 1. If n = 2, i.e., the 2n-gon is a square, then the claim holds
because the base sides of white triangles are the opposite sides of the square
and the altitudes lie on the same line, so the total area of white triangles is a
half of the area of the square. Assume in the following that n > 2. Consider
the regular n-gon whose sides are obtained by prolonging all sides of the
2n-gon that belong to white triangles (see Fig. 16). Join the point chosen
inside the initial 2n-gon with all vertices of the n-gon. The altitude drawn in
any white triangle coincides with the altitude drawn in the corresponding
triangle in the n-gon, while the ratio of the corresponding base sides equals
the ratio of the side length of the 2n-gon and the side length of the n-gon,
denote it by c. Thus the total area of white triangles is cSn , where Sn is the
area of the n-gon. Analogously, the total area of black triangles is cSn , too.
Solution 2. The claim of the problem is equivalent to the statement that
the sum of the altitudes of black triangles drawn to the sides that coin-
cide to the sides of the 2n-gon is equal to that of white triangles. Let O be
21
A6 A5
A2 A1
A7 A4
A
A3
l α
A8 A3
O
A1 A2
A4
Fig. 16 Fig. 17
the center of the 2n-gon, A be the point chosen inside, α be the angle be-
tween line OA and the line perpendicular to a side, and l be the distance
between O and any side of the 2n-gon (see Fig. 17). Then the altitude of
the corresponding triangle is l − |OA| cos α. The altitude of the next tri-
angle of the same color can be expressed similarly but α is replaced with
◦
α + 360
n . Thus the sum of all altitudes of the triangles of the same color is
◦ ( n−1)360◦
nl − |OA| · (cos α + cos(α + 360
n ) + . . . + cos( α + n )). To show that
◦
the sum inside parentheses equals 0, multiply the sum by sin 360
2n . Since
◦ ◦ ( k + 1 )·360◦ ( k − 1 )·360◦
cos(α + k·360 360 1
n ) sin 2n = 2 (sin(α +
2
n ) − sin(α + 2
n )), a tele-
◦
scoping sum emerges and after reduction one obtains − sin(α − 360 2n ) +
1
( n− 2 )·360 ◦
sin α + n = 0. Hence for both colors, the sum of the altitudes
of all triangles of this color is nl.
◦ ( n−1)·360◦
Remark. The sum cos α + cos(α + 360 n ) + . . . + cos(α + n ) in So-
lution 2 can also be computed as follows. Denote z = cos α + i sin α, where i
◦ ◦
is the imaginary unit and let zk = cos k·360 n + i sin k·360
n , k = 0, 1, . . . , n − 1.
Then the sum under consideration is the real part of the complex number
z · z0 + z · z1 + . . . + z · zn−1 . Thus z · z0 + z · z1 + . . . + z · zn−1 = z · (z0 +
z1n −1
z1 + . . . + zn−1 ) = z · (z01 + z11 + z21 + . . . + z1n−1 ) = z · z1 −1 = z · 0 = 0,
whence the sum under consideration is equal to 0.
22
First day
TS-1. Two circles lie completely outside each other. Let A be the point
of intersection of internal common tangents of the circles and let K be the
projection of this point onto their external common tangent. The tangents,
different from the common tangent, to the circles through point K meet the
circles at M1 and M2 . Prove that the line AK bisects the angle M1 KM2 .
Solution 1. Let L1 and L2 be the points of tangency of the external com-
mon tangent of the circles, N1 and N2 be the points of tangency of an in-
ternal common tangent, and O1 and O2 be the centers of the two circles
(see Fig. 18). As all the lines O1 L1 , AK, and O2 L2 are perpendicular to the
line L1 L2 , they are parallel to each other and thus || LL1 K |
= ||O1 A|
. The trian-
2 K| O2 A |
gles O1 AN1 and O2 AN2 are similar because they are both right-angled and
|O A | |O N | |O L |
have the same vertical angles. Thus, |O1 A| = |O1 N1 | = |O1 L1 | . Therefore, the
2 2 2 2 2
right-angled triangles O1 L1 K and O2 L2 K are similar due to proportionality
of their legs. Hence, ∠ L1 KO1 = ∠ L2 KO2 . As ∠ L1 KM1 = 2∠ L1 KO1 and
∠L2 KM2 = 2∠L2 KO2 , we also get that ∠L1 KM1 = ∠L2 KM2 . Together with
the equality ∠ L1 KA = ∠ L2 KA = 90◦ this implies ∠ M1 KA = ∠ M2 KA.
Solution 2. Both of the circles appear at the same angle, when viewed
from the point A. To solve the problem, it is enough to show that both of the
circles also appear at the same angle, when viewed from the point K. Let the
centers of the circles have the coordinates O1 ( a1 , b1 ) and O2 ( a2 , b2 ) and let
r1 and r2 be the radii of the circles. The two circles appear at the same angle
from the point P( x, y) if and only if |Or1P| = |Or2P| , i.e., √ r1
2 2
=
1 2 ( x − a1 ) +( y −b1 )
√ r2
2 2
. Simple algebra shows that this equation is equivalent to
( x − a2 ) +( y −b2 )
(r12 − r22 ) x2 + (r12 − r22 )y2 + c1 x + c2 y + c3 = 0, where c1 , c2 , and c3 are some
L2
L1 K
N1
O1 O2
A
M1
M2
N2
Fig. 18
23
constants. If r1 = r2 , then the statement clearly holds. If r1 6= r2 , then the
last equation is that of a circle. Point A as well as the point D of intersection
of the external common tangents both lie on that circle, and from symmetry,
the diameter of that circle is AD. As AK is perpendicular to the external
common tangent of the circles, the point K also lies on that circle.
Remark. The statement would hold even if we swapped the internal and
external tangents of the circles and considered angle M1 KM2 as the angle
between the lines KM1 and KM2 instead.
TS-2. Let n be a positive integer. Prove that for each factor m of the num-
ber 1 + 2 + . . . + n such that m > n, the set {1, 2, . . . , n} can be partitioned
into disjoint subsets, the sum of the elements of each being equal to m.
Solution. For every positive integer k, denote Sk = {1, 2, . . . , k} and
k ·( k +1)
sk = 1 + 2 + . . . + k = 2 . Prove the claim by induction: assume there
exists the required partitions of S1 , . . . , Sn−1 and prove the same for Sn . Fix
an arbitrary m such that m | sn , m > n.
First assume m > 2n. Let d = smn . To construct d disjoint subsets of Sn
with equal sum, partition the set {n, n − 1, . . . , n − 2d + 1} into subsets
Mi = {n + 1 − i, n − 2d + i }, where i = 1, . . . , d. As m > 2n implies
( n−2d )·( n+1−2d)
> (n−2d
2d )·2d
s
n + 1 > 4d, one gets n−d 2d = 2d = n − 2d. Note
( n−2d )·( n+1−2d)
also that d | 2 , since (n − 2d) · (n + 1 − 2d) ≡ n(n + 1) ≡ 0
(mod 2d). Hence by the induction hypothesis, there exist disjoint subsets
L1 , . . . , Ld of Sn−2d with equal sum. Taking Mi ∪ Li for each i = 1, . . . , d
forms the desired partition of Sn .
Now assume n 6 m < 2n. If m = n, then the task is trivial (n must be
odd to be a divisor of sn , so take sets {i, n − i }, i = 1, 2, . . . , n− 1
2 , and {n}).
If m > n, then form the subsets Mi = {m − 1 − n + i, n + 1 − i } of the set
{n, n − 1, . . . , m − n} for i = 1, . . . , n − ⌈ m2−1 ⌉, the sum of the elements of
each being m. The solution is complete if the remaining numbers in Sn can
be divided into sets, the sum of the elements of each being also m.
If m is odd, then the set of remaining numbers is {1, 2, . . . , m − n − 1}.
For m = n + 1, this set is empty and the partition is trivial, so assume
( m−n−1)·( m−n)
m > n + 1. As m > m − n − 1 and m | 2 (the latter following
from (m − n − 1) · (m − n) ≡ (n + 1)n ≡ 0 (mod m) and the parity of m),
the desired partition exists by the induction hypothesis.
If m is even, then the set of remaining numbers of Sn also includes m2 .
But m2 | sm−n−1 again by m | (m − n − 1) · (m − n). The inequality m < 2n
implies m2 < n, so also m2 > m − n > m − n − 1. Here, m = n + 1 is
impossible, since n + 1 | sn implies 2 | n and 2 6 | n + 1. Hence the induction
hypothesis gives the existence of a partition of Sm−n−1 into subsets, the sum
s − n −1
of the elements of each being m2 . The number of these subsets is mm/2 =
( m−n−1)·( m−n)
m = m − 2n − 1 + 2 · (n2m
+1 ) n
, which is odd. Together with the
number m2 , the subsets can be grouped by two to form the desired partition.
24
TS-3. Does there exist an operation ∗ on the set of all integers such that
the following conditions hold simultaneously:
(1) for any integers x, y, z, ( x ∗ y) ∗ z = x ∗ (y ∗ z);
(2) for any integers x and y, x ∗ x ∗ y = y ∗ x ∗ x = y?
Answer: yes.
Solution. Define an operation ⊕ on the set of all non-negative integers,
which maps two non-negative integers a and b to a non-negative integer
a ⊕ b, such that for all i = 0, 1, . . . , ( a ⊕ b)i = ( ai + bi ) mod 2, where ni
stands for the binary digit corresponding to 2i in the binary representa-
tion of n. This operation satisfies condition (1) for all non-negative integers
because addition modulo 2 satisfies it. The operation also satisfies condi-
tion (2) because if x, y ∈ {0, 1}, then ( x + x + y) mod 2 = y mod 2 =
(y + x + x ) mod 2. As the set of non-negative integers as well as the set
of all integers are countable, there exists one-to-one correspondence f be-
tween these sets (e.g. mapping a non-negative integer x to the integer
(−1) x ⌊ x + 1
2 ⌋). Every integer can therefore be uniquely expressed in the form
f (n), where n is a non-negative integer. Therefore we can define the opera-
tion ∗ by the formula f ( x ) ∗ f (y) = f ( x ⊕ y). Following from the construc-
tion, both conditions (1) and (2) still hold.
Remark. This problem was inspired by problem 5 on Baltic Way 2006.
Second day
TS-4. Let a, b, c be positive real numbers such that 2a2 + b2 = 9c2 . Prove
that
2c c √
+ > 3.
a b
Solution 1. Using the AM-GM inequality for three terms twice, one gets
√ √
2c c (2b + a)c (2b + a) 2a2 + b2 ( b + b + a ) a 2 + a 2 + b2
+ = = = >
a b ab p √3ab 3ab
√ √ √ √
3 3
3 b2 a 3 a 4 b2
3
3 3 b2 a · a 2 b 3 3ab √
> = = = 3.
3ab 3ab 3ab
Solution 2. Using HM-QM inequality for a, a, b gives
√
r r r
3 3 a 2 + a 2 + b2 2a2 + b2 9c2
= 6 = = = 3c.
2 1 1 1 1 3 3 3
+ + +
a b a a b
Thus
2 1

3 √
+ · c > √ = 3,
a b 3
which implies the desired inequality.
25
Solution 3. By AM-GM for two terms,
2c c 2 (2b + a)2 c2 (2b + a)2 (2a2 + b2 )
+ = 2 2
= =
a b a b 9a2 b2
(( a2 + b2 ) + 4ab + 3b2 )(a2 + ( a2 + b2 )) (6ab + 3b2 )( a2 + 2ab)
= 2 2
> =
9a b 9a2 b2
(2a + b)( a + 2b) 2a2 + 5ab + 2b2 4ab + 5ab
= = > = 3.
3ab 3ab 3ab
Solution 4. The square of the l.h.s. of the desired inequality is
2c c 2 (2b + a)2 (2a2 + b2 ) 1 a 2 b2
+ = = 2 + 2 + .
a b 9a2 b2 9 b a2
Denoting b = x, the desired inequality reduces to 9 (2 + x )2 (2 + x12 ) > 3,
a 1
which is equivalent to (2 + x )2 (2x2 + 1) > 27x2. This in turn is equivalent

to 2x4 + 8x3 − 18x2 + 4x + 4 > 0, that is 2( x − 1)2 ( x2 + 6x + 2) > 0 after fac-
torization. This inequality holds, since on positive arguments the quadratic
polynomial x2 + 6x + 2 is positive.
Remark. There are also plenty of solutions via derivative.
TS-5. Prove that if n and k are positive integers such that 1 < k < n − 1,
then the binomial coefficient (nk) is divisible by at least two different primes.
Solution 1. Assume w.l.o.g. that n > 2k (if n < 2k, then interchange the
roles of k and n − k). Let p be an arbitrary prime number. Consider the
numbers that remain into the numerator of the expression
n · ( n − 1) · . . . · ( n − k + 1)

n
=
k k · ( k − 1) · . . . · 1
after reducing all factors by the highest power of p by which they are di-
visible. Suppose that some two of the k factors resulting after this step are
equal. Then the corresponding initial factors are of the form s · pi and s · p j ,
where i > j. But then n > s · pi > p · s · p j > p · (n − k) > 2 · (n − k),
which contradicts the assumption n > 2k. Hence the k new factors are pair-
wise different. As 1 < k < n − 1, the numerator initially contains at least
two consecutive numbers, at least one of which is not divisible by p. This
number does not change in the process described above. By the assumption
n > 2k, this number is greater than k. Consequently, the product remaining
in the numerator after elimination of powers of p is greater than the de-
nominator k · (k − 1) · . . . · 1. This means that the powers of p in the original
numerator cannot be completely cancelled out with the denominator. So
the canonical representation of (nk) cannot consist of a power of p only.
Solution 2. Suppose that for some n and k,
n · ( n − 1) · . . . · ( n − k + 1)

n
= = pt,
k k · ( k − 1) · . . . · 1
where p is a prime number and t is some positive integer. Let m be a num-
ber in {n, n − 1, . . . , n − k + 1}, in the canonical representation of which
26
the exponent of p is the largest. Then the exponent of p in the canonical
representation of n, n − 1, . . . , m + 1 coincides with that in the canonical
representation of n − m, n − m − 1, . . . , 1, respectively. Similarly, the expo-
nent of p in the canonical representation of m − 1, . . . , n − k + 1 coincides
with that in the canonical representation of 1, . . . , m − 1 − n + k, respec-
tively. Consequently, the exponent of p in the canonical representation of
the product n(n − 1) . . . (m + 1)(m − 1) . . . (n − k + 1) equals to that in the
canonical representation of the product (n − m)! (m − 1 − n + k)!. Since
( k − 1) ! k−1

k!
= k· =k·
(n − m)! (m − 1 − n + k)! (n − m)! (k − 1 − n + m)! n−m
is clearly an integer, the exponent of p in the canonical representation of
(n − m)! (m − 1 − n + k)! does not exceed that in the canonical representa-
tion of k!. Hence, the exponent of p in the canonical representation of (nk)
does not exceed that in the canonical representation of m. As the assump-
tions of the problem imply (nk) > (n2 ) > n > m, this leads to a contradiction.
Remark. More straightforward solutions can be presented using either
Legendre’s formula or Kummer’s theorem.
TS-6. On a square board with m rows and n columns, where m 6 n, some

squares are colored black in such a way that no two rows are alike. Find the
biggest integer k such that for every possible coloring to start with one can
always color k columns entirely red in such a way that no two rows are still
alike.
Answer: n − m + 1.
Solution 1. Prove that if m 6 n, one of the columns can always be colored
red. Then, when excluding this column, we can continue the process until
the number of columns is smaller than the number of rows, i.e. n − m + 1
times. Suppose we cannot color a single column red such that no two rows
still appear alike. Then, for every column there are at least two rows that
differ by only one square in that column. Consider those two rows for every
column. Now consider a graph with its vertices being the rows and its
edges being the pairs of rows of interest. As this graph has at least as many
edges as vertices, the graph contains a cycle. Consider an arbitrary row (i.e.
vertex) x of the cycle. Then the row it is followed by differs from row x
exactly by one square, assume this square is located at the column y. Every
next row differs from the previous one in exactly one column. As all these
columns differ from column y, all the other edges of this cycle correspond to
passing from one row to another such that the square in column y remains
the same. Therefore, the square in column y remains the same in all the
rows after x and it has to remain the same at the last passage which takes
us back to the row x. This, however, means that all squares in row x have
to be of the same color as the square at the row after x, a contradiction.
In general, no more rows can be colored red. Assume that all the squares
are colored black to start with, apart from the diagonal of an m × m-sub-
27
squareboard. If we color n − m + 2 columns red, there are always at least
2 columns which lie in that subsquareboard. But this means that we will
over-color the only two white squares in two rows and thus we end up
with two equally colored rows.
Solution 2. Consider the columns of the squareboard one by one. The
first column divides the set of all the rows into two subsets: one of them
consists of the rows which have the square at the first column white and
the other consists of those rows that have their first square black. If the first
column has all its squares white or all its squares black, then we can color
it red. Similarly, for every next column divide the subsets even further de-
pending on whether there is a black or a white square in that row in that
column. If no subsets are divided further at a particular step, we can color
that column red. Therefore, for all columns, either the number of subsets in-
creases by at least one or this column is colored red. As we started off with
one set and ended up with m subsets each containing one row only, there
has to be no more than m − 1 columns that were not colored red and thus
at least k = n − m + 1 that were. All these rows are still different from each
other because we colored only those columns which had no new informa-
tion about the differences between rows compared with previous columns.
Similarly to Solution 1 we can show that there always exists a coloring
for which no more columns can be colored red.
Solution 3. Prove by induction on the number of rows that we can always
color at least k = n − m + 1 columns red. If m = 1 then we can color all
the columns red, i.e. k = n = n − m + 1. Assume that for m = l we can
color at least n − l + 1 columns red such that no two rows appear alike.
Assume now that m = l + 1. Following from the induction hypothesis we
can color at least n − l + 1 columns red such that the first l rows remain
different. If after that coloring process the last row is different from the rest,
all the conditions required are satisfied and we can color at least n − m + 2
columns red. If the last row however is similar to any of the rows above it
(there can be only one of those) then as these rows were all different to start
with, there should exist a column at which the last row and the row that
appears similar after coloring actually differ. If we do not color this column,
all the rows will appear different and we can color n − m + 1 columns red.
Similarly to Solution 1 we can show that there always exists a coloring
for which no more columns can be colored red.

Number theory: OC-4, OC-7, OC-8, OC-16, FR-7, FR-8, FR-11, FR-12, TS-5
Algebra: OC-1, OC-11, OC-13, FR-1, FR-3, FR-4, FR-9, FR-13, TS-3, TS-4
Geometry: OC-2, OC-5, OC-9, OC-12, OC-14, FR-5, FR-10, FR-14, FR-16, TS-1
Discrete mathematics: OC-3, OC-6, OC-10, OC-15,FR-2, FR-6,FR-15, TS-2,TS-6
28
2011/2012

Tartu 2012
WE THANK:
University of Tartu
Problem authors Juhan Aru, Maksim Ivanov, Urve Kangro, Oleg Koshik,
Toomas Krips, Tuan Le (USA), Jim Leahy (Ireland), Härmel Nestra,
Heiki Niglas, Erik Paemurru, Reimo Palm
Translators Urve Kangro, Erik Paemurru
Editor Reimo Palm

The Estonian Mathematical Olympiad is held annually in three rounds: at
In each round of the Olympiad, separate problem sets are given to the
students of each grade. Students of grade 9 to 12 compete in all rounds,
students of grade 7 to 8 participate at school and regional levels only. Some
towns, regions and schools also organize olympiads for even younger stu-
dents. The school round usually takes place in December, the regional
round in January or February and the final round in March or April in
Tartu. The problems for every grade are usually in compliance with the
school curriculum of that grade but, in the final round, also problems re-
quiring additional knowledge may be given.
and the Seniors. In the first category, students up to the 10th grade can
participate; the other category has no restriction. Being successful in the
open contests generally assumes knowledge outside the school curriculum.
Based on the results of all competitions during the year, about 20 IMO
April or May, lasting two days; each day, the contestants have 4.5 hours to
solve 3 problems, similarly to the IMO. All participants are given the same
problems. Some problems in our selection contest are at the level of diffi-
culty of the IMO but somewhat easier problems are usually also included.
The problems of previous olympiads are available at the Estonian Math-
Besides the above-mentioned contests and the quiz “Kangaroo” other
regional and international competitions and matches between schools are
held as well.
*
This booklet presents the problems of the open contests, the final round
of national olympiad and the team selection contest. For the open contests
and the final round, selection has been made to include only problems that
have not been taken from other competitions or problem sources and seem
interesting enough. The team selection contest is presented entirely.
3
O1. (Juniors.) Find all four-digit numbers, which after deleting any one
digit turn into a three-digit number that is a divisor of the original number.
Answer: 1100, 1200, 1500, 2200, 2400, 3300, 3600, 4400, 4800, 5500, 6600,
7700, 8800, 9900.
Solution. Let abcd be a such number. Since abcd is divisible by abc, we
have d = 0. Since abcd = abc0 is divisible by abd = ab0, we have c = 0.
Since abcd = ab00 is divisible by acd = a00 and by bcd = b00, the number
ab is divisible by a and b. So b = ax and 10a = by with integer x and y.
Therefore 10a = axy, whence xy = 10. If x = 1, y = 10, then a = b, which
gives 9 possible numbers 1100, 2200, 3300, 4400, 5500, 6600, 7700, 8800, 9900.
If x = 2, y = 5, then 2a = b, which gives 4 possibilities 1200, 2400, 3600,
4800. If x = 5, y = 2, then 5a = b, which gives 1 number 1500. The case
x = 10, y = 1 is impossible, since a and b must be one-digit numbers.
O2. (Juniors.) Find the minimum number of colours required to paint all
points with integer coordinates in the plane in such a way that no two
points which are exactly five units apart have the same color.
Answer: 2.
Solution. Obviously at least 2 colors are necessary. Color all points ( x, y)
with even sum of coordinates with one color and all other points with an-
other color. All points that are at distance 5 from ( x, y) are ( x ± 4, y ± 3),
( x ± 3; y ± 4), ( x ± 5, y), ( x, y ± 5). In each case, the sum of coordinates has
the parity different from that of ( x, y). Therefore they are colored differently
from ( x, y).
O3. (Juniors.) A hiking club wants to hike around a lake along an exactly
circular route. On the shoreline they determine two points, which are the
most distant from each other, and start to walk along the circle, which has
these two points as the endpoints of its diameter. Can they be sure that,
independent of the shape of the lake, they do not have to swim across the
lake on any part of their route?
Answer: No.
Solution. Suppose the shape of the lake is an equilateral triangle. Then
the two points which are the most distant from each other are two vertices
of the triangle. The circle, which has these two points as the endpoints of
its diameter, does not cover the whole triangle,
√
because the distance of the
3
third vertex from the center of the circle is 2 of the length of the side of the
1
triangle, but the radius of the circle is only 2 of this length.
O4. (Juniors.) Two circles c and c′ with centers O and O′ lie completely
outside each other. Points A, B, and C lie on the circle c and points A′ ,
4
C′
C
P O O′
A
A′
c B
c′ B′
Fig. 1
B′ , and C ′ lie on the circle c′ so that segment AB k A′ B′ , BC k B′ C ′ , and

∠ ABC = ∠ A′ B′ C ′ . The lines AA′ , BB′ , and CC ′ are all different and inter-
sect in one point P, which does not coincide with any of the vertices of the
triangles ABC or A′ B′ C ′ . Prove that ∠ AOB = ∠ A′ O′ B′ .
Solution. The triangles ABP and A′ B′ P are similar, because their corre-
| AB | | BP |
sponding sides are parallel (Fig. 1). Hence | A′ B′ | = | B′ P| . Likewise the tri-
| BC| | BP | | AB | | BC|
angles BCP and B′ C ′ P are similar, hence | B′ C′ | = | B′ P| . Thus | A′ B′ | = | B′ C| ,
and since ∠ ABC = ∠ A′ B′ C ′ , the triangles ABC and A′ B′ C ′ are also similar.
From the equality of the angles ACB and A′ C ′ B′ the equality of the central
angles AOB and A′ O′ B′ now follows.
−→ −−→
Remark. Figure 1 corresponds to the case when the vectors AB and A′ B′
have the same direction. If they have the opposite directions, then the figure
is different (the intersection point P lies on the segments AA′ , BB′ and CC ′ )
but the argument is still correct.
O5.h (Juniors.)
i Let n be a positive integer and a1 , . . . , a2n be real numbers
1 1
in − 2 , 2 . Leaving out any one of the numbers, the sum of the remaining
2n − 1 numbers is always an integer. Prove that a1 = . . . = a2n .
Solution. Assume that there exist ai and a j which are not equal. Let
S = a1 + . . . + a2n . Since S − ai and S − a j are integers, their difference
(S − ai ) − (S − a j ) = a j − ai is also an integer. Since a j − ai 6= 0, and they
h i
belong to − 21 , 21 , their difference can be only ±1, this happens when ai
and a j are 12 and − 21 in any order. Let ak be any of the given numbers. Since
(S − ai ) − (S − ak ) = ak − ai is an integer, ak must also be either 12 or − 21 .
Hence all numbers ai are either 12 or − 12 . It follows that the sum of any two
numbers ai and a j is an integer. As we have an even number of them, the
sum of all the numbers S is also an integer. But then S − ai = S ± 21 cannot
be an integer, a contradiction. Therefore all numbers ai must be equal.
5
O6. (Juniors.) Is it possible that the perimeter of a triangle whose side
lengths are integers, is divisible by the double of the longest side length?
Answer: no.
Solution. Let the side lenghts of the triangle be integers a, b, c. Without
loss of generality we may assume that c ≥ a and c ≥ b. Suppose that the
perimeter of the triangle a + b + c is divisible by double of the longest side
length 2c. Since 0 < a + b + c ≤ 3c < 2 · 2c, the perimeter a + b + c can be
divisible by 2c only in the case when a + b + c = 2c. But then a + b = c,
which violates the triangle inequality a + b > c.
O7. (Seniors.) For any positive integer n let an be the largest power of 2 that
divides n (e.g. a2011 = 1, a2012 = 4). Prove that for any positive integers i
and j with i < j, the sum a1 + a 1 + . . . + a1 is a fractional number.
i i +1 j
Solution. First prove that the largest power of 2 among the numbers ai ,
ai+1 , . . . , a j is unique. Let 2s be the largest of the numbers ai , ai+1 , . . . , a j .
If there were k and l with i ≤ k < l ≤ j such that ak = al = 2s , then they
must be of the form k = 2s u and l = 2s v, where u and v are odd numbers.
Since k < l, we have u < v and u + 1 < v. Since u + 1 is even, the num-
ber m = 2s (u + 1) has a divisor 2s+1, and k < m < l, which contradicts
the choice of s. Thus the largest power of 2 appears only once among the
numbers ai , ai+1 , . . . , a j . Converting the fractions to the common denomi-
nator the fraction with the largest denominator gives 1 in the numerator, all
others give a positive power of 2, i.e. an even number. Consequently the
numerator is odd and cannot cancel with the denominator.
O8. (Seniors.) Let a be a real number, 0 ≤ a ≤ 1. Prove that for any

nonnegative integer n the inequality (n + 1) a ≤ n + an+1 holds.
Solution 1. The inequality is equivalent to the inequality na − n ≤ an+1 −
a, or the inequality n( a − 1) ≤ a( a − 1)( an−1 + an−2 + . . . + 1). If a = 1,
then the inequality obviously holds. If a < 1 then a − 1 < 0, and dividing
both sides of the inequality by a − 1 we get an equivalent inequality n ≥
a( an−1 + an−2 + . . . + 1), or n ≥ an + an−1 + . . . + a. Since a < 1, in the last
sum all terms are less than 1, hence the sum does not exceed n.
Solution 2. If n = 0 then the inequality is a ≤ a, which obviously holds.
Suppose that the inequality holds for n = k and prove that it then holds
for n = k + 1 as well. From the inequality (k + 1) a ≤ k + ak+1 we get the
inequality (k + 2) a ≤ k + 1 + ak+2 by adding the inequality a ≤ 1 + ak+2 −
ak+1 . The last inequality is equivalent to the inequality (1 − a)(1 − ak+1 ) ≥
0, which obviously holds, because both factors are nonnegative.
Solution 3. If a = 0, then the inequality is 0 ≤ n, which obviously holds.
If a > 0, then AM-GM gives
1 + . . . + 1 + a n +1 √
n +1
≥ 1 · . . . · 1 · a n +1 ,
n+1
which is obviously equivalent with the original inequality.
6
A L C A
A
O L
M M L
K O M
O
B B K C B K C
Fig. 2 Fig. 3 Fig. 4
Remark 1. There are also solutions via derivative.

Remark 2. Actually the inequality holds for all nonnegative real numbers
a and n.
O9. (Seniors.) Let ABC be a triangle with median AK. Let O be the circum-
center of the triangle ABK.
a) Prove that if O lies on a midline of the triangle ABC, but does not
coincide with its endpoints, then ABC is a right triangle.
b) Is the statement still true if O can coincide with an endpoint of the
midsegment?
Solution. a) Let L and M be the midpoints of the sides CA and AB,
respectively. If O lies on the segment KM (Fig. 2), then the segment KM
and the perpendicular bisector of AB have two different common points O
and M, hence KM is the perpendicular bisector of AB. Since KM is parallel
to AC and is perpendicular to AB, the angle at vertex A must be right. If
O lies on the segment LM (Fig. 3), then we get similarly that the angle at
vertex B must be right. If O lies on the segment KL (Fig. 4), then on one
hand ∠ ABC is acute, because OK and MB are perpendicular to MO, the
perpendicular bisector of AB, and |OK | < | LK | = | MB|. On the other
hand, ∠ ABK must be obtuse, since the circumcenter O of the triangle ABK
lies outside of the triangle, a contradiction. Thus this case is not possible.
b) If the triangle ABC is equilateral, then the median AK is also the alti-
tude and ABK is a right triangle with the hypotenuse AB. The circumcenter
O of the last triangle is the midpoint of AB, i.e. an endpoint of a midseg-
ment of the triangle, but ABC is not a right triangle.
O10. (Seniors.) Determine the least number of the dark squares

which remain visible if one covers the n × n squared paper with
2 × 2 squares shown on the figure (they can be turned) so that all
of the squares on the paper are covered at least once?
Answer: n.
Solution. No matter how we cover the squared paper, there must be at
least one dark square in each column, because on the last figure we put
on this column one dark square always remains visible. Hence there must
always be at least n dark squares visible. On the other hand, we can leave
7
exactly n dark squares visible by covering the paper as follows. First put
one figure on the lower left corner of the paper. Then put two figures so
that they cover the dark squares of the first figure and their dark squares
form a diagonal of length 3. Then add 3 figures so that they cover the dark
squares of the previous figures and their dark squares form a diagonal of
length 4. Repeat this until we have a diagonal of length n from one corner
of the paper to the opposite corner. To cover all the squares repeat the same
steps starting from the upper right corner of the paper.
O11. (Seniors.) The teacher drew a 3 × 3 table in Juku’s exercise book and
wrote a number in every position of the table. Then he gave Juku the fol-
lowing task.
1) Turn the next page and draw a similar table. Write in the first row
the numbers obtained by subtracting the numbers in the third row of
the corresponding column from the numbers in the second row of the
corresponding column in the previous table. Similarly, the numbers
in the second and third row are obtained as differences of the third
and the first, and the first and the second row.
2) Turn the next page and draw a new table. Write in the first column
the numbers obtained by subtracting the numbers in the third column
from the numbers in the second column in the corresponding row in
the previous table. Similarly, the numbers in the second and third
column are obtained as differences of the third and the first, and the
first and the second column.
Repeat in turns steps 1 and 2 until you reach a table where all the numbers
are zeroes. Juku has reached the end of the third page and has not yet
reached the table with all zeroes in it. Prove that his task never ends.
Solution. First note that after step 1 we get a table where the column
sums of the table are 0, and after step 2 we get a table where the row sums
of the table are 0. Suppose that after some steps we reach the table with all
zeroes in it. By symmetry we can consider the case where we get this table
after step 2. Then the table on the previous step was
a a a
b b b
c c c
where at least one of the numbers a, b, c is not zero. This table was obtained
after step 1, hence a + b + c = 0. The table on the previous step was
d e f
d−c e−c f −c
d+b g+b f +b
This was also written by Juku, because he computed at least 2 tables. Since
this table was obtained after step 2, we must have d + e + f = d − c + e −
8
c + f − c = d + b + g + b + f + b = 0. From the first equality it follows that
c = 0, from the equality of the first and third expression b = 0, and since
a + b + c = 0, we have a = 0, which contradicts the assumption that at least
one of the numbers a, b, c is not zero.
O12. (Seniors.) Prove that for any positive integer n the sum of the first n
primes is greater than n2 .
Solution. First notice that the n-th prime p n satisfies the inequality pn ≥
2n − 1. Indeed, the claim holds for the first prime p1 = 2. Since all other
primes are odd and there is exactly n − 1 odd numbers between 2 and 2n,
there are at most n prime numbers less or equal to 2n − 1, hence pn ≥ 2n −
1. Now consider the sum of the n first primes P = p1 + p2 + . . . + pn . Since
pk ≥ 2k − 1 for any k, and additionally p1 = 2 > 1, the sum P is strictly
greater than the sum of n first odd numbers S = 1 + 3 + . . . + 2n − 1 =
= (12 − 02 ) + (22 − 12 ) + . . . + (n2 − (n − 1)2 ) = n2 . So P > S = n2 .
O13. (Seniors.) Find all triples ( a, b, c) of positive integers such that

abc + bca + c ab = 3abc.
Answer: (1, 1, 1), (1, 2, 3), (1, 3, 2), (2, 1, 3), (3, 1, 2), (2, 3, 1), (3, 2, 1).
Solution. First assume a ≥ 2, b ≥ 2, c ≥ 2. W.l.o.g., let c be the greatest
among the three numbers. Then abc + bca + c ab ≥ a4 + b4 + c4 > b4 + c4 ≥
2b2 c2 = 2b · c · bc > 3 · a · bc. Thus there are no solutions in this case.
It remains to study triples that contain 1. W.l.o.g., let a = 1. The equation
reduces to 1 + bc + cb = 3bc. Assume b ≥ 3, c ≥ 3. W.l.o.g., c ≥ b, leading
to 1 + bc + cb ≥ 1 + b3 + c3 > c3 ≥ 3 · b · c. Thus there are no solutions
in this case either. Now assume b ≥ 2, c ≥ 2 and one of the numbers is
2. W.l.o.g. let b = 2. The equation reduces to 1 + 2c + c2 = 6c which
p be interpreted as a quadratic equation w.r.t. c that leads to c = 3 ±
can
9 − (2c + 1). Hence 8 − 2c is a perfect square. The only candidates for
this are 4 and 0 that give c = 2 and c = 3, respectively, but c = 2 leads
to contradiction (the above formula would give c = 1 or c = 5). The case
c = 3 gives the solution (1, 2, 3) of the original equation. By symmetry,
also (1, 3, 2), (2, 1, 3), (2, 3, 1), (3, 1, 2), (3, 2, 1) are solutions. If one of the
numbers b and c is 1 then, w.l.o.g., b = 1. The equation reduces to 1 + 1 +
c = 3c, whence c = 1. This gives the trivial solution (1, 1, 1).
O14. (Seniors.) Let ABC be an acute triangle and D an interior point of its
side AC. We call a side of the triangle ABD friendly, if the excircle of ABD
tangent to that side has its center on the circumcircle of ABC. Prove that
there are exactly two friendly sides of ABD if and only if | BD | = | DC |.
Solution. Let E, F and G be the centers of excircles touching BD, AD and
AB respectively, and let ω be the circumcircle of ABC (see Fig. 5). To prove
the assertion of the problem, we will show that F and G cannot both lie on
ω and that E ∈ ω ⇐⇒ | BD | = | DC | ⇐⇒ F ∈ ω.
9
As AF and AG are bisectors of the two E
complementary angles of BAD, the point G B
A lies on the segment FG. Thus, only one
of the rays AF and AG can cut the circle ω
again, and therefore only one of F and G
can lie on ω.
To show that E ∈ ω iff | BD | = | DC |, A D C
we first note that E lies on ω iff ∠CAE = F ω
∠CBE. Since ∠CAE = 12 ∠BAD = 21 (π −
∠ ADB − ∠ ABD ) = 12 (π − (π − 2∠BDE) − Fig. 5
π π
(π − 2∠DBE)) = ∠BDE + ∠DBE − 2 = 2 − ∠BED, we see that E ∈ ω
is equivalent to ∠CBE = π2 − ∠ BED, i.e BC and DE being perpendicular.
Since DE is the bisector of BDC, this occurs iff | BD | = | DC |.
It remains to show that | BD | = | DC | iff F ∈ ω. Point F lies on ω iff
∠ AFB = ∠BCD. Using the fact that ∠BAF = ∠BAD + 21 (π − ∠BAD ) =
1 π −∠ ABD −∠BAD
2 ( π + ∠ BAD ), we get ∠ AFB = π − ∠ ABF − ∠ BAF = 2 =
∠ ADB ∠BCD +∠CBD
2 = 2 . Thus ∠ AFB = ∠ BCD is equivalent to ∠ BCD =
∠CBD, i.e., | BD | = | DC |.
Remark. The claim holds in the case of right or obtuse triangle, too. The
problem with the above proof is that if ∠ ABC > 90◦ then point E may fall
inside triangle ABC, whence equality ∠CAE = ∠CBE is no more equiva-
lent to E ∈ ω. Nevertheless, one can show that if | BD | = | DC |, then E must
lie outside triangle ABC, extending the validity of the claim to the obtuse
case.
O15. (Seniors.) Let k be a positive integer. Determine the

largest number of snakes, consisting of four squares (see fig-
ure), which can be placed on a (2k + 1) × (2k + 1) chessboard
so that the snakes neither overlap nor stick out across the
edges of the chessboard. The snakes can be turned and reflected.
Answer: k2 .
Solution. First show that k2 snakes can be placed on a (2k + 1) × (2k + 1)
chessboard. Divide the chessboard into strips of width 2 (one strip of width
1 remains). On any strip we can place k snakes, one after another; so on k
strips, it is possible to place k2 snakes.
It remains to prove that one can not place more than k2 snakes on the
chessboard. Write numbers 0, 1, 0, 1, . . . , 0 in the odd rows, and numbers 2,
3, 2, 3, . . . , 2 in the even rows. Notice that no matter how we place the snake
on the board, it always covers numbers 0, 1, 2 and 3. Since all numbers 3
are in the squares with even row and column numbers, there is exactly k2
of them, hence there can be at most k2 snakes.
10
F1. (Grade 9.) Integers a, b, c are such that a + b + c is divisible by 6, and
a2 + b2 + c2 is divisible by 36. Does it imply that a3 + b3 + c3 is divisible by
a) 8; b) 27?
Solution. a) As the sum of a, b, and c is divisible 6, and is therefore even,
there must be either 0 or 2 odd numbers among the three. If we had 2 odd
numbers, the sum of the squares a2 + b2 + c2 would give a remainder of
0 + 1 + 1 = 2 when dividing by 4. But this is not possible, since the sum is
divisible by 36, and therefore also by 4. So, all the numbers a, b, c are even.
Hence, all the numbers a3 , b3 , c3 are divisible by 8, and so is their sum.
b) If a = 8, b = c = 2, then all of the premises are fulfilled: 8 + 2 + 2 = 12
is divisible by 6 and 82 + 22 + 22 = 72 is divisible by 36. But 83 + 23 + 23 =
528 is not divisible by 9, and therefore, it is not divisible by 27.
F2. (Grade 9.) Let ABC be an isosceles triangle with | AB| = | AC |. The
bisector of angle ABC meets the side AC at the point D.
a) Is the triangle ABD isosceles whenever the triangle BCD is isosceles?
b) Is the triangle BCD isosceles whenever the triangle ABD is isosceles?
Solution. a) Denote ∠ BAC = α and ∠ ABC = C
∠ ACB = β (Fig. 6). Assume that the triangle BCD β
is isosceles. If |CB| = |CD |, then the angles ∠CBD, D
∠CDB and ∠BCD would be 2β , β2 and β respectively, β
β β 2
which implies 2 · + β =
2 180◦ giving β = 90◦ .This α 2
is impossible, since the triangle ABC has two angles A B
β
equal to β. If | DB| = | DC |, then we would get 2 = Fig. 6
β, which is also impossible. This leaves the only
option | BC | = | BD |. Then, the triangles ABC and BCD are similar, since all
corresponding angles are the same. Therefore ∠ DBA = ∠ DBC = ∠ BAC =
∠BAD, showing that the triangle ABD is isosceles with
C | DA| = | DB|.
b) If the angles of the triangle ABC are 73 · 180◦ , 27 ·
180◦ , 27 · 180◦ (Fig. 7), then ∠ ADB = 180◦ − ∠ BAD −
D
∠ ABD = 180◦ − 37 · 180◦ − 71 · 180◦ = 37 · 180◦ = ∠BAD,
which shows that the triangle ABD is isosceles with
| BA| = | BD |. At the same time, the angles in the trian-
A B
gle BCD are 17 · 180◦ , 72 · 180◦ , 74 · 180◦ , which are pair-
Fig. 7 wise different, so the triangle BCD is not isosceles.
11
F3. (Grade 9.) An equilateral triangle with side length 3 is
divided into 9 equilateral triangles with side length 1. An in-
teger from 1 to 10 is written into every point that is a vertex
of a small triangle (colored vertices on the figure), such that
all numbers are written exactly once. For every small trian-
gle, the sum of the numbers in its three vertices is written inside it. Prove
that at least three of those sums are greater than 11.
Solution. In a triangle, which has 10 at one vertex, the sum is at least
13. If 10 is not at one of the vertices of the large triangle, the number of
triangles with sum greater than 12 is at least 3 and the problem is solved. If
10 is at the vertex of the large triangle, then look, where is the number 9. If
9 does not lie at a vertex of the large triangle, then there are at least 3 small
triangles, with a sum of at least 12, and the problem is solved. If 9 is at a
vertex a the large triangle, then look, where is the number 8. If 8 does not
lie at the vertex of a large triangle, then it is at a vertex of at least 3 small
triangles. In at least two of them the sum is at least 12, and as at most one
of these can overlap with one of the triangles found earlier, the problem is
solved. If 8 is at a vertex of the large triangle, then either the sum in that
triangle is at least 12, meaning the problem is solved, or the other vertices
in the triangle have numbers 1 and 2. In the last case, the numbers 3, 4
and 5 are the smallest numbers whose positions are not set, but these give
a sum of 12. So, we can simply choose any triangle who vertices we have
not looked at yet as our third triangle (there are three of these triangles).
F4. (Grade 9.) Jüri wishes to draw n circles and any number of lines on
the plane such that all the lines meet at one point, and for every two circles
there exist two lines that touch both of these circles.
a) Is it possible for Jüri to solve this problem for any n ≥ 2?
b) For which natural numbers n is it possible to solve this problem if in
addition all the circles must have the same radius?
Answer: a) yes; b) n ≤ 4.
Solution. a) Jüri can draw two lines and draw any number of circles such
that they touch both of the lines (Fig. 8).
b) Assume that Jüri has solved the problem for some n, where n > 1.
Fig. 8 Fig. 9
12
Let O be the intersection point of all the lines. Look at any circle c. From
the premises of the problem we see that the circle c touches two of the lines
drawn by Jüri, which we call k and l. But any one circle can only touch up
to two lines drawn from one point. So, the circle c does not have any more
lines touching it. If c′ is any other circle drawn by Jüri, then the common
tangents of c and c′ can only be k and l. Therefore, k and l are the common
tangents of all the circles. Two lines divide the plane into four sectors, inside
each can be only one circle with the previously set radius (Fig. 9). So, for
n > 4 the problem has no solution but for n ≤ 4, it obviously has.
F5. (Grade 10.) Find all pairs (n, m) of positive integers such that the arith-
metic and geometric means of m and n are different two-digit numbers con-
sisting of the same digits.
Answer: (32, 98), (98, 32).
Solution. Let 10a + b be the arithmetic mean of the given numbers, where
a and b are decimal digits. Let 10a + b + x and 10a + b − x be the numbers
we are searching for. Then, by the premises
q
(10a + b + x )(10a + b − x ) = 10b + a,
which after squaring and simplifying gives x2 = 99( a2 − b2 ). So, x2 is
divisible by 99, implying x2 is divisible by 3 and 11. Since 3 and 11 are
primes, x itself is divisible by 3 and 11, and therefore by 33. Denoting x =
33z, we get
99 · 11z2 x2
11z2 = = = a2 − b2 = ( a + b)( a − b),
99 99
from which we see, that the product ( a + b)( a − b) is divisible by 11. Since
11 is a prime, either a + b or a − b is divisible by 11. Since a − b 6= 0 and
a and b are single-digit numbers, we must have a + b = 11. Therefore
a − b = z2 . Since a − b and a + b are either both odd or both even, z must be
odd. So, z = 1, since z ≥ 3 implies x ≥ 99, but 10a + b − x must be positive.
So, a = 6, b = 5, x = 33, and the corresponding pair is (98, 32).
F6. (Grade 10.) We say that two real numbers r and s are close if |r − s| =
10u for some integer u. Let y = ax + b be a linear function, for which there
exist close numbers x1 and x2 so that the corresponding y1 and y2 are also
close. Prove that for any close numbers x1′ and x2′ , the corresponding y1′ and
y2′ are also close.
Solution. From the premises we get | x1 − x2 | = 10u and |y1 − y2 | =
|( ax1 + b) − ( ax2 + b)| = 10v for some integers u, v. So,
10v = |( ax1 + b) − ( ax2 + b)| = | a( x1 − x2 )| = | a| · | x1 − x2 | = | a| · 10u ,
v
10
which gives | a| = 10 v − u . Let x ′ , x ′ be any close real numbers, | x ′ −
u = 10 1 2 1
x2 | = 10 . Then |y1 − y2 | = |( ax1′ + b) − ( ax2′ + b)| = | a( x1′ − x2′ )| =
′ w ′ ′
13
| a| · | x1′ − x2′ | = 10v−u · 10w = 10w+v−u. Since u, v, w are integers, w + v − u
is also an integer, which shows that y1′ , y2′ are close.
F7. (Grade 10.) Let ABC be a triangle on the plane. The angle bisector from
the vertex A meets the side BC at P, and the median from the vertex B meets
the side AC at M. The lines AB and MP meet at the point K. Prove that if
| PC|
| BP |
= 2, then AP and CK are perpendicular.
Solution 1. Let K ′ be a point on the ray AB, such
that B is the midpoint of the line segment AK ′ (Fig. C
10). Then CB is the median of the triangle ACK . ′
As P divides this line segment in the ratio 2 : 1, P
must be the centroid of the triangle ACK ′ . So, K ′ M,
M
which is also a median of the triangle ACK ′ , must
pass through the point P. Therefore, K = K ′ . So, B is P
the midpoint of the line segment AK. As AP passes
through the point P, AP is also a median of the trian- A B K′
gle ACK. By the premises it is also an angle bisector.
So, the triangle ACK is isosceles with | AC | = | AK | Fig. 10
and AP is its height. Therefore, AP ⊥ CK.
Solution 2. As in solution 1, we show that B is the midpoint of AK. So
BM joins the midpoints of sides in the triangle ACK and therefore BM k
CK. An the angle bisector divides the opposite side in the same ratio as
| AC| | PC|
the corresponding sides, so | AB| = | BP| = 2. As M is the midpoint of AC,
we have | AB| = | AM |. An angle bisector, drawn from the vertex opposite
the base in an isosceles triangle, is also the height in that triangle, giving
AP ⊥ BM. It follows that AP ⊥ CK.
F8. (Grade 10.) An equilateral triangle with side length 3 is

divided into 9 equilateral triangles with side length 1. An in-
teger from 1 to 10 is written into every point that is a vertex
of a small triangle (colored vertices on the figure), such that
all numbers are written exactly once. For every small trian-
gle, the sum of the numbers in the three vertices is written inside it. Prove
that there exist three small triangles such that the sum of the numbers inside
them is at least 48.
Solution 1. Any three small triangles, from which no two have common
vertices, take up nine of the ten numbers written into the vertices of the
small triangles. So, the sum of the numbers inside those small triangles
is 55 − a, where a is the number at the last vertex. Now it is sufficient to
prove that we can choose the three small triangles, from which no two have
common vertices, in four different ways, always leaving a different vertex
out. So, in at least one case, the number at the last vertex is at most 7, and
the sum of the numbers in the three chosen triangles is at least 55 − 7 = 48.
14
Fig. 11 Fig. 12 Fig. 13 Fig. 14
Indeed, the three triangles can be chosen so that they leave uncovered the
central number (Fig. 11) or one of the corner numbers (Fig. 12, 13, 14).
Solution 2. Let m be the number in the center of the large triangle. Then,
when adding the sums of the three corner triangles we get the sum of all
the numbers from 1 to 10, except m, so the sum of the corner triangles is
55 − m. If m ≤ 7, then the sum is at least 48.
In the rest of the cases, consider any three triangles around the center
point, such that no two of them share a side. Adding the numbers in them,
we get the sum of all the numbers from 1 to 10, except the three numbers
in the corners, while we add the center number three times. So the sum of
those triangles is at least 21 + 3m. If m ≥ 9, then the sum is at least 48.
This leaves the case m = 8. The sum of the numbers in the triangles in
the corners is at least 55 − 8 = 47, so at least one of them contains a sum
that is at least 16. If none of the triangles contains a sum 17 or greater, the
number 16 must occur in two different triangles. The sum of the numbers
in any three triangles around the center point, chosen like above, is at least
21 + 3 · 8 = 45. So, in both triples at least one of the triangles contains a
sum of at least 15, and since triangles sharing an edge cannot contain the
same sum as the corresponding sums differ by exactly one term, at least
one of the six triangles around the center point contains a sum of at least 16.
Thus we can pick the three desired triangles from among either one corner
triangle and two central triangles or two corner triangle and one central
triangle that contain the largest numbers.
F9. (Grade 10.) Find all triples of positive integers ( x, y, z), for which
x · y! + 2y · x! = z!.
Answer: (2, 1, 3) and (n, n + 1, n + 2) for every positive integer n.
Solution. Since the left-hand side is greater than both x! and y!, obviously
z > x and z > y. So, both sides of the equation are divisible by both x! and
y!. Therefore, x · y! is divisible by x!, which means that y! is divisible by
( x − 1)!, giving y ≥ x − 1. Analogously, 2y · x! is divisible by y!, meaning
2 · x! is divisible by (y − 1)!. The case x = 1, y = 3 is not a solution, the case
x > 1 gives 2 · x! < ( x + 1)!, which implies x ≥ y − 1. This leaves us to look
through the cases −1 ≤ y − x ≤ 1.
• If y = x − 1, then the equation simplifies to (2x − 1) · x! = z!. As
( x + 1)( x + 2) > 2x − 1, we have 2x − 1 = x + 1 and z = x + 1. This
gives the solution x = 2, y = 1, z = 3.
15
• If y = x, the equation simplifies to 3x · x! = z!. As ( x + 1)( x + 2) > 3x,
we have 3x = x + 1, but this does not give integer solutions.
• If y = x + 1, the equation simplifies to ( x 2 + 3x + 2) · x! = z! or ( x +
2)! = z!. From here we get a family of solutions x = n, y = n + 1,
z = n + 2.
F10. (Grade 11.) In his last research, professor P was concentrating on nat-
ural numbers with a certain property. It is known that whenever a natural
number x has this property, all multiples of x also have this property. Let
a1 , . . . , an be positive integers such that all their divisors that are greater
than one have the property professor P studied. Is it true that all divisors
greater than one of the product a1 . . . an definitely have this property?
Answer: yes.
Solution 1. Let k > 1 be any divisor of the product a1 . . . an . Then k has
a prime divisor p, which is also a divisor of the product a1 . . . an . As p is a
prime, there exists i, such that p is a divisor of ai . As all the divisors of ai
greater than 1 have the property, p also has this property. By the premise,
all the multiples of p have the property, so k has the property.
Solution 2. Let k > 1 be any divisor of the product a1 . . . an . If k were
relatively prime to all ai , it would be relatively prime to the product a1 . . . an ,
but gcd(k, a1 . . . an ) = k > 1. Hence gcd(k, ai ) > 1 for some ai . As a divisor
of ai , the number gcd(k, ai ) has the property studied by professor P. As a
multiple of gcd(k, ai ), also k has the same property.
F11. (Grade 11.) a) Find all positive integers n, such that the sum of all
integers from 1 to n + 1 can be represented as the sum of n consecutive
integers.
b) Find all positive integers n, for which there exists an integer a, such
that the sum of the integers from a to a + n is equal to the sum of the integers
from a + n + 1 to a + 2n.
Answer: a) 1; b) all positive integers.
Solution 1. a) Clearly the sum of the first two positive integers can be
represented as the sum of one positive integer. Now, let n ≥ 2 and let us
show that the sum of the n + 1 first positive integers cannot be represented
as a sum of n consecutive integers. Indeed, on one hand 1 + 2 + . . . + n +
(n + 1) > 2 + 3 + . . . + n + (n + 1), on the other hand 1 + 2 + . . . + n + (n +
1) < 1 + 2 + 3 + . . . + n + ( n + 1) + 1 = 3 + . . . + n + ( n + 1) + 2 + 2 ≤
3 + . . . + n + (n + 1) + (n + 2). So the sum of n + 1 first positive integers
1 + . . . + (n + 1) lies between 2 + . . . + (n + 1) and 3 + . . . + (n + 2), which
both are consecutive sums of n consecutive integers. So, the number 1 +
. . . + (n + 1) is not a sum of n consecutive integers.
b) Let n be any positive integer. To solve the problem, it suffices to see
that n2 + (n2 + 1) + . . . + (n2 + n) = n2 · (n + 1) + (1 + . . . + n) = n · (n2 +
n ) + ( 1 + . . . + n ) = ( n2 + n + 1 ) + . . . + ( n2 + n + n ) .
16
Solution 2. We use the formula for the sum of arithmetic progression.
a) If n is odd, then the sum of n consecutive integers is divisible by n.
So, if the number 1 + 2 + . . . + (n + 1) = n+ 1
2 · ( n + 2) was the sum of n
consecutive integers, it would be divisible by n+ 1
2 and by n. As n + 1 and
n are relatively prime, the same clearly holds for n+ 1
2 and n. Therefore,
n +1 n +1
2 · ( n + 2) should be divisible by 2 · n, meaning that n + 2 should be
divisible by n.
If n is even, the sum of n consecutive integers is divisible by n2 . So, if
1 + 2 + . . . + ( n + 1) = ( n + 1) · n+ 2
2 was the sum of n consecutive integers,
it would be divisible by both n + 1 and n2 . As n + 1 and n are relatively
prime, also n + 1 and n2 are relatively prime. Thus, (n + 1) · n+ 2
2 should be
n
divisible by (n + 1) · 2 , implying that n + 2 is divisible by n.
So, in all cases n + 2 is divisible by n, which is equivalent to saying 2 is
divisible by n. So, n = 1 or n = 2. Clearly 1 + 2 is the sum of one integer,
but 1 + 2 + 3 = 6, being an even number, cannot be represented as the sum
of two consecutive integers.
b) If a is the first of the two consecutive integers, then the problem can be
represented as the equation ( a + a + n)(n + 1)/2 = ( a + n + 1 + a + 2n)n/2.
By simplifying we see that it is equivalent to a = n2 . This means that the
sum of n + 1 consecutive integers, first of which is n2 , is the sum of the next
n consecutive integers. So, the desired numbers exist for every n.
F12. (Grade 11.) The sides AB and AC of the triangle ABC touch the circle
c respectively at points B′ and C ′ . The center L of the circle c lies on the
side BC. The circumcenter O of triangle ABC lies on the shorter arc B′ C ′ of
the circle c. Prove that the circumcircle of ABC and the circle c meet at two
points.
Solution. Let r be the circumradius of ABC,
let s be the radius of c and α = ∠ BAC (Fig. 15). C
By tangency, | AB′ | = | AC ′ |. Thus ∠C ′ B′ A = C ′ c
∠B′ C ′ A = π2 − α2 whence, by property of in-
L
scribed angle, ∠ B′ OC ′ = π − π2 − α2 = π2 + α2 .

′ ′
Clearly ∠ B OC > ∠ BOC = 2α, leading to 2 + π
α O
α π A
2 > 2α. Hence α < 3 . Now let K be the mid- B′
B
point of side BC. From the right triangle KOC,
one gets |KO| = |OC | cos ∠KOC = r cos α. By
the inequality obtained above, cos α > cos π3 = Fig. 15
1
2 . On the other hand, | KO | ≤ | LO | = s, leading
to 12 r < r cos α = |KO| ≤ s or r < 2s. As c passes through the circumcenter
of ABC, this inequality shows that these circles must intersect.
Remark. This problem, proposed by Estonia, appeared in the IMO 2011
shortlist as G1.
17
F13. (Grade 11.) A finite grid is covered with 1 × 2 cards in such a way that
the edges of the cards match with the lines of the grid, no card lies over the
edge of the grid, and every square is covered by exactly two cards. Prove
that one can remove some of the cards in such a way that every square will
be covered by exactly one card.
Solution. Choose any square covered by two cards, and choose one of
these cards. Move that card to a neighbouring square, and choose the other
card that is covering that square. From there we move to the next square,
etc., until we return to the first square. We cannot return to any other square
visited previously, since in all squares except the first one, both cards have
been chosen already. If we color the rectangular grid like a chessboard, then
after an odd number of moves, we reach a square with the opposite color,
and after an even number of moves, we reach a square with the same color.
Therefore, the number of chosen cards is even. So, we can remove every
second chosen card. All of the remaining squares we passed through will
be covered by exactly one card. If after this, there are still squares that are
covered by two cards, we repeat the process with a new randomly chosen
square which is covered by two cards. We can never move from a square
covered by two cards to a square covered by exactly one card, since all the
squares covered by exactly one card were previously connected to squares
now covered by exactly one card. So, after a finite number of steps we can
find a new cycle, from which we can remove every second card. We repeat,
until all squares are covered by exactly one card.
F14. (Grade 11.) There are 2012 points marked in a square with side length
11. Prove that one can choose an equilateral triangle with side length 12
which covers at least 671 points.
Solution. Place two equilateral triangles with side
lengths 12 on the square in such a way that both have
one vertex lie on the side of the square and the oppo-
site sides of these vertices partially coincide with the
other side of the square and with each other (Fig. 16).
The area common to both triangles forms an equi- Fig. 16
lateral triangle of side length 1. Position the third
equilateral triangle with side length 12 between the two triangles, turned
180◦ , such that the lowermost vertex of that triangle coincides with the up-
permost vertex of the small triangle. To show that the square is fully cov-
ered by these triangles, we must show that the sum of the heights of the
large
√
and the small triangle is at least 11, which is equaivalent to showing
3
√
2 · (12 + 1) > 11. As we can simplify this equation to 13 3 > 22 and
3 · 169 > 484, we see that it holds. Therefore at least a third of the 2012
points or at least 671 points lie in one of the three chosen triangles.
18
F15. (Grade 12.) Find all pairs ( x, y) of positive integers such that
1 249 1 1
+ + 2 = .
x2 xy y 2012
Answer: (503, 1006), (1006, 503).
Solution 1. Let gcd ( x, y) = d and x = ad, y = bd. Then the equation can
2 + b2
be written as a +a249ab
2 b2 d 2
1
= 2012 or
a2 b2 d2 = 2012( a2 + 249ab + b2 ).
As a and b are relatively prime, a2 and b2 are both relatively prime to a2 +
249ab + b2 and therefore both they must be divisors of 2012. As 2012 = 22 ·
503 and 503 is a prime, the possible cases are ( a, b) = (1, 1), ( a, b) = (1, 2),
( a, b) = (2, 1). If we substitute ( a, b) = (1, 1) into the last equation, we
get d2 = 2012 · 251, which is not solvable in integers. The other two cases
give 4d2 = 2012 · 503, from which d = 503. This leads to the solutions
( x, y) = (503, 1006) and ( x, y) = (1006, 503).
Solution 2. Multiplying both of the sides by 2012x 2 y2 , we get
2012x2 + 249 · 2012xy + 2012y2 = x2 y2 .
From the left-hand side we see that both sides of the equation must be di-
visible by 503. As 503 is a prime, one of the numbers x and y must be
divisible by 503. So x2 or y2 is divisible by 5032, giving that both sides
of the equation are divisible by 5032. If x is divisible by 503, the sum-
mands 2012x2 and 249 · 2012xy on the left-hand side are divisible by 5032,
meaning that 2012y2 is divisible by 5032. Therefore y is divisible by 503.
Analogously, we get that if y is divisible by 503, then x is also divisible by
503. Consequently, both x and y are divisible by 503. Denote x = 503a,
y = 503b. Then the equation, after dividing both sides by 5033 , simplyfies
to 4a2 + 996ab + 4b2 = 503a2 b2 . Assume a ≥ b. If b ≥ 2, then 503a2 b2 ≥
503a2 · 2b = 1006a2 b = 4a2 b + 4a2 b + 998a2 b > 4a2 + 4b2 + 996ab, so the
last equation cannot hold. Therefore b = 1. Now we get a quadratic equa-
tion 499a2 − 996a − 4 = 0 with respect to a, whose only positive solution is
a = 2. From here we obtain the solution (1006, 503) to our original equa-
tion. The case b ≥ a is symmetrical and gives the solution (503, 1006).
F16. (Grade 12.) a) Prove that for every real number x the arithmetic mean
√ √
of 1 + sin x and 1 − sin x is equal to one of the following: sin 2x , cos 2x ,
− sin x2 , − cos 2x .
b) Can one leave out one of the four numbers listed in part a) in such a
way that the claim still holds?
Answer: b) no.
Solution 1. a) Denote the arithmetic mean given in the problem by A( x ).
As
x x x x x x 2
1 + sin x = sin2 + cos2 + 2 sin cos = sin + cos ,
2 2 2 2 2 2
19
x x x x x x 2
1 − sin x = sin2 + cos2 − 2 sin cos = sin − cos ,
2 2 2 2 2 2
we get
√ √
1 + sin x + 1 − sin x | sin x2 + cos x2 | + | sin 2x − cos x2 |
A( x ) = = .
2 2
Depending on the signs of the numbers sin x2 + cos x2 and sin x2 − cos x2 , one
of the trigonometric functions in the numerator cancels out and the other
one is doubled, with either a positive or a negative sign. Therefore, A( x ) is
equal to one of the numbers sin x2 , cos 2x , − sin x2 , − cos 2x .
b) Clearly A( x ) = 1, whenever x is one of the numbers 0, π, 2π, 3π.
Nevertheless, each of these four values makes a unique expression among
sin 2x , cos x2 , − sin 2x , − cos x2 evaluate to 1. Therefore, none of these four can
be left out.
Solution 2. Part a) can also be proven as follows. Let A( x ) be the same
as in the first solution. Then
√ √ !2 p
1 + sin x + 1 − sin x 2 + 2 1 − sin2 x 1 + | cos x |
= = ,
2 4 2
q q
1+| cos x | 1+cos x
so that A( x ) = 2 . Therefore, if cos x ≥ 0, then A ( x ) = 2 =
q
± cos x2 ; if cos x < 0, then A( x ) = 1−cos 2
x
= ± sin x2 .
F17. (Grade 12.) In an acute triangle ABC, a point P is chosen such that all
points symmetrical to P with respect to the sides of ABC lie on the circum-
circle of ABC. Prove that P is the orthocenter of ABC.
Solution. Let A′ , B′ , C ′ be points symmetric to
the point P with respect to the sides BC, CA, AB
C A′
(Fig. 17). Then |C ′ A| = | PA| = | B′ A|, giving that
′ ′ B ′
the arcs AC and AB of the circumcircle of the tri-
angle ABC are equal. Since A and C ′ are on the
same half-plane from the line BB′ , and C on the P
′ ′
other one, we have ∠C CA = ∠ B CA = ∠ PCA.
A B
Since P and C ′ are on the same side from the line
AC, the points P, C, C ′ are collinear. Since PC ′ ⊥
AB, we must also have PC ⊥ AB, that is, the point C′
P lies on the height drawn from the vertex C in the
triangle ABC. Analogously we see that P is on the Fig. 17
other two heights.
Remark 1. The converse—the points symmetric to the orthocenter with
respect to the sides of the triangle lie on the circumcircle of the triangle—is
a known result in elementary geometry that can also be used to solve this
problem. Namely, the point P lies on the reflections of arcs AB, BC, CA from
the corresponding lines AB, BC, CA of the circumcircle. By the theorem
20
mentioned, the intersection point of the height lies on the same arcs. But
the circles, whose arcs are the reflections, already meet twice pairwise at
the points A, B, C. Therefore, they cannot have two common intersection
points.
Remark 2. The claim of the problem, as well as the theorem given in
Remark 1, hold for all triangles, not just acute ones.
F18. (Grade 12.) There are 2n soldiers standing in a line, where n is a posi-
tive integer. The soldiers can rearrange themselves into a new line only in
the following way: the soldiers standing at odd numbered positions move
to the front of the row, keeping their positions with respect to each other,
and the soldiers previously standing at even numbered positions move to
the end of the row, keeping their positions with respect to each other. Prove
that after n rearrangements the soldiers stand in the same ordering as in the
beginning.
Solution 1. The last soldier does not change its position. The rest of the
soldiers regroup just as in the case, when the last soldier was not there, and
the number of the soldiers was 2n − 1. So, it suffices to prove the claim
for 2n − 1 soldiers. We show that after n rearrangements the soldiers are
in positions, which can be found in the original line by counting cyclically
every 2i -th soldier (after the last soldier we go to the first one). Indeed,
after 0 rearrangements, the claim clearly holds, and every rearrangement
makes us cyclically count every second soldier in the previous line (after the
last soldier we go to the second one), the first soldier will still be counted
first. After n rearrangements the soldiers in the new line can be found by
counting every 2n -th soldier in the old line with 2n − 1 soldiers. Since the
remainder of 2n when divided by 2n − 1 is 1, this is equivalent to simply
counting the soldiers. This means that we get back the original line.
Solution 2. Enumerate the soldiers starting from 0, and write the num-
bers in binary form (adding leading zeros to make the lengths of the binary
codes equal; for example for n = 3 we have the numbers 000, 001, 010, 011,
100, 101, 110, 111). After a rearrangement the soldiers stand in such a way
that when reinterpreting the last digit as the first one (but leaving the order
of the rest of the digits the same), the soldiers are again enumerated by con-
secutive numbers. After n rearrangements the binary code of the soldiers
has returned to the original, so every soldier’s position corresponds to their
original position in the line.
F19. (Grade 12.) a) Does there exist a function from real numbers to real
numbers, which is not constantly zero and whose derivative’s graph can be
obtained by reflecting the graph of the original function with respect to the
y-axis?
b) Does there exist a function from real numbers to real numbers, which
is not constantly zero and whose derivative’s graph can be obtained by
21
shifting the graph of the original function towards the positive side of the
x-axis by one unit?
Answer: a) yes; b) yes.
Solution. a) What the problem asks is equivalent to finding a function f
such that f ′ ( x ) = f (− x ) for all real numbers x. This is so, for example, for
the function f ( x ) = sin x + cos x, since f ′ ( x ) = cos x − sin x = cos(− x ) +
sin(− x ) = f (− x ).
b) The premise about the graphs is equivalent to saying that for every
real number x, we have f ′ ( x ) = f ( x − 1). Assume that we have a > 1
such that ln a = a−1 . Then, defining f ( x ) = a x , we get f ′ ( x ) = a x ln a =
a x · a−1 = a x −1 = f ( x − 1). It remains to make sure that such a number
a > 1 exists. Since ln 1 = 0 < 1 = 1−1 and ln e = 1 > e−1 , the graphs of the
continuous functions g( x ) = ln x and h( x ) = x −1 intersect at some point
a > 1. This is the number we were looking for.
Remark. One can prove that in part a) precisely all functions of the
form f ( x ) = c · (sin x + cos x ), where c 6= 0, satisfy the√
premises. The an-
2
swers can be written in a different form, for example 2 · (sin x + cos x ) =
sin x + π4 .

First day
S1. Prove that for any positive integer k there exist k pairwise distinct
integers for which the sum of their squares equals the sum of their cubes.
Solution. For any integer m > 1 the numbers 2m2 + 1, m(2m2 + 1),
−m(2m2 + 1) satisfy the conditions of the problem, because they are pair-
wise different and
(2m2 + 1)2 + (m(2m2 + 1))2 + (−m(2m2 + 1))2
= (1 + m2 + m2 ) · (2m2 + 1)2 = (2m2 + 1)3 = (1 + m3 − m3 ) · (2m2 + 1)3
= (2m2 + 1)3 + (m(2m2 + 1))3 + (−m(2m2 + 1))3.
With m growing, the numbers in these triples get arbitrarily large, hence
for any set of these triples one can find a new triple, where all numbers are
larger than the ones already used.
Any positive integer k can be written as k = 3q + r with 0 ≤ r < 3.
Choose q triples as above so that the numbers in them do not coincide. If
r = 1, then add 0, and if r = 2, then add 0 and 1. Since for each group
the sum of the squares of the numbers equals the sum of the cubes of the
numbers, the same property holds for the whole set.
Remark 1. One can find these triples by looking for three numbers where
two of them are opposites of each other. This gives the equation x2 + 2y2 =
22
x3 , or 2y2 = x 2 ( x − 1). Let d = gcd( x, y) and x = dn, y = dm. Then
2m2 = n2 (dn − 1). If n had a nontrivial prime divisor, then it must also
divide m, a contradiction. Hence n = 1 and the equation is 2m2 = d − 1, or
d = 2m2 + 1. By choosing m freely we get the triples above.
Remark 2. One can also solve the problem by first showing that there
exist infinitely many quadruples (−m, m, −n, n + 1) with 2m2 = n2 + n that
satisfy the conditions of the problem.
S2. For a given positive integer n one has to choose positive integers a0 ,
a1 , . . . so that the following conditions hold:
(1) ai = ai+n for any i;
(2) ai is not divisible by n for any i;
(3) ai+ ai is divisible by ai for any i.
For which positive integers n > 1 is this possible only if the numbers a0 , a1 ,
. . . are all equal?
Answer: for all primes.
Solution. Let n be a prime. By condition (1) the sequence a0 , a1 , . . . con-
tains only finitely many different numbers. If am is maximal of them, then
by condition (3) am+ am must also be maximal. Let us prove that if am is
maximal of the numbers, then am+k· am is also maximal for any k ≥ 0. This
holds for k = 0. If the claim holds for k, then am+( k+1)· am = am+k· am + am =
am+k· am + am+k · am = am+k· am = am . This proves the claim. By condition (2) am
is not divisible by n. Since n is prime, the numbers am and n are relatively
prime. Hence among the numbers m + k · am , where 0 ≤ k < n, there is one
in each congruence class modulo n. Hence all members of the sequence are
maximal, i.e. they are equal.
Suppose n is a composite number; let m be its divisor with 1 < m < n.
For any k < m choose ak = m + k · n and continue the sequence with period
m. Condition (1) holds, since n is a multiple of m. Condition (2) holds,
since all members of the sequence are congruent to m modulo n. For the
condition (3) notice that all members of the sequence are divisible by m.
Hence i and i + ai are always congruent modulo m, therefore ai = ai+ ai . At
the same time not all the numbers are equal.
S3. In a cyclic quadrilateral ABCD we have | AD | > | BC | and the vertices

C and D lie on the shorter arc AB of the circumcircle. Rays AD and BC
intersect at point K, diagonals AC and BD intersect at point P. Line KP
intersects the side AB at point L. Prove that ∠ ALK is acute.
Solution 1. From the properties of cyclic quadrilaterals we get ∠KAB =
∠KCD and ∠KBA = ∠KDC. Let A′ , B′ , K ′ be the foots of the altitudes of
the triangle ABK drawn from the vertices A, B, K, respectively, and let H
be the orthocenter of the triangle ABK (Fig. 18). The points A, B, A′ , B′ lie
on a common circle, hence ∠KA′ B′ = ∠KAB if A′ 6= B′ . Therefore A′ and
23
A
A
D X
B′
H D
K K′
L K′
P K P L
A′
C C
B
B Y
Fig. 18 Fig. 19
B′ lie on a line parallel to CD. Denote this line by A′ B′ (even in the case
A′ = B′ = K).
Let d( X, l ) be the distance of point X from line l, and let S∆ be the area of
triangle ∆. By two angles, △ ACK ∼ △ BDK and △ PAD ∼ △ PBC, whence
| AK |
| BK |
= || BD
AC |
|
| AP | | AD |
and | BP| = | BC| . At the same time
d( A, CD ) S△ ACD | AC | · | AD | · sin ∠CAD | AC | | AD |
= = = · ,
d( B, CD ) S△ BCD | BD | · | BC | · sin ∠CBD | BD | | BC |
d( A, KP) S△ AKP | AK | · | AP| · sin ∠KAP | AK | | AP|
= = = · .
d( B, KP) S△ BKP | BK | · | BP| · sin ∠KBP | BK | | BP|
Therefore
| AL| d( A, KP) d( A, CD )
= = .
| LB| d( B, KP) d( B, CD )
Considering instead of the cyclic quadrilateral ABCD the quadrilateral de-
termined by points A, B, A′ , B′ , and instead of P and L the points H and
K ′ correspondingly, we get similarly that
| AK ′ | d( A, KH ) d( A, A′ B′ )
= = .
|K ′ B| d( B, KH ) d( B, A′ B′ )
This equality holds also in the special case A′ = B′ = K. Indeed, let the
projections of points A and B to the line A′ B′ be X and Y correspondingly
(Fig. 19), then ∠ AKX = ∠KDC = ∠KBA = ∠ AKK ′ , ∠ BKY = ∠KCD =
∠KAB = ∠BKK ′ , whence △ AKX ∼ = △ AKK ′ and △ BKY ∼ = △ BKK ′ . It
′ ′ | AK ′ | d ( A,A ′ B ′ )
follows that | AK | = | AX |, | BK | = | BY | and |K ′ B| = d( B,A′ B′ ) .
24
Since C and D lie on the shorter arc AB, we A
have ∠ BCA = ∠ BDA > π2 . Thus the line A′ B′ α
is farther from the points A and B than the line δ
CD. Since | AD | > | BC |, we have ∠ ABD > D
∠CAB and also ∠KBA > ∠KAB, which implies β
|KA| > |KB|. Hence d( A, CD ) + d(K, CD ) >
d( B, CD ) + d(K, CD ), or d( A, CD ) > d( B, CD ). K
P ξ L
All together
| AL| d( A, CD ) d( A, A′ B′ ) | AK ′ | α
= > ′ ′
= ′ . δ
| LB| d( B, CD ) d( B, A B ) |K B| C β
Hence L lies farther from A than K′
on the seg- B
ment AB, therefore ∠ ALK < ∠ AK ′ K = π2 , i.e.
∠ ALK is acute. Fig. 20
Solution 2. Denote ∠KAB = ∠KCD = α, ∠KBA = ∠KDC = β, ∠KAC =
∠KBD = δ and ∠ ALK = ξ (Fig. 20). Then ∠KDB = α + β − δ = ∠KCA.
The condition | AD | > | BC | is equivalent to β > α, and points C and D
being located in the shorter arc AB is equivalent to the inequality α + β −
|KD | sin α
δ < π2 . In triangle KCD we get |KC| = sin β . From triangles KDP and
|KP |
KCP we obtain sin( α+ β− δ )
= sin(ξ|−( KD |
,
|KP |
β− δ )) sin( α+ β− δ )
= sin(ξ|+(
KC |
α− δ ))
, respec-
|KD | sin( ξ −( β− δ )) |KD |
tively. Consequently, |KC| = sin( ξ +( α−δ)) . Expressions of |KC| together yield
sin α sin( ξ −( β− δ )) sin ξ cos( β− δ )−cos ξ sin( β− δ )
sin β = sin( ξ +( α− δ )) = sin ξ cos ( α− δ )+cos ξ sin( α− δ ) , that in turn implies
sin ξ (sin β cos( β − δ) − sin α cos(α − δ))

= cos ξ (sin β sin( β − δ) + sin α sin(α − δ)) . (1)
Clearly sin ξ > 0 and sin β sin( β − δ) + sin α sin(α − δ) > 0 because ξ < π
and β > δ, α > δ. By the formula sin x cos y = 12 (sin( x + y) + sin( x − y)),
we get sin β cos( β − δ) − sin α cos(α − δ) = 21 (sin(2β − δ) − sin(2α − δ)).
Now α + β − δ < π2 implies (2β − δ) + (2α − δ) < π, i.e., there exists a
triangle whose two angles are 2β − δ and 2α − δ. But 2β − δ > 2α − δ since
β > α, therefore the law of sines in that triangle implies sin(2β − δ) >
sin(2α − δ) (the larger the angle, the large its opposite side in a triangle).
Hence the second factor in the l.h.s. of equation (1) is positive. Altogether,
we obtain cos ξ > 0, whence ξ < π2 .
Remark. This problem can also be solved by coordinates.
Second day
S4. Let ABC be a triangle where | AB| = | AC |. Points P and Q are differ-
ent from the vertices of the triangle and lie on the sides AB and AC, respec-
tively. Prove that the circumcircle of the triangle APQ passes through the
circumcenter of ABC if and only if | AP| = |CQ|.
25
Solution. Without loss of generality, we A
can assume that | AP| ≤ | AQ|. Let O be
the circumcenter of ABC. Let R be the in- P
tersection point of the bisector of ∠ BAC
with the circumcircle of the triangle PAQ
— we then have | RB| = | RC | (Fig. 21). R Q
Also, ∠ APR = 180◦ − ∠ AQR = ∠CQR O
and | RP| = | RQ| (since ∠ RAP = ∠ RAQ). B C
So, | AP| = |CQ| ⇐⇒ △ APR ∼ = △CQR
⇐⇒ | RA| = | RC | ⇐⇒ R = O (where Fig. 21
∼
| RA| = | RC | ⇒ △ APR = △CQR by two sides and obtuse angle).
Remark. This problem has been taken from the booklet “The Coins of
Harpland and 20+10 other maths problems from Ireland” (edited by Bernd
Kreussler), its author is Jim Leahy. The solution here is new.
S5. Let x, y, z be positive real numbers whose sum is 2012. Find the max-
imum value of
( x2 + y2 + z2 )( x3 + y3 + z3 )
.
( x 4 + y 4 + z4 )
Answer: 2012.
( x 2 +y2 +z2 )( x 3 +y3 +z3 )
Solution. If x = y = z = 2012 3 , then ( x 4 + y 4 + z4 )
= 2012. Now we
prove that for all x, y, z satisfying the premises we have
( x2 + y2 + z2 )( x3 + y3 + z3 )
≤ 2012.
( x 4 + y 4 + z4 )
It suffices to show that ( x2 + y2 + z2 )( x 3 + y3 + z3 ) ≤ 2012( x4 + y4 + z4 ), or
( x2 + y2 + z2 )( x3 + y3 + z3 ) ≤ ( x + y + z)( x4 + y4 + z4 ). Multiplying out,
simplifying and rearranging the terms gives xy( x − y)( x2 − y2 ) + xz( x −
z)( x2 − z2 ) + yz(y − z)(y2 − z2 ) ≥ 0. Since the differences in the brackets
in every product have equal signs, the products are non-negative, showing
that the necessary inequality holds.
Remark. The inequality that we get after multiplying out and canceling
x5 , y5 , z5 , is a special case of Muirhead inequality (with exponent vectors
(4, 1, 0) and (3, 2, 0)).
S6. On an m × m board, at the midpoints of the unit squares there are some
ants. At the time 0 each ant starts moving with speed 1 parallel to some
edge of the board until it meets an ant moving in the opposite direction
or until it reaches the edge of the board. When two ants moving in the
opposite direction meet each other, both turn 90◦ clockwise and continue
moving parallel to another edge of the board. Upon reaching the edge of
the board the ant falls off the board.
a) Prove that eventually all the ants will have fallen off the board.
b) Find the latest possible moment for the last ant to fall off the board.
26
Answer: b) 32 m − 1.
Solution 1. Let the lower left corner of the board be the origin. Divide
the units of time and space by 2; then the squares are of dimensions 2 × 2,
the coordinates of the midpoints of the squares are odd positive integers,
and the speed of the ants is still 1.
We prove by induction that at integer time moments the coordinates of
the ants are integers and the sum of the coordinates for any fixed ant has
the same parity as the time moment. In addition, the ants can meet only at
integer time moments. At time t = 0 all coordinates of the ants are odd,
so their sum is even. Suppose that at an integer time moment t = k the
coordinates of the ants are integers and the sum of the coordinates for any
fixed ant has the same parity as the time moment. If two of the ants were
to meet each other within the next time unit, they have to move toward
each other from time t = k, hence one of their coordinates must be the
same. Since the parity of the sum of their coordinates was the same at time
t = k, another of their coordinates had to differ by at least 2. Hence they
cannot meet before time t = k + 1. Between time moments t = k and
t = k + 1 every ant has changed only one of its coordinates by 1, hence at
time t = k + 1 the parity of the sum of the coordinates is again the same as
the parity of the time moment.
Next we will prove by induction that for any point with integer coor-
dinates ( x, y) there are no collisions at this point after the time moment
t = x + y − 2. For x = y = 1 this is obviously true, since there are no colli-
sions in the middle of the lower left square (otherwise one of the ants has to
arrive to this point from the edge of the board). Let ( x, y) be arbitrary and
suppose that the claim holds for all points with the sum of the coordinates
less than x + y. Suppose that a collision takes place at point ( x, y) at time t.
One of the participants had to arrive from a point, where one of the coordi-
nates was smaller; w.l.o.g. we can assume that this was the x-coordinate. If
this ant has not collided with anyone before, then t ≤ x − 1 ≤ x + y − 2. If
the last collision of this ant occurred at time t′ < t, then the coordinates
of the last collision were ( x − (t − t′ ), y). By the induction assumption
t′ ≤ x − (t − t′ ) + y − 2, hence t ≤ x + y − 2.
By symmetry the claim holds when another corner is chosen as the ori-
gin. Let the last collision of a particular ant occur at the point ( x, y), where
the coordinates are taken with respect to the nearest corner. W.l.o.g., we
can assume x ≤ y. The time from the last collision to the falling off the
edge of the ants participating in the collision is at most 2m − x, hence the
time elapsed from the start is at most x + y − 2 + 2m − x ≤ 3m − 2. By this
time all ants have fallen off the edge. With respect to the original units the
maximal time is 32 m − 1.
For any m the maximal time can be achieved, if in the beginning there
are 2 ants at the adjoining corners of the board moving toward each other.
At the moment t = m− 1
2 the pair collides and one of the ants starts moving
27
toward the center, falling off the board at time t = 32 m − 1.
Solution 2. Part a) can also be solved as follows. For each ant consider
the distance to the edge in the direction of its motion. After an ant falls
this distance will remain 0. Observe that as long as an ant moves without
collision, this distance decreases with speed 1.
Consider now the sum of all such distances. When a collision happens,
the sum of the distances of the two corresponding ants is m, both right
before and right after the collision. Thus as long as there are ants left on
the board, the total sum decreases with the speed of at least 1. Since in the
beginning this sum is a finite number, after some time this sum will become
0 and thus all ants will have fallen off the board.
Remark. This problem, proposed by Estonia, appeared in the IMO 2011
shortlist as C5.

Number theory: O1, O7, O12, F1, F5, F9, F10, F15, S2
Algebra: O6, O8, O13, F6, F11, F16, F19, S1, S5
Geometry: O3, O4, O9, O14, F2, F4, F7, F12, F17, S3, S4
Discrete mathematics: O2, O5, O10, O11, O15, F3, F8, F13, F14, F18, S6
28
2012/2013
Tartu 2013
WE THANK:
University of Tartu
Problem authors: Juhan Aru, Maksim Ivanov, Kairi Kangro, Urve Kangro,
Oleg Koshik, Toomas Krips, Härmel Nestra, Uve Nummert, Erik Paemurru
Translators: Urve Kangro, Härmel Nestra
Editor: Härmel Nestra

may be given.
curriculum.
contest lasts two days; each day, the contestants have 4.5 hours to solve 3 problems, sim-
ilarly to the IMO. All participants are given the same problems. Some problems in our
*
round, selection has been made to include only problems that have not been taken from
other competitions or problem sources and seem to be interesting enough. The team
selection contest is presented entirely.
1
1 1 1
O-1. (Juniors.) Nonzero integers a, b and c satisfy + + = 0. Prove that among a,
a b c
b, c there are two integers which have a common divisor larger than 1.
Solution: Multiplying the given equation by abc we get bc + ca + ab = 0. If a, b, c were
all odd, then bc, ca and ab were also odd and their sum could not be 0. If one of the
numbers a, b, c was even and the others were odd, then two of the numbers bc, ca and
ab were even and one odd, which also would not add up to 0. Hence at least two of the
numbers a, b, c are even, which satisfy the conditions.
O-2. (Juniors.) Teacher tells Jüri two nonzero integers a and b such that b is divisible by
a. Jüri has to find a nonzero integer c such that c is divisible by b and all solutions of the
quadratic equation ax2 + bx + c = 0 are integers. Can Jüri always solve the problem?
Answer: Yes.
Solution: By the conditions of the problem there is an integer q such that b = aq. Let
c = −2aq2 ; then c 6= 0 and c is divisible by b. The quadratic equation ax2 + bx + c = 0
or ax2 + aqx − 2aq2 = 0 has solutions q and −2q.
O-3. (Juniors.) Inside a circle c with the center O there are two circles c1 and c2 which
go through O and are tangent to the circle c at points A and B crespectively. Prove that
the circles c1 and c2 have a common point which lies in the segment AB.
Solution: The radius AO of the circle c is perpendicular to the common tangent to cir-
cles c and c1 at the point A, hence AO is a diameter of the circle c1 . Similarly BO is a
diameter of the circle c2 . If the circles c1 and c2 are tangent at the point O (Fig. 1), then
the diameters AO and BO are both perpendicular to the common tangent to c1 and c2 at
the point O, whence the lines AO and BO coincide, i.e. O lies in the segment AB. If the
circles c1 and c2 intersect at O (Fig. 2), then let M be the other intersection point of the
circles. Since ∠ AMO = 90◦ and ∠ BMO = 90◦ (angles at the circumference supported
by a diameter), the lines AM and BM coincide and M lies in the segment AB.
O-4. (Juniors.) Numbers 1, . . . , 200 are written on a blackboard in one line. Juku has to
write in front of each number plus or minus sign so that for any positive integer n ≤ 100
the number itself and one of its multiples have different signs. Which numbers must he
assign a minus sign in order to get the maximal possible value of the expression?
Answer: The numbers 51, . . . , 100.
Solution: If Juku writes a minus in front of the number 51, . . . , 100 and a plus in front
of the others, then the conditions of the problem are satisfied: for 51 ≤ n ≤ 100, the
numbers n and 2n have different signs; for n ≤ 50 there is at least one multiple of n
among the numbers 51, . . . , 100.
To show that this arrangement of the signs gives the maximal value of the expression,
2
B
B
M
N
O
B A A
O P
M
A K L C
consider an arbitrary arrangement of signs satisfying the conditions of the problem.

Then always when 100 ≥ n ≥ 67 and n has a plus sign, 2n must have a minus sign.
Also when 66 ≥ n ≥ 51 and n has a plus sign, then either 2n or 3n must have a minus
sign. If we change all the pluses in front of the numbers n with 100 ≥ n ≥ 51 to minuses,
and the minuses in front of the corresponding 2n or 3n to pluses, then changing minus
m m
to plus in front of m corresponds to changing plus to minus in front of or or
2 3
m m
both. Since + < m, the changes increase the value of the expression. Then we
2 3
can also change all remaining minuses in front of the numbers 1, . . . , 50 and 101, . . . , 200
to pluses, which also can only increase the value of the expression. This results in the
arrangement of the signs described in the beginning. Hence this arrangement gives the
maximal value of the expression.
1 1
O-5. (Juniors.) Kärt writes the fractions and on the blackboard and Märt writes
2 3
10 positive integers on the paper, which he does not show to Kärt. Then Kärt starts to
write fractions on the blackboard by the following rule: on each step she chooses two
a c
fractions and which are already on the blackboard and writes on the blackboard
b d
a+c
the fraction after reducing. Can Kärt always choose the fractions so that after a
b+d
number of steps she writes on the blackboard a fraction whose denominator is coprime
with all the numbers Märt has written on the paper?
Answer: Yes.
2
Solution: The first fraction that Kärt adds to the blackboard has to be . On every fol-
5
1
lowing move, let Kärt pick as one fraction and the latest written fraction as the other
2
fraction. Ignoring the reducing step, this means that the denominator of every added
fraction is larger than the previous fraction by 2, or that the denominators of the frac-
tions are consecutive odd numbers. This is indeed the case, because all fractions added
k
in this way are irreducible (these fractions have the form , and k is always coprime
2k + 1
with 2k + 1 because any common divisor would also divide (2k + 1) − 2k = 1).
Therefore the denominators of the fractions that Kärt writes include all prime numbers
except 2. Since there are infinitely many prime numbers, Kärt will eventually write a
3
fraction with a prime denominator that is larger than all of the numbers written by Märt,
and hence coprime with them.
O-6. (Juniors.) Publisher Soothsayer published a reference book claiming that for each
real number x and positive even number n the equality (1 + x )n ≥ 2n x holds. Is this
claim true?
Answer: No.
1
Solution: The inequality does not hold for example when x = and n = 4.
2
Remark: This is a plausible mistake, because a similar inequality (1 + x )n ≥ 2nx holds
for any real number x and positive even integer n.
O-7. (Juniors.) In an isosceles right triangle ABC the right angle is at vertex C. On the
side AC points K, L and on the side BC points M, N are chosen so that they divide the
corresponding side into three equal segments. Prove that there is exactly one point P
inside the triangle ABC such that ∠KPL = ∠ MPN = 45◦ .
Solution: Without loss of generality let the points on the side AC be in the order A, K,
L, C and on the side BC in the order C, M, N, B (see Fig. 3). Choose the point P so that the
quadrilateral LCMP is a square. Then |KL| = | LC | = | LP| and | MN | = |CM | = | MP|,
i.e. KLP and PMN are isosceles right triangles, so ∠KPL = ∠ MPN = 45◦ . Since
∠KPN = 45◦ + 90◦ + 45◦ = 180◦ , the point P lies inside the segment KN, whose all
points except the endpoints are inside the triangle ABC.
To show that P is the only point with the required properties, let P0 be an arbitrary point
inside the triangle ABC which satisfies ∠KP0 L = ∠ MP0 N = 45◦ . Since P and P0 are on
the same side of the line KL and ∠KPL = ∠KP0 L, the point P0 lies on the circumcircle
of the triangle KPL; similarly it also lies on the circumcircle of the triangle MPN. Since
∠KLP = ∠ PMN = 90◦ , the segments KP and PN are the diameters of the circles.
Since the diameters KP and PN lie on the same straight line KN, they have a common
perpendicular at the point P which is tangent to both circles at this point. Hence the
point P is the only common point of these circles, i.e. P0 = P.
O-8. (Juniors.) The numbers 1, 2, . . . , 2012 are written on the blackboard in some
order, each of them exactly once. Between each two neighboring numbers the absolute
value of their difference is written and the original numbers are erased. This is repeated
until only one number is left on the blackboard. What is the largest possible number
that can be left on the blackboard?
Answer: 2010.
Solution: The largest number on the blackboard cannot increase on any step, because the
absolute value of the difference of two nonnegative numbers cannot be greater than the
maximum of these two numbers. Since in the beginning all the numbers are different
and positive, after the first step the largest possible number is 2011 and the smallest
possible number is 1. After the second step the largest possible number is 2010 and
hence the number left on the blackboard in the end cannot be larger than 2010.
The number 2010 can be left on the blackboard, for example when in the begin-
ning the numbers are written in the order 2012, 1, 2, 3, . . . , 2011. Then after the first
4
step there are the numbers 2011, 1, 1, . . . , 1, and after the second step the numbers
2010, 0, 0, . . . , 0. On each following step the number of zeroes decreases by one and
in the end only the number 2010 remains.
O-9. (Juniors.) Find all pairs of integers ( a, b) satisfying ( a + 1)(b − 1) = a2 b2 .

Answer: (0, 1) and (−1, 0).
Solution: Since a and a + 1 are coprime, a2 and a + 1 are also coprime. Similarly b2
and b − 1 are coprime. Hence the equality can hold only in the case a + 1 = ±b2 and
b − 1 = ± a2 , where the signs in both equations are the same.
Let both signs be pluses. Then from the first equation we get a = b2 − 1 = (b − 1)(b + 1).
The second equation implies b − 1 = a2 , whence a = a2 ( a2 + 2). If a = 0, then b = 1, i.e.
( a, b) = (0, 1). If a 6= 0, then by dividing by a we get 1 = a( a2 + 2); since a2 + 2 > 1, this
equation does not have integer solutions.
If both signs are minuses then by multiplying by −1 we get −b + 1 = a2 and − a − 1 =
b2 . These are the same equations with respect to −b and − a which we had previously
with respect to a and b, hence the only solution is −b = 0, − a = 1 i.e. ( a, b) = (−1, 0).
O-10. (Seniors.) Find all positive integers which are exactly 2013 times bigger than the
sum of their digits.
Answer: 36234.
Solution: Note that the minimal value of a k-digit number is 10k−1 and the maximal value
of the cross-sum multiplied by 2013 is 9k · 2013. Since 9 · 7 · 2013 = 126819 < 1000000
we can consider only numbers with up to 6 digits. Since then the cross-sum is at most
54, it is enough to consider numbers in the form n · 2013 with 1 ≤ n ≤ 54.
Since 2013 is divisible by 3, n · 2013 and its cross-sum are divisible by 3. Since the cross-
sum must be equal to n, n · 2013 is divisible by 9. But then its cross-sum and hence also n
is divisible by 9. It remains to consider the cases n = 9, 18, . . . , 54 which can be checked
by hand and see that only n = 18 satisfies the conditions.
O-11. (Seniors.) Find all remainders which one can get when dividing by 6 an integer
n which satisfies n3 = m2 + m + 1 for some integer m.
Answer: 1.
Solution: Numbers n and n3 give the same remainder when dividing by 6. Also, m2 +
m + 1 is odd and gives the remainder 0 or 1 when dividing by 3. The only possibility to
get 0 as the remainder is when m = 3k + 1, but then
n3 = (9k2 + 6k + 1) + (3k + 1) + 1 = 9k2 + 9k + 3 = 3(3k2 + 3k + 1)
which leads to a contradiction, since if n3 is divisible by 3, it is also divisible by 33 , but

3k2 + 3k + 1 is not divisible by 3. Hence the remainder of n3 is 1 both when dividing by
2 or 3, consequently its remainder when dividing by 6 is 1.
The remainder 1 is possible: take n = 1 and m = 0 (or n = 7 and m = 18).
O-12. (Seniors.) Prove that for any integer n ≥ 3 we have (2n)! < n2n .
5
Solution 1: For n = 3 the claim holds: (2n)! = 6! = 720 and n2n = 36 = 729.
Suppose n ≥ 4. Divide the numbers 2, 3, . . . , 2n − 2 into pairs (k, 2n − k ) with 2 ≤ k ≤
n − 1, leaving n alone. For each pair we have
k(2n − k ) = (n − (n − k ))(n + (n − k )) = n2 − (n − k)2 < n2 .
Hence 2 · 3 · . . . · (2n − 2) < (n2 )n−2 · n = n2n−3 , therefore
(2n)! < 1 · n2n−3 · (2n − 1) · (2n) < n2n−3 · (2n)2 = 4n2n−1 ≤ n2n .
Solution 2: For n = 3 the claim holds. Suppose the claim holds for n; to show that it also
(n + 1)2n
holds for n + 1 it is enough to show the inequality (2n + 1)(2n + 2) < 2n
( n + 1)2 .
n
2 2 (n + 1)2n
Since (2n + 1)(2n + 2) < (2n + 2) = 4(n + 1) , it is enough to show that > 4.
n n2n
1
This is equivalent with 1 + > 2 which holds for all n ≥ 2.
n
O-13. (Seniors.) Inside a circle c there are circles c1 , c2 and c3 which are tangent to
c at points A, B and C correspondingly, which are all different. Circles c2 and c3 have
a common point K in the segment BC, circles c3 and c1 have a common point L in the
segment CA, and circles c1 and c2 have a common point M in the segment AB. Prove
that the circles c1 , c2 and c3 intersect in the center of the circle c.
Solution: Take a point X on the common tangent to the circles c1 and c which lies on the
other side of the line AB from the point C. Then ∠ ALM = ∠XAM = ∠XAB = ∠ ACB
| AM|
(Fig. 4). Consequently ML k BC. Similarly KM k CA and LK k AB. If = λ, then
| AB|
| BK | | BM| |CL| |CK | | AM|
= = 1 − λ and = = 1 − (1 − λ) = λ, whence λ = =
| BC | | BA| |CA| |CB| | AB|
| AL| 1
= 1 − λ. Hence λ = , therefore the triangles AML, MBK and LKC are all similar
| AC | 2
1
to ABC with the factor . Thus the radii of their circumcircles c1 , c2 and c3 are equal to
2
half of the radius of the circumcircle c of the triangle ABC. Since the circles c and c1 are
tangent, the diameter of c1 and the radius of c, both drawn from the tangent point A,
coincide. Hence the circle c1 goes through the center of the circle c; similarly the circles
c2 and c3 go through the center of the circle c.
O-14. (Seniors.) For which positive integers m and n is it possible to write the numbers
1, 2, . . . , 2mn into the white squares of a 2m × 2n checkerboard in such a way that the
sum of the numbers in every row is the same, and the sum of the numbers in every
column is the same?
Answer: For all even m and n, except when m = n = 2.
Solution: All the numbers sum up to mn(2mn + 1). For odd m this is not divisible by 2n,
breaking the equality of all column sums. Thus m and likewise also n cannot be odd.
In the case m = n = 2 we cannot write the numbers as required, because the numbers
6
B
X
M
K
A
a ∗
L
∗
C
Figure 4 Figure 5
in the squares marked by ∗ in Fig. 5 must be equal.

Let us show that in the white squares of a 4 × 8 checkerboard we can write the numbers
k + 1, k + 2, . . . , k + 8 and 2mn − k − 7, 2mn − k − 6, . . . , 2mn so that the sums of the
numbers in rows are the same, and the sums of the numbers in columns are the same.
One possibility is the following, where P stands for 2mn:
k+1 P−k−2 k+8 P−k−5
k+2 P−k−3 k+7 P−k−4
P−k k+3 P−k−7 k+6
P−k−1 k+4 P−k−6 k+5
Here the sums in the columns are 2mn + 1 and the sums in the rows are 4mn + 2.
One can also write the numbers 1, 2, . . . , 12 and 2mn − 11, 2mn − 10, . . . , 2mn on a 4 × 12
checkerboard in the required way, where P stands for 2mn again:
1 6 12 P−3 P−4 P−9
3 8 9 P−1 P−6 P − 10
P P−5 P − 11 4 5 10
P−2 P−7 P−8 2 7 11
If one of the numbers m and n is even and the other is divisible by 4, then we can
cover the 2m × 2n checkerboard with 4 × 8 checkerboards and fill them as above, taking
k = 0, 8, . . . , mn − 8 in different small checkerboards. If neither m nor n is divisible by
4 and one of them is at least 6 then we can cover the checkerboard with one 4 × 12 and
4 × 8 checkerboards and in the 4 × 8 checkerboards take k = 12, 20, . . . , mn − 8.
O-15. (Seniors.) Let a and b be positive integers such that b is divisible by a and writing
b
a and b one after another in this order gives ( a + b)2 . Prove that = 6.
a
Solution: Let n be the number of digits of b and let b = ka. Then by the conditions of the
problem, 10n · a + ka = ( a + ka)2 , or
10n + k
a= . (1)
( k + 1)2
If k were odd, then the numerator on the r.h.s. of (1) would be odd and the denominator
even, so a could not be an integer. Hence k is even.
7
If k = 2 then the cross-sum of 10n + 2 is 3, which is not divisible by (2 + 1)2 = 9. The
case k = 4 also leads to a contradiction, since 10n + 4 ends with 4, hence cannot be
divisible by (4 + 1)2 = 25. Thus k ≥ 6.
In the following we show first that k ≤ 8 and finally that k 6= 8. The assumptions
ka = b ≥ 10n−1 give 10ka ≥ 10n . Equality (1) implies
10n = (k + 1)2 · a − k = k2 a + 2ka + a − k = (k + 2) · ka + a − k .
Thus 10ka ≥ (k + 2) · ka + a − k, whence
(8 − k) · ka ≥ a − k . (2)
As a is positive, (8 − k ) · ka > −k. As both sides of this inequality are divisible by k, this
implies (8 − k ) · ka ≥ 0. Consequently 8 − k ≥ 0, i.e., k ≤ 8.
10n + 8
If k = 8, the inequality (2) implies a ≤ 8 whereas the equality (1) reduces to a = .
81
Hence a ends with digit 8, leaving a = 8 and b = 8 · 8 = 64 as the only possibility. But
864 6= (8 + 64)2 , contradicting the conditions of the problem.
Remark: It is not hard to show that the smallest numbers satisfying the conditions of
1036 + 6
the problem are a = = 20408163265306122448979591836734694 and b = 6a =
49
122448979591836734693877551020408164.
x2 + 4xy + y2
O-16. (Seniors.) Let x and y be different positive integers. Prove that is
x 3 − y3
never an integer.
Solution 1: By symmetry we can assume that x > y. If x − y = 1, then
x2 + 4xy + y2 x2 + 4xy + y2 ( x − y)2 + 6xy

= = =
x 3 − y3 ( x − y)( x2 + xy + y2 ) ( x − y) ( x − y)2 + 3xy
1 + 6xy 3xy
= = 1+ ,
1 + 3xy 1 + 3xy
which is clearly not an integer. If x − y ≥ 2, then
x2 + 4xy + y2 x2 + 4xy + y2 x2 + 4xy + y2

= ≤ <
x 3 − y3 ( x − y)( x2 + xy + y2 ) 2( x2 + xy + y2 )
2x2 + 2xy + 2y2
< = 1,
2( x2 + xy + y2 )
where the last inequality follows from x2 − 2xy + y2 = ( x − y)2 > 0.

x2 + 4xy + y2
Solution 2: If were an integer, then
x 3 − y3
x2 + 4xy + y2 3xy
3 3
· ( x − y) − 1 = 2
x −y x + xy + y2
8
F2
D c2
F1 F2
F1 Q
A1
A2
Q
O1 c1
O2 D
A2 A1
c1 O1 O2
P
E1 c
c2
B
E2
E1 P
c
B E2
Figure 6 Figure 7
would also be an integer. As 3xy > 0 and x2 + xy + y2 > 0, we have 3xy ≥ x2 + xy + y2 ,

whence ( x − y)2 ≤ 0. Hence x = y, which contradicts the conditions of the problem.
O-17. (Seniors.) Circles c1 , c2 with centers O1 , O2 , respectively, intersect at points P

and Q and touch circle c internally at points A1 and A2 , respectively. Line PQ intersects
circle c at points B and D. Lines A1 B and A1 D intersect circle c1 the second time at
points E1 and F1 , respectively, and lines A2 B and A2 D intersect circle c2 the second time
at points E2 and F2 , respectively. Prove that E1 , E2 , F1 , F2 lie on a circle whose center
coincides with the midpoint of line segment O1O2 .
r
Solution: Let the radii of c1 , c2 and c be r1 , r2 and r, respectively. Homothety of ratio
r1
with center A1 takes circle c1 to circle c and points E1 , F1 to points B, D, respectively.
r
Thus it takes line E1 F1 to line BD. Analogously, homothety of ratio with center A2
r2
takes line E2 F2 to line BD. Consequently, lines E1 F1 and E2 F2 are parallel to line BD
(Fig. 6).
Furthermore, note that | BE1 | · | BA1 | = | BP| · | BQ| and | BE2 | · | BA2 | = | BP| · | BQ|,
implying | BE1 | · | BA1 | = | BE2 | · | BA2 |. Thus triangles BE1 E2 and BA2 A1 are similar
and
1
∠ BE1 E2 = ∠ BA2 A1 = ∠ BDA1 = ∠E1 F1 A1 = ∠E1O1 A1 = 90◦ − ∠O1 E1 A1 , (3)
2
whence
∠E2 E1O1 = 180◦ − ∠ BE1 E2 − O1 E1 A1 = 90◦ . (4)
Analogously, ∠E1 E2O2 = 90◦ . Hence the quadrilateral E1 E2O2O1 is a right-angled

trapezoid (or rectangle in the case r1 = r2 ) and the midpoint of the line segment O1O2
lies on the perpendicular bisector of the line segment E1 E2 , thus being equidistant from
E1 and E2 . Analogously, the midpoint of the line segment O1O2 is also equidistant from
F1 and F2 .
9
As line O1O2 is perpendicular to BD, line O1O2 is also perpendicular to E1 F1 . Thus the
line segment O1O2 entirely lies on the perpendicular bisector of E1 F1 . This means that
the midpoint of line segment O1O2 is equidistant from E1 and F1 .
Altogether, we have shown that these four points lie on a circle with its center at the
midpoint of the line segment O1O2 .
Remark: The chains of equations (3) and (4) hold as given in the situation depicted in
Fig. 6, where F1 and O1 lie at the same side from line A1 E1 . There are other situations
where F1 and O1 lie at different sides from A1 E1 or O1 lies on the line A1 E1 or circle c lies
inside circles c1 and c2 (see Fig. 7). Despite the equations having a little different form,
the final result ∠E2 E1O1 = 90◦ still holds.
O-18. (Seniors.) Eha and Koit play the following game. In the beginning of the game
at each vertex of a square there is an empty box. At each step each player has two
possibilities: either add one stone to an arbitrary box, or to move each box clockwise to
the next vertex of the square.
Koit begins and they make in turns 2012 steps (each player 1006). Then Koit marks one
of the vertices of the square and lets Eha make one more step. Koit wins if after this step
the number of stones in some box is larger than the number of stones in the box at the
vertex Koit marked; otherwise Eha wins. Which player has a winning strategy?
Answer: Eha.
Solution: First show that Eha can guarantee that before Koit’s move the number of stones
in the boxes lying at opposite corners of the square are equal. In the beginning it is true,
since all the boxes are empty. Let before Koit’s move the numbers of stones in the boxes
be ( a, b, a, b). If Koit adds one stone to a box, then Eha can add a stone to the box at the
opposite corner; if Koit moves the boxes cyclically, Eha also moves the boxes cyclically,
so the condition still holds. Hence Eha can make the moves so that after step 2012 the
numbers of the stones in the boxes are ( a, b, a, b). Without loss of generality we can
assume that a ≥ b. If Koit marks a vertex with a box with a stones then Eha adds one
stone to the box and wins. If Koit marks a vertex with a box with b stones, then Eha
moves the boxes cyclically and still wins.
O-19. (Seniors.) Find all functions f from the set of all positive integers to the same set
such that, for all positive integers a1 , . . . , ak with k > 0, the sum a1 + . . . + ak divides the
sum f ( a1 ) + . . . + f ( ak ).
Answer: All functions given by f (n) = an, a ∈ N.
Solution: Suppose that f is a function that satisfies the conditions of the problem. We
claim that f (n) = f (n − 1) + f (1) for all integers n > 1. Indeed, for any integer m > n,
we have m | f (n) + f (m − n) and m | f (n − 1) + f (1) + f (m − n) by conditions of the
problem. Hence the difference f (n) − ( f (n − 1) + f (1)) is also divisible by m. As m was
arbitrary, this implies that f (n) − ( f (n − 1) + f (1)) is divisible by an infinite number of
different integers, i.e., is equal to 0. This completes the proof of the claim.
Easy induction now gives that necessarily f (n) = n f (1). It is straightforward to verify
that all functions of the form f (n) = an satisfy the conditions of the problem.
10
Olympiad
F-1. (Grade 9.) Consider hexagons whose internal angles are all equal.
(i) Prove that for any such hexagon the sum of the lengths of any two neighbouring
sides is equal to the sum of the lengths of their opposite sides.
(ii) Does there exist such a hexagon with side lengths 1, 2, 3, 4, 5 and 6 in some order?
Answer: ii) Yes.

Solution 1:
(i) Let the hexagon be ABCDEF. It suffices to show that | AB| + | BC | = | DE| + | EF |.
Let K be the intersection point of rays FA and CB and let L be the intersection
point of rays FE and CD (Fig. 8). The size of every internal angle of the hexagon
is 120◦ , whence triangles KAB and LDE are equilateral. The quadrilateral FKCL
is a parallelogram since its opposite sides are parallel. This implies |KC | = | LF |
or |KB| + | BC | = | LE| + | EF |, which together with |KB| = | AB| and | LE| = | DE|
implies the desired equality | AB| + | BC | = | DE| + | EF |.
(ii) Take a parallelogram with side lengths 7 and 5 and internal angles 60◦ and 120◦ ,
and cut off equilateral triangles with side lengths 1 and 2 at its acute angles. This
gives rise to a hexagon with all internal angles having size 120◦ and side lengths 1,
4, 5, 2, 3, 6 (Fig. 9).
Solution 2:
(i) Note that the external angles of the hexagon have size 60◦ . Any two opposite sides
of the hexagon are parallel, as they are separated by exactly three external angles.
Consider a line s perpendicular to opposite sides CD and FA of the hexagon
ABCDEF (Fig. 10). As all other sides form the same angle 30◦ with line s, the
lengths of these sides are proportional to the lengths of the projections of the sides
to line s. The sum of the lengths of the projections of sides AB and BC is equal
to the distance between the parallel lines CD and FA and the same holds also for
the opposite sides DE and EF. Therefore the sum of the lengths of the projections
of sides AB and BC is equal to that of sides DE and EF, whence the sums of the
lengths of the sides are equal as well.
C D L
5 2
B
4 2 2
E
K 1 1 3
A F 1 6
Figure 8 Figure 9
11
C D
B
s
E
A F
Figure 10 Figure 11
x d d
c c
e e
x
b
b
f b−x f
x a x a
(ii) Figure 11 shows a hexagon in a triangular grid with distance between neighbouring
nodes being 1. The side lengths of the hexagon are 1, 4, 5, 2, 3 and 6.
Solution 3: Consider two types of transformations on hexagons that maintain the prop-
erty that all internal angles are of the same size.
(1) Prolonging two opposite sides by the same quantity x (Fig. 12); this causes the side
lengths to change according to the template
( a, b, c, d, e, f ) ←→ ( a+ x, b, c, d+ x, e, f ) .
(2) Prolonging the two neighbouring sides of one particular side by the same quantity
x (Fig. 13); this causes the side lengths to change according to the template
( a, b, c, d, e, f ) ←→ ( a+ x, b− x, c+ x, d, e, f )
A straightforward check shows that both transformations maintain the desired prop-
erty, no matter of in which direction the transformations are applied.
(i) As the internal angles of all regular hexagons are equal, it suffices to show that
an arbitrary hexagon with all internal angles equal can be turned into a regular
hexagon by a finite sequence of the transformations above.
Indeed, let the side lengths of a given hexagon with all internal angles equal be
( a, b, c, d, e, f ). Assume w.l.o.g. that d ≥ a and f ≥ c. Choose a quantity s such
that s ≥ max( a, b, c); by applying the transformation (1) thrice, we can obtain a
hexagon with three consecutive sides having the same length:
(1) (1) (1)

( a, b, c, d, e, f ) −→ (s, b, c, d0 , e, f ) −→ (s, s, c, d0 , e0 , f ) −→ (s, s, s, d0 , e0 , f 0 ) .
12
By the assumption made above we have d0 ≥ s and f 0 ≥ s. W.l.o.g., assume also
d0 ≤ f 0 . The transformation
(2)
(s, s, s, d0 , e0 , f 0 ) −→ (s, s, s, s, e00 , f 00 )
leads to a hexagon with four consecutive sides of equal length. But this must be
regular since all of its internal angles are equal (Fig. 14).
(ii) Such a hexagon can be obtained from a regular hexagon with side length 1 by the
following transformations:
(1) (1) (2)

(1, 1, 1, 1, 1, 1) −→ (1, 4, 1, 1, 4, 1) −→ (1, 4, 5, 1, 4, 5) −→ (1, 4, 5, 2, 3, 6) .
F-2. (Grade 9.) Two children are playing noughts and crosses with changed rules.
In each move, either of the players may draw into an empty square of a 3 × 3 board
either a nought or a cross according to one’s wish. Moves are made alternately and the
winner is the one after whose move a row, a column or a long diagonal becomes filled
with three similar signs. Is there a player with a winning strategy, and if yes then who?
Answer: Yes, the first player.
Solution: The first player may play the first move into the middle square and, later
on, make the immediately winning move if there is any and play symmetrically to the
opponent’s last move w.r.t. the center of the board otherwise.
Suppose that the opponent wins. As the central square is occupied, the winning move
must be played either into a corner or in the middle of an edge of the board. According
to the first player’s strategy, the position before the winning move was symmetric w.r.t.
the center of the board. Consequently, the square symmetric to the winning move is
empty in the final position, which in turn implies that the three signs of the same type
appear along an edge of the board. Before the winning move, there must already have
been two of these signs present and, by symmetry, similarly also at the opposite edge.
Three of these four signs must already have been there before the last move of the first
player. As two of these three must have been in one line, the first player could win in
her last move, which contradicts the chosen strategy.
Remark: It is also easy to argue by case study.
F-3. (Grade 10.) Can 2013 be represented as the difference of two cubes of integers?
Answer: No.
Solution 1: Suppose that 2013 = x3 − y3 where x and y are integers. Note that
x3 − y3 = ( x − y)3 + 3x2 y − 3xy2 = ( x − y)3 + 3xy( x − y) .
As 2013 is divisible by 3 and so is 3xy( x − y), the difference ( x − y)3 must be divisible
by 3. Thus also x − y is divisible by 3 as 3 is prime. Consequently, 3xy( x − y) is divisible
by 32 and ( x − y)3 is divisible by 33 , whence the sum x3 − y3 is divisible by 32 . But 2013
is not divisible by higher powers of 3. The contradiction shows that 2013 cannot be
represented as the difference of two cubes of integers.
13
A B
A B A B
P P O
D C
D C D C
Solution 2: Suppose that 2013 = x3 − y3 where x and y are integers. Then x3 ≡ y3

(mod 3). As integers are congruent to their cubes modulo 3, this gives x ≡ y (mod 3).
Now write 2013 = ( x − y)( x2 + xy + y2 ). By the congruence obtained above, the first
factor in the r.h.s. is divisible by 3 and the terms x2 , xy and y2 in the second factor are
all congruent modulo 3 whence x2 + xy + y2 is divisible by 3. Altogether, the product
( x − y)( x2 + xy + y2 ) must be divisible by 9, but 2013 is not.
Remark: It is possible to solve this problem by brute force in several ways.
F-4. (Grade 10.) The bases of trapezoid ABCD are AB and CD, and the intersection
| PA| | PB|
point of its diagonals is P. Prove that if = then the trapezoid is isosceles.
| PD | | PC |
| PA| | PD |
Solution 1: By assumptions, = . As the bases AB and CD are parallel, we
| PB| | PC |
| PA| | PC |
have also = (Fig. 15). Hence | PC | = | PD |. Similarity of triangles APD and
| PB| | PD |
| AD | | PD |
BPC implies = = 1, thus | AD | = | BC | as needed.
| BC | | PC |
Solution 2: By assumptions, triangles APD and BPC are similar. Thus ∠ ADB = ∠ ACB
(Fig. 16), showing that quadrilateral ABCD is cyclic. But if a quadrilateral with parallel
opposite sides has a circumcircle, the bisectors of these sides coincide as they have the
same direction and both pass through the circumcenter of the quadrilateral (Fig. 17). By
symmetry w.r.t. this line, the other pair of opposite sides have equal lengths.
F-5. (Grade 10.) Each unit square in a 5 × 5 table is coloured either blue or yellow.
Prove that there exists a rectangle with sides parallel to the edges of the table, such that
the four unit squares in its corners have the same colour.
Solution 1: Each row contains at least 3 squares with the same colour. Similarly, the
dominating colour must be the same in at least 3 rows. W.l.o.g., suppose that the first 3
rows contain at least 3 blue squares each. If the first two rows contain two blue squares
in the same columns then the desired rectangle exists. Otherwise, each column contains
at least one blue square in these two rows (in Fig. 18, the blue squares are consecutive
w.l.o.g.). Thus in one of the first two rows there are two blue squares that are in the
same columns with the blue squares in the third row. These form the desired figure.
Solution 2: The first row containes three squares of the same colour, say, blue. If in
some of the remaining four rows there are two blue squares aligned to the blue squares
in the first row then the desired rectangle exists. Otherwise, each of these four rows
14
contains two yellow squares aligned with the blue squares in the first row. But two
yellow squares can be placed into three columns in 3 different ways only. Thus there
exist two rows where these two yellow squares are in the same columns. These squares
form the desired figure.
F-6. (Grade 10.) Jüri draws a circle c with radius 3 and n circles with radius 1 on a
paper. Find the minimal n for which he can draw the circles in such a way that it would
not be possible to draw inside the circle with radius 3 any new circles with radius 1
having at most one common point with each of the previously drawn circles.
Answer: 3.
Solution: Let O be the center of circle c. Suppose that circles c1 , c2 , c3 with radius 1
and centers O1 , O2 , O3 , respectively, are placed in such a way that they touch circle c
internally and O1 , O2 , O3 are vertices of an equilateral triangle (coloured dark in Fig. 19).
No new circle can be placed to the same line with two existing circles with radius 1
because the three circles would require free area with length 6 which is possible only
along the diameter of circle c. The other ways of placement are in the middle or to the
other side of the narrower area between two circles (uncoloured in Fig. 19).
The circle with radius 1 and center O touches all three circles, hence the location is
fixed. The circle with radius 1 and center O10 being symmetric to point O w.r.t. line O2O3
touches circles c2 and c3 and has at least one common point also with circle c, as it is
in one line with the circle in the middle and circle c1 . Hence this location is fixed, too.
Note that after moving the circle c2 towards point O the circle will intersect the circle
with center O10 (since ∠OO2O10 < 90◦ ) and also the circle in the middle. Thus if all circles
c1 , c2 , c3 are moved a bit towards point O, chances to add new circles disappear.
Consequently, if n ≥ 3 then Jüri can draw the circles in such a way that new circles
cannot be added in the required fashion. Show now that in the case n ≤ 2 a new
circle can always be added; it suffices to consider the case n = 2. Let c1 and c2 be
the given circles with radius 1 and centers O1 and O2 . W.l.o.g., O1 6= O. Choose AB
as the diameter of c which is perpendicular to O1O (Fig. 20). Consider two circles with
radius 1, touching the circle c at points A and B, respectively. Circle c1 does not preclude
drawing either of them as it is located inside the strip with width 2 surrounding the line
O1O (dark in Fig. 20), where neither of the two circles outreach. Circle c2 can preclude
at most one of the two circles since it cannot outreach to both sides of the strip.
O2 c
O1 O
O1
O O1′
O3
15
F-7. (Grade 11.) Let a finite decimal fraction be given. Juku starts appending digits to
this fraction in such a way that each new digit equals the remainder of the sum of all
digits existing so far in division by 10. (For instance, if the initial fraction is 27.35 then
the digits added to the end are 7, 4, 8 etc.)
Prove that the infinite decimal fraction obtained this way represents a rational number.
Solution: It suffices to show that the infinite decimal fraction is periodic. For that, note
that each new digit except for the first digit is congruent to twice the previous digit
modulo 10. Indeed, let a1 , . . . , ak−1 , ak be the existing at some time moment digits where
ak is already added by Juku. Then the next digit ak+1 satisfies
ak+1 ≡ a1 + . . . + ak = ( a1 + . . . + ak−1 ) + ak ≡ ak + ak = 2ak (mod 10) .
Hence each new digit is uniquely determined by the last existing digit. As there are only
a finite number of different digits, some digit must be added repeatedly. According to
the fact just proven, all following digits are repeated as well.
Remark: This solution can be reformulated without mentioning the fact that each new
digit except for the first is congruent to twice the previous one. After noting that Juku
must add some digit repeatedly, denote their position numbers in the decimal fraction
by m and n. Hence the digits before the mth digit and the digits before the nth digit sum
up to congruent numbers modulo 10. Adding the mth and the nth digit, respectively,
to the sums maintains the congruence. This means that the next digits are also equal.
Thus the digits start repeating periodically.
F-8. (Grade 11.) Let n > 1 be an integer and a1 , a2 , . . . , an some real numbers, the sum
of which is 0 and the sum of the absolute values of which is 1. Prove that
n−1
| a1 + 2a2 + . . . + nan | ≤ .
2
Solution 1. According to the assumptions, for each k = 1, . . . , n − 1 it holds that
| a 1 + . . . + a k | = | a k +1 + . . . + a n | ,
| a 1 + . . . + a k | + | a k +1 + . . . + a n | ≤ | a 1 | + . . . + | a k | + | a k +1 | + . . . + | a n | = 1 .
1
Consequently, | ak+1 + . . . + an | ≤ for each k = 1, . . . , n − 1. Now
2
| a1 + 2a2 + . . . + nan |
= |( a1 + . . . + an ) + ( a2 + . . . + an ) + . . . + ( an−1 + an ) + an |
≤ | a 1 + . . . + a n | + | a 2 + . . . + a n | + . . . + | a n −1 + a n | + | a n |
n −1
≤ 0 + 21 + 12 + . . . + 21 + 21 = 2 .
Solution 2. Let A+ and A− be the sum of positive numbers and the sum of negative
numbers, respectively. By the assumptions, A+ + A− = 0 and A+ − A− = 1, implying
1 1
A+ = and A− = − .
2 2
16
B B
β β
C
F α F C
G G′
E E
β D
α D α
A A
Figure 21 Figure 22
By increasing the coefficients of positive terms and decreasing the coefficients of nega-
tive terms, the whole sum can only increase, and by decreasing the coefficients of posi-
tive terms and increasing that of negative terms, the sum can only decrease. Thus
n −1
a1 + 2a2 + . . . + nan ≤ n · A+ + 1 · A− = 2 ,
a1 + 2a2 + . . . + nan ≥ 1 · A+ + n · A− = − n−2
1
.
n−1
Consequently, | a1 + 2a2 + . . . + nan | ≤ .
2
n−1 1
Remark: The bound can be achieved for every n by taking a1 = − , a2 = . . . =
2 2
1
an−1 = 0 and an = .
2
F-9. (Grade 11.) A convex quadrilateral ABCD where ∠ DAB + ∠ ABC < 180◦ is
given on a plane. Let E be a point different from the vertices of the quadrilateral on
the line interval AB such that the circumcircles of triangles AED and BEC intersect
inside the quadrilateral ABCD at point F. Point G is defined so that ∠ DCG = ∠ DAB,
∠CDG = ∠ ABC and triangle CDG is located outside quadrilateral ABCD. Prove that
the points E, F, G are collinear.
Solution: Denote ∠ DAB = α and ∠ ABC = β (Fig. 21). From cyclic quadrilaterals AEFD
and BEFC one obtains
∠ DFE = 180◦ − ∠ DAE = 180◦ − α ,
∠CFE = 180◦ − ∠CBE = 180◦ − β ,
respectively. Thus ∠CFD = 360◦ − (180◦ − α) − (180◦ − β) = α + β. But ∠CGD =

180◦ − (α + β) by the choice of G. Hence the quadrilateral CFDG is cyclic. Conse-
quently, ∠ DFG = ∠ DCG = α = 180◦ − ∠ DFE, which implies that the points E, F, G
are collinear.
Remark 1: The argumentation can also be turned around in the following way: Let G 0 be
defined as the other intersection point of the circumcircle of triangle CFD and line EF
(Fig. 22). Then the quadrilateral CFDG 0 is cyclic, whence
∠ DCG 0 = ∠ DFG 0 = 180◦ − ∠ DFE = α ,

∠CDG 0 = ∠CFG 0 = 180◦ − ∠CFE = β .
17
These equalities imply that G 0 = G. Thus G belongs to line EF.
Remark 2: The claim of the problem holds also if point F does not have to be inside the
quadrilateral ABCD. Then G may also be located between E and F.
F-10. (Grade 11.) A (2k + 1) × (2k + 1) table, where k is a positive integer, contains
one real number in each entry, where these numbers are pairwise different. After each
row, one writes the median of the row, i.e., the number occurring in this row such that
the row contains the same amount of numbers less than it and greater than it. Let m be
the median of the column of medians. Prove that more than a quarter of the numbers
initially in the table are less than m.
Solution: Each row contains k numbers less than the median and k numbers greater than
the median. Thus k + 1 numbers in each row do not exceed the median of that row. In
rows whose median does not exceed m, these k + 1 numbers do not exceed m either.
There are k + 1 such rows. Consequently, there are at least (k + 1)2 numbers in the table
that do not exceed m. Only one of them is equal to m, whence (k + 1)2 − 1 = k2 + 2k
numbers are less than m. As k is positive by assumption, we have 1 < 4k and 4k + 1 <
k2 + 2k k2 + 2k k2 + 2k 1
4k + 4k = 8k. Now 2
= 2
> 2
= and we are done.
(2k + 1) 4k + 4k + 1 4k + 8k 4
F-11. (Grade 11.) For which natural numbers n ≥ 3 is it possible to cut a regular n-
gon into smaller pieces with regular polygonal shape? (The pieces may have different
number of sides.)
Answer: 3, 4, 6, 12.
Solution: A regular triangle can be partitioned into four regular triangles of equal size
(Fig. 23), a regular quadrilateral can be partitioned into four regular quadrilaterals with
equal size (Fig. 24) and a regular hexagon can be partitioned into six regular triangles
of equal size (Fig. 25). By building alternately equilateral triangles and squares onto the
sides of a regular 12-gon, just a regular hexagon remains (Fig. 26), whence also a regular
12-gon can be partitioned in the required way.
Show now that other regular polygons cannot be partitioned into smaller regular poly-
gons. For that, consider an arbitrary polygon that is partitioned into regular polygons.
As the size of an internal angle of a regular polygon is less than 180◦ and not less than
60◦ , at most two regular polygons can meet at each vertex.
If a vertex of the big n-gon is filled by just one smaller polygon then this piece is an n-
gon itself. Beside it, there must be space for at least one regular polygon. No more than
two regular polygons can be placed there since the sum of the internal angles of these
polygons and the n-gon itself would exceed 180◦ . Two new pieces can be placed only
if all these three pieces are triangular, which gives n = 3. It remains to study the case
where there is exactly one polygon beside the n-gonal piece. The size of the internal
angle of the n-gon being at most 120◦ implies n ≤ 6. The case n = 5 is impossible as its
external angles are of size 72◦ but no regular polygon has internal angles of size strictly
between 60◦ and 90◦ .
If each vertex of the big n-gon is the meetpoint of two smaller regular polygons then one
of them must be a triangle since other regular polygons have internal angles of size 90◦
or more. Beside a triangle, there is space for a triangle, a quadrilateral or a pentagon.
18
Figure 23 Figure 24 Figure 25 Figure 26
60◦
108◦ 108◦
84◦
Figure 27
In the first two cases, the size of the internal angles of the n-gon will be 120◦ and 150◦ ,
respectively, covering the cases n = 6 and n = 12. It remains to show that the third
case with a triangle and a pentagon meeting at each vertex is impossible. Indeed, the
side length of the pentagon must coincide with the side length of the initial big n-gon,
because it is impossible to place a regular polygon beside the pentagon along one side.
For the same reason, another pentagon must be built to the second next side along the
boundary of the initial polygon. These two pentagons meet at the third vertex of the
triangle built to the side between (Fig. 27). But the ulterior angle between the sides of
the pentagons at the meeting point has size 360◦ − 2 · 108◦ − 60◦ = 84◦ , which cannot
be filled with interior angles of regular polygons.
F-12. (Grade 12.) Find the smallest natural number n for which there exist integers
a1 , . . . , an (that do not have to be different) such that a41 + . . . + a4n = 2013.
Answer: 14.
Solution: Note that the fourth powers of even numbers are divisible by 16 and the fourth
powers of odd numbers are congruent to 1 modulo 16. As 2013 ≡ 13 (mod 16), the
desired representation must contain at least 13 odd summands.
Suppose that no more summands are needed. As 74 = 2401 > 2013, each summand
must be 14 = 1, 34 = 81 or 54 = 625. There can be at most 3 summands 625 since
4 · 625 > 2013. Therefore the number of summands not divisible by 5 is at least 10.
The fourth power of an integer not divisible by 5 is congruent to 1 modulo 5, whereas
2013 ≡ 3 (mod 5). Hence the number of summands not divisible by 5 must be at least
13. This shows that the representation contains only summands 1 and 81, but 13 such
numbers sum up to at most 13 · 81 which is less than 2013. Thus representations with
13 summands are impossible.
On the other hand, 14 fourth powers is enough as 64 + 54 + 34 + 11 · 14 = 2013.
Remark: The fact that 2013 cannot be represented as the sum of 13 odd fourth powers
can also be proved without calculations modulo 5. Suppose that
a41 + . . . + a413 = 2013 , (5)
19
where ai = 2bi + 1 for every i = 1, . . . , 13. We may assume that each bi is 0, 1 or 2. As

4 4 3 2 4 3 3 2 1
(2x + 1) = 16x + 32x + 24x + 8x + 1 = 16 x + 2x + x + x + 1
2 2
and
3 1 1 1 1
x4 + 2x3 + x2 + x = x4 + 2x3 + x2 + x2 + x = x2 ( x + 1)2 + x ( x + 1) ,
2 2 2 2 2
2013 − 1 − 1 − . . . − 1
the equality (5) reduces to f (b1 ) + f (b2 ) + . . . + f (b13 ) = = 125
16
1
where f ( x ) = x2 ( x + 1)2 + x ( x + 1). Thus 125 should be representable as the sum of
2
13 integers, each of which is f (0) = 0, f (1) = 5 or f (2) = 39. Obviously the number of
summands 39 is at most 3 and, as 125 − 13 · 5 = 60, at least 2. The cases with two and
three summands 39 give no solution.
F-13. (Grade 12.) Real numbers x1 , x2 , x3 , x4 in [0; 1] are such that the product
K = | x1 − x2 | · | x1 − x3 | · | x1 − x4 | · | x2 − x3 | · | x2 − x4 | · | x3 − x4 |
1 4
is as large as possible. Prove that>K> .
27 243
Solution 1: If some two numbers among x1 , x2 , x3 , x4 are equal then K = 0 which is not
maximal. Thus assume w.l.o.g. that x1 > x2 > x3 > x4 . Applying AM-GM for x1 − x2 ,
x2 − x3 and x3 − x4 gives
( x1 − x2 ) + ( x2 − x3 ) + ( x3 − x4 ) x − x4 1
q
3
( x1 − x2 )( x2 − x3 )( x3 − x4 ) ≤ = 1 ≤ ,
3 3 3
1
i.e., | x1 − x2 | · | x2 − x3 | · | x3 − x4 | ≤
. Among the remaining factors | x1 − x3 |, | x1 − x4 |,
27
| x2 − x4 |, at least one is less than 1. Hence we conclude the left-hand inequality needed.
3 1
For the second inequality, note that if x1 = 1, x2 = , x3 = , x4 = 0 then
4 4
1 3 1 3 1 9 4
K= · ·1· · · = > ,
4 4 2 4 4 512 243
since 9 · 243 = 2187 > 2048 = 4 · 512.
Solution 2: W.l.o.g., assume the inequalities x1 > x2 > x3 > x4 . In addition, assume that
x1 = 1 and x4 = 0 as otherwise K can be made larger. Substituting x2 = y and x3 = z
for simplicity, one obtains
K = (1 − y)(1 − z)(y − z)yz = (y − z) · (1 − y)z · (1 − z)y .
Consider pairs (y, z) with y − z fixed. The sums (1 − y) + z = 1 − (y − z) and (1 −

z) + y = 1 + (y − z) are then also fixed. The product of two numbers with fixed sum
is the largest if the numbers are equal; thus the product (1 − y)z is the largest in the
20
case y + z = 1 and the product (1 − z)y is the largest in the same case y + z = 1.
Consequently, also K obtains its largest value in the case y + z = 1. Substituting 1 − z at
place of y, one gets K = z2 (1 − z)2 (1 − 2z) = (z(1 − z))2 (1 − 2z).
Let f (z) = (z(1 − z))2 (1 − 2z), then f 0 (z) = 2z(1 − z)(1 − 5z + 5z2 ). The roots of√f 0
5− 5
within (0, 1) are the roots of the quadratic polynomial 5z2 − 5z + 1, namely
√ √ 10
5− 5

5+ 5 1
and = 1− . As f (0) = f = 0 and f (z) > 0 whenever 0 < z <
10 10 2 √
1 5− 5
, the maximum of f is achieved at z = . Thus the maximum value of K is
2 √ ! √ √ 10
5− 5 5 5
f = . This number satisfies both inequalities of the problem.
10 125 125
F-14. (Grade 12.) The midpoints of sides C2 C3 , C3 C1 and C1 C2 of a triangle C1 C2 C3

are K1 , K2 and K3 , respectively. The centers of circles c1 , c2 and c3 are C1 , C2 and C3 ,
respectively, and the centers of circles k1 , k2 , k3 are K1 , K2 , K3 , respectively. No two of
the given six circles intersect in two points nor are they inside each other. Circles k1 , k2
and k3 touch each other externally.
(i) Prove that the sum of the radii of circles c1 , c2 and c3 does not exceed one quarter
of the perimeter of the triangle C1 C2 C3 .
(ii) Prove that if the sum of the radii of circles c1 , c2 and c3 equals one quarter of the
perimeter of the triangle C1 C2 C3 then the triangle C1 C2 C3 is equilateral.
Solution: Let the radii of the circles c1 , c2 , c3 be r1 , r2 , r3 , and the radii of the circles k1 , k2 ,
k3 be R1 , R2 , R3 , respectively (Fig. 28). By assumptions,
1
R 1 + R 2 = | K1 K2 | = 2 |C1 C2 | ,
1
R 2 + R 3 = | K2 K3 | = 2 |C2 C3 | ,
1
R 3 + R 1 = | K3 K1 | = 2 |C3 C1 | ,
1
which sum up to 2R1 + 2R2 + 2R3 = (|C1 C2 | + |C2 C3 | + |C3 C1 |). The assumptions
2
also imply inequalities
1 1
r1 + R3 ≤ 2 |C1 C2 | , R3 + r2 ≤ 2 |C1 C2 | ,
1 1
r2 + R1 ≤ 2 |C2 C3 | , R1 + r3 ≤ 2 |C2 C3 | ,
1 1
r3 + R2 ≤ 2 |C3 C1 | , R2 + r1 ≤ 2 |C3 C1 | ,
which sum up to 2r1 + 2r2 + 2r3 + 2R1 + 2R2 + 2R3 ≤ |C1 C2 | + |C2 C3 | + |C3 C1 |.
1
(i) Altogether, we obtain the inequality 2r1 + 2r2 + 2r3 ≤ (|C1 C2 | + |C2 C3 | + |C3 C1 |),
2
1
which implies r1 + r2 + r3 ≤ (|C1 C2 | + |C2 C3 | + |C3 C1 |) as desired.
4
21
C3
c3
2 4 2 4 2 6
6 8 8 6 8 4
K2 K1
k2 k1
Figure 29
c1
c2 2 9 4
C1 K3 C2
7 5 3
k3 6 1 8
Figure 28 Figure 30
1
(ii) Suppose that r1 + r2 + r3 = (|C1 C2 | + |C2 C3 | + |C3 C1 |). Then all inequalities
4
1
above must hold as equalities. The equalities r1 + R3 = |C1 C2 | = r2 + R3 im-
2
ply r1 = r2 , analogously r1 = r3 . Denoting r = r1 = r2 = r3 , we get
1
r + R3 = 2 |C1 C2 | = R1 + R2 ,
1
r + R2 = 2 |C1 C3 | = R1 + R3 ,
where summing side-by-side gives 2r + R2 + R3 = 2R1 + R2 + R3 , i.e., R1 = r.

Analogously, R2 = R3 = r. Thus all sides of the triangle C1 C2 C3 have length 4r.
F-15. (Grade 12.) Define magic square as a 3 × 3 table where each cell contains one
number from 1 to 9 so that all these numbers are used and all row sums and column
sums are equal. Prove that any two magic squares can be obtained from each other via
the following transformations: interchanging two rows, interchanging two columns,
rotating the square, reflecting the square w.r.t. its diagonal.
Solution: As all the transformations are invertible, it suffices to show that every magic
square can be turned to one particular magic square by these transformations.
The sum of all numbers in a magic square is 45, whence the numbers in each row and
each column must sum up to 15. As this is odd, exactly 0 or 2 of the three summands
must be even. There are 4 even numbers in use, hence 2 even numbers must be in some
two rows and 0 even number in the remaining one. The same holds for columns.
Hence the even numbers 2, 4, 6, 8 occur in the corners of some rectangle with sides
parallel to the edges of the table. By interchanging rows or columns one can move the
even numbers to the corners of the whole table. There are 3 possibilities to locate these
four numbers into the corners, that can not be obtained from each other by rotations and
reflections of the table (Fig. 29). The last two of them cannot occur in the magic square
because the missing numbers in the first and third column would coincide. Hence only
the first possibility remains. Its completion to a magic square is unique (Fig. 30).
22
First day
S-1. Find all prime numbers p for which one can find a positive integer m and non-
negative integers a0 , a1 , . . . , am less than p such that
a0 + a1 p + . . . + am−1 pm−1 + am pm = 2013,

a0 + a1 + . . . + am−1 + am = 11.
Answer: 2003.
Solution: Subtracting the second equation from the first one gives
a1 ( p − 1) + . . . + am ( pm − 1) = 2002.
As the l.h.s. of the obtained equality is divisible by p − 1, 2002 = 2 · 7 · 11 · 13 must

also be divisible by p − 1. Thus p − 1 equals one of 1, 2, 7, 11, 13, 14, 22, 26, 77, 91, 143,
154, 182, 286, 1001 and 2002. Since p is prime, only 2, 3, 23 and 2003 remain. The first
equation of the given system is the p-ary representation of 2013, whence the coefficients
ai are uniquely determined by p.
Now we study all cases.
1. If p = 2 then m = 10 as 210 < 2013 < 211 . The second equation implies that all
ai s must be ones, but 1 + 2 + 22 + . . . + 210 = 211 − 1 = 2047. Hence there is no
solution in this case.
2. Let p = 3. As 2013 = 2 · 3 + 32 + 2 · 33 + 2 · 35 + 2 · 36 whereas 2 + 1 + 2 + 2 + 2 =
9 6= 11, this case gives no solution either.
3. Let p = 23. As 2013 = 12 + 18 · 23 + 3 · 232 while 12 + 18 + 3 > 11, this case gives
no solution either.
4. For p = 2003, we get 2013 = 10 + 2003 and 10 + 1 = 11, so the conditions are
satisfied.
Consequently, 2003 is the only prime number with the desired property.
S-2. For which positive integers n ≥ 3 is it possible to mark n points of a plane

in such a way that, starting from one marked point and moving on each step to the
marked point which is the second closest to the current point, one can walk through
all the marked points and return to the initial one? For each point, the second closest
marked point must be uniquely determined.
Answer: for all n ≥ 4.

2π
Solution: To find a construction for any n ≥ 4, choose ε < . Place the points A1 , A2 ,
n3
. . . , A n −1 on a circle in such a way that the angle between the radii drawn to the points
23
2π
Ai and Ai+1 for i = 1, . . . , n − 2 is equal to αi = − (n − 2 − i )ε (see Fig. 31 for
n−2
n = 6). The angle between the radii drawn to the points An−1 and A1 is then equal to
(n − 2)(n − 3)
α n −1 = ε. Place the point An outside of the circle on the extension of the
2
radius containing An−1 at the same distance d from An−1 as the distance between points
A1 and A2 . It is straightforward to verify that the points A1 , . . . , An satisfy the condition
of the problem.
On the other hand, suppose that there exists a construction for n = 3. Let the cyclic
walk be A1 → A2 → A3 → A1 . Then d( A1 , A2 ) > d( A1 , A3 ), d( A2 , A3 ) > d( A2 , A1 )
and d( A3 , A1 ) > d( A3 , A2 ), where d( X, Y ) denotes the distance between X and Y. But
these three inequalities cannot hold simultaneously.
S-3. Let x1 , . . . , xn be non-negative real numbers, not all of which are zeros.
(i) Prove that

x2 x3 xn
x1 + + +...+ · ( x1 + 2x2 + 3x3 + . . . + nxn ) ( n + 1)2
1≤ 2 3 n ≤ .
( x1 + x2 + x3 + . . . + x n )2 4n
(ii) Show that, for each n ≥ 1, both inequalities can hold as equalities.
Solution: Applying AM-GM gives

! ! ! !
n n n n
x 1 nx
∑ kk ∑ kxk =
n
· ∑ kk ∑ kxk ≤
k =1 k =1 k =1 k =1
!2
n n
1 1 nxk
≤ ·
n 4 ∑ k + ∑ kxk =
k =1 k =1
2
! !2
n n
1 n ( n + 1)2
=
4n ∑ xk k + k ≤
4n ∑ xk .
k =1 k =1
n
(The last inequality is proved by + k ≤ n + 1, as it is equivalent to (n − k)(k − 1) ≥ 0.)
k
This gives us the necessary upper bound; this bound is achieved for instance if x1 =
xn = 1 and x2 = . . . = xn−1 = 0.
For the lower bound, estimate the numerator by Cauchy-Schwarz inequality:
! ! !2 !2
n n n n
r
x xk p
∑ kk ∑ kxk ≥ ∑ k
· kxk = ∑ xk ;
k =1 k =1 k =1 k =1
the equality holds here if exactly one of xi s is non-zero.
24
O C
A3 A2
τ
α2 D β B
α3 α1 d A
α4
α5
A4 A5 A1
A6 Q
Figure 31 Figure 32
Second day
S-4. Let D be the point different from B on the hypotenuse AB of a right triangle ABC
such that |CB| = |CD |. Let O be the circumcenter of triangle ACD. Rays OD and CB
intersect at point P, and the line through point O perpendicular to side AB and ray CD
intersect at point Q. Points A, C, P, Q are concyclic. Does this imply that ACPQ is a
square?
Solution: As OQ is the perpendicular bisector of AD, one has ∠QAD = ∠ ADQ =

∠ BDC = ∠CBD (Fig. 32). Therefore AQ k BC, whence ∠QAC = 180◦ − ∠ ACB = 90◦ .
From the cyclic quadrilateral APCQ one also gets ∠CPQ = ∠ PQA = 90◦ , i.e., ACPQ is
a rectangle.
As ∠ DOC = 2∠ DAC, one obtains
∠ DOC = 2∠ BAC = 2(90◦ − ∠CBA) = 180◦ − 2∠CBA =

= 180◦ − ∠CBD − ∠ BDC = ∠ DCB ,
which implies that isosceles triangles BDC and DCO are similar. Thus ∠ BDC = ∠ DCO,
i.e., OC k AB, whence ∠QOC = 90◦ = ∠QAC. So O lies on the circle determined by A,
C, P, Q. Therefore
1 1 1
∠ ACQ = ∠ AOQ = ∠ AOD = ∠ AOP = ∠ ACP.
2 2 2
Consequently, the diagonal of the rectangle ACPQ bisects the angle of the rectangle,
whence ACPQ is a square.
Remark: It turns out from the solution that the conditions of the problem determine the
3
shape of the triangle ABC, namely ∠ ABC = π.
8
25
S-5. Call a tuple (bm , bm+1 , . . . , bn ) of integers perfect if both following conditions are
fulfilled:
1. There exists an integer a > 1 such that bk = ak + 1 for all k = m, m + 1, . . . , n;
2. For all k = m, m + 1, . . . , n, there exists a prime number q and a non-negative integer

t such that bk = qt .
Prove that if n − m is large enough then there is no perfect tuples, and find all perfect
tuples with the maximal number of components.

Answer: 20 + 1, 21 + 1, 22 + 1, 23 + 1, 24 + 1 .
Solution: Clearly (20 + 1, 21 + 1, 22 + 1, 23 + 1, 24 + 1) is a perfect tuple with length 5.

Show in the rest that there are no other perfect tuples with length 5 or larger.
For that, let ( am + 1, am+1 + 1, . . . , an + 1) be an arbitrary perfect tuple with length at
least 5. There must exist at least two odd exponents among m, m + 1, . . . , n; let k and
k + 2 be the two largest odd exponents. As ak + 1 and ak+2 + 1 are prime powers while
having a common divisor a + 1, these two integers must be powers of the same prime
q. Thus the larger of them, ak+2 + 1, is divisible by the smaller one, ak + 1, which shows
that ak + 1 divides also the difference a2 · ( ak + 1) − ( ak+2 + 1) = a2 − 1. Hence ak + 1 ≤
a2 − 1, implying k < 2. So k = 1 as k is odd. By choice of k, the only odd exponents in our
perfect tuple are 1 and 3 and the tuple is of the form ( a0 + 1, a1 + 1, a2 + 1, a3 + 1, a4 + 1).
a3 + 1
As a + 1 and a3 + 1 are powers of the same prime number q, also the ratio =
a+1
a2 − a + 1 is a power of q. Note that a2 − a + 1 ≥ 2a − a + 1 = a + 1 by a ≥ 2, hence
a2 − a + 1 is divisible by a + 1. Thus the difference ( a2 − a + 1) − ( a + 1)( a − 2) = 3 is
divisible by a + 1. This filters out the only possibility a = 2.
S-6. A class consists of 7 boys and 13 girls. During the first three months of the school
year, each boy has communicated with each girl at least once. Prove that there exist two
boys and two girls such that both boys communicated with both girls first time in the
same month.
Solution 1: Call the first communication between a boy and a girl their acquaintance.
During the 3 months, there are 7 · 13 = 91 acquaintances in total. Thus there exists a
month when there was at least 31 aquaintances. Let the boys be denoted by p1 through
p7 and let Ti , i = 1, . . . , 7, be the set of girls to whom pi acquainted in this month. We
have to show that there exist distinct i and j such that Ti ∩ Tj contains at least 2 girls.
W.l.o.g., assume the inequalities | T1 | ≥ | T2 | ≥ . . . ≥ | T7 |. Consider two cases.
1. The case | T1 | + | T2 | + | T3 | + | T4 | ≥ 20. Suppose that all intersections T1 ∩ T2 , T1 ∩

T3 , . . . , T3 ∩ T4 contain at most one girl. Let k ≤ 6 be the number of non-empty
intersections. Then the first four boys acquainted with
| T1 ∪ T2 ∪ T3 ∪ T4 | ≥ | T1 | + | T2 | + | T3 | + | T4 | − k ≥ 20 − 6 = 14
26
2)
2)
2)
∪T
∪T
∪T
T1
T1
T1
1
1
\T
\T
\T
\(
\(
\(
T1
T2
T3
T1
T2
T3
T1
T2
T3
p1 p2 p3 p1 p2 p3 p1 p2 p3
Figure 33
girls in the month under consideration. (For proving the first inequality, note that
girls that belong to one or two subsets count once in the r.h.s., girls belonging to
three subsets not count and girls in all four subsets count −2 times.)
This is a contradiction since there are only 13 girls.
2. The case | T1 | + | T2 | + | T3 | + | T4 | ≤ 19. As | T5 | + | T6 | + | T7 | ≥ 12, we have | T5 | ≥ 4.

Now | T4 | ≤ 4 implies | T4 | = | T5 | = 4 and hence also | T6 | = | T7 | = 4. Now
| T1 | + | T2 | + | T3 | = 15 in order to get 31 in total.
Suppose that all intersections Ti ∩ Tj , i, j = 1, . . . , 7, contain at most one girl. If boys
p1 , p2 and p3 altogether acquainted with all girls in this month then at least one
intersection Ti ∩ T4 , i = 1, 2, 3, contains at least two girls. Otherwise, | T1 | + | T2 \
T1 | + | T3 \ ( T1 ∪ T2 )| = 15 − 1 − 2 = 12 (Fig. 33 shows all possibilities), since only
then there is a girl, say t13 , with whom none of p1 , p2 , p3 acquainted. Clearly all boys
p4 , p5 , p6 , p7 must have acquainted with her, and each of them also acquainted with
one girl from sets T1 , T2 \ T1 and T3 \ ( T1 ∪ T2 ). As the last set contains at most three
elements, two of the four boys acquainted with the same girl from T3 \ ( T1 ∪ T2 ).

7
Solution 2: Let A be the set of all combinations of two boys, | A| = = 21. Say
2
that girl t determines an element { p1 , p2 } of A if t acquainted with p1 and p2 in the same
month. If in the ithmonth
a girlt acquainted
with exactly ni boys, where i = 1, 2, 3, then
n1 n2 n3
t determines + + elements of A. Applying Jensen’s inequality for
2 2 2
x ( x − 1)

n1 n2 n3 n1 + n2 + n3 7 2
f (x) = gives + + ≥ 3· f = 3· f =4 .
2 2 2 2 3 3 3
n1 n2 n3 n1 n2 n3
As + + is an integer, + + ≥ 5. Hence all girls
2 2 2 2 2 2
determine at least 13 · 5 = 65 elements of A in total.
As 65 ≥ 3 · 21 + 1, an element { p1∗ , p2∗ } of A is determined by at least 4 girls by the
pigeonhole principle. Consequently there exists a month in which this couple of boys is
determined by at least two girls.
Solution 3: Suppose that there is no required pairs of boys and girls. Like in Solution
1, consider a month with 31 or more acquaintances. Let gi be the number of girls who
acquainted with exactly i boys in this month, i = 0, 1, . . . , 7. We get a system of inequal-
27
ities

 g2 + 3g3 + 6g4 + 10g5 + 15g6 + 21g7 ≤ 21,
g1 + g2 + g3 + g4 + g5 + g6 + g7 ≤ 13, (6)
g1 + 2g2 + 3g3 + 4g4 + 5g5 + 6g6 + 7g7 ≥ 31.

If at least one girl aquainted with 5 or more boys then g5 + g6 + g7 ≥ 1. The system (6)
then reduces to

 g2 + 3g3 + 6g4 + 10( g5 + g6 + g7 − 1) ≤ 11,
g1 + g2 + g3 + g4 + ( g5 + g6 + g7 − 1) ≤ 12,
g1 + 2g2 + 3g3 + 4g4 + 7( g5 + g6 + g7 − 1) ≥ 24.

Summing the first two inequalities gives
g1 + 2g2 + 4g3 + 7g4 + 11( g5 + g6 + g7 − 1) ≤ 23,
which contradicts the third inequality.

Thus g5 = g6 = g7 = 0 and the system of inequalities reduces to

 g2 + 3g3 + 6g4 ≤ 21,
g1 + g2 + g3 + g4 ≤ 13, (7)
g1 + 2g2 + 3g3 + 4g4 ≥ 31.

Suppose that g4 ≥ 1. As in the previous case, (7) implies


 g2 + 3g3 + 6( g4 − 1) ≤ 15,
g1 + g2 + g3 + ( g4 − 1) ≤ 12,
g1 + 2g2 + 3g3 + 4( g4 − 1) ≥ 27.

The first two inequalities sum up to g1 + 2g2 + 4g3 + 7( g4 − 1) ≤ 27. In the light of the
third inequality, this is possible only if g3 = 0 and g4 = 1. Then the second and third
inequalities give g1 + g2 ≤ 12 and g1 + 2g2 ≥ 27, which contradict each other since
g1 + 2g2 ≤ 2( g1 + g2 ).
Hence also g4 = 0 and our system of inequalities reduces to

 g2 + 3g3 ≤ 21,
g1 + g2 + g3 ≤ 13,
g1 + 2g2 + 3g3 ≥ 31.

Subtracting the first inequality from the third one, we obtain g1 + g2 ≥ 10. Subtracting
twice the second inequality from the third one, we get g3 − g1 ≥ 5. The two inequalities
obtained sum up to g2 + g3 ≥ 15, contradicting the second inequality.
Remark 1: After decreasing either the number of boys or the number of girls, the claim
of the problem would not hold anymore.
Remark 2: This problem is a variant of the problem F-5 from the Final Round for the 10th
grade. In terms of that problem, here we take 7 × 13 table instead of 5 × 5 and use three
colours instead of two.
28
Number theory: O-1, O-5, O-10, O-11, O-15, F-3, F-7, F-12, S-1, S-5
Algebra: O-2, O-6, O-9, O-12, O-16, O-19, F-8, F-13, S-3
Geometry: O-3, O-7, O-13, O-17, F-1, F-4, F-6, F-9, F-14, S-4
Combinatorics: O-4, O-8, O-14, O-18, F-2, F-5, F-10, F-11, F-15, S-2, S-6
Singapore Mathematical
Olympiad Team
Selection Test
And Trainings
1. SiMO Team Selection Test

1994/1995-2007/2008
2. SiMO Trainings 2003 and 2001
Singapore International Mathematical Olympiad
National Team Selection Tests 1994/95 - 2003/2004
1994/95
1.1.* Let N = {1, 2, 3, ...} be the set of all natural numbers and f : N → N be a function.
Suppose f (1) = 1, f (2n) = f (n) and f (2n + 1) = f (2n) + 1 for all natural numbers
n.
(i) Calculate the maximum value M of f (n) for n ∈ N with 1 ≤ n ≤ 1994.
(ii) Find all n ∈ N, with 1 ≤ n ≤ 1994, such that f (n) = M .
1.2. ABC is a triangle with ∠A > 90o . On the side BC, two distinct points P and Q are
chosen such that ∠BAP = ∠P AQ and BP · CQ = BC · P Q. Calculate the size of
∠P AC.
1.3. In a dance, a group S of 1994 students stand in a big circle. Each student claps the
hands of each of his two neighbours a number of times. For each student x, let f (x) be
the total number of times x claps the hands of his neighbours. As an example, suppose
there are 3 students A, B and C. A claps hand with B two times, B claps hand with
C three times and C claps hand with A five times. Then f (A) = 7, f (B) = 5 and
f (C) = 8.
(i) Prove that {f (x) | x ∈ S} =
6 {n | n is an integer, 2 ≤ n ≤ 1995}.
(ii) Find an example in which
{f (x) | x ∈ S} = {n | n is an integer, n 6= 3, 2 ≤ n ≤ 1996}.
1
2.1. Let f (x) = 1+x where x is a positive real number, and for any positive integer n, let
gn (x) = x + f (x) + f (f (x)) + · · · + f (f (. . . f (x))),
the last term being f composed with itself n times. Prove that
(i) gn (x) > gn (y) if x > y > 0.
(ii) gn (1) = F1
F2 + FF23 + · · · + FFn+1
n+2
, where F1 = F2 = 1 and Fn+2 = Fn+1 + Fn for n ≥ 1.
2.2. Let ABC be an acute-angled triangle. Suppose that the altitude of 4ABC at B
intersects the circle with diameter AC at P and Q, and the altitude at C intersects
the circle with diameter AB at M and N . Prove that P, Q, M and N lie on a circle.
2.3. Show that a path on a rectangular grid which starts at the northwest corner, goes
through each point on the grid exactly once, and ends at the southeast corner divides
the grid into two equal halves: (a) those regions opening north or east; and (b) those
regions opening south or west.
•.................................................•.......................•.................•..................•.................•..................................................•........................•........
......................... ......................... ..
•...........................................................................•...............................................................•...........................................................•...............................................................•.......... •..........................................................................•.............. •.........
......................................................................................... ........................ ...
............................ .......
•...............................•.................................•.................................•.............................................................................•.............. •................................•..... •........
.................. ...
.......
•................................................................•....................................................................•.....................................................................•..................................................................................•............. •...............................................................•.....................................................................•........
............................................................. ..............................
............................. ..............
•...............................•.................................•.................................•...................................•..... •................................•................................•...
(The figure above shows a path meeting the conditions of the problem on a 5 × 8 grid.
The shaded regions are those opening north or east while the rest open south or west.)
1995/96
1.1. Let P be a point on the side AB of a square ABCD and Q a point on the side BC.
Let H be the foot of the perpendicular from B to P C. Suppose that BP = BQ.
Prove that QH is perpendicular to HD.
1.2. For each positive integer k, prove that there is a perfect square of the form n2k − 7,
where n is a positive integer.
1.3. Let S = {0, 1, 2, . . . , 1994}. Let a and b be two positive numbers in S which are
relatively prime. Prove that the elements of S can be arranged into a sequence
s1 , s2 , s3 , . . . , s1995 such that si+1 − si ≡ ±a or ± b (mod 1995) for i = 1, 2, . . . , 1994.
2.1. Let C, B, E be three points on a straight line l in that order. Suppose that A and D
are two points on the same side of l such that
(i) ∠ACE = ∠CDE = 90o and
(ii) CA = CB = CD.
Let F be the point of intersection of the segment AB and the circumcircle of 4ADC.
Prove that F is the incentre of 4CDE.
2.2. Prove that there is a function f from the set of all natural numbers to itself such that
for any natural number n, f (f (n)) = n2 .
2.3. Let S be a sequence n1 , n2 , . . . , n1995 of positive integers such that n1 + · · · + n1995 =
m < 3990. Prove that for each integer q with 1 ≤ q ≤ m, there is a sequence
ni1 , ni2 , . . . , nik , where 1 ≤ i1 < i2 < · · · < ik ≤ 1995, ni1 + · · · + nik = q and k
depends on q.
1996/97
1.1. Let ABC be a triangle and let D, E and F be the midpoints of the sides AB, BC
and CA respectively. Suppose that the angle bisector of ∠BDC meets BC at the
point M and the angle bisector of ∠ADC meets AC at the point N . Let M N and
CD intersect at O and let the line EO meet AC at P and the line F O meet BC at
Q. Prove that CD = P Q.
1.2. Let an be the number of n-digit integers formed by 1, 2 √ and 3 which do not contain
any consecutive 1’s. Prove that an is equal to ( 21 + √13 )( 3 + 1)n rounded off to the
nearest integer.
1.3. Let f : R −→ R be a function from the set R of real numbers to itself. Find all such
functions f satisfying the two properties:
(a) f (x + f (y)) = y + f (x) for all x, y ∈ R,
f (x)

(b) the set : x is a nonzero real number is finite.
x
2.1. Four integers a0 , b0 , c0 , d0 are written on a circle in the clockwise direction. In the first
step, we replace a0 , b0 , c0 , d0 by a1 , b1 , c1 , d1 , where a1 = a0 − b0 , b1 = b0 − c0 , c1 =
c0 − d0 , d1 = d0 − a0 . In the second step, we replace a1 , b1 , c1 , d1 by a2 , b2 , c2 , d2 ,
where a2 = a1 − b1 , b2 = b1 − c1 , c2 = c1 − d1 , d2 = d1 − a1 . In general, at the
kth step, we have numbers ak , bk , ck , dk on the circle where ak = ak−1 − bk−1 , bk =
bk−1 − ck−1 , ck = ck−1 − dk−1 , dk = dk−1 − ak−1 . After 1997 such replacements, we
set a = a1997 , b = b1997 , c = c1997 , d = d1997 . Is it possible that all the numbers
|bc − ad|, |ac − bd|, |ab − cd| are primes ? Justify your answer.
2.2. For any positive integer n, evaluate
b n+1
2
c
X n−i+1
,
i
i=0
m!

m
where = and b n+1
2 c is the greatest integer less than or equal to
k k!(m − k)!
n+1
2 .
2.3. Suppose the numbers a0 , a1 , a2 , . . . , an satisfy the following conditions:
1 1
a0 = , ak+1 = ak + a2k for k = 0, 1, . . . , n − 1.
2 n
1
Prove that 1 − < an < 1.
n
1997/98
1.1. Let ABCDEF be a convex hexagon such that AB = BC, CD = DE and EF = F A.

Prove that
BC DE FA 3
+ + ≥ .
BE DA F C 2
When does the equality occur?
1.2. Let n ≥ 2 be an integer. Let S be a set of n elements and let Ai , 1 ≤ i ≤ m, be
distinct subsets of S of size at least 2 such that
Ai ∩ Aj 6= ∅, Ai ∩ Ak 6= ∅, Aj ∩ Ak 6= ∅ imply Ai ∩ Aj ∩ Ak 6= ∅.
Show that m ≤ 2n−1 − 1.

1.3. Suppose f (x) is a polynomial with integer coefficients satisfying the condition
0 ≤ f (c) ≤ 1997 for each c ∈ {0, 1, . . . , 1998}.
Is is true that f (0) = f (1) = · · · = f (1998)?
2.1. Let I be the centre of the inscribed circle of the non-isosceles triangle ABC, and let
the circle touch the sides BC, CA, AB at the points A1 , B1 , C1 respectively. Prove
that the centres of the circumcircles of 4AIA1 , 4BIB1 and 4CIC1 are collinear.
2.2. Let a1 ≥ · · · ≥ an ≥ an+1 = 0 be a sequence of real numbers. Prove that
v
u n n √
uX X √ √
t ak ≤ k ak − ak+1 .
k=1 k=1
2.3. Let p and q be distinct positive integers. Suppose p2 and q 3 are terms of an infinite
arithmetic progression whose terms are positive integers. Show that the arithmetic
progression contains the sixth power of some integer.
1998/99
1.1. Find all integers m for which the equation
x3 − mx2 + mx − (m2 + 1) = 0
has an integer solution.

1.2. Is it possible to use 2 × 1 dominoes to cover a 2k × 2k checkerboard which has 2
squares, one of each colour, removed?
1.3. Find the number of 16-tuples (x1 , x2 , . . . , x16 ) such that
(i) xi = ±1 for i = 1, . . . , 16,

(ii) 0 ≤ x1 + x2 + · · · + xr < 4, for r = 1, 2, . . . , 15,
(iii) x1 + x2 + · · · + x10 = 4.
2.1. Let M and N be two points on the side BC of a triangle ABC such that BM =
M N = N C. A line parallel to AC meets the segments AB, AM and AN at the
points D, E and F respectively. Prove that EF = 3DE.
2.2. Find all possible values of
x−p −x − 1

+ ,
p p
where x is a real number and p is a nonzero integer.
Here bzc denotes the greatest integer less than or equal to z.
2.3. Let f (x) = x1998 − x199 + x19 + 1. Prove that there is an infinite set of prime numbers,
each dividing at least one of the integers f (1), f (2), f (3), f (4), · · · .
1999/2000
1.1. In a triangle ABC, AB > AC, the external bisector of angle A meets the circumcircle
of triangle ABC at E, and F is the foot of the perpendicular from E onto AB. Prove
that 2AF = AB − AC.
1.2. Find all prime numbers p such that 5p + 12p is a perfect square.
1.3. There are n blue points and n red points on a straight line. Prove that the sum of
all distances between pairs of points of the same colour is less than or equal to the
sum of all distances between pairs of points of different colours.
2.1. Find all functions f : R −→ R such for any x, y ∈ R,
(x − y)f (x + y) − (x + y)f (x − y) = 4xy(x2 − y 2 ).
2.2. In a triangle ABC, ∠C = 60◦ , D, E, F are points on the sides BC, AB, AC respec-
tively, and M is the intersection point of AD and BF . Suppose that CDEF is a
rhombus. Prove that DF 2 = DM · DA.
Let n be any integer ≥ 2. Prove that 1/pq = 1/2, where the summation is over all
P
2.3.
integers p, q which satisfy 0 n, (p, q) = 1.
2000/2001
1.1. Let a, b, c, d be four positive integers such that each of them is a difference of two
squares of positive integers. Prove that abcd is also a difference of two squares of
positive integers.
1.2. Let P, Q be points taken on the side BC of a triangle ABC, in the order B, P, Q, C.
Let the circumcircles of 4P AB, 4QAC intersect at M (6= A) and those of 4P AC,
4QAB at N . Prove that A, M, N are collinear if and only if P and Q are symmetric
in the midpoint A0 of BC.
1.3. A game of Jai Alai has eight players and starts with players P1 and P2 on court and
the other players P3 , P4 , P5 , P6 , P7 , P8 waiting in a queue. After each point is played,
the loser goes to the end of the queue; the winner adds 1 point to his score and stays
on the court; and the player at the head of the queue comes on to contest the next
point. Play continues until someone has scored 7 points. At that moment, we observe
that a total of 37 points have been scored by all eight players. Determine who has
won and justify your answer.
2.1. In the acute triangle ABC, let D be the foot of the perpendicular from A to BC,
let E be the foot of the perpendicular from D to AC, and let F be a point on the
line segment DE. Prove that AF is perpendicular to BE if and only if F E/F D =
BD/DC.
2.2. Determine all the integers n > 1 such that
n
X n−1
X
x2i ≥ xn xi
i=1 i=1
for all real numbers x1 , x2 , . . . , xn .

2.3. Let L(n) denote the least common multiple of {1, 2 . . . , n}.
(i) Prove that there exists a positive integer k such that
L(k) = L(k + 1) = · · · = L(k + 2000).
(ii) Find all m such that L(m + i) 6= L(m + i + 1) for all i = 0, 1, 2.

2001/2002
1.1. Let A, B, C, D, E be five distinct points on a circle Γ in the clockwise order and let
the extensions of CD and AE meet at a point Y outside Γ. Suppose X is a point on
the extension of AC such that XB is tangent to Γ at B. Prove that XY = XB if
and only if XY is parallel DE.
1.2. Let n be a positive integer and (x1 , x2 , . . . , x2n ), xi = 0 or 1, i = 1, 2, . . . , 2n be a
sequence of 2n integers. Let Sn be the sum
Sn = x1 x2 + x3 x4 + · · · + x2n−1 x2n .
If On is the number of sequences such that Sn is odd and En is the number of

sequences such that Sn is even, prove that
On 2n − 1
= n .
En 2 +1
1.3. For every positive integer n, show that there is a positive integer k such that
2k 2 + 2001k + 3 ≡ 0 (mod 2n ).
2.1. Let x1 , x2 , x3 be positive real numbers. Prove that
(x21 + x22 + x23 )3

≤ 3.
(x31 + x32 + x33 )2
2.2. For each real number x, bxc is the greatest integer less than or equal to x. For
example b2.8c = 2. Let r ≥ 0 be a real number such that for all integers m, n, m|n
implies bmrc|bnrc. Prove that r is an integer.
2.3. Find all functions f : [0, ∞) −→ [0, ∞) such that f (f (x)) + f (x) = 12x, for all x ≥ 0.
2002/2003
1.1. Determine whether there exists a positive integer n such that the sum of the digits
of n2 is 2002.
1.2. Three chords AB, CD and EF of a circle intersect at the midpoint M of AB. Show
that if CE produced and DF produced meet the line AB at the points P and Q
respectively, then M is also the midpoint of P Q.
1.3. In how many ways can n2 distinct real numbers be arranged into an n × n array (aij )
such that maxj mini aij = mini maxj aij ?
2.1. Let A = {3 + 10k, 6 + 26k, 5 + 29k, k = 1, 2, 3, 4, · · · }. Determine the smallest

positive integer r such that there exists an integer b with the property that the set
B = {b + rk, k = 1, 2, 3, 4, · · · } is disjoint from A.
2.2. Let M be a point on the diameter AB of a semicircle Γ. The perpendicular at M
meets the semicircle Γ at P . A circle inside Γ touches Γ and is tangent to P M at Q
and AM at R. Prove that P B = RB.
2.3. Determine all functions f : Z −→ Z, where Z is the set of integers, such that
f (m + f (f (n))) = −f (f (m + 1)) − n
for all integers m and n.
2003/2004
1.1. Let N be the fourth root of a product of 8 consecutive positive integers. Prove that
the greatest integer less than or equal to N is even.
1.2. Let Γ be a circle with center I, the incenter of triangle ABC. Let D, E, F be points of
intersection of Γ with the lines from I that are perpendicular to the sides BC, CA, AB
respectively. Prove that AD, BE, CF are concurrent.
1.3. Find all pairs of integers (x, y) satisfying x5 + y 5 = (x + y)3 .
2.1. Let A, B, C, D be four distinct points arranged in order on a circle. The tangent to
the circle at A meets the ray CB at K and the tangent to the circle at B meets the
ray DA at H. Suppose BK = BC and AH = AD. Prove that the quadrilateral
ABCD is a trapezium.
2.2. Determine the smallest constant k > 0 such that
ab bc ca
+ + ≤ k(a + b + c),
a + b + 2c b + c + 2a c + a + 2b
for all a, b, c > 0.
2.3. Consider an n × n square lattice with points colored either black or white. A square
path is a closed path in the shape of a square with edges parallel to the edges of the
lattice. Let M (n) be the minimum number of black points needed for an n × n square
lattice so that every square path has at least one black point on it. Prove that
2 2
(n − 1)2 ≤ M (n) ≤ n2 .
7 7
(*The numbering 1.1 refers to the first question of the selection test in the first day, while 2.1 refers
to the first question of the selection test in the second day.)
Solutions to National Team Selection Tests
Prepared by Tay Tiong Seng and Wong Yan Loi
1994/95
1.1 It can be proved by induction that f (n) is the number of ones in the binary represen-
tation of n.
(i) There can be at most 10 ones in the binary representation of a natural number if it is
less than or equal to 1994 = 11111001010(2) . Hence M = 10.
(ii) For any natural number n less than or equal to 1994, f (n) = 10 if and only if n is
1023 = 1111111111(2) ,
1535 = 10111111111(2) ,
1791 = 11011111111(2) ,
1919 = 11101111111(2) ,
1983 = 11110111111(2) .
1.2. Stewart’s theorem. In 4ABC, D is a point on BC such that AD bisects ∠A. Then
AB : BD = AC : CD.
1st solution
A
...............
....... .... .....
....... ... .. ....
....... .... ..... .........
.......
.......
.
......
....
.
...
.
... ..
....
.....
R
.....
.....
....... ... .... . .... .. .........
....
........ .
... . .
.. ..
.....................................................................................................................
.
B P Q C
AB BP
Applying Stewart’s theorem to 4ABQ, we have = .
AQ PQ
BC AB
Given BP · CQ = BC · P Q, it follows that = .
CQ AQ
Now let R be the point on AC such that QR is parallel to BA.
AB BC AB
Then = = .
RQ CQ AQ
Hence RQ = AQ and ∠QAR = ∠QRA.
Therefore ∠P AC = ∠P AQ + ∠QAR = 12 (∠BAQ + ∠QAR + ∠QRA) = π2 .
2nd solution
CB PB AB
Since = = , by Stewart’s theorem, AC is the external angle bisector of ∠BAQ.
CQ PQ AQ
π
Hence ∠P AC = .
2
P
1.3. (i) Note that twice the total number of clappings is equal to x∈S f (x) which cannot
be the odd number 2 + 3 + 4 + · · · + 1995.
(ii) Let n ≥ 2. For a group Sn of 4n − 2 students, the following configuration gives an

example in which {f (x) | x ∈ Sn } = {2, 4, 5, ...., 4n}.
....... ....... ....... ....... ....
. ....... ... ...
................. ...... ..... ....
.............
.................
..... .... ...
.... ... . .
..... 4n−6....
3 8 ...
...
.......... ....... ..
..
..
.... .
.... 2n−3
......... ......... .
............. ...........
.....
.. .....
.... ..... .............
.......................... ..... .....
.... ... ... ...
..... ... ..... 4n−4....
6 ...
..... . . .... ...
. ....... ....... . .
. ..
.................. ....... ...
3 ...
. . . .. ...
...
...
2n−1
... ... ..
. . ................. ....... ..........
.... . . ...
. ...... ...
... ..
..... . .
... 4n−2....
... 4
...... ......... .
...
... ..
. ...
............. . .............. .
.. ...
.. ...
1 ..
... ...
... 2n−1
.... ...
...................... ......................
. ... .. ...
.... .. .
. .
.... 4n−1...
...
... 2
...... .......
..
. ...
...... .......
..
........... ...........
... ..
... ..
... ...
1 ...
... ....
.. 2n
.......... ...............
...... ........ ...... ...
... ... ... ..
..... . .... 4n ...
... 5
...... ..... .
. .. ...
...... ........
.
............. .. ..... ..
... ... .
... ....
... ...
4 ... ... 2n
..................... .......................
.... ...
... .... ...
..... . . ..
... 4n−3... .
7 ...
.... ..
...
.........................
...
.. ...
..
......................
.....
..... ..
. ......
...... ................... .................. .........
...... .... ..... ....
3 .. ... ... . 2n−3
.... . .... 4n−5...
9 ...
...... ...... ....... ..... . .
..
.... ...... .....
..
........... ....... .
...... .......... ...
....... ....... .. ....... ....... .......
.....
Each circle in the diagram represents a student x and the number in the circle represents
f (x). The number on each edge represents the number of times the two adjacent students
clap hands with each other. Taking n = 499 gives an example of the problem.
2.1. (i) Denote the function f (x) composed with itself n times by f (n) (x). Also let g0 (x)
be the identity function. Note that f (2) (x) is strictly increasing for x > 0. We shall prove
by induction on n that gn (x) is strictly increasing for x > 0. It can easily be checked that
g1 (x) is strictly increasing for x > 0.
Suppose for n ≥ 2, g1 (x), ..., gn−1 (x) are strictly increasing. Let x > y > 0. We have
gn (x) − gn (y) = (x − y) + (f (x) − f (y)) + (f (2) (x) − f (2) (y)) + · · · + (f (n) (x) − f (n) (y))
= (g1 (x) − g1 (y)) + (gn−2 (f (2) (x)) − gn−2 (f (2) (y))) > 0.
By induction, gn (x) is strictly increasing.
F1 Fi Fi+1 F1 Fn+1
(ii) Note that = 1 and f ( )= . Hence + ··· + = gn (1).
F2 Fi+1 Fi+2 F2 Fn+2
B
2.2. Since 4ADP is similar to 4AP C, we have ............
... .... .......
.
... .. ........
AP /AD = AC/AP . Hence AP 2 = AD · AC = .... .......P
............................
...................................
.... .......
...... .... .. ......... ..... ......
.......... ..... ........... .......... ........ ........
(BD cot A) · AC = 2(ABC) cot A, where (ABC) is ...
... ... ..... ...
.
...... .....
...... ..... .....
.
... ....... .... ...... ..... ..
the area of 4ABC. Similarly, AM 2 = ... ........ ... ...... .... ...
p2(ABC) cot A.
........... ..
............. ............. ........... ...
......... ..
...... .. ..
...........................................................................................................
Hence AP = AQ = AM = AN = 2(ABC) cot A. A ... ... .. C
...
... D .
.
....
...
..
... ..
This shows that P,
pQ, M, N lie on the circle centered
...
...
.....
.
.
....
.. .....
...
. ..
.....
...... .... .....
at A with radius 2(ABC) cot A. ........ .. .....
.......
..................................
2.3. Let such a path be given. First the following facts are observed.
(i) The number of edges of the path is nm − 1.
(ii) By induction, each region with s squares is adjacent to 2s + 1 edges of the path.
(iii) Each edge on the north or east side of the grid which is not included in the path
corresponds to exactly one shaded region.
Let the number of shaded regions be k and let s1 , s2 , ..., sk be the number of squares in each
of these regions. From (iii), it follows that the number of edges of the path on the north
and east side of the grid is (m − 1) + (n − 1) − k. Hence by (ii), the total number of edges
k
X
of the path is (2si + 1) + [(m − 1) + (n − 1) − k]. By (i), we have
i=1
k
X
(2si + 1) + [(m − 1) + (n − 1) − k] = nm − 1.
i=1
k
X 1
From this the total number of shaded squares is si = (m − 1)(n − 1).
i=1
2
This problem appears in the American Mathematical Monthly. (See The American Mathe-
matical Monthly, Vol.104, No.6, June-July 1997, p572-573.)
1995/96
B .........................................Q
............................................................ C
1.1. Let BH intersect AD at F . Then 4AF B is congruent to .........
.
...
...
...
.........
.........
.. ...
...
... ... ... ......... ...
... .... ..... ................
4BP C. Hence AF = BP = BQ. Therefore F D = QC and ... .... .... .............
... ..................
...
...
...
............. .......... ...
QCDF is a rectangle. Since ∠CHF = 90o , the circumcircle P H ...
...
... .......
......
... ......
...
...
...
... ...
...... ...
... ......
of the rectangle QCDF passes through H. As QD is also a ...
...
... ......
......
.
.......
...
...
...
...
diameter of this circle, we have ∠QHD = 90o . ... ...... ...
... ......
... ...... ....
... . ... ..
.
........................................................................................................
A F D
1.2. Suppose there is a perfect square a2 of the form n2k − 7 for some positive integer n.
Then a is necessarily odd. We shall show how to produce a perfect square of the form
n0 2k+1 − 7 for some positive integer n0 . If n is even, then a2 = (n/2)2k+1 − 7 is of the
required form. Suppose that n is odd. We wish to choose a positive integer m such that
(a + m)2 is of the desired form.
Consider (a + m)2 = a2 + 2am + m2 = −7 + n2k + m(m + 2a). If we choose m = 2k−1 , then
m(m + 2a) is an odd multiple of 2k . Consequently, (a + m)2 is of the form n0 2k+1 − 7 for
some positive integer n0 . Now the solution of this problem can be completed by induction
on k.
1.3. Let p be the smallest positive integer such that pa ≡ 0 (mod 1995), i.e. pa = 1995k for
some positive integer k. Let q = 1995/p. Then q is an integer and it divides a. We claim
that
S = {ma + nb (mod 1995) | m = 0, 1, . . . , p − 1, n = 0, 1, . . . , q − 1}

First note that there are pq = 1995 elements in the set on the right hand side. It suffices
to prove that the elements are distinct. Suppose that ma + nb ≡ m0 a + n0 b (mod 1995).
Then (m − m0 )a + (n − n0 )b = 1995` for some integer `. Since q divides 1995 and a, and
q is relatively prime to b, we have q divides (n − n0 ). But |n − n0 | ≤ q − 1, so n − n0 = 0.
Consequently, m = m0 . This completes the proof of the claim.
Consider the following sequence:
a, a, . . . , a, b, −a, −a, . . . , −a, b, a, a, . . . , a, b, . . . , (−1)q a, (−1)q a, . . . , (−1)q a, b
| {z } | {z } | {z } | {z }
p terms p terms p terms p terms
In this sequence, there are q blocks of a, a, . . . , a, b or −a, −a, . . . , −a, b making a total of
pq = 1995 terms. For each i = 1, 2 . . . , 1995, let si be the sum of the first i terms of this
sequence. Then by the result above, {s1 , s2 . . . , s1995 } = S and si+1 − si = ±a or ± b
(mod 1995).
2.1. Since ∠CDF = ∠CAF = 45o , we have ∠F DE = A ........

............
.............
...........................
........
......
......
..... .. ..... .....
..... .... ........
∠CDE − ∠CDF = 45o = ∠CDF . Hence DF bisects .
.
...
... ..
. ... .....
...
...
......
.
......
....
...
...
...
..
. . .
...... ...
∠CDE. As CB = CD, we have ∠CBD = ∠CDB. Hence ....
... ...
...
...
..
.....
........
...
...
...
.... ... .....
D
∠F BD = ∠CBD − 45o = ∠CDB − 45o = ∠F DB. There- ...
...
...
...
...
...
.....
.....
..... .
....
..... ........... ............
...
.............
... ... ......... ... . ...
.
fore F D = F B. This shows that 4BCF is congruent to ...
...
...
...
...
... .....
......
.. .
...... ......... ....... .. .....
.
........ .. ..
. .
.
... ...... .. ........ . ..
...... .... ........ ........ .. .....
4DCF . Hence ∠BCF = ∠DCF and CF bisects ∠DCE. ....
.....
.....
...... .... ............... .... ..
...
.
... ...
..
...... .. ....
....
.F.
..... ..... .. ..
..... ...............................................
....... .............................................................. ... .
Therefore F is the incentre of 4CDE. ....................
................................. B E
C
2.2. Let N be the set of all natural numbers. Let A = {n2 | n ∈ N}. Let N \ A =
{n1 , n2 , n3 , . . . }. Define f as follows:


 1 if n = 1,
n2i if n = n2i−1 , i = 1, 2, . . .





 2
f (n) = n2i−1 if n = n2i , i = 1, 2, . . .
k 2k
n22i

if n = n2i−1 , k = 1, 2, . . .





 2k+1 k
n2i−1 if n = n22i ,

k = 1, 2, . . .
Then f : N −→ N satisfies the requirement f (f (n)) = n2 .
(Note: The function above comes from the following consideration. First, f (1) must be 1.
Let f (2) = 3. Then f (3) = 22 , f (22 ) = 32 , f (32 ) = 24 etc.. Next, let f (5) = 6. Continuing
as before, we have f (6) = 52 , f (52 ) = 62 , f (62 ) = 54 etc..)
2.3. Let N = {1, 2, ..., 1995}. Let q be an integer with 1 ≤ q ≤ m. We shall prove the
following statement S(q) by induction (on q):
P
S(q): There exists a subset Iq of N such that i∈Iq ni = q.
S(1) is true because one of the ni ’s must be 1. Now assume that for some q with 1 ≤ q < m,
S(i) is true for i ≤ q. Then |Iq | ≤ q and 1994.
If ni > q + 1 for all i ∈ N \ Iq , then i∈N ni ≥ q + (q + 2)(1995 − |Iq |) = (1996 − |Iq |)q +
P
2(1995 − |Iq |) ≥ 2q + 2(1995 − |Iq |) ≥ 2q + 2(1995 − q) = 3990, which is a contradiction.

Hence, there exists j ∈ N \Iq such that nj ≤ q +1. Let a = min{ni : i 6∈ Iq }. Then a ≤ q +1
and a − 1 ≤ q. Thus S(a − 1) is true. By the choice of a, there exists J ⊆ Iq such that
a − 1 = i∈J ni . Therefore, q + 1 = q + a − (a − 1) = i∈Iq \J ni + a. Thus, S(q + 1) is true.
P P
This problem appears in the American Mathematical Monthly with 1995 replaced by k and
3990 replaced by 2k. The proof above works for the general case too. See (The American
Mathematical Monthly, Vol.105, No.3, March 1998, pg 273-274.)
1996/97
1.1. Since DM and DN are angle bisectors of

∠BDC and ∠ADC respectively, by Stewart’s theo-
rem, we have
BM DB AN AD C .......
...........
= and = . .
.
..... .. ...
..... .... ....
MC DC NC DC .....
.
.... .. ....
.... .... ...
...
..... ... ...
.... ..
BM AN .
....
.
...
.
.
..
..
.
.
...
...
As AD = DB, we have = . F .....
.
..
..
.
.. ...
E
..
..................................................................................
. .
MC NC . .
..
.
.
..
.. .
.... .............. O .
.
.
........ .. ........... . ....... .
...... ....
..
.
N ..... .... .... .... ....................................... .... .... ..... .......
.. .. ...
.
M
..... ...
Hence N M//AB and 4ABC ∼ 4N M C. .
..
..... . ..........
.... ....................
.. ..... ..
... .. ..........
...
. ...........
. ........ ....
..
........ .... . .....
................ ... .... ...... ..
P .
....................................................................................................................................
....
Q
AB AC BC .
....
.
...
. ..
. ..
... .
.
.
.
. .... .... ...
...
Therefore = = . . ...
..................................................................................................................................................
.
NM NC MC A D B
BM DB DB + DC BM + M C BC AB
Since = , we have = = = .
MC DC DC MC MC NM
F E + DC 2F E
On the other hand, F E = 12 AB = DB. Therefore, = .
DC NM
1 1 2
Consequently, + = .
F E DC NM
Applying Menelaus’s Theorem to 4CM N for the lines EP and F Q and using the fact that
OM = ON , we have
CP OM CE CE CQ ON F C FC
= · = and = · = .
PN ON M E ME QM OM F N FN
CE FC CQ CP
Since F E//AB//N M , we have = . Therefore = so that F E//P Q.
ME FN QM PN
Hence P QEF is a trapezoid and O is the intersection point of its two diagonals.
1 1 2
From this, it follows that + = . Consequently, P Q = DC.
FE PQ NM
1.2. It can be shown that an satisfies the recurrence relation: an = 2an−1 + 2an−2 with
a1 = 3 and a2 = 8. Solving this difference equation gives
1 1 √ 1 1 √
an = ( + √ )(1 + 3)n + (−1)n+1 ( √ − )( 3 − 1)n .
2 3 3 2
√
Next we shall show that ( √13 − 12 )( 3 − 1)n < 0.5 for n ≥ 1. This is because
1 1 √ 1 1 √ 1
for n ≥ 1, 0 < ( √ − )( 3 − 1)n ≤ ( √ − )( 3 − 1) < (1 − )(2 − 1) = 0.5.
3 2 3 2 2
1 1 √
Thus an = ( + √ )(1 + 3)n rounded off to the nearest integer.
2 3
1.3. 1st solution
Let x ∈ R. By letting x = y + f (0), we obtain
f (f (x)) = f (f (y + f (0))) = f (0 + f (y)) = y + f (0) = x.
Hence for any t1 , t2 ∈ R, f (t1 + t2 ) = f (t1 + f (f (t2 ))) = f (t1 ) + f (t2 ).

Next, consider any positive integer m such that m 6= −f (x). We have
f (m + f (x)) x + f (m) x + mf (1)

= = .
m + f (x) m + f (x) m + f (x)
f (t)
Since the set { | t 6= 0} is finite, there exist distinct positive integers m1 , m2 with
t
6 −f (x) such that
m1 , m2 =
f (m1 + f (x)) f (m2 + f (x))

= .
m1 + f (x) m2 + f (x)
x + m1 f (1) x + m2 f (1)
Hence = . From this, we have f (x)f (1) = x.
m1 + f (x) m2 + f (x)
By letting x = 1, we obtain [f (1)]2 = 1 so that f (1) = ±1. Consequently, f (x) = ±x. Also
the functions f (x) = x and f (x) = −x clearly satisfy the two given conditions.
2nd solution
(i) First we prove that f (0) = 0. Putting x = 0 = y, we have f (f (0)) = f (0). If f (0) = a,
then f (0) = f (f (0)) = f (a). Thus a + f (0) = f (0 + f (a)) = f (f (0)) = f (0), whence a = 0.
(ii) Putting x = 0, we have f (f (y) = y for all y.
(iii) We will prove that f (x) = ±x for all x.
Suppose for some p, f (p) = cp for some constant c 6= ±1. Then f (p + f (p)) = p + f (p).
Let q = p + f (p). Then q 6= 0 and f (q) = q. Thus f (q + f (q)) = q + f (q) and f (2q) = 2q.
Inductively we have f (nq) = nq for any positive integer n. Now f (nq + f (p)) = p + f (nq).
So f (nq + cp) = p + nq. Thus f (nq + cp)/(nq + cp) = 1 − (c − 1)p/(nq + cp). Since c − 1 6= 0
and there are infinitely many choices for n so that nq + cp 6= 0, this gives an infinite number
of members in the set {f (x)/x} contradicting the second condition. Thus c = ±1.
(iv) For f (p) = p, we will prove that f (x) = x for all x.
If f(−p) = p, then −p = f (f (−p)) = f (p) = p which is impossible. Thus f (−p) = −p.
Suppose there exists r such that f (r) = −r. Then f (r+f (p)) = p+f (r), i.e., f (r+p) = p−r.
Therefore f (r + p)/(r + p) = (p − r)/(r + p) 6= ±1. (Note that the denominator is not zero.)
(v) From the above we conclude that either f (x) = x for all x or f (x) = −x for all x.
Clearly these functions satisfy the two given conditions. Thus these are the only two
functions required.
2.1. Let a, b, c, d represent the numbers at any stage subsequent to the initial one. Then
a + b + c + d = 0 so that d = −(a + b + c). It follows that
bc − ad = bc + a(a + b + c) = (a + b)(a + c),

ac − bd = ac + b(a + b + c) = (a + b)(b + c),
ab − cd = ab + c(a + b + c) = (a + c)(b + c).
Hence, |(bc − ad)(ac − bd)(ab − cd)| = (a + b)2 (a + c)2 (b + c)2 .

Therefore the product of the three quantities |bc − ad|, |ac − bd|, |ab − cd| is the square of
an integer. However the product of three primes cannot be the square of an integer, so the
answer to the question is “NO”.

n−i+1
2.2. is equal to the number of i-subsets of the set S = {1, 2, . . . , n} containing
i
no consecutive integers. Hence the required sum is just the number an of subsets of S
containing no consecutive integers. It can be shown easily that an satisfies the recurrence
relation: an = an−1 + an−2 with a0 = 1 and a1 = 2. This can also be derived from the
identity:
n−i+1 (n − 1) − i + 1 (n − 2) − (i − 1) + 1
= + .
i i i−1
From this, we obtain
b n+1
2
c √ √ !n √ √ !n
5+3 5 1+ 5 5−3 5 1− 5

X n−i+1
= + .
i 10 2 10 2
i=0
2.3. We shall prove by induction on k that

n+1 n
< ak < for k = 1, 2, . . . , n.
2n − k + 2 2n − k
For k = 1, we have
1 2 2n + 1
a1 = a0 + a = ,
n 0 4n
Hence
n+1 n
< a1 < ,
2n + 1 2n − 1
so the induction hypothesis is true for k = 1.
Now suppose the induction hypothesis is true for k = r < n, then
1 1

ar+1 = ar + a2r = ar 1 + ar .
n n
Hence we have
n+1 1 n+1

ar+1 > 1+ ·
2n − r + 2 n 2n − r + 2
n+1 n+1
> = .
2n − r + 1 2n − (r + 1) + 2
On the other hand,

n 1 n n(2n − r + 1) n

ar+1 < 1+ · = 2
< ,
2n − r n 2n − r (2n − r) 2n − (r + 1)
since (2n − r)2 > (2n − r + 1)(2n − (r + 1)). Hence the induction hypothesis is true for
k = r + 1. This completes the induction step.
When k = n, we get
1 1 n+1 n
1− <1− = < an < = 1,
n n+2 n+2 2n − n
the required inequality.
1997/98
1.1. Let AC = a, CE = b, AE = c. Applying the Ptolemy’s Theorem1 for the quadrilateral

ACEF we get
AC · EF + CE · AF ≥ AE · CF.
FA c DE b BC a
Since EF = AF , it implies FC ≥ a+b . Similarly, DA ≥ c+a and BE ≥ b+c . It follows that
BC DE FA a b c 3
+ + ≥ + + ≥ . (1)
BE DA F C b+c c+a a+b 2
The last inequality is well known2 . For equality to occur, we need equality to occur at
every step of (1) and we need an equality each time Ptolemy’s Theorem is used. The latter
happens when the quadrilateral ACEF , ABCE, ACDE are cyclic, that is, ABCDEF is
a cyclic hexagon. Also for the equality in (1) to occur, we need a = b = c. Hence equality
occurs if and only if the hexagon is regular.
1.2. We will prove the statement by induction on n. It obviously holds for n = 2. Assume
that n > 2 and that the statement is true for any integer less than n. We distinguish two
cases.
Case 1. There are no i and j such that Ai ∪ Aj = S and |Ai ∩ Aj | = 1.
Let x be an arbitrary element in S. The number of sets Ai not containing x is at most
2n−2 − 1 by the induction hypothesis. The number of subsets of S containing x is 2n−1 .
At most half of these appear as a set Ai , since if x ∈ Ai , then there is no j such that
Aj = (S − Ai ) ∪ {x} for otherwise |Ai ∩ Aj | = 1. Thus the number of sets Ai is at most
2n−2 − 1 + 2n−2 = 2n−1 − 1.
Case 2. There is an element x ∈ S such that A1 ∪ A2 = S and A1 ∩ A2 = {x}.
Let |A1 | = r + 1 and |A2 | = s + 1. Then r + s = n − 1. The number of sets Ai such that
Ai ⊆ A1 is at most 2r − 1 by the induction hypothesis. Similarly the number of sets Ai such
that Ai ⊆ A2 is at most 2s − 1.
If Ai is not a subset of A1 and A2 , then A1 ∩ Ai 6= ∅, A2 ∩ Ai 6= ∅. Since A1 ∩ A2 6= ∅, we
have A1 ∩ A2 ∩ Ai 6= ∅. Thus A1 ∩ A2 ∩ Ai = {x}. Thus Ai = {x} ∪ (Ai − A1 ) ∪ (Ai − A2 ),
and since the nonempty sets Ai − A1 and Ai − A2 can be chosen in 2s − 1 and 2r − 1 ways,
respectively, the number of these sets is at most (2s − 1)(2r − 1). Adding up these partial
results we obtain the result that the number of Ai ’s is at most 2n−1 − 1.
1.3. 1st solution

Note that for any a, b, we have (a − b)| ± (F (a) − F (b)). Thus 1998 divides F (1998) − F (0),
whence F (1998) = F (0) as |F (1998) − F (0)| ≤ 1997. Also we have 4 = 1998 − 1994 divides
F (1994) − F (1998) = F (1994) − F (0), and 1994|(F (1994) − F (0)). Thus LCM (4, 1994) =
3988 divides F (1994) − F (0) which implies F (1994) = F (0). By reversing the role of 4 and
1998, we have F (4) = F (0). By considering 5 and 1993, we also have F (1993) = F (5) =
F (0). Then for any a, 1 ≤ a ≤ 1997, we have (x − a)|F (0) − F (a) for x = 4, 5, 1993, 1994.
The least common multiplier of the 4 numbers x − a is larger than 1998. Thus F (a) = F (0).
2nd solution
We shall prove that the statement holds for any integer k ≥ 4, not just k = 1998. Consider
any polynomial F (x) with integer coefficients satisfying the given inequality 0 ≤ F (c) ≤ k
for every c ∈ {0, 1, . . . , k + 1}. Note that F (k + 1) = F (0) because F (k + 1) − F (0) is a
multiple of k + 1 not exceeding k in absolute value. Hence
F (x) − F (0) = x(x − k − 1)G(x)
where G(x) is a polynomial with integer coefficients. Thus
k ≥ |F (c) − F (0)| = c(k + 1 − c)|G(c)| for each c ∈ {1, 2 . . . , k}. (2)
The inequality c(k + 1 − c) > k holds for each c ∈ {1, 2, . . . , k − 1} which is not an empty
set if k ≥ 3. Thus for any c in this set, |G(c)| < 1. Since G(c) is an integer, G(c) = 0. Thus
2, 3, . . . , k − 1 are roots of G(x), which yields
F (x) − F (0) = x(x − 2)(x − 3) · · · (x − k + 1)(x − k − 1)H(x). (3)
We still need to prove that H(1) = H(k) = 0. For both c = 1 and c = k, (3) implies that
k ≥ |F (c) − F (0)| = (k − 2)! · k · |H(c)|.
Now (k − 2)! > 1 since k ≥ 4. Therefore |H(c)| < 1 and hence H(c) = 0.
For k = 1, 2, 3 we have the following counterexamples:
F (x) = x(2 − x) for k = 1
F (x) = x(3 − x) for k = 2
F (x) = x(4 − x)(x − 2)2 for k = 3
2.1. 1st solution

C
.....................
............ ..... .... .............
............
............. ... ....
. .......
....................... ... ...
.......
.......
.......................
..
.............
.
.
.
...
.
.. ...
... I .......
.......
.......
................... ..
. ...... ...
.
.......
.......
.................. .
...
... .... ... .... ....
. .....
........ . .... . ... ... ....
...
......... .. . .... .... ............
................... .
... ....
. ..
. .... .... ........
......... ... ... ... .... ...
.. ........................................................................................................................................................................ ...................................................................................................................................
..
........ ..
.
C2 A C1 B
Let the line perpendicular to CI and passing through C meet AB at C2 . By analogy,
we denote the points A2 and B2 . It’s obvious that the centres of the circumcircles of
AIA1 , BIB1 and CIC1 are the middle points of A2 I, B2 I and C2 I, respectively. So it’s
sufficient to prove that A2 , B2 and C2 are collinear. Let’s note that CC2 is the exterior
C2 A B2 A
bisector of ∠ACB, and so C 2B
CA
= CB . By analogy B 2C
BA
= BC and A 2B AB
A2 C = AC . Thus
C2 A B2 C A2 B CA BC AB 3
C2 B B2 A A2 C = CB BA AC = 1 and by Menelaus’ Theorem , the points A2 , B2 and C2 are
collinear.
2nd solution
Let A0 , B 0 , C 0 be the midpoints of AI, BI, CI, respectively. Let the perpendicular bisectors
of AI and BI meet at C 00 . A00 and B 00 are similarly defined.
A
........
....... ....... ..............
.................... ...
B 00
.. .... ......... ........ ..... ....
... .... ....... ............. ............
... ......... ....... ............. . ...
.. ..... ....... .. .............. ........ ...
A0
.
. . ................. . ... ..
. .. .. ...
... ...... .. ....... ..
...
. ......
. . ....... ................ ............. .. ... ........ .
...
. .. ....... ....... .
... ............................ .... .................. ...
..................
. .. ......
...................... .... C 00
.............
..
... ... ....
... .... ....
.
. ..
....
.... ............
.
............
....... .............
....... ...
.
...
..
.... . ... ................
... .. ............... .... .... ....
. . ...
........ ....
....... ..
.... ....
.... ....
.... .... ... ...
......
............. .. .. ............
. . . .
... .......................................
I ....
....
.......
....... ....
....... ..
........
.
... ... ....
. .. ...... .. .
..
. . . .
.
. .... .... . .
....
.. .. .......
.... .... . . . ....... .... ............. .. .
...
.......
.... .... ... ..... .. ....... .... ........ .............
............. . ....
.. .......
. .... .... .... ... ... ............. ... ... ... ............. .... ..
..
.......
.......
... .
.. .. .. . .. .
.. . . . .. . ... ..
............... . .. ..
.
... ... .
. ...
. ...........
.
.. .
... ... ...
. .....
. .
.......
.......
.... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... ......... .... .... ............... .... .... .... .... ......... ..... .... .... .... .... .... .... .... .... .... .... . ............. .......
. .
.. . . . . . .. .. .. .
. .............
A000 ....
... . .... ... ..... ... .
. ....
.......
..
B0
.
. .......... . ... . . . .
.............
..... .. . ............. ............
C0
.
. . .
..... .. . ..
. . .. .......... ...
. . . .
. .
.. ............. .......
......... .....
. ... .. ..... .
..............................................................................................................................................................................................................................................................................................................
..
B
...
...
...
...
A1...
...
..
.
...
C
...
... ...
... ...
... ..
... ... .
...
... ...
... ... ...
... ... ...
... ... ..
... .
. .
...
... ...
... ... ...
... ... ...
... ... ..
... .
. ...
.
... ...
... ... ...
...
... ... ...
... .. ...
... ..... ....
... ..
... ... ..
... ... ...
... .. ...
... ... ...
... ... ...
... .. ...
... ... ..
...... ..
..... ..
......
A00
Then the circumcentre A000 of AIA1 is the intersection of B 00 C 00 with B 0 C 0 . Likewise the
circumcentre B 000 of BIB1 is the intersection of A00 C 00 with A0 C 0 and the circumcentre C 000
of CIC1 is the intersection of A00 B 00 with A0 B 0 .
First we note that the circumcentre of AIB lies on the line CI. To prove this, let the
circumcircle of AIB meet CI at another point X. Then ∠XAB = ∠XIB = 12 (∠B + ∠C).
Thus ∠XAI = ∠XAB + ∠BAI = 90◦ . Thus XI is a diameter and the circumcentre which
is C 00 is on the line CI. Similarly, A00 is on AI and B 00 is on BI.
Now we consider the triangles A0 B 0 C 0 and A00 B 00 C 00 . The lines A0 A00 , B 0 B 00 , and C 0 C 00 are
concurrent (at I), thus by Desargues’ Theorem4 , the three points, namely, the intersections
of B 00 C 00 with B 0 C 0 , A00 C 00 with A0 C 0 and A00 B 00 with A0 B 0 are collinear.
3rd solution (By inversion)

Let c be the incircle of 4ABC of radius r. The image of a point X under the inversion
about c is the point X ∗ such that IX·IX ∗ = r2 . Inversion about a circle c has the following
properties:
(a) If X lies on c, then X ∗ = X.
(b) I ∗ = ∞.
(c) If s is a circle intersecting c at two points P, Q and s passes through I, then s∗ is a
straight line passing through P and Q.
Now A∗ = Ao , where Ao is the midpoint of B1 C1 . Also, A1 ∗ = A1 and I ∗ = ∞. Hence,
the inversion of the circumcircle of 4AIA1 is the line A1 Ao . Similarly, the inversion of the
circumcircle of 4BIB1 is the line B1 Bo and the inversion of the circumcircle of 4CIC1 is
the line C1 Co , where Bo is the midpoint of C1 A1 and Co is the midpoint of A1 B1 . Note that
the 3 medians A1 Ao , B1 Bo , C1 Co of 4A1 B1 C1 are concurrent. Furthermore, they meet at
∞. This means that the circumcircles under consideration pass through two points. (one
of them is I.) Thus they are coaxial and hence their centres are collinear.
2.2. 1st solution

We need to prove that
v
u n n−1
uX X√ √ √ √
t ak ≤ k( ak − ak+1 ) + nan .
k=1 k=1
We prove this by induction on n. For n = 1 the void sum has value zero and the result is
clear. Assume that the result holds for a certain n ≥ 1. Consider a1 ≥ · · · ≥ an+1 ≥ an+2 =
0. Write S = nk=1 ak and b = an+1 . It suffices to prove that
P
√ √ √ q
S+b− S ≤ − nb + (n + 1)b.
√
This holds trivially when b = 0. And if b > 0, division by b takes it into the form
√ √ √ √
U + 1 − U ≤ n + 1 − n,
where U = S/b; equivalently:

1 1
√ √ ≤√ √ .
U +1+ U n+1+ n
Since b = an+1 ≤ S/n, we have U ≥ n, whence the last inequality is true and the proof is
complete.
2nd solution
√ √
Set xk = ak − ak+1 , for k = 1, . . . , n. Then
a1 = (x1 + · · · + xn )2 , a2 = (x2 + · · · + xn )2 , . . . , an = x2n .
Expanding the squares we obtain

n
X n
X X
ak = kx2k + 2 kxk x` . (3)
k=1 k=1 1≤k<`≤n
Note that the coefficient of xk x` (where k < `) in the last sum is equal to k. The square of
the right-hand side of the asserted inequality is equal to
n √
X 2 n
X X √
kxk = kx2k + 2 k`xk x` . (4)
k=1 k=1 1≤k<l≤n
And since the value of (3) is obviously not greater than the value of (4), the result follows.
√ √
3rd solution Let ck = k − k − 1, then the inequality can be transformed to
v
u n n
uX X √
t ak ≤ ak ck .
k=1 k=1
By squaring both sides, this is in turn equivalent to

n
X X √
ak (c2k − 1) + 2 ai aj ci cj ≥ 0.
k=2 0≤i<j≤n
√ p p p
Note that ci cj = ij − i(j − 1) − (i − 1)j + (i − 1)(j − 1). Thus for k = 3, . . . , n,
k−1 k−2
X √
X √ q √ √
2 ai ak ci ck = 2 ik − i(k − 1) ai ak − ai+1 ak
i=1 i=1
√ q
= +2 ak−1 ak k(k − 1) − (k − 1)
√ q √
≥ 2 ak−1 ak k(k − 1) − (k − 1) = ak−1 ak 1 − c2k .

√ √
Also 2 a1 a2 c1 c2 = a1 a2 (1 − c22 ). Hence
n
X X √
ak (c2k − 1) + 2 ai aj ci cj
k=2 0≤i<j≤n
n n
X X √
ak c2k − 1 + ak−1 ak 1 − c2k

≥
k=2 k=2
n
2 √
X
= 1 − ck ak−1 ak − ak ≥ 0.
k=2
√
since ak−1 ak − ak ≥ 0 and 1 − c2k ≥ 0. This completes the proof.
From solutions 2 and 3, we can conclude that equality holds if and only if there exists an
index m such that a1 = · · · = am and ak = 0 for k > m.
2.3. 1st solution

We prove by induction on h, the common difference of the progression. If h = 1, there
is nothing to prove. Fix h > 1 and assume that the statement is true for progressions
whose common difference is less than h. Consider an arithmetic progression with first term
a, and common difference h such that both x2 and y 3 are terms in the progression. Let
d = gcd(a, h). Write h = de. If an integer n satisfies n ≡ a (mod h) and n ≥ a, then n is a
term in the progression. Thus it suffices to prove that there is a z satisfying z 6 ≡ a (mod h)
as this implies (z + kh)6 ≡ a (mod h) for any positive integer k and one can always choose
a large k so that (z + kh)6 ≥ a.
Case 1. gcd(d, e) = 1: We have x2 ≡ a ≡ y 3 (mod h), hence also (mod e). The number
e is coprime to a, hence to x and y as well. So there exists an integer t such that ty ≡ x
(mod e). Consequently (ty)6 ≡ x6 (mod e), which can be rewritten as t6 a2 ≡ a3 (mod e).
Dividing by a2 (which is legitimate because gcd(a, e) = 1), we obtain t6 ≡ a (mod e). As
gcd(d, e) = 1, it follows that t + ke ≡ 0 (mod d) for some integer k. Thus
(t + ke)6 ≡ 0 ≡ a (mod d).
Since t6 ≡ a (mod e), we get from the Binomial Formula
(t + ke)6 ≡ a (mod e).
And since d and e are coprime and h = de, the latter two equations imply
(t + ke)6 ≡ a (mod h).
Case 2. gcd(d, e) > 1. Let p be a prime divisor of d and e. Assume that pα is the greatest
power of p dividing a and pβ is the greatest power of p dividing h. Recalling that h = de
with e being coprime to a, we see that β > α ≥ 1. If follows that for each term of the
progression (a + ih : i = 0, 1, . . .), the greatest power of p which divides it is pα . Since x2
and y 3 are in the progression, α must be divisible by 2 and 3. So α = 6γ for some integer
γ; hence α ≥ 6.
The progression (p−6 (a + ih) : i = 1, 2, . . .) with common difference h/p6 < h has integer
terms and contains the numbers (x/p3 )2 and (y/p2 )3 . By the induction hypothesis it con-
tains a term z 6 for some integer z. Thus (pz)6 is a term in the original progression. This
completes the induction.
2nd solution
We use the same notation as in the first solution.
The assertion is proved by induction on h. The case d = 1 is trivially true.
(a) gcd(a, h) = 1. (a−1 exists mod h.) In this case, we have (y/x)6 ≡ a (mod h).
(b) gcd(a, h) = r > 1. Pick a prime p dividing r and let α be the largest positive integer
such that pα divides r. If α ≥ 6, then
x 3 a y 2 a d
( ) ≡ 6, ( ) ≡ 6 (mod ).
p2 p p3 p p6
a d
By induction hypothesis, there exists z such that z 6 ≡ (mod 6 ). Then (zp)6 ≡ a
p6 p
(mod h). So we suppose 0 < α < 6. From x3 ≡ a, y 2 ≡ a (mod h), we have
x3 a y2 a d
α
≡ α, ≡ α (mod α ). (∗)
p p pα p p
h d
(i) gcd(p, ) = 1. (p−1 exists mod α .) Multiply both sides of (*) by pα−6 . We have
pα p
x 3 a y 2 a d
( ) ≡ 6, ( ) ≡ 6 (mod ).
p2 p p3 p pα
a d
By induction hypothesis, there exists z such that z 6 ≡ (mod ). Write a = pα a0 ,
p6 pα
then there is an integer m such that
h
(pz)6 − pα a0 = m .
pα
Since α < 6, pα divides the left hand side of the equation. Thus it also divides m, whence
(pz)6 ≡ pα a0 = a (mod h).
h
(ii) gcd(p, α ) = p. Then pα is the largest power of p dividing a. Furthermore, α is a
p
multiple of 3. To see this write x = pβ x0 , where p does not divide x0 and let x = a + kh
for some positive integer k. Then p3β x0 3 = x3 = a + kh = pα (a0 + pkh0 ) for some integer
a0 , h0 with gcd(a0 , p) = 1. Consequently, α = 3β. Similarly, α is a multiple of 2. Therefore,
α ≥ 6, and this case does not arise.
Footnotes
1. Ptolemy’s Theorem. For any quadrilateral ABCD, we have
AB · CD + BC · DA ≥ AC · BD
and equality occurs if and only if ABCD is cyclic.

2. Proof of the inequality. Let x = a + b, y = a + c, z = b + c, then
a b c 1 x y x z y z 3

+ + ≥ + + + + + −3 ≥ .
b+c c+a a+b 2 y x z x z y 2
3. Menelaus’ Theorem. Three points X, Y and Z on the sides BC, CA, and AB
(suitably extended) of triangle ABC are collinear if and only if
BX CY AZ
· · = 1.
XC Y A ZB
4. Desargues’ Theorem. Given any pair of triangles ABC and A0 B 0 C 0 , the following are
equivalent: (i) The lines AA0 , BB 0 and CC 0 are concurrent. (ii) The points of intersection
of AB with A0 B 0 , AC with A0 C 0 , BC with B 0 C 0 are collinear.
1988/89
1.1. Suppose p is an integer such that p3 − mp2 + mp − (m2 + 1) = 0. It follows that
(p2 + m)(p − m) = 1. Since p and m are integers, we have either
(1) p2 + m = p − m = −1, or
(2) p2 + m = p − m = 1.
In case (1), we have m = p + 1, and so p2 + p + 1 = −1 or p2 + p + 2 = 0 which has no real
solution.
In case (2), we have m = p − 1, and so p2 + p − 1 = 1 or p2 + p − 2 = 0 which has the
solutions p = −2 and 1.
Hence, m = −3 and 0 are the integer values of m for which the given equation has an
integer solution.
..........................................................................................................................................................................................................................................................................
..................................... .........
................................................
........
...................................................
.........
.............................................. ....
..................................... ...................................... ........................................ ..................................... ...
..................................... . ........ .......... .......... .
....................................................................................................................................................................................................................................................................................................................
........ ......... ......... ........
.... ................................................... .............................................. ................................................ ...................................................
... ........................................ ..................................... ...................................... ........................................
1.2. It is only necessary to partition the checker- ...
.
........................................
..........
.....................................
..........
......................................
. ........
........................................
..........
................................................................................................................................................................................................................................ ...................................................................................
board into a closed path one square wide. One way .....................................
.....................................
............................
.
.
....................................
.
. ........
.............................
.
.
......................................
.
. .........
...........................
.
.
...................................
.
. ........
....
...
..................................... .................... ..................... ................... .
..............................................................................................................................................................................................................................................................................................
to do this is shown in the diagram. The squares ..
...
..............................
........................................
...........................
.....................................
.............................
......................................
..............................
........................................
... ........................................ ..................................... ...................................... ........................................
... .............................. ............................. ............................ ..............................
lie with alternating colours along the closed path. ................................................................................................................................................................................................................................................................................................................
.................................... ..................................... ........................................ ..................................... ...
.................................... ..................................... ........................................ ..................................... ...
The removal of two squares of opposite colours .................................... ........................... ..............................
........................................................................................................................................................................................................................................................................................................
............................. ...
... ........................................ ..................................... ..................................... ........................................
... ........................................ ..................................... ..................................... ........................................
from any two positions along the path will cut the ... ............................. ............................ ...........................
......................................................................................................................................................................................................................................................................................................................
.............................
.......................................... ..................................... ........................................ ..................................... ...
path into two open-ended segments (or one seg- ....................................
....................................
.....................................
.....................................
........................................
........................................
.....................................
.....................................
...
...
................................ .................................... ........................................ .................................... ..
ment if the removed squares are adjacent on the ...........................................................................................................................................................................................................................................................................................
... ........................................ ..................................... ..................................... ........................................
... ........................................ ..................................... ..................................... ........................................
.. ............................. ............................ ........................... .............................
path). ................................................................................................................................................................................................................................................
Each segment must consist of an even number of squares, so each segment must be com-
pletely covered by dominoes.
1.3. 1st solution

More generally, let Sn be the set of all n-tuples (x1 , x2 , . . . , xn ) such that
(i) xi = ±1 for i = 1, . . . , n,
(ii) 0 ≤ x1 + x2 + · · · + xr < 4, for r = 1, 2, . . . , n − 1,
(iii) x1 + x2 + · · · + xn = 4.
Also let Sn0 be the set of all n-tuples (x1 , x2 , . . . , xn ) such that
(i) xi = ±1 for i = 1, . . . , n,
(ii) −2 ≤ x1 + x2 + · · · + xr < 2, for r = 1, 2, . . . , n − 1,
(iii) x1 + x2 + · · · + xn = 2.
Let |Sn | = an and |Sn0 | = bn . First note that an = 0 if n is odd. Hence we consider only
even values of n.
Let (x1 , x2 . . . , x2k ) be an element in S2k . Then (x1 , x2 ) = (1, 1) or (1, −1). If (x1 , x2 ) =
0
(1, 1), then (x3 , x4 . . . , x2k ) is an element of S2k−2 . If (x1 , x2 ) = (1, −1), then (x3 , x4 . . . , x2k )
is an element of S2k−2 . Conversely, if each element of S2k−2 0 is augmented at the beginning
by two ones, it gives rise to an element of S2k . Similarly, if each element of S2k−2 is
augmented at the beginning by 1 and -1, it gives rise to an element of S2k . This shows that
a2k = b2k−2 + a2k−2 .
Next consider an element (x1 , x2 . . . , x2k ) in S2k 0 . Then (x , x ) = (1, −1), (−1, 1) or
1 2
(−1, −1). Hence, (x3 , x4 . . . , x2k ) is an element of S2k−2 0 if (x1 , x2 ) = (1, −1) or (−1, 1),
and it is an element of S2k−2 if (x1 , x2 ) = (−1, −1). By augmenting each element of S2k−2 0
0
at the beginning by either 1,-1 or -1,1, it gives rise to an element of S2k . Similarly, by
augmenting each element of S2k−2 at the beginning by −1, −1, we get an element of S2k 0 .
This shows that b2k = a2k−2 + 2b2k−2 .

By eliminating the b2k ’s in the above two difference equations, we have a2k −3a2k−2 +a2k−4 =
0. The initial conditions are a2 = 0 and a4 = 1. By solving this difference equation, it gives
√ !2k−2 √ !2k−2
1 3+ 5 1 3− 5
a2k =√ − √ .
5 2 5 2
2nd solution
Let Sn00 be the set of all n-tuples (x1 , x2 , . . . , xn ) such that
(i) xi = ±1 for i = 1, . . . , n,
(ii) 0 ≤ x1 + x2 + · · · + xr < 4, for r = 1, 2, . . . , n − 1,
(iii) x1 + x2 + · · · + xn = 2.
Let |Sn00 | = cn . Again, we only have to consider even values of n. Note that each sequence
in S2k must end in two “1”s. By dropping these two ones, we obtain a sequence in S2k−2 00 .
00
Conversely, each sequence in S2k−2 can be augmented at the end by two “1”s to get a
sequence in S2k . Hence a2k = c2k−2 .
Let’s examine the end terms of each sequence in S2k 00 . The last three terms of any sequence
00
in S2k are as follow:
−1 1 1, 1 − 1 1, 1 1 − 1, −1 − 1 1, −1 1 − 1. (∗)
For the first three cases, one can replace them by a single “1” to get a member of S2k−2 00 .
For the last two cases, one can drop the last two terms to get a member of S2k−2 . 00
Conversely, for any sequence in S2k−200 , which ends in a “1”, one can replace the “1” by any
one of the first three endings in (*) to get a sequence in S2k00 .
Let the number of sequences in S2k−200 which end in a “1” be x. Let the number of sequences
00
in S2k−2 which end in a “-1” be y. In the latter case, observe that if this last “-1” is replaced
by a “1”, then a sequence in S2k−2 is obtained. Hence, y = a2k−2 .
00
Similarly, for any sequence in S2k−2 which ends in a “-1”, it can be replaced by any one of
the last two endings in (*) to get a sequence in S2k 00 .
Therefore, c2k = 3x + 2y = 3(x + y) − y = 3c2k−2 − y. That is a2k+2 = 3a2k − a2k−2 . This

is the same difference equation in solution 1.
.
A
............
..............
... ... ... .....
.. ... .... ........
2.1. Let H and J be the points on AB such D .
.......... ... .... .......
...... ................ .... .........
E
that HE and JF are parallel to BC. H ...
.
... ... ....... ....
... ....... ....
.....
.....
.....
... ... ..... ...
..... ..
.....
... .
. ........
.....
.....
... ..
.
.... ... ... ... ... .... ... ... ... ... ......... ... F .....
GC FA JF JF J .
.
...
.
..
.
........
. ..
........
.....
.....
.....
..
. ..
. . ..
... ...... .....
Then = = = . .
.
..
.. ..
.
..
.
... .....
.
.
... ..... ..
.....
.....
.....
NC NA BN 2N C .
.
...
.. .
.
.
..
.
.
...
.
...
.
.......
.
..
. ...
.......
.....
.......
..........
... ..
. ... ..
..... .....
.....
..
. ..
. .... ....... .....
... ..
. . ....... .....
Hence JF = 2GC. ... .
. ...
.......................................................................................................................................................................
. .
. ...
B M N G C
EH AE GC GC
Also = = = . This shows that GC = 2EH. Therefore,
MB AM MC 2M B
DF FJ
JF = 4EH. As 4DEH is similar to 4DF J, we have = = 4.
DE EH
Consequently, EF = DF − DE = 4DE − DE = 3DE.
2.2. Let r = x−bxc. Write bxc = lp+q, where l, q ∈ Z and 0 ≤ q < |p|. Hence x = lp+q +r.
Now
x−p −x − 1 q+r −q − r − 1
b c+b c=b c+b c − 1.
p p p p
q+r −q − r − 1
So it suffices to find the value of the expression A ≡ b c+b c.
p p
(i) (p > 0.)
(
q+r −q − r − 1 −2 if q = p − 1 and r > 0
Then b c = 0 and b c= .
p p −1 otherwise
(
−2 if q = p − 1 and r > 0
Hence, A = .
−1 otherwise
(ii) (p = −1.) (
q+r −q if r = 0 −q − r − 1
Then b c= and b c = q + 1.
p −q − 1 if r > 0 p
(
1 if r = 0
Hence, A = .
0 if r > 0
(iii) (p < −1.)
( In this case, we have (
q+r 0 if q + r = 0 −q − r − 1 1 if q + r + 1 ≥ −p
b c= and b c= .
p −1 if q + r > 0 p 0 if q + r + 1 < −p
(
0 if q + r = 0 or q + r ≥ −p − 1
Hence, A = .
−1 if 0 < q + r < −p − 1
Therefore the possible values of the expression A − 1 are -3,-2,-1, and 0.
2.3. More generally, we can prove the following result.

Let f (x) be a nonconstant polynomial with integer coefficients. Then the numbers f (1),
f (2), f (3), · · · , contain infinitely many prime factors.
Let f (x) = an xn + an−1 xn−1 + · · · + a1 x + a0 , where a0 , a1 , . . . , an ∈ Z and an 6= 0 for some
n ≥ 1.
(1) If a0 = 0, then f (k) = k(an k n−1 + an−1 k n−2 + · · · + a1 ). As k can be any prime number,
the numbers f (1), f (2), f (3), · · · , contain infinitely many prime factors.
(2) Consider the case a0 6= 0. Suppose the numbers f (1), f (2), f (3), · · · , contain only finitely
many prime factors, p1 , p2 , . . . , pm . Let y be any integer. We have
f (p1 p2 · · · pm a0 y) = an (p1 p2 · · · pm a0 )n y n + an−1 (p1 p2 · · · pm a0 )n−1 y n−1

+ . . . + a1 (p1 p2 · · · pm a0 )y + a0 ,
= a0 (An y n + An−1 y n−1 + · · · + A1 y + 1),
where Ai = ai (p1 p2 · · · pm )i ai−1

0 , i = 1, 2, . . . , n and An 6= 0.
Let g(y) = An y n + An−1 y n−1 + · · · + A1 y + 1. As p1 p2 · · · pm divides Ai for all i = 1, 2, . . . n,

we have
g(y) = multiple of p1 p2 · · · pm + 1.
Therefore, p1 , p2 , p3 , . . . , pm are not the factors of g(y).
As the equation g(y) = ±1 has at most 2n roots, we can pick an integer y0 such that
a0 y0 > 0 and g(y0 ) 6= ±1. Then the integer g(y0 ) must have a prime factor p distinct from
p1 , p2 , . . . , pm . Consequently, f (p1 p2 · · · pm a0 y0 ) = a0 g(y0 ) has a prime factor different from
p1 , p2 , . . . , pm . This contradicts the assumption that p1 , p2 , . . . , pm are all the prime factors
of the numbers f (1), f (2), f (3) · · · .
1999/2000
....................................................
....... .......... ........ 0
C ......
......
.......
...... E
......... ... .. ...........
... ......... .......... .. . .....
... .. ..... .. .... ........
.... . .....
... .... .....
..... ...
. ...
...
...
... .. ..... ...
...
1.1. Let A0 be the point on AB such that .. ..... . .. ...
. . ..... ...
. .. ... .....
..... .
... ...
...
... ..
. ...... . .. . ...
.... .... ..... ..
A0 F = F A. Then 4AEA0 is isosceles. Ex- ...
... ...
...
......
.
..........
.....
...
..
...
...
...
... ... ..
. ..... ... ...
tend EA0 meeting the circumcircle of 4ABC ....
...
...
...
...
... ....
. .....
.....
.....
.....
.
.
.
...
. ..
.
..
...
..... .
at E 0 . Join BE 0 and BE. Since ∠ABC = ... .. . .. ..... ..
.
... ... . ..... .. ..
.....
... ... . ..... ... ....
... .. ... .....
.... .. ...
∠EBC − ∠ABE = ∠AA0 E − ∠ABE = F ...... .
......................................................................................................................................................................
A B ........
... ........
........
..
.
... 0 . .......
....... ... .....
.
0 0 0 A ... .... ...... ...
.
... ........... .... .....
∠E EB, we have BE = AC. Also, 4AEA is ...
.... ........ .. ...
........ ... ... ....... .......
...............
. .
... .
. ....... ...
.... ...
. ..
.. ..
.....
similar to 4E 0 BA0 implies that A0 B = BE 0 = .. ......
... ...... .....
......
Q E .......
.........
............. .................
.......
AC. Hence, 2AF = AB − A0 B = AB − AC.

............................... . .
(Remark: Let P A be the tangent at A with P inside the sector of ∠QAE. As AB > AC,
we have ∠C > ∠B. Hence, ∠P AB = ∠C > ∠B = ∠QAP . This implies that E is on the
arc AB not containing C. Also, ∠EBF = ∠P AE < ∠EAB so that BF > AF . Hence, A0
is between F and B.)
1.2. The problem can be changed to find all integers m such that 5m + 12m is a perfect
square. Again the only answer is m = 2. We shall give the solution in this more general
case. (The solution of the original problem is easy by considering mod 5 or mod 10.)
One solution is p = 2 and we assert that it is the only solution. If p = 2k + 1 is odd, then
52k+1 + 122k+1 ≡ 22k+1 ≡ 2 · 4k ≡ 2(−1)k ≡ 2 or 3 (mod 5). However the square of an
integer can only be 0, 1 or 4 (mod 5). So 5p + 12p is not a square when p is odd.
Now suppose that 52n + 122n = t2 with n ≥ 2. Then
52n = t2 − 122n = (t − 12n )(t + 12n ).
If 5 divides both factors on the right, it must also divide their difference which means it
divides 12. But this is impossible. Thus t − 12n = 1 and
52n = 2 · 12n + 1 or 22n+1 3n = (5n − 1)(5n + 1).
If n is odd, then 3 | 5n + 1 and 3 - 5n − 1. Thus 5n + 1 = 2 · 3n and 5n − 1 = 4n which cannot

hold for n > 1. If n is even, then 5n − 1 = 2 · 3n and 5n + 1 = 4n , which again cannot hold
for n ≥ 2. Thus there is no solution for p = 2n, n ≥ 2.
1.3. 1st solution by Tan Chee Hau

We shall prove the assertion using induction on n. Let x1 , x2 , . . . , xn be the coordinates of
the n red points on the real line. Similarly, let y1 , y2 , . . . , yn be the coordinates of the n blue
points on the real line. Let An be the sum of distances of points of the same colour, Bn the
sum of distances of points of different colours. If n = 1, then A1 = 0 and B1 = |x1 − y1 |.
Clearly, B1 ≥ A1 . Now suppose Bn−1 ≥ An−1 .
n
X n
X
An − An−1 = (xn − xi ) + (yn − yi ) = (xn − yi ) + (yn − xi ).
i=1 i=1
n
X
Bn − Bn−1 = |xn − yn | + |xn − yi | + |yn − xi |.
i=1
Hence, Bn −Bn−1 ≥ An −An−1 . It follows from this and induction hypothesis that Bn ≥ An .
2nd solution by Lim Yin

Take 2 consecutive points A and B with the coordinate of A less than the coordinate of B.
Suppose that there are k blue points and l red points with their coordinates less than or
equal to the coordinate of A. Then the segment AB is covered (n − k)k + (n − l)l times by
segments whose endpoints have the same colour, and (n − k)l + (n − l)k times by segments
whose endpoints have different colours. Since (n − k)k + (n − l)l ≤ (n − k)l + (n − l)k, the
assertion follows by summing the lengths of all these segments over all pairs of consecutive
points.
3rd solution by Julius Poh

Let S be the total length of the segments whose endpoints are of the same colour and D be
the total length of the segments whose endpoints are of different colour. Move the leftmost
point to the right by a distance x. Then S decreases by (n − 1)x while D decreases by nx.
Thus D decreases more than S. Continue to move this point until it hits the next point.
If these two points are of different colour, then deleting them causes S and D to decrease
by the same amount. If they are of the same colour, then continue to move the pair to
the right and in the process D decreases more than S does. We continuing this process,
when the block that we are moving (all points in the block are of the same colour) hits a
point which is of different colour, remove a pair of points of different colour. If it hits a
point of the same colour, then add the point to the block and continue moving to the right.
Eventually all the points will be removed and both S and D have decreased to 0. Thus at
the beginning D ≥ S.
2.1. 1st solution

Let x = y = 1. We have f (0) = 0. Let a = x + y and b = x − y. Then the given functional
equation is equivalent to bf (a) − af (b) = (a2 − b2 )ab. This holds for all real numbers a and
b. For nonzero a and b, this can be rewritten as
f (a) f (b)
− a2 = − b2 .
a b
Hence, for any nonzero real number x, f (x) 2

x − x = f (1) − 1. Let α = f (1) − 1. We have
f (x) = x + αx, for all x 6= 0. As f (0) = 0, we thus have f (x) = x3 + αx for all x ∈ R.
3
Clearly f (x) = x3 + αx satisfies the given relation.

y
2.2. Set up a coordinate system with CA on the x- ....
........ B ..
.... .........
... .. ..
axis and C = (0, √
0). Let A = (a,√0) with a > 0, F ...
...
...
.... .... .....
.. . ..
... .. ..
... ... .... .....
=(1, 0), D = ( 2 , 2 ), and E = ( 32 , 23 ). Then,
1 3 ...
... .
..
...
... ..
...
..
...
...
...
... ... .. ...
... .
. .
.
... ...
√ ...
... .
...
... ..
.
.
.
.
..
...
...
...
a 3a ... . ..
.
.
.
...
.......................................................................
...
.
B=( , ) and D
...
... .
.
.. .
... ................ ..
. E .......
. ....
2(a − 1) 2(a − 1) ...
...
..
.
...
.. ... ......
. ... ........ ..
...
M ...... ..
.
.
.
...
...
. .. ...
...
... ...
... .......
........ ....
...
... .. ... .
. ...
. ............
√ ... ....
... ....
. ...
... ..
... .... ..... ...........
...
...... .....
...
...
a(1 + a) 3a(a − 1) 60
... ...
... ...... ◦
... .. ...
... ... ...
.
...... ..
. ... ..
M =( , ). ..........................................................................................................................................................................................
... x
2(1 − a + a ) 2(1 − a + a2 )
2
C F A
Hence, DF = 1, DA2 = ( 12 − a)2 + 3

4 = 1 − a + a2 , and
√ √
2 a(1 + a) 1 2 3a(a − 1) 3 2 1
DM = ( 2
− ) +( 2
− ) = .
2(1 − a + a ) 2 2(1 − a + a ) 2 1 − a + a2
2nd solution by Tay Kah Keng

Since DE is parallel to CA, 4DEB is similar to 4F AE so that DB : DE = F E : F A.
As CDEF is a rhombus, we have DE = F E = DF . Hence, DB : DF = F D : F A.
Also, ∠BDF = ∠DF A = 120◦ . This shows that 4BDF is similar to 4DF A. Therefore,
∠DF B = ∠F AD. This implies that 4DM F is similar to 4DF A. Consequently, DF 2 =
DM · DA.
2.3. Let f (n) be the given sum. The summands that appear in f (n) but not in f (n − 1)
are those of the form ap = 1/pn where 1 ≤ p < n, (p, n) = 1; the summands in f (n − 1)
but not in f (n) are those of the form bp = 1/p(n − p) where 1 ≤ p < n − p, (p, n − p) = 1,
equivalently (p, n) = 1. (For example, if n = 10, those summands in f (10) but not in f (9)
1 1 1 1 1 1
are 1×10 , 3×10 , 7×10 , 9×10 , while those which are in f (9) but not in f (10) are 1×9 , 3×7 .)
Hence summing only over values of p such that (p, n) = 1, we have
X X X
f (n) − f (n − 1) = ap − bp = (ap + an−p − bp ).
p<n 2p<n 2p<n
But ap + an−p − bp = 0; hence f (n) = f (n − 1) for all n ≥ 3, and the result follows.
2000/2001
1.1 It suffices to prove that the product of two differences of two squares is also a difference
of two squares. Let a = x2 − y 2 and b = r2 − s2 . Then, ab = (x − y)(r − s)(x + y)(r + s) =
(xr + ys − yr − xs)(xr + ys + yr + xs) = (xr + ys)2 − (yr + xs)2 .
There is another characterization of a difference of two squares. Namely, a positive integer
n is a difference of two squares of positive integers if and only if n 6= 1, 2, 4, and n 6≡ 2
(mod 4). The result also follows from this characterization.
....................................................
......... .......
....... ......
.... ...... ......
..... .....
..
..
....... .....
..... ..
..... ..
. ..
...
. . . ....... .. ....... ....... ..... ....
.
....
...
...
......
. A
...... ....
. .. ..
....... ..........................................
... .. ....... ............. ......
1.2. Let K and L be points of intersection of ...
....
..
.....
...
....... ......................................... . .......
... .
.. . ..... ..
.......
......
.....
.....
... ... .. .
.. .......... ... . ... ......... .
. . . .....
.. ...
the line BC with the lines AM and AN re- ....
... ...
..
..
.
....
.
...
..
........ .... ... .... .... ........
... ... . ... ... ....
.
...... ... .. ... ... .......
..
...
...
...
...
... .
. .... .
. ...
... ... ... ... ... ....... .... ...
spectively. Suppose that the line BC is the ...
...
...
.
.
..
.. ........ ..
. . . .. .
... ... ...
.. .... .... ....
.....
.
.
.
.
...
....... ...
...
...
...
..
. . ...... . ..... ... .
. ..
.. .
... ..
x-axis of a coordinate system with origin B, ...
...
...
...
. .
..
.
.
...
. .
.......
..
.
...
.
.
. .... .. .... ... .. ..
.
. .... .. ....
.
.
. ...
. ..
..
. ..
.
...
and let c, p, q, k and l denote the x-coordinates

...
.....
.....
.
...
.....
.
.... ..
..... ...
...
. .
M
... ..... .. ...... ....
...........
. .
...
..
..
...
...
...
.
... .....
.
...
.
..... .. ..... ... .. ... ...

... .. ........
...... ..
...... .. ........
. .
. .
. .. ............. ....... ... ..
. .
. . .
........ ........ . ..... . ..... .. ... . ....
of C, P, Q, K and L respectively. ...........................................................................................................................................................................................
............................................. ..... . ..
. .........................................
...
B ....... ......
...
Q ...... ..... C
....... P
....... ....... ............
. ... ......
. .
.
....... ....... ....... ..
.
....
..
N
The point K is on the radical axis of the circumcircles of 4P AB and 4QAC, hence its
powers k(k − p) and (k − q)(k − c) with respect to these two circles are equal. It follows
that k = cq/(c + q − p). Similarly, we have l = cp/(c + p − q), interchanging the roles of p
and q. We easily find that l = k if and only if p + q = c and the result follows.
1.3 Each time a player loses a match, he has to wait six games before his turn comes again.
If x is the number of games before his first turn, then the player will win if x + 7r + 7 = 37,
where r ≥ 0 is an integer and 0 ≤ x ≤ 6. Here r counts the number of times he lost. From
this, we obtain x = 2 and r = 4. Thus the second player in the queue wins. That is P4
wins.
2.1. Let G be the point on CE such that DG is parallel to
BE. Then ∠EBD = ∠GDC. Also EG/GC = BD/DC. ......
A
.........
.. .........
Note that 4ADE is similar to 4DCE. Then, .... .... .........
.
. .. .
... .... .... .....
... ... ... ...
... ... ... ....
.... ... ... ...
... ... ... ....
F E/F D = BD/DC ..
.
...
.
. ... ...
.....
...
...
...
...
...
.... ... ... ...
⇐⇒ EG/GC = F E/F D ..
.
..
.. .
.
..
.
.
.
...
...
...
...
...
.... .... ... ...
. ..
⇐⇒ 4ADF is similar to 4DCG ..
.
...
. .
...
.
.
...
...
...
...
...
...
.
. .
.
... .
... E
.....
................ ................ .....
.
...
⇐⇒ ∠DAF = ∠GDC .. . . ...
.
....... .
..
. . . .
. . ..
................... ... ........ .
. .
.
.
.
.
.
.
G
...
..
... ............. .... ....... ... ... ......
.. ............. ....... ... ... ... ... ...
F
⇐⇒ ∠DAF = ∠EBD B ..
. ................... ...... .
. .. ... . ...
. ... . . .
..................................................................................................................................................................... C
...
.
⇐⇒ AF ⊥BE. D
2nd Solution Let A = (0, a), B = (−b, 0), C = (c, 0), D = (0, 0), E = (x, y), F = (tx, ty),
where a, , b, c, t > 0. DE⊥AC implies that (x, y) = (sa, sc) for some s. E lies on AC implies
that s = ac/(a2 + c2 ). Hence, x = a2 c/(a2 + c2 ) and y = ac2 /(a2 + c2 ).
Then,
AF ⊥BE
⇐⇒ (tx, ty − a) · (x + b, y) = 0
⇐⇒ tx2 + txb + ty 2 − ay = 0
ta4 c2 2c ta2 c4 a 2 c2
⇐⇒ (a2 +c2 )2
+ atba
2 +c2 + (a2 +c2 )2 − a2 +c2 = 0
⇐⇒ ta2 c + tb(a2 + c2 ) + tc3 − c(a2 + c2 ) = 0

⇐⇒ −a2 c(1 − t) − c3 (1 − t) + tb(a2 + c2 ) = 0
⇐⇒ (a2 + c2 )tb = c(1 − t)(a2 + c2 )
⇐⇒ b/c = (1 − t)/t
⇐⇒ BD/DC = F E/F D.
2.2 For n ≥ 6, take xn−5 = xn−4 = xn−3 = xn−2 = xn−1 = 1/2 and xn = 1 and zero
for other xi . Then the left hand side of the inequality is 9/4, while the right hand side is
5/2. So the inequality is not valid for n ≥ 6. We shall prove that the inequality holds for
n = 2, 3, 4, 5. The cases n = 2 and 3 can be verified easily. Let’s consider the case n = 5.
(The case n = 4 can be proved in a similar way.) The inequality to be proved is equivalent
to
x25 − (x1 + x2 + x3 + x4 )x5 + (x21 + x22 + x23 + x24 ) ≥ 0.
Regard this as a quadratic equation in x5 . It suffices to prove that its discriminant is less
than or equal to zero. The discriminant is equal to (x1 + x2 + x3 + x4 )2 − 4(x21 + x22 + x23 + x24 )
which can simplified to −[(x1 −x2 )2 +(x1 −x3 )2 +(x1 −x4 )2 +(x2 −x3 )2 +(x2 −x4 )2 +(x3 −x4 )2 ].
It is obviously less than or equal to zero.
2.3 If n is a prime-power pα , where p is a prime and α is a positive integer, then L(n) =

pL(n − 1) since pα−1 < n ensures that pα−1 divides L(n − 1). On the other hand, if n is not
a prime-power, it is greater than every prime-power which divides it, so L(n) = L(n − 1).
Thus L(n) = L(n − 1) if and only if n is not a prime-power.
(i) We shall prove that there are arbitrarily long sequences of consecutive positive integers
with the same value of L(n). For any n, let P (n) be the product of all distinct primes
p ≤ n. If 2 ≤ r ≤ n, then r divides L(n), so L(n)P (n) + r is a multiple of r. However, it
is not a prime-power, for if pα is a maximal prime-power factor of r, then pα+1 is a factor
of L(n)P (n), so that L(n)P (n) + r is greater than r but has pα as a maximal prime-power
factor. Therefore, {an = L(n)P (n) + r : 1 ≤ r ≤ n} is a sequence of n consecutive positive
integers with the same value of L(an ). Now take n = 2001.
(ii) We know that m + 1, m + 2, m + 3 are all prime powers. One of them is a power of 2
and another is a power of 3 and they must be adjacent. Since the equation
2x + 1 = 3 y (1)
has two solutions in integers, (x, y) = (3, 2), (1, 1) and the equation
2x − 1 = 3 y (2)
has the solution (x, y) = (2, 1). By examining L(1), . . . , L(11), we see that the only solutions
are:
m = 1, 2, 6.
[(1) has solutions (x, y) = (3, 2), (1, 1). Assume that x > 3. As 3y − 1 = 2x = 16(2x−4 ) is
divisible by 16, it implies that y ≡ 0 (mod 4). Write y = 4k. Thus 34k − 1 = 81k − 1 =
80(81k−1 + 81k−2 + · · · + 1) which is not a power of 2.
(2) has a solution (x, y) = (2, 1). Now assume that y > 1. Then 2x − 1 is divisible by 9.
This implies that x must be even. Write x = 2x1 . Then 2x = 4x1 . For 4x1 ≡ 1 (mod 9), we
have x1 = 3k. Thus 2x − 1 = 64k − 1 = 63(64k−1 + 64k−2 + · · · + 1). Thus it cannot be a
power of 3.]
2001/2002
1.1 Suppose XY = XB. Then XY 2 = B
........
............
......................................................................................................................................
........
....... .. .
...... ....
X
....... ......
XB 2 = XC · XA so that XY : XC = ...............
.
......
.....
.....
C
..... ..... ...
...
......
...... ...
...
..... .......... ..
... ..
XA : XY . This shows that 4XCY is sim- ...
.. .
......
. .........
...... ..........
.
.
...
... ...... ...... ...
... ...... ......
D ..
ilar to 4XY A. Hence ∠EDY = ∠XAY = ....
.. ......
. . . .. . ...... ...
.
..........
..
.
.
...
.. ...... ........ ...
.. ...... .. ... ..
∠XY C. Therefore, XY is parallel to DE. ..... ...... ..
....
....... .. ... .....
.
. ..
.
.
... . .
...... ... ... .... ...
... ...... ... ... .... ....
The converse is similar. ... ......
... ...... .. .. ... ..
... .. .
........ .
....... ... ...
... .......... ... .. ..
A ....................................................................................................................................................................
...
... ...
... Y
.....
..... ....... E
..... .....
...... .....
....... ......
........
............ . . . ............
..................................
1.2 Use Induction. We can prove by induction that On = 22n−1 − 2n−1 and En = 22n−1 +
2n−1 . We merely have to note that
On+1 = En + 3On , En+1 = 3En + On .
2nd Solution Using generating function:

Let yi = x2i−1 x2i . Then Sn = y1 + · · · + yn . There are three ways for each yi to be 0 and 1
way for it to be 1. Thus if f (x) = (3+x)n = ai xi , then ai is the number of sequences with
P
Sn = i. Thus On = a1 +a3 +· · · = [f (1)−f (−1)]/2 and En = a0 +a2 +· · · = [f (1)+f (−1)]/2.

The result thus follows.
3rd Solution Direct computation:

Let yi = x2i−1 x2i . Then Sn = y1 + · · · + yn . There are three ways for each yi to be 0 and 1
way for it to be 1. Thus
! ! !
n−1 n n n
On = 3 + 3n−3 + 3n−5 + ···
n−1 n−3 n−5
! !
n n−2 n n
En = 3 + 3 + 3n−4 + ···
n−2 n−4
From here we have En + On = (1 + 3)n and En − On = (3 − 1)n . The result then follows.
1.3 We show more generally that ak 2 + bk + c ≡ 0 (mod 2n ) has a solution for all n
whenever b is odd and a or c is even. For n = 1, take k = 0 if c is even and k = 1
if c is odd. Now suppose the claim is true for all n. If c is even, then, by assumption,
the congruence 2at2 + bt + c/2 ≡ 0 (mod 2n ) has some solution t. Letting k = 2t we get
ak 2 + bk + c = 2(2at2 + bt + c/2) ≡ 0 (mod 2n+1 ). If c is odd, then a is even, so a + b + c is
even; hence, by assumption, the congruence 2at2 + (2a + b)t + (a + b + c)/2 ≡ 0 (mod 2n )
has some solution t. Letting k = 2t + 1 yields
ak 2 + bk + c = 2[2at2 + (2a + b)t + (a + b + c)/2] ≡ 0 (mod 2n+1 ).
Thus, whether c is even or odd, the claim is true for n + 1, and so by induction for all n.
2nd Solution by Tan Kiat Chuan, Tay Wei En Joel, Leung Ngai-Hang Zachary,
Kenneth Tay
It suffices to show that 2k 2 + 2001k + 3, k = 0, 1, · · · , 2n − 1 forms a complete residue
class modulo 2n . Suppose there are distinct integers k1 , k2 , 0 ≤ k1 , k2 ≤ 2n − 1 such that
2k12 + 2001k1 + 3 ≡ 2k22 + 2001k2 + 3 (mod 2n ). That is (k1 − k2 )(2k1 + 2k2 − 2001) ≡
(mod 2n ). Since 2k1 + 2k2 + 2001 is odd, we must have 2n divides k1 − k2 . Thus k1 ≡ k2
(mod 2n ). Since 0 ≤ k1 , k2 ≤ 2n − 1, we have k1 = k2 . Therefore, 2k 2 + 2001k + 3
k = 0, 1, · · · , 2n − 1 forms a complete residue class modulo 2n .
2.1 The inequality can be written in the form
x21 + x22 + x23 1 x3 + x32 + x33 1

( )2 ≤ ( 1 )3
3 3
This is known as the Power Mean Inequality. There are several proofs of this inequality.
First recall Hölder’s inequality: Let p, q be real numbers such that p, q > 1 and p1 + 1q = 1.
Then for any 2n positive real numbers a1 , b1 , . . . , an , bn ,
n n 1
n 1
api ) p ( bqi ) q .
X X X
ai bi ≤ (
i=1 i=1 i=1
In our case, take p = 3, q = 32 , n = 3, a1 = a2 = a3 = 1 and b1 = x21 , b2 = x22 , b1 = x23 . We

then have
1 2
x21 + x22 + x23 ≤ 3 3 (x31 + x32 + x33 ) 3 .
That is
(x21 + x22 + x23 )3
≤ 3.
(x31 + x32 + x33 )2
3
Alternatively, Consider the function f (x) = x 2 for x > 0. f 00 (x) = √3x > 0 for x > 0. Hence,
f is concave upward. By Jensen’s Inequality, for any three positive numbers z1 , z2 , z3 ,
z1 + z2 + z3 f (z1 ) + f (z2 ) + f (z3 )

f ≤ .
3 3
Now take z1 = x21 , z2 = x22 and z3 = x23 . We have

! 32
x21 + x22 + x23 x31 + x32 + x33
≤ .
3 3
That is
(x21 + x22 + x23 )3
≤ 3.
(x31 + x32 + x33 )2
2nd Solution by Lim Yin

The given inequality is equivalent to
[x61 + 2x31 x32 − 3x41 x22 ] + [x61 + 2x31 x33 − 3x41 x23 ] + [x62 + 2x32 x31 − 3x42 x21 ]
+[x62 + 2x32 x33 − 3x42 x23 ] + [x63 + 2x33 x31 − 3x43 x21 ] + [x63 + 2x33 x32 − 3x43 x22 ]
+[2x31 x32 + 2x31 x33 + 2x32 x33 − 6x21 x22 x23 ] ≥ 0.
Each term in the square brackets is non-negative by the AM-GM inequality. So the result
follows.
3rd Solution by Leung Ngai-Hang Zachary

The given inequality is equivalent to:
[2x31 x32 + 2x31 x33 + 2x32 x33 − 6x21 x22 x23 ] + [x61 + x62 + 4x31 x32 − 3x41 x22 − 3x21 x42 ]
+[x61 + x63 + 4x31 x33 − 3x41 x23 − 3x21 x43 ] + [x62 + x63 + 4x32 x33 − 3x42 x23 − 3x22 x43 ] ≥ 0.
The first term is nonnegative by rearrangement inequality. The next three can be shown to
be nonnegative by using rearrangement inequality as follows:
x61 + x62 + 4x31 x32 = x31 (x31 + x32 ) + x32 (x32 + x31 ) + 2x31 x32
≥ x31 (x21 x2 + x1 x22 ) + x32 (x1 x22 + x21 x2 ) + 2x31 x32
= x51 x2 + x41 x22 + x1 x52 + x21 x42 + 2x31 x32
= (x51 x2 + x31 x32 ) + (x1 x52 + x31 x32 ) + x41 x22 + x21 x42
≥ 3x41 x22 + 3x21 x42 .
Alternatively,
x61 + x62 + 4x31 x32 = (x61 + x31 x32 + x31 x32 ) + (x62 + x31 x32 + x31 x32 ) ≥ 3x41 x22 + 3x21 x42 .
2.2 Suppose r is not an integer, choose an integer a such that ar 6= barc > 1. (Note that
r > 0. If r is irrational, choose any large positive integer a. If r = p/q is rational, choose a
large positive integer a such that (a, q) = 1.) Let k be the unique integer such that
1 1
≤ ar − barc < .
k+1 k
Then
k+1
1 ≤ (k + 1)(ar − barc) < ≤ 2.
k
Since
b(k + 1)arc = (k + 1)barc + b(k + 1)(ar − barc)c = (k + 1)barc + 1
we see that barc does not divide b(k + 1)arc. Thus m = a, n = (k + 1)a form a counter
example.
2nd Solution by Charmaine Sia

Suppose that r is not an integer. We may assume r > 1. (If 0 < r < 1, we may choose a
positive integer p such that pr > 1 and pr is not an integer. Then consider r0 = pr.) Now
choose a positive integer m such that k + 21 ≤ mr < k + 1 for some positive integer k > 1.
Then bmrc = k and b2mrc = 2k + 1. Thus bmrc does not divide b2mrc.
3rd Solution by Tan Weiyu Colin

Given r, choose m ∈ N so that mr > 10 and mr 6∈ N. Let mr = a + 10b j + 10xj+1 where j ∈ N,
a = bmrc > 10, b is a positive integer less than 10 and x is a nonnegative real number, also
less than 10. Let k = 10j and n = mk. Then bmrc = a, bnrc = ka + b and bmrc = a - bnrc.
2.3 Fix any x ≥ 0. Let f [0] (x) = x and f [1] (x) = f (x). For n ≥ 1, let f [n] (x) = f (f [n−1] (x)).
Then the above functional equation gives
f [n+2] (x) + f [n+1] (x) = 12f [n] (x).
Solving this difference equation, we have
f [n] (x) = C1 3n + C2 (−4)n .
Using the initial conditions f [0] (x) = x and f [1] (x) = f (x), we have C1 = (f (x) + 4x)/7 and
C2 = (3x − f (x))/7. Therefore,
1 1
f [n] (x) = (f (x) + 4x)3n + (3x − f (x))(−4)n .
7 7
Since f (x) ≥ 0, f [n] (x) ≥ 0 for all n ≥ 0. By taking n even, we have 17 (f (x) + 4x)3n +
1 n 1 n
7 (3x − f (x))4 ≥ 0. From this, 3x − f (x) ≥ 0. By taking n odd, we have 7 (f (x) + 4x)3 −
1 n
7 (3x − f (x))4 ≥ 0. From this, 3x − f (x) ≤ 0. Consequently, f (x) = 3x. One can easily
verify that f (x) = 3x satisfies the given functional equation.
Alternatively, suppose for some a, f (a) = 3a + c where c ∈ R. Then f [n] (a) = 3n a + kn c.
One can obtain a recurrence relation in kn and use it to prove that c = 0. Finally check
that f (x) = 3x satisfies the given condition.
2002/2003
1.1 Take n = 10222 − 3. Then
(10222 − 3)2 = 10444 − 6 · 10222 + 9 = |9 ·{z

· · 9} 40 · · · 09.
221 nines
The sum of the digits of n2 = 222 × 9 + 4 = 2002.
1.2 This is in fact the Butterfly Theorem. There are many proofs of this result. See the
discussion on page 45 in Geometry Revisited by Coxeter and Greitzer. Here we give two
proofs. ........................................
............. ........ .........
C .......
........................
.......
......
First, apply sine rule to .
..
...
.... ..
......... .......... ........................
. α .....
.
.............
. ..
.. .............
......
.....
.....
.....
... ...
. ..... . ....
. . ...
4CM P , .
..
... ..
.
.. .. ......
......
......
................
. .. .. ...................
.
E .....................
.. ...
. ..... γ β ............
. .. ........ ........................
.. ...... .
... .. .....
................. ............ .. ..
. .............
.............
.. ... . .. ..
MP CP Q A .
. M ... .. ........ .
θ .... . . ... .........
........................................................................................................................................................................................................................................................................................................................................................
.......... . . .... .. . .
. . P
= − (1) ..........
..........
..........
.
.. ..
.. ..
.
.......... ... .. .................... ..... .... . .
θ
....... .. .... .... . ...... ...................
γ
.....
..
B.
.
.
.
. ...
...
...
sin α sin(γ + θ) ........................
............... α
.....
.....
.....
.
..
..
.
...
..... ..
F ... .............
... ..........
..........
.....
.
.....
..
. ....
... .......... ..... .. ...
... .......... .. ...
Apply sine rule to 4EM P , ...
...
...
..........
..........
..........
.....
.....
.....
..... .
.. ....
. ..
... .......... ..... ... ....
..........
...
... β
.......... .
.......... ......... .. .....
........... .........
.
....
MP EP .....
.....
...........
...
.
= − (2) .....
...... D
......
.....
..
...... ......
sin β sin θ .......
.........
.............
............................... .. .. ... .
. ..........
.......
MP2 CP · EP
(1) × (2) gives = . – (3)
sin α sin β sin(γ + θ) sin θ
QM QD
For 4QDM , = . – (4)
sin β sin(γ + θ)
QM QF
For 4QF M , = . – (5)
sin α sin θ
QM 2 QD · QF
(4) × (5) gives = . – (6)
sin α sin β sin(γ + θ) sin θ
CP · EP QD · QF
By (3) and (6), 2
=
MP QM 2
(M P + AM )(M P − AM ) (QM + AM )(QM − AM )

This implies 2
= .
MP QM 2
M P 2 − AM 2 QM 2 − AM 2
Or equivalently, = .
MP2 QM 2
That is M P = QM .
Second solution Set up a rectangular coordinate system with M as the origin and QP as
the x-axis. Let the equation of the circle be x2 + (y + c)2 = r2 . Let the coordinates of C, D,
F and E be (p1 , ap1 ) and (p2 , ap2 ) (q1 , bq1 ), (q2 , bq2 ), respectively. Direct calculation shows
that the x-intercept of CE at the point P is p1 q2 (b−a)/(bq2 −ap1 ) and the x-intercept of DF
at Q is p2 q1 (a − b)/(ap2 − bq1 ). It suffices to verify that p1 q2 /(bq2 − ap1 ) = p2 q1 /(ap2 − bq1 ).
This equation can be rearranged to
ap1 p2 (q1 + q2 ) = bq1 q2 (p1 + p2 ). (∗)
The line CD has equation y = ax. Thus the x-coordinates of C and D are the roots of the
equation x2 + (ax + c)2 = r2 . That is p1 and p2 are the roots of the quadratic equation
(1 + a2 )x2 + 2acx + (c2 − r2 ) = 0. Similarly, q1 and q2 are the roots of the quadratic equation
(1 + b2 )x2 + 2bcx + (c2 − r2 ) = 0. Using the relations between roots and coefficients, we have
p1 + p2 = −2ac/(1 + a2 ) and p1 p2 = (c2 − r2 )/(1 + a2 ). Similarly, q1 + q2 = −2bc/(1 + b2 )
and q1 q2 = (c2 − r2 )/(1 + b2 ). With these, (∗) can be easily verified.
1.3 If maxj mini aij = mini maxj aij = aαβ , then clearly aαβ is at once the largest number
in the αth row and the smallest numbers in the βth column, and hence
aαj < aαβ < aiβ for all i 6= α and for all j 6= β. (∗)
Conversely, if (∗) holds for some aαβ , then mini aij ≤ aαj < aαβ for all j 6= β and maxj aij ≥
aiβ > aαβ for all i 6= α would imply that maxj mini aij = aαβ = mini maxj aij . To obtain a
required configuration, it is therefore necessary and sufficient to choose any 2n − 1 of the
given n2 numbers, say x1 < x2 < · · · < x2n−1 . Put xn anywhere in the array. Then put
x1 , x2 , . . . , xn−1 in the same row as xn and put xn+1 , xn+2 , . . . , x2n−1 in the same column
as xn . The remaining n2 − 2n + 1 numbers can be used to fill up the remaining n2 − 2n + 1
positions. Therefore, the total numbers of such configurations is
!
n2 (n2 )!(n!)2
· n2 · [(n − 1)!]2 · (n2 − 2n + 1)! = .
2n − 1 (2n − 1)!
2.1 The answer is 290. First observe that the arithmetic sequences {p + mk, k ∈ N} and
{q + nk, k ∈ N} are disjoint if and only if p − q 6= ln − km for all integers k, l, which holds if
and only if gcd (m, n) does not divide p − q. Therefore, the required r cannot be relatively
prime to 10 = (2)(5), 26 = (2)(13) and 29. We start by choosing r to be the smallest
lcm of d1 , d2 , d3 where d1 , d2 , d3 are factors (greater than 1) of 10, 26 and 29 respectively.
The smallest such r is 58 = (2)(29). In this case, gcd (10, 58) = 2, gcd (26, 58) = 2 and
gcd (29, 58) = 29. We also require b − 3 6≡ 0 (mod 2), b − 6 6≡ 0 (mod 2) and b − 5 6≡ 0
(mod 29). But there is no solution for b from the first two equations. Therefore we cannot
take r = 58. The next smallest lcm would be 290 = (5)(2)(29). In this case, a simple
checking using the above criterion shows that {1 + 290k, k ∈ N} is disjoint from A
...................................................
2.2 Let the radius of Γ be r1 and the radius ........... ........................ S
........ ...........
....... .......
.......
....... ..... .. ....
... ..............
.
..... ..... P
of the inscribed circle be r2 and its center be .
....
.
.
.
....
.
..
..
.
..
.
..
...
.....
.. .........
..........
...........
... .. .. ... ..... ....
F . First we have P B 2 = AB · BM because Γ .
.
..
.
...
.
...
.
...
..
F ....
·
.
...
.
..
.
..
.
..
·Q
.. ..... .....
...
...
...
. .
... ... ... ... .... ...
4AP B is similar to 4P M B. ...
....
...
...
... .
..
.
..
...
.. . . ...
...
..
.. ...
.....
......
...
...
...
... ... .. .. ..
Using Pythagoras’ Theorem, RO2 = F O2 − ...
...
..... ...
...... ..
........ .
........
.. ..
......
.
.
..
... ...
...
....
...............................................................................................................................................................................................
...
...
.
A B
F R2 = (r1 − r2 )2 − r22 = r12 − 2r1 r2 = AO2 − O R M
2AO · RM .
Thus, AO2 − RO2 = 2AO · RM = AB · RM . Therefore, AB · RB − RB 2 = AR · RB =
AO2 − RO2 = AB · RM . From this, we have AB · (RB − RM ) = RB 2 . Therefore,
P B 2 = AB · BM = AB · (RB − RM ) = RB 2 and so P B = RB.
Note that since ∠P AM = ∠BP M , we have P B = RB if and only if ∠BP R = ∠BRP
if and only if ∠AP R = ∠M P R if and only if P R bisects ∠AP M . With this observation,
if we inscribe another circle in the curvilinear triangle P BM touching M B at a point
R0 , then ∠RP R0 = 45◦ . Note also that S, Q, B are in fact collinear. Using this, we can
obtain another solution as follow. Let the extension of P M meet the circle at M 0 . Then
BR2 = BQ · BS = BQ2 + BQ · QS = BQ2 + P Q · QM 0 = BQ2 + (P M − M Q)(M 0 M +
M Q) = BQ2 + (P M − M Q)(P M + M Q) = BQ2 + P M 2 − M Q2 = M B 2 + P M 2 = P B 2 .
There is an even shorter proof due to Colin Tan. The fact that S, Q, B are collinear gives
∠ASQ = 90◦ . Thus 4ABS is similar to 4QBM so that AB/BS = BQ/BM . Therefore,
P B 2 = AB · BM = BS · BQ = BR2 .
2.3 Replacing m by f (f (m)), we have
f (f (f (m)) + f (f (n))) = −f (f (f (f (m)) + 1)) − n. (1)
Interchanging m and n in (1), we get
f (f (f (m)) + f (f (n))) = −f (f (f (f (n)) + 1)) − m. (2)
Put m = 1 in the original functional equation and denote for simplicity f (f (2)) by k. We
obtain f (f (f (n)) + 1) = −k − n. Using this and equating (1) and (2), we get f (−m − k) −
f (−n − k) = m − n. Letting m = −p + k and n = −p + k + 1, we have f (p) − f (p − 1) = −1.
Inductively, we obtain f (p) = f (0)−p. Thus, f (f (p)) = f (f (0)−p) = f (0)−(f (0)−p) = p.
Substituting this into the original functional equation, we have f (m + n) = −m − 1 − n. In
other words, f (n) = −n − 1 for all n. Indeed, f (n) = −n − 1 satisfies the given functional
equation.
(2nd solution by Teo Wei Hao) Setting m = 0, we obtain f 3 (n) = −f 2(1) − n. This
functional relation immediately shows that f is bijective, because f f 2 (n) = −f 2 (1) − n
and f (p) = f (q) =⇒ f 3 (p) = f 3 (q) =⇒ p = q. Now we may let f 2 (n0 ) = 1 for some n0 .
The original functional equation becomes f (m + 1) = −f 2 (m + 1) − n0 . Using the fact
that f is bijective, we may let f (m + 1) = x, so that f (x) = −x − n0 . Applying this on
f 2 (n0 ) = 1 gives n0 = 1. Therefore, f (x) = −x − 1.
(3rd solution by Colin Tan) Let f 2 (1) = c. Put n = 1, m = 0 and replace n by f (n), we
get, respectively,
f (m + c) = −f 2 (m + 1) − 1 (1)
3
f (n) = −c − n (2)
3 2
f (m + f (n)) = −f (m + 1) − f (n) (3)
From (1) and (3), we get
f (m + c) + 1 = f (m − c − n) + f (n) (4)
Put n = −2c + 1 and replace m by m + 1 − c in (4), we get
f (m + 1) + 1 = f (m) + f (−2c + 1).

Thus f is linear. Write f (x) = ax + b and from the original equation, we get a = b = −1.
Thus f (x) = −x − 1. From the above computation, we see that this function satisfies the
original functional equation.
2003/2004
1.1 Let M = x(x + 1)(x + 2)(x + 3)(x + 4)(x + 5)(x + 6)(x + 7), where x is a positive integer.
Then M = (x2 + 7x)(x2 + 7x + 6)(x2 + 7x + 10)(x2 + 7x + 12). Let a = x2 + 7x + 6 =
(x + 1)(x + 6) ≥ 2 × 7 = 14. Thus M = (a − 6)a(a + 4)(a + 6) = a4 + 4a(a + 3)(a − 12) > a4 .
Also (a + 1)4 − M = 42a2 + 148a + 1 > 0. Therefore, a4 < M < (a + 1)4 . Consequently,
1
bN c = bM 4 c = a = (x + 1)(x + 6) which is an even integer.
C ...
..................
........... ... ...........
.............................. .....γ
1.2 Let the intersection of AD, BE, CF .
z
.. ..... . . .... ...................γ
. . .. .
.
.........................
.. .
.. .....................
............... .........
.................. ..
with BC, CA, AB be D0 , E 0 , F 0 respec- .
...........
.
.. .....
..............
.....
.. .
.....
. . .
.
.
.
.
..
... ..............
...
...
..........
..........
..........
E ............
. ..................................................... .... ..........
tively. It is easy to establish that ∠F AF 0 = ..
. . ... . ........ ... ........
........ ............... ................ .
.
. .. .
..... ... .......
........
.
........... .....
.. . ..
............. .
.
.
. .
. . .
........ D
........
....
..... . . .. . . .... ............... .........
.
∠EAE 0 = α, F BF 0 = ∠DBD0 = β,
.
... . .
.... .................. .. .
.
. .. . .
. . . . .
. ......
.. ...... 0 ...... .......
. . .........
..
.
.....
.
.
. .. ........ .
. E ...... .......
.. .. .
.
.
.
.
.... .
..
..
.
... .
...
........... ....
. . ......
......
....... .. ... ..... ...... .. ............ 0 ......
∠DCD0 = ∠ECE 0 = γ. Also AE = AF =
.
. ..... ......
..
.
. .. ....... ...
.
. ..
. ..
..
..... ...... .. ..............
..... .....................
.
.
. . . . . .. ..
... . .. D .....
.
.
.
... ...
......
.... ... ....... ..
. .... ..
. . .
. . ......
.... ............ ... . .. . . ................. ......... .. ...
. . .....
.... . . .. ...... . .... ..
x, BF = BD = y, CD = CE = z. The .
...
.. ...... ....
. .
.
..
.. ...
...
... ........ ............. ... .. . I .
... .
. .
.
. .. ......
........ ...
.
. ... .......
... .....
.... ........ .... . . . .. . ..
.
. . . ........ .. .
. .
... .. ....
. . .
ratio AF 0 /F 0 B equals to the ratio of the al- .
... .....
...α.....
... ........ ... .
...
.
... ............... .....
.........
........
...
.
.... ....
... ..
. ...
. . . .
......
. ..
... .......... .........β....
.
.... ..
.
. . . ..... . ... .......... ... ...
............ ............... ..... ..... ... ..
. .. .. .
........ ........
.
titudes from A and B on CF of the trian- . . ...... ........
... ...
.................
. ....
..
....
.........
..... ..........
...
.
.
. .
........
... . .......
...
. ... . ..............
... ..
A ............................................................................................................................................................................................................................................................................ B
................ α . ...... . . .
gles AF C and BF C and hence as the ratio ................
................ ...........
. ...
...................... F 0....
. .
.... .
.. . .
.. .
. .
. . . . β
................... . . . . .
.. . . .
x .................. ..............
..................... .... ......................................... y
of their areas. .....................................
F
AF 0 Area4AF C xAC sin(∠A + α)

Therefore, 0
= = .
FB Area4BF C yBC sin(∠B + β)
BD0 yAB sin(∠B + β) CE 0 zBC sin(∠C + γ)

Similarly, 0
= and 0
= .
DC zAC sin(∠C + γ) EA xAB sin(∠A + α)
AF 0 BD0 CE 0
It follows that = 1, so by Ceva’s Theorem, AD, BE and CF are concurrent.
F 0 B D0 C E 0 A
1.3 The integer pair (x, y) is a solution of the given equation if and only if x + y = 0 or
(x, y) = (0 ± 1), (±1, 0), or ±(2, 2). Clearly, if x + y = 0, then (x, y) is a solution. Assume
now that (x, y) is a solution with x + y 6= 0.
We first show that xy ≥ 0. Dividing both sides of x5 + y 5 = (x + y)3 by x + y yields
x4 − x3 y + x2 y 2 − xy 3 + y 4 = (x + y)2 .
This is equivalent to
(x2 + y 2 )2 + x2 y 2 = (x + y)2 (xy + 1),
and it follows that xy ≥ 0.
Next we show that |x + y| ≤ 4. The convexity of the function f (t) = t5 on [0, ∞) implies
that for nonnegative x and y,
5
x5 + y 5 x+y 1

≥ , or equivalently, x5 + y 5 ≥ (x + y)5 .
2 2 16
If x + y > 4, then x5 + y 5 > (x + y)3 . Similarly, if x and y are both nonpositive with
x + y < −4, then x5 + y 5 < (x + y)3 .
Finally, examining the cases where xy ≥ 0 and |x + y| = 1, 2, 3 or 4, we find the solutions
(x, y) = (0 ± 1), (±1, 0), or ±(2, 2).
...............................
............... .........
......... .......
2.1 We have to show that ei- ....... ......
...... ......
.. .. ..... .....
. ....
. .....
.....
ther AB is parallel to CD or ...
..... .
...
...
A..
.
.
...............................................................................................................................................................................................................................................................................................
...
D
AD is parallel to BC. Us- H .......
.......
....... .
... ....
.... ...
.. ...
.. ....
.
. .
....... .... .. .. ...
ing the powers of H nd K re- .......
.......
.......
.......
........ .....
... ...
...
...
.
... ....
...
.
...
....... ... ... ... ... ...
spective to the circle, we have .......
.......
.......
....... ..
... .....
.. .
...
...
.... ..
..
..
. .
.
..
..
.
.
. . . .
. .
.
HB 2 = HA · HD = 2HA2 and
....... ... ...
....... .... ... ... ... ...
....... ..
. ... ... ... ...
.......... ... ... ... ..
..
.. ............ ..
KA2 = KB · KC = 2KB 2 . ...
...
. ....... ...
...
....... ...
....... ..... ....
...
... .
...
... ....
...
..
.
.
... ............ ... .. .....

........... .. .. ..........
Thus HB/HA = KA/KB. ... . .
. .
... . .
..........
. .
.
.............................................................................................................................................................................
..
..
K ..................................................
B C
Note that ∠HBA = ∠KAB. Applying sine rule to triangles ABK and BAH, we obtain
KA/KB = sin(∠ABK)/ sin(∠KAB) and HB/HA = sin(∠HAB)/ sin(∠HBA). Thus
sin(∠ABK) = sin(∠HAB). Therefore, either ∠ABK = ∠HAB or ∠ABK + ∠HAB =
180◦ . Consequently, AB is parallel to CD or AD is parallel to BC.
2.2 The smallest such value of k is 1/4. First note that for x, y > 0,
1 4xy 1 (x + y)2 1 1 1 1

= · ≤ · = + ,
x+y x + y 4xy (x + y) 4xy 4 x y
ab bc ca
with equality if and only if x = y. We then have + +
a + b + 2c b + c + 2a c + a + 2b
ab 1 1 bc 1 1 ca 1 1 1

≤ + + + + + = (a + b + c),
4 a+c b+c 4 c+a b+a 4 c+b a+b 4
with equality if and only if a + b = b + c = c + a, that is a = b = c.
2.3 Consider an n × n square lattice with M (n) black

points so that every square path has at least one black ◦...........................◦...........................◦...........................◦•...........................◦...........................◦•...........................◦.....
point on it. ....
...
....
...
....
...
....
...
....
...
....
...
....
...
.. ... ... ... ... ... ...
◦ ....
..................◦•
....
....
.
...
...
...
...
...
...
... ◦ ....
.
...
...
...
...
...
...
... ◦
....
.
...
...
...
...
...
...
...
.... ◦ .
...
...
...
...
...
...
...
.... ◦ .
...
...
...
...
...
...
...
.... ◦•
Let P be a black point in the lattice, and suppose S ...
...
...
...
...
...
...
...
...
...
...
...
...
...
... ... ... ... ... ... ..
is a 2 × 2 square path that passes through P . Assign ◦ . ...................... . ...................... . ...................... . ...................... . ...................... . ...................... ..
.. ◦ .. ◦• .. ◦ .. ..◦• .. ◦ .. ◦
... ... ... ... ... ... ...
P a “credit” of k1 if S passes through exactly k black ...
...
..
...
...
..
...
...
..
...
...
..
...
...
..
...
...
..
...
...
..
points. Let T (P ) be the sum of all credits assigned •◦ ....
◦ ....
◦
...................... ...................... ...................... ...................... ...................... ......................
....
◦ ....
◦ ....
•◦
.... ....
◦
... ... ... ... ... ... ...
... ... ... ... ... ... ...
to P as S varies over all 2 × 2 square paths that pass ◦
...
◦•
...
◦
...
...................... ...................... ...................... ...................... ...................... ...................... . ◦•
...
◦
...
◦
... ..
◦
... ... ... ... ... ... ...
... ... ... ... ... ... ...
through P . ...
....
...
....
...
....
...
....
...
....
...
....
...
..
. ...................... . ...................... . ...................... . ...................... . ...................... . ...................... ...
◦ ..
...
◦ ..
...
◦ ..
...
◦ ..
...
◦• ..
...
◦ ..
...
◦• ..
...
Note that the sum of T (P ) as P varies over all black ...
...
...
...
...
...
...
...
...
...
...
...
...
...
points in the square lattice is (n − 1)2 since each of •◦
...
◦
...
•◦
...................... ...................... ...................... ...................... ...................... ...................... .
...
◦
...
◦
...
◦
...
◦
..
the (n − 1)2 two by two square path contributes 1 to

the total.
It is clear that T (P ) ≤ 1 if P is at a corner, and T (P ) ≤ 2 if P is on an outer edge. Suppose
P is a point in the interior of the lattice. It lies on exactly four 2 × 2 square paths, and there
must be at least one black point on the 3×3 square path surrounding P . Thus, for such a P ,
T (P ) ≤ 7/2. Therefore, in all cases, 27 M (n) ≥ (n − 1)2 , or equivalently, 27 (n − 1)2 ≤ M (n).
On the other hand, the pattern shown in the figure for a 7 × 7 lattice (2/7 of the points are
black and every square path passes through a black point) can be extended to an arbitrary
n × n lattice by tiling an m × m lattices, m = 7dn/7e, with copies of the lattice in the figure,
and then removing (m − n) rows and columns from the top and left respectively, so that
the number of black point is less than or equal to 27 n2 .
National Team Selection Test Day 1
Time allowed: 4.5 hours 16 April 2005
No calculator is allowed
1. In 4ABC, ∠C = 90◦ . D is a point on the line AB. E and F are, respectively, the
feet of the perpendiculars from A and B to the line DC. ` and m are, respectively,
lines through E and F parallel to CB and CA. Prove that the lines AB, ` and m
are concurrent.
Soln: Let AB intersect ` and m at I1 and I2 respectively. Let P1 , P2 be the perpen-

dicular projection of I1 , I2 onto CD.
......
B
..............
m `
... ......
....
........ ..... ..
........ ......... ....
.. ........ .... ....
....
... ......... .....
... .... ..
.
..........
..
.. . .
.
I
... ....................
. .
..
. ..
...... ..
...
... .
. .
....... .. .... .... ..
...
.
... ... . .... .
.... ..
. ...... .... ...
. .
...
..... .... .. .
.. ...
..
..
..
.... ... .. . . ... ..... ..
. .
... . .. .
...
..
.. ....A
... ........... .
.. .
..
. .. . ... ..
.
. .
..
...
...
. ..
..
. .... .
...
.
...
..
..
..
........ .......... . ..
. . .
. .....
..
. ..
. ...
.
......... .
.
.....
..... .. .. ... ...
. ..
.......... .
. . ..
.. .
... .
. ...
. ...
........ .
. ..... .. ... .. ...
....... . . .
.....
. .. ..
. .. . . ..
.......
. .
. .
.
. ........ ...
. .. .. ... .
...
.... .
. .. .
... ... ....
..
........ . ....
.
.
.....
.....
. .
. .
.... .. ..... .... ... ..
...
..... .
. ...
α ... .. . ...
...................................................................................................................................................................................................................................................
D E C P F
Let ∠ACE = α. Then
EP1 + P2 F = EF
⇔ EI1 sin α + F I2 cos α = AC cos α + BC sin α
⇔ (F I2 − AC) cos α = (BC − EI1 ) sin α
CF CE
⇔ AC cos α = BC sin α
CD CD
CF CE
⇔ cos α = sin α
BC AC
⇔ sin α cos α = cos α sin α
Since the last line is true, we have EP1 + P2 F = EF , i.e., I1 = I2 and we are done.
DF DF
¡ −CD ¢ AC·CF
Note IF = AC · CD , so that IF − AC = AC · CD − AC = AC DFCD = CD .
2. Let a, b, c be positive real numbers. Prove that

· µ ¶ µ ¶¸ 13 1 1
a+b a+b+c a + (ab) 2 + (abc) 3
a· · ≥ .
2 3 3
1
Soln: By taking the geometric mean of the terms on each side of the inequalities
1 a+b 1 1 a+b+c 1
a= (a + a + a), ≥ (a + (ab) 2 + b) and = (a + b + c),
3 2 3 3 3
we have
· µ ¶ µ ¶¸ 13
a+b a+b+c 1 1 1 1 1
a· · ≥ (a + a + a) 3 (a + (ab) 2 + b) 3 (a + b + c) 3 .
2 3 3
By Cauchy-Schwarz inequality, we have
1 1 3 1
(a + (ab) 2 + b)(a + b + c) ≥ (a + a 4 b 4 + (bc) 2 )2 .
Thus, we have
· µ ¶ µ ¶¸ 13
a+b a+b+c 1 1 1 3 1 2
a· · ≥ (a + a + a) 3 (a + a 4 b 4 + (bc) 2 ) 3 .
2 3 3
Now by Hölder’s inequality, we have
2 2
1 1 3 1 2 1 2 1 1 3 1 1 1 1
(a+a+a) 3 (a+a 4 b 4 +(bc) 2 ) 3 ≥ a 3 a 3 +a 3 (a 4 b 4 ) 3 +a 3 ((bc) 2 ) 3 = a + (ab) 2 + (abc) 3 .
Consequently,
· µ ¶ µ ¶¸ 13 1 1
a+b a+b+c a + (ab) 2 + (abc) 3
a· · ≥ .
2 3 3
Also equality holds if and only if a = b = c.
The following inequality
GM(AM(x1 ),AM(x1 , x2 ),· · ·,AM(x1 , · · · , xn )) ≥ AM(GM(x1 ),GM(x1 , x2 ),· · ·,GM(x1 , · · · , xn ))
was conjectured by F. Holland and proved by K. Kedlaya in The American Math-

ematical Monthly Vol 101, No 4 April 1994, p355-357.
Level. Medium
2
3. Let n be a positive integer such that the sum of all its positive divisors (inclusive
of n) equals to 2n + 1. Prove that n is an odd perfect square.
Soln: Let n = 2q r, where r is odd. Thus 2n + 1 = σ(n) = σ(2q r) = σ(2q )σ(r) =

(2q+1 − 1)σ(r). Suppose r is not a perfect square. Let d1 , d2 , . . . , dk be all the distinct
√
divisors of r such that 1 ≤ di < r, i = 1, 2, . . . , k. Note that for every divisor di of r,
r √
di is also a divisor of r. Likewise, if√ d is a divisor of r such that r < d ≤ r, then dr
is also a divisor of r. Since 1 ≤ dr < r, it follows that dr is one of d1 , d2 , . . . , dk . Thus
di , dri , for i = 1, 2, . . . , k are all the distinct divisors of r. Note that di , dri are both odd,
Xk
r
so σ(r) = (di + ) is even. (Contradiction as 2n + 1 is odd.) Thus r must be a
i=1
di
perfect square.
Now, from (2q+1 −1)σ(r) = 2n+1 = 2q+1 r+1, we have 2q+1 r ≡ −1 (mod 2q+1 −1)
and 2q+2 r ≡ −2 (mod 2q+1 − 1). Note that either 2q+1 r or 2q+2 r is a perfect square,
depending of whether q is odd or even respectively.
Note that ∀ q ≥ 1, 2q+1 − 1 ≡ 3 (mod 4), thus ∃ some prime p ≡ 3 (mod 4)
dividing 2q+1 − 1. Note that −1 is a quadratic non-residue (mod p), thus 2q+1 r
cannot be a perfect square. Thus 2q+2 r is a perfect square, and so q is even.
If q = 0 then we are done. Suppose not, then since 2q+2 r ≡ −2 (mod 2q+1 − 1)
is a quadratic residue and (−1) is a quadratic non-residue (mod 2q+1 − 1), it follows
(2q+1 −1)2 −1 q+1 2
that (−1)( 8 +1)
= 1 ⇔ (2 −1)8
−1
is odd ⇒ 22q−1 − 2q−1 is odd. Note that
2q−1 q−1
2q − 1 ≥ 1, thus 2 is even. ⇒ 2 is odd and so can only be 1, which means q = 1.
(Contradiction since q must be even).
∴ n must be an odd perfect square.
3
National Team Selection Test Day 2
4. Positive real numbers x and y satisfy
3
x + 2x2 + · · · + 7x7 + 45x10 = = y + 2y 2 + · · · + 7y 7 + 8y 8 + 9y 9 .
2
Determine with proof if x or y is larger.
Soln: Note that y < 1/2. Indeed, if y ≥ 1/2, then
3
≥ y + 2y 2 + 3y 3 + 4y 4
2
1 1 1 1
≥ + 2( )2 + 3( )3 + 4( )4 ,
2 2 2 2
which is absurd. Therefore,
45y 10 − 9y 9 − 8y 8 = y 8 (15y − 8)(3y + 1) < 0.
Suppose that y ≥ x. Then
3
= x + 2x2 + · · · + 7x7 + 45x10
2
≤ y + 2y 2 + · · · + 7y 7 + 45y 10
3
< y + 2y 2 + · · · + 7y 7 + 8y 8 + 9y 9 = ,
2
a contradiction. Thus y < x.
5. Two circles Γ1 and Γ2 lying outside each other are tangent to a straight line ` at
points E and F respectively such that both are on the same side of `. The two
internal common tangents `1 and `2 to the two circles intersect at the point A such
that `1 meets ` at B and is tangent to Γ2 at H, and `2 meets ` at C and is tangent
to Γ1 at G. Suppose the extensions of EG and F H intersect at P . Prove that P A
is perpendicular to `.
4
Soln: Let the extension of P A meet ` at D. Let ∠GP A = x1 and ∠HP A = x2 .
ED P E sin x1
Applying sine rule to 4P ED and 4P F D, we have = · .
DF P F sin x2
P
.......
.........
.
.... ........
..
... ... ...
............. ...
... ..............
.
.
...
. ... .. x
...
...
....1 .....
... ...2 x
.
.. ... ...
.
. ... ...
... ...
...
. ..... ...
..
. .
... ...
.
.. ... ...
.
.. . ...
.
. .
... ...
....
. ..
. ...
..
. ... ...
.
.. .
. ...
. .. .... ...
...
.
.. .
..
.. . ...
..
. ... ...
... ... ...
... ... ...
2 ` . ..
.... .
.
.
.
...
...
..
. ....
.... .
...
.. . ..................................... ...
....... ..
. .
......
G ..
.
.
.
.
...
...
...
...
.
.
..
.....
.
... ........
. .........
. .
.........
.
.
...
...
...
... .......
` 1
.
...
.
...
..
... ...... ..
.
. H ... ..............
.. .. ... ...... . .
.
. ...............................
.. .
. . . .. ...
.. .
...
... ..... ... ....... ... ......
.. ... .....
..... ... ........... .. .....
.... O ... A ....... ... .. ...
... 1...... ... ........... ... ... ... ... ........... ... ... ... .......................................... ...... ...... ...
...
... . ... .. ...
. . . ....
. .
.. . . . .
. .
. ...
... ......... ... ....... ... ... ..... ...
...
...
.. ...
. ...
. .
.
..
..
... .
... .. ....... .. . ...... ..
. ..... .. . .
. O
... ..
.
. 2 ...
.
Γ 1 .......
... . ...
.
.. ....
.. ... ...
. ..
. ....
.................
..
...
. .
..
.
.
.. ..... ...
.......
. . .
... .
... .. .
.
...
.
.
Γ2
..... .
....... . .
...... ..... .
.
......
...... ....... .. ..... .... .....
...... ....... . .... ......
...............................................................................................................................................................................................................................................
`
EB D CF
sin x1 AG sin x2 AH
By sine rule on 4P AG and 4P AH, we have = and = .
sin ∠AGP PA sin ∠AHP PA
Thus
sin x1 AG sin ∠AGP AG sin ∠P EF O1 E P F
= · = · = · ,
sin x2 AH sin ∠AHP AH sin ∠P F E O2 F P E
where O1 , O2 are the centres of Γ1 and Γ2 respectively. Therefore,
ED P E sin x1 O1 E
= · = .
DF P F sin x2 O2 F
Since O1 A : O2 A = O1 E : O2 F , AD is parallel to O1 E. Thus, P A is perpendicular to
`.
6. Let k ∈ N and the numbers {1, 2, 3, . . . , 4k} are placed arbitrarily at 4k points
on the circumference of a circle. Prove that there exist 2k nonintersecting chords
joining these 4k points in pairs so that the difference between the numbers at the
endpoints of each chord is ≤ 3k − 1.
Soln: f (k) = 3k − 1
Lemma. Given any 2n-gon with n red vertices and n blue vertices, we can always
find n non-intersecting line segment with different coloured endpoints.
Induction. Clearly, There are 2 consecutive points of different colours. Draw a line
segment linking these 2 points. Now induct.
S
Now, Set A = {1, 2, . . . , k} {3k + 1, 3k + 2, . . . , 4k}. Set B = {k + 1, k + 2, . . . , 3k}.
5
By the above lemma, we can find a set of 2k lines satisfying given condition.
To prove 3k − 2 not possible, consider the clockwise arrangement of points {1, 3k +
1, 2, 3k + 2, 3, 3k + 2, . . . k, 4k} at the top and {k + 1, k + 2, k + 3, . . . , 3k} at the bottom.
Since no two consecutive vertices can be connected with a line at the top, all 2k line
must joint a top point and a bottom point. Hence, points 1 and 3k must be joined,
contradicting the assumption that the largest difference is 3k − 1.
Remark for question 3.
Consider the equation
(2q+1 − 1)σ(r) = 2q+1 r + 1 (∗)
Suppose q is an odd positive integer. Taking modulo 4, we get −σ ≡ 1 (mod 4), or

equivalently, σ(r) ≡ 3 (mod 4). Thus there exists a prime p such that p ≡ 3 (mod 4)
and p divides σ(r). Now taking modulo p in (*), we obtain 2q+1 r ≡ −1 (mod p). Since
2q+1 r is a perfect square, this contradicts the fact that −1 is a quadratic non-residue of
p as p ≡ 3 (mod 4).
6
National Team Selection Test I 2006
Time allowed: 4.5 hours 12 May 2006

1. Let AN C, CLB and BKA be triangles erected on the outside of the triangle
ABC such that ∠N AC = ∠KBA = ∠LCB and ∠N CA = ∠KAB = ∠LBC.
Let D, E, G and H be the midpoints of AB, LK, CA and N A respectively.
Prove that DEGH is a parallelogram.
2. Let n be an integer greater than 1 and let x1 , x2 , . . . , xn be real numbers such

that
|x1 | + |x2 | + · · · + |xn | = 1 and x1 + x2 + · · · + xn = 0.
Prove that
x1 x2 xn 1 1

+ + · · · + ≤ 1 − .

1 2 n
2 n
3. A pile of n pebbles is placed in a vertical column. This configuration is

modified according to the following rules. A pebble can be moved if it is at
the top of a column which contains at least two more pebbles than the column
immediately to its right. (If there are no pebbles to the right, think of this as
a column with 0 pebbles.) At each stage, choose a pebble from among those
that can be moved (if there are any) and place it at the top of the column
to its right. If no pebbles can be moved, the configuration is called a final
configuration. For each n, show that, no matter what choices are made at each
stage, the final configuration obtained is unique. Describe that configuration
in terms of n.
National Team Selection Test II 2006
Time allowed: 4.5 hours 13 May 2006

1. In the plane containing a triangle ABC, points A0 , B 0 and C 0 distinct from

the vertices of 4ABC lie on the lines BC, AC and AB respectively such that
AA0 , BB 0 and CC 0 are concurrent at G and AG/GA0 = BG/GB 0 = CG/GC 0 .
Prove that G is the centroid of 4ABC.
2. Let S be a set of sequences of length 15 formed by using the letters a and b

such that every pair of sequences in S differ in at least 3 places. What is the
maximum number of sequences in S?
3. Let n be a positive integer such that the sum of all its positive divisors (in-
clusive of n) equals to 2n + 1. Prove that n is an odd perfect square.
1. Let AN C, CLB and BKA be triangles erected on the outside of the trian-
gle ABC such that ∠N AC = ∠KBA = ∠LCB and ∠N CA = ∠KAB =
∠LBC. Let D, E, G and H be the midpoints of AB, LK, CA and N A re-
spectively. Prove that DEGH is a parallelogram.
Solution. The given condition means that 4AN C, 4CLB and 4BKA are
all similar.
C L
......................................
........................... ... ......
......... ..............
........... ....... ... .......
.
.................. .......
.......
... ......
.....
.
........ ....... ...
........
...
.. .....
.... ....... .....
.. .... .
.... .......
. . . .. ....... ..... .....
.....
... .... .
....... .....
......
G.......
. ..
.....
.... ... ... ... ...
R .. .. .
................ .....
.....
. . . ...
.. . .. .....
N .. .
..... ................
.. . ... ... ... ... .
.. ... ... ... ..
. ... ... ... ... ...... ...... ... . .
.
...... .. ........
.
........ .....
... ............. . ........ ..........
.
........ ....... ... E
........... ......
. ....
... .......... ............ ....... .....
....... .....
H ... .............. ... ... ... . . ...
.... ...... ... ..... ....... .....
....... ..... . P
.. ... .. ... ... ... ...
............ ... ....... ............ ......
......... ... ... ... ... ...
. ........... . ... ......... .........
A ..........................................................................................................................................................................................................................................................................................
............... ............... ..... ..
...............
............... D ....................
............... ......... . .
......
.... B
...............
............... ... .......................... ............
............... ... ........
..
............... ... ........
...............
............... .
.
..............................
....F
...
...
K
Let F be the midpoint of BK. Note that DF is parallel to AK and EF is

parallel LB. Since ∠DF E = 180◦ − ∠DF K − ∠EF B = 180◦ − (∠KAB +
∠ABK) − ∠LBP = 180◦ − ∠LBC − ∠ABK − ∠LBP = ∠ABC and
DF/EF = AK/LB = AB/CB, we see that 4DF E is similar to 4ABC.
Thus ∠BAC − ∠ACN = ∠F DE − ∠F DB so that DE is parallel to N C
and HG. Next, we also have ∠EDR = ∠F DR − ∠F DE = ∠F DB +
∠BDR − ∠F DE = ∠KAB + ∠BAC − ∠F DE = ∠KAB, and DE/DR =
2DE/AC = 2DF/AB = AK/AB so that 4EDR is similar to 4KAB. That
means 4EDR is similar to 4N CA. Therefore, DE/HG = 2DE/N C =
2DR/AC = 1. Consequently, DE is parallel and equal to HG. This shows
that DEGH is a parallelogram.
[Remark by Lim Wei Quan] Let M be the 4th vertice of the parallelogram
AKBM . Let F be the midpoint of BC. Then since triangles BM D, BLF ,
CN G are similar, M L = DF ×BL/BF = AC/2×2CN/CA = CN Similarly,
M N = CL Thus, M N CL is a parallelogram This gives, DE = M L/2 =
CN/2 = GH Also, DE||M L||CN ||GH Therefore, DEGH is a parallelogram.
2. Let n be an integer greater than 1 and let x1 , x2 , . . . , xn be real numbers such

that
|x1 | + |x2 | + · · · + |xn | = 1 and x1 + x2 + · · · + xn = 0.
Prove that
x1 x2 xn 1 1

+ + ··· + ≤ 1− .

1 2 n 2 n
Solution. The following lemma can be proved by direct simplification.
Lemma. Let Sk = a1 + a2 + · · · + ak . Then

n
X n−1
X
ak b k = S n b n + Sk (bk − bk+1 ).
k=1 k=1
Let Si = x1 + x2 + · · · + xi . By the given condition, Sn = 0 and |Si | ≤ 12 for
i = 1, · · · , n − 1. To see this, suppose |Si | > 12 . Then 1 = |x1 | + |x2 | + · · · +
|xn | ≥ |x1 + · · · + xi | + |xi+1 + · · · + xn | = |Si | + | − Si | = 2|Si | > 1, which is
a contradiction. By the lemma, we have
n
xk 1 n−1 1 1
X X
= Sn · + Sk − .
k=1 k n k=1 k k+1
Thus
n n−1 n−1
X x X
1 1

1 1

k X
= Sk − ≤ |Sk | −

k=1 k k k + 1 k=1 k k+1

k=1
n−1
X 1 1 1 1 1

≤ − = 1− .
k=1 2 k k+1 2 n
Second Solution. The inequality is achievable when x1 = ± 12 and xn = ∓ 12

and the rest of xi = 0. So the inequality can be proved by the smoothing
principle.
Let a1 ≥ · · · ≥ ak ≥ 0 be the nonnegative terms among the xi ’s and b1 ≤
b2 ≤ · · · ≤ bl < 0 be the negative terms among the xi ’s. Then we have
a1 + · · · + ak = 1/2 and b1 + · · · + bl = −1/2.
Without loss of generality, we can assume that the contribution from the
nonnegative terms are greater than the contributions from the negative terms
in the LHS. Note that for 0 < i < j, and x, y ≥ 0, we have x/i + y/j ≤
(x + y)/i + 0/j. Applying this, we see that the LHS is less than or equal to
Pk Pk
ai 0 0 i=1 bi 1 1

i=1
+ + ··· + + = 1− .
1 2 n−1 n 2 n
3. A pile of n pebbles is placed in a vertical column. This configuration is

modified according to the following rules. A pebble can be moved if it is
at the top of a column which contains at least two more pebbles than the
column immediately to its right. (If there are no pebbles to the right, think
of this as a column with 0 pebbles.) At each stage, choose a pebble from
among those that can be moved (if there are any) and place it at the top
of the column to its right. If no pebbles can be moved, the configuration is
called a final configuration. For each n, show that, no matter what choices
are made at each stage, the final configuration obtained is unique. Describe
that configuration in terms of n.
Solution. At any stage, let pi be the number of pebbles in column i for

i = 1, 2, . . . , where column 1 denotes the leftmost column. We will show that
in the final configuration, for all i for which pi > 0 we have pi = pi+1 + 1,
except that for at most one i∗ , pi∗ = pi∗ +1 . Therefore, the configuration
looks like the figure shown below, where there are c nonempty columns and
there are from 1 to c pebbles in the last diagonal row of the triangular
configuration. In particular, let tk = 1 + 2 + · · · + k = k(k + 1)/2 be the kth
triangular number. Then c is the unique integer for which tc−1 < n ≤ tc . Let
s = n − tc−1 . Then there are s pebbles in the rightmost diagonal, and so the
two columns with the same height are columns c − s and c − s + 1 (except if
s = c, in which case no nonempty columns have equal height).
.......
... ......
... .......
... .....
... .....
.....
........ .....
.. .....
..... ......... .....
.....
... ..... .....
... ..... .....
.....
.... ..... .....
• ...
...
...
.....
.....
.....
.....
.....
.....
.....
.....
... ..... .....
.
..... .....
... ..... .....
• ...
...
.....
• .....
.....
......
.....
.....
.....
.....
.....
..... .....
.. ..... .....
.... ..... .....
• ....
...
..
• • • .....
.....
.....
.....
.....
.....
.....
.....
..... .....
... ..... .....
... ..... .....
.....
• ...
•
...
...
• • • .....
.....
......
....
.....
.....
.....
.............................................................................................................................................
Final Configuration for n = 12
Another way to say this is

(
c−i if i ≤ c − s,
pi = (1)
c − i + 1 if i > c − s.
To prove this claim, we show that
(a) At any stage of the process, p1 ≥ p2 ≥ · · · .
(b) At any stage, it is not possible for there to be i < j for which pi = pi+1 ,
pj = pj+1 , and pi+1 − pj ≤ j − i − 1 (that is, the average decrease per
column from column i + 1 to column j is 1 or less).
(c) At any final configuration, pi − pi+1 = 0 or 1, with at most one i for

which pi > 0 and pi − pi+1 = 0.
In the proofs of (a), (b) and (c), we use the following terminology. Let a
k-switch be the movement of one pebble from column k to column k + 1, and
for any column i let a drop be the quantity pi − pi+1 .
To prove (a), suppose a sequence of valid moves resulted in pi < pi+1 for the
first time at some stage. Then the move leading to this stage must have been
an i-switch, but it would be contrary to the condition that column i have at
least 2 more pebbles than column i + 1, to allow switches.
To prove (b), if such a configuration were obtainable, there would be a mini-
mum value of j − i overall such obtainable configurations, and we now show
that there is no minimum. Suppose p1 , p2 , . . . was such a minimal configura-
tion. It cannot be that j = i + 1, for what would columns i, i + 1, i + 2 look
like just before the move that made the height equal? The move must have
been a k-switch for i − 1 ≤ k ≤ i + 2, but if so the configuration before the
switch was impossible (not decreasing).
Now suppose j > i + 1. Consider the first configuration C in the sequence for
which columns i, i + 1, j, j + 1 are at their final heights. Note that from pi+1
to pj the column decrease by exactly one each time in C, because if there
was a drop of 2 or more at some point, there would have to be another drop
of 0 in this interval to obtain an average of 1 or less, and thus j − i is not
minimal. The move leading to C was either an i-switch or a j-switch. If it
was the former, at the previous stage columns i + 1 and i + 2 had the same
height, violating the minimality of j − i. A similar contradiction arises if the
move was a j-switch.
Finally, to prove (c), if any drop is 2 or more, the configuration isn’t final.
However, if all drops are 0 or 1, and there were two drops of 0 between
nonempty columns (say between i and i + 1 and between j and j + 1), then
(b) would be violated. Thus a final configuration that satisfied (b) also
satisfies (c).
4. In the plane containing a triangle ABC, points A0 , B 0 and C 0 distinct from

the vertices of 4ABC lie on the lines BC, AC and AB respectively such that
AA0 , BB 0 and CC 0 are concurrent at G and AG/GA0 = BG/GB 0 = CG/GC 0 .
Prove that G is the centroid of 4ABC.
Solution. We take all segments to be directed segments: thus AG = −GA,

etc. We are given the condition
AG BG CG
= = . (1)
GA0 GB 0 GC 0
Now B, A0 and C are collinear points, one on each side (extended if necessary)
of triangle AGB 0 . By Menelaus’ theorem, we have
AA0 GB B 0 C
· · = −1. (2)
A0 G BB 0 CA
Similarly, from triangle CGB 0 and collinear points C 0 , B and A, we have
CC 0 GB B 0 A
· · = −1. (3)
C 0 G BB 0 AC
C0
....
... ...
... .....
.. .. ...
A A .
...
...
...
...
........... ..... ...
... ... ...... .... ............... ...
... .. ...... ... ..........
...........
...
.... ..... ........ .... . ..
..... .....
...
..
.
... ... .....
..... ...
. ...... ..... .....
. ...
... ... ..... ... . .
..... ..... ....
..
... ... ..... .... . .
..... ....... ....
.
.... ... ..... ..
. ...... ....... ...
... ... ..... .
. ... ..... ... ..... ......... G
0 ..... ... ..... 0 ... ..... ........
C B .... .. .. .....
....
.
. .........
.
.
... ...........
.
...
.
B 0................... ..............
...
.
. ........
....
...
. .
.. ..
. ....... ........ ....
..... ....... ......... ..............
........ .. ....... . ..... ..... . ..... ... ......
.
... ........ .. ....... ..... .... ...
. . . ..... ... ......
... ...... ..... . ......
... ............. ... ....... ..... ... ......
....... ..... ................ ..... ... ....... ..... ... .....
.. ......... .. .....
... ......... ..... ... .....
.
.
...
.
..
............
. G .
...
.
.
.. ..........
..
..........
.....
..... .
...
.
...........
.. . ..... ...
..... ...
.....
.....
.....
.
.. ........... ... .... ........ ......... ...
. .
.......... ..
. ..
........ .....
.... ........... ...
. . ........... ...... ... .
.......... .. . .....
.....
.. .............. ... ....... ... . . .. ........... ........
. .
.... .....
.
.......................................................................................................................................................................... . .... .
...........................................................................................................................................................................................................
B A0 C B C A0
By adding 1 to each member of AG/GA0 = CG/GC 0 from (1), we have
AA0 CC 0
= . (4)
GA0 GC 0
An easy combination of (2),(3),(4) gives
CB 0 = B 0 A. (5)
Since these are directed segments, (5) means that B 0 is the midpoint of AC.
Similarly, by selecting other triangles, we obtain C 0 as midpoint of AB, and
A0 as midpoint of BC. Thus G is the centroid of 4ABC.
So (1) implies that A0 , B 0 , C 0 lie in the open segments BC, CA, AB, respec-
tively, and the “open segments” restriction in the hypothesis is unnecessary.
In order for any other conclusion to hold, we must interpret the symbols in
(1) to mean undirected segments. Then, upon replacing them by directed
segments we see that either all ratios have the same sign - the case already
discussed - or two have one sign and the third the opposite sign. Suppose,
say,
AG CG BG
0
= 0
=− . (6)
GA GC GB 0
By the original argument, the first equation implies (5) so that B 0 is the
midpoint of AC. From AG/GA0 = −BG/GB 0 , we have
AG GA0 BG GB 0
+ − 1 = − − + 1,
GA0 GA0 GB 0 GB 0
whence
AA0 BB 0
= 2 − . (7)
GA0 GB 0
Now (5) implies that B 0 C/CA = −1/2, whence (2) becomes
AA0 GB
· = 2. (8)
A0 G BB 0
Putting GB/BB 0 = GB 0 /BB 0 − 1 in (8) and eliminating AA0 /A0 G from (7)
and (8), we obtain
2r2 − r + 1 = 0, (9)
where r = GB 0 /BB 0 . But the roots of (9) are not real, so that this case
cannot hold. Thus the case of medians is the only conclusion.
5. Let S be a set of sequences of length 15 formed by using the letters a and b

such that every pair of sequences in S differ in at least 3 places. What is the
maximum number of sequences in S?
Solution. The answer is 211 = 2048.

We may identify a as 0 and b as 1. Then S is simply the set of binary 15-
tuples satisfying the condition that any two tuples in S differ in at least 3
places. For each element s of this set S, there are exactly 15 + 1 = 16 tuples
(including itself) that differ from it in at most 1 place. Let Bs denote the
set of these tuples. For any distinct s, t ∈ S, we must have Bs ∩ Bt = ∅,
otherwise s and t would differ in at most two places. Hence |S| · 16 ≤ 215 ,
and so |S| ≤ 211 . An explicit S meeting the upper bound is then given as
follows:
For each of the 211 tuples (a1 , . . . , a11 ), associate to it the tuple a = (a1 , . . . , a15 )
in S, where
a12 = a5 + a6 + a7 + a8 + a9 + a10 + a11 (mod 2)

a13 = a2 + a3 + a4 + a8 + a9 + a10 + a11 (mod 2)
a14 = a1 + a3 + a4 + a6 + a7 + a10 + a11 (mod 2)
a15 = a1 + a2 + a4 + a5 + a7 + a9 + a11 (mod 2)
To verify that this S is indeed valid, simply check that if any one of the first
eleven values of a are changed, then at least two of a12 , a13 , a14 , a15 must also
be changed, and if any two of the first eleven values of a are changed, then
at least one of a12 , a13 , a14 , a15 must also be changed.
6. Let n be a positive integer such that the sum of all its positive divisors
(inclusive of n) equals to 2n + 1. Prove that n is an odd perfect square.
Solution. Let n = 2q r, where r is odd. Thus 2n + 1 = σ(n) = σ(2q r) =

σ(2q )σ(r) = (2q+1 − 1)σ(r).
Suppose r is not a perfect √ square. Let d1 , d2 , . . . , dk be all the distinct divisors
of r such that 1 ≤ di < r, i = 1, 2, . . . , k. Note that for every divisor √ di of r,
r
is also a divisor of r. Likewise, if d is a divisor of r such that r < d ≤ r,
di √
then dr is also a divisor of r. Since 1 ≤ dr < r, it follows that dr is one of
d1 , d2 , . . . , dk . Thus di , dri , for i = 1, 2, . . . , k are all the distinct divisors of r.
k
X r
Note that di , dri are both odd, so σ(r) = (di + ) is even. (Contradiction
i=1 di
as 2n + 1 is odd.) Thus r must be a perfect square.
There are two ways to complete.
(i) From (2q+1 −1)σ(r) = 2n+1 = 2q+1 r+1, we have 2q+1 r ≡ −1 (mod 2q+1 −
1) and 2q+2 r ≡ −2 (mod 2q+1 −1). Note that either 2q+1 r or 2q+2 r is a perfect
square, depending of whether q is odd or even respectively.
∀ q ≥ 1, 2q+1 − 1 ≡ 3 (mod 4), thus ∃ some prime p ≡ 3 (mod 4) dividing
2q+1 − 1. Note that −1 is a quadratic non-residue mod p, thus 2q+1 r cannot
be a perfect square. Thus 2q+2 r is a perfect square, and so q is even.
If q = 0 then we are done. Suppose q ≥ 2, then since 2q+2 r ≡ −2 (mod 2q+1 −
1) is a quadratic residue and −1 is a quadratic non-residue (mod 2q+1 − 1).
In other words, 2 is a quadratic non-residue (mod 2q+1 − 1).
(2q+1 −1)2 −1 q+1 2
It follows that (−1)( 8
+1)
= 1 ⇔ (2 −1)8
−1
is odd ⇒ 22q−1 − 2q−1 is
odd. Note that 2q − 1 ≥ 1, thus 22q−1 is even so that 2q−1 is odd and so can
only be 1, which means q = 1. (Contradiction since q must be even).
As q ≥ 2, 2q+1 − 1 ≡ 7 (mod 8) so that it cannot be a power of 3. Hence
there exists a prime factor p of 2q+1 − 1 such that p ≡ 1, 5, or 7 (mod 8).
(ii) If q = 0, then we are done. If not, then from (2q+1 − 1)σ(r) = 2n + 1 =
2q+1 r + 1, we have 2q+1 r ≡ r ≡ −1 (mod 2q+1 − 1). Now 2q+1 − 1 has a prime
factor p ≡ −1 (mod 4). Thus r ≡ −1 (mod 4). But r is a perfect square
and −1 is not a quadratic residue mod 4, a contradiction. So q = 0.
Therefore, n must be an odd perfect square.

1. Find all pairs of nonnegative integers (x, y) satisfying (14y)x + y x+y = 2007.
2. Let ABCD be a convex quadrilateral inscribed in a circle with M and N

the midpoints of the diagonals AC and BD respectively. Suppose AC bisects
∠BM D. Prove that BD bisects ∠AN C.
3. Let a1 , a2 , . . . , a8 be 8 distinct points on the circumference of a circle such

that no three chords, each joining a pair of the points, are concurrent. Every
4 of the 8 points form a quadrilateral which is called a quad. If two chords,
each joining a pair of the 8 points, intersect, the point of intersection is called
a bullet. Suppose some of the bullets are coloured red. For each pair (i, j),
with 1 ≤ i < j ≤ 8, let r(i, j) be the number of quads, each containing ai , aj
as vertices, whose diagonals intersect at a red bullet. Determine the smallest
positive integer n such that it is possible to colour n of the bullets red so that
r(i, j) is a constant for all pairs (i, j).

4. Two circles Γ1 and Γ2 touch externally at the point C and internally at points
A and B respectively with another circle Γ centred at O. Suppose the common
tangent of Γ1 and Γ2 at C meets Γ at P with P A = P B. Prove that P O is
perpendicular to AB.
5. Prove the inequality

X ai aj n X
≤ ai aj
i<j ai + aj 2(a1 + a2 + · · · + an ) i<j
for all positive real numbers a1 , a2 , . . . , an .
6. Let A, B, C be 3 points on the plane with integral coordinates. Prove that

there exists a point P with integral coordinates distinct from A, B and C such
that the interiors of the segments P A, P B and P C do not contain points with
integral coordinates.
Solution to National Team Selection Test 2007
Solution. Suppose x and y are nonnegative integers satisfying (14y)x + y x+y =
2007. Note that y ≥ 1 because y = 0 does not satisfy the equation. For x ≥ 3,
we have (14y)x + y x+y ≥ 14x ≥ 143 = 2744 > 2007. Thus x = 0, 1 or 2. If
x = 0, then the equation becomes 1 + y y = 2007. That is y y = 2006 = (2)(17)(59)
which obviously has no solution in y. If x = 1, then 14y + y y+1 = 2007. That is
y(14 + y y ) = 2007 = 32 (223). Thus y = 1, 3, 9 or 223. But none of these satisfy
14y + y y+1 = 2007. If x = 2, then y 2 (142 + y y ) = 32 (223). Since y = 1 does not
satisfy this equation, we must have y = 3. Consequently x = 2 and y = 3 and they
do satisfy the given equation. Therefore (x, y) = (2, 3) is the only solution to the
equation.
2. Let ABCD be a convex quadrilateral inscribed in a circle with M and N the

midpoints of the diagonals AC and BD respectively. Suppose AC bisects ∠BM D.
Prove that BD bisects ∠AN C.
Solution. Let O be the center of the circle. Let the extensions of CN and DM
meet the circle at E and F respectively. Join AE and BF .
E .....................
..................................................
...........
..........
......... .. .... .........
. .. . . ........ .. ..
........
.......
.. . .. .... .. . ......
......
..... .. .. ......
. . . . .
. .. ..... . . .. .....
.....
....... .. . ..
.. .....
..
.... .. . .....
.. ..... . . ..
..
.....
....
.. .. .. . ...
.... .
.. ..
..
...
...
... . ...
... .. .. ...
. . . .
..
..
. . . .. ...
... . . ...
.. .. .. ...
. . .. ...
... .. .. ...
.... .. ... ...
... .. .. ...
... .. . ...
... . .. .. ...
. .. ...
... .. .. ...
...
·O
.. ... ...
....
...
... ..
..
.. ..
..
.. .
.....
. .
.. ..
........................
. .
...
D
... .. .. ................................. .........
... .. ............. ....... . .
.........
... .. ............ .... .............. ... ... ..
.. ............ .
... ............. ........
....... ...
.. .. ..
...
.
..
. ........ .
.............. ......
. .
... .... ....
... .. ...
..
. ............. .......
.. . ..
... .. ..
... ............ ....... ..
.. .... ........
... .. ............. .......
............. .......
...
...
... .. ..
..
........ . . . . . . .. . . . ............
.......
............. N .
................................................... ......... .......
.. ... ... ..
....
... .....
... .. ........................ ......................................................
....................................................................................................................................................................................................................................................................................................................................................
..................... . . .
A ...
...
................
................
................ M ... ..
....
. .. . . .. ....... .. .
.......
....... R
.
... .......
.. .......
.
...
.
C
...
..... ....................... .......
....... .. ........
..... . . . ... .......................... ....... ..... ................
..... .. . ................... .. ....... . .
.. .. ..
..... .... ................ .
................ .............. ................
..... ... .. ...............
..... ........ ...... ..
.......... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .................................................
...... ....
F .......
.......
........
.......... ........
.......
......
..
B
............ ..........
...................................................................
First note that F is the reflection of B about the line OM so that M F = M B.

Thus ∠BF M = ∠BM C and F B is parallel to M C. Therefore ∠BAD = ∠BF M =
∠BM C. As ∠ADB = ∠M CB, we have 4ADB is similar to 4M CB. Therefore,
BD DA DA AC DA DA
BC = CM = AC/2 so that BC = BD/2 = N B . As ∠DAC = ∠N BC, we see that
4DAC is similar to 4N BC so that ∠BN C = ∠ADC = ∠AEC. This means AE
is parallel to BD. As N O is perpendicular to BD, we have the extension of N O
is perpendicular to AE and meets AE at its midpoint. Thus 4AN E is isosceles
with ∠N EA = ∠N AE. Consequently, ∠CN R = ∠N EA = ∠N AE = ∠AN R.
2nd Solution. Set up a coordinate system in which M is the origin, AC is the

x-axis and the center of the circle lies on the y-axis with coordinate (0, d). Let
the radius of the circle be r and let the gradients of the lines M√D and M B be
α√and −α respectively. Then the coordinates of A and C are (− r2 − d2 , 0) and
( r2 − d2 , 0) respectively. By solving the equation of the circle x2 + (y − d)2 = r2
and the equation y = αx of the line M D, we get the coordinate of D equal to
√ √ !!
αd + r2 − d2 + α2 r2 αd + r2 − d2 + α2 r2
,α .
1 + α2 1 + α2
Replacing α by −α, we obtain the coordinate of B as

√ √ !!
−αd + r2 − d2 + α2 r2 −αd + r2 − d2 + α2 r2
, −α .
1 + α2 1 + α2
√ !
r2 − d2 + α2 r2 α2 d
Therefore, the midpoint N of BD has coordinate , . Let
1 + α2 1 + α2
r2 −d2
AC intersect BD at R. Then the x-coordinate of R is found to be √
r2 −d2 +α2 r2
.
Thus a direct calculation gives
2 2
2r2 − 2d2 + α2 r2 + 2 (r2 − d2 )(r2 − d2 + α2 r2 )
p
RA NA

= 2 p = .
RC 2r − 2d2 + α2 r2 − 2 (r2 − d2 )(r2 − d2 + α2 r2 ) NC
RA NA
That is RC = NC . Using the Angle Bisector Theorem, we have ∠AN B = ∠CN B.
3. Let a1 , a2 , . . . , a8 be 8 distinct points on the circumference of a circle such that no

three chords, each joining a pair of the points, are concurrent. Every 4 of the 8
points form a quadrilateral which is called a quad. If two chords, each joining a
pair of the 8 points, intersect, the point of intersection is called a bullet. Suppose
some of the bullets are coloured red. For each pair (i, j), with 1 ≤ i < j ≤ 8, let
r(i, j) be the number of quads, each containing ai , aj as vertices, whose diagonals
intersect at a red bullet. Determine the smallest positive integer n such that it is
possible to colour n of the bullets red so that r(i, j) is a constant for all pairs (i, j).
Solution. First note that each bullet is determined by a unique quad. Thus a
red quad is a quad whose associated bullet is red. For such an n, let the constant
r(i, j) be t. Form an incidence matrix whose rows are indexed by pairs (i, j) and
whose columns are indexed by the n red quads. The ((ij), (pqrs)) entry is 1 if
{i, j} ⊆ {p, q, r, s} and is 0 otherwise. Then each row has t 1’s and each column
has 6 1’s. Thus 6n = t 82 , i.e., 3n = 14t. The smallest n is n = 14 and t = 3.
The following 14 red quads have the desired property:
(1234),(2345),(3456),(4567),(5678),(6781),(7812),
(8123),(1256),(2367),(3478),(4581),(1357),(2468)
4. Two circles Γ1 and Γ2 touch externally at the point C and internally at points A and
B respectively with another circle Γ centred at O. Suppose the common tangent
of Γ1 and Γ2 at C meets Γ at P with P A = P B. Prove that P O is perpendicular
to AB.
Solution. Let r1 , r2 and r be the radii of Γ1 , Γ2 and Γ respectively. It suffices to
prove that r1 = r2 . Then P, O, C are collinear and P O is perpendicular to AB.
P
...........................................................
......... ..... ... .......
....... ......
...... ... ... ... ......
........ .. .. ... .....
......
.
.
.... .... .... .....
....
. ... .. ... .....
. ...
.
.. .. .
.
. ... ...
.
.... ..
. ..
. ...
...
...
...
... ..
. ..
. . ...
.. ...
.... ..
. . ... ...
... ... ..
. .
... ...
... ... ..
. ... ...
...
... X ..
..... ... .........
...
...........
. ....... .. .......
..
.
.
.
. .. ...
. ..
.
.........
. .. .
.
Y
... ........
...
..
.
... .... .
. . .
. . . . ..
... .
..... ... ..... .. ....
. . .. ... .....
. ....
... .. . ... . .. ... ... .
... .. .
. .. .
....... .
... .... .... . .
.
... ... ... ...... ... ... ...
... ... ... ... ... ... ...
... ..
...... ....
..... ..
.. C .
.
.
.....
. ....
...
... ........
... .....
. ..
..... .. ..
. .... ... ..... .
....... .
.. ..... . ..
..... .... .....
............. .............. .......
...................................
A ................
........
........... ........
.....
B
.........................................
First AX/AP = r1 /r so that P X/P A = (r − r1 )/r. Also Pq C 2 = P X · P A.

2 2 r
Thus P A /P C = P A/P X = r/(r − r1 ). Therefore, P A = P C r−r 1
. Similarly,
q
r
PB = PC r−r1 . As P A = P B, we have r1 = r2 .

X ai aj n X
≤ ai aj
i<j
ai + aj 2(a1 + a2 + · · · + an ) i<j
P
Solution. Let S = i ai . Denote by L and R the expressions on the left and right
hand side of the proposed inequality. We transform L and R using the identity
X X
(ai + aj ) = (n − 1) ai . (1)
i<j i
And thus:
!
X ai aj X1 (ai − aj )2 n−1 1 X (ai − aj )2
L= = ai + aj − = ·S − . (2)
i<j
ai + aj i<j
4 ai + aj 4 4 i<j ai + aj
X
To express R we express the sum ai aj in two ways; in the second transformation
i<j
identity (1) will be applied to the squares of the numbers ai :
!
X 1 X
ai aj = S2 − a2i ;
i<j
2 i
X 1 X 2 n−1X 2 1X
ai aj = ai + a2j − (ai − aj )2 = ai − (ai − aj )2 .
i<j
2 i<j 2 i
2 i<j
Multiplying the first of these equalities by n − 1 and adding the second one we
obtain
X n−1 1X
n ai aj = · S2 − (ai − aj )2 .
i<j
2 2 i<j
Hence
n X n−1 1 X (ai − aj )2
R= ai aj = ·S− . (3)
2S i<j 4 4 i<j S
Now compare (2) and (3). Since S ≥ ai + aj for any i < j, the claim L ≥ R results.
2nd Solution. Let S = a1 + a2 + · · · + an . For i 6= j,

X X
ai ak + aj ak + 2ai aj
4ai aj (ai − aj )2 (ai − aj )2 k6=i k6=j
= ai + aj − ≤ ai + aj − = .
ai + aj ai + aj S S
The statement is obtained by summing up these inequalities for all pairs i, j:
 
X ai aj 1 X X ai aj 1 X X X X
= ≤ ai ak + aj ak + 2ai aj 
i<j
ai + aj 2 i j6=i ai + aj 8S i j6=i k6=i k6=j
 
1 X X X XXX XX
= ai ak + aj ak + 2 ai aj 
8S k i6=k j6=i k j6=k i6=j i j6=i
 
1 X X XX XX
= (n − 1)ai ak + (n − 1)aj ak + 2 ai aj 
8S k i6=k k j6=k i j6=i
n XX n X
= ai aj = ai aj .
4S i j6=i 2S i<j
6. Let A, B, C be 3 points on the plane with integral coordinates. Prove that there
exists a point P with integral coordinates distinct from A, B and C such that
the interiors of the segments P A, P B and P C do not contain points with integral
coordinates.
Solution. Let A = (a1 , a2 ), B = (b1 , b2 ), C = (c1 , c2 ) where a1 , a2 , b1 , b2 , c1 , c2 are
integers. If a lattice point X = (x1 , x2 ) is between A = (a1 , a2 ) and B = (b1 , b2 ),
then there exist relatively prime positive integers u and v such that (u + v)xi =
uai + vbi , i = 1, 2. That is (u + v)(xi − bi ) = u(ai − bi ), i = 1, 2. This implies
that (u + v) divides ai − bi , for i = 1, 2. That is a1 − b1 and a2 − b2 have a
common factor. Therefore, it suffices to find a lattice point P = (x1 , x2 ) such that
gcd(x1 − a1 , x2 − a2 ) = 1, gcd(x1 − b1 , x2 − b2 ) = 1 and gcd(x1 − c1 , x2 − c2 ) = 1.
Let p = 2 or 3. There are at least 4 distinct values (m1 , m2 ) modulo p. Thus
there exists (r1 , r2 ) such that (r1 , r2 ) 6≡ (a1 , a2 ), (b1 , b2 ), (c1 , c2 ) (mod 2). Similarly,
there exists (s1 , s2 ) such that (s1 , s2 ) 6≡ (a1 , a2 ), (b1 , b2 ), (c1 , c2 ) (mod 3). By the
Chinese Remainder Theorem, there exists (p1 , p2 ) such that p1 ≡ r1 (mod 2),
p1 ≡ s1 (mod 3); and p2 ≡ r2 (mod 2), p2 ≡ s2 (mod 3). We can also require
p1 6= a1 , b1 , c1 and p2 6= a2 , b2 , c2 . This ensures that (p1 , p2 ) 6≡ (a1 , a2 ), (b1 , b2 ),
(c1 , c2 ) (mod 2) and (mod 3).
Suppose p is a prime larger than 3 such that p divides p2 − a2 , p2 − b2 , or p2 − c2 .
There are only a finite numbers of such primes p. Let the set of all such primes be Y .
For each p ∈ Y , pick an integer zp 6≡ a1 , b1 , c1 mod p. Since p > 3, this is possible.
Apply the Chinese Remainder Theorem to find an x1 such that x1 ≡ p1 (mod 2),
x1 ≡ p1 (mod 3) and x1 ≡ zp (mod p), p ∈ Y . Take x2 = p2 . Then P = (x1 , x2 )
is the desired point.

1. In 4ABC, D is a point on AB and E is a point on AC such that BE and CD are

bisectors of ∠B and ∠C respectively. Let Q, M and N be the feet of perpendiculars
from the midpoint P of DE onto BC, AB and AC, respectively. Prove that
PQ = PM + PN.
2. Let x1 , x2 , . . . , xn be positive real numbers such that x1 x2 · · · xn = 1. Prove that

1 1 1
+ + ··· + ≤ 1.
n − 1 + x1 n − 1 + x2 n − 1 + xn
P103 p−1 .
3. Find all odd primes p, if any, so that p divides n=1 n

4. Let C be a circle centred at O, and let ABP be a line segment such that A, B lie
on C and P is a point outside C. Let C be a point on C such that P C is tangent
to C and let D be the point on C such that CD is a diameter of C and intersects
AB inside C. Suppose that the lines DB and OP intersect at E. Prove that AC is
perpendicular to CE.
5. Find all functions f : R −→ R so that
(x + y)(f (x) − f (y)) = (x − y)f (x + y)
for all x, y ∈ R.
6. Fifty teams participate in a round robin competition over 50 days. Moreover, all
the teams (at least two) that show up in any day must play against each other.
Prove that on every pair of consecutive days, there is a team that has to play on
those two days.
Solution to National Team Selection Test 2007
Solution. Suppose x and y are nonnegative integers satisfying (14y)x + y x+y = 2007.
Note that y ≥ 1 because y = 0 does not satisfy the equation. For x ≥ 3, we have
(14y)x + y x+y ≥ 14x ≥ 143 = 2744 > 2007. Thus x = 0, 1 or 2. If x = 0, then the equation
becomes 1 + y y = 2007. That is y y = 2006 = (2)(17)(59) which obviously has no solution in
y. If x = 1, then 14y + y y+1 = 2007. That is y(14 + y y ) = 2007 = 32 (223). Thus y = 1, 3, 9
or 223. But none of these satisfy 14y + y y+1 = 2007. If x = 2, then y 2 (142 + y y ) = 32 (223).
Since y = 1 does not satisfy this equation, we must have y = 3. Consequently x = 2 and
y = 3 and they do satisfy the given equation. Therefore (x, y) = (2, 3) is the only solution
to the equation.
2. Let ABCD be a convex quadrilateral inscribed in a circle with M and N the midpoints
of the diagonals AC and BD respectively. Suppose AC bisects ∠BM D. Prove that BD
bisects ∠AN C.
Solution. Let O be the center of the circle. Let the extensions of CN and DM meet the
circle at E and F respectively. Join AE and BF .
E ......................
................................................
...........
........... .........
......... .. .... ........
. . .. . . ........ .. .. .......
. .. . ..... . .. ..
......
......
... . .... . . .
. ......
. .. ..... .
. .. .....
.....
....... .. . ..
.. .....
.....
...
. .. . . .....
.... . ..
. .. . . . ....
. . .. .. .. ...
... . . .. ...
... .. ...
. . . .. .. ...
...
. . . . .
..
...
...
..
. . . ...
... ... .. ...
... .. .. ...
.. . .. ...
. . .. ...
.... .. ... ...
... .. . ...
... .. .. ...
... . .. .. ...
. ..
... .. .. ...
... .. .. ...
·O
... .. .. ...
...
... ..
.. ..
.. . .
... . . . . .
.........................
. .. D
.. . .
... .
.................. .........
... .. ............ .. ... .. ... .
... .. ..... .... . . ..
. ......
. . ........
.. .. ........
... .. ............. ..
.
... ... ...
... .. ............. ......... ... ... ...
... .. ............ ....... ...
.. ............. ....... ... .... ....
...
... .. . .
..................... ............. ..
.. ..
. .. ..
........ ...... .
.. .... ........
... .. ............. .......
... .. ............. .......
...
... .. ..
....... . . . . . . . .. . . ............
................... N ..
............................................. ........ ...... .. ..
.. .. .. ..
.
. ..
.
...
.. .
.....
... .. ........................ ......................................................
....................................................................................................................................................................................................................................................................................................................................................
..................... . .. . .. .....
A ...
...
................
. . . .................
................ ......... M . .. . ..... ................
........
....... R ... ........
.. .......
C
...
..... .................
. . .
.......
. .
. .. ...............
..... ... ................ ....... . .......
..... ..... ................
................
.......
. ... ..........
..... .... ................ .............. ................
..... ..... ............. ..... .............
..... ........
...... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ..................................................
...... ...
F ....... ...... B
....... .......
........ ........
.........
............ .. .
. ..............
....................... ......
...................................
First note that F is the reflection of B about the line OM so that M F = M B. Thus
∠BF M = ∠BM C and F B is parallel to M C. Therefore ∠BAD = ∠BF M = ∠BM C. As
∠ADB = ∠M CB, we have 4ADB is similar to 4M CB. Therefore, BD DA DA
BC = CM = AC/2
AC DA DA
so that BC = BD/2 = N B . As ∠DAC = ∠N BC, we see that 4DAC is similar to
4N BC so that ∠BN C = ∠ADC = ∠AEC. This means AE is parallel to BD. As N O
is perpendicular to BD, we have the extension of N O is perpendicular to AE and meets
AE at its midpoint. Thus 4AN E is isosceles with ∠N EA = ∠N AE. Consequently,
∠CN R = ∠N EA = ∠N AE = ∠AN R.
2nd Solution. Set up a coordinate system in which M is the origin, AC is the x-axis and
the center of the circle lies on the y-axis with coordinate (0, d). Let the radius of the circle
be r and let the gradients of the √ lines M D and M √B be α and −α respectively. Then the
coordinates of A and C are (− r2 − d2 , 0) and ( r2 − d2 , 0) respectively. By solving the
equation of the circle x2 + (y − d)2 = r2 and the equation y = αx of the line M D, we get
the coordinate of D equal to
√ √ !!
αd + r2 − d2 + α2 r2 αd + r2 − d2 + α2 r2
,α .
1 + α2 1 + α2
Replacing α by −α, we obtain the coordinate of B as
√ √ !!
−αd + r2 − d2 + α2 r2 −αd + r2 − d2 + α2 r2
, −α .
1 + α2 1 + α2
√ !
r 2 − d2 + α 2 r 2 α 2 d
Therefore, the midpoint N of BD has coordinate , . Let AC
1 + α2 1 + α2
2 2
r −d
intersect BD at R. Then the x-coordinate of R is found to be √r2 −d 2 +α2 r 2
. Thus a direct
calculation gives
2 p 2
2r2 − 2d2 + α2 r2 + 2 (r2 − d2 )(r2 − d2 + α2 r2 )

RA NA
= p = .
RC 2r2 − 2d2 + α2 r2 − 2 (r2 − d2 )(r2 − d2 + α2 r2 ) NC
RA NA
That is RC = NC . Using the Angle Bisector Theorem, we have ∠AN B = ∠CN B.
3. Let a1 , a2 , . . . , a8 be 8 distinct points on the circumference of a circle such that no three

chords, each joining a pair of the points, are concurrent. Every 4 of the 8 points form a
quadrilateral which is called a quad. If two chords, each joining a pair of the 8 points,
intersect, the point of intersection is called a bullet. Suppose some of the bullets are
coloured red. For each pair (i, j), with 1 ≤ i < j ≤ 8, let r(i, j) be the number of quads,
each containing ai , aj as vertices, whose diagonals intersect at a red bullet. Determine the
smallest positive integer n such that it is possible to colour n of the bullets red so that
r(i, j) is a constant for all pairs (i, j).
Solution. First note that each bullet is determined by a unique quad. Thus a red quad is
a quad whose associated bullet is red. For such an n, let the constant r(i, j) be t. Form
an incidence matrix whose rows are indexed by pairs (i, j) and whose columns are indexed
by the n red quads. The ((ij), (pqrs)) entry is 1 if {i, j} ⊆ {p, q, r, s} and is 0 otherwise.
Then each row has t 1’s and each column has 6 1’s. Thus 6n = t 82 , i.e., 3n = 14t. The
smallest n is n = 14 and t = 3.
The following 14 red quads have the desired property:
(1234),(2345),(3456),(4567),(5678),(6781),(7812),
(8123),(1256),(2367),(3478),(4581),(1357),(2468)
4. Two circles Γ1 and Γ2 touch externally at the point C and internally at points A and B
respectively with another circle Γ centred at O. Suppose the common tangent of Γ1 and
Γ2 at C meets Γ at P with P A = P B. Prove that P O is perpendicular to AB.
Solution. Let r1 , r2 and r be the radii of Γ1 , Γ2 and Γ respectively. It suffices to prove
that r1 = r2 . Then P, O, C are collinear and P O is perpendicular to AB.
P
.......................................................
.......... .. .......
....... ... .. .... ......
.......... ... ... .. ......
.....
.....
. .. .. ... .....
.
...... .... .... .... .....
.... ... ... .... ...
... ...
.... ..
. ..
. .... ...
... ..
. ..
. ... ...
..
. ... ..
. .
...
...
...
.... ..
. ..
. ... ...
... ... ..
. .
... ...
... ... ..
. . ...
...
...
X . . . . . ........
.
..
.
............
.
..
.
.
.
. .... .
. .
. .
.
...
...........
. .
Y
... . .
..
..
... ... ..... ...
. ...... .. .......
.. . .
.
... .......
. . ..
... .. ... ..
.
..... . ...
. .
. .. ... .....
. ....
... .... .. ... .. .. ... ... ..
... .. . ........ ... .... .... .. .
... ... ... ..... ... ... ..
... ... ... ... . ..
......
..... ....
.. C ..
...
......
...
... ........
.. ...
...... ..
. . . ....
. ... ......
....... . . .......
..... .... ........ ..
...... ..... ....
A ..................................
.......
.........
.......
.................................
.........
B
................. .................... . .
.
..........
First AX/AP = r1 /r so that P X/P A = (r − r1 )/r. Also qP C 2 = P X · P A. Thus
r
P A2 /P C 2 = P A/P X = r/(r − r1 ). Therefore, P A = P C r−r 1
. Similarly, P B =
q
r
P C r−r 1
. As P A = P B, we have r1 = r2 .

X ai aj n X
≤ ai aj
a + aj
i<j i
2(a1 + a2 + · · · + an ) i<j

P
Solution. Let S = i ai . Denote by L and R the expressions on the left and right hand
side of the proposed inequality. We transform L and R using the identity
X X
(ai + aj ) = (n − 1) ai . (1)
i<j i
And thus:
X1 (ai − aj )2 1 X (ai − aj )2

X ai aj n−1
L= = ai + aj − = ·S− . (2)
a + aj
i<j i i<j
4 ai + aj 4 4 i<j ai + aj
X
To express R we express the sum ai aj in two ways; in the second transformation identity
i<j
(1) will be applied to the squares of the numbers ai :
!
X 1 X
ai aj = S2 − a2i ;
i<j
2 i
X 1X 2 n−1X 2 1X
ai aj = ai + a2j − (ai − aj )2 = ai − (ai − aj )2 .
i<j
2 i<j 2 i
2 i<j
Multiplying the first of these equalities by n − 1 and adding the second one we obtain
X n−1 1X
n ai aj = · S2 − (ai − aj )2 .
i<j
2 2 i<j
Hence
n X n−1 1 X (ai − aj )2
R= ai aj = ·S− . (3)
2S i<j 4 4 i<j S
Now compare (2) and (3). Since S ≥ ai + aj for any i < j, the claim L ≥ R results.
2nd Solution. Let S = a1 + a2 + · · · + an . For i 6= j,

X X
ai ak + aj ak + 2ai aj
4ai aj (ai − aj )2 (ai − aj )2 k6=i k6=j
= ai + aj − ≤ ai + aj − = .
ai + aj ai + aj S S
The statement is obtained by summing up these inequalities for all pairs i, j:

 
X ai aj 1 X X ai aj 1 X X X X
= ≤  ai ak + aj ak + 2ai aj 
a
i<j i
+ aj 2 i
ai + a j 8S i
j6=i j6=i k6=i k6=j
 
1  X X X X X X X X
= ai ak + aj ak + 2 ai aj 
8S i j6=i
k i6=k j6=i k j6=k i6=j
 
1  X X X X X X
= (n − 1)ai ak + (n − 1)aj ak + 2 ai aj 
8S i j6=i
k i6=k k j6=k
n XX n X
= ai aj = ai aj .
4S i 2S i<j
j6=i
6. Let A, B, C be 3 points on the plane with integral coordinates. Prove that there exists a
point P with integral coordinates distinct from A, B and C such that the interiors of the
segments P A, P B and P C do not contain points with integral coordinates.
Solution. Let A = (a1 , a2 ), B = (b1 , b2 ), C = (c1 , c2 ) where a1 , a2 , b1 , b2 , c1 , c2 are integers.
If a lattice point X = (x1 , x2 ) is between A = (a1 , a2 ) and B = (b1 , b2 ), then there exist
relatively prime positive integers u and v such that (u + v)xi = uai + vbi , i = 1, 2. That
is (u + v)(xi − bi ) = u(ai − bi ), i = 1, 2. This implies that (u + v) divides ai − bi , for
i = 1, 2. That is a1 − b1 and a2 − b2 have a common factor. Therefore, it suffices to find
a lattice point P = (x1 , x2 ) such that gcd(x1 − a1 , x2 − a2 ) = 1, gcd(x1 − b1 , x2 − b2 ) = 1
and gcd(x1 − c1 , x2 − c2 ) = 1.
Let p = 2 or 3. There are at least 4 distinct values (m1 , m2 ) modulo p. Thus there exists
(r1 , r2 ) such that (r1 , r2 ) 6≡ (a1 , a2 ), (b1 , b2 ), (c1 , c2 ) (mod 2). Similarly, there exists (s1 , s2 )
such that (s1 , s2 ) 6≡ (a1 , a2 ), (b1 , b2 ), (c1 , c2 ) (mod 3). By the Chinese Remainder Theorem,
there exists (p1 , p2 ) such that p1 ≡ r1 (mod 2), p1 ≡ s1 (mod 3); and p2 ≡ r2 (mod 2),
p2 ≡ s2 (mod 3). We can also require p1 6= a1 , b1 , c1 and p2 6= a2 , b2 , c2 . This ensures that
(p1 , p2 ) 6≡ (a1 , a2 ), (b1 , b2 ), (c1 , c2 ) (mod 2) and (mod 3).
Suppose p is a prime larger than 3 such that p divides p2 − a2 , p2 − b2 , or p2 − c2 . There
are only a finite numbers of such primes p. Let the set of all such primes be Y . For each
p ∈ Y , pick an integer zp 6≡ a1 , b1 , c1 mod p. Since p > 3, this is possible. Apply the
Chinese Remainder Theorem to find an x1 such that x1 ≡ p1 (mod 2), x1 ≡ p1 (mod 3)
and x1 ≡ zp (mod p), p ∈ Y . Take x2 = p2 . Then P = (x1 , x2 ) is the desired point.
Training Problems
18 January 2003
1. Let M be a point on the segment AB. Squares AM CD and M BEF are erected on
the same side of AB with F lying on M C. The circumcircles of AM CD and M BEF
meet at a second point N . Prove that N is the intersection of the lines AF and BC.
2. Let DM be the diameter of the incircle of a triangle ABC where D is the point at
which the incircle touches the side AC. The extension of BM meets AC at K. Prove
that AK = CD.
3. Tangents P A and P B are drawn from a point P outside a circle Γ. A line through P
intersects AB at S and Γ at Q and R. Prove that P S is the harmonic mean of P R
and P Q.
4. (IMO 1981)Three circles of equal radius have a common point O and lie inside a given
triangle. Each circle touches a pair of sides of the triangle. Prove that the incenter
and the circumcenter of the triangle are collinear with the point O.
1. Let M be a point on the segment AB. Squares AM CD and M BEF are erected on
the same side of AB with F lying on M C. The circumcircles of AM CD and M BEF
meet at a second point N . Prove that N is the intersection of the lines AF and BC.
Solution
...................................
............ ........
....... .......
......
D.....................................................................................................................C
...... ........
... .. .........
... .. ... .. ...
... .... ... .. ....
... .
. ... .. ...
.... .... ... .. ...
. ...
... ... ...
... .. ..
... .. .....
...
...
....
..
... N
... ................................
... .
. F .
................. .... ........ E
. ................................................
... .... ... ......... . . .. ...
...
.... ... ... ... .... ... .. .. .
...
. ..
. ... ... .
. ... .. .. ... .....
...
... .... . ... ... .
.
. .
. ..
. .. .. .
... .. ... ... . . ... ... ..
. . ..
. .. .... ....
... ... . ... ... ... ... .. ... ...
... .. .. ... ..... ... .. ...
. .... ...
....... .... . . .
........................................................................................................................................................
...... . .
A ...... ...... ............................... B
....... ...... M
.......... .......
..........................................
Let AF intersect BC at N 0 . We wish to show that N = N 0 . As 4AM F is congruent

to 4CM B, we have ∠AN 0 C = 90◦ so that N 0 lies on the circle with AC as diameter.
That is N 0 lies on the circumcircle of AM CD. Similarly, N 0 lies on the circumcircle
of M BEF . Thus N = N 0 .
2. Let DM be the diameter of the incircle of a triangle ABC where D is the point at
which the incircle touches the side AC. The extension of BM meets AC at K. Prove
that AK = CD.
Solution
B
..
.......
...... ..
... .. ...
.
.... .... .....
.. . ...
... .. ...
... ... ...
... .... ...
. ... . ...
....
...
. ...
... . ...
. .
.. ............................
.
. M .
.
..
............. ....
. ........ .....
. . . ........ ...
..........
.
. .....
. ..... ...
..... ... ... .....
.....
.... ... ... ....
.... .. .. ...
.. ....... ... ....
. ...
.. .. ... .
.. ... .....
.
... ... ... ... ......
... ... . ..
. ....
.... ... .
.
. .... ... ....
..... ...
. .
... .
.
. .. ....
.
.
.. ...
...
.
... . ....
... ...
.. ... ...
... ...
..... ....
... ... ...
... ... ....
....
. ..... .
..
. .. .
. . . ........ ...
...
.. ....... .
. ....... .
. . . ... . ..
A............................................................................................................................................................................................
. . . . . ...
C
.. ........ .......... ...
..
.
.
.. ........... ..... .
.
K D . .... ........
. ..
...
....... ...
... ....... ...... ...
............... ...... ...
.......... .....
..
.........
. ..... .....
.
.. ....... ...
.....
.
. .
... .... ....
... ... ..
... ... ..
....
. ......
......
..
...
. .....
......
. .....
....... ...
. . ..... ...
.. ...
. ...
.
.. ... ...
.. ....
... ...
. ......
... ...
. ......
.... ... ......
.. ... ...
.... ... ...
... . ..
... 0 ... ...
... O ... ....
... ... ....
...
... ... ...
...
... ... .
... ...
... ...
... ...
... .. .
...
... ...
... ..
... ...
... .....
.... ... .
..... .....
..... .....
..... .....
.....
...... .........
.
...... ......
....... ......
....... .......
......... ........
...........
..................... . . . . ..... . . . ............
.......................
2
Consider a homothety centered at B carrying the incircle to the excircle. The diameter
M D of the incircle is mapped to the diameter M 0 D0 of the excircle. Since M D is
perpendicular to AC, M 0 D0 is also perpendicular to AC. Therefore M 0 must be the
point K. That is the excircle touches AC at K. Therefore, AK = (a + b − c)/2 = CD.
3. Tangents P A and P B are drawn from a point P outside a circle Γ. A line through P
intersects AB at S and Γ at Q and R. Prove that P S is the harmonic mean of P R
and P Q.
Solution Since 4AP Q is similar to 4RP A, we have P Q/P A = AQ/AR. Also

4BP Q is similar to 4RP B, we have P R/P B = RB/QB. Dividing the second
equation by the first equation and using the fact that P A = P B, we obtain P R/P Q =
(RB/AQ) · (AR/QB) = (RS/AS) · (AS/QS) = SR/SQ. This shows that the ratio
that S divides QR internally is the same as the ratio that P divides QR externally.
This determines the position of S on the segment QR.
A
.............................................
.......... ... .......................
Q .......
...... ... ...... ..........
................
. ...
..... ...........
..... ..........
.... .............................
..
.
.
.. ................ .... ....
...
..........
..........
. ......................... ..........
.... . ................ .....
R .................
..........
..........
..
. S ..
. .
.
... .................. ..........
.... .... ... ................
................
..........
.
... ... ... ................ ...................
... ... ... ................ .........
... ................ ..........
.... ... ... ........................
.............
... ... ..... P
... ... ... ..........
... ... .. ..........
... .
. ...
.
.
.............. .
..
.
... ..
. ... ........
.......
.
...
.... ...
. ..........
... ..........
... ... ... ..........
... ... .. ..........
... .
. .....
......
. ............
.....
.... .....
.. ......
..... ..........
...... ... ..... ..........
...... ...............
.......
.......... ... .......................
.................................. ... .
..
B
Thus
PR PQ
SR = QR · , SQ = QR · .
PR + PQ PR + PQ
Also
QR · P R (P Q − P R) · P R 2P R · P Q
P S = P R + RS = P R + = PR + = .
PR + PQ PR + PQ PR + PQ
That is P S is the harmonic mean of P R and P Q.
(Second Solution by Colin Tan) Let M be the midpoint of QR. Then to prove P S =
2P R·P Q/(P R+P Q) is equivalent to prove that P S ·P M = P R·P Q. Or equivalently,
P S·P M = P B 2 , since P R·P Q = P B 2 . Therefore, we have to show that P B is tangent
to the circumcircle of 4SM B. Let O be the centre of Γ. Then O, M, P, B are concyclic
and OP is perpendicular to AB. Hence, ∠P BA = ∠P OB = ∠SM B. Therefore, P B
is tangent to the circumcircle of 4SM B.
3
...........
....................................
.
A
........ ... ........................
Q ...........
...... ... ................
..... ...........
.....
.... .............................. M ... ..... ..........
.
.. .
. .... ..........
....
...
. ................
.
.
....• .
.
.... ................................ ....
..
...
.................. R
..........
..........
..........
..........
..
... .. ..
. .. ...
.
S . .. .. . . .
... ...................... ..........
..........
. ... .. ................. ......
.... .. .. .... .
... ................
................ ..................
... .. .. ... ... ................ ..........
... .. .. .. . .............. ........
.... ..... ... ....... ..... ... ... ... ... ... ... ... ......... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...............................................
... .. .. ...
.. . .. ..
..
.......
..........
P
... O .. . .. . . .. ..........
... .. ..........
... .. .. ..
. . .
. .... ...
. ..............
.
..
... . .. ..
...
. ..........
... ..
.. .. ....
.
... ..........
... .........
... .. .. .. .. ..........
... .. .. ... ..... ....
...............
..... .. ..... .. ......
..... ..... ..........
...... ....... ..... ...........
......
....... ....... ..............................
.......... ................
....................................
B
(Third Solution) Applying Stewart’s Theorem to 4AP B, we have
(AS + SB) · P S 2 + (AS + SB) · AS · SB = AP 2 · SB + BP 2 · AS.
Since AP = BP , we may cancel the common factor (AS + SB), thus obtaining
P S 2 + AS · SB − AP 2 = 0.
Since AS · SB = QS · SR = (P Q − P S)(P S − P R) = (P Q + P R) · P S − P S 2 − P Q · P R
and P A2 = P Q · P S, we have (P Q + P R) · P S = 2P Q · P R. Thus, P S is the harmonic
mean of P R and P Q.
4. (IMO 1981)Three circles of equal radius have a common point O and lie inside a given
triangle. Each circle touches a pair of sides of the triangle. Prove that the incenter
and the circumcenter of the triangle are collinear with the point O.
Solution
.................
........................................................... A
..............
............ .. ...
.......... ........ .................
... .......... ... . ... .......
.. ... . .... .
..... ... .... .......
......
......... .... .. ... ......
......... .. .. .. .... ......
......
........ .. .. .. ...
... .....
....... ...
. .
. . .....
.....
....... ..
.. .
.
... .....
.... .. . ... .....
... .. . .
... .....
...
... ... . ....
... ... .. ... ...
.. . . . ....
...
. .. .. ..
...
...
...
...
. .... .................................. . ...
... . .. ............ . ...........
.
... ...
.
.......... ...... .. .
...
. .. .
. ........ .. . ...
...
... .........
. .
.
. .
. .
.... ... ...
... .... . . . . ...
.. ... . ...... ...
. . .
. ..
... ...
..
. . ...
.
. . ... ...
... . . ..... .
. . ...
... ......
. .
. ...
. ...
.... ... ... . 0 ...... ...
... ..
.
.. ... .
.
. A .....
. . ...
... .... .... . .........
. .
.. ... ...
... ..
. ... .... .. ... . ....
.
.
...
... ... .. . ... . ... ...
... .... ... ..
.. .. .... . . ..
. .... ...
... .. . ... .. .
. ...
... .... . .... .. ... . ... ... ...
.. ... . . ... ............. ... ...
C
·O
... .. .. .......... . . . .. .. ... .
........... .. .. .
. . .. . .. .. .
........... . ...
.... .. .. ............... .
. . ... . . . .. . .
................ .
.. .. . . .
... ..
. .......... ..... ........ . ... .... ... . .
. .. ...... ...
. ..
... . .. . . .
...... ..... .... .. . ... ...... . . ..
..... .. . ..
.......
...
...
...
...
... ....
......
..
...
........
...
...
...
...
.. .
··I
.. . . .
.
. . ..
... .....
.. .........
...
...... ...........
.
.
.. ............................... ......................... .
...
...
...
... ..
.
.
........
.
.....
......
...
... ..
...
.
...
....
... ........ . ... ............

. ... .... .... ... ...
... .
.. . .
.. ... .. . .
... ..... ... . ..... .
... ... .. ... .... ... . ... .. ...... ...
... ... ... .................................................................................................... .. ...
... ... .... .... ... . ... .... ...
... ..... ... ... ..... 0 B .
... ....
C 0 ....
.. .
.. .
. ...
... .. ... .... .... .
... ... ... ..... ... ... .... ... ... ...
... ... ........ ... .. .... .. .... ...
...
... ... . ... .... .... .... ... ..
... .
. ... .. ........... ... .... .... ..
... ...
....
... .. .... .... ..... ..... ..
...
.
... ... ....... ..... .... ..... ..... ....... .....
... ... .. ..... ..... .......... ..... .... .. .....
... ............. ...... ...... ....... ...... .... ... ...
.......... .......
........ .... . ... ....... .......... .. .. . .. .
. ....... ... .
. ..
. . .
........................................................................................................................................................................................................................................................................... .
..... ...
B ..... ..... C
..... .....
.....
..... ........
.
..... ....
..... .....
...... ......
...... ......
......
....... ...........
....... .......
........ .......
......... ........
.......... .........
.............
....................... ......... . . . .............
..................................... .
4
Let A0 , B 0 , C 0 be the centres of the circles inside 4ABC. As AA0 , BB 0 , CC 0 are angle
bisectors, they meet at the incenter I of triangle ABC. I is also the incenter of the
triangle A0 B 0 C 0 . The circles are of the same radii. Thus A0 and B 0 are of equal distance
from AB so that AB is parallel to A0 B 0 . Similarly, BC is parallel to B 0 C 0 and A0 C 0 is
parallel to AC. That is 4ABC is similar to 4A0 B 0 C 0 . Consider a homothety centred
at I sending A0 to A, B 0 to B and C 0 to C. Thus the circumcentre O of 4A0 B 0 C 0
is mapped to the circumcentre C of 4ABC under this homothety. Therefore, I, C, O
are collinear.
5
Training Problems
25 January 2003
1. In a quadrilateral ACGE, H is the intersection of AG and CE, the lines AE and CG

meet at I and the lines AC and EG meet at D. Let B be the intersection of the line
IH and AC. Prove that AB/BC = AD/DC, or equivalently, DB is the harmonic
mean of DA and DC.
2. The extensions of the chords QR and Q0 R0 of a circle Γ intersect at a point P outside

Γ. Tangents P A and P A0 are drawn from P to Γ. Prove that A, X, A0 are collinear
where X is the intersection of QR0 and Q0 R.
3. In a quadrilateral ABCD, E is a point on CD, BE intersects AC at F and the

extension of DF meets BC at G. Suppose that AC bisects ∠BAD. Prove that
∠GAC = ∠EAC.
4. (Crux 2333) Points D and E are on the sides AC and AB of 4ABC. Suppose F and G
are points of BC and ED, respectively, such that BF : F C = EG : GD = BE : CD.
Prove that GF is parallel to the angle bisector of ∠BAC.
5. (Balkan Math Olympiad 2002) Let O be the center of the circle through the points
A, B, C and let D be the midpoint of AB. Let E be the centroid of triangle ACD.
Prove that the line CD is perpendicular to the line OE if and only if AB = AC.
1. In a quadrilateral ACGE, H is the intersection of AG and CE, the lines AE and CG
meet at I and the lines AC and EG meet at D. Let B be the intersection of the line
IH and AC. Prove that AB/BC = AD/DC, or equivalently, DB is the harmonic
mean of DA and DC.
Solution Apply Menelaus’ Theorem to 4ACI, 4AEC and 4CEI with transversals
EGD, IHB and AHG respectively. We have
CD AE IG AB CH EI CG IA EH
= 1, = 1, = 1.
DA EI GC BC HE IA GI AE HC
The result is obtained by multiplying these three equations together.
I .
.......
... .......
... .....
.. .........
.
.. ... ...
.. ... ...
... ... ...
... ... ...
... ... ...
... ... ....
... ... ...
. ... ... ...
... ... ...
.. ... ...
. ... ...
... ... ....
...
E .
.. .
... ....................
...
...
...
...
... .............. ... ...
.. .............. . ...
. .
. ..
.
...... ........ .
... ...
... .. ...
...... ....... ... F ...
..
. ... . . .. . ...
...... ..........
... ...... ........ .
...
. ...... ............. .....
.. ...... .. ....... ..
. . . ...... .. . G ........
..
H . . . .. ......................
.. .
. .....
.
... ................. ... ........
.. .. .. .
... .. . .............. ................ . .......
.
..............
... .. ... ....... .... .......
... .............. ... ...... ... .......
.......
..
. ............. ... ........... .......
...................... . . .
.........................................................................................................................................................................................
.
. . ...
.
A B C D
Alternatively, by Ceva’s Theorem applied to 4ACI, we have

IE AB CG
= 1.
EA BC GI
Next by Menelaus’ Theorem applied to 4ACI with transversal EGD, we have
AD CG IE
= 1.
DC GI EA
Thus, AB/BC = AD/DC.
2. The extensions of the chords QR and Q0 R0 of a circle Γ intersect at a point P outside

Γ. Tangents P A and P A0 are drawn from P to Γ. Prove that A, X, A0 are collinear
where X is the intersection of QR0 and Q0 R.
Solution Let AA0 intersect P Q at S and P Q0 at S 0 . We wish to prove that S, X, S 0
are collinear. We know that P S is the harmonic mean of P Q and P R or equivalently
P Q : P R = SQ : SR. (See problem 3 in the training problems on 18 January 2003.)
Thus
2P Q · P R 2P Q 2P Q 2P Q · SR
PS = = P Q
= SQ
= .
PR + PQ 1 + PR 1 + SR QR
PS 2P Q PS P S SR 2P Q P R 2P R
Therefore, = . Also = = = .
SR QR SQ SR SQ QR P Q QR
2
Note that all these ratios are equivalent to each other and they are just different ways
to express the location of S. Similarly,
P S0 2P Q0 P S0 2P R0
= and = .
S 0 R0 Q0 R0 S 0 Q0 Q0 R0
............
................................... A
........ ...................
Q..................... ...
...
................
..... ..........
..... .........
.................................. ... ..... ..........
. ... ........ ................ . ... ..........
. ..... ... ... ................. . ... ..........
.... .. . . ......... . ..........
..
.....
..... . .............................
. R ..........
..
.
. .....
......
S .
.
.
. .......................
.. ...
.... ...
................
..........
..........
.... ..... .. ..... ... ................ .......
..... .
.. ...... . ... ................ ...................
... .....
..... .
. .... ... .
... ................
...................................
... ..... .. ..... .
. .. . ... .........................
... ..... .. .....
..... ....... ... ................
... ........ ..
. ................
.......................
P
...
... X ....
. ..
......... ......... ..
. .........
....................................
. .
...............................
... ..... ... .......... ... ..............
... ..... ..... .............. ...................
.
... ..... .... ..... ....
..... .. ...........................
..... ... ...
...
... ......
. . ................... ...............
Γ ...
.... .
. .
.......
S 0 .. ........................ 0
.
. ......
.. .. .
......
. ................
..... .....
.
..... ......... ...........................
.................
....
R
..... ...................
... ..
...... .................. ..... ..........
.........
. ... ............................
...
0 ................. .
.
.. ....
...
Q ................................
A0
Apply Menelaus’ Theorem to 4P RQ0 and the transversal QXR0 . We get
XR P R0 QR P R QR
0
= 0 0
= ,
XQ PQ Q R P Q0 Q0 R0
because 4P RQ0 is similar to 4P R0 Q. This characterizes the position of X along RQ0 .

Now consider 4P RQ0 and the transversal SXS 0 . Combining the above results, we
have
P S RX Q0 S 0 P S P R QR Q0 S 0 2P Q P R QR Q0 R0 PQ PR
= = = = 1.
SR XQ0 S 0 P SR P Q0 Q0 R0 S 0 P QR P Q0 Q0 R0 2P R0 P Q0 P R0
By Menelaus’ Theorem, S, X, S 0 and hence A, X, A0 are collinear.
3. In a quadrilateral ABCD, E is a point on CD, BE intersects AC at F and the

extension of DF meets BC at G. Suppose that AC bisects ∠BAD. Prove that
∠GAC = ∠EAC.
Solution Let A be the origin of a rectan- y

..
gular coordinates system with AC as the x- ...
.......
..
axis. Let C = (c, 0), F (f, 0), D = (xD , kxD ), ..
..
D ...
.. .......... .......
..........
.. .......... .....
.......... .......
B = (xB , −kxB ). Then the equation of the ..
..
..........
..........
... ..
..... ......
... ...
..
.. ......
. .........
E . ..
................
line DF is .. .. ...
..................
.. ..........................................................................
........... ................................ .................. ... . .
... ......
...... ...
C . .
............................................................ ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ........... ......... ... ... ............... x
A .......... ..............................
.......... ............................ .... ..
...... ....
f − xD ..........
..........
..........
............................ F
............................ .. ..... .
.
...........
x−f + y=0 (1) ..........
..........
.......... G .. ..
... ...
kxD ..........
..........
..........
.......... .
... ...
.....
..
.......... ......
........
The equation of the line BC is

c − xB
x−c+ y=0 (2)
−kxB
3
By taking c × (1) − f × (2), we have
1 1 1

(c − f )x + cd + − (c + f ) y = 0 (3)
k xD xB
This is the equation of a line passing through A (as there is no constant term) and
the intersection of DF and BC. Hence, it is the equation of the line AG. Similarly,
the equation of the line AE is given by
1 1 1

(c − f )x − cd + − (c + f ) y = 0 (4)
k xD xB
From (3) and (4), we see that the slopes of the lines AG and AE are negative of each
other. Therefore, ∠GAC = ∠EAC.
(Second Solution by Colin Tan and Meng Dazhe)

Pick a point G0 on BC such that ∠G0 AC = ∠BAC (let this angle is a). Join BD to
intersect AC at H. Let ∠BAG0 = ∠EAD be b.
Consider the Cevians CH, BE, DG0 of 4BCD. Now EG0 /G0 C = [ABG0 ]/[AG0 C] =
(AB sin a)/(AC sin b). Similarly, we find that CE/ED = (AC sin b)/(AD sin a) and
DH/HB = (AD sin(a+b))/(AB sin(a+b)). Thus, (BG0 /G0 C)(CE/ED)(DH/HB) =
1.
So by Ceva’s theorem, we conclude that BE, CH and DG0 are concurrent (at F ).
Hence, G0 , F, D are collinear implying that G0 = G. Therefore ∠GAC = ∠EAC as
required.
4. (Crux 2333) Points D and E are on the sides AC and AB of 4ABC. Suppose F and G
are points of BC and ED, respectively, such that BF : F C = EG : GD = BE : CD.
Prove that GF is parallel to the angle bisector of ∠BAC.
C
.....
... ..
... ...
Solution Let A be the origin of a rectangular ...... ....
. ...
... ...
... ...
coordinates system. For each of the points in the .....
... ...
...
.
... ...
question, we use the small case letter in bold face . ... ....
....... ....
F
D ...
. ..
....
.
.... ....
...... ...
...
.. .... ......
.
to denote the position vector of that point. First .....
..
... .................
....
.... ...
...
...
.... G ... ...
we have e = pb and d = qc for some p, q in (0, 1). .
......................................................................................
.
A E B
tc + b td + e tqc + pb
Let t = BF/F C. Then f = and g = = . Since BE = tCD, so
t+1 t+1 t+1
(1 − p)|b| = t(1 − q)|c|. Thus,
t(1 − q) 1−p (1 − p)|b| c b

f −g = c+ b= + .
t+1 t+1 t+1 |c| |b|
c b
This is parallel to + , which is in the direction of the angle bisector of ∠BAC.
|c| |b|
4
5. (Balkan Math Olympiad 2002) Let O be the center of the circle through the points
A, B, C and let D be the midpoint of AB. Let E be the centroid of triangle ACD.
Prove that the line CD is perpendicular to the line OE if and only if AB = AC.
A
............
............... ........ ......................
........ ... .. .......
...... .. ... ......
......
Solution Set the origin at O. As in the last .
............
..
.
.
... ....
...
...
.....
.....
....
...
.... ... . ...
question, for each of the points in the question, ..
.
...
.
..
.
..
...
.
...
...
...
...
...
..
. ..
. . ...
.. ...
·O·
we use the small case letter in bold face to denote .... . E ... ...
... D ...... . . ...
.... ... ...... ........ ... .
.. ........ ... ...
the position vector of that point. Then d = 12 (a+ ...
...
... ..
.
.
..
. .....
.....
..
.....
....
...
.
... ..
.
...
.
... ... ...
b) and e = 31 (a + c + d) = 16 (3a + b + 2c), ... ..... ..
... ... .....
..... ..... ...
... ... ..... .. ..
...
... ... ..... ... ..
d − c = 12 (a + b − 2c). ..... ... ..... .. .
..... .. ........ ........
....... ... ..
...........................................................................................
....... ..
.. ...
B .........
...........................................
. C
Hence CD is perpendicular to OE if and only if (a + b − 2c) · (3a + b + 2c) = 0. Since

a · a = b · b = c · c, this is equivalent to a · (b − c) = 0, which is the same as OA is
perpendicular to BC, or AB = AC.
5
Training Problems
8 February 2003
1. Determine whether there exist an integer polynomial f (x) together with integers a, b
and c satisfying the following conditions.
(i) ac 6= bc.
(ii) f (a) = a, f (b) = b, c2 + f (c)2 + f (0)2 = 2cf (0).
2. Let p(x) = x4 + ax3 + bx2 + cx + d, where a, b, c, d are real constants. Suppose

p(1) = 827, p(2) = 1654 and p(3) = 2481. Find the value of (p(9) + p(−5))/4.
3. How many integer polynomials of the form x3 + ax2 + bx + c = 0 having a, b, c as roots

are there?
1. Determine whether there exist an integer polynomial f (x) together with integers a, b
and c satisfying the following conditions.
(i) ac 6= bc.
(ii) f (a) = a, f (b) = b, c2 + f (c)2 + f (0)2 = 2cf (0).
Solution The second condition in (ii) implies that (c − f (0))2 + f (c)2 = 0. That
is f (c) = 0 and f (0) = c. As f is an integer polynomial and f (a) = a, we have
(a − c)|(f (a) − f (c)). That is (a − c)|a. Also (a − 0)|(f (a) − f (0)) so that a|(a − c).
By (i), c 6= 0. Thus, a − c = −a giving c = 2a. Similarly, c = 2b. But this gives a = b,
contradicting (i). Therefore, no such integer polynomial exists.
2. Let p(x) = x4 + ax3 + bx2 + cx + d, where a, b, c, d are real constants. Suppose

p(1) = 827, p(2) = 1654 and p(3) = 2481. Find the value of (p(9) + p(−5))/4.
Solution Let q(x) = p(x) − 827x. Then q(x) is a polynomial of degree 4. As q(1) =
q(2) = q(3) = 0, we have q(x) = (x − 1)(x − 2)(x − 3)(x − r), for some r. Therefore,
1 1 1
4 (p(9) + p(−5)) = 4 (q(9) + q(−5)) + 827 = 4 ((8)(7)(6)(9 − r) + (6)(7)(8)(5 + r)) =
1176 + 827 = 2003.
3. How many integer polynomials of the form x3 + ax2 + bx + c = 0 having a, b, c as roots

are there?
Solution From the relation between roots and coefficients of a polynomial equation,
we have
a+b+c = −a (1)
ab + bc + ac =b (2)
abc = −c (3)
From (3), c = 0 or ab = −1.

Case 1. c = 0. Substituting this into (1), we obtain b = −2a. Using (2), ab = b. Thus,
a = 1, b = −2 or a = b = 0.
Case 2. c 6= 0 and ab = −1. As a and b are integers, we must have a = 1, b = −1 or
a = −1, b = 1. When a = 1, b = −1, we get c = −1 using (1). These values of a, b, c
also satisfy (2). When a = −1, b = 1, we get c = 1. But then these values of a, b, c do
not satisfy (2).
Therefore, there are altogether 3 such polynomials.
2
Training Problems
15 February 2003
1. Let n be an odd integer which is not a multiple of 5. Prove that there exists a strictly
positive integer k such that n divides a string of k 1’s, i.e.
n| 11...11
| {z } .
k1’s
2. Determine all natural numbers (k, m, n) such that
n! = mk .
3. Show that for all integers A, B, there exists an integer C such that the following sets
M1 = {x2 + Ax + B : x ∈ Z} and M2 = {2x2 + 2x + C : x ∈ Z} are disjoint.
4. Let m be a strictly positive integer. Show that there exists infinitely many pairs of integers
(x, y) such that
(a) x, y are relatively prime
(b) y divides x2 + m
(c) x divides y 2 + m
(d) x + y ≥ m + 1
5. Let m and k be positive integers such that gcd(m, k) = a.

(a) Suppose that a = 1. Show that there exists integers a1 , a2 , ..., am and b1 , b2 , ..., bk such
that each of the products ai bj (i = 1, 2, ..., m, j = 1, 2, ..., k) gives a different remainder
modulo mk.
(b) Suppose that a > 1. Show that for all integers a1 , a2 , ..., am and b1 , b2 , ..., bk there exists
two products ai bj and as bt ((i, j) 6= (s, t)) such that they have the same remainder
modulo mk.
6. Let n be a non negative integer. Suppose that there exists rational numbers p, q, r such
that
n = p2 + q 2 + r 2 .
Prove that there exists integers a, b, c such that
n = a2 + b2 + c2 .
Solutions
1. From the given conditions gcd(n, 10) = 1. But gcd(9, 10) = 1 and hence gcd(9n, 10) = 1.
Thus by Euler’s Theorem,
10φ(9n) ≡ 1 (mod 9n),
which implies the desired result.
2. Using Bertrand’s Postulate, there exists a prime p satisfying n2 n, hence p only has a single power in n!, i.e. k = 1. Hence (m, n, k) =
(n!, n, 1) is a solution triplet. If n = 2, we have 2 = mk , hence we must have m = 2, k = 1.
If n = 1, we must have 1 = mk , or m = 1, k ∈ ℵ thus (m, n, k) = (1, 1, k) is another solution
triplet. Thus the only solutions to the equation are
(m, n, k) = (1, 1, k), (n!, n, 1), n, k ∈ ℵ.
3. If A is odd, x2 + Ax + B ≡ x(x + A) + B ≡ B (mod 2), but 2x2 + 2x + C ≡ C (mod 2).

So we may choose C = B + 1.
2 2 2
If A is even, x2 + Ax + B = (x + A2 )2 + B − A4 ≡ B − A4 or B − A4 + 1 (mod 4), but
2
2x2 + 2x + 1 ≡ C (mod 4), so we may choose C = B − A4 + 2 in this case.
4. Note that (1, 1) satisfies the given conditions. Now if (x, y) is a solution with y ≥ x, consider
(x1 , y) where
y 2 + m = xx1
All common divisors of x1 and y must by the above a divisor of m, and since y|x2 +xx1 −y 2 ,
we must have y|x(x + x1 ), and since gcd(x, y) = 1, we must have y|(x + x1 ), and hence the
common divisor of y and x1 must divide x too, but gcd(x, y) = 1, we have gcd(x1 , y) = 1.
It is clear that x1 |y 2 + m, and
x2 (x21 + m) = (y 2 + m)2 + x2 m = y 4 + 2my 2 + m(x2 + m),
but y|(x2 + m) implies that y|x2 (x21 + m), but gcd(x, y) = 1 implies that y|(x21 + m). Now
x1 > y ≤ x. Repeat the same argument to generate y1 , but instead consider
x21 + m = yy1 .
Then (x1 , y1 ) is also a solution, with x1 + y1 > x + y. Continue this process to generate
(x2 , y2 ),... and since m is fixed, xn + yn ≥ m + 1 for some n, thus (xn , yn ), (xn+1 , yn+1 ), ...
is a set of infinitely many solution pairs which satisfies all given conditions.
5. (a) Consider ai = ki + 1, bj = mj + 1. Suppose that two of the residues are the same.
Then mk divides ai bj − as bt = (ki + 1)(mj + 1) − (ks + 1)(mt + 1) = km(ij − st) +
m(j − t) + k(i − s), and thus m|k(i − s) but gcd(m, k) = 1, hence m|(i − s), and since
|i − s| < m, we must have i = s and similarly j = t and we are done.
(b) Suppose all the residues are distinct. Then 0 is one the residues. WLOG, suppose
mk|a1 b1 . Hence there exists a0 , b0 such that a0 |a1 , b0 |b1 and mk = a0 b0 . Suppose now
that for i 6= s, a0 |(ai − as ). Then we have mk = a0 b0 |(ai b1 − as b1 ), which is a contra-
diction. Hence all the ai ’s cannot have the same residue modulo a0 , similarly, all the
bj ’s cannot have the same residue modulo b0 . Thus we must have a0 ≥ m, b0 ≥ k thus
a0 = m, b0 = k.
Now let p be a prime divisor of m and k. p > 1 since gcd(m, k) > 1. Since all the ai ’s
form a distinct set of residues modulo m, there are m − m p between them which are
k
not divisible by p. Similarly, there are k − p bj ’s which are not divisible by p. On the
other hand all the ai bj ’s form a set of reduced residues modulo mk by our assumption,
and hence between them, there are mk − mk p which are not divisible by p. But
m k mk

m− k− = mk −
p p p
if and only if m = 0, k = 0 or p = 1, which is a contradiction.
6. If n = 0, the result is clear. So suppose n > 0. Suppose the set of points (x1 , x2 , x3 ) which
lies on the sphere
n = x2 + y 2 + z 2
are all rational points. We will obtain a contradiction. Now there exists an integer point
u = (u1 , u2 , ..., un ) such that ad = u, where d ≥ 2. Suppose that a and u are chosen such
that d is minimal. Then let x0 , y 0 z 0 be the integers closest to x, y, z, where a = (x, y, z).
Then |x − x0 | ≤ 21 , |y − y 0 | ≤ 12 and |z − z 0 | ≤ 12 , hence ||a − a0 || < 1, where a0 = (x0 , y 0 , z 0 ).
Now consider the line connecting a and a0 . This will intersect the sphere x2 + y 2 + z 2 = n
at two points, one at a and the other which we call b. The equation of the line is given by
a0 + λ(a − a0 ). Now b lies on the sphere so
n = ||b||2 = ||a0 ||2 + 2λ < a0 , a − a0 > +λ2 ||a − a0 ||2 .
One of the solutions to this equation is given by λ = 1, which correspond to the point a.
0 ||2 −n
The other thus is given by λ = ||a
||a−a0 ||2
. Now
2 d1
||a − a0 ||2 = ||a0 ||2 + ||a||2 − 2 < a0 , a >= ||a0 ||2 + n − < a0 , u >= ,
d d
d(||a0 ||2 −n)
where d1 ∈ ℵ and since ||a − a0 ||2 < 1 we have d1 < d. Hence λ = d1 and we have
b = a0 + λ(a − a0 )
||a0 ||2 − n
= a0 + (u − da0 )
d1
v
=
d1
where v is an integer point. Now b = vd1 with d1 < d contradicts our assumption that d is
minimal.
Note that a generalisation is not possible using this method since ||a − a0 ||2 < 1 will NOT
be satisfied for higher dimension spaces. For a one dimensional space, i.e. the real line,
this result is obvious. For a two dimensional space, i.e. the plane, this argument works.
Training Problems
08 March 2003
1. Let p be a prime number congruent to 1 modulo 4. Define the set
S = {(a, b, c) ∈ N2 × (Z − {0}), 4ab + c2 = p}.
Now consider
S1 = {(a, b, c) ∈ S, a > b + c}.
Assume that |S1 | is odd. Show that p can be expressed as a sum of two squares.
2. Prove that the circle with equation x2 + y 2 = 1 contains an infinite number of points with
rational coordinates such that the distance between each pair of the points is irrational.
3. Let f : R → R be a function such that

(a) f (2x) = 2 [f (x)]2 − 1, for all x ∈ R;
(b) There exists a real number x0 such that f (2n x0 ) ≤ 0 for all natural numbers n.
Determine the value of f (x0 ).

Solutions
1. S is non empty since (k, 1, 1) ∈ S and |S| is also finite since if (a, b, c) ∈ S, a, b, |c| are all
bounded by p. Now consider
S1 = {(a, b, c) ∈ S, a > b + c} S2 = {(a, b, c) ∈ S, a < b + c}.
Then S1 and S2 are clearly disjoint, and if a = b + c, we have p = 4b(b + c) + c2 = (c + 2b)2 ,
which is a contradiction. Hence S is the disjoint union of S1 and S2 . Consider now
(a, b, c) → (b, a, −c). This is a bijection from S1 to S2 . Hence |S1 | = |S2 |.
Since |S1 | is odd, we have |S| ≡ 2 (mod 4). Now consider
S3 = {(a, b, c) ∈ S, a 6= b}.
Then for all (a, b, c) ∈ S, with a 6= b, we can associate 4 distinct triplets
(a, b, c), (a, b, −c), (b, a, c), (b, a, −c)
in S. Hence |S3 | is divisible by 4. This shows that S3 ⊂ S (strict inclusion). Hence there
exists a triplet (a, a, c) ∈ S. Thus p = 4a2 + c2 = (2a)2 + c2 , as required. Note that the
assumption that |S1 | is odd can be proven. (As Andre pointed out, consider the map f
which maps (a, b, c) → (a − b − c, b, −2b − c) from S1 to itself. f 2 = Id, and we can check
that a ’fixed point’ is ( p−1
4 , 1, −1). A pairing up will give you the result.)
2. Let p be a prime and p ≡ 1 (mod 4). Then there exists integers a and b such that p = a2 +b2 ,
and by Dirichlet’s Theorem, there exists infinitely primes of the form 4k + 1, thus consider
the set of points given by !
a2 − b2 2ab
xp = ,
p p
All these points lie on the unit circle, and are irreducible fractions. If p 6= q, we have
2|ad − bc|
||xp xq || = √
pq
and since p 6= q are primes, the above fraction is irrational and we are done.
3. Note that if f (x0 ) is a solution, −f (x0 ) is also a solution. It is clear that f (x0 ) 6= 0. So
2
assume WLOG f (x0 ) = p < 0. Then we have f (2n−1 x0 ) = 2 f (2n−2 x0 ) − 1 ≥ −1,

and thus −1 ≤ f (2n−1 x0 ) < 0 for all natural numbers n. Now we obtain a better upper
bound. We claim that we have −1 ≤ f (2n−1 x0 ) ≤ − 41 . Suppose that there exists a natural
number m such that 0 > f (2m−1 x0 ) > − 41 . Then we have −1 ≤ f (2m x0 ) < − 78 , and hence
17
1 > f (2m+1 x0 ) > 32 > 0, a contradiction. Now from the given functional, we have
1 1 1

f (2x) + = 2 f (x) − f (x) + ,
2 2 2
and hence for all natural numbers n, we have

f (2n x0 ) + 1 = 2 f (2n−1 x0 ) − 1 f (2n−1 x0 ) + 1 ≥ 2 1 + 1 f (2n−1 x0 ) + 1 ,

2 2 2 2 4 2
or equivalently
f (2n−1 x0 ) + 1 ≤ 2 f (2n x0 ) + 1 .

2 3 2
Hence we have
n+1 n+1 n
f (x0 ) + 1 ≤ 2 f (2x0 ) + 1 ≤ ... ≤ 2 f (2n+1 x0 ) + 1 ≤ 2 1 2

1+ ≤ ,
2 3 2 3 2 3 2 3
and letting n → ∞, we have
f (x0 ) + 1 → 0,

2
and thus f (x0 ) = − 12 . Hence we have f (x0 ) = ± 12 .
Singapore International Mathematical Olympiad 2003
National Team Training
Geometry
(8) Let P and Q be the feet of the perpendiculars from the orthocenter on
triangle ABC onto the internal and external bisectors of ∠A. Assume that
∠A is not a right angle, show that the line through P and Q bisects BC.
Solution. Let H be the orthocenter and consider the circle C with diameter
AH. Observe that P, S, Q, D all lie on this circle. Since ∠P AQ is a right
angle, P Q is another diameter of C. Also ∠DAQ = ∠SAQ, so the arcs DQ
and SQ on C have the same length. Thus the diameter P Q must be the
perpendicular bisector of the segment DS. Now consider the circle C with
diameter BC. S and D both lie on C . Hence P Q, which is the perpendicu-
lar bisector of the chord DS of C , must pass through the center of C , which
is the midpoint of BC.
(9) Let ABCD be a convex quadrilateral whose diagonals AC and BD

intersect at O. If P and Q are the centroids of triangles AOD and BOC
respectively, and R and S are the orthocenters of triangles AOB and DOC
respectively, show that P Q ⊥ RS.
Solution. Construct P and Q so that P AOD nd Q COB are parallelo-
grams. We first claim that P Q is parallel to P Q . In fact, using vectors,
OP = OA + OD and OP = (OA + OD)/3. So OP = 3OP . Similarly,
OQ = 3OQ. Thus P Q = OQ − OP = 3(OQ − OP ) = 3P Q. This proves
the claim.
1
2
Hence it suffices to show that RS is perpendicular to P Q . Draw all four

lines through B and D parallel and perpendicular to AC. From here, I
use coordinate geometry, you may be able to find a better way. Set up a
coordinate system with D as the origin and DP as the y-axis. Set A = (a, b),
B = (c, d) and C = (a, e). We find that
ad d(b − d)
P = (0, b − ) R = (a + , d)
c c
ed (c − a)d
S = (a + , 0) Q = (c, e + ).
c c
Hence the slope of P Q is (e + d − b)/c and the slope of RS is c/(b − d − e).
This shows that P Q and RS are perpendicular.
(10) Let E be an interior point on the median AD of acute triangle ABC.

Label the foot of the perpendicular from E onto BC as F . From and interior
point M of the segment EF , drop perpendiculars onto the sides AB and AC
and let the feet of these perpendiculars be P and N respectively. Show that
the angle bisectors of ∠P M N and ∠P EN are either parallel or coincide.
Solution. Observe that AP M N are concyclic. Hence ∠P M N = 180◦ − ∠A.
Therefore, the bisector of ∠P M N makes an angle of 90◦ − ∠A/2 with the
segment P M , which impies that it makes an angle of ∠A/2 with the line
AP . Thus the bisector of ∠P M N is parallel to the bisector of ∠A. We now
show that the bisector of ∠P EN is also parallel to the bisector of ∠A. Let
the line through E parallel to BC meet the sides AB and AC at H and K
respectively. Since E lies on the median from A, HE = EK. Also, EHP M
are concyclic and EKN M are concyclic. Therefore,
∠EP H = ∠EM H = ∠EM K = ∠EN K.
Considering the quadrilateral AP EN ,
360◦ − ∠P EN + ∠A + ∠EP H + ∠EN K = 360◦ .
It follows that ∠P EN = ∠A + 2∠EP H. Thus the bisector of ∠P EN makes
and angle of ∠EP H + ∠A/2 with P E. Hence it is parallel to the bisector
of ∠A, as required.
Training problems 27 March 2003
1. Find all integer solutions in x and y of the equation x3 + 27xy + 2009 = y 3 .

Solution. Let y = x + a. Then the equation becomes
(27 − 3a)x2 + (27a − 3a2 )x − a3 + 2009 = 0.
As a quadratic equation in x, its discriminant is
(27a − 3a2 )2 − 4(27 − 3a)(−a3 + 2009) = −3(a − 14)(a − 9)(a2 + 41a + 574).
The factor a2 + 41a + 574 is always positive. Therefore the equation has integer solution
in x only when a = 9, 10, 11, 12, 13, 14. When a = 9, the equation becomes −3x2 − 30x +
1009 = 0 which has no integer solution in x. Similarly for a = 10,11,12,13, the resulting
quadratic equations do not give integer solution in x. When a = 14, the equation becomes
−15(x + 7)2 = 0. Thus x = −7, y = a + x = 14 − 7 = 7 is a solution to the given equation.
2. Let ABC be a given triangle, and M, N , and P be arbitrary points in the interiors of
the line segments BC, CA, and AB respectively. Let the lines AM, BN, and CP intersect
the circumcircle of ABC in points Q, R, and S respectively. Prove that
AM BN CP
+ + ≥ 9.
MQ NR PS
Solution. Let F be the foot of the perpendicular from A onto BC. If Q0 is the midpoint of
the arc BC and A0 is the midpoint of BC, it is clear that AM/M Q = AF/M 0 Q ≥ AF/A0 Q0 .
AM BN CP
Therefore, the minimum value of M Q + N R + P S will be obtained uniquely where Q, R, S
are the midpoints of the arcs BC, CA and AB respectively. Thus, we will henceforth
assume that Q, R, S are positioned so that AQ, BR, QS are the angle bisectors of angles
A, B, C respectively.
A
.............
..........
......
.................... ..............................
........
........................... .......
.... ......
...... ... .. ... ...... .....
.
..... .
.. ... ... ...... .....
.....
..
... .
.. .. ... ...... .....
.
.... .
.. .. ... ......
... ..... ...
... ... ... . ...
.
. .. . ... ..... ...
.. .
. .
.. ... ......
... ... . ... ...... ...
... ... .
. ... ..
..... ...
.. .
. .
. .... ...... ...
... ..
.. ..
. ... ...... ...
.. .. .
. ... .
..... ...
.... ...
.. .
.
.. ... .
..... ...
... .
. . ... ...... ...
.. .
.. .
. ... ....... ...
... .
. . .
. ... ...... ...
... .. ..
. . ...... .
. ...
...
..
.
.
. .
.
.
. ... ......
. .
....
... .. . ... ..... .
... ... . . . .... .
.... 0 ...... 0 ............ .....
.
... ...
...... .. F A . MM
..................................................................................................................................................................................
.
... ... .... ...
B ...
...
...
...
..
...
...
...
...
..
... ....
... C
... . ... .. . .
.
... .
..... ... .. ...
..... ... ... .. .....
.....
...... .. ... ... .........
...... . . .
..... ....... ..
....... ... ..
........ ... ........
........... ........
....................................................
Q
Q0
We know that the angle-bisector AQ cuts the side BC of length a in the ratio c : b so
BM = ca/(b + c) and M C = ba/(b + c). Therefore,
AM AM 2 AM 2 AM 2 (b + c)2
= = = . − − − (∗)
MQ AM · M Q BM · M C a2 bc
1
From Stewart’s Theorem,
c2 M C + b2 BM − aBM · M C − aAM 2 = 0.
It follows that
ca2 b
AM 2 = bc − .
(b + c)2
Substituting this into (*), we have
µ ¶2
AM b+c
= − 1.
MQ a
Similarly,
µ ¶2 µ ¶2
BN a+c CP a+b
= − 1, = − 1.
NR b PS c
Thus we have
µ ¶2 µ ¶2 µ ¶2
AM BN CP b+c a+c a+b
+ + ≥ + + − 3.
MQ NR PS a b c
By the convexity of f (x) = x2 , we have

µ ¶2 µ ¶2 µ ¶2 µ ¶2
b+c a+c a+b 1 b+c a+c a+b
+ + ≥ + + .
a b c 3 a b c
Also, using AM ≥ GM , and the inequality x + 1/x ≥ 2 for x > 0, we get

·µ ¶ ³c µ ¶¸2
AM BN CP 1 b a a´ c b
+ + ≥ + + + + + ≥ 9.
MQ NR PS 3 a b a c b c
Equality holds if and only if a = b = c; that is to say, if and only if triangle ABC is
equilateral and M.N.P are the midpoints of the sides.
3. Does there exist a convex pentagon, all of whose vertices are lattice points in the xy-
plane, with no lattice point in the interior? (A point in the xy-plane is called a lattice
point if it has integer coordinates.)
Solution. The answer is No. A convex lattice pentagon must have an interior lattice
point. To see this, note that every lattice point (x, y) belongs to one of the four classes
K00 , K01 , K10 and K11 , where the index pair ij is determined by taking i ≡ x and j ≡ y
modulo 2. A convex lattice pentagon has five vertices, so two of them, say P and Q belong
to the same class, which implies that their midpoint R is also a lattice point. If P and Q
are endpoints of a diagonal of the pentagon, then R is an interior lattice point. If P and
Q are the endpoints of an edge, say edge AB of the pentagon ABCDE, we continue by
considering the convex lattice pentagon ARCDE and so on. This case cannot continue
indefinitely, because if so, there would be an infinite sequence of distinct lattice points
within a finite region of the coordinate plane, which is not true.
2
4. A piece of cardboard in the shape of a square is to be cut into n acute- angled triangles.
Find the smallest n for which this can be done. Show at least one way to do it this
minimum n.
Solution. Suppose the square has been cut into n acute-angled triangles. Form a graph
whose vertices are the vertices of the square and the vertices of the triangles. Any two
consecutive vertices on a side of a triangle or on a side of a square are joined by an edge.
There are no other edges. There are three types of vertices:
(a) Vertices of the square: These are of degree at least 3.
(b) Vertices which are in the interior of a side of a triangle or the square. These are
of degree at least 4. Denote the number of vertices of these type by b
(c) Vertices of the triangles which do not lie in the interior of a side. These are of
degree at least 5. Denote the number of vertices of these type by c
Let m be the number of edges. Then
2m ≥ 4 × 3 + 4b + 5c.
But 2m = 4 + 3n + b. Euler’s formula gives
(4 + b + c) + (n + 1) − m = 3, or m = b + c + n + 3.
Substitute this into the inequality, we get c ≥ 2. Each vertex of type (c) is associated with
at least 5 triangles and the triangles associated to 2 vertices can have a overlap of at most
2. Thus we get n ≥ 8. See the picture below for a construction which gives n = 8.
............................................................................................................................................................................
.. ............... .... ...
........ ..... ......... .....
.. ......... ... . . ......... ......
...... ......... ...
............ . .... .... ........................ .
.. .. .......... .... .. . .. . ... .. ..
... .. ..................... .. .................... ... ....
.. ..... . .. .......................... .. ....
.. ....... ..
.. .. .
.
.. . . . . .. ... .
... .
. .
....
... ....... ...... ....... .... .... ...... .... ....... ...... ...
.. . ..... .... .... ...
.. .
.
. ..... ... ... ...
... ... ... .... ... ...
.. ... ...... .... ... ...
.. .
. ... ...
... ... ..... ... ... ...
.. .
.. ... ... .. ...
.. .. . . .
.
... .. ...
.... .. .... .....
...
.. ..... ... ... .. ... ..
...
.. ...... .. ... ...
.. ... ......... ...
... .. . .... ... ...
.. ... . .. .
... ...
.. .
. . . . .
... ..
. ..... ..... ... ....
.. .
. . . ... ....
.. .. .. .. . . . ..
...
.. ........... ....... ...
. ... .... ..... .. ..... .
. ....... ........ ...
..
.
. .... . .. . ...... ......
.. .... .
.......... ....... ... ...
. . . ...... ....
... . ....... . . . .
...... ..
.. ... ... . . . . ... ... . ...
.. ... ..... ......... .. ...... ... ... ...
.........
..... ...
....... ......
. .. .
.
.
.
.
. .
.....
... ...
... ...
..... ...... ....... .... ... ...
......................................................................................................................................................
5. (Prize Problem) Let x be a positive rational number. Prove that there exist a unique
set of integers a1 , a2 . . . , ak , with 0 ≤ an ≤ n − 1 for n > 1 such that
a2 a3 ak
x = a1 + + + ··· + .
2! 3! k!
Show also that there exist a set of integers, 106 < b1 < b2 < · · · bm such that
1 1 1
x= + + ··· + .
b1 b2 bm
3
Solution. First note that x can be written in the form m/n!, where m, n are positive
integers. We shall prove by induction on n that
m a2 an
= a1 + + ··· +
n! 2! n!
where 0 ≤ ai ≤ i − 1 for i ≥ 2. For n = 1, a1 = m, and ai = 0 for i > 1 works. Now
assume that it holds for some n. Let
m = (n + 1)q + r, 0≤r <n+1
Then
m q r
= + .
(n + 1)! n! (n + 1)!
The result then follows by applying the induction hypothesis on q/n! and putting r = ak+1 .
Next we prove uniqueness. Suppose
a2 a3 ak b2 b3 bk
x = a1 + + + ··· + = b1 + + + ··· + .
2! 3! k! 2! 3! k!
Then
ak ≡ bk ≡ x(k!) (mod k)
and it follows that ak = bk . By considering x − (ak /k!), we can show that ak−1 = bk−1 .
Thus uniqueness follows.
To prove the second part we first note that if ai > 0, then ai | i! and thus ai /i! is
the reciprocal of a positive integer. Moreover, (i − 1)! < i!/ai ≤ i!. Thus all the positive
integers are different.
Now if x < 1, then a1 = 0 and so x can be expressed as the sum of reciprocals of
different positive integers.
Now we consider the general case. Let x be any positive rational number and m be
an integer > 106 such that m > 1/x and n be the largest integer ≥ m such that
1 1 1
x≥ + + ··· + .
m m+1 n
Let
1 1 1
y =x− − − ··· − .
m m+1 n
Then 0 ≤ y < 1/(n + 1). Thus y can be written as a sum of reciprocals of different positive
integers:
1 1
y= + ··· + .
q1 qj
Since 1/qi < y < 1/(n + 1), qi > (n + 1). Thus
1 1 1 1 1
x= + + ··· + + + ··· +
m m+1 n q1 qj
4
as required.
6. Given a segment AB of length 1, define the set M of points as follows: A, B ∈ M and if
X, Y ∈ M , then M contains the point Z in the segment XY for which Y Z = 3XZ. Prove
that M does not contain the midpoint of AB.
Solution. Represent A by 0 and B by 1 on the number line. Denote by Mn the set of points
of the segment AB obtained from A, B by not more than n iterations. It can be proved
by induction that Mn consists of all points in [0, 1] represented by 3k/4n and (3k − 2)/4n ,
where k is an integer. Thus M consists of numbers of the form 3k/4n and (3k − 2)/4n . To
prove our assertion, we need to show that 1/2 cannot be expressed in this form. Suppose
1/2 = 3k/4n , then 4n = 6k which has no solution. Suppose 1/2 = (3k − 2)/4n , then
6n = 4n + 4 which has no solution. Thus 1/2 6∈ M .
2nd soln (Joel): Represent each point by a coordinate in the set [0, 1] with A = 0,
B = 1. For any two points x, y ∈ M , the point (3x + y)/4 ∈ M . Now take decimal
representations. Note that each point is a finite decimal. It’s easy to see that the digit
sum of the decimal representation ≡ 0or1 (mod 3) for any point in M since it initially
holds for A, B and if it holds x, y then it also holds for (3x + y)/4. Since midpoint of AB
has decimal representation of 0.5, it is not in M .
7. (Prize Problem) Let a1 , a2 , . . . , an , n ≥ 1, be real numbers ≥ 1 and A = 1 + a1 +
· · · + an . Define xk , 0 ≤ k ≤ n by
1
x0 = 1, xk = , 1 ≤ k ≤ n.
1 + ak xk−1
Prove that
n2 A
x1 + x2 + · · · + xn > .
n2 + A2
(Hint: Let yk = 1/xk )
ak
Solution. Let yk = 1/xk . We then have yk = 1 + yk−1 . From yk−1 ≥ 1, ak ≥ 1 we obtain
µ ¶
1 ak 1
− 1 (ak − 1) ≤ 0 ⇔ 1 + ≤ ak + .
yk−1 yk−1 yk−1
ak 1
So yk = 1 + yk−1 ≤ ak + yk−1 . We have
n
X n
X n
X n
X n−1
X 1 X 1 n−1 X 1 n
1 1
yk ≤ ak + = ak + + =A+ <A+ .
yk−1 y0 yk yk yk
k=1 k=1 k=1 k=1 k=1 k=1 k=1
Pn Pn
Let t = k=1 1/yk . Then k=1 yk ≥ n2 /t. So for t > 0,
n2 t < A + t ⇔ t2 + At − n2 ≥ 0
√
−A + A2 + 4n2 2n2
⇔ t> = √
2 A + A2 + 4n2
2n2 n2 A
≥ = .
A + A + (2n2 /A) n 2 + A2
5
Training problems 1 April 2003
8. In a group of interpreters each one speaks one or several languages, 24 of them speak
Japanese, 24 Chinese and 24 English. Prove that it is possible to select a subgroup in
which exactly 12 interpreters speak Japanese, exactly 12 speak Chinese and exactly 12
speak English.
Solution: Suppose that in a group of interpreters n speak Japanese, n speak Chinese and
n speak English. Denote these groups by A, B, C. Put p = |A ∩ B c ∩ C c |, q = |Ac ∩ B ∩ C c |,
r = |Ac ∩ B c ∩ C|, a = |Ac ∩ B ∩ C|, b = |A ∩ B c ∩ C|, c = |A ∩ B ∩ C c |, d = |A ∩ B ∩ C|.
A group of interpreters is called a k-group if exactly k interpreters speak Japanese,
exactly k speak Chinese and exactly k speak English.
We shall prove by induction on n, that for n ≥ 2, it’s possible to find a 2-group inside
an n-group.
When n = 2, it’s trivially true. Now suppose n > 2 is an integer and that for each k,
2 ≤ k < n, the result is true.
1. a, b, c > 0: It’s enough to select one interpreter from each of the sets:
Ac ∩ B ∩ C, A ∩ B c ∩ C, A ∩ B ∩ C c.
2. p, q, r > 0: Select one from each of the sets:

Ac ∩ B c ∩ C, Ac ∩ B ∩ C c , A ∩ Bc ∩ C c
and then apply induction on the remaining people.
3. d > 0: It’s enough to select one from A ∩ B ∩ C apply the induction hypothesis to
the remaining group.
4. None of the above hold: One of a, b, c is 0, say a = 0; d = 0 and one of p, q, r is 0.
We have q + c = b + r = p + b + c = n. If q = 0, then c = r = n and p = b = 0. Thus
c > 0, r > 0. We can choose one from each of A ∩ B ∩ C c , Ac ∩ B c ∩ C and then apply
the induction hypothesis. The case r = 0 is similar. The final case if p = 0 and r, q > 0.
Then since b + c = n, one of them is positive, say b > 0. Then choose one from each of
A ∩ B c ∩ C, Ac ∩ B ∩ C c and then apply the induction hypothesis.
Thus from n = 24, we can choose k as long as k is an even number less than 24.
Solution: 2nd soln by Colin. Let L1 , L2 , L3 be the three languages. Divide the interpreter
into 7 groups S(l1 , l2 , l3 ) where li = 1 if the people from the group speak Li and li = 0
otherwise. Let a1 , . . . , a7 be,respectively, the number of people in the S(1, 0, 0), S(0, 1, 0),
S(0, 0, 1), S(0, 1, 1), S(1, 0, 1), S(1, 1, 0), S(1, 1, 1).
We shall prove that for any n ≥ 2, it’s possible to find a 2-group inside an n-group.
We have 3 equations by considering the interpreters who can speak each of the lan-
guages in turn.
a1 + a5 + a6 + a7 = n
a2 + a4 + a6 + a7 = n
a3 + a4 + a6 + a7 = n
1
Without loss of generality, we can assume that a1 ≤ a2 ≤ a3 . The solutions are of the
form
(a1 , . . . , a7 ) = (a, a + b, a + c, d, d + b, d + c, n − (a + b + c + 2d))
for (independent) nonnegative integers a, b, c, d.
The set {S(1, 0, 0), S(0, 1, 0), S(0, 0, 1)} gives a 1-groups. The set {S(0, 1, 0), S(1, 0, 1)}
gives b 1-groups. The set {S(0, 0, 1), S(1, 1, 0)} gives c 1-groups. The set S(1, 1, 1) gives
24 − (a + b + c + 2d) 1-groups. The set {S(0, 1, 1), S(1, 0, 1), S(1, 1, 0)} gives d 2-groups.
If (a) + (b) + (c) + (24 − (a + b + c + 2d)) = n − 2d ≥ 2, then there are 2 1-groups
which will combine to give a 2-group. Otherwise, n − 2d ≤ 1, or 2d ≥ n − 1 or d ≥ 1. We
still have a 2-group.
Apply this result to the case n = 24, we have a 2-group. Remove this 2-group, we are
left with the case with n = 22. Continuing this way, we can find 6 distinct 2-groups and
they combine to give a 12-group as desired.
9. Points P1 , . . . , Pn are placed inside or on the boundary of a disk of radius 1 in such a
way that the minimum distance dn between any two of these points has its largest possible
value Dn . Calculate Dn for n = 2, . . . , 7. Justify your answers.
Solution: Suppose n ≤ 6. Decompose the disk by its radii into n congruent regions so that
one of the points is on one of the radii. Then there is one region (including its boundary)
which contains 2 of the points. Since the distance between any two points in a region is at
most 2 sin π/n, then dn ≤ 2 sin π/n. If points Pj are placed in the vertices of regular n-gon
inscribed in the boundary of the disk, then dn = 2 sin π/n. Therefore Dn = 2 sin π/n.
For n = 7, we have D7 ≤ D6 = 1. If one of the given points is placed in the center of
the disk and if the other 6 points are placed at the vertices of the regular hexagon inscribed
in the boundary of the disk, then d7 = 1. Thus D7 = 1.
10. Prove that in any triangle, a line passing through the incentre cuts the perimeter of
the triangle in half if and only if it halves the area of the triangle.
Solution: Let ABC be the triangle and O, r denote t he incentre and inradius. Let l be a
line passing through O. It intersects one side, say BC, at an interior point. Without loss
of generality, let it intersect the side AC at P . (Note that P may coincide with A.) Let
x = P C, y = QC. For a triangle XY Z, denote its area by [XY Z]. Then
r(a + b + c) r(x + y)
[ABC] = , [CP Q] =
2 2
the latter because the altitudes of 4OCP, 4OCQ from O are both r. The line l halves
the area iff
a + b + c = 2(x + y)
iff l halves the perimeter.
11. Nine positive integers a1 < a2 < · · · < a9 are such that all the sums (of at least one
and at most nine different terms) that can be made up of them are different. Prove that
a9 > 100.
Solution: Assume that a9 ≤ 100. Let S be the set of those sums ≥ a4 of 8at
most 5 terms
8
out of a1 , . . . , a8 . The number of sums of at most 5 terms is 1 + · · · + 5 = 218. Those
2
that can be less than a4 are made up of a1 , a2 , a3 and there are at most 7 of them. Thus
|S| ≥ 211. The greatest sum in S is a4 + · · · + a8 < a4 + 4a9 and therefore all the sums
are in [a4 , a4 + 4a9 ]. The inequality |S| ≥ 2a9 implies that there are 3 numbers which are
congruent mod a9 . Thus 2 of them must have a difference of a9 , a contradiction.
12. The quadrilateral ABCD inscribes in a circle with centre O. Let BA meet CD at P ,
AD meet BC at Q and AC meet BD at M . Show that O is the orthocentre of triangle
P QM .
Solution: Let R be the radius of the circle. As ∠QM D > ∠CBD = ∠DAM , one can
extend QM to QF such that ∠F AD = ∠QM D. Then A, D, M, F are concyclic. Also
∠QBD = ∠DAM = ∠DF M so that B, F, D, Q are concyclic.
P
.......
...... ..
....
. ..... ....
.
...... ...
..... ..
............................ A ...... ...
... ................... .
... .............
........ .
.
.
....... ........ ...
...... ...... .......... ...
..... ..... ..... ........ ...
.......
. .......... .. .... .............. .
.
. .. .. ... ...
.
... ...... ..
... ...... .. .... .......... ...
..... .. .... ... .. ...
..... ...
....... .
. . . ... .
. .
.
... ......
. ...... ........ ... ........
... ... ... ... .......
...
..
......
......
..... F .
..... ............. ......
. ...
..........................
D
........
.... ......
..
....
........ ......... ... .... .... .... .....
.. ... ......
.
... ...........
.
... ....................................
.... . ............
·
......
M ... ....... ... ... ....
... ...... . ...
B .................... O ... ..........

... .... ............
...
...
... .................
.. .
. .. .
.. ..... .....
... ............. . . . ..... .
... . ............... .
.
... .. ....
. .
. ..... .....
.
... ............. ... ... ... ..... ...
... ............. . ..... ...
... .
..... ................. ...... .
. .. .. ..... ...
.... ................ ..... ..
..... ...
.....
..... . . . ..... ........................
. ..... ...
.......
......
.......
........ ....... ..
... .
....
. . .
C ............
.
. ............ ...............
.................
.......
............. . ...
.
.............................
Q
Thus, QM · QF = QD · QA = QO2 − R2 , and QM · M F = M B · M D = R2 − M O2 .

Subtracting, we get QM (QF − M F ) = QO2 + M O2 − 2R2 . That is QM 2 = QO2 + M O2 −
2R2 Similarly, P M 2 = P O2 + M O2 − 2R2 . Subtracting again, we have P M 2 − QM 2 =
P O2 − QO2 .
It follows from this that OM is perpendicular to P Q. To see this, suppose the exten-
sion of OM meets P Q at E and ∠P EM > ∠QEM . By cosine rule, P M 2 = EP 2 +M E 2 −
2EP · M E cos ∠P EM > EP 2 + M E 2 . Similarly, P O2 > EP 2 + OE 2 , QM 2 < EQ2 + M E 2
and QO2 < EQ2 +OE 2 . Subtracting, we obtain P M 2 −QM 2 > EP 2 −EQ2 > P O2 −QO2 ,
a contradiction.
Similarly, P M is perpendicular to OQ and QM is perpendicular to OP . Therefore,
O is the orthocentre of triangle P QM .
(Alternate Solution) Let OM meet P Q at W . From W draw tangents to the circle touching
it at H and G Then H, M, G are collinear as M is the pole of P Q. Let HG meet P Q at
Z. Then the cross ratio (H, G; M, Z) = −1. Since W M bisects ∠HW G, we have OW or
M W is perpendicular to P Q.
13. Suppose a1 , a2 , · · · , an are n ≥ 3 positive numbers such that (a21 + a22 + · · · + a2n )2 >
(n − 1)(a41 + a42 + · · · + a4n ). Prove that any three such a0i s form the lengths of the sides of
a triangle.
Solution: First we prove the assertion when n = 3. Let’s write the numbers as a, b, c.
They satisfy the inequality : (a2 + b2 + c2 )2 > 2(a4 + b4 + c4 ). We may assume without
3
loss of generality that a ≥ b ≥ c. Then
0 ≤ (a2 + b2 + c2 )2 − 2(a4 + b4 + c4 ) = −[a2 − (b + c)2 ][a2 − (b − c)2 ].
Thus, |b − c| < a < |b + c|. Therefore, a, b, c are the lengths of the sides of a triangle.
In the general case, we can simply show that a1 , a2 , a3 are the lengths of the sides
of a triangle. Using Cauchy-Schwarz inequality, we have (n− 1)(a41 + a42 + · · · + a4n ) ≤
2 2 2 2 2
a21 +a22 +a23 a21 +a22 +a23

a +a +a
1· 2 +1· 2 + 1 · a4 + · · · + 1 · an ≤ (n−1) 2 1 22 3 a44 + · · · + a4n .
2 2
From this, we get (a21 + a22 + a23 )2 > 2(a41 + a42 + a43 ). Using the case for n = 3, a1 , a2 , a3
are the lengths of the sides of a triangle.
4
Training problems 10 April 2003
14. Let A be a set with 8 elements. Find the maximum number of distinct 3-element
subsets of A such that the intersection of any two of them is not a 2-element set.
Solution. Let G be a graph with vertex set A. Two vertices are adjacent if and only if they
belong to the same 3-element subset. Thus 3 vertices that belong to the same 3-element
subset will form a 3-cycle (C3 ). The question asks for the maximum number of C3 s that
G can contain subject to the condition that every 2 of these C3 s do not have a common
edge. (The graph may, however, contain other C3 . But these will share an edge with the
C3 s we want.) We may further assume that every edge belongs to some C3 .
Since at most 3 C3 can share a common vertex, every vertex if od degree at most 6.
Thus G has at most (6 × 8)/2 = 24 edges and so at most 8 C3 s.
An example with 8 sets is:
123, 145, 167, 246, 278, 348, 357, 568.
2nd solution. This is not easier than the first but it introduces a very useful idea of
an incidence matrix.
Let B1 , . . . , Bn be 3-element subsets of A such that |Bi ∩ Bj | 6= 2. Form an inci-
dence matrix with rows indexed by B1 , . . . , Bn and with columns indexed by the elements
a1 , . . . , a8 of A. An entry (Bi , aj ) is 1 if aj ∈ Bi and is 0 otherwise. Then there are 3 ones
in every row. Call a pair of ones in the same column a 1-pair. The given condition states
that there is at most 1 1-pair in every pair of rows. Suppose there is a column, say column
1, that has 4 ones, say in the first 4 rows. Then in the submatrix formed by the first 4
rows and the last 7 columns, there are 8 ones. Thus there is at least one 1-pair. Hence
some pair of rows has two 1-pairs, a contradiction.
Thus every column has at most 3 ones. Counting the total number of ones in the
incidence matrix in 2 different ways, we conclude that n ≤ 8.
It’s not hard to get an example with 8 sets. Thus the answer is 8.
15. Find all primes p for which p(2p−1 − 1) is the kth power of a positive integer for some
k > 1.
Solution. Let p(2p−1 − 1) = xk for some positive integers x, k. It’s clear that p 6= 2.
Thus p = 2q + 1. Write x = py. Then (2q + 1)(2q − 1) = pk−1 y k . Since at least one of
2q + 1 and 2q − 1 is the kth power of an integer since they are coprime.
Case (1): 2q − 1 = z k . Then 2q = z k + 1. If k is even, then z k + 1 is not divisible by
4. Hence q = 1, p = 3 and p(2p−1 − 1) = 32 .
If k = 2` + 1, i.e., odd, then
2q = (z + 1)(z 2` − z 2`−1 + · · · − z + 1) i.e z + 1 = 2α , 0 ≤ α < q.
On the other hand,
2q = (2α − 1)2`+1 + 1 = A22α + 2α (2` + 1), A is an integer
1
The last equality contradicts with α < q.
Case (2): 2q + 1 = z k . Then 2q = z k − 1. If k is odd we get a contradiction as before.
If k = 2`, then (z ` −1)(z ` +1) = 2q and since gcd(z ` −1, z ` +1) = 2, we have z ` −1 = 2,
i.e., q = 3, p = 7, p(2p−1 − 1) = 212 .
Thus the answers are p = 3, 7.
16. Let k be a given real number. Find all functions f : (0, ∞) → (0, ∞) such that the
following equality holds for all positive real number x:
x
kx2 f (1/x) + f (x) = .
x+1
Solution. Divide by x we get
1 1
kxf (1/x) + f (x) =
x x+1
Replace x by 1/x we get

k x
f (x) + xf (1/x) = .
x 1+x
Solve the system of equations with unknowns f (x) and f (1/x), we get
(1 − k 2 )f (x) 1 − kx
= .
x x+1
If k 6= ±1, then there is a unique solution
x 1 − kx
f (x) =
x + 1 1 − k2
It’s easy to see that this expression satisfies the given functional equation.
If k = ±1, there is no solution. For f (x) > 0 for all positive x, we also need −1 < k <
0. Thus such a function exists only when −1 < k < 0. In this case the unique solution is
x 1−kx
f (x) = x+1 1−k2 .
17. Let n be an integer, n ≥ 3. Let a1 , . . . , an be real numbers, where 2 ≤ ai ≤ 3 for

i = 1, . . . , n. if s = a1 + · · · + an . prove that
a21 + a22 − a23 a2 + a23 − a24 a2 + a21 − a22

+ 2 + ··· + n ≤ 2s − 2n.
a1 + a2 − a3 a2 + a3 − a4 an + a1 − a2
Solution. Write
a2i + a2i+1 − a2i+2 2ai ai+1

Ai = = ai + ai+1 + ai+2 − .
ai + ai+1 − ai+2 ai + ai+1 − ai+2
2
Since (ai − 2)(ai+1 − 2) ≥ 0, −2ai ai+1 ≤ −4(ai + ai+1 − 2) and
µ ¶
ai+2 − 2
Ai ≤ ai + ai+1 + ai+2 − 4 1 + .
ai + ai+1 − ai+2
Since 1 = 2 + 2 − 3 ≤ ai + ai+1 − ai+2 ≤ 3 + 3 − 2 = 4,

µ ¶
ai+2 − 2
Ai ≤ ai + ai+1 + ai+2 − 4 1 + = ai + ai+1 − 2.
4
P
Hence Ai ≤ 2s − 2n.
18. Two chords U V and RS of a circle C centred at O intersect at the point N . Suppose
AB is a line segment outside the circle C such that AU, AV, BR and BS are tangent to
the circle C at U, V, R and S respectively. Prove that ON is perpendicular to AB.
Solution. ..............
.....................................
.........
......... .......
....... ......
.
......... ......
...
... .....
.....
...... .....
..... .....
.... ...
. .. ...
... ...
... ...
...
. ..
..
. ...
.
. ...
... ...
U• .
.
. .....
..... .........
...
...
.
. .
........
......
.....
..
O . .. .
...
...
..
. ...
.
.
.....
. ..... ..... ..
. ..... .. .. ...
.... .... ..... .. ....
.. . .... ...
.
. .
. ....... ... .
... .. ..... ..... .. ...
...
.
.
.. ...
... ..... P
.....
..... ........
..
. .
..... .. ...
.. .....
. . . .. ..
.......
....... .....
•S
. ... .. ... . .
... ..... ....... ... . ...
....... .....
... ..... ......... ... ... ... ..... ...
..
...
...
... ....
..... ..... Q
.. ... ... .. .... .............
....... ......
.......
.... ...
..... ........ .... .
........ . . . .
. . ..
.... ..... . ... .
. ..... . ......... ... ..
... ......... ..... .. ............. ..... .... ..
.. ..... ........... ..... ...
..... .......... ... ..... ....
....
. .....
....
.
.. ......
.......
.
..
... . ....
.
....... ... ........
N
..... .... ......
..........
....
..
... ..... .................... ..... .. ...
...
.
. ... .
..
. ..... • .........
.. .....
.... ... .. .
.. .
..
........ ................. •.
...
...
... .
..........................
.
. ... ..... .....
... .... R .. . ...
.............. ................. ...
..
..
... .........
. .
.....
.. .. . ..... .
...............
..............
.
V
.. ............
..
.. ..
...
... ....
.
...
. ......... .
.. ..... .............. .. ......... .... ....
....... ................. . .. .
.........................................................................................................................................................................................
A M B
Join ON and extend it to meet AB at M . Let OA intersect U V at P and OB

intersect RS at Q. Join P Q. Then ∠OP N = ∠OQN = 90◦ . Hence, O, P, N and Q are
concyclic. As OP · OA = OU 2 = OS 2 = OQ · OB, we have A, B, Q and P are concyclic.
Therefore, ∠OAM = ∠OQP = ∠ON P . This shows that P, A, M and N are concyclic.
Hence, ∠AM O = ∠OP N = 90◦ .
(2nd solution by Colin Tan) Extend BN to W such that ∠N W S = ∠BSN (= ∠BRN ).
This is possible as ∠BN S > ∠BRN . Thus BS is tangent to circumcircle of SN W and
SW RB cyclic. This gives the relations BN · BW = BS 2 = OB 2 − OR2 and BN · N W =
SN · N R so BN 2 = BN · BW − BN · N W = OB 2 − OR2 − SN · N R. Get a similar
expression for AN 2 , and this would give BN 2 − AN 2 = OB 2 − OA2 which implies that
ON is perpendicular to AB. (Compare the proof of this with question 12. Also the same
proof using projective geometry as in question 12 can be applied here.)
19. Let ABCD be a cyclic quadrilateral. Prove that
|AC − BD| ≤ |AB − CD|.
3
When does equality hold?
Solution. Let E and F be the midpoints of the diagonals AC and BD. In every
quadrilateral the following relation due to Euler holds:
AC 2 + BD2 + 4EF 2 = AB 2 + BC 2 + CD2 + DA2 .
Since ABCD is a cyclic quadrilateral, we have Ptolemeus identity
AB · CD + AD · BC = AC · BD.
Hence,
(AC − BD)2 + 4EF 2 = (AB − CD)2 + (AD − BC)2 .
Let us prove that 4EF 2 ≥ (AD − BC)2 . This will implies the stated inequality. Let M
be the midpoint of AB. In the triangle M EF , we have AD = 2M F , BC = 2M E, and
from triangle’s inequality, EF ≥ |M E − M F |, hence 2EF ≥ |BC − AD| and 4EF 2 ≥
(AD − BC)2 .
The equality holds if and only if the points M, E, F are collinear, which happens if
and only if AB is parallel to CD, that is ABCD is either an isosceles trapezium or a
rectangle.
20. Let Γ be a convex polygon with 2000 sides and P an interior point which does not lie
on any diagonal of Γ. Prove that P is in the interior of an even number of triangles formed
using the vertices of Γ.
Solution. First observe that if P lies in a quadrilateral, then it is contained in the interiors
of two triangles. Next, if a triangle 4 contains P , then any quadrilateral containing 4
also contains P . As each triangle in Γ is contained in 1997 quadrilaterals, the point P ∈ 4
is contained in 1997 quadrilaterals. Let m be the number of quadrilaterals containing P
and n the number of triangles containing P . Then 2m = 1997n. Hence, n must be even.
Here, we are counting the number of pairs (4, Q), where P lies in the triangle 4 which is
inside the quadrilateral Q.
4
Singapore International Mathematical Olympiad Committee
22-9-2001
1. Consider the set of integers A = {2a 3b 5c : 0 ≤ a, b, c ≤ 5}. Find the smallest number n
such that whenever S is a subset of A with n elements, you can find two numbers p, q in
A with p | q.
2. “Words” are formed with the letters A and B. Using the words x1 , x2 , . . ., xn we can
form a new word if we write these words consecutively one next to another: x1 x2 . . . xn .
A word is called a palindrome, if it is not changed after rewriting its letters in the reverse
order. Prove that any word with 1995 letters A and B can be formed with less than 800
palindromes.
3. Let n ≥ 2 be an integer and M = {1, 2, . . . , n}. For every k ∈ {1, 2, . . . , n − 1}, let
1 X
xk = (min A + max A).
n + 1 A⊂M
|A|=k
Prove that x1 , x2 , . . . , xn−1 are integers, not all divisible by 4.
4. The lattice frame construction of 2 × 2 × 2 cube is formed with 54 metal shafts of length
1 (points of shafts’ connection are called junctions). An ant starts from some junction A
and creeps along the shafts in accordance with the following rule: when the ant reaches
the next junction it turns to a perpendicular shaft. t some moment the ant reaches the
initial junction A; there is no junction (except for A) where the ant has been twice. What
is the maximum length of the ant’s path?
5. Let n black and n white objects be placed on the circumference of a circle, and define
any set of m consecutive objects from this cyclic sequence to be an m-chain.
(a) Prove that for each natural number k ≤ n, there exists a chain of 2k consecutive
pieces on the circle of which exactly k are black.
(b) Prove that there are at least two such chains that are disjoint if
√
k≤ 2n + 2 − 2
1
29-9-2001
6. We are given 1999 rectangles with sides of integer not exceeding 1998. Prove that
among these 1999 rectangles there are rectangles, say A, B and C such that A will fit
inside B and B will fit inside C.
7. We are given N lines (N > 1) in a plane, no two of which are parallel and no three
of which have a point in common. Prove that it is possible to assign, to each region of
the plane determined by these lines, a non-zero integer of absolute value not exceeding N ,
such that the sum of the integers on either side of any of the given lines is equal to 0.
8. Let S be a set of 2n + 1 points in the plane such that no three are collinear and no four
concyclic. A circle will be called good if it has 3 points of S on its circumference, n − 1
points in its interior and n − 1 in its exterior. Prove that the number of good circles has
the same parity as n.
2
6-10-2001
9. Let ARBP CQ be a hexagon. Suppose that ∠AQR = ∠ARQ = 15o , ∠BP R =

∠CP Q = 30o and ∠BRP = ∠CQP = 45o . Prove that AB is perpendicular to AC.
10. Γ1 and Γ2 are two circles on the plane such that Γ1 and Γ2 lie outside each other.
An external common tangent to the two circles touches Γ1 at A and Γ2 at C and an
internal common tangent to the two circles touches Γ1 at B and Γ2 at D. Prove that the
intersection of AB and CD lie on the line joining the centres of Γ1 and Γ2
11. Let E and F be the midpoints of AC and AB of 4ABC respectively. Let D be a

point on BC. Suppose that
(i) P is a point on BF and DP is parallel to CF ,
(ii) Q is a point on CE and DQ is parallel to BE,
(iii) P Q intersects BE and CF at R and S respectively.
Prove that P Q = 3RS.
12. Let AD, BE and CF be the altitudes of an acute-angled triangle ABC. Let P be a
point on DF and K the point of intersection between AP and EF . Suppose Q is a point
on EK such that ∠KAQ = ∠DAC. Prove that AP bisects ∠F P Q.
13. In a square ABCD, C is a circular arc centred at A with radius AB, P and M are
points on CD and BC respectively such that P M is tangent to C. Let AP and AM
intersect BD at Q and N respectively. Prove that the vertices of the pentagon P QN M C
lie on a circle.
3
13-10-2001
14. In triangle ABC, the angle bisectors of angle B and C meet the median AD at points
E and F respectively. If BE = CF , prove that 4ABC is isosceles.
15. Let ABCD be a convex quadrilateral. Prove that there exists a point E in the plane
of ABCD such that 4ABE is similar to 4CDE.
16. Let P, Q be points taken on the side BC of a triangle ABC, in the order B, P, Q, C.
Let the circumcircles of P AB, QAC intersect at M (6= A) and those of P AC, QAB at N
(6= A). Prove that A, M, N are collinear if and only if P, Q are symmetric in the midpoint
A0 of BC.
17. About a set of four concurrent circles of the same radius r, four of the common tangents
are drawn to determine the circumscribing quadrilateral ABCD. Prove that ABCD is a
cyclic quadrilateral.
18. Three circles of the same radius r meet at common point. Prove that the triangle
having the other three points of intersections as vertices has circumradius equal to r
4
27-10-2001
19. For any positive real numbers a, b, c,

a b c 3
+ + ≥ .
b+c c+a a+b 2
20. Let a, b, c be positive numbers such that a + b + c ≤ 3. Prove that

1 1 1 3
+ + ≥ .
1+a 1+b 1+c 2
21. Prove that for any positive real numbers a, b, c,

a b c 1
+ + ≥ .
10b + 11c 10c + 11a 10a + 11b 7
22. Prove that for any positive real numbers a, b, c, d, e,

a b c d
+ + +
b + 2c + 3d + 4e c + 2d + 3e + 4a d + 2e + 3a + 4b e + 2a + 3b + 4c
e 1
+ ≥ .
a + 2b + 3c + 4d 2
23. Let n be a positive integer and let a1 , a2 , . . ., an be n positive real numbers such that
a1 + a2 + · · · + an = 1. Is it true that
a41 a42 a4n 1
2 2 + 2 2 + · · · + 2 ≥ ?
a1 + a2 a2 + a3 a2n + a1 2n
24. Let a, b, c, d be nonnegative real numbers such that ab + bc + cd + da = 1. Prove that

a3 b3 c3 d3 1
+ + + ≥ .
b+c+d a+c+d a+b+d a+b+c 3
25. (IMO 95) Let a, b and c be positive real numbers such that abc = 1. Prove that
1 1 1 3
+ 3 + 3 ≥ .
a3 (b + c) b (a + b) c (a + b) 2
Cauchy’s inequality is useful for these questions.

(x21 + x22 + · · · + x2i )(y12 + y22 + · · · + yi2 ) ≥ (x1 y1 + x2 y2 + · · · + xi yi )2 .
Equality holds iff xj = tyj for all j, where t is some constant.
5
3-11-2001
26. A sequence of natural numbers {an } is defined by a1 = 1, a2 = 3 and
an = (n + 1)an−1 − nan−2 (n ≥ 2).
Find all values of n such that 11|an .
27. Let{xn }, n ∈ N be a sequence of numbers satisfying the condition

p
−xn + 3 − 3x2n
|x1 | < 1, xn+1 = , (n ≥ 1).
2
(a) What other condition does x1 need to satisfy so that all the numbers of the
sequence are positive?
(b) Is the given sequence periodic?
28. Suppose that a function f defined on the positive integers satisfies f (1) = 1, f (2) = 2,
and
f (n + 2) = f (n + 2 − f (n + 1)) + f (n + 1 − f (n)), (n ≥ 1).
(a) Show that 0 ≤ f (n + 1) − f (n) ≤ 1.
(b) Show that if f (n) is odd, then f (n + 1) = f (n) + 1.
(c) Determine, with justification, all values of n for which f (n) = 210 + 1.
29. Determine the number of all sequences {x1 , x2 , . . . , xn }, with xi ∈ {a, b, c} for i =
1, 2, . . . , n that satisfy x1 = xn = a and xi 6= xi+1 for i = 1, 2, . . . , n − 1.
30. Given is a prime p > 3. Set q = p3 . Define the sequence {an } by:
½
n for n = 0, 1, 2, . . . , p − 1,
an =
an−1 + an−p for n > p − 1.
Determine the remainder when aq is divided by p.
31. A and B are two candidates taking part in an election. Assume that A receives m
votes and B receives n votes, where m, n ∈ N and m > n. Find the number of ways in
which the ballots can be arranged in order that when they are counted, one at a time, the
number of votes for A will always be more than that for B at any time during the counting
process.
6
10-11-2001
32. Find all prime numbers p for which the number p2 + 11 has exactly 6 different divisors
(including 1 and the number itself.)
33. Determine all pairs (a, b) of positive integers such that ab2 + b + 7 divides a2 b + a + b.
34. Let p be an odd prime. Prove that

µ ¶p−2
p−2 p−2 p−2 p−1 2 − 2p
1 +2 +3 + ··· + ≡ (mod p).
2 p
35. (10th grade) Let d(n) denote the greatest odd divisor of the natural number n. Define
the function f : N → N by f (2n − 1) = 2n , f (2n) = n + 2n/d(n) for all n ∈ N. Find all k
such that f (f (. . . (1) . . .)) = 1997 where f is iterated k times.
36. Given three real numbers such that the sum of any two of them is not equal to 1,
prove that there are two numbers x and y such that xy/(x + y − 1) does not belong to the
interval (0, 1).
7
17-11-2001
37. In the parliament of country A, each MP has at most 3 enemies. Prove that it is
always possible to separate the parliament into two houses so that every MP in each house
has at most one enemy in his own house.
38. Let T (x1 , x2 , x3 , x4 ) = (x1 − x2 , x2 − x3 , x3 − x4 , x4 − x1 ). If x1 , x2 , x3 , x4 are not equal

integers, show that there is no n such that T n (x1 , x2 , x3 , x4 ) = (x1 , x2 , x3 , x4 ).
39. Start with n pairwise different integers x1 , x2 , x3 , xn , n > 2 and repeat the following
step:
x1 + x2 x2 + x3 xn + x1
T : (x1 , x2 , · · · , xn ) → ( , ,···, ).
2 2 2
Show that T, T 2 , · · ·, finally leads to nonintegral component.
40. Is it possible to transform f (x) = x2 + 4x + 3 into g(x) = x2 + 10x + 9 by a sequence

of transformations of the form
f (x) → x2 f (1/x + 1) or f (x) → (x − 1)2 f (1/(x − 1))?
41. Is it possible to arrange the integers 1, 1, 2, 2, · · ·, 1998, 1998, such that there are
exactly i − 1 other numbers between any two i’s?
42. A rectangular floor can be covered by n 2 × 2 and m 1 × 4 tiles, one tile got smashed.
Show that one can not substitute that tile by the other type (2 × 2 or 1 × 4).
43. In how many ways can you tile a 2 × n rectangle by 2 × 1 dominoes?
44. In how many ways can you tile a 2 × n rectangle by 1 × 1 squares and L trominoes?
45. In how many ways can you tile a 2 × n rectangle by 2 × 2 squares and L trominoes?
46. Let a1 = 0, |a2 | = |a1 + 1|, · · · , |an | = |an−1 + 1|. Prove that
a1 + a2 + · · · + an 1
≥− .
n 2
47. Find the number an of all permutations σ of {1, 2, . . . , , n} with |σ(i) − i| ≤ 1 for all i.
48. Can you select from 1, 12 , 14 , 18 , . . . an infinite geometric sequence with sum (a) 15 ? (b) 17 ?
49. Let x0 , a > 0, xn+1 = 12 (xn + a

xn ). Find lim an .
n→∞
8
50. Let 0 < a < b, a0 = a and b0 = b. For n ≥ 0, define
p an + bn
an+1 = an bn , bn+1 = .
2
Show that lim an = lim bn .
n→∞ n→∞
51. Let a0 , a1 = 1, an = 2an−1 + an−2 , n > 1. Show that 2k |an if and only if 2k |n.
9
24-11-2001
52. Determine all α ∈ R such that there exists a nonconstant function f : R → R such
that
f (α(x + y)) = f (x) + f (y).
53. Let f : N → N be a function satisfying
(a) For every n ∈ N, f (n + f (n)) = f (n).
(b) For some n0 ∈ N, f (n0 ) = 1.
Show that f (n) = 1 for all n ∈ N.
54. Suppose that f : N → N satisfies f (1) = 1 and for all n,
(a) 3f (n)f (2n + 1) = f (2n)(1 + 3f (n)),
(b) f (2n) < 6f (n).
Find all (k, m) such that f (k) + f (m) = 2001.
55. Let f : R → R be a function such that for all x, y ∈ R,

f (x3 + y 3 ) = (x + y)((f (x))2 − f (x)f (y) + (f (y))2 ).
Prove that for all x ∈ R, f (2001x) = 2001f (x).
56. Find all functions f : N → N with the property that for all n ∈ N,
1 1 1 f (f (n))
+ + ··· + = .
f (1)f (2) f (2)f (3) f (n)f (n + 1) f (n + 1)
57. Define f : N → N and g : N → Z such that
(a) f (x, x) = x,
(b) f (x, y) = f (y, x),
(c) f (x, y) = f (x, x + y),
(d) g(2001) = 2002,
(e) g(xy) = g(x) + g(y) + mg(f (x, y)).
Determine all integers m for which g exists.
10
1-12-2001
a2n
58. Let a0 = 2001 and an+1 = an +1 . Find the largest integer smaller than or equal to
a1001 .
59. Given a polynomial f (x) = x100 − 600x99 + · · · with 100 real roots and that f (7) > 1,
show that at least one of the roots is greater than 7.
60. We define S(n) as the number of ones in the binary representation of n. Does there
2
exist a positive integer n such that S(n ) 501
S(n) < 2001 ?
61. A semicircle S is drawn on one side of a straight line l. C and D are points on S. The
tangents to S at C and D meet l at B and A respectively, with the center of the semicircle
between them. Let E be the point of intersection of AC and BD, and F be the point on
l such that EF is perpendicular to l. Prove that EF bisects ∠CF D.
62. At a round table are 2002 girls, playing a game with n cards. Initially, one girl holds
all the cards. In each turn, if at least one girl holds at least two cards, one of these girls
must pass a card to each of her two neighbours. The game ends when each girl is holding
at most one card.
(a) Prove that if n ≥ 2002, then the game cannot end.
(b) Prove that if n < 2002, then the game must end.
11
8-12-2001
m
63. Suppose {ai }i=n is a finite sequence of integers, and there is a Pprime p and some
m
k, n ≤ k ≤ m such that p|ak but p 6 |aj , n ≤ j ≤ m, j 6= k. Prove that i=n a1i cannot be
an integer.
Pm 1
64. Prove that for any choice of m and n, m, n > 1, i=n i cannot be a positive integer.
65. There is a tournament where 10 teams take part, and each pair of teams plays against
each other once and only once. Define a cycle {A, B, C} to be such that team A beats
team B, B beats C and C beats A. Two cycles {A, B, C} and {C, A, B} are considered the
same. Find the largest possible number of cycles after all the teams have played against
each other.
66. Consider the recursions xn+1 = 2xn + 3yn , yn+1 = xn + 2yn with x1 = 2, y1 = 1.
Show that for each integer n ≥ 1, there is a positive integer Kn such that
³ ´
2
x2n+1 = 2 Kn2 + (Kn + 1) .
67. Suppose that ABCD is a tetrahedron and its four altitudes AA1 , BB1 , CC1 , DD1
intersect at the point H. Let A2 , B2 , C2 be points on AA1 , BB1 , CC1 respectively such that
AA2 : A2 A1 = BB2 : B2 B1 = CC2 : C2 C1 = 2 : 1. Show that the points H, A2 , B2 , C2 , D1
are on a sphere.
12
15-12-2001
68. The nonnegative real numbers a, b, c, A, B, C and k satisfy a + A = b + B = c + C = k.

Prove that aB + bC + cA ≤ k 2 .
69. Find the least constant C such that the inequality
x1 x2 + x2 x3 + ... + x2000 x2001 + x2001 x1 ≤ C

P2001
holds for any i=1 xi = 2001, x1 , ..., x2001 ≥ 0. For this constant C, determine the
instances of equality.
70. Let D be a point inside an acute triangle ABC such that
DA · DB · AB + DB · DC · BC + DC · DA · CA = AB · BC · CA.
Determine the geometric position of D.
71. Consider the polynomial p(x) = x2001 + a1 x2000 + a2 x1999 + ... + a2000 x + 1. where
all the ai ’s are nonnegative. If the equation p(x) = 0 has 2001 real roots, prove that
p(2001) ≥ 20022001 .
72. Let a, b, c, d be nonnegative real numbers such that a + b + c + d = 1. Prove that

1
bcd + cda + dab + abc ≤ 27 + 176
27 abcd. Determine when equality holds.
73. A sequence {ak } satisfies the following conditions:
1
(a) a0 = 2001 ,
(b) ak+1 = ak + n1 a2k , k = 0, 1, 2, ..., n.
1
Prove that 1 − 2000n < 2000an < 1.
13
22-12-2001
74. Let ABC be an equilateral triangle. Draw the semicircle Γ which has BC as its
diameter, where Γ lies on the opposite side of BC as A. Show that the straight lines
drawn from A that trisect the line BC also trisect Γ when they are extended.
75. ABC is an equilateral triangle. P is the midpoint of arc AC of its circumcircle, and
M is an arbitrary point of the arc. N is the midpoint of BM and K is the foot of the
perpendicular from P to M C. Prove that AN K is an equilateral triangle.
76. Two concentric circles are given with a common centre O. From a point A on the
outer circle, two tangents to the inner circle are drawn, meeting the latter at D and E
respectively. AD and ED are extended to meet the outer circle at C and B respectively.
AB 2 BE
Show that ( BC ) = BD .
77. Ali Baba the carpet merchant has a rectangular piece of carpet whose dimensions
are unknown. Unfortunately, his tape measure is broken and he has no other measuring
instruments. However, he finds that if he lays it flat on the floor of either of his storerooms,
then each corner of the carpet touches a different wall of that room. He knows that the
sides of the carpet are integral numbers of feet, and that his two storerooms have the
same (unknown) length, but widths of 38 and 50 feet respectively. What are the carpet’s
dimensions?
78. Let I be the incentre of the non-isosceles triangle ABC. Let the incircle of ABC
touch the sides BC, CA and AB at the points A1 , B1 and C1 respectively. Prove that the
circumcentres of ∆AIA1 , ∆BIB1 and ∆CIC1 are collinear.
14
Solutions for 22-9-2001
1. Consider the set of integers A = {2a 3b 5c : 0 ≤ a, b, c ≤ 5}. Find the smallest number n
such that whenever S is a subset of A with n elements, you can find two numbers p, q in
A with p | q.
Soln. Identify a number 2a 3b 5c in A by its triple of indices abc. Thus 000 represents
20 30 50 while 213 represents 22 31 53 . Consider the following arrangement of numbers into
16 rows.
000
100
200 110
300 111 210 120
400 211 310 130 220
500 311 410 140 320 302 221
510 411 420 240 321 312 330 222
520 511 430 340 421 412 331 322
530 521 440 341 431 512 332 422
540 531 441 342 432 513 333 522
550 541 442 352 532 514 433
551 542 443 353 524
552 543 444 534
553 544
554
555
Turn the first column in 6 by considering the six permutations and turn each of the other
columns into 3 by considering the cyclic permutations. Note that (i) every number is one
of the rows, (ii) every two numbers in the same row do not divide each other (iii) for every
two numbers in the same column, one must divided the other. Since the row with the most
elements is 27, the answer is 28.
(Joel’s soln) The problem is equivalent to the following: Consider the ordered triples
of the form {(a, b, c) : 0 ≤ a, b, c ≤ 5}. Find the smallest number n such that if there are
n such triples, you can always find two (p, q, r) and (x, y, z) such that p ≥ x, q ≥ y, and
r ≥ z. If this is the case, we say that the two triples are comparable.
First we note that any two triples are comparable if their second and third elements
are the same. Let A be a set of pairwise noncomparable triples. Since there are 36 possible
combinations for (b, c), A has at most 36 triples. Examine the sequence of 11 triples in
order:
(a, 0, 0), (a, 0, 1), (a, 0, 2), (a, 0, 3), (a, 0, 4), (a, 0, 5), (a, 1, 5), (a, 2, 5), (a, 3, 5), (a, 4, 5), (a, 5, 5).
15
For these to be in A, the first elements have to be strictly decreasing. Thus at most 6
of these can be in A or at least 5 cannot be in A. Similarly, at least three of the following
triples
(a, 1, 0), (a, 2, 0), (a, 3, 0), (a, 4, 0), (a, 5, 0), (a, 5, 1), (a, 5, 2), (a, 5, 3), (a, 5, 4)
and at least 1 of the following triples:
(a, 1, 1), (a, 1, 2), (a, 1, 3), (a, 1, 4), (a, 2, 4), (a, 3, 4), (a, 4, 4)
are not in A. Thus A can have at most 27 triples and if 28 triples are chosen, then at least
two of them are comparable. Below is a collection of 27 pairwise noncomparable triples
(the sum of the 3 components is 7):
(0, 2, 5), (0, 3, 4), (0, 4, 3), (0, 5, 2), (1, 1, 5), (1, 2, 4), (1, 3, 3),
(1, 4, 2), (1, 5, 1), (2, 0, 5), (2, 1, 4), (2, 2, 3), (2, 3, 2), (2, 4, 1),
(2, 5, 0), (3, 0, 4), (3, 1, 3), (3, 2, 2), (3, 3, 1), (3, 4, 0), (4, 0, 3),
(4, 1, 2), (4, 2, 1), (4, 3, 0), (5, 0, 2), (5, 1, 1), (5, 2, 0)
So the answer is 28.
2. (Byelorrussian MO 95) “Words” are formed with the letters A and B. Using the words
x1 , x2 , . . ., xn we can form a new word if we write these words consecutively one next to
another: x1 x2 . . . xn . A word is called a palindrome, if it is not changed after rewriting its
letters in the reverse order. Prove that any word with 1995 letters A and B can be formed
with less than 800 palindromes.
Soln. Official solution: (The key idea is to find the longest word that can be formed using
at most 2 palindromes.) First of all, it is easy to check that any 5-letter word may be
formed with at most two palindromes. Indeed, (A and B are symmetric).
AAAAA = AAAAA, AAAAB = AAAA + B, AAABA = AA + ABA,
AAABB = AAA + BB, AABAA = AABAA, AABAB = AA + BAB,
AABBA = A + ABBA, AABBB = AA + BBB, ABAAA = ABA + AA,
ABAAB = A + BAAB, ABABA = ABABA, ABABB = ABA + BB,
ABBAA = ABBA + A, ABBAB = ABBA + B, ABBBA = ABBBA,
ABBBB = A + BBBB.
Let us consider an arbitrary word with 1995 letters and divide it into words with
5 letters each. Each of these 1995/5 = 399 words may be formed with at most two
palindromes. Thus any 1995-letter word may be formed with at most 399 × 2 = 798
palindromes.
3. (49th Romania National MO 1998) (10th Form) Let n ≥ 2 be an integer and M =

{1, 2, . . . , n}. For every k ∈ {1, 2, . . . , n − 1}, let
1 X
xk = (min A + max A).
n + 1 A⊂M
|A|=k
16
Prove that x1 , x2 , . . . , xn−1 are integers, not all divisible by 4.
¡ n−i ¢
Soln. For each i ∈ {1, 2, . . . , n}, there are k−1 subsets, each with k elements and contains
¡n¢ ¡ i−1 ¢
i as the minimum element. (Note that k = 0 if k > n.) Also there are k−1 subsets,
each with k elements and contains i as the maximum element. Thus
1 X
xk = (min A + max A)
n + 1 A⊂M
|A|=k
· µ ¶ µ ¶ µ ¶
1 n−1 n−2 k−1
= 1 +2 + · · · + (n + 1 − k)
n+1 k−1 k−1 k−1
µ ¶ µ ¶ µ ¶¸
n−1 n−2 k−1
+n + (n − 1) + ··· + k
k−1 k−1 k−1
µ ¶
n
=
k
¡ ¢ ¡ n ¢
Thus x1 , . . . , xn−1 are all integers. Since x1 + · · · + xn−1 = n1 + · · · + n−1 = 2n − 2, not
all of x1 , . . . , xn−1 are divisible by 4.
(2nd Solution) For each k element subset A = {a1 , . . . , ak } with ai < aj if i < j, with
a1 = 1 + p and ak = n − q, let B = {b1 , . . . , bk }, where bi = ai + q − p. We have ¡ ¢
min A + min B + max A + max B = 2n + 2. Since the number of k element subsets is nk ,
we have µ ¶ µ ¶
1 X 1 n 2n + 2 n
xk = (min A + max A) = = .
n + 1 A⊂M n+1 k 2 k
|A|=k
Here’s a sketch/hint for 3: For each set A, consider the set A0 :
A0 = {n + 1 − x|x ∈ A}
Then, min A = n + 1 − max A0 . Do the usual summation rewriting, and you can
compute xk explicitly to be some binomial function. The other bit follows readily too
using another standard trick.
4. (Byelorrussian MO 95) The lattice frame construction of 2 × 2 × 2 cube is formed with

54 metal shafts of length 1 (points of shafts’ connection are called junctions). An ant starts
from some junction A and creeps along the shafts in accordance with the following rule:
when the ant reaches the next junction it turns to a perpendicular shaft. At some moment
the ant reaches the initial junction A; there is no junction (except for A) where the ant
has been twice. What is the maximum length of the ant’s path?
Soln. Official solution: The maximum length of an ant’s path is equal to 24. First we
prove that the path along which the ant creeps, has at most 24 junctions of the shafts of
17
the cube frame. By the condition, any two consecutive shafts in the path (except possibly
for the first and the last shafts) are perpendicular. In particular, the ant’s path passes at
most one shaft on all but one edge of the cube. Thus there are at most 13 shafts along the
edges of the cube. However, each vertex in the path requires two shafts. Thus the path
misses at least two vertices of the cube. Hence the ant’s path passes through at most 25
junctions of the shafts. The ant’s path consists of an even number of junctions. This is
easily seen to be true by taking the starting point as the origin and the three mutually
perpendicular lines passing through it as the axes and assume that each shaft is of unit
length. Then each move by the ant causes exactly a change of one unit in one of the
coordinates. Thus the total number of moves is even. Thus the length of the path is at
most 24. A path of length 24 is shown in the figure.
•••••••••••••••••••••••••••••••••••••••••••••••• ••
••••• ••••••••
........................................................................................................................
...... . .
.... ... .... ... .... ...
•• • ••• ..... .
..... .... •
•• ..... .
..... ....
..... .
..... ....
•
••••••••••••••••••••••••••••••••••••••••••••••••• ••••• •
•••••• •••
..
.. ..
.. ..
..
......................................................................................................................................... ....
....... .... ....
.... ..
..... .... • ••• •
• .. ••• •
.
.... ..
..... .... ..
.
.... ..
..... ....
...
..
..
•• ••• ••• ••
.. ..
•••••••••••••••••••••••••••••••••••••••••••••••••
.
.... . . .
.... . . .
.... . ..
..
. . .
. ..
. . .
. ..
. . ...
................................................................................................................................. .
•
• .... .... •
••• ... ••• •• .... ... ... ....
.
.
...
..
..
• ••••••••••••••••••••••••••••••••••••••••••••••••••
•
.. . . . .
. . . .
. .
....
• .........................................................................................................................................
•
•
...
..
•••••• ..
... ............
.. . . .
.. .
.... . . .
....
• ...
•• ••• .... . .. .
. .
. .
. .
...
•••• •
. ... .
... ......... .... ... ......... ....
• .. .. ...... ... ..
••••••• •••
...
•
•
...
..
... ...... .
••
.. . ...
...............................................................................................................................................
•
. . ... ..
...
•
•
...
.. ......... ....
•• ••
..... . .
...
••
••• .
...
... ..
.. ..
. . .
.
.
.
... .
.. ..
.. ..
. . ..
..
• ... ..... .. .
••
••
....
. ...... ...
.
... ..... . .
.
.
. ...... ...
.
... ..... . ...
••••••••••••••••••••••••••••••••••••••••••••••••••• •••••
........ ... .
•...........................................................................................................................................
.. ... ... ..
. .
...
.
... ..
. .
..
.
...
..
..
...
••• •• ••
...
..
...
.. ••• ..
..
...
..
..
...
..
..
...
..
..
...
..
.
••• •••••••••••••••••••••••••••••••••••••••••••••••• •••
...
..
...
.. .. .. ................................................................................................................................
. . . . .
.
.
•••• ••
... ......... .. ..
•••• .... .. . . . .
...
•• .
... ........
..
.. .
... .
.... ........
..
..
•• •••••••••••••••••••••••••••••••••••••••••••••••••••••
.. .. ...... ..
...
• •
•
...
..
....... ..
.........................................................................................................................................
...... ......
••• ••• •••••••

..
... ..
••
•••••
. ... . ... .
.. ........ ..... .....
.... ......... .... .........
•• •• ... ......
. .. •
• •••••••• . .. . ..
•••••••••••••••••••••••••••••••••••••••••••
................................................................................................................................
5. Let n black and n white objects be placed on the circumference of a circle, and define
any set of m consecutive objects from this cyclic sequence to be an m-chain.
(a) Prove that for each natural number k ≤ n, there exists a chain of 2k consecutive
pieces on the circle of which exactly k are black.
(b) Prove that there are at least two such chains that are disjoint if
√
k ≤ 2n + 2 − 2
Soln. (a) Label the positions 1, 2, . . . , 2n in the clockwise direction. For each i, if the
object Pis i is black, let ai = 1. Otherwise, let ai = −1. Fix k ≤ n. Define the function
i+2k−1
g(i) = j=i aj . Then there is a chain of 2k consecutive pieces of which exactly k are
P2n
black iff g(i) = 0 for some i. Since i=1 g(i) = 0, and since g(i + 1) − g(i) = ±2, 0, such a
g(i) exists.
(b): Suppose on the contrary that there does not exist 2 disjoint chains. WLOG,
we may assume from (a) that g(2k) = 0 and g(2k + 1), . . . g(2n) are all nonzero, and
18
by the same continuity argument, they must all be of the same sign. WLOG, assume
g(2k + 1), . . . , g(2n) < 0 and thus g(2k), . . . , g(2n − 1) ≤ −2. Therefore, we have:
g(1) + . . . + g(2k) = −(g(2k + 1) + . . . + g(2n)) ≥ 2(2n − 2k)
Thus, the sum on the left is maximum when g(1) = 0, g(1) = 2, . . . , g(k) = g(k + 1) =
2(k − 1), . . . , g(2k) = 0, that is, g(1), . . . , g(k) forms an increasing arithmetic progression
with a difference of 2, and g(k + 1), . . . , g(2k) forms a decreasing arithmetic progression
with a difference of −2. This yields:
2(k − 1)2 ≥ 2(2n − 2k) =⇒ k 2 + 4k + 4 > 2n + 2
which yields the required result.
6. We are given 1999 rectangles with sides of integer not exceeding 1998. Prove that
among these 1999 rectangles there are rectangles, say A, B and C such that A will fit
inside B and B will fit inside C.
Soln. (T& T Spring 1989 Q6, modified slightly). We partition the given set of rectangles
into 999 suitably chosen (pairwise disjoint) subsets S1 , . . ., S999 . These are defined as
follows:
For each i ∈ {1, . . . , 999}, let Si be the set of rectangles which have the following
properties:
(a) the shorter side has length at least i,
(b) the longer side has length at most 1998 − i,
(c) either the shorter side has length i or the longer side has length 1998 − i.
Any three rectangles in Si can always be arranged with the desired inclusion property.
Thus the result follows by the pigeon hole principle.
7. We are given N lines (N > 1) in a plane, no two of which are parallel and no three
of which have a point in common. Prove that it is possible to assign, to each region of
the plane determined by these lines, a non-zero integer of absolute value not exceeding N ,
such that the sum of the integers on either side of any of the given lines is equal to 0.
Soln. (Tournament of Towns Spring 1989 Q5). First note the regions can be painted
in two colours so that two regions sharing a common side have different colours. This is
19
trivial for N = 1. Assume that it is true for N = k. When the (k + 1)st line is drawn we
simply reverse the colours on exactly one side of this line.
Assign to each region an integer whose magnitude is equal to the number of vertices
of that region. The sign is + is the region is one of the two colours and − if it is the other
colour. Let L be one of the given lines. Consider an arbitrary vertex on one side of L. If
this vertex is on L, then it contributes = 1 to one region and −1 to a neighbouring region.
If it is not on L, it contributes = 1 to two regions and −1 to two regions. Thus the sum
of the numbers of one side of L is 0.
8. (APMO 99) Let S be a set of 2n + 1 points in the plane such that no three are collinear
and no four concyclic. A circle will be called good if it has 3 points of S on its circumference,
n − 1 points in its interior and n − 1 in its exterior. Prove that the number of good circles
has the same parity as n.
Soln. For any two points A and B, let P1 , P2 , . . ., Pk be points on one side of the line
AB and Pk+1 , . . ., P2n−1 be points on the other side. We shall prove that the number of
good circles passing through A and B is odd. Let
½
∠APi B if i = 1, . . . , k
θi =
180◦ − ∠APi B if i = k + 1, . . . , 2n − 1
It is easy to see that Pj is in the interior of the circle ABPi , if and only if
½
θj > θi for 1 ≤ j ≤ k
.
θj < θi for k + 1 ≤ j ≤ 2n − 1
Arrange the points Pi in increasing order of their corresponding angles θi . Colour the
points Pi , i = 1, . . . , k, black and the points Pi , i = k + 1, . . . , 2n − 1, white. For any point
X (different from A and B), let BX be the number of black points less than X minus the
number of black points greater than X and WX be the corresponding difference for white
points. (Note that black points which are greater than X are interior points of the circle
ABX while the white points greater than X are exterior points.) Define DX = BX − WX .
From the forgoing discussion we know that 4ABX is good if and only if DX = 0. We call
such a point good . If X < Y are consecutive points, then DX = DY if X and Y are of
different colours. (It is easy to show that DY − DX = −2 if X and Y are both white and
DY − DX = 2 if X and Y are both black. But we do not need these.)
If all the points are of the same colour, there is only one good point, namely the
middle point among the Pi ’s.
Now we suppose that there are points of either colour. Then there is a pair of adjacent
points, say X, Y , with different colour. Since DX = DY , either both are good or both are
not good. Their removal also does not change the value of DZ for any other point Z. Thus
the removal of a pair of adjacent points of different colour does not change the parity of
20
the number of good points. Continue to remove such pairs until only points of the same
colour are left. When this happens there is only one good point. Thus the number of good
circles through A and B is odd.
Now let gAB be the number of good circles through A and P B. Since each good circle
contains exactly three points, i.e., three pairs of points. Then gAB = 3g where g is total
number of good circles. Since there are a total of n(2n + 1) terms in the sum, and each
term is odd, we have g ≡ n (mod 2).
19. For any positive real numbers a, b, c,
a b c 3
+ + ≥
b+c c+a a+b 2
Soln. Let x = a + b, y = a + c, z = b + c, then

µ ¶
a b c 1 x y x z y z 3
+ + ≥ + + + + + −3 ≥ .
b+c c+a a+b 2 y x z x z y 2
Soln 2.
· ¸
a b c
+ + [a(b + c) + b(c + a) + c(a + b)] ≥ (a + b + c)2 .
b+c c+a a+b
Thus
a b c (a + b + c)2
+ + ≥ .
b+c c+a a+b 2(ab + ac + bc)
Also
(a + b + c)2 = (a2 + b2 + c2 ) + 2(ab + bc + ac) ≥ 3(ab + bc + ac).
Thus
(a + b + c)2 3
≥ .
a(b + c) + b(c + a) + c(a + b) 2
Soln 3. Since the inequality is symmetric about a, b, c, we may assume that a ≥ b ≥ c > 0.
Then,
1 1 1
≥ ≥ .
b+c a+c a+b
21
By rearrangement inequality, we have
a b c a b c
+ + ≥ + + ,
b+c c+a a+b c+a a+b b+c
a b c a b c
+ + ≥ + + .
b+c c+a a+b a+b b+c c+a
The inequality follows by adding these two inequalities.
20. Let a, b, c be positive numbers such that a + b + c ≤ 3. Prove that
1 1 1 3
+ + ≥ .
1+a 1+b 1+c 2
Soln. Apply AM ≥ HM on the three numbers a + 1, b + 1, c + 1 we have
1 1 1 9 3
+ + ≥ ≥ .
1+a 1+b 1+c (1 + a) + (1 + b) + (1 + c) 2
Soln 2. Using Cauchy’s inequality and denoting the denominators by x, y, z:

³1 1 1´
+ + (x + y + z) ≥ 9.
x y z
Thus
1 1 1 9 3
+ + ≥ ≥ .
x y z 3+a+b+c 2
21. Prove that for any positive real numbers a, b, c,
a b c 1
+ + ≥ .
10b + 11c 10c + 11a 10a + 11b 7
Soln. We shall prove a more general inequality:
a b c 3
+ + ≥ ,
mb + nc mc + na ma + nb m+n
where m, n are positive real numbers.
Write the denominators as A, B, C respectively. By Chauchy-Schwarz inequality, we

have · ¸
a b c
+ + [aA + bB + cC] ≥ (a + b + c)2 .
A B C
22
Since aA + bB + cC = (m + n)(ab + bc + ca) and 3(ab + bc + ca) ≤ (a + b + c)2 , the result
follows.
Soln 2. The inequality can also be proved by clearing the denominator and moving all
terms to the left hand side. After some simplification, the resulting inequality to be proved
is equivalent to
770(a3 + b3 + c3 ) − 253(a2 b + b2 c + c2 a) − 510(a2 c + b2 a + c2 b) − 21abc ≥ 0.
By rearrangement inequality, we have a3 + b3 + c3 ≥ a2 b + b2 c + c2 a and a3 + b3 + c3 ≥

a2 c + b2 a + c2 b. Using these and a3 + b3 + c3 ≥ 3abc, the above inequality follows.
22. Prove that for any positive real numbers ai , i = 1, . . . 5,

5
X ai 1
≥
i=1
ai+1 + 2ai+2 + 3ai+3 + 4ai+4 2
where the subscripts are to be taken mod 5.
Soln. Use the same method as in the previous problem. The only extra thing that you
have to note is that X 1 X
a2i ≥ ai aj .
2
1≤i<j≤5
23. (SMO2001) Let n be a positive integer and let a1 , a2 , . . ., an be n positive real numbers
such that a1 + a2 + · · · + an = 1. Is it true that
a41 a42 a4n 1

2 2 + 2 2 + · · · + 2 2 ≥ ?
a1 + a2 a2 + a3 an + a1 2n
Soln. The answer is yes. First observe that

µ ¶ µ ¶
a41 a42 a4n a42 a43 a41
+ 2 + ··· + 2 − + 2 ··· + 2 = 0.
a21 + a22 a2 + a23 an + a21 a21 + a22 a2 + a23 an + a22
Thus
a41 a42 a4n
+ + · · · +
a21 + a22 a22 + a23 a2n + a21
µ 4 ¶
1 a1 + a42 a42 + a43 a4n + a41
= + 2 + ··· + 2
2 a21 + a22 a2 + a23 an + a21
1
≥ [(a21 + a22 ) + · · · + (a2n + a21 )]
4
1 1
= (a21 + · · · + a2n ) ≥
2 n
23
which completes the proof. The last inequality follows as
r
a21 + · · · + a2n a1 + · · · + an 1
≥ = .
n n 2n
Soln 2. Denote the denominators by A1 , A2 , . . . , An . Then

· 4 ¸
a1 a42 a4n
+ + ··· + [A1 + A2 + · · · + An ] ≥ (a21 + a22 + · · · + a2n )2 .
A1 A2 An
But
A1 + A2 + · · · + An = 2(a21 + · · · + a2n ).
Thus we only need to prove
1
a21 + a22 + · · · + a2n ≥ .
n
24. Let a, b, c, d be nonnegative real numbers such that ab + bc + cd + da = 1. Prove that

a3 b3 c3 d3 1
+ + + ≥ .
b+c+d a+c+d a+b+d a+b+c 3
Soln. (NT-ST 2001, IMO proposed 1991) By Cauchy-Schwarz Inequality,

a2 + b2 + c2 + d2 ≥ ab + bc + cd + da = 1.
Assume without loss of generality that a ≥ b ≥ c ≥ d ≥ 0. Then

1 1 1 1
≥ ≥ ≥ > 0.
b+c+d a+c+d a+b+d a+b+c
Using Chebyshev’s Inequality twice followed by AM ≥ GM , and writing x = b + c + d, y =
c + d + a, z = d + a + b, w = a + b + c, we have
µ ¶
a3 b3 c3 d3 a3 + b3 + c3 + d3 1 1 1 1
+ + + ≥ + + +
x y z w 4 x y z w
2 2 2 2
µ ¶
(a + b + c + d )(a + b + c + d) 1 1 1 1
≥ + + +
16 x y z w
µ ¶
a+b+c+d 1 1 1 1
≥ + + +
16 x y z w
µ ¶
(x + y + z + w) 1 1 1 1
≥ + + +
3 × 16 x y z w
1 µ ¶
4(xyzw) 4 1 1 1 1
≥ + + +
3 × 16 x y z w
1
4 × 4(xyzw) 4 1 1
≥ 1 ≥
3 × 16 (xyzw) 4 3
24
Soln 2. Let the denominators be A, B, C, D. Then
· 3 ¸
a d3
+ ··· + [aA + · · · + dD] ≥ (a2 + · · · + d2 )2 .
A D
Thus we need to prove

(a2 + · · · + d2 )2 1
≥ .
2 + 2ac + 2bd 3
Since 2ac ≤ a2 + c2 and 2bd ≤ b2 + d2 , we need to prove
x2 1
≥ , or 3x2 − x − 2 = (3x + 2)(x − 1) ≥ 0
2+x 3
which is obviously true since x = a2 + · · · + d2 ≥ 1.
25. (IMO 95) Let a, b and c be positive real numbers such that abc = 1. Prove that
1 1 1 3
+ 3 + 3 ≥ .
a3 (b+ c) b (a + b) c (a + b) 2
Soln. Let S be the right hand side and T = a(b + c) + b(a + c) + c(a + b) = 2(ab + ac + bc) ≥
6(abc)2/3 = 6. Then by Cauchy’s inequality, we have
µ ¶2
1 1 1 T2
ST ≥ + + = .
a b c 4
Thus S ≥ T /4 ≥ 3/2 as desired.
Second Solution Let x = 1/a. y = 1/b, z = 1/c. Then xyz = a and
x2 y2 z2
S= + + .
y+z x+z x+y
By Cauchy’s inequality,
[(x + y) + (z + x) + (x + y)]S ≥ (x + y + z)2

(x+y+z)
or S ≥ 2 ≥ 32 (xyz)1/3 = 3
2 as desired.
26. A sequence of natural numbers {an } is defined by a1 = 1, a2 = 3 and
an = (n + 1)an−1 − nan−2 (n ≥ 2).
25
Find all values of n such that 11|an .
Soln. Rearranging the recurrence relation, we obtain: an − an−1 = n(an−1 − an−2 ).

Substituting bn = an − an−1 , we find that bn = nbn−1 = n(n − 1)bn−2 = . . . = n!. Hence,
an = 1! + 2! + . . . + n!. One can easily verify that 11|a4 , a8 , a10 . Since 11|a10 and 11|i! for
all i ≥ 11, this implies that 11|an for all n ≥ 10. So the required answers are: n = 4, 8 and
n ≥ 10.
27. Let{xn }, n ∈ N be a sequence of numbers satisfying the condition

p
−xn + 3 − 3x2n
|x1 | < 1, xn+1 = , (n ≥ 1).
2
(a) What other condition does x1 need to satisfy so that all the numbers of the
sequence are positive?
(b) Is the given sequence periodic?
Soln.
√
(a) For the sequence to be positive, we require 3 − 3x2n ≥ x2n , ie that |xn | ≤ 23 . One
√
can easily check by plugging into the recurrence relation that when |xn | ≤ 23 ,
√
then |xn+1 | ≤ 23 also, so the entire sequence will be positive. Hence, the required
√
condition is |x1 | ≤ 23 .
(b) Substitute xn = sin θn into the

√
recurrence relation. After some simplification, we
find that xn+1 = − 2 sin θn + 2 cos θn = sin( π3 − θn ). Hence, xn+2 = xn and the
1 3
sequence is periodic.
28. Suppose that a function f defined on the positive integers satisfies f (1) = 1, f (2) = 2,
and
f (n + 2) = f (n + 2 − f (n + 1)) + f (n + 1 − f (n)), (n ≥ 1).
(a) Show that 0 ≤ f (n + 1) − f (n) ≤ 1.
(b) Show that if f (n) is odd, then f (n + 1) = f (n) + 1.
(c) Determine, with justification, all values of n for which f (n) = 210 + 1.
Soln. The key is to show that given n = 2k + m, where 0 ≤ m ≤ 2k − 2, then f (n) =

f (m + 1) + 2k−1 . Also, if n = 2k + 2k − 1 = 2k+1 − 1, then f (n) = 2k . This can be done
using strong induction on k. Notice that this helps us to express all the values of f (n) from
n = 2k to 2k+1 , in terms of the values of f (n) from n = 0 to n = 2k − 1. (We can define
26
f (0) = 0 for convenience; this definition is consistent with the recurrence relation.) Once
we have established this claim, parts (a) and (b) follow essentially by induction. Also,
the answer to (c) is n = 211 . (Endnote: The key to solving this question would be to
list out the values and try to find a pattern, which can then be rigorously justified using
induction.)
29. Determine the number of all sequences {x1 , x2 , . . . , xn }, with xi ∈ {a, b, c} for i =
1, 2, . . . , n that satisfy x1 = xn = a and xi 6= xi+1 for i = 1, 2, . . . , n − 1.
Soln. Let the number of sequences for which x1 = xn = a be An . Define Bn to be the

number of sequences with x1 = a, xn = b, and let Cn be the number of sequences with
x1 = a, xn = c. Notice by symmetry that Bn = Cn (just simply switch the b’s and c’s in the
sequence). Notice that An = Bn−1 + Cn−1 = 2Bn−1 , since any sequence of length n that
ends with a is generated by a sequence of length n − 1 which ends with b or c. (Conversely,
given a sequence of length n − 1 that ends with b or c, we can simply append a to obtain
a sequence of length n which ends with a.) Similarly, Bn = An−1 + Cn−1 = An−1 + Bn−1 ,
since any sequence of length n that ends with b must be generated by a sequence of length
n − 1 generated with a or c. Having established this simultaneous system of recurrence
relations, we substitute and solve for Bn = Bn−1 + 2Bn−2 , B1 = 0, B2 = 1. Solving, we
obtain that An = 2Bn−1 = 32 [ 2n−2 + (−1)n−1 ] .
30. Given is a prime p > 3. Set q = p3 . Define the sequence {an } by:
½
n for n = 0, 1, 2, . . . , p − 1,
an =
an−1 + an−p for n > p − 1.
Determine the remainder when aq is divided by p.
Soln. (This remains an open question. I suspect the key is to prove that ap2 −1 ≡ a0 (mod
p), from which it follows that ap3 ≡ p − 1 (mod p).)
31. A and B are two candidates taking part in an election. Assume that A receives m
votes and B receives n votes, where m, n ∈ N and m > n. Find the number of ways in
which the ballots can be arranged in order that when they are counted, one at a time, the
number of votes for A will always be more than that for B at any time during the counting
process.
Soln. We model the vote counting process on the Cartesian plane. Starting at (0, 0), we
move to (1, 1) if the first vote goes to A, else we move to (1, −1) if the first vote goes to B
instead. Notice that this walk will end at (m + n, m − n). The number of ways of counting
in which the number of votes A gets is always more than that which B has is equal to the
number of paths on the plane from (0, 0) to (m+n, m−n) that do not cross the x-axis. We
count the complement of this set, ie the number of paths that do cross the x-axis. Notice
that the first vote must go to A. Starting from (1, 1), the number of paths that cross the
27
x-axis and end at (m+n, m−n) is equal to the number of paths (unrestricted) from (1, −1)
to (m + n, m − n). To see this, take any given path from (1, 1) to (m + n, m − n) and let
X be the first point of intersection with the x-axis. Reflect the portion of this path to the
left of X about the x-axis and we have a path from (1, −1) to (m + n, m − n). Vice versa,
we can take any such path from (1, −1) to (m + n, m¡− n) and ¢ reflect it back uniquely. The
number of paths from (1, −1) to (m + n, m − n) is m+n−1 m , since you need to choose m
steps upwards out¡ of a possible
¢ m + n − 1. Similarly, the number of paths from (1, 1) to
m+n−1
(m + n, m − n) is m−1 . Hence, the number of paths from (0, 0) to (m + n, m − n) that
¡ ¢ ¡m+n−1¢ m−n ¡m+n¢
do not cross the x-axis is m+n−1
m−1 − m = m+n m .
32. (Auckland MO 98) Find all prime numbers p for which the number p2 + 11 has exactly
6 different divisors (including 1 and the number itself.)
Soln. If p = 3, p2 + 11 has exactly 6 divisors. Now let p > 3. Then p is odd and p2 ≡ 1
(mod 4), thus p2 + 11 ≡ 0 (mod 4). Also p2 ≡ 1 (mod 3), thus p2 + 11 ≡ 0 (mod 3).
Thus p2 + 11 ≡ 0 (mod 12). Since every p2 + 11 > 12 and every divisor of 12 is also a
divisor of p2 + 11, it follows that p2 + 11 has more divisors than 12. Thus p2 + 11 has more
than 6 divisors. The only prime number with the desired property is therefore 3.
33. (IMO 98) Determine all pairs (a, b) of positive integers such that ab2 + b + 7 divides
a2 b + a + b.
Soln. Since ab2 + b + 7|b(a2 b + a + b) and a2 b2 + ab + b2 = a(ab2 + b + 7) + (b2 − 7a), we

have either b2 − 7a = 0 or b2 − 7a is a multiple of ab2 + b + 7. The former implies that
b = 7t and a = 7t2 . Indeed these are solutions for all positive t.
For the second case, we note that b2 − 7a < ab2 + b + 7. Thus b2 − 7a < 0. If b ≥ 3,
then ab2 + b + 7 > 7a − b2 . Thus for ab2 + b + 7 to divide 7a − b2 , b = 1, 2. The case b = 1
requires that 7a − 1 be divisible by a + 8. The quotients are less than 7. Testing each of
the possibilities yields a = 49, 11. These are indeed solutions.
The case b = 2 requires that 7a − 4 be divisible by 4a + 11. The quotient has to be 1

and this is clearly impossible.
34. Let p be an odd prime. Prove that

µ ¶p−2
p−2 p−2 p−2 p−1 2 − 2p
1 +2 +3 + ··· + ≡ (mod p).
2 p
28
Soln. All the congruences are taken mod p. For each 0 < i < p, we have ip−1 ≡ 1. Thus
ip−2 ≡ 1i . So
µ ¶p−2
p−2 p−2 p−2 p−1 1 1 1
1 +2 +3 + ··· + ≡ + + ··· + .
2 1 2 (p − 1)/2
On the other hand

p−1 µ ¶ p−1 p−1 p−1 (p−1)/2
2 − 2p 2 − (1 + 1)p 1X p X
i1
X 1 X 1 X 1
= = − ≡ (−1) ≡ − +2 ≡ .
p p p i=1 i i=1
i i=1
i i=1
2i i=1
i
1 1
The last congruence follows because i + p−i ≡ 0.
Soln 2. First, for each i = 1, 2, · · · , p−1

2 ,
µ ¶
2i p (p − 1)(p − 2) · · · (p − (2i − 1)) (−1)(−2) · · · (−(2i − 1))
= ≡ ≡ −1 (mod p).
p 2i (2i − 1)! (2i − 1)!
Hence
(p−1)/2
X (p−1)/2
X µ ¶
p−2 p−2 2i p
i ≡− i
i=1 i=1
p 2i
(p−1)/2 µ ¶
2 X p−1 p
≡− i
p i=1 2i
(p−1)/2 µ ¶
2 X p
≡− (mod p) by Fermat’s Theorem.
p i=1 2i
The last summation counts the even-sized nonempty subsets of a p-element set, of which
there are 2p−1 − 1.
35. (Ukrainian MO 97) (10th grade) Let d(n) denote the greatest odd divisor of the natural
number n. Define the function f : N → N by f (2n − 1) = 2n , f (2n) = n + 2n/d(n) for all
n ∈ N. Find all k such that f (f (. . . (1) . . .)) = 1997 where f is iterated k times.
Soln. Let a1 = 1 and an+1 = f (an ) for n ≥ 1. Then a1 = 1, a2 = 2, a3 = 3 and so on.

After a few more terms, it is easy to notice that if am = 2j , then am+j+1 = 2j+1 . This
can be proved as follows. Let am = 2j . Then
am+1 = 2j−1 + 2j = 3 · 2j−1 ;

am+2 = 3 · 2j−2 + 2j−1 = 5 · 2j−2 ;
Thus we can ofrmulate the following induction hypothesis:
am+i = (2i + 1)2j−i , i = 0, 1, . . . , j.
29
am+i+1 = (2i + 1)2j−i−1 + 2j−i = (2(i + 1) + 1)2j−i−1 .
Thus the result follows by induction. Since a1 = 20 , we can find write down a formula
for an as follows. If n = (0 + 1 + 2 + · · · + p) + 1 + q, where p ≥ 0, 0 ≤ q ≤ p, then
an = (2q + 1)2p−q . If (2q + 1)2p−q = 1997, then p = q = 998. Thus a499500 = 1997. Thus
k = 499499.
36. (Byelorrussian MO 95) Given three real numbers such that the sum of any two of
them is not equal to 1, prove that there are two numbers x and y such that xy/(x + y − 1)
does not belong to the interval (0, 1).
Soln. Official solution: Let a, b, c be the given numbers. Suppose that each of the numbers
ab ac bc
A= , B= , C= .
a+b−1 a+c−1 b+c−1
belongs to (0, 1). Then A > 0, B > 0, C > 0 and
a2 b2 c2
ABC = > 0.
(a + b − 1)(a + c − 1)(b + c − 1)
Hence
D = (a + b − 1)(a + c − 1)(b + c − 1) > 0. (∗)
On the other hand, A − 1 < 0, B − 1 < 0, C − 1 < 0. Thus (A − 1)(B − 1)(C − 1) < 0. It
is easy to verify that
(a − 1)(b − 1) (a − 1)(c − 1) (b − 1)(b − 1)

A−1= , B−1= , C −1= .
a+b−1 a+c−1 b+c−1
Consequently,
(a − 1)2 (b − 1)2 (c − 1)2

(A − 1)(B − 1)(C − 1) = <0
(a + b − 1)(a + c − 1)(b + c − 1)
contradicting (∗). This proves the statement.
52. Determine all α ∈ R such that there exists a nonconstant function f : R → R such
that
f (α(x + y)) = f (x) + f (y).
αx
Soln. For α = 1, f (x) = x satisfies the functional equation. For α 6= 1, let y = 1−α , but
this implies f (y) = f (x) + f (y), hence f (x) = 0.
30
53. Let f : N → N be a function satisfying
(a) For every n ∈ N, f (n + f (n)) = f (n).
(b) For some n0 ∈ N, f (n0 ) = 1.
Show that f (n) = 1 for all n ∈ N.
Soln. Note that f (n0 + 1) = f (n0 + f (n0 )) = f (n0 ) = 1. Hence f (n) = 1 for all n ≥ n0 .
Let S = {n ∈ N|f (n) 6= 1}. If S = ∅, then we are done. So suppose S 6= ∅, let
M = supS = maxS, since S is a finite set. Then f (M + f (M )) = f (M ) 6= 1. Hence
M + f (M ) ∈ S, but M + f (M ) > M , contradicting the maximality of M . Hence S = ∅,
f (n) = 1 for all n ∈ N
54. Suppose that f : N → N satisfies f (1) = 1 and for all n,
(a) 3f (n)f (2n + 1) = f (2n)(1 + 3f (n)),
(b) f (2n) < 6f (n).
Find all (k, m) such that f (k) + f (m) = 2001.
Soln. Note that gcd(3f (n), 1 + 3f (n)) = 1, hence 3f (n)|f (2n). But f (2n) < 6f (n),
hence f (2n) = 3f (n), f (2n + 1) = 3f (n) + 1.Claim: If n = (b1 b2 · · · bn )2 , then f (n) =
(b1 b2 · · · bn )3 .This is a simple exercise in induction (induct on n, consider n odd and n
even separately) and is left to the reader. Then to find integers k, m such that f (k) +
f (m) = 2001, let us first write 2001 in its tenary representation. Now 2001 = 22020103 .
We note that the tenary representations of f (k), f (m) can only contain 0 and 1’s, hence
there are also no carry-overs from the addition. Hence the possible pairs of k, m are only
(11010002 , 11010102 ) and (11010102 , 11010002 ), i.e. (104, 106) and (106, 104).
55. Let f : R → R be a function such that for all x, y ∈ R,

f (x3 + y 3 ) = (x + y)((f (x))2 − f (x)f (y) + (f (y))2 ).
Prove that for all x ∈ R, f (2001x) = 2001f (x).
Soln. Let x = y = 0 and we have f (0) = 0. Setting y = 0, we have f (x3 ) = x(f (x))2 ,
or equivalently, f (x) = x1/3 f (x1/3 )2 . In particular, x and f (x) have the same sign.Let S
be the set S = {a > 0 : f (ax) = af (x)∀x ∈ <}.Clearly 1 ∈ S, and we will show a1/3 ∈ S
whenever a ∈ S. In fact, axf (x)2 = af (x3 ) = f (ax3 ) = f ((a1/3 x)3 ) = a1/3 xf (a1/3 x)2 .
Hence (a1/3 f (x))2 = f (a1/3 x)2 . Since x and f (x) have the same sign, we conclude that
f (a1/3 x) = a1/3 f (x).Now we show that if a, b ∈ S, then a + b ∈ S. Indeed, we have
f ((a + b)x) = f ((a1/3 x1/3 )3 + (b1/3 x1/3 )3 )
= (a1/3 + b1/3 )x1/3 (f (a1/3 x1/3 )2 − f (a1/3 x1/3 )f (b1/3 x1/3 ) + f (b1/3 x1/3 )2 )
= (a1/3 + b1/3 )(a2/3 − a1/3 b1/3 + b2/3 )x1/3 f (x1/3 )2
= (a + b)f (x).
31
Hence in particular, we have f (2001x) = 2001f (x).
56. Find all functions f : N → N with the property that for all n ∈ N,
1 1 1 f (f (n))
+ + ··· + = .
f (1)f (2) f (2)f (3) f (n)f (n + 1) f (n + 1)
Soln. Let n = 1 and we have f (f (1))f (1) = 1; hence f (1) = 1. Replacing the equality for
n into one for n + 1 we obtain
f (f (n)) 1 f (f (n + 1))
+ = .
f (n + 1) f (n + 1)f (n + 2) f (n + 2)
This is equivalent to
f (f (n))f (n + 2) + 1 = f (f (n + 1))f (n + 1).
Note that f (n + 1) = 1 implies f (f (n + 1)) = 1, hence f (f (n))f (n + 2) = 0, which is

impossible. Hence f (n) > 1 for n > 1.We use induction to show that f (f (n)) < f (n + 1).
The inequality is true for n = 1, since f (2) > 1 = f (f (1)). Also, if f (n + 1) > f (f (n)),
then f (n + 1) ≥ f (f (n)) + 1. Hence
f (f (n))f (n + 2) + 1 ≥ f (f (n + 1))f (f (n)) + f (f (n + 1)).
Since n+1 > 1, f (n+1) > 1 and thus f (f (n+1)) > 1, which implies that f (n+2) > f (f (n+
1)).The problem is now reduced to finding all functions f such that f (f (n)) < f (n + 1).
f has a unique minimum at n = 1, for if n > 1, we have f (n) > f (f (n − 1)). By the same
reasoning, we see that the second smallest value is f (2), etc. Hence
f (1) < f (2) < f (3) < · · · .
Now since f (1) ≥ 1, we have f (n) ≥ n. Now suppose for some natural number n, we have
f (n) > n. Then f (n) ≥ n + 1. Since f is increasing, f (f (n)) ≥ f (n + 1), a contradiction.
Hence f (n) = n for all n ∈ N.
57. Define f : N → N and g : N → Z such that
(a) f (x, x) = x,
(b) f (x, y) = f (y, x),
(c) f (x, y) = f (x, x + y),
(d) g(2001) = 2002,
32
(e) g(xy) = g(x) + g(y) + mg(f (x, y)).
Determine all integers m for which g exists.
Soln. We claim that f (x, y) = gcd(x, y). This is a simple exercise in induction. (induct
on the sum x + y).Now let a = b and we have g(a2 ) = (m + 2)g(a). Applying this again
we get g(a4 ) = (m + 2)g(a2 ) = (m + 2)2 g(a).On the other hand,
g(a4 ) = g(a) + g(a3 ) + mg(a)

= (m + 1)g(a) + g(a3 )
= (m + 1)g(a) + g(a) + g(a2 ) + mg(a)
= (2m + 2)g(a) + g(a2 ) = (3m + 4)g(a).
Now let a = 2001, and we have (m + 2)2 = 3m + 4 and hence m = 0, −1. For m = 0, an
example is given by
g(pα αn
1 · · · pn ) = α1 h(p1 ) + · · · + αn h(pn ),
1
where k is a prime factor of 2001, h(3) = 2002 and h(p) = 0 for all primes p 6= 3. For
m = −1, an example is given by
g(pα αn
1 · · · pn ) = h(p1 ) + · · · + h(pn ).
1
It is easy to check that these functions do indeed satisfy the given requirements.

m
63. Suppose {ai }i=n is a finite sequence of integers, and there is a Pprime p and some
m
k, n ≤ k ≤ m such that p|ak but p 6 |aj , n ≤ j ≤ m, j 6= k. Prove that i=n a1i cannot be
an integer.
Pm
Soln. Suppose i=n a1i = L is an integer. Then after finding common denominator and
multiplying both sides by an · · · am , we have
Lan an+1 · · · am = an+1 an+2 · · · am + · · · + an · · · b

ak · · · am + · · · + an+1 an+2 · · · am−1 .
But p divides the left hand

Pmside and all except the kth term of the right hand side. This
contradiction shows that i=n a1i cannot be an integer.
Pm 1
64. Prove that for any choice of m and n, m, n > 1, i=n i cannot be a positive integer.
Soln.
Pm 1Clearly,
Pm it 1is meaningful to consider the cases where m > n. If ¥mm ¦≤ 2n, then
i=n i < i=n n = 1. If m > 2n, then therePis a prime between m and 2 , which we
m
call p. From question 1, we can conclude that i=n 1i can never be an integer.
33
65. There is a tournament where 10 teams take part, and each pair of teams plays against
each other once and only once. Define a cycle {A, B, C} to be such that team A beats
team B, B beats C and C beats A. Two cycles {A, B, C} and {C, A, B} are considered the
same. Find the largest possible number of cycles after all the teams have played against
each other.
Soln. We label the teams T1 , T2 , ..., T10 . Let the number of times Ti wins be wi and the
number of losses be li . Instead of looking at the cycles as it is, we first look at the total
number of 3-combinations of teams and the number of 3-combinations which cannot be
rearranged to form a cycle. The total number of 3-combinations is 10 C3 = 120. Next,
we look at the 3-combinations that can be made to form cycles and those which cannot
(from here on, we call these non-cycles). The non-cycles consists of a single team beating
the other two teams and a single team losing to the other 2 teams, and if a 3-combination
consists of a single team beating the other two teams or a single team losing to the other
2 teams, it is a non-cycle. We can then count the non-cycles in a different manner. By
focusing
P10 ¡on ¢this¡lproperty
¢ of non-cycles, we can deduce that the number of non-cycles
wi
is + i
. Clearly, we have wi + li = 9, and rearranging the formula gives
P10 i=1
¡wi ¢ 2 ¡li ¢ 2 P10 ¡ 2 ¢
i=1 2 + 2 = i=1 wi − 9wi + 36 . And the number of non-cycles and the number
of cycles sum up to 120. We aim to minimise the number of non-cycles. This can be
achieved if wi = 4 or 5, and is certainly attainable. (The answer to this question is 40.)
The description of such a case is:
(i) If Ta and Tb are such a and b have the same parity, and a > b, then Ta beats Tb .
(ii) If Ta and Tb are such a and b have different parity, and a > b, then Ta loses to Tb .
66. Consider the recursions xn+1 = 2xn + 3yn , yn+1 = xn + 2yn with x1 = 2, y1 = 1.
Show that for each integer n ≥ 1, there is a positive integer Kn such that
³ ´
2
x2n+1 = 2 Kn2 + (Kn + 1) .
Soln. Now, after some rearrangement, we have

³ ´
2 2 2
x2n+1 = 2 Kn + (Kn + 1) = (2Kn + 1) + 1.
Therefore, it suffices to show that x2n+1 − 1 is an odd perfect square. Expressing xn+2
and yn+2 in terms of xn and yn , we have xn+2 = 7xn + 12yn and yn+2 = 4xn + 7xn .
As we try to express xn−2 in terms of xn and yn , we have xn−2 = 7xn − 12yn . Hence
xn+2 = 14xn − xn−2 .
We define a1 = 1 and a3 = 5 and an+4 = 4an+2 − an . We can prove by induction that

xn − a2n = 1 for all odd integers n. Then we can proceed to prove by induction that an are
all odd for all odd n. Therefore, xn − 1 = a2n and we are done.
34
To elaborate certain details. With xn+2 = 14xn − xn−2 , x1 = 2 and x3 = 26, we can
solve for xn to get:
1³ √ ´³ √ ń 1 ³ √ ´³ √ ń
x2n+1 = 2+ 3 7+4 3 + 2− 3 7−4 3
2 2
1³ √ ´³ √ ´2n 1 ³ √ ´³ √ ´2n
= 2+ 3 2+ 3 + 2− 3 2− 3
2 2
1 ³ √ ´ ³ √ ´ 2n 1³ √ ´³ √ ´2n
= 4+2 3 2+ 3 + 4−2 3 2− 3 −1+1
4 4
µ ³ ¶ µ ³ ¶
1 √ ´ 2³ √ ´2n 1 √ ´ 2³ √ ´2n
= 1+ 3 2+ 3 + 1− 3 2− 3 −1+1
2 2
µ ³ ¶ µ ³ ¶
1 √ ´³ √ ń 2 1 √ ´³ √ ń 2
= 1+ 3 2+ 3 + 1− 3 2− 3 −1+1
2 2
µ ³ ¶
1 √ ´³ √ ń 1 ³ √ ´³ √ ń 2
= 1+ 3 2+ 3 + 1− 3 2− 3 +1
2 2
67. Suppose that ABCD is a tetrahedron and its four altitudes AA1 , BB1 , CC1 , DD1
intersect at the point H. Let A2 , B2 , C2 be points on AA1 , BB1 , CC1 respectively such that
AA2 : A2 A1 = BB2 : B2 B1 = CC2 : C2 C1 = 2 : 1. Show that the points H, A2 , B2 , C2 , D1
are on a sphere.
Soln. Let G be the centroid of the triangle ABC. Prove that ∠HA2 G = ∠HB2 G =
∠HC2 G = ∠HD1 G = 90o . Thus, The centre of the sphere that we are looking for is at
the midpoint of H and G.
35
IMO Winter and Summer
Camp Trainings Test
Winter Camp 2008 Buffet Contest
Saturday, January 5, 2008
List of problems
A1. Find all functions f : R → R such that for all real numbers x and y,
f (xf (y) + x) = xy + f (x).
A2. Let x, y, z be positive real numbers. Prove that

x y z
p + p + p ≤ 1.
x + (x + y)(x + z) y + (y + z)(y + x) z + (z + x)(z + y)
A3. Let p(x) be a polynomial with integer coefficients. Does there always exist a positive integer k such
that p(x) − k is irreducible?
(An integer polynomial is irreducible if it cannot be written as a product of two nonconstant integer
polynomials.)
.....................................................................................................
C1. Let X be a finite set of positive integers and A a subset of X. Prove that there exists a subset B of X
such that A equals the set of elements of X which divide an odd number of elements of B.
C2. Let B be a set of more than 2n+1 /n distinct points with coordinates of the form (±1, ±1, . . . , ±1) in
n-dimensional space with n ≥ 3. Show that there are three distinct points in B which are the vertices
of an equilateral triangle.
C3. Let S be a set of n points on a plane, no three collinear. A subset of these points is called polite if they
are the vertices of a convex polygon with no points of S in the interior. Let ck denote the number of
polite sets with k points. Show that the sum
n
X
(−1)i ci
i=3
depends only on n and not on the configuration of the points.

.....................................................................................................
G1. Let ABC be an acute triangle. The points M and N are taken on the sides AB and AC, respectively.
The circles with diameters BN and CM intersect at points P and Q respectively. Prove that P, Q and
the orthocenter H are collinear.
G2. Let ABC be a triangle with AC 6= AB, and select point B1 on ray AC such that AB = AB1 . Let ω be
the circle passing through C, B1 , and the foot of the internal bisector of angle CAB. Let ω intersect
the circumcircle of triangle ABC again at Q. Prove that AC is parallel to the tangent to ω at Q.
G3. Let OAB and OCD be two directly similar triangles (i.e., CD can be obtained from AB by some
rotation and dilatation both centered at O). Suppose their incircles meet at E and F . Prove that
∠AOE = ∠DOF .
.....................................................................................................
N1. Let n > 1 be an odd integer. Prove that n does not divide 3n + 1.
N2. Let S be a finite set of integers, each greater than 1. Suppose that for each integer n there is some
s ∈ S such that gcd(s, n) = 1 or gcd(s, n) = s. Show that there exist s, t ∈ S such that gcd(s, t) is
prime.
N3. Let a positive integer k be given. Prove that there are infinitely many pairs of integers (a, b) with
|a| > 1 and |b| > 1 such that ab + a + b divides a2 + b2 + k.
Winter Camp 2008 Buffet Contest
Saturday, January 5, 2008
Solutions
A1. Find all functions f : R → R such that for all real numbers x and y,
f (xf (y) + x) = xy + f (x).
Solution: Putting x = 1, y = −1 − f (1) and letting a = f (y) + 1, we get
f (a) = f (f (y) + 1) = y + f (1) = −1.
Putting y = a and letting b = f (0), we get
b = f (xf (a) + x) = ax + f (x),
so f (x) = −ax + b. Putting this into the equation, we have
a2 xy − abx − ax + b = xy − ax + b.
Equating coefficients, we get a = ±1 and b = 0, so f (x) = x or f (x) = −x. We can easily check both
are solutions.
A2. Let x, y, z be positive real numbers. Prove that
x y z
p + p + p ≤ 1.
x+ (x + y)(x + z) y+ (y + z)(y + x) z+ (z + x)(z + y)
√ √
Solution: By Cauchy, we have (x + y)(x + z) ≥ ( xy + xz)2 . Hence,
√
X x X x X x
p ≤ √ √ = √ √ √ = 1.
cyc x+ (x + y)(x + z) cyc
x + xy + xz cyc
x + y+ z
A3. Let p(x) be a polynomial with integer coefficients. Does there always exist a positive integer k such
that p(x) − k is irreducible?
(An integer polynomial is irreducible if it cannot be written as a product of two nonconstant integer
polynomials.)
Solution: Yes. Choose k such that the constant term of p(x)−k is −q, where q is some large prime (to
be specified later). Now, suppose p(x) − k = f (x)g(x) for some nonconstant integer polynomials f, g.
Since q is prime, looking at the constant term, we must have that either |f (0)| = 1 or |g(0)| = 1. Assume
without loss of generality |f (0)| = 1. Then if r1 , . . . , rk are the roots of f , we have |r1 · · · rk | = 1/|a| ≤ 1,
where a is the leading coefficient of f . Therefore, there is some ri such that |ri | ≤ 1.
Consequently, we must have that p(ri ) − k = 0. Let p(x) − k = an xn + an−1 xn−1 + · · · + a1 x − q, then
n
n n
X X X
k
q= ak ri ≤ |ak ||ri |k ≤ |ak |.

k=1 k=1 k=1
Pn
Therefore, if we pick k large enough so that q > k=1 |ak |, then we have a contradiction, which means
that p(x) − k is irreducible.
Source: MOP 2007
.....................................................................................................
1
C1. Let X be a finite set of positive integers and A a subset of X. Prove that there exists a subset B of X
such that A equals the set of elements of X which divide an odd number of elements of B.
First solution: We construct B in stages. Set B = ∅ and consider every number in X, starting
with the largest and going down. For each element x ∈ X, see whether it divides the correct parity of
elements in B. (That is, if x ∈ A, x divides an odd number of elements in B; if x ∈ X − A, x divides an
even number of elements in B.) If it does not, add it to B. Thus the first element added to B is the
largest element of A. Now, this procedure will not change the divisibility condition for any element
greater than x, and will fulfill the condition for x. Thus when all elements of X have been examined,
the divisibility conditions will be satisfied by all elements of X, and B will be as desired.
Second solution: Given B a subset of X, it is clear that there is a unique A such that A equals the
set of elements of X which divide an odd number of elements of B. Now, given two distinct subsets,
B1 , B2 ⊆ X, let us show that the corresponding subsets A1 , A2 ⊆ X are distinct too. If B1 and B2
are not disjoint, then we can simply replace B1 by B1 − B1 ∩ B2 and replace B2 by B2 − B1 ∩ B2 ,
as this would not change the distinctness of the two resulting A1 , A2 . So we assume that B1 and B2
are disjoint. Let n be the largest element in B1 ∪ B2 . Say that n ∈ B1 . Then n ∈ A1 but n ∈/ A2 . It
follows that B1 6= B2 implies A1 6= A2 .
It follows that the map sending B to A is a bijection from the subsets of X to itself. Thus for every
A, we can find a corresponding B.
Source: 102 Combinatorics Problems
C2. Let B be a set of more than 2n+1 /n distinct points with coordinates of the form (±1, ±1, . . . , ±1) in
n-dimensional space with n ≥ 3. Show that there are three distinct points in B which are the vertices
of an equilateral triangle.
Solution: Let S be the set of points with all coordinates equal to ±1. For each P ∈ B, let SP be the
set of points in S which differ from P in exactly one coordinate. Since there are more than 2n+1 /n
points in B, and each SP has n elements, the cardinalities of the sets SP sum to more than 2n+1 ,
which is to say, more than twice the number of points in S. By the Pigeonhole Principle, there must
be a point of S in at least three of the sets, say in SP , SQ , SR . But then any
√ two of P, Q, R differ
in exactly two coordinates, so P QR is an equilateral triangle of side length 2 2, by the Pythagorean
Theorem.
Source: Putnam 2000
C3. Let S be a set of n points on a plane, no three collinear. A subset of these points is called polite if they
are the vertices of a convex polygon with no points of S in the interior. Let ck denote the number of
polite sets with k points. Show that the sum
n
X
(−1)i ci
i=3
depends only on n and not on the configuration of the points.
Solution: Consider the sum X

(−1)|T | . (†)
T ⊆S,|T |≥3
This clearly does not depend on the configuration of the points, and so we may call it f (n), where
n = |S|. (It is easy to see that f (n) = − 21 (n − 1)(n − 2), but we will not need this.)
Alternatively, let us calculate (†) by first grouping subsets of S into collection of subsets that share the
same convex hull. If the convex hull has t vertices on the boundary, and k points of S inside, then its
contribution to the sum in (†) will be
k
t
X k i
(−1) (−1) ,
i=0
i
2
which equals to zero if k ≥ 1 and (−1)t otherwise (consider the expansion of (1 − 1)k ). In other words,
if T is a subset of S whose vertices form a convex polygon, then the collection that T belongs to
contributes (−1)t to the sum if T is polite and 0 otherwise. Therefore,
n
X X
(−1)i ci = (−1)|T | = f (n).
i=3 T ⊂S,|T |≥3
Source: Iran 2006 Round 3

.....................................................................................................
G1. Let ABC be an acute triangle. The points M and N are taken on the sides AB and AC, respectively.
The circles with diameters BN and CM intersect at points P and Q respectively. Prove that P, Q and
the orthocenter H are collinear.
Solution: We need to show that H lies on the radical axis of the two circles, and thus it suffices to
show that it is equal powers with respect to the two circles. Let Y and Z be the feet of the altitudes
from B and C, respectively. Since ∠BY N = 90◦ , Y lies on the circle with diameter BN . Since BY
passes through H, it follows that the power of H with respect to this circle is HB · HY . Similarly, the
power of H with respect to the other circle is HC · HZ. On the other hand, HB · HY = HC · HZ
since B, C, Y, Z are concyclic. Thus, H has equal powers with respect to the two circles.
Source: Leningrad 1988
G2. Let ABC be a triangle with AC 6= AB, and select point B1 on ray AC such that AB = AB1 . Let ω be
the circle passing through C, B1 , and the foot of the internal bisector of angle CAB. Let ω intersect
the circumcircle of triangle ABC again at Q. Prove that AC is parallel to the tangent to ω at Q.
Solution: Let the angle bisector of ∠BAC meet BC at E and ω again at D. We have ∠ADB1 =
∠ADB = ∠ACB (this is true in both configurations) and it follows that C, B1 , E, D are concyclic.
Thus Q = D. Let ` be the line tangent to ω at Q. Then we have ∠(`, DA) = ∠ECD = ∠BCD =
∠BAD = ∠CAD. It follows that ` is parallel to AC.
Source: Russia 2001
G3. Let OAB and OCD be two directly similar triangles (i.e., CD can be obtained from AB by some
rotation and dilatation both centered at O). Suppose their incircles meet at E and F . Prove that
∠AOE = ∠DOF .
Solution: Let Ω1 be the incircle of OAB and Ω2 the incircle of OCD. Suppose p that Ω1 touches OA
at X, and Ω2 touches OC at Y . Consider an inversion about O with radius |OX| · |OY |. Suppose
that Ω1 gets sent to Ω01 and Ω2 gets sent to Ω02 . Note that the choice of the radius of inversion implies
that the radii of Ω1 and Ω02 are equal, and that the radii of Ω01 and Ω2 are equal. It follows that Ω01 is
the reflection of Ω2 about the angle bisector of ∠AOD, and likewise with Ω1 and Ω02 .
Let E 0 denote the image of E, so that E 0 is an intersection point of Ω01 and Ω02 . Then, E 0 is the
reflection of F across the angle bisector of ∠AOD. But O, E, E 0 are collinear. It follows that ∠AOE =
∠AOE 0 = ∠DOF .
Source: Tournament of Towns 2004 Fall
.....................................................................................................
3
N1. Let n > 1 be an odd integer. Prove that n does not divide 3n + 1.
Solution: Assume to the contrary that there is a positive odd integer n that divides 3n + 1. Let p be
the smallest prime divisor of n. Then p divides 3n + 1; that is, 3n ≡ −1 (mod p), so 32n ≡ 1 (mod p).
By Fermat’s little theorem, we also have 3p−1 ≡ 1 (mod p). It follows that
3gcd(2n,p−1) ≡ 1 (mod p).
Since p is the smallest prime divisor of n, gcd(n, p − 1) = 1. Because n is odd, p − 1 is even. Hence
gcd(2n, p − 1) = 2. It follows that 32 ≡ 1 (mod p), or p divides 8, which is impossible as p is odd.
N2. Let S be a finite set of integers, each greater than 1. Suppose that for each integer n there is some
s ∈ S such that gcd(s, n) = 1 or gcd(s, n) = s. Show that there exist s, t ∈ S such that gcd(s, t) is
prime.
Solution: Let n be the smallest positive integer such that gcd(s, n) > 1 for all s in n; note that n has
no repeated prime factors. By the condition on S, there exists s ∈ S which divides n.
On the other hand, if p is a prime divisor of s, then by the minimality of n, n/p is relatively prime to
some element t of S. Since n cannot be relatively prime to t, t is divisible by p, but not by any other
prime divisor of s (any such prime divides n/p). Thus gcd(s, t) = p, as desired.
Source: Putnam 1999
N3. Let a positive integer k be given. Prove that there are infinitely many pairs of integers (a, b) with
|a| > 1 and |b| > 1 such that ab + a + b divides a2 + b2 + k.
Solution: By inspection, we see that if (a, b) = (0, 1), then a2 + b2 + k is divisible by ab + a + b:
a2 + b2 + k
= k + 1. (†)
ab + a + b
Let us rearrange this as a quadratic in a:
a2 − (k + 1)(b + 1)a + (b − 1)(b − k) = 0.
As a quadratic in a, the sum of the roots is (k + 1)(b + 1). Hence, if (a, b) is a solution to (†), then so
is ((k + 1)(b + 1) − a, b), and hence by symmetry of (†), so is (b, (k + 1)(b + 1) − a).
Define a sequence (an ) as follows: a1 = 0, a2 = 1, and
an = (k + 1)(an−1 + 1) − an−2
for all n ≥ 2. Then by the above reasoning, (a, b) = (an , an+1 ) is a solution of (†) for all n ≥ 0.
Furthermore, the sequence (an ) is increasing, giving an infinite number of positive integer solutions.
4
IMO Training Camp Buffet Contest
June 30, 2008
1. Let A be a subset of {1, 2, . . . , 2008}, such that for all x, y ∈ A with x 6= y, the sum x + y is not divisible
by 1004. Find, with proof, the maximum possible size of A.
Solution: We can group the 2008 numbers in 1003 pairs of the form {k, 2008−k}, with k ∈ {1, 2, . . . , 1003},
and the pair {1004, 2008}. Observe that A cannot contain two elements from the same pair, as they would
add up to a number divisible by 1004. Thus |A| ≤ 1004.
To see that |A| = 1004 is possible, construct A by taking all the elements of {1, 2, . . . , 2008} whose remainder
upon division by 1004 is less than or equal to 502, and removing the elements 1506 and 2008 from the
resulting set. This yields a 1004-element set A satisfying the desired properties (why?).
2. Find, with proof, all real number solutions to the following:
(a2 + 1)(b2 + 1) = (ab + 1)(a + b).
Solution: We have
2(a2 + 1)(b2 + 1) − 2(ab + 1)(a + b)

= 2(a2 b2 + a2 + b2 + 1 − a2 b − ab2 − a − b)
= (a2 b2 − 2a2 b + a2 ) + (a2 b2 − 2ab2 + b2 ) + (a2 − 2a + 1) + (b2 − 2b + 1)
= a2 (b − 1)2 + b2 (a − 1)2 + (a − 1)2 + (b − 1)2
= (b2 + 1)(a − 1)2 + (a2 + 1)(b − 1)2 ,
which is positive unless a = 1 and b = 1, which forms a solution. Therefore, the only solution is a = b = 1.
Remark: Alternatively, note that by Cauchy-Schwarz inequality, we have
(a2 + 1)(b2 + 1) ≥ (ab + 1)2 and (a2 + 1)(1 + b2 ) ≥ (a + b)2
and multiplying together yields [(a2 + 1)(b2 + 1)]2 ≥ [(ab + 1)(a + b)]2 , so we can find all the solutions to
the equation by consider the equality cases in the above inequalities.
3. Find all ordered pairs (x, y) of positive integers such that 2x = 3y + 7.
Solution: In mod 3, we have (−1)x ≡ 1 (mod 3), so that x is even. Let x = 2a. Now, in mod 4, we have
0 ≡ (−1)y − 1 (mod 4), so that y is even. Let y = 2b. Then 22a − 32b = 7, so that
(2a + 3b )(2a − 3b ) = 7.
Since 2a + 3b > 0, 2a + 3b > 2a − 3b , and 7 is prime, we must have 2a + 3b = 7 and 2a − 3b = 1. So 2a = 4

and 3b = 3. Thus (a, b) = (2, 1) and so (x, y) = (2a, 2b) = (4, 2).
4. To clip a convex n-gon means to choose a pair of consecutive sides AB, BC and to replace them by the
three segments AM , M N , and N C, where M is the midpoint of AB and N is the midpoint of BC. In
other words, one cuts off the triangle M BN to obtain a convex (n + 1)-gon. A regular hexagon P6 of area
1 is clipped to obtain a heptagon P7 . Then P7 is clipped (in one of the seven possible ways) to obtain an
octagon P8 , and so on. Prove that no matter how the clippings are done, the area of Pn is at least 21 , for
all n ≥ 6.
Solution: The key observation is that for any n ≥ 6 and any side of P6 , some subsegment of this
side is a side of Pn (this can be easily proven using induction). So, for any Pn , we can select points
P1 , P2 , . . . , P6 on its perimeter so that Pi lies on the i-th side of P6 . Since Pn is convex, it contains the
1
1
hexagon P1 P2 P3 P4 P5 P6 . Therefore, it suffices to prove that the area of P1 P2 . . . P6 is at least 2 whenever
Pi lies on the i-th side of P6 for each i.
Consider this problem as a minimization problem, where we want to minimize the area of P1 P2 P3 P4 P5 P6
subject to the above condition. Observe that as Pi moves along the i-th side of P6 , the area of P1 P2 P3 P4 P5 P6
changes monotonically (in fact, linearly) as Pi moves form one end to the other. Therefore, the minimum
must occur when Pi coincides with a vertex of P6 . Therefore, we simply needs to search through the set
of (possibly degenerate) hexagons P1 P2 P3 P4 P5 P6 with the property that each Pi is one of the endpoints
of the i-th side of P6 . We wish to find the one with the minimum area. After some work, we see that the
minimum occurs when P1 P2 P3 P4 P5 P6 is an equilateral triangle, and its area is 21 .
Therefore, the area of Pn is at least 12 .
Remark: More elegantly, the bound can be proven using the inequality
x1 (1 − x2 ) + x2 (1 − x3 ) + x3 (1 − x4 ) + x4 (1 − x5 ) + x5 (1 − x6 ) + x6 (1 − x1 )
≤ (1 − x2 ) + x2 + (1 − x4 ) + x4 + (1 − x6 ) + x6
≤ 3.
I’ll leave the details to you.
Also, the bound 21 is not optimal. Can you find a better bound? What’s the best bound? (I don’t know
the answer to the last question.)
π
5. Let n be a positive integer. Suppose that θ1 , θ2 , . . . , θn are angles with 0 < θi < 2 for each i such that
2 2 2
cos θ1 + cos θ2 + · · · + cos θn = 1.
Prove that
tan θ1 + tan θ2 + · · · + tan θn ≥ (n − 1)(cot θ1 + cot θ2 + · · · + cot θn ).
Solution: Let ai = cos θi . Then

√
q
sin θi 1− cos2
θi a21 + a22 + · · · + a2i−1 + a2i+1 + · · · + a2n
tan θi = = =
cos θi cos θi ai
a1 + a2 + · · · + ai−1 + ai+1 + · · · + an
≥ √
ai n − 1
by the Power-Mean inequality. Summing the above inequality for i = 1, 2, . . . , n yields
n n X
X 1 X aj √ X aj
tan θi ≥ √ = n−1 ,
i=1
n − 1 i=1 j6=i ai i6=j
ai
ai
as each ratio aj appears n − 1 times.
On the other hand, we have
cos θi cos θi ai
cot θi = =√ =q
sin θi 1 − cos2 θi a1 + a2 + · · · + a2i−1 + a2i+1 + · · · + a2n
2 2

1 1 1 1 1 1
≤ ai + + · · · + + · · ·
(n − 1)3/2 a1 a2 ai−1 ai+1 an
again by the Power-Mean inequality. Summing the above inequalities for i = 1, 2, . . . , n yields
n n X
X 1 X aj √ X aj
cot θi ≤ = n − 1 .
i=1
(n − 1)3/2 i=1 j6=i ai i6=j
ai
Therefore,
n n
X √ X aj X
tan θi ≥ n−1 ≥ (n − 1) cot θi .
i=1
ai i=1
i6=j
2
6. Let AA1 , BB1 , CC1 be the altitudes of an acute triangle ABC. Let O be an arbitrary point inside A1 B1 C1 .
Denote the feet of the perpendiculars from O to the lines AA1 and BC by M and N , respectively; the ones
from O to the lines BB1 and CA by P and Q, respectively; the ones from O to the lines CC1 and AB by
R and S, respectively. Prove that the lines M N , P Q, and RS are concurrent.
Solution: Consider a dilation about O with ratio 2. Let M 0 , P 0 , R0 be the image of M, P, R respectively.
The dilation sends line M N to line A1 M 0 , line P Q to line B1 P 0 , line RS to line C1 R0 . So it suffices to
prove that A1 M 0 , B1 P 0 , C1 R0 are concurrent.
Observe that in triangle A1 B1 C1 , A1 A is an angle bisector (why?), and ∠M 0 A1 A = ∠M N O = ∠OA1 A.
That is, line A1 M 0 is the reflection of A1 O across the angle bisector of ∠B1 A1 C1 . Therefore, A1 M 0 , B1 P 0 , C1 R0
concur at the isogonal conjugate of O with respect to triangle A1 B1 C1 .
7. Positive integers a and b are given such that 2a + 1 and 2b + 1 are relatively prime. Find all possible values
of the greatest common divisor of 22a+1 + 2a+1 + 1 and 22b+1 + 2b+1 + 1.
Solution: We begin with a well-known lemma:

Lemma. For any positive integers k and n, we have gcd(2k − 1, 2n − 1) = 2gcd(k,n) − 1.
Proof. We use induction on the quantity k + n. If k + n = 1, the claim is obvious. Now, using the fact
that gcd(a, b) = gcd(a − b, b), we have (for k ≥ n)
gcd(2k − 1, 2n − 1) = gcd(2k − 2n , 2n − 1) = gcd(2n (2k−n − 1), 2n − 1)

= gcd(2k−n − 1, 2n − 1) = 2gcd(k−n,n) − 1 = 2gcd(k,n) − 1.
where we applied the induction hypothesis to the pair (k − n, n).
Recall the well-known factorization identity
4x4 + 1 = (2x2 + 1)2 − 4x2 = (2x2 + 2x + 1)(2x2 − 2x + 1).
Applying it to x = 2a gives us
24a+2 + 1 = (22a+1 + 2a+1 + 1)(22a+1 − 2a+1 + 1).
It follows that the greatest common divisor d of 22a+1 + 2a+1 + 1 and 22b+1 + 2b+1 + 1 divides the greatest
common divisor of 24a+2 + 1 and 24b+2 + 1, and hence it divides the greatest divisor of 28a+4 − 1 and
28b+4 − 1, which, by the lemma, equals to 2(8a+4,8b+4) − 1 = 24 − 1 = 15, as we are given that 2a + 1 and
2b + 1 are relatively prime. Therefore, d divides 15.
Since 22a+1 + 2a+1 + 1 ≡ 2a+1 6≡ 0 (mod 3), 3 does not divide d. Hence d = 1 or d = 5. Both cases are
possible. For a = 4, b = 8 we obtain d = 5, and for a = 2, b = 3 we obtain d = 1.
Second solution: Suppose a prime p divides 22a+1 + 2a+1 + 1 and 22b+1 + 2b+1 + 1. Then 22a+1 ≡ 2a+1 + 1
(mod p). Squaring gives
24a+2 ≡ 22a+2 + 2a+2 + 1 ≡ 2(22a+1 + 2a+1 + 1) − 1 ≡ −1 (mod p).
Similarly, 24b+2 ≡ −1 (mod p). Since 2a + 1 and 2b + 1 are relatively prime, there are integers j, k such
that j(2a + 1) + k(2b + 1) = 1 (where j and k necessarily have opposite parity). Hence, we get
22 ≡ 22((2a+1)j+(2b+1)k) ≡ (−1)j+k ≡ −1 (mod p).
Hence, p = 5.
On the other hand, suppose 25 divides 2(22x+1 + 2x+1 + 1) = (2x+1 + 1)2 + 1. Since the only solutions
to y 2 + 1 ≡ 0 (mod 25) are y ≡ ±7 (mod 25), we must have 2x+1 + 1 ≡ ±7 (mod 25). But it is easily
checked that 2 is a primitive root mod 25, as its order is not φ(25)/2 = 10 or φ(25)/5 = 4. It follows that
there is there a unique solution in mod 20 for x for each of 2x+1 + 1 ≡ 7 (mod 25) and 2x+1 + 1 ≡ −7
3
(mod 25), namely x ≡ 7 (mod 20) for the former and x ≡ 12 (mod 20) for the latter. In either case,
we have 5 | 2x + 1. This means that if 25 divides the greatest common divisor of 22a+1 + 2a+1 + 1 and
22b+1 + 2b+1 + 1, then 5 must divide both 2a + 1 and 2b + 1, contradicting the hypothesis that they are
relatively prime.
It follows that the only possible values for the greatest common divisor are 1 and 5. They can be achieved
by (a, b) = (4, 8) and (2, 3) respectively.
Source: St. Petersburg 2002
8. Let X be a finite set, and suppose A1 , . . . , Am and B1 , . . . , Bm are subsets of X with |Ai | = r and |B
i| = s
for each i, such that Ai ∩ Bi = ∅ for every i and Ai ∩ Bj 6= ∅ whenever i 6= j. Prove that m ≤ r+s r .
Solution: (by Béla Bollobàs in 1965) Let |X| = n. There are n! ways to label the elements of X with
{1, 2, . . . , n}. Let us pick a random labeling (i.e. permutation), so that every labeling is equally likely to
be chosen. For 1 ≤ i ≤ m, let Ei denote the event that the highest label in Ai is less than the lowest label
in Bi (which we denote by Ai < Bi ).
−1
First, observe that Pr(Ei ) = r+s r (why?). Now, we claim that if i 6= j, then Ei ∩ Ej = ∅. That is, no
labeling of X will result in both Ai < Bi and Aj < Bj . Indeed, if this were the case, and, say, the highest
label of Ai is at least as great as the highest label of Aj (we may switch i and j needed), then we must
have Aj < Bi so that Ai ∩ Bj = ∅, contradiction.
Thus, E1 , E2 , . . . , Em are mutually disjoint events. It follows that
−1
r+s
1 ≥ Pr(E1 ∪ E2 ∪ · · · ∪ Em ) = Pr(E1 ) + Pr(E2 ) + · · · + Pr(Em ) = m .
r
Therefore, |X| = m ≤ r+s

r .
4
Winter Camp 2008 Mock Olympiad
Monday, January 7, 2008
1. Find all triples of positive integers x, y, z satisfying
1 + 2 x 3y = z 2 .
2. For positive real numbers a, b, c such that abc ≤ 1, prove that

a b c
+ + ≥ a + b + c.
b c a
3. In acute triangle ABC, ∠A < 45◦ . Point D lies in the interior of triangle ABC such that
BD = CD and ∠BDC = 4∠A. Point E is the reflection of C across line AB, and point F is
the reflection of B across line AC. Prove that AD ⊥ EF .
4. Let m, n be two positive integers with m ≥ n. Prove that

n
k m−k n
X
(−1) = 1.
n k
k=0
Winter Camp 2008 Mock Olympiad
Monday, January 7, 2008
Solutions
1. Find all triples of positive integers x, y, z satisfying
1 + 2 x 3y = z 2 .
Solution: Since z = 1, 2 or 3 does not yield a solution, we assume z ≥ 4. Factoring we

obtain (z + 1)(z − 1) = 2x 3y . Since the numbers z − 1 and z + 1 differ by 2, at most one of
them is divisible by 3. Also, since the product is even, the factors are both even, and exactly
one of them is divisible by 4. So either z + 1 = 2 · 3y and z − 1 = 2x−1 or z + 1 = 2x−1 and
z − 1 = 2 · 3y .
By subtracting the two equations, the first case yields 2 = 2 · 3y − 2x−1 ; hence 3y − 2x−2 = 1.
This has a solution x = 3, y = 1, for which z = 5. For solutions with x ≥ 4, looking mod 4
we conclude that y must be even, say y = 2y1 . Factoring yields (3y1 + 1)(3y1 − 1) = 2x−2 . So
both factors on the left are powers of two, and since they differ by two, the first one is 4 and
the second is 2. This gives the solution x = 5, y = 2, and consequently z = 17.
The second case leads to the equation 2x−2 − 3y = 1. By looking at mod 3 we see that x is
even, so let x − 2 = 2x1 . Then we have (2x1 − 1)(2x1 + 1) = 3y . So 2x1 − 1 and 2x1 + 1 are
two powers of 3 that differ by 2, which can only be 1 and 3. From this we get the solution
x = 4, y = 1, z = 7.
In summary, the solutions are (x, y, z) = (3, 1, 5), (5, 2, 17), (4, 1, 7).
Source: Romanian TST 1984
2. For positive real numbers a, b, c such that abc ≤ 1, prove that

a b c
+ + ≥ a + b + c.
b c a
First solution: Using the weighted AM-GM, we have

r
2a b 2
3 a
+ ≥3 ≥ 3a.
b c bc
2b c 2c a
Similarly, c + a ≥ 3b and≥ 3c. Adding these three inequalities gives the result.
a + b
√
Second solution: Replacing a, b, c by ta, tb, tc with t = 1/ 3 abc preserves the LHS while
increases the RHS and makes at·bt·ct = abct3 = 1. Hence we may assume wolog that abc = 1.
Then there exists positive real numbers x, y, z such that
y z x
a= , b= , ,c = .
x y z
The Rearrangement Inequality gives
x3 + y 3 + z 3 ≥ x2 y + y 2 z + z 2 x.
1
Thus
a b c x3 + y 3 + z 3 x2 y + y 2 z + z 2 x
+ + = ≥ = a + b + c,
b c a xyz xyz
as desired.
3. In acute triangle ABC, ∠A < 45◦ . Point D lies in the interior of triangle ABC such that
BD = CD and ∠BDC = 4∠A. Point E is the reflection of C across line AB, and point F is
the reflection of B across line AC. Prove that AD ⊥ EF .
First solution: All angles are directed. Let RA be the rotation by angle 2∠CAB centered
at A and let RD be the rotation by angle ∠BDC centered at D. Then T = RA ◦ RD ◦ RA
is a translation because the sum of angles of RA , RD , RA is zero.
Note that RA (F ) = B, RD (B) = C, RA (C) = E, so composing them gives T(F ) = E.

−−→
Therefore, T is the translation by F E. Let M = RD (A), N = RA (M ). Then because triangle
ADM is isosceles and ∠M DA = 4∠CAB, we have ∠DAM = 90◦ − 2∠CAB. Therefore,
∠DAN = ∠DAM + ∠M AN = 90◦ − 2∠CAB + 2∠CAB = 90◦ .

−−→ −−→
Since T(A) = N , we have AN = F E, and therefore AD ⊥ EF.
Second solution: We use complex numbers. Use lower case letters to denote the complex
number corresponding to a point. Place D at the center, and let B and C be on the unit circle
so that B and C are symmetric about the real axis. Let α denote the unit complex number
with argument ∠CAB. Then, b = α2 , c = α−2 . Since AE = AC and ∠CAE = 2∠CAB, we
see that e − a = α2 (c − a), which gives us
e = a + α2 (c − a) = a(1 − α2 ) + cα2 = a(1 − α2 ) + 1.
Similarly, f = a(1 − α−2 ) + 1. So f − e = a(α2 − α−2 ). Note that α2 − α−2 is purely imaginary
as α lies on the unit circle. It follows that EF is perpendicular to AD.
Source: MOP 2007
2
4. Let m, n be two positive integers with m ≥ n. Prove that
n
k m−k n
X
(−1) = 1.
n k
k=0
First solution: (Finite differences) Let f be the polynomial

m−x (m − x)(m − 1 − x) · · · (m − n + 1 − x)
f (x) = = .
n n!
Then
n
X n
LHS = (−1)k f (k) = (−1)n ∆n f (0),
k
k=0
where ∆ denotes the finite difference operator, which transforms a function g(x) into g(x +
1) − g(x). For polynomials, if P is a polynomial of degree d, then ∆P is a polynomial of
degree d − 1. Since deg(f ) = n, we see that ∆n f (x) is a constant. So
n
k n
X
n n n n
LHS = (−1) ∆ f (0) = (−1) ∆ f (m − n) = (−1) f (m − n + k).
k
k=0
Since f (m − n + 1) = f (m − n + 2) = · · · = f (m) = 0, the only nonzero term is f (m − n) = 1,

and thus the original sum equals to 1.
Second solution: (Combinatorial) For i = 1, 2, . . . , n, let Si denote the set of all n-element
subsets of {1, 2, . . . , m} that does not contain the element i. That is
Si = {A ⊂ {1, 2, . . . , m} : |A| = n, i ∈
/ A} .
Since S1 ∪ S2 ∪ · · · ∪ Sn contains all n-element subset of {1, 2, . . . , m} except for the subset
{1, 2, . . . , n}, we have
m
|S1 ∪ S2 ∪ · · · ∪ Sn | = − 1.
n
Also, the intersection Sa1 ∩ Sa2 ∩ · · · ∩ Sai , where 1 ≤ a1 < a2 < · · · < ai ≤ n, contains all
n-element subsets of {1, 2, . . . , m}\{a1 , a2 , . . . ai }. So

m−i
Sa1 ∩ Sa2 ∩ · · · ∩ Sai = .
n
By the Inclusion-Exclusion Principle,
X X X
|S1 ∪S2 ∪· · ·∪Sn | = |Si |− |Si ∩Sj |+ |Si ∩Sj ∩Sk |−· · ·−(−1)n |S1 ∩S2 ∩· · ·∩Sn |,
1≤i≤n 1≤i<j≤n 1≤i<j<k≤n
which is equivalent to

m m−1 n m−2 n m−3 n n m−n n
−1= − + − · · · − (−1) ,
n n 1 n 2 n 3 n n
and rearranging gives
n
X
k m−k n
(−1) = 1,
n k
k=0
3
as desired.
Third solution: (Generating function) We see that given sum equals to the coefficient of
xm in the product F (x)G(x), where
∞ ∞
k n k
X X
F (x) = (−1) xk , and G(x) = xk .
k n
k=0 k=0
Using binomial theorem, we have

n
X n k
F (x) = (−1)k x = (1 − x)n ,
k
k=0
and
∞ ∞ ∞
X k k X n+k k X −n − 1 k
G(x) = x = xn x = xn (−1)k x = xn (1 − x)−n−1 .
n n k
k=n k=0 k=0
And thus
F (x)G(x) = xn (1 − x)−1 = xn + xn+1 + xn+2 + · · · .
So the coefficient of xm is 1, and thus the sum equals to 1.
Source: MathLinks Contest
4
IMO Training Camp Mock Olympiad #2
July 3, 2008
Time limit: 4.5 hours
1. Given an isosceles triangle ABC with AB = AC. The midpoint of side BC is denoted by M .
Let X be a variable point on the shorter arc M A of the circumcircle of triangle ABM . Let
T be the point in the angle domain BM A, for which ∠T M X = 90◦ and T X = BX. Prove
that ∠M T B − ∠CT M does not depend on X.
2. Find all functions f : R+ → R+ such that
f (x + f (y)) = f (x + y) + f (y)
for all x, y ∈ R+ . (Symbol R+ denotes the set of all positive real numbers.)
3. For a prime p and a positive integer n, denote by νp (n) the exponent of p in the prime
factorization of n. Given a positive integer d and a finite set {p1 , . . . , pk } of primes. Show
that there are infinitely many positive integers n such that d | νpi (n!) for all 1 ≤ i ≤ k.
July 5, 2008
1. A unit square is dissected into n > 1 rectangles such that their sides are parallel to the sides
of the square. Any line, parallel to a side of the square and intersecting its interior, also
intersects the interior of some rectangle. Prove that in this dissection, there exists a rectangle
having no point on the boundary of the square.
2. The diagonals of a trapezoid ABCD intersect at point P . Point Q lies between the parallel
lines BC and AD such that ∠AQD = ∠CQB, and line CD separates points P and Q. Prove
that ∠BQP = ∠DAQ.
3. Let n be a fixed positive integer. Find the maximum value of the expression
(ab)n (bc)n (ca)n

+ +
1 − ab 1 − bc 1 − ca
where a, b, c ≥ 0 and a + b + c = 1.
July 6, 2008
1. Let b, n > 1 be integers. Suppose that for each integer k > 1 there exists an integer ak such
that b − ank is divisible by k. Prove that b = An for some integer A.
2. Consider those functions f : N → N which satisfy the condition
f (m + n) ≥ f (m) + f (f (n)) − 1
for all m, n ∈ N. Find all possible values of f (2007).

(N denotes the set of all positive integers.)
3. Point P lies on side AB of a convex quadrilateral ABCD. Let ω be the incircle of triangle
CP D, and let I be its incenter. Suppose that ω is tangent to the incircles of triangles AP D
and BP C at points K and L, respectively. Let lines AC and BD meet at E, and let lines
AK and BL meet at F . Prove that points E, I and F are collinear.
July 8, 2008
1. Given non-obtuse triangle ABC, let D be the foot of the altitude from A to BC, and let I1 , I2
be the incenters of triangles ABD and ACD, respectively. The line I1 I2 intersects AB and
AC at P and Q, respectively. Show that AP = AQ if and only if AB = AC or ∠A = 90◦ .
2. Let λ be the positive root of the equation t2 − 2008t − 1 = 0. Define the sequence x0 , x1 , . . .
by setting
x0 = 1 and xn+1 = bλxn c for n ≥ 0.
Find the remainder when x2008 is divided by 2008.
3. Let A0 = (a1 , . . . , an ) be a finite sequence of real numbers. For each k ≥ 0, from the sequence
Ak = (x1 , . . . , xn ) we construct a new sequence Ak+1 in the following way.
We choose a partition {1, . . . , n} = I ∪ J, where I and J are two disjoint sets, such that the
expression

X X

xi − xj
i∈I j∈J
attains the smallest possible value. (We allow the sets I or J to be empty; in this case the
corresponding sum is 0.) If there are several such partitions, one is chosen arbitrarily. Then
we set Ak+1 = (y1 , . . . , yn ), where yi = xi + 1 if i ∈ I, and yi = xi − 1 if i ∈ J.
Prove that for some k, the sequence Ak contains an element x such that |x| ≥ n2 .
July 11, 2008
1. Let ABC be a triangle, and let M be the midpoint of side BC. Triangles ABM and ACM
are inscribed in circles ω1 and ω2 , respectively. Points P and Q are midpoints of arcs AB
and AC (not containing M , on ω1 and ω2 respectively). Prove that P Q ⊥ AM .
2. Let a1 , . . . , an and b1 , . . . , bn be two sequences of distinct real numbers such that ai + bj 6= 0

for all i, j. Show that if
n
(
X cjk 1 if i = k
=
a + bj 0 otherwise,
j=1 i
then
n X
X n
cjk = (a1 + · · · + an ) + (b1 + · · · + bn ).
j=1 k=1
3. Let X be a subset of Z. Denote
X + a = {x + a|x ∈ X}.
Show that if there exist integers a1 , a2 , . . . , an such that X + a1 , X + a2 , . . . , X + an form a

partition of Z, then there is an non-zero integer N such that X = X + N .
July 13, 2008
1. Let T be a finite set of real numbers satisfying the property: For any two elements t1 and
t2 in T , there is a element t in T such that t1 , t2 , t (not necessarily in that order) are three
consecutive terms of an arithmetic sequence. Determine the maximum number of elements
T can have.
2. Let ABM be an isosceles triangle with AM = BM . Let O and ω denote the circumcenter
and circle of triangle ABM , respectively. Point S and T lie on ω, and tangent lines to ω at S
and T meet at C. Chord AB meet segments M S and M T at E and F , respectively. Point X
lies on segment OS such that EX ⊥ AB. Point Y lies on segment OT such that F Y ⊥ AB.
Line ` passes through C and intersects ω at P and Q. Chords M P and AB meet R. Let Z
denote the circumcenter of triangle P QR. Prove that X, Y, Z are collinear.
3. Let a1 , a2 , . . . be a sequence of positive integers satisfying the condition 0 < an+1 − an ≤ 2008
for all integers n ≥ 1 Prove that there exist an infinite number of ordered pairs (p, q) of
distinct positive integers such that ap is a divisor of aq .
IMO Training Camp Mock Olympiad #2 Solutions
July 3, 2008
1. Given an isosceles triangle ABC with AB = AC. The midpoint of side BC is denoted by M .
Let X be a variable point on the shorter arc M A of the circumcircle of triangle ABM . Let
T be the point in the angle domain BM A, for which ∠T M X = 90◦ and T X = BX. Prove
that ∠M T B − ∠CT M does not depend on X.
Solution: See IMO Shortlist 2007 Problem G2
2. Find all functions f : R+ → R+ such that
f (x + f (y)) = f (x + y) + f (y)
for all x, y ∈ R+ . (Symbol R+ denotes the set of all positive real numbers.)
Solution: See IMO Shortlist 2007 Problem A4
3. For a prime p and a positive integer n, denote by νp (n) the exponent of p in the prime
factorization of n. Given a positive integer d and a finite set {p1 , . . . , pk } of primes. Show
that there are infinitely many positive integers n such that d | νpi (n!) for all 1 ≤ i ≤ k.
Solution: See IMO Shortlist 2007 Problem N7
1
July 5, 2008
1. A unit square is dissected into n > 1 rectangles such that their sides are parallel to the sides
of the square. Any line, parallel to a side of the square and intersecting its interior, also
intersects the interior of some rectangle. Prove that in this dissection, there exists a rectangle
having no point on the boundary of the square.
Solution: See IMO Shortlist 2007 Problem C2
2. The diagonals of a trapezoid ABCD intersect at point P . Point Q lies between the parallel
lines BC and AD such that ∠AQD = ∠CQB, and line CD separates points P and Q. Prove
that ∠BQP = ∠DAQ.
3. Let n be a fixed positive integer. Find the maximum value of the expression
(ab)n (bc)n (ca)n

+ +
1 − ab 1 − bc 1 − ca
where a, b, c ≥ 0 and a + b + c = 1.
Answer: When n = 1, then maximum is 38 . When n ≥ 2, the answer is 1

3·4n−1
.
a+b 2
≤ 14 , and

Solution: First consider the case when n ≥ 2. By AM-GM, we have ab ≤ 2
similarly ab, bc, ca ≤ 14 . So
(ab)n (bc)n (ca)n 4

+ + ≤ ((ab)n + (bc)n + (ca)n ) .
1 − ab 1 − bc 1 − ca 3
Thus, we have to prove that (ab)n + (bc)n + (ca)n ≤ 41n . Without loss of generality, suppose
that a is the maximum among a, b, c. We have a(1 − a) ≤ 14 , so
1
≥ an (1 − a)n = an (b + c)n ≥ an bn + an cn + nan bn−1 c ≥ an bn + bn cn + cn an .
4n
So we have proved that in this case the maximum is at most 3·41n−1 . Furthermore, this bound
can be attained when a = b = 12 , c = 0. Therefore, the maximum value of the expression is
1
3·4n−1
when n ≥ 2.
Now, suppose that n = 1. The value of 83 can be attained by a = b = c = 13 . So it remains to
prove that
ab bc ca 3
+ + ≥ .
1 − ab 1 − bc 1 − ca 8
Since a + b + c = 1, it is equivalent to prove that
ab bc ca 3
2
+ 2
+ 2
≥ .
(a + b + c) − ab (a + b + c) − bc (a + b + c) − ca 8
2
Now, clearing the denominators and simplifying (details omitted here . . . ), we see that the
above inequality is equivalent to
3[6, 0, 0] + 14[5, 1, 0] + 2[4, 2, 0] + 10[4, 1, 1] ≥ 6[3, 3, 0] + 10[3, 2, 1] + 13[2, 2, 2], (†)
where we adopt the notation

X
[p, q, r] = ap bq cr = ap bq cr + ap br cq + aq bp cr + aq br cp + ar bp cq + ar bq cp .
sym
Now, (†) is true as it is the sum of the following inequalities, each of which is true due to
Muirhead’s Inequality:
3[6, 0, 0] ≥ 3[3, 3, 0], 3[5, 1, 0] ≥ 3[3, 3, 0], 10[4, 1, 1] ≥ 10[3, 2, 1],

11[5, 1, 0] ≥ 11[2, 2, 2], 2[4, 2, 0] ≥ 2[2, 2, 2].
3
July 6, 2008
1. Let b, n > 1 be integers. Suppose that for each integer k > 1 there exists an integer ak such
that b − ank is divisible by k. Prove that b = An for some integer A.
Solution: See IMO Shortlist 2007 Problem N2
2. Consider those functions f : N → N which satisfy the condition
f (m + n) ≥ f (m) + f (f (n)) − 1
for all m, n ∈ N. Find all possible values of f (2007).

(N denotes the set of all positive integers.)
Solution: See IMO Shortlist 2007 Problem A2
3. Point P lies on side AB of a convex quadrilateral ABCD. Let ω be the incircle of triangle
CP D, and let I be its incenter. Suppose that ω is tangent to the incircles of triangles AP D
and BP C at points K and L, respectively. Let lines AC and BD meet at E, and let lines
AK and BL meet at F . Prove that points E, I and F are collinear.
4
July 8, 2008
1. Given non-obtuse triangle ABC, let D be the foot of the altitude from A to BC, and let I1 , I2
be the incenters of triangles ABD and ACD, respectively. The line I1 I2 intersects AB and
AC at P and Q, respectively. Show that AP = AQ if and only if AB = AC or ∠A = 90◦ .
Solution: First, we prove that if AB = AC or ∠A = 90, then AP = AQ. If AB = AC,

then the result is obvious, so assume that ∠A = 90◦ . Let P 0 , Q0 be the points on AB, AC,
respectively, such that AP 0 = AQ0 = AD, so ∠AP 0 Q0 = ∠AQ0 P 0 = 45◦ . Let I10 and I20
be the intersections of the bisectors of ∠BAD and ∠CAD, respectively, with P 0 Q0 . Then
triangles AP 0 I1 and ADI10 are congruent, and triangles and AQ0 I20 and ADI20 are congruent,
so ∠ADI10 = ∠ADI20 = 45◦ . Therefore I10 lies on the angle bisector of ∠BDA and I20 lies
on the angle bisector of ∠CDA. Hence, I10 and I20 are the incenters of ABD and ACD,
respectively, so I10 = I1 and I20 = I2 . This means that P 0 = P and Q0 = Q, so AP = AQ.
P I2
I1
B C
D
Conversely, suppose that AP = AQ. Let D0 be the point on AD such that AD0 = AP = AQ.
If D 6= D0 , then triangles AP I1 and AD0 I1 are congruent, and triangles AQI2 and AD0 I2
are congruent, so ∠I1 D0 A = ∠I2 D0 A. Then ∠I1 D0 D = ∠I2 D0 D, and since we also know
that ∠I1 DD0 = ∠I2 DD0 (they are either both 45◦ or both 135◦ , depending on whether D0
is on the segment AD or the ray AD past D), we have that triangles I1 D0 D and I2 D0 D
are congruent. Thus triangles AD0 I1 and AD0 I2 are congruent, so in particular ∠BAD =
2∠I1 AD0 = 2∠I2 AD0 = ∠CAD, hence triangle ABC is isosceles with AB = AC.
If D = D0 , then ∠AP I1 = ∠ADI1 = ∠ADI2 = ∠AQI2 = 45◦ , so ∠A = 90◦ .
Comment: Various trigonometric solutions are available, and they are generally pretty easy.
6
2. Let λ be the positive root of the equation t2 − 2008t − 1 = 0. Define the sequence x0 , x1 , . . .
by setting
x0 = 1 and xn+1 = bλxn c for n ≥ 0.
Find the remainder when x2008 is divided by 2008.
Solution: Because λ2 − 2008 − 1 = 0, we have λ = 2008 + λ1 . Note also that xn is an integer.

We conclude that
j xn k jx k jx k
n n
xn+1 = bxn λc = 2008xn + = 2008xn + ≡ (mod 2008).
λ λ λ
5
Because xn = bxn−1 λc and xn−1 is an integer and λ is irrational, we have xn = xn−1 λ − ,
where 0 < < 1 is the fractional part of xn−1 λ. Since λ > 1, we have 0 ≤ λ < 1, and so
jx k j k
n
= xn−1 − = xn−1 − 1.
λ λ
It follows that xn+1 ≡ xn−1 −1 (mod 2008). Therefore, by induction x2008 ≡ x0 −1004 ≡ 1005
(mod 2008).
3. Let A0 = (a1 , . . . , an ) be a finite sequence of real numbers. For each k ≥ 0, from the sequence
Ak = (x1 , . . . , xn ) we construct a new sequence Ak+1 in the following way.
We choose a partition {1, . . . , n} = I ∪ J, where I and J are two disjoint sets, such that the
expression

X X

xi − xj
i∈I j∈J
attains the smallest possible value. (We allow the sets I or J to be empty; in this case the
corresponding sum is 0.) If there are several such partitions, one is chosen arbitrarily. Then
we set Ak+1 = (y1 , . . . , yn ), where yi = xi + 1 if i ∈ I, and yi = xi − 1 if i ∈ J.
Prove that for some k, the sequence Ak contains an element x such that |x| ≥ n2 .
Solution: See IMO Shortlist 2007 Problem C4
6
July 11, 2008
1. Let ABC be a triangle, and let M be the midpoint of side BC. Triangles ABM and ACM
are inscribed in circles ω1 and ω2 , respectively. Points P and Q are midpoints of arcs AB
and AC (not containing M , on ω1 and ω2 respectively). Prove that P Q ⊥ AM .
First solution: By arc midpoints, we have that P A = P B and QA = QC, ∠P M A =

∠P M B, and ∠QM A = ∠QM C. Pick point D such that A, D lie on the same side of M and
M B = M C = M D. By SAS, 4P AM ∼ = 4P DM , and 4QM C ∼ = 4QM D; consequently,
P D = P A and QD = QA. Therefore, P DQA is a kite, and so P Q ⊥ AD =⇒ P Q ⊥ AM .
Second solution: Consider a rotation about P that brings B to A, and let the image of M be
T . Since ∠P BM + ∠P AM = 180◦ , we see that M, A, T are collinear and M B = AT = M C.
Similarly, consider a rotation about Q that brings C to A, this must also bring M to T as
the triangles QCM and QAT have equal side lengths.
Now, notice that in quadrilateral M P T Q we have P M = P T and QM = QT , so it is a kite,
and thus P Q ⊥ M T . The result follows from the fact that A lies on line M T .
2. Let a1 , . . . , an and b1 , . . . , bn be two sequences of distinct real numbers such that ai + bj 6= 0

for all i, j. Show that if
n
(
X cjk 1 if i = k
=
a + bj 0 otherwise,
j=1 i
then
n X
X n
cjk = (a1 + · · · + an ) + (b1 + · · · + bn ).
j=1 k=1
Pn
Solution: Let ri = k=1 cjk . Then
n n
X X cjk n X X cjk n n
X rj
= = = 1, (1)
ai + bj ai + bj ai + bj
j=1 j=1 k=1 k=1 j=1
for all i = 1, 2, . . . , n. We wish to determine r1 + · · · + rn . Let

n
X rj
R(x) = . (2)
x + bj
j=1
Then R(x) = P (x)/Q(x) where Q(x) = (x + b1 )(x + b2 ) · · · (x + bn ) and P (x) has degree at
most n − 1. By (1), R(a1 ) = R(a2 ) = · · · = R(an ) = 1, so if we write
S(x)
R(x) = 1 − ,
Q(x)
7
then S(x) is a monic polynomials of degree n and S(a1 ) = S(a2 ) = · · · = S(an ) = 0. Hence
S(x) = (x − a1 )(x − a2 ) · · · (x − an ).
Consider the coefficient of xn−1 in P (x) = Q(x)−S(x). Form (2), this coefficient is r1 +· · ·+rn .
On the other hand,
Q(x) = (x + b1 )(x + b2 ) · · · (x + bn ) and S(x) = (x − a1 )(x − a2 ) · · · (x − an ).
From this we see that coefficient xn−1 in Q(x) − S(x) equals to (a1 + a2 + · · · + an ) + (b1 +
b2 + · · · + bn ). Hence we have our desired result.
3. Let X be a subset of Z. Denote
X + a = {x + a|x ∈ X}.
Show that if there exist integers a1 , a2 , . . . , an such that X + a1 , X + a2 , . . . , X + an form a

partition of Z, then there is an non-zero integer N such that X = X + N .
Solution:
Assume wlog that 0 = a1 < a2 < · · · < an . Define a map f : Z → {a1 , a2 , . . . , an } thus: for
r ∈ Z, f (r) is the unique ai such that r = X + ai .
Suppose we are given f (r), f (r + 1), f (r + 2), . . . , f (r + an ) for some r ∈ Z. Then the value
of f (r + an + 1) is determined: if there exists ai > 0 such that f (r + an + 1 − ai ) = 0, then
f (r + an + 1) = ai ; otherwise, f (r + an + 1) = 0 = a1 . The value of f (r − 1) is also determined:
if there exists ai < an such that f (r − 1 + an − ai ) = an , then f (r − 1) = ai ; otherwise,
f (r − 1) = an .
By the Pigeonhole principle, there must exist distinct integers r and s such that
(f (r), f (r + 1), . . . , f (r + an )) = (f (s), f (s + 1), . . . , f (s + an )).
The preceding paragraph, along with a straightforward induction argument shows that that
f is periodic with period |r − s|, and therefore X = X + |r − s|.
8
July 13, 2008
1. Let T be a finite set of real numbers satisfying the property: For any two elements t1 and
t2 in T , there is a element t in T such that t1 , t2 , t (not necessarily in that order) are three
consecutive terms of an arithmetic sequence. Determine the maximum number of elements
T can have.
Solution: The answer is five.

Let m denote this maximum number. It is not difficult to check that the set S = {−3, −1, 0, 1, 3}
satisfies the conditions of the problem. Hence m ≥ 5. It suffices to show that m ≤ 5. We
approach indirectly by assuming on the contrary that m ≥ 6 and that T is a set satisfies
the given property and that T has m elements. Note that adding or multiplying a common
number to each element in T will not effect the property of the set. Hence, we may assume
that −3 = t1 < t2 < · · · < tm = 3. Then taking t1 = −3 and tm = 3, we know that 0 is an
element of T .
Because T has m ≥ 6 elements. There is an element in t in T such that t is not an element
in S. By symmetry, we may assume that 0 < t < 3. Because t is not in S, t 6= 1. Let t be the
element in T such that t > 0, t 6= 1, and |t − 1| is minimal. (This is possible because such t
exists by our assumption and that T is finite.) Taking elements −3 and t shows that t−3
2 ∈ T.
Because t−3 < 0, taking t−3
and 3 shows that t 0 = 1 t−3 + 3 = t+3 ∈ T . Note that t0 > 0,
2 2 2 2 4
and that |t0 − 1| = 41 |t − 1|, violating the minimality of |t − 1|. Thus, our assumption was
wrong and thus m = 5.
2. Let ABM be an isosceles triangle with AM = BM . Let O and ω denote the circumcenter
and circle of triangle ABM , respectively. Point S and T lie on ω, and tangent lines to ω at S
and T meet at C. Chord AB meet segments M S and M T at E and F , respectively. Point X
lies on segment OS such that EX ⊥ AB. Point Y lies on segment OT such that F Y ⊥ AB.
Line ` passes through C and intersects ω at P and Q. Chords M P and AB meet R. Let Z
denote the circumcenter of triangle P QR. Prove that X, Y, Z are collinear.
Solution: Consider a homothety centred at S that sends segment OM to segment XE. Let
ω0 be the image of ω under the homothety. We see that ω0 is centred at X and passes through
E and S. Furthermore, because ω is tangent to CS, ω0 is tangent to CS as well.
9
M
A R E B
X
Z S
Q
P
Since ∠M AR = ∠M BA = ∠M P A, we see that triangles M AR and M P A are similar, from

which we get that M A2 = M R · M P . Similarly, triangles M EA and M AS are similar, so
M A2 = M E · M S. Thus
M R · M P = M E · M S.
Also, by Power of a Point, we have
CP · CQ = CS 2 .
It follows that both M and C have equal powers with respect to ω0 and the circumcircle
of P QR. That is, M C is the radical axis of these two circles. Since the radical axis is
perpendicular to the line joining the centres of the two circles, we have that ZX ⊥ M C.
Similarly, ZY ⊥ M C. Therefore, X, Y, Z are collinear.
Source: China TST 2007
3. Let a1 , a2 , . . . be a sequence of positive integers satisfying the condition 0 < an+1 − an ≤ 2008
for all integers n ≥ 1 Prove that there exist an infinite number of ordered pairs (p, q) of
distinct positive integers such that ap is a divisor of aq .
Solution: Consider all pairs (p, q) of distinct positive integers such that ap is a divisor of aq .
Assume, by way of contradiction, that there exists a positive N such that q < N for all such
pairs.
We prove by induction on k that for each k ≥ 1, there exist
• a finite set Sk ⊂ {aN , aN +1 , . . . }, and

• a set Tk of 2008 consecutive positive integers greater than or equal to aN ,
such that at least k elements of Tk are divisible by some element of Sk .

For k = 1, the sets S1 = {aN } and T1 = {aN , aN +1 , . . . , aN +2007 } suffice.
Given Sk and Tk (with k ≥ 1), define
Y
Tk+1 = {t + s | t ∈ Tk }.
s∈Sk
10
Tk+1 , like Tk , consists of 2008 consecutive positive integers greater than or equal to aN —
in fact, greater than or equal to Q max Sk . Also, at least k elements of Tk+1 are divisible by
some element of Sk : namely, t + s∈Sk s for each of the elements t ∈ Tk which are divisible
by some element of Sk .
By the given condition 0 < an+1 − an ≤ 2008, and because the elements of Tk+1 are greater
than or equal to aN , we have that aq ∈ Tk+1 for some q ≥ N . Because the elements of Tk+1
are greater than max Sk , we have aq 6∈ Sk . Thus, by the definition of N , no element of Sk
divides aq .
Hence, at least k + 1 elements of Tk+1 are divisible by some element of Sk ∪ {aq }: at least k
elements of Tk+1 are divisible by some element of Sk , and in addition aq is divisible by itself.
Therefore, setting Sk+1 = Sk ∪ {aq } completes the inductive step.
Setting k = 2009, we have the absurd result that T2009 is a set of 2008 elements, at least 2009
of which are divisible by some element of S2009 . Therefore, our original assumption was false,
and for each N there exists q > N and p 6= q such that ap | aq . It follows that there are
infinitely many ordered pairs (p, q) with p 6= q and ap | aq .
Source: Vietnam TST 2001
11
Winter Camp 2009
Pre-camp problem set
1. Prove that n − 1 < √ 1√ + √2+3 √5 + √5+5√10 + . . . + √ 2n−1 √

< n for every positive
1+ 2 (n−1)2 +1+ n2 +1
integer n.
a+b
2. Prove that c+d is irreducible if ad − bc = 1.
3. The first five terms of a sequence are 1, 2, 3, 4, 5. From the sixth term on, each term is 1 less
than the product of all the preceding ones. Prove that the product of the first 70 terms is
equal to the sum of their squares.
4. Prove that the equation y 2 = x5 − 4 has no integer solutions.
5. Let ABC be an equilateral triangle of altitude 1. A circle, with radius 1, and center on the
same side of AB as C, rolls along the segment AB. Prove that the length of the arc of the
circle that is inside the triangle remains constant.
6. What real-valued functions f satisfy the inequalities
f (x) ≤ x, f (x + y) ≤ f (x) + f (y)
for all real x, y?
7. Let S be a finite set of points in the plane, and let r be a positive real number. Suppose it is
possible to place several circular discs of radius r on the plane such that: (a) every point in
S is covered by exactly one disc, and (b) the center of every disc is a point in S. Prove that,
no matter how this is done, the number of discs used will be constant.
8. Let a, b, c, d be positive real numbers such that a + b + c + d = 1. Prove that
a3 b3 c3 d3 1
+ + + ≥ .
b+c c+d d+a a+b 8
9. Fix a prime p > 2. An integer x is called a quadratic residue mod p if there exists an integer
y so that y 2 − x is a multiple of p.
(a) Determine the number of integers x in {0, 1, 2, . . . , p − 1}, for which both x and x − 1
are quadratic residues mod p.
p−1
(b) Prove that −1 is a quadratic residue mod p if and only if 4 is an integer.
10. Prove that there exists a unique function f defined on the positive reals such that f (f (x)) =
6x − f (x) and f (x) > 0 for all positive x.
1
11. Let O be the circumcenter of an acute-angled triangle ABC and let A1 be a point on the
smaller arc BC of the circumcircle of 4ABC. Let A2 and A3 be points on the sides AB and
AC respectively, such that ∠BA1 A2 = ∠OAC and ∠CA1 A3 = ∠OAB. Show that the line
A2 A3 passes through the orthocenter of 4ABC.
12. The numbers 1, 2, . . . , 22008 are stored in 2009 memory locations of a computer. Two pro-
grammers take turns choosing five memory locations, and then subtracting 1 from each of
these locations. If any location ever acquires a negative number, the computer breaks and
the guilty programmer pays for the repairs. Which programmer can ensure himself financial
security, and how?
13. (Pascal’s Theorem) Let A, B, C, D, E, F be points on a circle, and let P , Q, R respectively

be the intersection of AB and DE, BC and EF , CD and F A respectively.
(a) Let ω be the circumcircle of 4CF R, and let G and H be the second intersections of
lines BC and EF with ω. Prove that triangles RGH and P BE have parallel sides.
(b) Prove that P , Q, and R lie on a line.
14. A positive integer is written in each square of an 8 × 8 chessboard. One is allowed to choose
a 3 × 3 or a 4 × 4 square on the chessboard and increase all numbers in it by 1. Is it always
possible, applying such operations several times, to arrive at a situation where all squares
contain multiples of 10?
15. Let P and Q be points in the plane and let ω1 , ω2 , and ω3 be circles passing through both.
If A, B, C, D, E, and F are points on a line in that order so that A and D lie on ω1 , B and
E lie on ω2 , and C and F lie on ω3 , prove that AB · CD · EF = BC · DE · F A.
16. Given x, y, z ≥ 0 with xy + yz + zx = 2. Prove that
7(x + y + z)3 − 9(x3 + y 3 + z 3 ) ≤ 108.
Determine when equality holds.
17. Find all positive integers m, n such that m|n2 + 1 and n|m2 + 1.
18. For every finite set of positive integers A, define the collection A2 to be
A2 = {a + b|a, b ∈ A, a 6= b}.
For example, if A = {1, 2, 3, 4}, then A2 = {3, 4, 5, 5, 6, 7}. (Note that we are allowed to
repeat elements in a collection.) Suppose there exist two finite sets A, B such that A2 = B2
but A 6= B. Prove that the number of elements in A and in B are the same and this number
is a power of 2.
2
Solutions
2k−1 √
√ p
1. Note that √ = k 2 + 1 − (k − 1)2 + 1. Therefore,
(k−1)2 +1+ k2 +1
n
X n p
X p p
2k − 1
p √ = k 2 + 1 − (k − 1)2 + 1 = n2 + 1 − 1,
k=1
(k − 1)2 + 1 + k 2 + 1 k=1
√
and the result follows from the fact that n < n2 + 1 < n + 1.
2. d(a + b) − b(c + d) = ad − bc = 1, which implies a + b and c + d are relatively prime.

P Q
3. Let xn denote the nth term in the sequence, and let sn = ni=1 x2i , pn = ni=1 xi . For n ≥ 6,
sn − pn = sn−1 + (pn−1 − 1)2 − pn−1 (pn−1 − 1)

= sn−1 − pn−1 + 1
Since s5 − p5 = 55 − 120 = −65, it follows that s70 − p70 = 0.
4. For any integers x, y, one can check y 2 ∈ {0, 1, 3, 4, 5, 9} mod 11 and x5 ∈ {−1, 0, 1} mod 11.
Therefore, the equation has no solutions mod 11, and hence no integer solutions.
5. Suppose the circle is centered at O. We assume without loss of generality that O is closer to
B than A. Since O and C are both distance 1 from AB, we know CO is parallel to AB, and
hence ∠OCB = 60◦ .
Now let P and Q be where the circle hits CA and CB respectively. Also let Q0 be the point
−−→
on CB such that ∠P OQ0 = 60◦ . Note that ∠P OQ0 = 60◦ = ∠P CQ0 , so P COQ0 is cyclic.
Therefore, ∠OP Q0 = ∠OCQ0 = ∠OCB = 60◦ , which implies that 4OP Q0 is equilateral and
−−→
OQ0 = OP = 1. However, we know Q is the unique point on CB such that OQ = 1, which
means Q0 = Q, and hence ∠P OQ = ∠P OQ0 = 60◦ .
Therefore, the arc contained within 4ABC is always one-sixth of the circumference, regardless
of where O is.
6. Setting x = y = 0, we have f (0) ≤ f (0) + f (0) =⇒ f (0) ≥ 0. We are given f (0) ≤ 0, so it

follows that f (0) = 0. Setting y = −x, we have 0 = f (0) ≤ f (−x) + f (x). On the other hand,
f (−x) + f (x) ≤ 0 with equality only if f (−x) = −x and f (x) = x. It follows that f (x) = x
for all x.
7. Let X and Y be two sets of discs, each fulfilling the given requirements. We define a function
f : X → Y as follows. Fix a disc C ∈ X. Then, its center must be covered by some unique
disc D in Y . Define f (C) = D. Suppose f (C) = f (C 0 ) = D for some discs C, C 0 . Then, the
center of D is within distance r of the centers of C and C 0 , meaning C and C 0 both cover
the center of D, which is impossible. Therefore, f is one-to-one and |Y | ≥ |X|. Similarly,
|X| ≥ |Y |, and the result follows.
µq q q q ¶
a3 b3 c3 d3
8. By the Cauchy-Schwarz inequality applied to the vectors b+c , c+d , d+a , a+b and
3
√ √ √ √
( b + c, c + d, d + a, a + b), we have
¡ 1.5 ¢2
a3 b3 c3 d3 a + b1.5 + c1.5 + d1.5
+ + + ≥
b+c c+d d+a a+b (b + c) + (c + d) + (d + a) + (a + b)
µ 1.5 ¶2
a + b1.5 + c1.5 + d1.5
= 8· ,
4
¡ a+b+c+d ¢3
which, by the power-mean inequality is at least 8 · 4 = 18 .
9. Let n denote the number of consecutive quadratic residues mod p, and let m denote the
number of solutions (x, y) to x2 − y 2 ≡ 1(mod p). Since every quadratic residue except for 0
has exactly two square roots mod p, every pair of consecutive quadratic residues corresponds
to 4 solutions to x2 −y 2 ≡ 1(mod p), unless one of the residues is 0, in which case it corresponds
to 2 solutions. Therefore, m = 4n − 4 if −1 is a quadratic residue and m = 4n − 2 otherwise.
We now calculate m. Note that x2 −y 2 ≡ 1(mod p) is equivalent to there existing z 6= 0 so that
−1
x + y ≡ z(mod p) and x − y ≡ z −1 (mod p), which in turn is equivalent to x ≡ z+z2 (mod p)
−1
and y ≡ z−z2 (mod p). Thus, there is one unique solution (x, y) for each z 6= 0, and m = p−1.
Therefore, n = p+3 p+1
4 or 4 depending on whether −1 is a quadratic residue. Since § n¨ is an
integer, −1 is a quadratic residue if and only if p ≡ 1(mod 4), and either way, n = p4 .
10. For any x > 0, define the sequence {xn } by x0 = x and xn = f (xn−1 ) for n > 0. Then,
f (f (xn−2 )) = 6xn−2 − f (xn−2 ) =⇒ xn = 6xn−2 − xn−1 for all n. It follows1 that there exist
constants A, B so that xn = A · 2n + B · (−3)n . Since xn > 0 for all n, we must have B = 0,
and since x0 = x, we then have A = x. Therefore, f (x) = x1 = 2x. Conversely, it is easy to
check f (x) = 2x does satisfy the given condition.
11. Let B 0 and C 0 be where the altitudes from B and C hit the circumcircle of 4ABC. Then,
∠BA1 A2 = ∠OAC = 90◦ − ∠B = ∠BCC 0 = ∠BA1 C 0 , so A2 is the intersection of AB and
A1 C 0 . Similarly, A3 is the intersection of AC and A1 B 0 . Now, Pascal’s theorem (see #13),
applied to the “hexagon” ABB 0 A1 C 0 C, states that A2 , H, and A3 are collinear.
Remark: Instead of applying Pascal’s theorem, one can show HA2 = C 0 A2 and HA3 = B 0 A3 ,
and then angle-chase.
12. Let us call the programmers Alice and Biff. Also label the first 2005 memory locations as
good, and the remaining ones as bad. Alice can guarantee a win as follows. On her first turn,
she chooses the first good memory location and the four bad ones. From then on, she always
copies Biff’s last move.
This ensures that at the end of every one of Alice’s turns, all good memory locations have
even parity. On the other hand, there are only 4 bad memory locations, so each turn must
decrease a good memory location. Therefore, a good location will become negative in at most
1 + 1 + 2 + 4 + . . . + 22004 = 22005 turns. Conversely, no bad location can become negative
that quickly.
Since all good memory locations are even at the end of every one of Alice’s turns, it follows
that Biff must be the first one to turn one of these location negative.
1
We
Pare using a general theorem here: let c1 , c2 , . . . , cnn be P
constants, and let {x1 , x2 , . . .} be a sequence satisfying
xm = n i=1 ci xm−i for m ≥ n. Suppose the polynomial Pnx = i=1
n
ci xn−i has n distinct roots α1 , α2 , . . . , αn . Then,
m
there exist constants C1 , C2 , . . . , Cn such that xm = i=1 Ci · αi for all m.
4
13. It is important to correctly handle all possible configurations while doing part (a).
(a) Claim: Let circles ω1 and ω2 intersect at X and Y . If AB is a chord on ω1 , and C and
D are the second intersections of AX and BY with ω2 , then AB and CD are parallel.
Proof: Using directed angles2 , we have ∠XAB = ∠XY B = ∠XY D = ∠XCD. There-
fore, lines AB and CD make the same angle with line AXC, and thus are parallel.
The result follows from applying this claim to chords BA, BE, and DE.
(b) Since triangles RGH and P BE have parallel sides, P R, BG, and EH all meet at a
common point, namely Q, so Q lies on P R.
14. It is not always possible. We consider all positions modulo 10. In this setting, adding 1 to a
square 10 times has no effect.
If a position A can reach the zero position by adding 1 to various squares, the zero position
can reach A by subtracting 1 from the same squares. Since there are 52 + 62 = 61 squares
to choose from, and each square can be subtracted from between 0 and 10 times, a total of
1061 < 1064 positions can be reached from the zero position in this way.
However, there are 1064 positions altogether modulo 10, so not all of them can be reached.
15. Let R = P Q ∩ AF , and let a = AR, b = BR, c = CR, d = DR, e = ER, f = F R denote
−−→
signed distances along the line AB. Also let x denote the signed product P R · QR. By power
of a point on each circle, we have ad = be = cf = x. Now:
AB · CD · EF − BC · DE · F A = (b − a) · (d − c) · (f − e) − (c − b) · (e − d) · (f − a)
= bdf − bde − bcf + bce − adf + ade + acf − ace −
cef + ace + cdf − acd + bef − abe − bdf + abd
= bdf − dx − bx + cx − f x + ex + ax − ace −
ex + ace + dx − cx + f x − ax − bdf + bx = 0.
Remark: A slicker solution is to perform an inversion around P , and then use angle-Ceva’s
theorem to exploit the fact that lines A0 D0 , B 0 E 0 , and C 0 F 0 concur at Q0 .
16. Let s = x + y + z and p = xyz. Then,
x3 + y 3 + z 3 = (x + y + z)3 − 3(x + y + z)(xy + yz + xz) + 3xyz = s3 − 6s + 3p,
which implies:
108 − 7(x + y + z)3 + 9(x3 + y 3 + z 3 ) = 108 + 2s3 − 54s + 27p

= 2(s − 3)2 (s + 6) + 27p ≥ 0.
Equality holds iff s = 3 and p = 0, or equivalently, x, y, z are the roots of w3 − 3w2 + 2w = 0.

This means equality holds iff (x, y, z) = (0, 1, 2) or a permutation thereof.
2
We define the directed angle ABC to be the counter-clockwise angle (mod 180◦ ) one has to rotate A around
B by in order to make A lie on line BC. We use the following key properties: ∠ABD = ∠ACD iff A, B, C, D are
concyclic, and ∠ABC = ∠ABD iff B, C, D are collinear. These properties hold regardless of configuration, which is
a big improvement over traditional angle-chasing!
5
17. Let Fk denote the Fibonacci sequence. Then, the full set of solutions for (m, n) is:
T = {(1, 1), (F2k−1 , F2k+1 ), (F2k+1 , F2k−1 )}.

2
That these solutions are all valid follows immediately from the fact that F2i+1 + 1 = F2i−1 ·
³ √ í ³ √ í
1+ 5 1− 5
F2i+3 . (This identity can be proven directly by substituting Fi = √15 2 − √1
5 2 ,
on both sides and then expanding).
Now suppose there is a solution not in T . Take a minimal such solution (m, n), and assume
without loss of generality that m ≤ n. If m = n, then gcd(n, m2 +1) = 1, implying m = n = 1,
2
and giving a contradiction. Otherwise, let x = m n+1 , which we assumed to be an integer.
2 +1
Note x ≤ mm+1 ≤ m < n. On the other hand, working modulo m, we have x2 + 1 ≡
x2 + (m2 + 1)2 ≡ x2 + n2 x2 ≡ x2 (n2 + 1) ≡ 0. Therefore, (x, m) is a smaller solution than
(m, n).
Since
³ 2 x ≤´m, and since³ (m, n) was
´ assumed to be the smallest solution not in T , we have
m +1 m2 +1
n , m = {1, 1} or n , m = {F2k−1 , F2k+1 }. In the former case, (m, n) = (1, 2). In
the latter case, (m, n) = (F2k+1 , F2k+3 ). Either way, we have contradicted the assumption
that (m, n) 6∈ T , and the proof is complete.
¡ ¢ ¡ ¢
18. Suppose A2 = B2 . Then |A2 | = |A| 2 and |B2 | = |B|
2 so |A| = |B|. Let n = |A| = |B|.
P P
Let f (x) and g(x) be the polynomials a∈A x and b∈B xb . Factor f (x) − g(x) as (x − 1)t ·
a
h(x) for a non-negative integer t and a polynomial h satisfying h(1) 6= 0. Then,
(x − 1)t · h(x) · (f (x) + g(x)) = f 2 (x) − g 2 (x)

X X X X
= x2a + 2 xa − x2b − 2 xb
a∈A a∈A2 b∈B b∈B2
X X
2a 2b
= x − x
a∈A b∈B
= f (x2 ) − g(x2 )
= (x2 − 1)t · h(x2 )
= (x − 1)t · h(x2 ) · (x + 1)t .
Dividing through by (x − 1)t , we have h(x) · (f (x) + g(x)) = h(x2 ) · (x + 1)t . Now, letting
x = 1 and dividing through by h(1) (which we know is not 0), we have 2n = f (1) + g(1) = 2t .
It follows that n is a power of 2.
6
Winter Camp 2009
Buffet contest
A1. Show that for any positive integer n, there exists a positive integer m such that
√ √ √
(1 + 2)n = m + m + 1.
A2. For every ordered pair of positive integers (x, y), define f (x, y) recursively as follows:

 f (x − y, y) + 1 for x > y,
f (x, y) = f (x, y − x) + 1 for y > x,

x for x = y.
For example, f (5, 3) = f (2, 3) + 1 = f (2, 1) + 2 = f (1, 1) + 3 = 4. If f (x, y) ≤ 15, show that
x + y < 2009.
A3. Let x, y, z ≥ 0 be such that x + y + z = 3. Prove that
x3 y3 z3 1 2
3
+ 3
+ 3
≥ + · (xy + yz + zx).
y +8 z +8 x +8 9 27
A4. We assign a real number tx,y between 0 and 1 to every point on the plane (x, y) with integer
t +t +t +t
coordinates. This is done in such a way that tx,y = x−1,y x,y−1 4 x+1,y x,y+1 for all x, y. Show
that all the numbers are equal.
C1. A deck contains 52 cards of 4 different suits. Vanya is told the number of cards in each suit.
He picks a card from the deck, guesses its suit, and sets it aside; he repeats until the deck is
exhausted. Show that if Vanya always guesses a suit having no fewer remaining cards than
any other suit, he will guess correctly at least 13 times.
C2. A mathematics competition has n contestants and 5 problems. For each problem, each
contestant is assigned a positive integer score which is at most seven. It turns out every
pair of contestants have at most one problem whose scores are common. Find the maximum
possible value of n.
C3. Let n ≥ 2 be an integer and Tn be the number of non-empty subsets S of {1, 2, . . . , n} with
the property that the average of the elements in S is an integer. Prove that Tn − n is always
even.
C4. For n an odd positive integer, the unit squares of an n × n chessboard are coloured alternately
black and white, with the four corners coloured black. A tromino is an L-shape formed by
three connected unit squares. For which values of n is it possible to cover all the black
squares with non-overlapping trominos? When it is possible, what is the minimum number
of trominos needed?
1
G1. In 4ABC, points D and F are selected on sides BC and AB respectively so that AD · BC =
AB · CF . Let AD and CF intersect at P . Prove that either quadrilateral BF P D is cyclic or
quadrilateral F ACD is cyclic.
G2. Let ABC be a scalene triangle and let A0 , B 0 , and C 0 (respectively) be the points of intersection
of the interior angle bisectors A, B, and C (respectively) with the opposite sides of the triangle.
Now let:
– A00 be the intersection of BC with the perpendicular bisector of AA0 ;

– B 00 be the intersection of CA with the perpendicular bisector of BB 0 ;
– C 00 be the intersection of AB with the perpendicular bisector of CC 0 .
Show that A00 , B 00 , and C 00 are collinear.
G3. Let ω1 and ω2 be concentric circles with ω2 inside ω1 . Let ABCD be a parallelogram with
B, C, D on ω1 and A on ω2 . If BA intersects ω2 again at E and CE intersects ω2 again at
P , prove that CD = P D.
G4. Convex hexagon ABCDEF has area 1. Prove that at least one triangle out of ABC, BCD,
CDE, DEF , EF A, and F AB has area at most 16 .
N1. Find all positive integers n less than 1000 such that n2 is equal to the cube of the sum of its
digits.
N2. Find all integers a, b, c greater than 1 for which ab − 1 is divisible by c, bc − 1 is divisible by
a, and ca − 1 is divisible by b.
N3. The sequence of natural numbers a1 , a2 , a3 , . . . , satisfies the condition an+2 = an+1 an + 1 for
all n. Prove that an − 22 is composite for all n > 10, no matter what a1 and a2 are.
N4. Find all positive integers that can be written in the form
a2 + b2 + 1
ab
where a, b are positive integers.
2
Solutions
√ n
A1. Applying
√ nthe binomial
√ theorem to√(1 n+ 2) , we√ see there exist integers a and b 2such that
(1 2
¡ + √2) = a√+ ¢bn 2 and (1 − 2) = a − b 2. Multiplying these, we get a − 2b =
(1 + 2)(1 − 2) = ±1.
√ √ √ √ √ √
Setting m = min(a2 , 2b2 ), we have m + m + 1 = a2 + 2b2 = a + b 2 = (1 + 2)n .
Source: Romanian Math Stars Competition 2007, #1. Also see CMO 1994, #2.
A2. Let Fn = {1, 1, 2, 3, 5, . . .} be the Fibonacci sequence. We prove by induction on n that

if f (x, y) = n > 1, then min(x, y) ≤ 2Fn−1 and max(x, y) ≤ 2Fn . When n = 2, it is
straightforward to check (x, y) must be one of (2, 2), (1, 2), or (2, 1), and the result holds.
Now assume the result holds for n = k, and consider x, y with f (x, y) = k + 1. If x = y, the
result is trivial. Otherwise, assume without loss of generality that x > y. Then f (x−y, y) = k,
so by our inductive hypothesis, x = max(x − y, y) + min(x − y, y) ≤ 2Fk−1 + 2Fk = 2Fk+1 ,
and y ≤ max(x − y, y) ≤ 2Fk , completing the proof of the inductive step.
It follows that if f (x, y) ≤ 15, then x+y = max(x, y)+min(x, y) ≤ 2F14 +2F15 = 2F16 < 2009.
q
3 y 2 −2y+4
A3. Since y ≥ 0, the AM-GM inequality implies y3x+8 + y+2 27 + 27 ≥ 3 · 3 x3
272
= x3 . Similarly,
y3 z+2 z 2 −2z+4 y z3 x+2 x2 −2x+4
+
z 3 +8 27 + 27 ≥ 3 and x3 +8
+ 27 + 27 ≥ z3 . Adding all three inequalities, we
have:
x3 y3 z3 x + y + z y + z + x 6 y 2 + z 2 + x2
+ + ≥ + − −
y 3 + 8 z 3 + 8 x3 + 8 3 27 9 27
2
4 (x + y + z) − 2xy − 2yz − 2zx
= −
9 27
1 2
= + · (xy + yz + zx).
9 27
2
For equality to hold, we must have y+227 =
y −2y+4
27 =⇒ y 2 − 3y + 2 = 0, so y equals 1 or 2.
The same holds for z and x. Since x + y + z = 3, the only possibility is x = y = z = 1, and
it is easy to check that equality does indeed hold in this case.
A4. Define dx,y,n = tx+n,y+n − tx,y . Let C, ² be constants so that dx,y,1 ≥ C for some x, y
but dx,y,1 ≤ C + ² for all x, y. For any x, y, note that tx,y,n = tx,y,1 + tx+1,y+1,1 + . . . +
tx+n−1,y+n−1,1 ≤ n(C + ²).
We prove by induction that for all n, there exist x, y so that dx,y,n ≥ nC − 3n ². For n = 1, the
claim is trivial. Now suppose the result holds for n, and choose x, y so that dx,y,n ≥ nC − 3n ².
Using the given relation on t, we have:
dx−1,y,n+1 + dx,y−1,n+1 + dx+1,y,n−1 + dx,y+1,n−1
= dx,y,n ≥ nC − 3n ²
4
dx−1,y,n+1 + dx,y−1,n+1
=⇒ ≥ 2nC − 2 · 3n ² − (n − 1)(C + ²) ≥ (n + 1)C − 3n+1 ².
2
Therefore, one of dx−1,y,n+1 or dx,y−1,n+1 is at least (n + 1)C + 3n+1 ², and the claim is proven.
Now suppose dx,y,1 = C > 0 for some x, y. Fix ² > 0 and let m be the largest integer so that
there exists x, y for which dx,y,1 ≥ C + m². Then, as shown above, for each n, there exist
3
x, y so that dx,y,n ≥ nC + m² − 3n ² ≥ nC − 3n ². Choosing n large and then ² small, we have
dx,y,n > 1, which is impossible.
Therefore, dx,y,1 ≤ 0 for all x, y. Similarly, dx,y,1 ≥ 0 for all x, y, and hence tx,y = tx+1,y+1
for all x, y. Similarly, tx,y = tx+1,y−1 for all x, y. The original relation now implies that
tx,y = tx−1,y = tx,y−1 = tx+1,y = tx,y+1 , and the result follows.
Source: Iberoamerican Olympiad, miscellaneous problem
Remark: You cannot assume that there exist x, y for which dx,y,n is maximal. However, there
is a theorem in analysis saying there exist real numbers Mn such that dx,y,n gets arbitrarily
close to Mn without exceeding Mn . If you know this theorem, the proof becomes a lot cleaner.
C1. Let M denote the maximum number of cards remaining in any single suit. As Vanya proceeds,
M will only decrease if the current card is in a suit with M cards remaining, and no other suit
has M cards remaining. In this case, however, Vanya will correctly guess that suit. Therefore,
Vanya will guess correctly every time M decreases.
52
Since M ≥ 4 = 13 initially and it is 0 by the end, Vanya will be correct at least 13 times.
Source: Russia, 1998
§ ¨
C2. There are 7 possible scores on each question. If n ≥ 50, then at least 50
7 = 8 contestants
got the same score on problem 1. But then two of those contestants must have gotten the
same score on problem 2, which is impossible.
Now, for 1 ≤ i, j ≤ 7, let xi,j,k denote the value in {1, 2, . . . , 7} that is congruent to i +
jk (mod 7). Consider 49 contestants Ci,j where contestant Ci,j receives score xi,j,k on problem
k. Suppose that two contestants Ci1 ,j1 and Ci2 ,j2 got the same scores on questions k1 and
k2 . Then i1 − i2 + (j1 − j2 )k1 ≡ i1 − i2 + (j1 − j2 )k2 ≡ 0 (mod 7). Subtracting, we have
(j1 − j2 )(k1 − k2 ) ≡ 0 (mod 7) =⇒ j1 ≡ j2 (mod 7) =⇒ j1 = j2 . But then we must also
have i1 = i2 , which is a contradiction.
Therefore, it is possible to satisfy the required condition with 49 contestants, and hence 49
is the maximum possible value for n.
C3. Let S 0 denote the subsets of {1, 2, . . . , n} with at least two elements and with integer average.
For each set X ∈ S 0 that contains its average x, we pair it with the set X\{x}, and conversely
for each set Y ∈ S 0 that does not contain its average y, we pair it with the set Y ∪ {y}. This
is a proper pairing, so S 0 must contain an even number of sets.
Therefore, Tn has the same parity as the number of singleton sets with integer average, of
which there are exactly n.
Source: Putnam 2002, A3
C4. Let Xn denote the minimum number of trominos required to cover an n × n board in this
2
way. We claim any n ≥ 7 is possible, and Xn = (n+1)
4 .
Indeed, let B denote the set of all squares that are an even number of rows and even number
2
of columns away from the bottom-left square. There are exactly (n+1)
4 such squares, and they
(n+1)2
are all black. Furthermore, each tromino can cover at most one square in B, so Xn ≥ 4 .
(n+1)2
However, if n ∈ {1, 3, 5}, then 3 · 4 > n2 , so it is impossible to place this many trominos
on the board.
2
It remains to show that if n ≥ 7, there exists a valid tiling with (n+1)
4 trominos. For n = 7,
a valid tiling is shown below. Now, suppose there is a valid tiling of an n × n square using
4
2
exactly (n+1)
4 trominos. For k > 1 odd, we can also tile a 2 × k rectangle with k+1
2 trominos
by using two tronominos for the first 2 × 3 rectangle, and 1 tromino for each following 2 × 2
rectangle. Since an (n+2)×(n+2) rectangle can be partitioned into an n×n rectangle, a 2×n
2 (n+3)2
rectangle, and a 2×(n+2) rectangle, it can therefore by tiled with (n+1)
4 + n+1 n+3
2 + 2 = 4
trominos. The result now follows by induction.
Source: IMO Shortlist 2002, C2
G1. Let R denote the circumradius of 4ABC. By the extended sine law, we have AD ·BC = AB ·
4R2 ·(sin A)·(sin B)·(sin C) 2
sin B
sin ADB · BC = sin ∠ADB . Similarly, AB · CF = 4R ·(sinsin
A)·(sin B)·(sin C)
∠CF B . Equating
these, we get sin ∠ADB = sin ∠CF B, which implies ∠ADB = ∠CF B or ∠ADB = 180◦ −
∠CF B. In the former case, F ACD is cyclic; in the latter case, BF P D is cyclic.
G2. Assume without loss of generality that A00 lies on the same side of A0 as C does. Then,
∠A00 AA0 = ∠A00 A0 A = ∠CA0 A = 180◦ − ∠A0 AC − ∠A0 CA = ∠A 2 + ∠B. It follows that
00 00
∠CAA = ∠B, and hence AA is tangent to the circumcircle ω of 4ABC at A. Therefore,
A00 is the intersection of BC and the tangent to ω at A. Similar statements hold for B 00 and
C 00 .
The problem is now equivalent to Pascal’s theorem on the degenerate hexagon AABBCC.
Source: Iberoamerican Olympiad 2004
G3. Let O denote the center of ω1 and ω2 . The perpendicular bisectors of AE and CD are parallel
and both pass through O, so they are in fact identical. Therefore, the quadrilateral AEDC
is symmetric about this line, and BC = AD = EC.
Now, let B 0 and C 0 denote the second intersections of BE and CE with ω1 . ∠EC 0 B 0 = ∠EBC
and ∠B 0 EC 0 = ∠CEB so 4C 0 B 0 E ∼ 4BCE, and hence, C 0 B 0 = EB 0 .
Applying the symmetry argument to B 0 EAB and C 0 EP C, we also have B 0 E = AB = DC
and C 0 E = P C. Also, ∠C 0 EB 0 = ∠P CD since B 0 E and DC are parallel. Therefore,
4C 0 B 0 E ∼
= 4P DC, and the result follows.
Remark: there are two configurations, depending on which of A or E is closer to B, but this
argument works without change in either case.
G4. Let P be the intersection of AD and BE, Q be the intersection of BE and CF , and R be the
intersection of CF and AD. Also assume without loss of generality that P is on the same
side of CF as A and B; i.e., P is between A and R. Then, it is easy to check that R must be
between C and Q, and Q must be between E and P .
In this case, triangles ABR, BCR, CDQ, DEQ, EF P , and F AP are all disjoint, so their
total area is at most 1. It follows that one of them has area at most 16 . Regardless of which
triangle it is, we have found four adjacent vertices P, Q, R, S on the hexagon and a point X on
5
segment P S for which 4QRX has area at most 16 . Note that the area of 4QRX is bounded
between the area of 4QRP and the area of 4QRS. Therefore, one of these triangles also
has area at most 16 , and the result follows.
N1. Let s denote the sum of the digits of n. Then s3 = n2 ≡ s2 (mod 9) =⇒ s2 (s − 1) ≡

0 (mod 9), which implies s ≡ 0 (mod 3) or s ≡ 1 (mod 9). Also, s ≤ 9 + 9 + 9 = 27, and s is
a perfect square since n2 = s3 .
This leaves only the possibilities s = 1 or s = 9, which lead to n = 1 and n = 27, both of
which are valid solutions.
Source: Iberoamerican Olympiad 1999, #1
N2. Note that a, b and c are all relatively prime, since if p|a, b, then a cannot divide bc − 1. Now
the given condition implies:
(ab − 1)(bc − 1)(ca − 1) ≡ 0 (mod abc)

=⇒ a2 b2 c2 − a2 bc − ab2 c − abc2 + ab + bc + ca − 1 ≡ 0 (mod abc)
=⇒ ab + bc + ca ≡ 1 (mod abc)
Since ab + bc + ca > 1, it follows that ab + bc + ca > abc, or equivalently, a1 + 1b + 1c > 1. Now

assume without loss of generality that a ≤ b ≤ c. If a > 2, then since a, b, c are relatively
prime, a1 + 1b + 1c ≤ 13 + 14 + 15 < 1. Therefore, a = 2. If b > 3, then a1 + 1b + 1c ≤ 12 + 15 + 17 < 1.
Therefore, b = 3, which leaves only the option of c = 5.
Conversely, it is easy to check that (a, b, c) = (2, 3, 5) is a valid solution, as are all permutations
of (2, 3, 5).
Source: American Math Olympiad Program 1998
N3. For any n ≥ 3, we have an ≡ 0 (mod an ) and an+1 = an an−1 +1 ≡ 1 (mod an ). We now apply
the recurrence to calculate the sequence {an , an+1 , . . . an+6 } ≡ {0, 1, 1, 2, 3, 7, 22} (mod an ).
Therefore, an+6 − 22 must be a multiple of an .
For n ≥ 3, we have an = an−1 an−2 + 1 > 1. It is also easy to check that an+6 − an > 22.
Therefore, an and an+6
an
−22
are both integers greater than 1, and hence an+6 − 22 is not prime.
a2 +b2 +1
N4. Suppose k can be expressed in this form, and let (a, b) be such that ab = k and a + b is
as small as possible.
a2 +1
Suppose a < b. Then b = ka − b is an integer. Denoting this quantity by b0 , we have
(b−1)2 +1
b0 ≤ b < b, and
³ ´2
a2 +1
a2 + (b0 )2 +1 a2 + b +1 a2 + b2 + 1
= = = k,
ab0 a· a2 +1 ab
b
which contradicts the minimality of (a, b). Similarly, b < a is impossible so we must have
2 2 +1
a = b. In this case, a +b
ab is only an integer if a = b = 1 and k = 3.
Therefore, 3 is the only integer that can be expressed in this form.
2 2
Remark: The equation for b0 is found by root-flipping. We interpret a +bab
+1
= k as a quadratic
0
equation in b, and note that if b is one root, then b is the other one.
6
Winter Camp 2009
Mock Olympiad
1. At a party, one or more pairs of people shook hands with each other. We say two people were
“close” if either they shook hands with each other or if there was a third person they both
shook hands with. Show there were two people who (a) were close, and (b) shook hands with
the exact same number of people.
2. Let ABCD be a cyclic quadrilateral such that AD + BC = AB. Prove that the bisectors of
the angles ADC and BCD meet on the line AB.
3. Find all integer-valued functions, f and g, defined on the integers, for which g is one-to-one
and
f (g(x) + y) = g(f (y) + x)
for all integers x, y.
4. Prove that any set of 10 positive integers, ranging between 1 and 2009, contains three distinct
elements a, b, c such that gcd(a, b) divides c.
1
Solutions
1. For each person p, let np denote the number of people that p shook hands with. Let M =
max np , let x0 be a person for which np = M , and let {x1 , x2 , . . . , xM } be the people that x0
shook hands with.
For each i, we have nxi ≥ 1 since each person xi shook hands with x0 , and nxi ≤ M by defi-
nition. Therefore, the Pigeonhole Principle implies that two of the values {nx0 , nx1 , . . . , nxM }
are equal. But if nxp = nxq , then xp and xq have the property we are looking for.
2. Choose E on AB so that AE = AD and hence BE = BC. Also let F denote the second
intersection of AB and the circumcircle of 4CDE. Note that F lies between A and B.1 Also
note that regardless of configuration, ∠AF D = ∠ECD.
Therefore, ∠ADC = 180◦ − ∠EBC = 2∠BCE = 2∠BCD − 2∠ECD = 360◦ − 2∠F AD −
2∠AF D = 2∠ADF . Similarly, ∠BCD = 2∠BCF , so F is the intersection of the bisectors of
∠BCD and ∠ADC.
We know that F lies on AB, however, so the result follows.
3. Let m = f (0) and n = g(0). Setting x = 0 in the given equation yields f (n + y) = g(f (y))
(1). Setting y = 0 in the given equation yields g(m + x) = f (g(x)) (2).
Setting y = g(x) in equation (1) gives us g(f (g(x))) = f (n + g(x)), which we know from
the original equation is equal to g(f (n) + x). Since g is one-to-one, it follows that f (g(x)) =
f (n) + x. Combining this with (2), we have g(m + x) = f (n) + x. In particular, there exists
a constant C so that g(x) = x + C.
Substituting this into (2) yields m + x + C = f (x + C), which implies there exists a constant
B so that f (x) = x + B.
Conversely, it is easy to check that f (x) = x + B and g(x) = x + C satisfies the original
functional equation.
Source: Latvia
4. Consider a set {x1 , x2 , . . . , x10 } of positive integers for which gcd(xi , xj ) does not divide xk
for all distinct i, j, k. We claim max(xi ) > 2009. If each xi is even, we can divide through by
2 to achieve a set with smaller elements. Therefore, we may assume without loss of generality
that the set contains an odd element, say x1 .
For i > 1, let gi = gcd(x1 , xi ). If gi |gj for any i 6= j, then gcd(x1 , xi )|xj , which is impossible.
Therefore, x1 has 9 factors (g2 , g3 , . . . , g9 ), none of which divide each other.
Now, let p1e1 · pe22 · . . . · pekk be the prime factorization of x1 , ordered so that e1 ≥ e2 ≥ . . . ≥ ek .
e e e
Each gi can be written in the form p1i,1 · p2i,2 · . . . · pki,k . If ei,t = ej,t for allQ
t > 1, then one of gi
or gj must divide the other, which is impossible. However, there are onlyQ j>1 (ej +1) possible
values for {ei,2 , ei,3 , . . . , ei,t }, so if these are all distinct, we must have j>1 (ej + 1) ≥ 9.
1
For example, if B, E, A, F occur on line BA in that order, then by continuity, there exists X between E and A
for which BX · XA = EX · XF . This point X must lie on the radical axis of circle ABCD and circle CDE, which
is CD. Therefore, CD intersects AB between A and B, which is impossible.
2
Now, if k = 2, then e2 ≥ 8 and x1 ≥ 38 · 58 . If k = 3, then e2 , e3 ≥ 2 and x1 ≥ 32 · 52 · 72 ,
or e2 ≥ 4 and x1 ≥ 34 · 54 · 7. If k = 4, then a2 ≥ 2 and x1 ≥ 32 · 52 · 7 · 11. If k = 5, then
x1 ≥ 3 · 5 · 7 · 11 · 13. Regardless, x1 > 2009, and the result is proven.
Source: Romanian Math Stars Competition 2007, #8, except they consider 27 integers instead
of 10. They also conjecture the result holds even for 6 integers.
3
1 Mock CMO
1. (Lithuania team contest 2000, #3)
Solve the equation
p p p
−3x2 + 18x + 37 + −5x2 + 30x − 41 = x2 − 6x + 109.
Solution: √ √ √
After setting t = (x − 3)2 , the equation reduces to 64 − 3t + 4 − 5t = t + 100. t = 0 is
one solution, and the left-hand side is decreasing in t while the right-hand side is increasing in t.
Therefore, t = 0 (and hence x = 3) is the unique solution.
2. (Ukraine 10th grade 2008, #2)
Let ABCD be a parallelogram with 6 BCD > 90◦ . Extend line BC to K so that DK = DC, and
extend line DC to L so that BL = BC. Let the bisectors of 6 CDK and 6 CBL meet at point Q. If
6 BQD = α and KL = a, express the circumradius of triangle AKL in terms of a and α.
Solution:
ADLB and ADKB are isosceles trapezoids, so K, L lie on the circumcircle of 4ABD. 6 BAD =
6 LCK = 180◦ − 6 BQD since BQ and QD are perpendicular to LC and CK. Therefore, ABLQKD
is a cyclic hexagon. Now, 6 LAK = 6 LAQ + 6 KAQ = 6 LDQ + 6 KBQ = 180◦ − 2α. Applying
the extended sine law, we have the radius is 2 sina 2α .
3. (Japan junior math olympiad 2008, #10)
Consider a board consisting of 9 squares of length 1, arranged in 3 rows and 3 columns as in the figure
below. Let us number squares as indicated in the figure.
1 2 3
8 4
7 6 5
Let A, B, C, D be points chosen from the interior (not on the boundary) of the square 1, 3, 5, 7,
respectively. Denote by X the sum of the areas of the intersection of the quadrilateral ABCD with the
squares 1, 3, 5, and 7. Denote also by Y the sum of the areas of the intersections of the quadrilateral
ABCD with the squares 2, 4, 6, and 8. Prove that X < Y .
Solution:
Suppose that ABCD intersects the boundary of square i and square i + 1 at point Xi , and that
Xi is distance xi from P8 the central square. Then, the area of ABCD within square 2i is exactly
x2i−1 +x2i 1
2 , so Y = 2 i=1 xi . Now let Yi and Zi denote the outer and inner corners of square i for
i = 1, 3, 5, 7. Then, the area of ABCD within square 2i − 1 is less than the area of quadrilateral
X2i−2 Y2i−1 X2i−1 Z2i−1 , which is exactly 1 − 1−x22i−2 − 1−x22i−1 = x2i−2 +x
2
2i−1
. Therefore, X <
1
P8
2 i=1 ix = Y .
4. (Lithuania team contest 2000, #10)

A positive integer n is said to be reducible if there exist positive integers m and d such that
m+1 m+2
n= + .
d+1 d+2
How many reducible numbers are there from the set {1, 2, 3, . . . , 2000}?
Solution:
m+1 m+2 (m+1)(d+2)+(m+2)(d+1)
d+1 + d+2 = (d+1)(d+2) . For this to be an integer, we must have (m + 2)(d + 1) ≡
0 (mod d + 2) =⇒ m + 2 ≡ 0 (mod d + 2) and (m + 1)(d + 2) ≡ 0 (mod d + 1) =⇒ m + 1 ≡
1
0 (mod d + 1). Equivalently m ≡ d (mod lcm(d + 1, d + 2)). Since d + 1 and d + 2 are relatively
prime, it follows that m = C(d + 1)(d + 2) + d, which gives n = C(2d + 3) + 2.
Clearly, n = 2 can be expressed in this form but n = 1 cannot. For n > 2, the condition is
equivalent to requiring that n − 2 have an odd divisor greater than 3. This fails only for n = 2 + 2k
and n = 2 + 3 · 2k . Altogether, that gives only 1 + 11 + 10 numbers that are not reducible, so the
answer is 1978.
5. (Japan math olympiad final 2008, #5)
Does there exist a positive integer n satisfying the following condition?
Condition: For an arbitrary rational number r, there exists an integer b and nonzero integers a1 , a2 , . . . , an
such that r = b + a11 + . . . + a1n .
Solution:
No. We prove by induction on n that for each integer n and rational number q, there exists a
rational number q 0 < q such that no number in the interval (q 0 , q) can be expressed in this form.
For n = 1, the claim is trivial.
Now suppose the claim has been proven for n − 1. Fix an arbitrary q, and let f denote the
fractional part of q (except if q is an integer, we take f = 1). Then, if we are to express some
number y ∈ (q − f2 , q) in the required form, we must have some a1i ≥ f2 =⇒ ai ≤ f2 . This leaves
only a finite number of options for ai . Once ai is fixed to some value u, we have reduced the problem
to the n − 1 case, so by the inductive hypothesis, there exists qu0 such that no number in (qu0 , q) can
be achieved, given that ai = u. The inductive step follows from taking q 0 = max(q − f2 , maxu qu0 ).
2
2 Mock IMO Day 1
A point O is placed inside triangle ABC so that 6 BOC = 90◦ and 6 BAO = 6 BCO. If M and N are
the midpoints of the segments AC and BC respectively, prove that 6 OM N = 90◦ .
Solution:
Let P be the midpoint of OC. N is the circumcenter of right triangle BOC since it is the midpoint
of the hypoteneuse, so 6 BCO = 6 N CP = 6 N OP . Since M P is parallel to AO and M N is parallel
to AB, we have 6 N M P = 6 BAO = 6 BCO = 6 N OP . Therefore, OM P N is a cyclic quadrilateral
and 6 OM N = 6 OP N = 6 BOC = 90◦ .
2. (Japan math olympiad final 2008, #4)
Determine all real-valued functions f defined on the real line, which satisfy
f (x + y)f (f (x) − y) = xf (x) − yf (y).
for all real numbers x and y.

Solution:
Taking y = 0 gives f (x)f (f (x)) = xf (x), so f (x) = 0 or f (f (x)) = x. Taking y = f (x) gives
f (x + f (x))f (0) = xf (x) − f (x)f (f (x)) = 0. Taking y = −x gives 0 = f (0)f (f (x) + x) =
xf (x) + xf (−x), which implies f (−x) = −f (x) for x 6= 0.
Now suppose f (x) = 0 and y, f (y) 6= 0. Then, f (x+y)f (−y) = −yf (y) = yf (−y) =⇒ f (x+y) = y.
In particular, f (x + y) 6= 0 so f (f (x + y)) = x + y, and f (y) = x + y (*). If x 6= 0, then f (−x) = 0
also and the same argument implies f (y) = −x + y, which is impossible. Therefore, if f (x) = 0 for
some x 6= 0, then f (x) = 0 for all x 6= 0. It is easy to check any such function is a solution.
Otherwise, f (x) 6= 0 for all x = 0. Then f (0)f (f (0)) = 0 implies f (0) = 0, and (*) implies that in
this case, f (y) = y for all y. This, in turn, is easily checked to be a valid solution.
3. (Bulgarian math olympiad team selection test 2008, #1)
The number -1 is written at k of the vertices of a regular 2009-gon and the number 1 is written at
the remaining 2009 − k vertices. A vertex is said to be good if, starting from this vertex and running
around the polygon in either direction, every partial sum is positive. Find the largest number k such
that there exists a good vertex for any arrangement of the 1’s and -1’s.
Solution:
Let p be the number of positive vertices, and n the number of negative vertices. If p ≤ 2n, put all
positive vertices together. Then, any positive vertex can reach the n negative vertices after going
through at most n positive vertices, so it is bad. Conversely, if p > 2n, then for each positive vertex,
there must exist a minimal interval with it as an endpoint and sum 0. The opposite endpoint is
negative, and each negative point is endpoints to two such intervals. Therefore, some vertex is
good. k = 669.
3
3 Mock IMO Day 2
4. (IMO short list 2008, N1) Let n be a positive integer and let p be a prime number. Prove that if a, b, c
are integers (not necessarily positive) satisfying the following equations
an + pb = bn + pc = cn + pa,
then a = b = c.
Solution:
If two values are equal, the claim is trivial. Otherwise, multiply the equations an − bn = p(c − b) to
n
−bn bn −cn cn −an
get a a−b · b−c · c−a = −p3 . If n is odd, an − bn has the same sign as a − b, and the left-hand
side is positive, which is impossible. Otherwise, two of a, b, c are the same parity, and in this case,
n
−bn
the corresponding a a−b term is even, which implies p = 2.
n n
−b
The original equation now implies a, b, c must all be the same parity, and hence a a−b is even (as
3
above). Since the product of these terms is −2 , each such term is ±2. Comparing with the original
equation, we have a − b = ±(b − c). If a − b, b − c, c − a are all the same sign, they are all equal,
but their sum is 0 so then a = b = c. Otherwise, a − b = c − b or some shift thereof, and two values
are equal. Either way, we have a contradiction.
Prove that the inequality
x y z √
p +p +p ≤ 3
x2 + y + z x + y2 + z x + y + z2
holds for any non-negative real numbers x, y, z satisfying x2 + y 2 + z 2 = 3.

Solution:
x(1+y+z)+y(1+z+x)+z(1+x+y)
The given condition implies x+y +z ≤ x2 +y 2 +z 2 = 3, and hence (x+y+z)2 ≤ 1.
1 1+y+z x y z
By Cauchy-Schwarz, x2 +y+z ≤ (x+y+z) 2 , so therefore, x2 +y+z + x+y 2 +z + x+y+z 2 ≤ 1. The result
now follows from x + y + z ≤ 3 and Cauchy-Schwarz.
6. (Romanian master in mathematics 2008, #4)
Prove that from among any (n + 1)2 points inside a square of side length positive integer n, one can
pick three determining a triangle with area at most 12 .
Solution:
If three of the N = (n + 1)2 points are collinear, the problem is trivial. Otherwise, let H denote the
convex hull of the points, and let k denote the number of points on the boundary of H. Then, H has
area at most n2 and perimeter at most 4n. We first triangulate H. This uses exactly 2(N − 1) − k
n2
triangles, so one triangle has area at most 2(N −1)−k . Next, there must exist consecutive sides of
H with lengths a, b such that a+b 2 ≤ 4n
k . The area of the triangle with these two sides is at most
1 1 a+b 2 8n2

2 · ab ≤ 2 · 2 ≤ k2 .

n2 2
Therefore, some triangle has area at most X = min 2(N −1)−k , 8n
k2 . The first term is increasing in
n2 8n2
k, and the second is decreasing, so X is maximized when 2(N −1)−k = k2 , or equivalently k = 4n.
In this case, X = 21 , and the result follows.
4
Mock Olympiad #1
1. Given trapezoid AY CD with parallel sides AY and CD, assume that there exist points E on
line AD outside segment AD, and F inside segment Y C such that ∠Y CE = ∠ADF . Denote
by I the point of intersection of AY and EF , and by J the point of intersection of CD and
EF . Let K be the midpoint of segment EF , assume it does not lie on line CD. Prove that
I belongs to the circumcircle of CDK if and only if K belongs to the circumcircle of AY J.
2. Let a0 , a1 , a2 , . . . be a sequence of positive integers such that the greatest common divisor of
any two consecutive terms is greater than the preceding term; in symbols, gcd(ai , ai+1 ) > ai−1 .
Prove that an ≥ 2n for all n ≥ 0.
3. Prove that for any four positive real numbers a, b, c, d the inequality
(a − b)(a − c) (b − c)(b − d) (c − d)(c − a) (d − a)(d − b)

+ + + ≥0
a+b+c b+c+d c+d+a d+a+b
holds. Determine all cases of equality.
1
Canada Day Mock Olympiad 2009
Solutions
1. (IMO Short lost 2008, G2)

We will assume D is between A and E, as shown in the diagram below. It could also be
that A is between D and E, which leads to a different configuration, but the same argument
applies in that configuration with very little change.1
C J D
A
I Y
Since ∠Y CE = ∠ADF , we have ∠F CE = ∠Y CE = ∠ADF , and hence F CED is a cyclic

quadrilateral. Therefore, ∠F EA = ∠F ED = ∠F CD. Since CD and Y A are parallel, this
implies in turn that ∠F EA = 180◦ − ∠F Y A. Hence, F EAY is also a cyclic quadrilateral. In
particular, power of a point now implies the following two identities: CJ · DJ = F J · EJ(1)
and F I · EI = Y I · AI(2).
Furthermore, I belongs to the circumcircle of CDK if and only if IJ · KJ = CJ · DJ, which
is equivalent to IJ · KJ = F J · EJ by (1). Writing IJ = IF + F J, EJ = F E − F J, and
KJ = 21 F E − F J, this occurs if and only if F J = 2FF I·F E
I+F E (3).
Similarly, K belongs to the circumcircle of triangle AY J if and only if JI · KI = Y I · AI,
1
You can usually avoid the issue of different configurations by using signed angles and distances, but in this case,
you have to make sure you get the sign right when converting the given condition ∠DAE = ∠CBF into a condition
on signed angles.
1
which is equivalent to JI · KI = F I · EI. Writing JI = F I + F J, KI = F I + 12 F E, and
EI = F I + F E, this occurs if and only if F J = 2FF I·F E
I+F E (4).
The result follows from the fact that conditions (3) and (4) are identical.
2. (IMO Short list 2008, N3)
We first prove the result for small values of n = 0, 1, 2, 3. Since ai ≥ gcd(ai , ai+1 ) > ai−1 , the
sequence is increasing, so a1 ≥ 2. For each i ≥ 1, we also have ai+1 ≥ gcd(ai , ai+1 ) + ai ≥
ai + ai−1 + 1, hence a2 ≥ 4, a3 ≥ 7. a3 = 7 is impossible, because then 2 ≤ a1 < gcd(a2 , 7),
which contradicts the fact that a2 < 7. So a3 ≥ 8 and the claim is proven for n = 0, 1, 2, 3.
We now proceed be induction.
Let n ≥ 3 and assume that ai ≥ 2i for i = 0, 1, . . . , n. We must show that an+1 ≥ 2n+1 .
We will in fact show that an+1 ≥ min(2an , 4an−1 , 8an−2 , 16an−3 ), which is clearly sufficient.
Suppose this is not the case. Then we must have an+1 < 2an , so gcd(an+1 , an ) < an+12 , and
an+1 an+1
< an−1 < gcd(an , an+1 ) < .
4 2
an+1 2an+1
So we must have that gcd(an+1 , an ) = 3 and an = 3 .
Now look at an−2 . By assumption,
an+1 3an−1 an−1
an−2 >
> > .
8 8 3
But we also know that an−2 < gcd(an−1 , an ), and so gcd(an−1 , an ) = an−12 . Hence
an an+1 an−1 an
< < an−2 < gcd(an−1 , an ) = < ,
6 8 2 4
an 2an 4an+1
so we conclude that gcd(an−1 , an ) = 5 , and an−1 = 5 = 15 .
Finally, we see that

an+1 4an+1
< an−3 < gcd(an−2 , an−1 ) = gcd an−2 , .
16 15
4an+1 an+1 2an+1
From before, we know that 45 < 8 < an−2 < gcd(an−1 , an ) = 15 , hence an−2 doesn’t
divide 4a15
n+1
, and so
4an+1 an−2 an+1
gcd an−2 , ≤ < ,
15 2 15
but then
4an+1 4an+1 an+1
an−3 < gcd an−2 , ≤ <
15 75 16
which is a contradiction. This completes the induction, so the claim is proven.
3. (IMO Short list 2008, A7)
2 (c−a)2
Note that 2(a−b)(a−c)
a+b+c = (a−c)
a+b+c +
(a−c)(a−2b+c)
a+b+c and 2(c−d)(c−a)
c+d+a = c+d+a + (c−a)(c−2d+a)
c+d+a . There-
2(a−b)(a−c) 2(c−d)(c−a)
fore, a+b+c + can be written as
c+d+a

2 1 1 a − 2b + c a − 2d + c
(a − c) · + + (a − c) · −
a+b+c c+d+a a+b+c a+d+c

2 2a + b + 2c + d 3a + 3c
= (a − c) · + (a − c) · (d − b) · .
(a + b + c)(c + d + a) (a + b + c)(c + d + a)
2
2(b−c)(b−d) 2(d−a)(d−b)
Similarly, b+c+d + d+a+b can be written as

a + 2b + c + 2d 3b + 3d
(b − d)2 · − (a − c) · (d − b) · .
(b + c + d)(d + a + b) (b + c + d)(d + a + b)
Letting S = a + b + c + d, it therefore suffices to prove the following:

S+a+c S+b+d
(a − c)2 · + (b − d)2 ·
(S − b)(S − d) (S − a)(S − c)

(a + c)(S − a)(S − c) − (b + d)(S − b)(S − d)
≥ 3(a − c)(b − d) · .
(S − a)(S − b)(S − c)(S − d)
Assume without loss + c) ≥ bd(b + d). The right-hand side simplifies

of generality that ac(a
ac(a+c)−bd(b+d) 3·|(a−c)(b−d)·ac(a+c)|
to 3(a − c)(b − d) · (S−a)(S−b)(S−c)(S−d) ≤ (S−a)(S−b)(S−c)(S−d) . By the AM-GM inequality,
the left-hand side is at least
p
(S + a + c)(S + b + d)(S − a)(S − b)(S − c)(S − d)
2 · |(a − c)(b − d)| ·
(S − a)(S − b)(S − c)(S − d)
p
2(a + c) · (a + c) · c · (a + c) · a · (a + c)
≥ 2 · |(a − c)(b − d)| · .
(S − a)(S − b)(S − c)(S − d)
where we used the fact here that b, d ≥ 0 in each factor. Furthermore (a + c)2 ≥ 4ac, so this
is at least
p √
8(a + c)2 · a2 c2 4 2 · |(a − c)(b − d) · ac(a + c)|
2 · |(a − c) · (b − d)| · = .
(S − a)(S − b)(S − c)(S − d) (S − a)(S − b)(S − c)(S − d)
3·|(a−c)(b−d)·ac(a+c)|
This is clearly at least (S−a)(S−b)(S−c)(S−d) , which completes the proof. For equality to hold
at the end, we must have a = c or b = d. For equality to hold during our initial application
of AM-GM, we must then have both a = c and b = d. Conversely, it is clear that if a = c and
b = d, then equality does indeed hold.
3
Mock Olympiad #2
1. Let p be the product of two consecutive integers greater than 2. Show that there are no
integers x1 , x2 , . . . , xp satisfying the equation
p
" p
#2
! 4 !
x2i − xi = 1.
4·p+1
i=1 i=1
2. Let S = {x1 , x2 , . . . , xk+l } be a (k + l)-element set of real numbers contained in the interval
[0, 1]; k and l are positive integers. A k-element subset A ⊂ S is called nice if
$ $
$ $
$1 ! 1 ! $ k+l
$ xi − xj $$ ≤
$k l 2kl
$
xi ∈A xj ∈S\A $
% &
2 k+l
Prove that the number of nice subsets is at least .
k+l k
3. In $ABC, let AA0 , BB0 , CC0 be altitudes. Let A1 be a point inside $ABC such that
∠A1 BC = ∠A1 AB, ∠A1 CB = ∠A1 AC.
Let A2 , B2 , C2 be midpoints of AA1 , BB1 , CC1 respectively. Prove that A2 A0 , B2 B0 , C2 C0

are concurrent.
1
Mock Olympiad #2 Solutions
July 4, 2009
1. (IMO Short list 1988, #22)

Suppose integers x1 , x2 , . . . , xp exist. We will show that p = 2 or 6. Note that:
p p
!2
X 4 X
LHS = x2i − · xi
4p + 1
i=1 i=1
 !2 
p p p
1 X 4 X X
= · x2i + · p · x2i − xi 
4p + 1 4p + 1
i=1 i=1 i=1
 
p
1 X 4 X
= · x2i + · (xi − xj )2  .
4p + 1 4p + 1
i=1 1≤i<j≤p
Suppose xi takes on at least 3 values, with a numbers taking on the minimum value, c
numbers taking on the maximum value, and b numbers taking on intermediate values. Then,
2 2 2 2
P
1≤i<j≤p (xi − xj ) ≥ ac · 2 + ab · 1 + bc · 1 ≥ 4a + b + c ≥ p + 3, and LHS > 1. Therefore,
xi can take on at most 2 different values.
pn 2
If all the xi are equal to some value n, then LHS = 4p+1 . If p = 1, then this has no solutions
because 5 is not a square. If p > 1, then the factor of p in the numerator can never be
canceled out, so there are no solutions in this case either. Assume now that p > 2.
Let a numbers take the value A, and b numbers take the value B. Suppose that |A − B| > 1,
then
4 16(p − 1)
LHS ≥ · (4ab) ≥ >1
4p + 1 4p + 1
since p > 1. So we can assume that |A − B| = 1.
Now assume that neither A nor B are equal to 0. Then
(p − 1) + 4 4(p − 1) 5p − 1
LHS ≥ + =
4p + 1 4p + 1 4p + 1
which is bigger than 1 since p > 2.
So we can further assume A = 0. Then
b + 4ab
LHS = .
4p + 1
1
4p−3
If b = 1, then a = p − 1 and LHS = 4p+1 6= 1. If b ∈ [2, p − 2], then p ≥ 4, ab ≥ 2p − 4, and
8p−14 5p−5
LHS ≥ 4p+1 > 1. If b = p − 1, then a = 1 and LHS = 4p+1 , which is 1 only if p = 6.
This proves that p must equal 2 or 6. Conversely, if p = 2, we can take {x1 , x2 } = {1, 2}, and
if p = 6, we can take {x1 , x2 , . . . , x6 } = {0, 1, 1, 1, 1, 1}.
2. (IMO Short list 2008, C5)

If k = l = 1, the claim is trivial, so we will assume that k + l > 2. Consider a permutation
{y1 , y2 , . . . , yk+l } of S. Now look at the k + l k-element subsets
Ai = {yi , yi+1 , . . . , yi+k−1 }, i = {1, 2, ..., k + l}
where all indices are taken mod k + l.

Claim 1: At least 2 of the Ai are nice.
X X
Define f (Ai ) = k1 xj − 1l xj .
xj ∈Ai xj ∈S\Ai
Notice that f (A1 ) + f (A2 ) + · · · + f (Ak+l ) = 0. (This is because each element xj appears in
Ai for k different values of i, and it appears in S\Ai for l different values of i.) Also,

xi+k − xi xi+k − xi 1 1
|f (Ai+1 ) − f (Ai )| = + ≤ + .
k l k l
Therefore if Ai and Ai+1 are of different signs1 , then either |f (Ai )| or |f (Ai+1 | is at most
1 1 1 k+l
2 · k + l = 2kl , and therefore one of Ai , Ai+1 is nice.
Since the sum of the f (Ai ) is 0, we must have at least 1 negative and 1 positive sign (unless
they’re all 0 which is silly). If there exist 2 disjoint sets {i, i + 1}, {j, j + 1} such that
f (Ai ), f (Ai+1 ) and f (Aj ), f (Aj+1 ) are of opposite signs, then by above we have at least two
nice sets. Otherwise, exactly one f (Ai ) is of a different sign from the rest. Assume wlog that
f (A1 ) ≥ 0 and for i 6= 1, f (Ai ) < 0. If A1 is not nice, than bothP Ak+l and A2 are nice,k+l so we
have found our two nice sets. Otherwise, f (A1 ) ≤ k+l 2kl , and i6=1 |f (A i )| = f (A 1 ) ≤ 2kl , so
every set must be nice. This finishes the proof of Claim 1.
Now, consider choosing a random permutation {y1 , y2 , . . . , yk+l }, and then choosing a random
Ai corresponding to this permutation. By Claim 1, this chooses a nice set with probability
2
at least k+l . On the other hand, this is equivalent to first choosing the shift i and then the
permutation {y1 , y2 , . . . , yk+l }, but once i is fixed, we will be equally likely to choose any
possible set. Therefore, this entire process chooses a set uniformly at random, so it follows
2
that at least k+l · k+l
k sets are nice.
3. Ukraine 2008,11.8
Solution 1:
Denote the angles of triangle ABC by a, b, c, and let ∠A1 BC = ∠A1 AB = x,∠A1 CB =
∠A1 AC = y. Then
∠ABA1 = b − x, and so ∠BA1 A = 180 − b. Similarly, ∠CA1 A = 180 − c, and so ∠BA1 C =
180 − a. Therefore, if we let H be the orthocenter of triangle ABC, CBA1 H are concyclic.
1
We consider 0 to be of positive sign.
2
Denote the circle they lie on by S1 . Do a dilation centered at A with factor 21 , and let
S1 transform to S2 . Then denoting the midpoints of ABC by A0 , B 0 , C 0 , we know that S2
contains B 0 , C 0 , and the midpoint of AH, so it must be the nine-point circle. So S2 passes
through B0 , B 0 , C0 and C 0 as well. Notice that since S1 passed through A1 , A2 lies on the
nine-point circle.
We will prove A2 A0 , B2 B0 , C2 C0 are concurrent using Sine-Ceva’s theorem on triangle A0 B0 C0 .
Let B 00 denote the image of B0 under the dilation centered at A with factor 2. We have
sin ∠A2 A0 B0 = sin ∠A2 B 0 B0 by concyclicity

00
= sin ∠A1 CB by dilating around A
= sin ∠A1 CA
AA1 · sin ∠AA1 C
= by sine law for AA1 C
AC
AA1
= sin c
AC
AA1
Similarly, sin ∠A2 A0 C0 = AB sin b.
Applying Sine-Ceva to A0 B0 C0 ,
sin ∠A2 A0 B0 sin ∠B2 B0 C0 sin ∠C2 C0 A0

= 1,
sin ∠A2 A0 C0 sin ∠B2 B0 A0 sin ∠C2 C0 B0
so the 3 lines are concurrent and we’re done.
A2
B0
C’ B’
C0 A1
H B’’
B C
A0
C’’
Solution 2:
Let A0 denote the intersection of AA2 and BC. The given condition implies that the circum-
circle of 4AA1 B is tangent to BC at B, and the circumcircle of 4AA1 C is tangent to BC at
3
C. Since A0 is on the radical axis of these two circles, it follows that BA0 = CA0 , and hence
A0 is the midpoint of BC. Also let B 0 and C 0 denote the midpoints of AC and AB.
Now, as in the other solution, note that A2 lies on the nine-point circle, and hence C0 , A2 , B0 , A0 ,
and A0 are concyclic. Therefore, ∠C0 A0 A2 = ∠C0 A0 A2 = ∠C0 A0 A. By the sine law,
0
sin ∠C0 A0 A = AC0 · sinA∠BAA
0C
0
. Now, BC0 C is a right triangle with circumcenter A0 so
0 0B
A0 C0 = A0 B, and sin ∠C0 A0 A = (AC cos A) · sin ∠BAA
A0 B = (AC cos A) · sin ∠AA
AB .
sin ∠AA0 C sin ∠C0 A0 A AC 2
Similarly, sin ∠B0 AA0 = (AB cos A) · AC , so sin ∠B0 A0 A = AB 2
. Therefore,
sin ∠C0 A0 A2 sin ∠A0 C0 C2 sin ∠B0 B0 B2 AC 2 AB 2 BC 2

· · = · · = 1.
sin ∠B0 A0 A2 sin ∠C0 B0 B2 sin ∠A0 C0 C2 AB 2 BC 2 AC 2
The result now follows from Sine Ceva on 4A0 B0 C0 .
A2
B0
C0 A1
B C
A0 A’
4
Mock Olympiad #3
1. Let n ∈ N. Find the number of permutations (a1 , a2 , . . . , an ) of {1, 2, . . . , n} for which k |

2(a1 + a2 + . . . + ak ), for every 1 ≤ k ≤ n.
2. Find all functions f : R+ → R+ such that for any x, y, z ∈ R+ with x + y ≥ z,

√ √
f (x + y − z) + f (2 xz) + f (2 yz) = f (x + y + z).
3. (IMO Short list 2001, N4) Let p ≥ 5 be a prime number. Prove that there exists an integer
a with 1 ≤ a ≤ p − 2 such that neither ap−1 − 1 nor (a + 1)p−1 − 1 is divisible by p2 .
1
Mock Olympiad #3 Solutions
July 6, 2009
1. (IMO Short list 2008, C2)

Let Xn be the number of permutations of {1, 2, . . . , n} that have the given property, and let
Yn be the number of permutations of {2, 3, . . . , n + 1} that have
the given property. Note
that 2(a1 + a2 + . . . + ak ) ≡ 2 (a1 − 1) + (a2 − 1) + . . . + (ak − 1) (mod k), so we can always
subtract one from each a1 , a2 , . . . , an without changing the problem. In particular, this means
that Xn = Yn .
It is easy to check that X1 = 1, X2 = 2, and X3 = 6. From now on, we will assume n ≥ 4. We
have 2(a1 + a2 + . . . + an ) = 2(1 + 2 + l · · · + n) = n(n + 1) ≡ 0(mod n), so a permutation need
only satisfy the given condition for k = 1, 2, . . . , n − 1. Examining the condition for n − 1, we
see that
(n − 1)|2(a1 + a2 + · · · + an − an ),
so
(n − 1)|n(n + 1) − 2an .
n+1
This says that 2an = 2(mod n − 1), so either an = 1, n or n is odd and an = 2 . Suppose
we’re in the latter case. Then examining the condition for n − 2, we get
(n − 2)|2(a1 + a2 + · · · + an − an − an−1 ),
so
(n − 2)|n2 − 1 − 2an−1 .
But this implies 2an−1 = 3(mod n − 2), and since n − 2 is odd, this has the unique solution
an−1 = n+12 , which contradicts an−1 6= an . So we must have an = 1 or an = n. This means
that Xn = Xn−1 + Yn−1 = 2Xn−1 . For n ≥ 3, we therefore have Xn = 3 · 2n−2 .
2. (Mongolia 2008 TST, #2.1)

Claim: For any positive real numbers a, b, c, d satisfying d2 = a2 + b2 + c2 , we have f (a) +
f (b) + f (c) = f (d).
d−a b2 c2
Proof: Set z = 2 ,x = 4z , y = 4z . These values are all positive since d > a. Further-
1
more,
b2 + c2 (d2 − a2 ) − (d2 + a2 − 2ad)
x+y−z = −z = = a,
4z 2(d − a)
r r
√ b2 √ c2
2 xz = 2 z · = b, 2 yz = 2 z · = c, and
4z 4z
b2 + c2 (d2 − a2 ) + (d2 + a2 − 2ad)
x+y+z = +z = = d.
4z 2(d − a)
Since x + y − z = a > 0, we can substitute x, y, z into the given equation to get f (a) + f (b) +
f (c) = f (d), which completes the proof of the claim.
√
Now let g(x) = f ( x). We have proven that g(a) + g(b) + g(c) = g(a + b + c) for all a, b, c > 0.
Taking a = b = c = x, we have g(3x) = 3g(x). Taking a = b = 3x and c = x, we then
have g(7x) = 7g(x). Taking a = b = 2x and c = 3x, we then have g(2x) = 2g(x). Now,
taking a = b = x and c = (n − 2)x, we have g(nx) = 2g(x) + g((n − 2)x), so it follows that
g(nx) = ng(x) for all positive integers n. For any positive integers p, q, we therefore have
g( pq ) = 1q · g(p) = pq · g(1) (*).
Also, if a < d, then g(d) − g(a) = g(b) + g(c) > 0, so g is increasing. Since the rationals
are dense in the reals, it now follows from (∗) that g(x) = Cx for some constant C > 0, and
√ √
hence f (x) = Cx2 . Conversely, if f (x) = Cx2 , then f (x + y − z) + f (2 xz) + f (2 yz) =
√ √
(x + y − z)2 + (2 xz)2 + (2 yz)2 = (x + y + z)2 = f (x + y + z).
Remark: It isn’t necessary to calculate x, y, z explicitly to prove the claim. One may also
proceed as follows: First fix the products xz and yz, and let z vary. Then if z is very small,
x + y − z will approach infinity, and if z is very big, x + y − z will approach minus infinity,
so by continuity of x + y − z as a function of z, x + y − z will take every possible value. If
you find this sort of argument confusing and want to understand it better, feel free to speak
to one of the trainers!
3. (IMO Short list 2002, N4

If p = 5, then a = 2 satisfies the given condition. From this point forward, we will assume
p ≥ 7.
Claim 1: If x ∈ [1, p − 1], then xp−1 6≡ (p − x)p−1 (mod p2 ).
Proof: By the binomial theorem,
(p − x)p−1 ≡ xp−1 − p(p − 1) · xp−2 6≡ xp−1 (mod p2 )
since p 6 |x, p − 1.
Claim 2: If x ∈ [1, p − 1], then xp−1 6≡ (x + 2p)p−1 (mod p2 ).
Proof: Again, the binomial theorem gives
(x + 2p)p−1 ≡ xp−1 + 2p(p − 1) · xp−2 6≡ xp−1 (mod p2 )
since p 6 |x, p − 1.
Let hus callia number x “good” if xp−1 ≡ 1(mod p2 ). By pairing up x and p − x for all
x ∈ 1, p−1 p−1
2 , Condition 1 implies that at most 2 numbers in [1, p−1] are good. Furthermore,
2
we already know that 1 is good. Now, if two consecutive numbers in [1, p] are not good, then
the problem is done. Otherwise, the good numbers in [1, p] must be exactly 1, 3, 5, . . . , p − 2.
Now, (3p − 6)p−1 ≡ 3p−1 · (p − 2)p−1 ≡ 1(mod p2 ), so 3p − 6 is good. However, we already
know p − 6 is good, so this contradicts Claim 2. The proof is now complete.
3
Mock Olympiad #4
July 9, 2009
1. Let a, b, c, d be positive real numbers such that
a b c d
abcd = 1 and a + b + c + d > + + + .
b c d a
Prove that
b c d a
a+b+c+d< + + + .
a b c d
2. In an acute triangle ABC, segments BE and CF are altitudes. Two circles passing through
the points A and F are tangent to the line BC at the points P and Q so that B lies between
C and Q. Prove that the lines P E and QF intersect on the circumcircle of triangle AEF .
3. For every n ∈ N, let d(n) denote the number of (positive) divisors of n. Find all functions
f : N → N with the following properties:
(a) d(f (x)) = x for all x ∈ N;

(b) f (xy) divides (x − 1)y xy−1 f (x) for all x, y ∈ N.
1
1 Solutions
1. (IMO 2008 Short list, A5)
We will prove that
X a b

2a + 2b + 2c + 2d ≤ + . (1)
cyc
b a
This is clearly sufficient. Using abcd = 1 to homogenize, (1) becomes

3 3 3 3
2a 4 2b 4 2c 4 2d 4 X a b

1 1 1 + 1 1 1 + 1 1 1 + 1 1 1 ≤ +
b4 c4 d4 a4 c4 d4 a4 b4 d4 b4 c4 a4 cyc
b a
Using AM-GM, we get

3
a a b a 4a 4
+ + + ≥ 1 1 1 , and
b b c d b4 c4 d4
3
a a d a 4a 4
+ + + ≥ 1 1 1 .
d d c b b4 c4 d4
Adding these inequalities cyclically, we get the desired result.
2. (IMO 2008 Short list, G4)
Q B P C
We claim that AEP Q is a cyclic quadrilateral. Let a, b, c denote the angles of 4ABC. By
power of a point, BP 2 = BF · BA = BQ2 . Therefore, CP · CQ = (CB + BQ) · (CB −
2

BP ) = CB 2 − BF · BA = CB 2 − (CB · cos b) · CB sin c 2 · sin(b+c)−(cos b)(sin c)

sin a = CB sin a =
(cos c)(sin b) CB sin b
CB 2 ·

sin a = (CB cos c) · sin a = CE · CA. This proves AEP Q is cyclic by power
of a point.
Now, ∠F RE = ∠QRP = 180◦ − ∠RP Q − ∠RQP = ∠QAC − ∠QAB. Here, we used the
fact that AEP Q is cyclic and the circumcircle of AF Q is tangent to BC at Q. Therefore,
∠F RE = ∠BAC = ∠F AE, which completes the proof.
Remark: Here is another way to prove AEP Q is cyclic. Let H be the orthocentre of ABC.
Note that AF HE is cyclic. By power of a point, BP 2 = BQ2 = BF · BA = BH · BE, so the
circumcircle of QEH is tangent to line BC. Likewise, the circumcircle of P EH is tangent to
line BC. Now invert around C, preserving circle AF HE. The circles passing through AF
tangent to line BC goes to the circle passing through EH tangent to line BC, so P and Q
go to each other.
3. (IMO 2008 Short list, N5)

Recall that if the prime factorization of n is
pe11 · pe22 · · · pemm ,
then d(n) = (e1 + 1)(e2 + 1) · · · (em + 1).

Now, if p is a prime, then by above there must exist a prime q such that f (p) = q p−1 . Suppose
q 6= p. Taking x = q, y = p gives f (pq)|(q − 1)ppq−1 f (q). There exists some prime r with
f (q) = rq−1 , so there are at most q − 1 factors of q in (q − 1)ppq−1 f (q), and hence also in
f (pq).
Taking x = p, y = q gives f (pq)|(p − 1)q pq−1 f (p) = (p − 1)q pq+p−2 , and since there are at
most q − 1 factors of q in f (pq), we get f (pq)|(p − 1)q q−1 . Now, d(p − 1) ≤ p − 1 < p, and
d(q q−1 ) = q, so
d(f (pq)) ≤ d((p − 1)q q−1 ) ≤ d(p − 1)d(q q−1 ) < pq,
which contradicts condition (a) for pq. Hence, f (p) = pp−1 for all primes p.
Now let n be a positive integer, and let p be the smallest prime dividing n. Taking x = p, y =
n
p , we get
n−1
n
f (n)|(p − 1) · f (p).
p
Now, since p is the smallest prime dividing n and d(f (n)) = n, the exponent of any prime
dividing f (n) must be at least (p − 1). Suppose a prime power rk divides f (n) but not n.
n−1
Then r doesn’t divide np f (p), so rk divides p − 1. But k ≥ p − 1, and so rk ≤ p − 1 ≤ k,
which is impossible. So all primes dividing f (n) must divide n. In particular, it follows
immediately that if n = pk is a prime power, then f (n) = pn−1 .
e1 e2 em −1
p −1 p −1
We claim that if n = pe11 · pe22 · · · pemm , then f (n) = p11 · p2 2 · · · ppmm . We proceed by
induction on n. d(f (1)) = 1, so f (1) = 1. Now assume it is true for all numbers less than
n. If n is a prime power, then we’ve already proven the claim. So assume n has at least 2
3
prime factors. Let r be a prime dividing
and let p 6= r be a prime dividing n. Taking
f (n),
n n n−1 n
x = p , y = p, we get f (n)| p − 1 p f p . We know r must divide n, so it doesn’t divide

n n−1 , and hence it divides f ( n ). By our induction claim, the exponent of r dividing
p −1 p p
e1 e2 em −1
p −1 p −1
f (n) is bounded by rk − 1, where rk ||n. Hence, f (n) divides p11 · p22 · · · ppmm . The
claim now follows from the fact that d(f (n)) = n.
It’s straightforward to check this f actually satisfies the two conditions.
4
Mock Olympiad #5
July 10, 2009
1. An eccentric mathematician has a ladder with n rungs that he always ascends and descends
in the following way: When he ascends, each step he takes covers a rungs of the ladder, and
when he descends, each step he takes covers b rungs of the ladder, where a and b are fixed
positive integers. By a sequence of ascending and descending steps he can climb from ground
level to the top rung of the ladder and come back to ground level again. Find, with proof,
the minimum value of n, expressed in terms of a and b.
2. Let A0 , B 0 , C 0 , D0 , E 0 , F 0 be midpoints of the sides AB, BC, CD, DE, EF, F A of convex hexagon
ABCDEF . Also let p denote the perimeter of ABCDEF and p0 denote the perimeter of
A0 B 0 C 0 D0 E 0 F 0 . If all inner angles of hexagon A0 B 0 C 0 D0 E 0 F 0 are equal, prove that
√
2· 3
p≥ · p1 .
3
3. Let X be a set of 2k elements and F a family of subsets of X each of cardinality k such that
each subset of X of cardinality k − 1 is contained in precisely one member of F. Show that
k + 1 is prime.
1
1 Solutions
1. (IMO 1990 Short list, #13)
We claim the minimum value of n is a + b − gcd(a, b). We will prove this first in the case
where a and b are relatively prime.
Claim: If the mathematician makes m moves beginning at some arbitrary rung and ends up
where he started, then m ≥ a + b (assuming a, b are relatively prime).
Proof of claim: If he ascends A times and descends B times, we must have Aa = Bb.
Since a and b are relatively prime, it follows that a|B and b|A, so B ≥ a, A ≥ b, and
m = A + B ≥ a + b.
Consider a ladder with n < a + b − 1 rungs, and suppose the mathematician can make a
sequence of moves and end up where he started. By the above claim, this number of moves
must be at least a + b > n + 1. However, there are only n + 1 different positions for the
mathematician, so this is impossible.
Next we show a ladder with n = a + b − 1 rungs is always possible. From any position, the
mathematician always has a legal move: if he is on rung 0, 1, . . . , b − 1, he can ascend, and if
he is on rung b, b + 1, . . . , a + b − 1, he can descend. Therefore, he can start on the ground
floor and continue to make moves until he returns to a position he has already been to. By
the original claim, he can only repeat a position after visiting all other a + b − 1 positions
first. Therefore, he must return to the ground floor before repeating another position, and he
must have visited the top rung during that time. Thus, n = a + b − 1 rungs is indeed always
possible, which completes our solution in the case where a, b are relatively prime.
Finally, suppose a and b have greatest common divisor g > 1. Then the mathematician can
only visit rungs with position a multiple of g. Ascending is equivalent to climbing up ag such
rungs, and descending is equivalent to going down gb such rungs. Since ag and gb are relatively
prime, we know the minimum possible n for which is possible to visit the top rung and return
to the start in this scenario satisfies ng = ag + gb − 1. Equivalently, n = a + b − gcd(a, b).
2. (Vietnam TST 2004, #5)

Let A0 , B0 , . . . , F0 be the projections of A, B, . . . , F onto the segments F 0 A0 , A0 B 0 , . . . , E 0 F 0
respectively. Then,
A0 A0 + A0 B0 = AA0 cos ∠AA0 A0 + A0 B cos ∠BA0 B0

AB
· cos ∠AA0 A0 + cos(60◦ − ∠AA0 A0 ) .

=
2
Since cos θ is concave on the range [0◦ , 60◦ ], it follows from Jensen’s inequality that
√
0 0 ◦ AB · 3
A0 A + A B0 ≤ AB · cos 30 = .
2
Applying
√ similar arguments for B 0 , C 0 , . . . , F 0 , and adding the resulting inequalities, we have
p0 ≤ 23 · p, as required.
2
For equality to hold, ∠AA0 A0 = ∠BA0 B0 = 30◦ , and so on. Therefore, ABCDEF is equian-
gular and triangles A0 BB 0 , B 0 CC 0 , . . . , F AA0 are all isosceles, so AA0 = A0 B = BB 0 = . . . =
F F 0 = F 0 A, and ABCDEF is equilateral. Thus, equality holds if and only if ABCDEF is
regular.
3. (China TST 2009, #4.3 and India TST 1998)

Claim: Let r be a positive integer such that 2 ≤ r ≤ k − 1 and let S be a subset of size k − r.
Then there are
(k + r)(k + r − 1) · · · (k + 2)
r!
sets from F that contains S.
Proof of claim: There are k + r elements not in S. Therefore, there are (k + r)(k + r −
1) · · · (k + 2) ways to add elements to S (in order) to make S size k − 1. Let the resulting set
be S 0 . By the property of F, there exists a unique element not in S 0 that we can add to S 0
so that the resulting set is in F. For each set F in F that contains S, there are r! ways to
add r elements to S to make the resulting set in F . Therefore, there are
(k + r)(k + r − 1) · · · (k + 2)
r!
sets from F that contains S. This proves the claim.
We now have that
(k + r)(k + r − 1) · · · (k + 2)
(1)
r!
is an integer for all 2 ≤ r ≤ k − 1. Let p be a prime factor of k + 1. Then p | k + 1 and p does
not divide any of the numbers k + 2, k + 3, · · · , k + p, which means
(k + p)(k + p − 1) · · · (k + 2)
(2)
p!
is not an integer. If k + 1 is not prime however, then 2 ≤ p ≤ k − 1, and (2) contradicts (1)
when r = p. We conclude that k + 1 must be prime.
3
IMO Winter Camp 2010 Warm-Up Problems 1
IMO Winter Camp 2010 Warm Up Problems

Algebra
A1 Let x, y, z ∈ R be non-negative real numbers such that 0 ≤ x, y, z ≤ 1. Find the maximum possible value
of
x + y + z − xy − yz − zx.
Determine all triples (x, y, z) for which this maximum is attained.
A2 For any positive integer k, prove that
2k p
X 1
2k 2 + k < k 2 + j < 2k 2 + k + .
j=1
2
A3 Let a, b, c be positive real numbers such that a2 + b2 + c2 = 3. Prove that

1 1 1 3
+ + ≥ .
1 + ab 1 + bc 1 + ca 2
A4 P (x) is a polynomial of odd degree with real coefficients. Show that the equation P (P (x)) = 0 has at
least as many distinct real roots, as the equation P (x) = 0.
A5 Find all functions f : R 7→ R such that ∀x, y, z ∈ R, we have that if x3 + f (y) · x + f (z) = 0, then
f (x)3 + y · f (x) + z = 0.
Number Theory
N1 Let n be a positive integer. The sum of the positive divisors of n is s. Prove that the sum of the reciprocal
of these divisors is ns .
N2 A rational number x is written on a blackboard. In each step, you erase x and replace it with either x + 1
or − x1 . (If x = 0, you must choose x + 1). Prove that for any rational number p, if p currently appears on
the blackboard, then you can make 0 appear after a finite number of steps.
N3 Find all triples of positive integers (a, b, c) such that a2 + 2b+1 = 3c .
N4 Find all positive integers that can be written in the form
m 2 + n2 + 1
,
mn
for all positive integers m, n.
N5 Find all prime numbers p such that the following statement is true: there are exactly p ordered pairs of
integers (x, y) such that 0 ≤ x, y < p and y 2 ≡ x3 − x mod p.
1
Geometry
G1 A circle C and a point P are given on the same plane. Given any point Q on the circumference of C, let
M be the midpoint of P Q. Find the locus of point M , i.e. find all possible locations for point M .
G2 Let A be a point outside of a circle C. Two lines pass through A, one intersecting C at B, C, with B closer
to A than C, and the other intersecting C at D, E, with D closer to A than E. The line passing through
D parallel to AC intersects C a second time at F and the line AF intersects C a second time at G. Let
M = EG ∩ AC. Prove that
1 1 1
= + .
|AM | |AB| |AC|
G3 Let ABC be a triangle and D be a point on side BC. The internal angle bisector of ∠ADB and that of
∠ACB intersect at P . The internal angle bisector of ∠ADC and that of ∠ABC intersect at Q. Let M
be the midpoint of P Q. Prove that |M A| < |M D|.
G4 A convex quadrilateral ABCD has |AD| = |CD| and ∠DAB = ∠ABC < 90◦ . The line through D and
the midpoint of BC intersects line AB at point E. Prove that ∠BEC = ∠DAC.
G5 Let ABC be a triangle with |AB| > |AC|. Let its incircle touch side BC at E. Let AE intersect this
incircle again at D. Let F be the second point on AE such that |CE| = |CF |. Let CF intersect BD at
G. Prove that |CF | = |F G|.
Combinatorics
C1 Find the number of subsets of {1, 2, · · · , 10} that contain its own size. For example, the set (1, 3, 6) has 3
elements and contains 3.
C2 A sequence of non-negative integers is defined by G0 = 0, G1 = 0 and Gn = Gn−1 + Gn−2 + 1 for every
n ≥ 2. Prove that for every positive integer m, there exists a positive integer a such that Ga , Ga+1 are
both divisible by m.
C3 A set S of ≥ 3 points in a plane has the property that no three points are collinear, and if A, B, C are
three distinct points in S, then the circumcentre of ∆ABC is also in S. Prove that S is infinite.
C4 Let n, k be positive even integers. A survey was done on n people where on each of k days, each person
was asked whether he/she was happy on that day and answered either ”yes” or ”no”. It turned out that
on any two distinct days, exactly half of the people gave different answers on the two days. Prove that
there were at most n − nk people who answered ”yes” the same number of times he/she answered ”no”
over the k days.
C5 There are n ≥ 5 people in a room, where each pair is classified as friends or strangers. No three people are
mutually friends. There also exist an odd number of people P1 , · · · , Pm such that Pi is friends with Pi+1
for all i ∈ {1, · · · , m}, where the indices are taken modulo m. Prove that there exists one person who is
friends with at most 2n/5 people.
2
IMO Winter Camp 2010 Warmup Solutions

Algebra
A1 Let x, y, z ∈ R be non-negative real numbers such that 0 ≤ x, y, z ≤ 1. Find the maximum possible value
of
x + y + z − xy − yz − zx.
Determine all triples (x, y, z) for which this maximum is attained.
Solution: The maximum possible value is 1 and is attained when (x, y, z) = (1, t, 0) for any 0 ≤ t ≤ 1, or
any of its permutation solutions.
Note that
x + y + z − xy − yz − zx = −(1 − x)(1 − y)(1 − z) + 1 − xyz ≤ 1.
Equality holds when at least one of x, y, z is equal to 1 and at least one of x, y, z is equal to 0. Clearly,
this is possible. Therefore, the maximum possible value of x + y + z − xy − yz − zx is 1 and is attained
when (x, y, z) = (1, t, 0) for any 0 ≤ t ≤ 1, and any of its permutation solutions.
Source: Diamonds by Tran Phuong
Comments: When the expressions x + y + z, xy + yz + zx, xyz appear, you should always consider poly-
nomials of degree three and/or terms of the form (c − x)(c − y)(c − z) for some constant c. This may even
help you factor terms in other similar problems.
A2 For any positive integer k, prove that

2k p
X 1
2k 2 + k < k 2 + j < 2k 2 + k + .
j=1
2
Solutions: By subtracting 2k 2 from all three terms, it suffices to show that

2k p
X 1
k< ( k 2 + j − k) < k + .
j=1
2
Note that
p j
k2 + j − k = p ,
k2 + j + k
and
j j j
<p < .
2k + 1 k2 + j + k 2k
Therefore,
2k 2k 2k
X j X j X j
< p < .
j=1
2k + 1 j=1 k 2 + j + k j=1
2k
3
Hence,
2k
2k(2k + 1) X j 2k(2k + 1)
< p < .
2(2k + 1) j=1
2
k +j+k 2(2k)
j
p
Since √ = k 2 + j − k, we conclude that
k2 +j+k
2k p
X 1
k< k2 + j − k < k + .
j=1
2
This solves the problem.
Source: Mongolian Mathematical Olympiad 2009
Comments: The appearance of the square root suggests that p you rationalize the pnumerator of some
you to consider the term k 2 + j − k, since k < k 2 + j < k + 1 for
expression. It should be natural for p
j j
1 ≤ j ≤ 2k. Bounding the term j/( k 2 + j + k) between 2k+1 and 2k should also be natural since you
j
are looking for rational lower and upper bounds of √ 2 .
k +j+k
A3 Let a, b, c be positive real numbers such that a2 + b2 + c2 = 3. Prove that

1 1 1 3
+ + ≥ .
1 + ab 1 + bc 1 + ca 2
a2 +b2 1 1
Solution 1: Note that ab ≤ 2 . Therefore, 1+ab ≥ 2 2 for all a, b ≥ 0. Therefore, by AM-HM
1+ a +b
2
inequality, we have
X 1 X 1 9 9 3
≥ a2 +b2 ≥ P a2 +b2 = 2 2 2
= ,
cyc
1 + ab cyc 1 + 2 (1 + 2 ) 3+a +b +c 2
cyc
as desired.
Solution 2: By adding the terms on the left-hand side and cross multiplying, the inequality becomes
equivalent to
3 + (ab + bc + ca) ≥ abc(a + b + c) + 3a2 b2 c2 .
By homogenizing (i.e. using the condition a2 + b2 + c2 = 3 to make every term have the same degree), we
have
(a2 + b2 + c2 )3 (a2 + b2 + c2 )2 (ab + bc + ca) (a2 + b2 + c2 )abc(a + b + c)
3· + ≥ + 3a2 b2 c2 .
27 9 3
By clearing denominators, this inequality becomes equivalent to
(a2 + b2 + c2 )3 + (a2 + b2 + c2 )2 (ab + bc + ca) ≥ 3(a2 + b2 + c2 )abc(a + b + c) + 27a2 b2 c2 .
For any non-negative real numbers x, y, z, let
X
[x, y, z] = ax by cz .
sym
4
We recall Muirhead’s Majorization inequality, which states that if x, y, z, u, v, w are non-negative real
numbers such that x ≥ y ≥ z, u ≥ v ≥ w, x + y + z = u + v + w, x ≥ u, x + y ≥ u + v, then
[x, y, z] ≥ [u, v, w]. [x, y, z] is said to majorize [u, v, w]. Note that
1
(a2 + b2 + c2 )3 = [6, 0, 0] + 3[4, 2, 0] + [2, 2, 2],
2
1
(a2 + b2 + c2 )2 (ab + bc + ca) = [5, 1, 0] + [4, 1, 1] + [3, 3, 0] + 2[3, 2, 1],
2
3
3(a2 + b2 + c2 )abc(a + b + c) = [4, 1, 1] + 3[3, 2, 1],
2
9
27a2 b2 c2 = [2, 2, 2].
2
Hence, after cancellation of terms, the previous inequality becomes equivalent to
1 7
[6, 0, 0] + 3[4, 2, 0] + [5, 1, 0] + [3, 3, 0] ≥ [4, 1, 1] + [3, 2, 1] + [2, 2, 2].
2 2
By Muirhead’s inequality, [5, 1, 0] ≥ [4, 1, 1], [3, 3, 0] ≥ [3, 2, 1], 21 [6, 0, 0] + 3[4, 2, 0] ≥ 21 [2, 2, 2] + 3[2, 2, 2] =
7
2 [2, 2, 2]. Hence, the inequality holds.
Source: Belarus Mathematical Olympiad 1999
Comments: You should always be able to solve any three-variable symmetric inequality, that can be
homogenized, has equality case a = b = c and/or a = b, c = 0 and its permutation solutions, and whose
variables are non-negative real numbers, using Muirhead’s and Schur’s inequality. Muirhead’s inequality
states that [x, y, z] ≥ [u, v, w] whenever [x, y, z] majorizes [u, v, w]. An equivalent version of Schur’s in-
equality is [x + 2, 0, 0] + [x, 2, 0] ≥ 2[x + 1, 1, 1]. For proofs of these inequalities, see Chapter 3.2 of [2].
You may be thinking that solving problems in this manner is not elegant and can be considered ugly.
But the truth of the matter is that a lot of mathematics is initially done by getting your hands dirty and
working with cumbersome expressions. Muirhead’s and Schur’s inequality is in the arsenal of every top
math olympian in the world and should always be used as a last resort to solve inequalities of this type.
To test your ability to use these inequalities, do try the following problem.
Exercise: Let a, b, c be positive real numbers such that a + b + c = 3. Prove that

1 1 1
2
+ 2 + 2 ≥ a2 + b2 + c2 .
a b c
A4 P (x) is a polynomial of odd degree with real coefficients. Show that the equation P (P (x)) = 0 has at
least as many distinct real roots, as the equation P (x) = 0.
Solution: Let x1 , · · · , xn be all of the distinct roots of the equation P (x) = 0. We want to show that
P (P (x)) = 0 has at least n distinct real roots.
For each i = 1, · · · , n, consider the equation P (x) = xi . It has at least one real root ai since P is a
polynomial of odd degree. Now, for i, j ∈ {1, 2, · · · , n}, i 6= j, if ai = aj , then P (ai ) = P (aj ). Therefore,
5
xi = xj , which is impossible since x1 , · · · , xn are pairwise distinct. Therefore, the ai ’s are also pairwise
distinct. But each ai is a solution to P (P (x)) = 0, since P (P (ai )) = P (xi ) = 0. Hence the equation
P (P (x)) = 0 has at least n distinct real roots a1 , a2 , ..., an . The result follows.
Source: Russian Mathematical Olympiad 2000
Comments: When polynomials in question have certain properties such as having odd degree, a question
you should ask yourself is what are the differences between polynomials of odd degree and those of even
degree? The answer will often give you clues as to how to approach a problem. If you have not used the
fact that the polynomial has odd degree, you may not be approaching the problem in a correct way.
A5 Find all functions f : R 7→ R such that ∀x, y, z ∈ R, we have if x3 + f (y) · x + f (z) = 0, then
f (x)3 + y · f (x) + z = 0.
Solution: We will call x3 + f (y) · x + f (z) = 0 relation (A) and f (x)3 + y · f (x) + z = 0 relation (B).
We first prove that f is surjective. Let m be an arbitrary real number. We will show that f (x) = m
for some x ∈ R. Since (A) is a cubic polynomial in x, given any fixed y, z ∈ R, there exists x ∈ R such
that (x, y, z) satisfies (A). We choose y = 1, z = −m3 − m. We can choose x ∈ R such that (x, y, z)
satisfies (A). Therefore, (x, y, z) satisfies (B), i.e. f (x)3 + f (x) − (m3 + m) = 0. This implies that
(f (x) − m)(f (x)2 + m · f (x) + m2 + 1) = 0. The latter factor, as a quadratic equation in f (x), has negative
discriminant, hence cannot be zero. Therefore, f (x) = m. This proves that f is surjective.
We now show that f (x) = 0 if and only if x = 0. Since f is surjective, f (r) = 0 for some r ∈ R.
Note that (x, y, z) = (0, a, r) satisfies (A) for any choice of a ∈ R. Hence, (0, a, r) satisfies (B), i.e.
f (0)3 + af (0) + r = 0 holds for all a ∈ R. Therefore, af (0) = r − f (0)3 for all a ∈ R. The right hand side
is a constant. Therefore, f (0) = 0. This implies r = 0. Hence, f (x) = 0 if and only if x = 0.
We now show that f is injective. Assume f (a) = f (b) for some a, b ∈ R. Take any z 6= 0. Then f (z) 6= 0.
There exists x for which x3 + f (a)x + f (z) = 0. Note that x 6= 0, since f (z) 6= 0. Then (x, a, z) and
(x, b, z) satisfy (A), since f (a) = f (b) and x3 + f (a)x + f (z) = 0. Therefore, (x, a, z), (x, b, z) satisfy (B),
i.e. x3 + af (x) + z = x3 + bf (x) + z = 0, hence (a − b)f (x) = 0. But x 6= 0, implying f (x) 6= 0. Hence
a = b. Therefore, f is injective.
We now show that (A) and (B) are equivalent, i.e. (x, y, z) satisfies (A) if and only if (x, y, z) satisfies (B).
From the condition given in the problem, (x, y, z) satisfies (A) implies (x, y, z) satisfies (B). To prove the
converse, let x, y, z ∈ R be a triple such that (x, y, z) satisfies (B), i.e. we have have f (x)3 +y ·f (x)+z = 0.
Let z 0 ∈ R such that (x, y, z 0 ) satisfies (A). This is possible since f is surjective. Therefore, (x, y, z 0 ) sat-
isfies (B), i.e. z 0 = −f (x)3 − yf (x). But this term is also equal to z. Therefore, z 0 = z, implying (x, y, z)
satisfies (A). Therefore, (A) and (B) are equivalent.
Take (x, y, z) such that x = 1, f (y) = −1, z = 0. Choosing such a y is possible since f is surjective. Then
(x, y, z) satisfies (A). Hence, f (1)3 + yf (1) = 0. Therefore, y = −f (1)2 , which implies f (−f (1)2 ) = −1,
by the definition of y. Take (x, y, z) such that x = 1, y = 0, f (z) = −1. Then (x, y, z) satisfies (A). Hence,
f (1)3 + z = 0, implying z = −(f (1))3 . Therefore, f (−f (1)3 ) = −1. But f is injective and f (z) = 1.
6
Therefore, −f (1)2 = −f (1)3 , or equivalently, f (1)2 (f (1) − 1) = 0. Since f is injective, f (1) 6= f (0) = 0.
Therefore, f (1) = 1.
Take x = 1, z = −y − 1. Since f (1) = 1, (x, y, z) satisfies (B). Since (A) and (B) are equivalent, (x, y, z)
satisfies (A), i.e.
f (−y − 1) = −f (y) − 1, ∀y ∈ R. (1)
Substitute y = 0 into (1) yields

f (−1) = −1. (2)
Take x = −1, z = y + 1. Since f (−1) = −1, (x, y, z) satisfies (B), which implies (x, y, z) satisfies (A), i.e.
f (y + 1) = f (y) + 1, ∀y ∈ R. Inductively, we can show that
f (y + k) = f (y) + k, ∀y ∈ R, k ∈ Z. (3)
Because f (0) = 0, from (3), we have that
f (x) = x, ∀x ∈ Z. (4)
From (1) and (3), we get that f (−y − 1) = −f (y) − 1 = −(f (y + 1) − 1) − 1 = −f (y + 1). Therefore,
f (y) = f (−y), ∀y ∈ R. (5)
Let x, y be any fixed real numbers and let z = −f (x)3 − yf (x). Then (x, y, z) satisfies (A). Hence, (x, y, z)
satisfies (B), i.e. x3 + f (y)x + f (z) = 0. Therefore, f (z) = −x3 − f (y)x. By the definition of z, we now
have
f (−f (x)3 − yf (x)) = −x3 − f (y)x, ∀x, y ∈ R. (6)
Using (5), we get that

f (f (x)3 + yf (x)) = x3 + f (y)x, ∀x, y ∈ R. (7)
If x is an integer, then by (4), f (x) and x3 are also integers. Therefore, by (3) and (7), we have x3 +xf (y) =
f (f (x)3 + yf (x)) = f (x)3 + f (yf (x)) = x3 + f (yx). (We substitute y ← yf (x) and k ← f (x)3 into (3) for
the second last assertion and the fact that f (x) = x for all x ∈ Z, for the last assertion.) Therefore,
xf (y) = f (yx), ∀x ∈ Z, y ∈ R. (8)
Now, if m, n are integers with n 6= 0, then by (8), we have

m nx + m f (nx + m) nf (x) + m m
f x+ =f = = = f (x) + .
n n n n n
Hence,
f (x + q) = f (x) + q, ∀x ∈ R, q ∈ Q. (9)
Substituting x = 0 into (9) gives us

f (q) = q, ∀q ∈ Q. (10)
Substituting y = −f (x)2 into (7) gives us 0 = f (0) = −x3 − f (−f (x)2 )x. Therefore, f (−f (x)2 ) = −x2 .
By (5), we have
f (f (x)2 ) = x2 , ∀x ∈ R. (11)
7
Let c > 0 be an arbitrary real number. Then since f is surjective, there exists d ∈ R such that c = f (d)2 .
Since c 6= 0, f (0) = 0 and f is injective, d 6= 0. Therefore, by (11), we have f (c) = f (f (d)2 ) = d2 > 0.
Hence,
c > 0 ⇒ f (c) > 0, ∀c ∈ R (12)
Assume for some x, we have f (x) < x. Then there exists a rational number r for which f (x) < r < x.
Then by (12), we have f (x − r) > 0 and by (9), we have r > f (x) = f (r + (x − r)) = r + f (x − r) > r.
This is a contradiction. So we cannot have f (x) < x. Similarly, we cannot have f (x) > x. So the only
possibility for f is f (x) = x for all x ∈ R. It is easy to check f (x) = x satisfies the conditions of the
problem and so is the only solution.
Source: German Team Selection Test 2009
Comments: This problem is difficult in the sense that the solution requires many steps. However, every
step demonstrates an important technique that you should know to solve functional equations.
The first thing you should almost always do is to find solutions to the functional equation by inspection.
The most common ones are f (x) = c for some constant c, f (x) = x + c for some constant c, quadratic
equations and f (x) = x1d for some integer d.
You next want to establish properties of the functional equation, i.e. what is f (0)? If f (x) = 0, does it
mean x = 0? Is f one-to-one? Is f surjective? (Don’t do the latter two questions if one of the solutions
to the functional equation is, say, f (x) = x2 . You already know that this function is neither one-to-one
or surjective over the reals.)
Another technique is solving the problem over the rationals first, and then use detailed (and careful)
continuity arguments to solve the problem over all of the reals.
8
Number Theory
N1 Let n be a positive integer. The sum of the positive divisors of n is s. Prove that the sum of the reciprocal
of these divisors is ns .
Solution: Let d1 , · · · , dt be the positive divisors of n with 1 = d1 < d2 < · · · < dt = n. Note that
n = di dt+1−i for each i. Therefore, d1i = dt+1−i
n . Hence,
t t t t
X 1 X dt+1−i 1X 1X s
= = dt+1−i = di = .
d
i=1 i i=1
n n i=1 n i=1 n
Source: Original
Comments: Always remember that the positive divisors√of n come in pairs as shown in this solution.
The exception is when n is a perfect square. In this case, n is paired with itself.
N2 A rational number x is written on a blackboard. In each step, you erase x and replace it with either x + 1
or − x1 . (If x = 0, you must replace x with x + 1). Prove that for any rational number p, if p currently
appears on the blackboard, then you can make 0 appear in a finite number of steps.
Solution: By adding 1 repeatedly to p (which is a legal step), we may assume that p ≥ 0. If p = 0,

we are already done. Let p = m n , where m, n > 0. I claim that after a finite number of steps, you can
obtain a rational number with a non-negative numerator non-negative which is strictly smaller than m.
−n
Let n = qm − r where 0 ≤ r < m. We erase m n and replace it with m . By adding 1 to this q times, we get
r
a fraction m and 0 ≤ r < m. We obtain a number with a non-negative numerator which is smaller than
m. Repeating this procedure gives us fractions with strictly smaller non-negative numerators. Hence, we
can eventually obtain a rational number with numerator 0, which is the number zero.
Source: John Conway’s Talk on Knot Theory
Comments: This is motivated by the Euclidean algorithm and using a general technique in number
theory called descent. A strategy for this type of problem generally involves solving this problem for
certain values of p with small numerator and denominator. It should be pretty natural to try to lower the
numerator and/or denominator of the number on the board. I will leave as an exercise that starting at 0,
you can make p appear after a finite number of steps.
Another note is that we used a variant of the division algorithm. Instead of writing n = qm + r with
0 ≤ r < m, which is the more traditional method, we wrote it as n = qm − r with 0 ≤ r < m to make the
problem easier. The other useful variant is writing n as n = qm + r where −m/2 < r ≤ m/2.
N3 Find all triples of positive integers (a, b, c) such that a2 + 2b+1 = 3c .
Solution: The answers are (1, 2, 2) and (7, 4, 4).
9
Since 2b+1 is even and 3c is odd, a2 is odd. Therefore, a2 ≡ 1 mod 4. Since 2b+1 ≡ 0 mod 4, 3c ≡ 1 mod
4. This implies 3c ≡ 1 mod 4. Hence, (−1)c ≡ 1 mod 4. We conclude that c is even. Therefore,
c c
2b+1 = (3 2 − a)(3 2 + a).
c c
Hence, 3 2 − a and 3 2 + a are both powers of 2. Since a is odd, both of these terms are even, so none
c c
of these terms are 1. Hence, we can let s, t be positive integers such that 3 2 − a = 2s and 3 2 + a = 2t .
t s t−1 s−1
Subtracting the former from the latter yields 2a = 2 − 2 . Hence, a = 2 − 2 . Since a is odd, s = 1.
c c
Therefore, a = 2t−1 − 1. Substituting this into 3 2 + a = 2t yields 3 2 + 2t−1 − 1 = 2t , or equivalently,
c
3 2 − 1 = 2t−1 .
c c
If t = 1, then 3 2 = 2, which has no integer solutions. If t = 2, then 3 2 − 1 = 2. Hence, c = 2,
a = 2t−1 − 1 = 22−1 − 1 = 1. Therefore, 1 + 2b+1 = 32 , implying b = 2. Hence, (1, 2, 2) is a solution to the
c
equation and is easily verified to be a solution. Otherwise, t > 2, implying 4 | 2t−1 . Hence, 3 2 ≡ 1 mod
c
4, which implies (−1) 2 ≡ 1 mod 4. Hence, 2c is even. Therefore,
c c
(3 4 − 1)(3 4 + 1) = 2t−1 .
c c
From this, we have that both 3 4 − 1, 3 4 + 1 are powers of 2 and are consecutive even integers. This is
c
only possible when the smaller integer is 2 and the larger integer is 4. Therefore, 3 4 − 1 = 2, implying
c = 4. Therefore, 32 − 1 = 2t−1 = a + 1, or equivalently, a = 7. Finally, 72 + 2b+1 = 34 , implying b = 4.
Therefore, (7, 4, 4) is a solution to the equation and is easily verified to be a solution.
Source: Italian Team Selection Test 2008
Comments: One technique to solve Diophantine equations is consider reduce everything modulo a cer-
tain positive integer. The existence of a square and a power of 2 suggests that mod 4 and mod 8 are
possibilities. Problems like this test your comfort level with divisibility problems. You should always try
to find as many solutions by inspection as you can as a first step. (However, I do not expect you to initially
find the solution (7, 4, 4).)
Try to use modular arithmetic to solve the following problem.
Exercise: Let m, n be positive integers. Find the minimum possible value of |12m − 5n |.
N4 Find all positive integers that can be written in the form

m2 + n2 + 1
,
mn
for all positive integers m, n.
Solution: The answer is 3.
Let S = {(m, n) | mn divides m2 + n2 + 1, m, n ∈ N} and
m2 + n2 + 1
f (m, n) = .
mn
10
Note that (m, n) ∈ S if and only if (n, m) ∈ S.
If m = 1, then n divides n2 + 2. This implies n | 2. Therefore, n = 1 or n = 2. Both yield the property

2 2
that m +n
mn
+1
= 3. This holds similar for n = 1. Therefore, if m = 1 or n = 1, then f (m, n) = 3.
We next consider the case m = n. If (m, n) ∈ S and m = n, then m2 divides 2m2 + 1. Therefore, m2 | 1.
Therefore, m = 1. Hence, if (m, m) ∈ S, then m = 1. This also yields that f (m, n) = 3.
m2 +n2 +1
Now, suppose (m, n) ∈ S with 1 < m < n. Suppose k = mn . I claim there exists 0 < n0 < m such
that (m, n0 ) ∈ S. Rewriting the equation gives us
n2 − kmn + (m2 + 1) = 0.
Hence, n is an integer root of the equation x2 − kmx + (m2 + 1) = 0. Let n0 be the other root. Since
n + n0 = km, which is an integer, n0 is an integer. Since nn0 = m2 + 1 > 0 and n > 0, n0 > 0. Note that
f (m, n) = f (m, n0 ), by rewriting the quadratic equation back to the form f (m, n0 ). Therefore, (m, n0 ) ∈ S.
I claim that n0 < m. Suppose on the contrary that n0 ≥ m. Since (m, n0 ) ∈ S and n, m > 1, n0 6= m by
the argument in the case m = n. Suppose n0 > m. Let n = m + a, n0 = m + b for some positive integers
a, b. Then nn0 = (m + a)(m + b) = m2 + 1, since the product of the roots of x2 − kmx + (m2 + 1) is m2 + 1.
Hence, m(a + b) + ab = 1, which is impossible since a, b ≥ 1. Therefore, n0 < m. This proves the claim.
Hence, (m, n0 ) ∈ S, which implies (n0 , m) ∈ S. Note that k = f (m, n) = f (m, n0 ) = f (n0 , m). If n0 = 1,
then from an earlier argument, k = 3. Otherwise, we repeat this process for 1 < n0 < m. Every time we
perform the previous paragraph, min{m, n} strictly decreases and remains positive and f (m, n) remains
constant. Hence, eventually, at least one of m, n = 1. By an earlier argument, this results in k = 3. Since
f (m, n) remains constant at every step, we conclude that f (m, n) = 3. This completes the problem.
Source: Arthur Engel’s Problem-Solving Strategies
Comments: The technique used in this problem is called Vieta-jumping or descent. It is extremely
helpful in solving Diophantine equations with quadratic terms. By letting S be a solution set, you can
claim statements of the form if (a, b) ∈ S, then (c, d) ∈ S for certain choices of a, b, c, d. By doing so, you
can create a series of solutions for which min{m, n} is decreasing. Try to solve the following problem from
the 1988 International Mathematical Olympiad using Vieta-jumping.
Exercise: Suppose m, n are positive integers such that mn + 1 divides m2 + n2 . Prove that
m 2 + n2
mn + 1
N5 Find all prime numbers p such that the following statement is true: there are exactly p ordered pairs of
integers (x, y) such that 0 ≤ x, y < p and y 2 ≡ x3 − x mod p.
Solution: The answers are p = 2 and all primes p such that p ≡ 3 mod 4.
11
If p = 2, then we can check that the only pairs that work are (0, 0) and (1, 0). There are exactly two
solutions. Hence p = 2 is an answer.
We may assume now that p is odd. For a ∈ {0, 1, · · · , p − 1}, we call a a quadratic residue modulo p if
and only if n2 ≡ a mod p for some integer n. Otherwise, we call a a quadratic non-residue. We recall
that the product of two quadratic residues is a quadratic residue, and the product of two quadratic non-
residues is a quadratic residue. Also, the product of a quadratic residue and a quadratic non-residue, is a
quadratic non-residue. Also recall that −1 is a quadratic residue of p if and only if p ≡ 1 mod 4 or p = 2. (*)
Note that (−1, 0), (0, 0), (1, 0) are the only solutions involving y = 0 and x ∈ {−1, 0, 1}. There are three
such solutions. We will henceforth only consider the cases when x ∈ {2, · · · , p − 2} and y 6= 0.
We will call x ∈ {2, · · · , p − 2} solvable if there exists y such that y 2 ≡ x3 − x mod p.
If p ≡ 1 mod 4, then note that x3 − x = −1 · ((−x)3 − (−x)). By the arguments made in (*), since −1 is
a quadratic residue mod 4, x is solvable if and only if −x is solvable. For each x ∈ {2, · · · , p − 2} which is
solvable, there are exactly two values of y such that (x, y) is a solution to the original equation. This is
because (x, y) is a solution if and only if (x, −y) is a solution. Note that x 6≡ −x and y 6≡ −y mod p since
p is odd and y 6= 0. Hence the number of values of x ∈ {2, · · · , p − 2} which is solvable is even and each
solvable x yields two values of y. Hence, the number of ordered pair solutions with x ∈ {2, · · · , p − 2} is
divisible by 4. Including the three aforementioned solutions, the number of solutions (x, y) is congruent
to 3 mod 4. But p ≡ 1 mod 4. Therefore, the number of solutions (x, y) cannot equal to p.
If p ≡ 3 mod 4, again, note that x3 − x = −1 · ((−x)3 − (−x)). Since −1 is a quadratic non-residue

mod p and the product of a quadratic residue and a quadratic non-residue is a quadratic non-residue, x is
solvable if and only if −x is not solvable. Hence, exactly half of the value of x ∈ {2, · · · , p − 2} are solvable
and using a similar argument as in the case p ≡ 1 mod 4, each solvable x yields two solutions. Therefore,
there are p−3
2 · 2 = p − 3 solutions. Adding the aforementioned three solutions, there are p solutions total.
Hence, p ≡ 3 mod 4 is a solution to the problem.
Source: Turkish Team Selection Test 2005
Comments: The facts in the paragraph marked (*) are important topics to learn for number theory,
especially the fact that n2 ≡ −1 mod p has a solution if and only if p = 2 or p ≡ 1 mod 4. Please read
Theorem 4.9 and Chapter 7 of [3] for more details.
Another important point is that the very least you should be able to do for this problem, even without
knowledge of quadratic residues and of fact (*), is to conjecture your answer for p by solving this problem
for small primes p, say up to p = 17. The motivation behind this given solution arises from the work you do
for small values of p. Only then would you be able to consider the differences between primes 1 and 3 mod
4. The least that you should be able to write, is something along the lines of ”I conjecture that the solution
to this problem are p = 2 and all primes p ≡ 3 mod 4”. Of course, now that you have read and under-
stood this solution, I will expect much more than this. :) Using fact (*), you can try the following exercise.
Exercise: Prove that there are no pairs of positive integers a, b such that 4ab − a − b is a perfect square.
12
Geometry
G1 A circle C and a point P are given on the same plane. Given any point Q on the circumference of C, let
M be the midpoint of P Q. Find the locus of point M , i.e. find all possible locations for point M .
Solution 1: The locus of M is exactly the dilation of C about point P with ratio 12 . The image is therefore
the circle with radius half of that of C whose centre is the midpoint of P and O, where O is the centre of C.
|P M | |P N |
Solution 2: Let O be the centre of C, r the radius of C and N the midpoint of P O. Since |M Q| = |N O| = 1,
|M N | 1 r r
we have |QO| = 2. Hence, |M N | = 2. Therefore, M lies on the circle with radius 2 centered at N .
Conversely, let M be any point on the circle centered at N with radius 2r . Let Q be on the circumference
of C such that QO is parallel to M N . Then |M N| |P N | 1
|QO| = |P O| = 2 , which implies P, M, Q are collinear and
|P M |
|P Q| = 12 . Therefore, M is the midpoint of P Q.
Source: Original
G2 Let A be a point outside of a circle C. Two lines pass through A, one intersecting C at B, C, with B closer
to A than C, and the other intersecting C at D, E, with D closer to A than E. The line passing through
D parallel to AC intersects C a second time at F and the line AF intersects C a second time at G. Let
M = EG ∩ AC. Prove that
1 1 1
= + .
|AM | |AB| |AC|
Solution: Note that ∠GAM = ∠F AM = ∠AF D = ∠GF D = ∠GED, the latter assertion following
|M A| |M E|
from that DEF G is cyclic. Therefore, ∆GAM ∼ ∆AEM . Hence, we have |M G| = |M A| , which implies
2
|M A| = |M G| · |M E| = |M B| · |M C|. The assertion follows from that DEF G is cyclic. We shall call
this property (*).
Note that M lies on the segment AB. Therefore, |AB| = |M A| + |M B| and |AC| = |M A| + |M C|. Hence,
1 1 1 |AB| + |AC| 2|M A| + |M B| + |M C|

= + = = .
|AM | |AB| |AC| |AB| · |AC| (|M A| + |M B|)(|M A| + |M C|)
This equality holds if and only if (|M A| + |M B|)(|M A| + |M C|) = 2|M A|2 + |M A||M B| + |M A||M C| ⇔
|M B| · |M C| = |M A|2 , which is true by property (*).
Source: Romanian Team Selection Test 2006
Comments: If you are not familiar with Power of a Point, please read Chapter 6 of [1]. It is an important
and powerful technique to solve certain geometry problems, especially those involving circles and tangent
lines. Power of a Point is the motivation for this solution.
13
G3 Let ABC be a triangle and D be a point on side BC. The internal angle bisector of ∠ADB and that of
∠ACB intersect at P . The internal angle bisector of ∠ADC and that of ∠ABC intersect at Q. Let M
be the midpoint of P Q. Prove that |M A| < |M D|.
Solution: Note that P is the excentre of 4ACD opposite C, and Q is the excentre of 4ABD opposite
B. Therefore, ∠P AD = 180−∠CAD
2 and ∠QAD = 180−∠BAD
2 .
Note also that ∠QDP = ∠QDA + ∠ADP = 12 (∠BDA + ∠CDA) = 90o . Therefore, |M P | = |M Q| =
|M D|. Consider the circle centred at M passing through P, Q, D. To prove that |M A| < |M D|, it suffices
to show that A is in the interior of the circle, i.e. ∠QAP is obtuse. But
180 − ∠BAD 180 − ∠CAD ∠BAD + ∠CAD
∠QAP = ∠QAD + ∠P AD = + = 180 − = 180 − ∠BAC
2 2 2
and this angle is greater than 90o (and less than 180o ). Therefore, |M A| < |M D|.
Source: Original and similar to one on Irish Mathematical Olympiad 2005
Comments: It is important to observe excentres in a diagram. Their existence, along with incentres,
often simplifies problems involving angle bisectors.
G4 A convex quadrilateral ABCD has |AD| = |CD| and ∠DAB = ∠ABC < 90◦ . The line through D and
the midpoint of BC intersects line AB in point E. Prove that ∠BEC = ∠DAC.
Solution: Let F be on side AB such that |DA| = |DF |, which is also equal to |DC|. Since ∠AF D =
∠DAF = ∠DAB = ∠ABC, DF ||BC. Extend DF to meet EC at G. Since DF ||BC and ED passes
through the midpoint of BC, D is the midpoint of F G. Hence, |DA| = |DC| = |DF | = |DG|.
Hence, ACGF is cyclic. Then ∠F AC = ∠F GC = ∠DGC = ∠BCE. Since ∠DAB = ∠ABC,
∠DAC = ∠DAB − ∠F AC = ∠ABC − ∠BCE = ∠BEC, as desired.
Source: Bulgarian Mathematical Olympiad 1998
Comments: Construction problems in geometry are fun. Of course, there are solutions that does not
involve extra constructions given in this solution. However, the construction of point F on AB is quite nat-
ural given that |AD| = |CD| and ∠DAB = ∠ABC. You will be given three equal lengths in AD, CD, F D
and that F D is parallel to AB. Many new properties of the diagram arise from this one point. Wow! :)
G5 Let ABC be a triangle with |AB| > |AC|. Let its incircle touch side BC at E. Let AE intersect this
incircle again at D. Let F be the second point on AE such that |CE| = |CF |. Let CF intersect BD at
G. Prove that |CF | = |F G|.
Solution: Let the line tangent to the incircle at D intersect BC (extended) at P . Since |P E| = |P D|
and |CE| = |CF |, CF ||P D.
14
By Menelaus’ Theorem on ∆BCG, we have that
|GD| |BE| |CF |

· · = 1.
|DB| |EC| |F G|
|GD| |CP |
Since CF ||P D, |DB| = |P B| . Therefore, to prove that |CF | = |F G|, it suffices to prove that
|CP | |CE|
= .
|P B| |EB|
Note that the polar of P with respect to the incircle is line DE, which also contains A. Therefore, the
polar of A contains P . Hence, the polar of A is XY , where X, Y are points where the incircle touches
AC, AB, respectively. Hence, P lies on XY . Since |AX| = |AY |, |BY | = |BE|, |CE| = |CX|, we have
that
AY BE CX
· · = 1. (13)
Y B EC XA
Hence, AE, BX, CY are concurrent by Ceva’s Theorem. Since X, Y, P are collinear, by Menelaus’ Theorem
on ∆ABC, we have
AY BP CX
· · = −1. (14)
Y B P C XA
Comparing equations (13) and (14), yields
|CP | |CE|
= ,
|P B| |EB|
as desired.
Source: China Team Selection Test 2008
Comments: Ceva’s and Menealus’ Theorem are essential pieces of knowledge for proving concurrency
and collinearity involving triangles. Please learn these theorems if you haven’t done so already.
Poles and polars have come to the rescue. Those familiar with poles and polars and harmonic conjugates
will realize that (E, P ) divides (B, C) harmonically. To learn more about these powerful techniques, please
consult Chapter 10 and 11 of [1].
15
Combinatorics
C1 Find the number of subsets of {1, 2, · · · , 10} that contain its own size. For example, the set (1, 3, 6) has 3
elements and contains 3.
Solution: The answer is 512.
For each k ∈ {0, 1, · · · , 10}, we want to find how many elements there are of size k. Note that k 6= 0
since 0 6∈ {1, · · · , 10}. Every
subset of size k must contain k. The other k − 1 elements can be anything.
9
Therefore, there are k−1 subsets of size k that contains k. Therefore, the number of subsets that contain
its own size is
10 X 9
X 9 9
= = 29 = 512.
k−1 k
k=1 k=0
Source: Original
C2 A sequence of non-negative integers is defined by G0 = 0, G1 = 0 and Gn = Gn−1 + Gn−2 + 1 for every

n ≥ 2. Prove that for every positive integer m, there exists a positive integer a such that Ga , Ga+1 are
both divisible by m.
Solution: Consider each pair (Gt , Gt+1 ) modulo m. Since there are at m2 pairs of integers (x, y) mod-
ulo m, the sequence (Gt , Gt+1 ) modulo m eventually repeats, by Pigeonhole Principle. It suffices to
prove that (Ga , Ga+1 ) = (0, 0) = (G0 , G1 ) modulo m, for some positive integer a. Suppose the sequence
{(Gt , Gt+1 } repeats the first time at index t = a. Then (Ga , Ga+1 ) = (Gj , Gj+1 ) modulo m for some
0 ≤ j < t. If j = 0, then (Ga , Ga+1 ) = (0, 0), implying Ga , Ga+1 are both divisible by m and we are
done. Otherwise, note that the pair that appears before (x, y) is (y, x + y − 1). In other words, given a
pair of consecutive terms in the sequence, the previous pair of consecutive terms is determined. Hence,
(Ga , Ga+1 ) = (Gj , Gj+1 ) implies (Gj−1 , Gj ) = (Ga−1 , Ga ), contradicting that the sequence repeats the
first time at index a. Therefore, Ga , Ga+1 are both divisible by m.
Source: Estonian Mathematical Olympiad 2007
Comments: Pigeonhole Principle is a powerful tool to solve certain existence type of combinatorics prob-
lems. It is an art to determine what are the pigeons and what are the holes.
C3 A set S of ≥ 3 points in a plane has the property that no three points are collinear, and if A, B, C are
three distinct points in S, then the circumcentre of ∆ABC is also in S. Prove that S is infinite.
Solution: Suppose on the contrary that S is finite. Consider the convex hull P of S, i.e. the smallest poly-
gon containing all of S (on the polygon’s boundary and interior). Since |S| ≥ 3, P contains at least three
vertices. Let AB be a side of P and C be any other vertex of P. Note that P is contained in the halfplane
of AB containing C and there are no points in S in the halfplane not containing C. If ∠ACB > 90◦ , then
the circumcircle of ∆ACB lies in the halfplane of AB not containing C, which does not contain any point
16
in S. Hence, the circumcircle of ∆ACB is not in S. This contradicts the assumption given in the problem.
Otherwise, ∠ACB is acute. We let O1 be the circumcentre of ∆ACB and recursively, define Oi+1 to be
the circumcentre of ∆AOi B. Hence, Oi ∈ S for all i ≥ 1. Hence, ∆AO1 B = 2∠ACB. We choose the
smallest index k such that ∠AOk B ≥ 90◦ . Therefore, ∠AOk B = 2∠AOk−1 B. Since ∠AOk−1 B ≤ 90◦ by
our choice of k, ∠AOk−1 B < 180◦ . If ∠AOk B = 90◦ , then Ok+1 lies on AB, contradicting the fact that
no three points in S are collinear. (Note that Oi 6= A, B for any i, since the circumcentre of a triangle
cannot be any of the vertices of the triangle.) Otherwise, 90 < ∠AOk B < 180◦ , which implies Ok+1 lies
on the halfplane of AB not containing Ok . Hence, Ok+1 cannot lie in S, but this contradicts the fact that
Oi ∈ S for all i ≥ 1. Hence, S cannot be finite, implying S is infinite.
Source: Unknown
Comments: When you are given a finite set of points in the plane in a problem, you should always
consider the convex hull of the set. It is likely to simplify the problem. Try the following problem from
the 1999 International Mathematical Olympiad by considering the convex hull.
Exercise: Find all finite set of points S such that for every pair of points A, B ∈ S, the perpendicular
bisector of A and B is an axis of symmetry for S.
C4 Let n, k be positive even integers. A survey was done on n people where on each of k days, each person was
asked whether he/she was happy on that day and answered either ”yes” or ”no”. It turned out that on any
two distinct days, exactly half of the people gave different answers on the two days. Prove that there were
at most n− nk people who answered ”yes” the same number of times he/she answered ”no” over the k days.
Solution: Consider a grid with k rows and n columns. Label the days from 1 to k and the people from
1 to n. We place a 1 in the ith row and j th column if person j is happy on day i and a 0 otherwise. Since
on any two days, exactly half of the people gave different answers on the two days, we have that for any
two rows, exactly half of the columns differ. Hence, the number of pairs of squares in the same column
that differ, is k2 · n2 . For each person that answered yes the same number of times as he/she answered no,
the number of pairs of squares in the person’s column that differ is ( k2 )2 . Hence, the maximum number of
people that could have answered yes the same number of times as he/she answered no, is
k n
nk(k−1)
2 2 4 n(k − 1) n
= k2
= =n− ,
( k2 )2 4
k k
as desired.
Source: Iran Mathematical Olympiad 2006
Comments: An important key to solving combinatorics problem is find out what you are suppose to be
counting. If you can count a certain object in two different ways, you can yield some extremely neat and
powerful identities and relationships. If you take the time to solve enough combinatorics problems, it will
become more natural as to what objects you should be counting. Use this technique to solve the following
problem from the 1998 International Mathematical Olympiad
17
Exercise: In a contest, there are m candidates and n judges, where n ≥ 3 is an odd integer. Each
candidate is evaluated by each judge as either pass or fail. Suppose that each pair of judges agrees on at
most k candidates. Prove that
k n−1
≥ .
m 2n
C5 There are n ≥ 5 people in a room, where each pair is classified as friends or strangers. No three people are
mutually friends. There also exist an odd number of people P1 , · · · , Pm such that Pi is friends with Pi+1
for all i ∈ {1, · · · , m}, where the indices are taken modulo m. Prove that there exists one person who is
friends with at most 2n/5 people.
Solution: Given a positive integer m ≥ 3, we will call an ordered set of m people P1 , · · · , Pm a cycle if
Pi is friends with Pi+1 for all i ∈ {1, · · · , m}, where the indices are taken modulo m. If m is odd, we call
a cycle an odd cycle. Given the condition in the problem, an odd cycle exists. Consider the odd cycle in
the room P1 , · · · , Pg with the smallest number of people. Since no three people who are mutually friends,
the odd cycle must contain at least five people, i.e. g ≥ 5
I claim that Pi is not friends with Pj if |i − j| =

6 1 mod g. Without loss of generality, suppose P1 is friends
with Pj for some j 6= 2, j 6= g. Consider the set of vertices (P1 , · · · , Pj ) and (P1 , Pj , Pj+1 , · · · , Pg ). These
are both cycles and the sum of the the two sets is j + (g − j + 2) = g + 2, which is odd. Therefore, one of
these cycles is odd, and smaller than g. This contradicts the minimality of g.
I claim that every person outside of the cycle is friends with at most two people inside the cycle. Suppose
there exists a person A outside of the cycle who is friends with three people in the cycle, say Pi , Pj , Pk
with i < j < k. Note that no two of Pi , Pj , Pk are friends, since otherwise, A and these two people are
three mutual friends, which is not allowed. Consider the cycle drawn as a graph with Pi , Pj , Pk appearing
in clockwise order. Let x be the distance of the path between Pi and Pj on the cycle not containing
Pk . Define y, z analogously. Therefore, x + y + z = g. Since no two of Pi , Pj , Pk are friends, x, y, z > 1.
Therefore, max{x, y, z} < g−2. Since g is odd, at least one of x, y, z is odd. Without loss of generality, sup-
pose x is odd. Then APi Pi+1 · · · Pj A is an odd cycle of length x + 2 < g, contradicting the minimality of g.
Therefore, every person outside of the cycle is friends with at most two people inside the cycle. Hence,
the number of pairs of friends with one person in the pair in the cycle and the other person in the
pair outside of the cycle, is at most 2(n − g). Therefore, there exists a person inside the cycle who
is friends with at most 2(n−g)
g friends outside of the cycle. Since this person is friends with exactly two
2(n−g) 2n 2n
people inside the cycle, this person is friends with at most g +2 = g ≤ 5 , since g ≥ 5, as desired.
Source: Unknown
Comments: Knowledge in graph theory is of utmost importance to have to solve this problem. Please
read [4] to learn what graphs are and their important properties.
18
Winter Camp 2010 Buffet Contest 1
IMO Winter Camp 2010 Buffet Contest

Algebra
A1 From an infinite arithmetic sequence a1 , a2 , a3 · · · of positive real numbers, some (possibly infinitely many)
terms are deleted, obtaining an infinite geometric sequence 1, r, r2 , r3 , · · · for some real number r > 0. Prove
that r is an integer.
A2 Find all functions f : R → R such that
f (f (x) + y) = f (x2 − y) + 4f (x)y
for all x, y ∈ R.
A3 Find all finite sets A of distinct non-negative real numbers for which:
(a) the set A contains at least four numbers.

(b) for any 4 distinct numbers a, b, c, d ∈ A, the number ab + cd ∈ A.
Number Theory
N1 Let a, b be rational numbers such that a + b and a2 + b2 are integers. Prove that a, b are both integers.
N2 Find all pairs of positive integers (a, b) such that the sequence of positive integers a1 , a2 , a3 , · · ·, formed
by a1 = a, a2 = b and
an−1 + an−2
an = ,
gcd(an−1 , an−2 )
for n ≥ 3, is bounded. (A bounded sequence is a sequence for which there exists a positive real number
M such that an ≤ M for all n ∈ N.)
N3 For a positive integer n, let f (n) be the largest prime divisor of n. Prove that there are infinitely many
positive integers n such that
f (n) < f (n + 1) < f (n + 2).
1
Combinatorics
C1 Let n ≥ 2 be a positive integer. An n × n grid is filled with the integers 1, 2, · · · , n2 . Let t be the maximum
of the (positive) difference of the entries of two neighbouring squares, where two squares are said to be
neighbours if they share at least one vertex. Determine the minimum possible value of t in terms of n.
C2 A chessboard is tiled with 32 dominoes. Each domino covers two adjacent squares, a white and a black
square. Show that the number of horizontal dominoes with the white square on the left of the black square
equals the number of horizontal dominoes with the white square on the right of the black square.
C3 Let n > 1 be a positive integer. On each of 2n points around a circle we place a disk with one white side
and one black side. We may perform the following move: select a black disk, and reverse its two neighbors.
Find all initial configurations from which some sequence of such moves leads to a position where all disks
but one are white.
Geometry
G1 The altitude from A of triangle ABC intersects the side BC at D. A circle is tangent to BC at D,
intersects AB at M and N , and intersects AC at P and Q. Prove that
AM + AN AP + AQ
= .
AC AB
G2 Let P be a convex 2010-gon. The 1005 diagonals connecting opposite vertices and the 1005 lines connect-
ing the midpoints of opposite sides are concurrent. (i.e. all 2010 lines are concurrent.) Prove that the
opposite sides of P are parallel and have the same length.
G3 Two circles meet at A and B. Line ` passes through A and meets the circles again at C and D respectively.
Let M and N be the midpoints of arcs BC and BD which do not contain A, and let K be the midpoint
of CD. Prove that ∠M KN = 90◦ .
2
IMO Winter Camp 2010 Buffet Contest Solutions

Algebra
A1 From an infinite arithmetic sequence a1 , a2 , a3 · · · of positive real numbers, some (possibly infinitely many)
terms are deleted, obtaining an infinite geometric sequence 1, r, r2 , r3 , · · · for some real number r > 0. Prove
that r is an integer.
Solution: The arithmetic sequence contains 1. Let 1, d + 1, 2d + 1, · · · be the given arithmetic sequence
starting at 1. This sequence contains 1, r, r2 . Then r = 1 + md and r2 = 1 + nd for some integers
2
m, n. If d = 0, then r = 1, which is clearly an integer. Otherwise, m = r−1 d and n = r d−1 are in-
2
tegers. But r d−1 = r−1d · (r + 1). This implies r + 1 is a rational number. Hence, r is rational. But
r is a term in the arithmetic sequence 1, d + 1, 2d + 1, · · ·. This implies this arithmetic sequence has a
rational common difference and consequently, contains only rational numbers. Let f be the denominator
of the rational common difference. Then all terms in the arithmetic sequence have denominator at most f .
If r is not an integer, then if we write r = ab with gcd(a, b) = 1 and b > 1, then the denominators of
1, r, r2 , r3 , · · · are strictly increasing, and will eventually exceed f . But then these terms whose denomi-
nator is larger than f cannot be in the given arithmetic sequence. This is a contradiction. Therefore, r
must be an integer.
Source: IberoAmerican Mathematical Olympiad 2000
A2 Find all functions f : R → R such that
f (f (x) + y) = f (x2 − y) + 4f (x)y
for all x, y ∈ R.
Solution: The answers are f (x) ≡ 0, x2 .
Subsituting y = −f (x) yields
f (0) = f (x2 + f (x)) − 4f (x)2 , ∀x ∈ R. (1)
Substituting y = x2 yields
f (f (x) + x2 ) = f (0) + 4f (x)x2 , ∀x ∈ R. (2)
Comparing equations (1) and (2) yields
4f (x)2 = 4f (x)x2
for all x ∈ R. Equivalently, f (x)(f (x) − x2 ) = 0. Therefore, for each x ∈ R, f (x) = 0 or f (x) = x2 . I
claim that exactly one of these conditions hold for all x ∈ R. Suppose f (a) = 0 for some a 6= 0. I claim
that f ≡ 0. Then substituting x = a yields
f (y) = f (a2 − y),

2
for all y ∈ R. Suppose f (y) = y 2 for some y 6= 0, a2 . Then f (a2 −y) = y 2 6= 0. Hence, (f (a2 −y) = (a2 −y)2 ,
2
implying (a2 −y)2 = y 2 . This simplifies to a2 (a2 −2ay) = 0. This contradicts a 6= 0 and y 6= a2 . Therefore,
2 2
f (y) = 0 for all y 6= a2 . Choose z such that z 6= ±a. By the same argument, f (y) = 0 for all y 6= z2 .
3
a2 z2 2
Since 2 6= 2 , f ( a2 ) = 0. Hence, f ≡ 0.
Therefore, f (x) ≡ 0, x2 are the only candidate solutions. It remains to verify that both solutions work. If
f (x) ≡ 0, then f (f (x) + y) = 0 and f (x2 − y) + 4f (x)y = 0. If f (x) ≡ x2 , then f (f (x) + y) = (x2 + y)2
and f (x2 − y) + 4f (x)y = (x2 − y)2 + 4x2 y = (x2 + y)2 , as desired.
Source: Iranian Mathematical Olympiad 1999
Comments: In the step f (x)(f (x) − x2 ) = 0, a costly mistake is to conclude here that f (x) ≡ 0 and
f (x) ≡ x2 are the only solutions. It is still possible that f is a function that takes on 0 for some non-trivial
values of x and x2 for the other values of x. You must handle this with care in similar problems. Try
the following problem from the 2008 International Mathematical Olympiad where you have to handle a
similar situation.
Exercise: Find all functions f : (0, ∞) → (0, ∞) ( so f is a function on the positive real numbers) such
that
(f (w))2 + (f (x))2 w2 + x2
= ,
f (y 2 ) + f (z 2 ) y2 + z2
for all positive real numbers w, x, y, z such that wx = yz.
A3 Find all finite sets A of distinct non-negative real numbers for which:
(a) the set A contains at least four numbers.

(b) for any 4 distinct numbers a, b, c, d ∈ A, the number ab + cd ∈ A.
Solution: The answer is {0, x1 , 1, x} for some x > 0.
We first prove a lemma.
Lemma: Let a, b, c, d be non-negative real numbers such that a 0 and
(ac + bd) − (ad + bc) = (b − a)(d − c) > 0. End Proof of Lemma 1
Suppose n = |A| ≥ 5. Let a1 , · · · , an ∈ A with a1 < a2 < · · · < an . Then by the Lemma,
a1 a4 + a2 a3 < a1 a2 + a3 a4 < a1 a2 + a3 a5 < · · · < a1 a2 + a3 an ,
and
a1 a2 + a3 an < a1 a2 + a4 an < · · · < a1 a2 + an−1 an ,
and
a1 a2 + an−1 an < a1 a3 + an−1 an < · · · < a1 an−2 + an−1 an
and
a1 an−2 + an−1 an < a2 an−2 + an−1 an < · · · < an−3 an−2 + an−1 an .
There are (n − 2) + (n − 4) + (n − 4) + (n − 4) = 4n − 14 elements in these inequalities. Since n ≥ 5,
4n − 14 > n. Hence, the 4n − 14 terms are pairwise distinct and all in A. This is impossible since |A| = n
1 This result is simply the rearrangement inequality on four variables.
4
and 4n − 14 > n.
Therefore, |A| = 4. Let A = {a, b, c, d} with a < b < c < d. Let x = ad + bc, y = ac + bd, z = ab + cd.
Then by the Lemma, x < y < z. By property (b), x, y, z ∈ A. Since A = {a, b, c, d}, {x, y, z} ⊆ {a, b, c, d}.
Therefore, x = a or x = b.
If x = a, then ad + bc = a. Hence, bc = a(1 − d). Since bc > 0, d < 1. Note that ac + bd = b or c, implying
ac+bd ≥ b. Then b−a ≤ (ac+bd)−(ad+bc) = (b−a)(d−c). Hence, d−c ≥ 1, contradicting d < 1. Hence,
x 6= a. Since x < y < z and x, y, z ∈ A and A = {a, b, c, d} and x 6= a, we conclude that x = b, y = c, z = d.
Hence, ad + bc = b, ac + bd = c, ab + cd = d. Rewriting these equations give us ad = b(1 − c), bd =

c(1 − a), ab = d(1 − c). Subtracting the first equation from the third equation yields (1 − c + a)(d − b) = 0.
Since d > b, c − a = 1. Substituting this into the first equation yields a(d + b) = 0. Since d > b, a = 0 and
c = 1. Substituting this into the second equation yields bd = 1. Therefore, b = d1 . Since b < d, d1 < 1 < d.
It is easy to verify that that the set {0, d1 , 1, d} satisfy the given condition for d > 1. This finishes the
problem.
Source: Bulgarian Mathematical Olympiad 1998
Number Theory
N1 Let a, b be rational numbers such that a + b and a2 + b2 are integers. Prove that a, b are both integers.
Solution 1: Let a = xz for some integers x, z with z, > 0 and gcd(x, z) = 1. Then b = k − xz = kz−x z
for some integer k. Since gcd(kz − x, z) = gcd(x, z) = 1, the denominator of b in lowest terms is also z.
Hence, we can let b = yz with gcd(y, z) = 1.
Since a + b and a2 + b2 , (a + b)2 − (a2 + b2 ) = 2ab is also an integer. Hence, 2xy

z 2 is an integer. Since
gcd(x, z) = gcd(y, z) = 1, gcd(xy, z 2 ) = 1. Hence, z 2 divides 2. This implies z = 1. Therefore, a, b are
integers.
Solution 2: Using the same notation as Solution 1, we note that 2(a2 + b2 ) − (a + b)2 = (a − b)2 is an
integer. Since a, b is rational, a − b is rational. Hence, a − b is an integer. Since a + b is an integer, 2a, 2b
are integers. If at least one of a, b is an integer, then since a + b is an integer, the other of a, b is also an
integer and we are done. Otherwise, a, b are both not integers, implying a, b have denominator 2 when
expressed in lowest terms. Let a = x2 , b = y2 , where x, y are odd integers. Since a2 + b2 is an integer, 4
divides x2 + y 2 . But x2 , y 2 ≡ 1 mod 4 since x, y are odd. Hence, x2 + y 2 ≡ 2 mod 4, implying 4 - x2 + y 2 .
This is a contradiction. Therefore, a, b are integers.
Source: Russian Math Olympiad 2006 Variant
N2 Find all pairs of positive integers (a, b) such that the sequence of positive integers a1 , a2 , a3 , · · ·, formed
by a1 = a, a2 = b and
an−1 + an−2
an = ,
gcd(an−1 , an−2 )
for n ≥ 3, is bounded. (A bounded sequence is a sequence for which there exists a positive real number
M such that an ≤ M for all n ∈ N.)
Solution: The answer is (a, b) = (2, 2).
5
First note if for two consecutive terms ak , ak+1 satisfy gcd(ak , ak+1 ) = 1, then gcd(ak+2 , ak+1 ) = gcd(ak +
ak+1 , ak+1 ) = gcd(ak , ak+1 ) = 1. Iteratively, we have that gcd(an , an+1 ) = 1 for all n ≥ k. Hence, for
all n ≥ k + 2, an = an−1 + an−2 and so the sequence is strictly increasing since each term is a positive
integer. Hence, the sequence is not bounded. Henceforth, we will assume that no two consecutive terms
in the sequence are coprime.
Then for all positive integers n, we have an+2 ≤ an +a2

n+1
≤ max(an , an+1 ) with equality iff an = an+1 and
gcd(an , an+1 ) = 2. Hence max(an , an+1 ) is non-increasing. Since this number is always an integer, eventu-
ally this max becomes constant, so for some m, max(am , am+1 ) = max(am+1 , am+2 ) = max(am+2 , am+3 ) =
· · · = t. If am 6= am+1 , then am+2 < t, am+3 < t so max(am+2 , am+3 ) < t, which is impossible.
Hence am = am+1 = am+2 = · · · = t and gcd(am , am+1 ) = 2. Therefore, t = 2. Finally, we now find the
am−1 +am am−1 +2
values of am−1 , am−2 , · · · , a1 . Since 2 = am+1 = gcd(am−1 ,am )
= gcd(am−1 ,2)
≥ am−1
2
+2
. Hence, am−1 ≤ 2.
If am−1 = 1, then gcd(am−1 , am ) = 1, which is not allowed. Therefore, am−1 = 2. Hence, iteratively, we
have that am−1 = am−2 = · · · = a1 = 2.
So the only such sequence is the sequence where all terms are equal to 2. Hence, (a, b) = (2, 2) is the only
solution.
Source: Russian Math Olympiad 1999
N3 For a positive integer n, let f (n) be the largest prime divisor of n. Prove that there are infinitely many
positive integers n such that
f (n) < f (n + 1) < f (n + 2).
k k
Solution 1: Let p ≥ 5 be an odd prime number and let mk = p2 − 1 and nk = p2 + 1 (= mk + 2).
Clearly, f (mk + 1) = p. Note also that
k−1
nk − 2 = mk = (p − 1)(p + 1)(p2 + 1) · · · (p2 − 1) = (p − 1)n0 n1 · · · nk−1 .
k
Note that nk = p2 + 1 ≡ 2 mod 4. Hence, 4 - nk and nk > 4. Therefore, nk contains an odd prime factor.
Suppose q is an odd prime such that q | nk , ni for some 0 ≤ i < k − 1. Then q|2, which is impossible.
Therefore, nk contains an odd prime factor which is not a prime factor of n0 , n1 , · · · , nk−1 . Hence, let k be
the smallest non-negative integer such that f (nk ) > p. Since n0 = p + 1 and p + 1 is composite, implying
f (n0 ) < p. Therefore, k ≥ 1. Hence, f (mk + 2) > p = f (mk + 1).
Finally, note that mk = (p − 1)n0 n1 n2 · · · nk−1 . By the choice of k and the fact that p - ni , we have
f (ni ) mk to generate
another integer n larger than mk such that f (n) < f (n + 1) < f (n + 2). We repeat this process similarly
to generate infinitely many such positive integers. This completes the problem.
Solution 2: Suppose the statement is false, i.e. there are only finitely many integers n such that
f (n) < f (n + 1) < f (n + 2). Consider the number n = 2k for some positive integer k. Note that f (n) = 2
and f (n + 1) > 2 = f (n). Since there are only finitely many k such that f (2k ) < f (2k + 1) < f (2k + 2),
there exists a positive integer m such that k ≥ m implies f (2k ) < f (2k + 1) and f (2k + 2) ≤ f (2k + 1).
Since gcd(2k + 1, 2k + 2) = 1, f (2k + 2) < f (2k + 1).
For k ≥ m, note that f (2k + 2) = f (2k−1 + 1), since every odd factor of 2k + 2 (which is ≥ 2) is also
an odd factor of 2k−1 + 1. Therefore, for k ≥ m, f (2k−1 + 1) < f (2k + 1). Hence, f (2m + 1), f (2m+1 +
1), f (2m+2 + 1), · · · is a strictly increasing sequence of integers. (*)
6
Let k ≥ m. Consider n = 22(2k−1) + 1 = (22k−1 + 1)2 − 2 · 22k−1 = (22k−1 − 2k + 1)(22k−1 + 2k + 1).

Therefore, f (22(2k−1) + 1) = f (22k−1 + 2k + 1) or f (22(2k−1) + 1) = f (22k−1 − 2k + 1). If the former holds,
then
f (22(2k−1) + 1) = f (22k−1 + 2k + 1) = f (2(22k−1 + 2k + 1)) = f ((2k + 1)2 ) = f (2k + 1).
Since 2(2k − 1) > k ≥ m, this contradicts (*). Therefore, f (22(2k−1) + 1) = f (22k−1 − 2k + 1) for all
k ≥ m. Since f (22(2k−1) + 1) > f (2k + 1) (since k ≥ m), we have that f (22k−1 − 2k + 1) > f (2k + 1).
Let n = 22k − 2k+1 = 2k+1 (2k−1 − 1). Then f (n) = f (2k−1 − 1), f (n + 1) = f (22k − 2k+1 + 1) =
f ((2k − 1)2 ) = f (2k − 1) and f (n + 2) = f (22k − 2k+1 + 2) = f (22k−1 − 2k + 1) > f (2k − 1) = f (n + 1).
If f (n) < f (n + 1) for infinitely many choices of k, then we are done. Otherwise, there exists a positive
integer M such that M > m and k ≥ M implies f (n) ≥ f (n + 1), i.e. f (2k−1 − 1) ≥ f (2k − 1). Since
(2k − 1) − 2(2k−1 − 1) = 1, gcd(2k−1 − 1, 2k − 1) = 1. Hence, f (2k−1 − 1) > f (2k − 1) for all k ≥ M .
Hence, for f (2M − 1), f (2M +1 − 1), f (2M +2 − 1), · · · is an infinite strictly decreasing sequence of positive
integers, which is absurd. Hence, the statement of the problem is indeed true.
Source: Brazilian Mathematical Olympiad 1995
Combinatorics
C1 Let n ≥ 2 be a positive integer. An n × n grid is filled with the integers 1, 2, · · · , n2 (with each number
used exactly once). Let t be the maximum of the (positive) difference of the entries of two neighbouring
squares, where two squares are said to be neighbours if they share at least one vertex. Determine the
minimum possible value of t in terms of n.
Solution: The answer is t = n + 1.
The number of squares in the shortest sequence of successive neighbouring squares between any two squares
is at most n − 1. Since both 1, n2 are in the grid and differ by n2 − 1, there are two neighbouring squares
2
−1
that differ by at least nn−1 = n + 1. Hence, t ≥ n + 1. Consider the grid filled in order row by row
1, 2, · · · , n , i.e. place (i − 1)n + j in the ith row and the j th column. Two entries in such a configuration
2
differ by at most n + 1. Hence, t = n + 1.
Source: Unknown
C2 A chessboard is tiled with 32 dominoes. Each domino covers two adjacent squares, a white and a black
square. Show that the number of horizontal dominoes with the white square on the left of the black square
equals the number of horizontal dominoes with the white square on the right of the black square.
Solution: Let Si be the set of horizontal dominoes contained in columns i and i + 1 (obviously, all Si are
disjoint). Then Si = Li ∪ Ri , where Li is the set of such dominoes with a white square on the left, and
Ri is the set of such dominoes with a white square on the right. Since each column has an even number
of rows, and an even number of those rows is taken up by vertical dominoes (each of them takes 2 rows),
the horizontal dominoes must also occupy an even number of rows, so |Si−1 ∪ Si | is even for each i. Since
|S1 | is even, we must have that |Si | is even for each i.
Consider the horizontal dominoes that intersect the first column of the board. The space between any
consecutive pair of horizontal dominoes is taken up by vertical dominoes. Each of those occupies 2 rows,
so there is an even number of rows between the horizontal dominoes. Thus, if one of those dominoes is in
L1 , the other is in R1 , and vice-versa. Since |S1 | is even, we must have |L1 | = |R1 |.
7
We prove by induction that |Li | = |Ri | for each i. The base case is i = 1, shown above. Suppose
|Li−1 | = |Ri−1 |. The set of horizontal dominoes that intersect column i is Si−1 ∪ Si . As in the i = 1 case,
any consecutive pair of these dominoes has an even number of rows between them. Thus, the number of
dominoes that use a black square in column i (|Li−1 ∪ Ri |) is the same as the number of dominoes that
use a white square in column i (|Ri−1 ∪Li |). However, since |Li−1 | = |Ri−1 |, we must have that |Li | = |Ri |.
Therefore, the total number of horizontal dominoes with a white square on the left (| ∪i Li |) equals the
total number of horizontal dominoes with a white square on the right (| ∪i Ri |).
Source: IMO Training 2006, General Problems
C3 Let n > 1 be a positive integer. On each of 2n points around a circle we place a disk with one white side
and one black side. We may perform the following move: select a black disk, and reverse its two neighbors.
Find all initial configurations from which some sequence of such moves leads to a position where all disks
but one are white.
Solution: The answer is when the number of black disks initially is odd.
Note that each move preserves the parity of the number of black disks on the circle. Hence, if the initial
number of black disks is even, then we can never have one black disk remaining.
Suppose there is an odd number of black disk. We define a group to be a maximal set of black disks that
appear on consecutive points on the circle. Let k be the number of groups initially. I claim that if k > 1,
after a finite number of moves, we can decrease the number of groups. Since there are an odd number of
black disks, one group contains an odd number of disks. Suppose there are 2m − 1 disks in this group for
some positive integer m. We label the disks 1, 2, · · · , 2m − 1 in clockwise order. By assumption, the disk
next to 1 (which is not 2), which we now call 0, is white and the disk next to 2m − 1 (which is not 2m − 2),
which we will call 2m, is also white. Note that since k > 1, disks 0 and 2m are distinct. We perform a
move on disks 1, 3, 5, · · · , 2m − 1, in this order. In doing so, the disks 2, 4, · · · , 2m − 2 are each flipped
twice, and therefore, remain black. The only disks flipped are 0 and 2m, which are now flipped to black.
Hence, this group increased by size at least 2. If it increased by more than 2, then this group merged with
another group and the number of groups decreased. Otherwise, this group still has odd cardinality and
we can repeat this process. Since there is another group, eventually, this group will merge with another
group. Hence, we have decreased the number of groups. We repeat this process (by choosing another odd
group). Eventually, the number of groups becomes one.
By performing the steps in the previous paragraph, we have one group of black disks. There are an
odd number of disks in this group, say 2m − 1 for some positive integer m. We again label the disks
1, 2, · · · , 2m − 1. If m = 1, then only one black disk remain and we are done. Otherwise, we choose the
disks 2, 4, · · · , 2m − 2. Hence, disks 3, 5, · · · , 2m − 3 are flipped twice, and thus remain the same colour.
But disks 1 and 2m − 1 are changed to white. Hence, the group size decreased by 2. We can repeat this
procedure until one black disk remains. We are done.
Source: Japanese Mathematical Olympiad 1998
Geometry
G1 The altitude from A of triangle ABC intersects the side BC at D. A circle is tangent to BC at D,
intersects AB at M and N , and intersects AC at P and Q. Prove that
AM + AN AP + AQ
= .
AC AB
8
Solution 1: By Power of a Point, we have that BM · BN = BD2 and CP · CQ = CD2 . Hence,

(AB − AM )(AB − AN ) = BD2 ⇒ AB 2 − AB(AM + AN ) + AM · AN = BD2 ⇒ AB 2 − BD2 =
AB(AM + AN ) − AM · AN . Similarly, AC 2 − CD2 = AC(AP + AQ) − AP · AQ. But AB 2 − BD2 =
AD2 = AC 2 − CD2 by Pythagorean Theorem and AM · AN = AP · AQ by Power of a Point. Hence,
AB(AM + AN ) = AC(AP + AQ), as desired.
Solution 2: Let O be the centre of the circle. Since the circle is tangent to BC at D and AD ⊥ BC, O
lies on AD. Let U , V be the feet of the perpendicular on AB, AC from O, respectively. Therefore, U, V
are midpoints of M N, P Q, respectively. This implies AM + AN = 2 · AU and AP + AQ = 2 · AV . Hence,
it suffices to show that
AU AV
= .
AC AB
AD AD
Since AU = AO cos ∠BAO, AC = cos ∠CAO , AV = AO cos ∠CAO, AB = cos ∠BAO , this equation is true,
as desired . 2
Source: IMO Correspondence program, 1995-96
G2 Let P be a convex 2010-gon. The 1005 diagonals connecting opposite vertices and the 1005 lines connect-
ing the midpoints of opposite sides are concurrent. (i.e. all 2010 lines are concurrent.) Prove that the
opposite sides of P are parallel and have the same length.
Solution: We first prove a lemma.
Lemma: Let ABCD be a convex quadrilateral, M the midpoint of AB, N the midpoint of CD. Suppose
AC, BD, M N are concurrent at a point P . Then AB||CD and ∆P AB ∼ ∆P CD.
Proof of Lemma: Let l be a line passing through B parallel to CD and intersecting P A at a point A0 . Let
M 0 be the midpoint of A0 B. Since ∆P A0 B ∼ ∆P CD, ∆P AM 0 ∼ ∆P CN . Hence, ∠A0 P M 0 = ∠CP N ,
i.e. M 0 lies on P N . Suppose A 6= A0 . Then M 6= M 0 . Since M also lies on P N , M M 0 is parallel to
AP . But they intersect at P , contradicting the fact that they are parallel. Therefore, A = A0 . Therefore,
AB||CD. Subsequently, ∠P BA = ∠P DC and ∠P AB = ∠P CD, we have that ∆P AB ∼ ∆P CD. End
Proof of Lemma
Let A1 , A2 , · · · , A2010 be the vertices appearing clockwise of the 2010-gon and Mi the midpoint of Ai and
Ai+1 . Then Ai and Ai+1005 are opposite vertices. Let P be the concurrent point of the 2010 lines. Since
Ai Ai+1005 , Ai+1 Ai+1006 , Mi M1005 are concurrent at P , consider the quadrilateral Ai Ai+1 Ai+1005 Ai+1006 .
Since the polygon is convex, this quadrilateral is convex. By the Lemma, we have that Ai Ai+1 is parallel
to Ai+1005 Ai+1006 . Hence, opposite sides of the polygon are parallel.
By the similarities, we have that

P A1 P A2 P A3 P A2010
= = = ··· = .
P A1006 P A1007 P A1008 P A1005
More importantly, we have
P Ai P Ai+1005 P Ai+1005
= = ,
Ai+1005 P Ai+2010 P Ai
since the indices are taken modulo 2010. Hence, |P Ai | = |P Ai+1005 |. Therefore, since by the Lemma that
∆P Ai Ai+1 ∼ ∆P Ai+1005 Ai+1006 , we have |Ai Ai+1 | = |Ai+1005 Ai+1006 |. Hence, the opposite sides of the
2 We AU AV
can also prove AC
= AB
by noting ∆AOU ∼ ∆ABD and ∆AOV ∼ ∆ACD.
9
polygon are equal in length.
Source: Brazilian Mathematical Olympiad 2006, modified in the obvious way
AP CP
Alternate Proof of Lemma: It suffices to prove BP = DP . By Sine Law, we have that
AP AP BM sin ∠AM P sin ∠BP M sin ∠BP M

= · = · = ,
BP AM BP sin ∠AP M sin ∠BM P sin ∠AP M
since ∠AM P + ∠BM P = 180◦ . Similarly, we have
CP sin ∠DP N sin ∠BP M
= = .
DP sin ∠CP N sin ∠AP M
AP CP
Therefore, BP = DP . End Proof of Lemma.
G3 Two circles meet at A and B. Line ` passes through A and meets the circles again at C and D respectively.
Let M and N be the midpoints of arcs BC and BD which do not contain A, and let K be the midpoint
of CD. Prove that ∠M KN = 90◦ .
Solution 1: Let X, Y be the midpoints of BC, BD, respectively. Note that M X ⊥ BC and N Y ⊥ BD.
Also note that KY ||BC and KX||BD. Hence, KXBY is a parallelogram.
Since |N B| = |N D| and |M B| = |M C|, ∠BN Y = 12 ∠BN D = 12 ∠BAC = ∠M AC = ∠M BC. Since

∠BXM = ∠N Y B = 90◦ , we conclude that ∆BXM ∼ ∆N Y B. Observe that
|M X| |M X| |BY | |XK|
= = = .
|KY | |XB| |Y N | |Y N |
Also, ∠M XK = 90◦ + ∠BXK = ∠90◦ + ∠KY B = ∠KY N . Hence, ∆M XK ∼ ∆KY N . Finally,
∠M KN = ∠XKY − ∠XKM − ∠N KY = ∠XKY − ∠XKM − ∠XM K
= ∠XKY − (180◦ − ∠M XK) = ∠XKY − 180 + 90 + ∠BXK = 180 − 180 + 90 = 90◦ ,

(with the second last assertion following from the fact that KY ||XB), as desired.
Solution 2: (Inversion solution:) Since M, N are midpoints of the arcs BC and BD, respectively, AM
bisects ∠BAC and AN bisects ∠BAD. Therefore, ∠M AN = 90◦ . To prove ∠M KN = 90◦ , it suffices to
prove that AM N K is cyclic.
We will invert the diagram about the point A with radius 1. For each object X, let X 0 be its image under
the inversion. Since ABM C is cyclic, B 0 , M 0 , C 0 are collinear. Since AM bisects ∠BAC, AM 0 bisects
∠B 0 AC 0 . Similarly, B 0 , N 0 , D0 are collinear and AN 0 bisects ∠B 0 AD0 . The point K 0 lies on C 0 D0 external
to segment C 0 D0 . To prove that M 0 , N 0 , K 0 are collinear, by Menelaus’ Theorem, it suffices to prove that
|C 0 M 0 | |B 0 N 0 | |D0 K 0 |
· · = 1.
|M 0 B 0 | |N 0 D0 | |K 0 C 0 |
|C 0 M 0 | |AC 0 | |B 0 N 0 | |AB 0 |
By angle bisector theorem, we have that |M 0 B 0 | = |AB 0 | and |N 0 D 0 | = |AD 0 | . Hence, it suffices to show
that
|C 0 A| |C 0 K 0 |
= .
|AD0 | |K 0 D0 |
10
By properties of inversion, we know that
|KC| · |AC 0 | |KD| · |AK 0 |

|K 0 C 0 | = , |K 0 D0 | = .
|AK| |AD|
Since K is the midpoint of CD, |KC| = |KD|. Hence,
|K 0 C 0 | |AC 0 | · |AD| |AC 0 | · |AD| |AC 0 |

= = = ,
|K 0 D0 | |AK| · |AK 0 | |AD0 | · |AD| |AD0 |
as desired.
Solution 3: (Harmonic Division solution:) Using the notation of Solution 2, it suffices to prove that
|C 0 A| |C 0 K 0 |
= ,
|AD0 | |K 0 D0 |
i.e. (K 0 , A) divides (C 0 , D0 ) harmonically. But since K is the midpoint of CD, (K, ∞) divides (C, D)
harmonically. Since the property that a harmonic quadruple is preserved under inversion of a point on
the same line as the quadruple3 , this implies (K 0 , A) divides (C 0 , D0 ) harmonically, as desired.
Source: Romanian Team Selection Test 1999
3 The proof of this fact is similar to the last step of Solution 2.
11
IMO Winter Camp Mock Olympiad 2010
Time: 4 Hours
1. Let ABCD be a parallelogram with AC/BD = k. The bisectors of the angles formed by AC and
BD intersect the sides of ABCD at K, L, M, N . Prove that the ratio of the areas of KLM N
and ABCD is 2k/(k + 1)2 .
2. Each edge of an m × n rectangular grid is oriented with an arrow such that
(a) the border is oriented clockwise, and

(b) each interior vertex has two arrows coming out of it, and two arrows going into it.
Prove that there is at least one square whose edges are oriented clockwise.
3. Let Z∗ denote the set of non-zero integers. A function f : Z∗ → Z≥0 satisfies the following
properties:
(a) f (m + n) ≥ min{f (m), f (n)} for all m, n ∈ Z∗ , m + n 6= 0.

(b) f (mn) = f (m) + f (n) for all m, n ∈ Z∗ .
(c) f (2010) = 1.
Determine the minimum and maximum possible value of f (2010!).
4. If a, b, c are positive real numbers such that a + b + c = 3, show that

1 1 1 3
+ + ≤
2 + a2 + b2 2 + b2 + c2 2 + c2 + a2 4
IMO Winter Camp Mock Olympiad 2010 Solutions
1. Let ABCD be a parallelogram with AC/BD = k. The bisectors of the angles formed by AC and
BD intersect the sides of ABCD at K, L, M, N . Prove that the ratio of the areas of KLM N
and ABCD is 2k/(k + 1)2 .
Solution: Suppose K, L, M, N are on AB, BC, CD, DA, respectively. Let P = AC ∩ BD. Since
ABCD is a parallelogram, AP = P C and BP = P D. For any polygon X1 X2 · · · Xn , we denote
its area by [X1 X2 · · · Xn ].
By the angle bisector theorem, AK/KB = AP/P B = AC/BD = k. Similarly, AN/N D = k.

k
Hence, AK/AB = AN/AD = k+1 . Consequently, KN is parallel to BD. Furthermore,
k2 k2
[AKN ] = [ABD] = [ABCD].
(k + 1)2 2(k + 1)2
Similarly,
1 k2 1
[BKL] = [ABCD], [CLM ] = [ABCD], [DM N ] = [ABCD].
2(k + 1)2 2(k + 1)2 2(k + 1)2
Hence,
[KLM N ] = [ABCD] − [AKN ] − [BKL] − [CLM ] − [DM N ]
k2

1 2k
= 1− − [ABCD] = [ABCD],
(k + 1)2 (k + 1)2 (k + 1)2
as desired.
Source: IMO Correspondence program, 1995-96 (from Ed Barbeau)
2. Each edge of an m × n rectangular grid is oriented with an arrow such that
(a) the border is oriented clockwise, and

(b) each interior vertex has two arrows coming out of it, and two arrows going into it.
Prove that there is at least one square whose edges are oriented clockwise.
Solution: Let a square be called half-clockwise if its top and left sides are oriented clockwise.
Suppose that there is no clockwise square in the grid.
Consider a half-clockwise square S. Since it is not a clockwise square, either the right or bottom
edge must be oriented counterclockwise. If it is the right edge, let the square to the right of S
be T (if there are no squares to the right of S, then the right edge of S is on the border and
thus must be oriented clockwise).
Consider the top right vertex of S - it is either in the interior or on the top border of the grid.
If it is an interior vertex, it has the top and right edges of S going into it, so its other edges are
going out, in particular the top edge of T . If it is on the top border, then the top edge of T is
clockwise. In both cases, T is a half-clockwise square.
If the bottom edge of S is oriented counterclockwise, we can similarly show that the square T
below S is a half-clockwise square.
Since the top left square in the grid is half-clockwise, we can make a path of half-clockwise
squares by stepping either to the right or down. This implies that the bottom right square in the
grid is half-clockwise. However, the bottom and right edges of this square are clockwise-oriented,
so the square is clockwise. Contradiction.
Source: IMO Training 2006, General Problems
3. Let Z∗ denote the set of non-zero integers. A function f : Z∗ → Z≥0 satisfies the following
properties:
(a) f (m + n) ≥ min{f (m), f (n)} for all m, n ∈ Z∗ , m + n 6= 0.

(b) f (mn) = f (m) + f (n) for all m, n ∈ Z∗ .
(c) f (2010) = 1.
Determine the minimum and maximum possible value of f (2010!).
Solution: The minimum possible value is 30 and the maximum possible value is 2002.
Substituting m = n = 1 into (b) yields f (1) = 0. Substituting m = n = −1 into (b) yields

f (−1) = 0. Substituting m = −1 into (b) yields f (n) = f (−n) for all n ∈ Z∗ .
Let p be the smallest integer such that f (p) = 1. This is well defined since f (2010) = 1 > 0. I
claim that p is a prime. Suppose p = ab for some 1 < a, b < p. Then f (p) = f (a) + f (b) by (b).
Hence, at least one of f (a), f (b) = 1. This contradicts p being the smallest integer such that
f (p) > 0. Therefore, p is prime and f (1) = f (2) = · · · = f (p − 1) = 0.
Let a be a positive integer not divisible by p. I claim that f (a) = 0. Write a = qp + r

where 0 ≤ r < p. Then f (r) = 0. Therefore, 0 = f (r) = f (a − qp) ≥ min{f (a), f (−qp)} =
min{f (a), f (qp)} ≥ 0. Hence, equality must hold throughout. Therefore, min{f (a), f (qp)} = 0.
Since f (qp) = f (q) + f (p) ≥ f (p) > 0, f (a) = 0.
For any positive integer m, write m = pr ·s where r ∈ Z≥0 and s - p. Then f (m) = f (pr )+f (s) =
r · f (p) = r. Hence, f (m) is the number of times p divides into m for a fixed prime p. We will
now verify that such a function satisfies properties (a) and (b).
Let m = pa · b and n = pc · d such that a, c ∈ Z≥0 and p - b, d. Then mn = pa+c · bd and
p - bd. Therefore, f (mn) = a + c = f (m) + f (n). Hence, f satisfies (b). Finally, without loss of
generality, suppose a ≤ c. Then m + n = pa · b + pc · d = pa (b + pc−a d). Hence p divides into
m + n at least a times. Therefore, f (m + n) ≥ a = min{a, c} = min{f (m), f (n)}. Hence, f
satisfies (a). Note that p can be chosen to be any prime number.
Since f (2010) = 1 and 2010 = 2 · 3 · 5 · 67, we conclude that all of the possible values of p are
2, 3, 5, 67. We want to find the number of times each of these values of p divide into 2010!. The
maximum and minimum possible value of f (2010!) is obtained by taking p = 2 and p = 67,
respectively.
If p = 2, then the number of times 2 divides into 2010! is

2010 2010 2010 2010 2010 2010 2010 2010 2010 2010
+ + + + + + + + + ,
2 4 8 16 32 64 128 256 512 1024
= 1005 + 502 + 251 + 125 + 62 + 31 + 15 + 7 + 3 + 1 = 2002.
If p = 67, then the number of times 67 divides into 2010! is 2010

67 = 30. Therefore, the minimum
and maximum possible value of f (2010!) is 30 and 2002, respectively.
Source: Original, but this is a well-known map called a valuation map.
4. If a, b, c are positive real numbers such that a + b + c = 3, show that

1 1 1 3
+ + ≤
2 + a2 + b2 2 + b2 + c2 2 + c2 + a2 4
Solution 1: Multiply both sides by 2 to get:

2 2 2 3
+ + ≤
2 + a2 + b2 2 + b2 + c2 2 + c2 + a2 2
⇔
2 2 2 3
3− 2 2
+ 2 2
+ 2 2
≥3−
2+a +b 2+b +c 2+c +a 2
⇔
X a2 + b2 3
≥ .
cyc
2 + a2 + b2 2
By Cauchy-Schwarz Inequality, we have

!2
P√
a2 + b2
Pp 2
(a2 + b2 ) + 2 (a + b2 )(a2 + c2 )
P
X a2+ b2 cyc cyc cyc
2 2
≥ P = .
(2 + a2 + b2 ) 6 + 2 a2
P
cyc
2+a +b
cyc cyc
By Cauchy-Schwarz Inequality,
p
(a2 + b2 )(a2 + c2 ) ≥ a2 + bc.
Hence, our expression is greater than or equal to
P 2
(a + b2 ) + 2 (a2 + bc) (3 a2 ) + (a + b + c)2 (3 a2 ) + 9
P P P
cyc cyc cyc cyc 3
P 2 = P 2 = P 2 = ,
6+2 a 6+2 a 6+2 a 2
cyc cyc cyc
as desired.
Solution 2: (Mixing Variables - With The Cumbersome Steps Omitted) For x, y, z > 0, let
1 1 1
f (x, y, z) = 2 2
+ 2 2
+ .
2+x +y 2+y +z 2 + z 2 + x2
Let t = a+b
2 . We will prove that f (a, b, c) ≤ f (t, t, c) ≤
3
4 to solve the problem. Without loss of
generality, suppose a ≤ b ≤ c.
Note that f (t, t, c) − f (a, b, c)

1 1 2 1 1
= − + − − .
2 + 2t2 2 + a2 + b2 2 + t2 + c2 2 + a2 + c2 2 + b2 + c2
We prove each of these two terms is non-negative to prove f (a, b, c) ≤ f (t, t, c).
1 1 1 1 (a − b)2
2
− 2 2
= 2 − = ≥ 0.
2 + 2t 2+a +b 2+2 2a+b 2 + a2 + b2 4 1+

a+b 2
(2 + a2 + b2 )
2
In the second term, it suffices to show that the numerator is non-negative. The numerator is
equal to
8 + 4(a2 + b2 + 2c2 ) + 2(a2 + c2 )(b2 + c2 ) − (2 + t2 + c2 )(4 + a2 + b2 + 2c2 )

= 2(a2 + b2 + a2 b2 ) − t2 (a2 + b2 + 4) + (a2 + b2 − 2t2 )c2 .
2 2 +2c2
= (a − b)2 · 4−a −4ab−b
4
2
It suffices to show that a2 +4ab+b2 −2c2 ≤ 4. Note that a2 +4ab+b2 −2c2 ≤ (a+b)2 + (a+b) 2 −2c2 .
By substituting a + b = 3 − c, the latter term being at most 4 is equivalent to (c + 19)(c − 1) ≥ 0.
Since a ≤ b ≤ c and a + b + c = 3, c ≥ 1. Hence, the statement is true. Hence, we conclude that
f (a, b, c) ≤ f (t, t, c).
It remains to show that f (t, t, c) ≤ 34 , i.e.
1 2 3
2
+ 2 2
≤ .
2 + 2t 2+t +c 4
This simplifies to 3t2 c2 + 3t4 + c2 − t2 − 6 ≥ 0. By substituting c = 3 − 2t, we simplify this to
proving (t − 1)2 (5t2 − 2t + 1) ≥ 0. Since the latter term has negative discriminant, this term is
indeed non-negative and therefore the inequality is true. Hence, f (t, t, c) ≤ 43 , as desired.
Source: Iranian Team Selection Test 2009

2010 Mock 1
Time: 4.5 Hours
1. Let AD be the angle bisector of ∆ABC. A line ` is tangent to the circumcircles of ∆ADB
and ∆ADC at points M, N , respectively. Prove that ` is also tangent to the circle passing
through the midpoints of BD, DC, M N .
Solution 1: (Using homothety:) Let ω1 , ω2 be circumcircles of ∆BAD, ∆CAD, respec-

tively with circumcentres O1 , O2 , respectively.
Let h be the homothety with positive ratio that maps ω1 to ω2 . Then h(M ) = N and
h(O1 ) = O2 . We claim that h(B) = D and h(D) = C. Note that ∠O1 BC = 90 − ∠BAD and
∠O2 DC = 90 − ∠DAC. Since ∠BAD = ∠DAC, ∠O1 BC = ∠O2 DC. Therefore, BO1 ||DO2 .
Consequently, h(B) = D. Therefore, the centre of h lies on BD. Hence, h(D) is a point
(different from D) on ω2 that lies on BC. Therefore, h(D) = C.
Let E, X, Y be the midpoints of M N, BD, DC, respectively and ω the circumcircle of ∆EXY .
Since N = h(M ) and D = h(B), M B||N D. Therefore, EX||M B||N D. Similarly, EY ||M D||N C.
Hence, the homothety h1 (with positive ratio) that maps M B to EX has the same centre as h
and maps D to Y . Since the circumcircle of ∆M BD is tangent to ` and ` passes through the
centre of h1 and h1 maps ∆M BD to ∆EXY , the image of this circumcircle is also tangent
to `, i.e. the circumcircle of ∆EXY is tangent to h. This completes the problem.
Solution 2: (Using angle chasing:) Let ω1 , ω2 be circumcircles of ∆BAD, ∆CAD, re-

spectively. We will consider the two tangent lines of ω1 , ω2 separately.
We first consider the case when ` intersects DA on ray DA. Using the same argument as
in Solution 1, AD passes through the midpoint of M N ; we will call this point E. I claim
that M D||N C and M B||N D. Let α = ∠BAD = ∠DAC, θ1 = ∠N M D, θ2 = ∠M N D.
Note that ∠BM D = ∠BAD = α and ∠DN C = ∠DAC = α. Then by properties of
tangents, ∠M BD = θ1 , ∠N CD = θ2 . Looking at the quadrilateral M N CB, we have
2(θ1 + θ2 + α) = 360◦ . Hence, θ1 + θ2 + α = 180◦ . Since ∠N M D = θ1 and ∠M N C = θ2 + α
and θ1 + θ2 + α = 180◦ , M D||N C. Since ∠BM D = ∠DN C = α, M B||N D.
Let X, Y be the midpoints of BD and DC. Since E is the midpoint of M N , M B||EX||N D

and M D||EY ||N C. Therefore, ∠M EX = ∠M N D = ∠N CD = ∠EY X. Hence, the circum-
circle of ∆M XY is tangent to `, as desired.
The case when ` intersects DA on ray AD is handled similarly. Using the same argument
as in Solution 1, AD passes through the midpoint of M N ; we will call this point E. I claim
that M D||N C and M B||N D. Let α = ∠BAD = ∠DAC, θ1 = ∠N M D, θ2 = ∠M N D.
Note that ∠BM D = 180 − ∠BAD = 180 − α and ∠DN C = 180◦ − ∠DAC = 180◦ − α.
Then by properties of tangents, ∠M BD = θ1 , ∠N CD = θ2 . Looking at the quadrilateral
M N CB, we have 2(θ1 + θ2 + 180◦ − α) = 360◦ . Hence, θ1 + θ2 + 180◦ − α = 180◦ . Since
∠N M D = θ1 and ∠M N C = 180 − α + θ2 and θ1 + θ2 + 180◦ − α = 180◦ , M D||N C. Since
∠BM D = ∠DN C = 180◦ − α, M B||N D.
Let X, Y be the midpoints of BD and DC. Since E is the midpoint of M N , M B||EX||N D

and M D||EY ||N C. Therefore, ∠M EX = ∠M N D = ∠N CD = ∠EY X. Hence, the circum-
circle of ∆M XY is tangent to `, as desired.
2. Let P (x) be a non-constant polynomial with integer coefficients. Prove that there is no func-
tion T : Z → Z such that the number of integers x with T n (x) = x is equal to P (n) for every
n ≥ 1, where T n denotes the n-fold application of T .
Solution: Suppose such a function T exists. For each positive integer n, let
A(n) = {x ∈ Z | T n (x) = x}
and
B(n) = {x ∈ Z | T n (x) = x, and n is the smallest positive integer with this property.}
Then P (n) = |A(n)| for all positive integers n. This implies that A(n) is finite for all n ∈ N.
Note that [
A(n) = B(d)
d|n
and the sets B(d), d|n are pairwise disjoint. Therefore,

X
|A(n)| = |B(d)|.
d|n
We claim that d divides |B(d)|. We represent this as a direct graph: let the set of integers be
represented by vertices and there is a direct edge from a to b if and only if b = T (a). Then
an element x ∈ B(d) if and only if x is in a cycle of length d. Since the number of elements
in a given cycle of length d is a multiple of d, the number of elements in a cycle of length d is
a multiple of d. Therefore, d divides |B(d)|.
Let m = deg P with P (x) = am xm + am−1 xm−1 + · · · + a1 x + a0 . Since P is not a constant,

m ≥ 1. Let p be any prime number. Note that P (1) = |A(1)| = |B(1)| and P (p) = |A(p)| =
|B(p)|+|B(1)| = |B(p)|+|A(1)| = |B(p)|+P (1). Since p divides |B(p)|, p divides P (p)−P (1).
Hence,
p divides am (pm − 1) + · · · + a1 (p − 1).
Therefore, p|am + am−1 + · · · + a1 . Since this is true for all primes p,
am + am−1 + · · · + a1 = 0 (1)
Finally, let p, q be any two distinct primes. Then P (pq) = |A(pq)| = |B(pq)|+|B(p)|+|B(q)|+
|B(1)| = |B(pq)|+(|A(p)|−B(1))+(|A(q)|−B(1))+|A(1)| = |B(pq)|+|A(p)|+|A(q)|−|A(1)| =
|B(pq)| + P (p) + P (q) − P (1). Therefore, |B(pq)| = P (pq) − P (p) − P (q) + P (1). Since pq
divides |B(pq)|, pq divides P (pq) − P (p) − P (q) + P (1). Therefore,
pq divides am ((pq)m − pm − q m + 1) + · · · + a1 (pq − p − q + 1)

= am (pm − 1)(q m − 1) + · · · + a1 (p − 1)(q − 1).
In particular, p divides this expression. Taking this equation modulo p and by (1), we have
−am (q m − 1) − · · · − a1 (q − 1) ≡ −(am q m + · · · + a1 q) ≡ 0 (mod p).
Therefore, every prime q is a root of the equation am xm + am−1 xm−1 + · · · + a1 x (mod p).
We will show that this is impossible. Choose p such that p > m + 1. By Dirchlet’s theorem
for prime, for each i ∈ {1, · · · m + 1}, there exists a prime qi such that qi ≡ i (mod p).
Since p > m + 1, q1 , · · · , qm+1 are pairwise distinct modulo p. Then q1 , · · · , qm+1 are pairwise
distinct roots of the polynomial am xm + am−1 xm−1 + · · · + ax (mod p). But this polynomial
contains at most m roots. This is a contradiction.
Therefore, no such function T exists.
3. On a 999 × 999 board a limp rook can move in the following way: From any square it can
move to any of its adjacent squares, i.e. a square having a common side with it, and every
move must be a turn, i.e. the directions of any two consecutive moves must be perpendicular.
A non-intersecting route of the limp rook consists of a sequence of pairwise different squares
that the limp rook can visit in that order by an admissible sequence of moves. Such a non-
intersecting route is called cyclic, if the limp rook can, after reaching the last square of the
route, move directly to the first square of the route and start over.
How many squares does the longest possible cyclic, non-intersecting route of a limp rook visit?
Solution: The answer is 4 · ((499)2 − 1).
Colour the board with four colours A, B, C, D according to the following rules: Given a square
in the ith row and j th column, colour the square A if i, j are both odd, B if i is odd and j is
even, C if i is even and j is odd and D if i, j are both even. Then a limp rook encounters the
same colour exactly every four steps. More specifically, D appears once every four steps.
Since there are only 4992 squares coloured D, a limp rook cycle contains at most 4(4992 ). We
claim that a limp rook cycle cannot go through every square coloured D. This will prove that
a limp rook cycle passes through at most 4(4992 − 1) squares. We will then construct such a
cycle to prove that this length is indeed the maximum.
Colour the squares marked D in a ”chessboard” manner in black and white. Suppose that a
limp rook cycle passes through every square marked D. Note that two consecutive squares
of a limp rook cycle that are coloured D, are ”neighbours” of each other (i.e. distance 2
horizontally and/or vertically of each other.) Since there are an odd number of squares
marked D, there are two consecutive squares in the limp rook cycle coloured the same colour,
marked D. By symmetry, we may assume that these two squares are (a, b), (a + 2, b + 2),
where (i, j) is the square in the ith row and the j th column, for some positive integers a, b.
Without loss of generality, suppose (a, b) is going toward (a + 2, b + 2). Also, assume that the
path goes
(a, b) → (a, b + 1) → (a + 1, b + 1) → (a + 1, b + 2) → (a + 2, b + 2).
It is easy to show that by parity, the limp rook cycle enters any square coloured D horizon-
tally and exits any square coloured D vertically. Consider the square (a, b + 2). This square
is coloured D. Then the limp rook cycle enters this square horizontally and exits this square
vertically. Therefore, the limp rook cycle goes from (a − 1, b + 2) → (a, b + 2) → (a, b + 3).
By the direction of the limp rook cycle, the limp rook goes from (a, b + 3) to (a, b) and from
(a + 2, b + 2) to (a − 1, b + 2). Let P1 , P2 be these two paths of the limp-rook cycle, respec-
tively. Note that P1 , P2 does not go through the interior of the quadrilateral whose vertices
are (a, b), (a + 2, b + 2), (a, b + 3), (a − 1, b + 2). Therefore, P1 and P2 must intersect. Hence,
the limp rook cycle must intersect. This is a contradiction.
Hence, a limp rook cycle contains at most 4(4992 − 1) squares. And now the construction.
2010 Mock 2
Time: 4.5 Hours
1. A positive integer N is called balanced, if N = 1 or if N can be written as a product of an
even number of not necessarily distinct primes. Given positive integers a and b, consider the
polynomial P defined by P (x) = (x + a)(x + b).
(a) Prove that there exist distinct positive integers a and b such that all the numbers
P (1), · · · , P (50) are balanced.
Solution: For each n ∈ N, let f : N → Z be a function such that f (n) = −1 if n

has an odd number of prime divisors and f (n) = 1 if n has an even number of prime
divisors. Then n is balanced if and only if f (n) = 1. Note that f is multiplicative.
Consider the 250 + 1 sequences (km + 1, · · · , km + 50) for k = 1, 2, · · · , 250 + 1. Since
there are 250 possible sequences containing ±1, two of the 250 + 1 sequences are the
same. Hence (f (um + 1), · · · , f (um + 50)) = (f (vm + 1), · · · , f (vm + 50)) for some
distinct u, v ∈ {1, · · · , 250 + 1}. Hence, f ((um + i)(vm + i)) = 1, which implies that
(um + i)(vm + i) is balanced for each i ∈ {1, · · · , 50}. Setting a = um, b = vm yields
P (1), · · · , P (50) each having an even number of prime divisors.
(b) Prove that if P (n) is balanced for all positive integers n, then a = b.
Solution: Suppose a 6= b. WLOG, suppose a < b. Then for any n > b, P (n − a) =

n(n + b − a) is balanced. Therefore, f (n) = f (n + (b − a)). Then the sequence
f (1), f (2), · · · is eventually periodic with period d for some d | b − a. Then for any
prime p, f (dp) = −f (dp2 ), since dp2 has exactly one more prime divisor (namely p) than
dp. Choose a sufficiently large p such that f (dp) is part of the sequence f (1), f (2), · · · for
which the sequence is periodic. Since dp2 and dp differs by a multiple of d, f (dp) = f (dp2 ).
This is a contradiction. Therefore, a = b.
2. In a triangle ABC with AB 6= AC, the incircle touches the sides BC, CA, AB at D, E, F ,
respectively. Line AD meets the incircle again at P . The line EF and the line through P
perpendicular to AD meet at Q. Line AQ intersects DE at X and DF at Y . Prove that A
is the midpoint of XY .
Solution 1: Let I be the incentre of 4ABC. Let AI intersect EF at M . Then clearly M is

the midpoint of EF and ∠AM Q = 90o . Since ∠AP Q = 90o , AP M Q is cyclic. Furthermore,
by Power of a Point, note that AP · AD = AF 2 = AM · AI (since 4AF M ∼ 4AIF .) Hence,
P M ID is also cyclic.
Then, ∠AQP = ∠AM P = ∠ADI, and so ∠QAD = 90o −∠AQP = ∠BDI−∠ADI = ∠BDA,
and so BC||AQ. Therefore, ∠AXE = ∠EDC = ∠CED = ∠AEX. Therefore, AE = AX.
Similarly, AF = AY . Since AE = AF , AX = AY , as desired.
Solution 2: We do the same as Solution 1 to prove that AQ||BC.
Since F P ED is cyclic, F P ED is harmonic. Therefore, the pencil D(D, F, P, E) is harmonic.

Intersecting it with line AQ we get a harmonic bundle (S, A; Y, X), where S is some point
on XY . Since we want to show A is the midpoint of XY , it suffices to show S is a point at
infinity, i.e. that AQ||BC. This completes the solution.
3. Find all functions f : R → R such that
f (xf (x + y)) = f (yf (x)) + x2 ,
for all x, y ∈ R.
Solution: The answers are f (x) = x, ∀x ∈ R and f (x) = −x, ∀x ∈ R.
Substituting x = 0 into the original equation yields f (0) = f (yf (0)), ∀y ∈ R. Suppose
f (0) 6= 0. Then yf (0) ranges over all reals as y varies over all reals. Hence, f (0) = f (z) for
all z ∈ R. Consequently, f is a constant. If f (x) = c, a constant, then substituting this into
the original equation yields c = c + x2 , or x2 = 0 for all x ∈ R. This is clearly absurd since
x = 1 violates this condition. Therefore,
f (0) = 0. (2)
Suppose f (z) = 0. We will prove that z = 0. Substituting x = z, y = −z into the original

equation yields f (zf (0)) = f (−zf (0)) + z 2 = z 2 . Since f (0) = 0 by (2), 0 = 0 + z 2 . Therefore,
z = 0. We have
f (z) = 0 ⇔ z = 0, ∀z ∈ R (3)
Since f (0) = 0, substituting y = 0 into the original equation yields
f (xf (x)) = x2 , ∀x ∈ R (4)
Substituting y = −x yields
f (−xf (x)) = −x2 , ∀x ∈ R (5)
Lemma 1: f is surjective.
Proof of Lemma 1: This follows from both (4) and (5). End Proof of Lemma 1
Lemma 2: f is injective.
Proof of Lemma 2: Suppose f (z) = f (z + r) for some z, r ∈ R. Substituting x = z, y = r
into the original equation yields f (zf (z + r)) = f (rf (z)) + z 2 . Since f (z) = f (z + r),
f (zf (z)) = f (rf (z)) + z 2 . By (4), f (zf (z)) = z 2 . Therefore, f (rf (z)) = 0. By (3), rf (z) = 0.
Therefore, f (z) = 0 or r = 0. If f (z) = 0, then z = 0, Hence, f (0) = f (0 + r). By (3), r = 0.
Hence, in either case, r = 0. Therefore, f is injective. End Proof of Lemma 2
Lemma 3: f (−x) = −f (x), ∀x ∈ R.
Proof of Lemma 3: Let z ∈ R. If z = 0, then clearly, f (−z) = −zf (z), since both are equal to
0. Otherwise, suppose z 6= 0. Substituting x = −z yields f (−zf (−z + y)) = f (yf (−z)) + z 2 .
Since z 6= 0, f (−z) 6= 0. Therefore, yf (−z) ranges over all reals as y varies over all reals. Since
f is surjective, we can let w ∈ R such that f (wf (−z)) = −z 2 . Then f (−zf (−z + w)) = 0.
Hence, −zf (−z + w) = 0. Since z 6= 0, f (−z + w) = 0. Hence, w = z. Therefore, by the
definition of w, f (zf (−z)) = −z 2 . But we also know that f (−zf (z)) = −z 2 by (5). Since f
is injective, zf (−z) = −zf (z). Since z 6= 0, f (−z) = −f (z), as desired.
Since f is surjective, there exist z ∈ R such that f (z) = 1. Then substituting x = z

into (4) yields f (z) = z 2 . Since f (z) = 1, z = ±1. Therefore, by Lemma 3, either
(f (1), f (−1)) = (1, −1) or (f (−1), f (1)) = (−1, 1).
If f (1) = 1 and f (−1) = −1, then substituting x = 1 yields
f (f (y + 1)) = f (y) + 1, ∀y ∈ R (6)
and substituting x = −1 yields f (−f (y − 1)) = f (−y) + 1, or by Lemma 3,
−f (f (y − 1)) = −f (y) + 1, ∀y ∈ R. (7)
Therefore, −f (f (y + 1)) = −f (y + 2) + 1, ∀y ∈ R. Adding this equation with (6) yields

0 = f (y) − f (y + 2) + 2. Hence, f (y + 2) − f (y) = 2. Similarly, f (y + 4) − f (y + 2) = 2.
Therefore,
f (y + 4) − f (y) = 4, ∀y ∈ R. (8)
Substituting x = 2 into the original equation fields f (2f (y + 2)) = f (yf (2)) + 4. By (8), we
have f (yf (2) + 4) = f (yf (2)) + 4. Therefore, f (2f (y + 2)) = f (yf (2) + 4). Since f is injective,
2f (y + 2) = yf (2) + 4. Hence, f (y) = f (2) f (2)
2 (y − 2) + 2. Let c = 2 . Then f (y) = cy + (2 − 2c).
Since y = 0 and f (0) = 0, c = 1. Then f (y) = y, ∀y ∈ R, which can be verified as a solution
to the functional equation.
If f (1) = −1 and f (−1) = 1, then substituting x = 1 yields
f (f (y + 1)) = f (−y) + 1 = −f (y) + 1, ∀y ∈ R (9)

and substituting x = −1 yields f (−f (y − 1)) = f (y) + 1, or by Lemma 3,
−f (f (y − 1)) = f (y) + 1, ∀y ∈ R. (10)
Therefore, −f (f (y + 1)) = f (y + 2) + 1, ∀y ∈ R. Adding this equation with (8) yields

0 = −f (y) + f (y + 2) + 2. Hence, f (y) − f (y + 2) = 2. Similarly, f (y + 2) − f (y + 4) = 2.
Therefore,
f (y) − f (y + 4) = 4, ∀y ∈ R. (11)
Substituting x = 2 into the original equation fields f (2f (y + 2)) = f (yf (2)) + 4. By (8), we
have f (yf (2) − 4) = f (yf (2)) + 4. Therefore, f (2f (y + 2)) = f (yf (2) − 4). Since f is injective,
2f (y + 2) = yf (2) − 4. Hence, f (y) = f (2) f (2)
2 (y − 2) − 2. Let c = 2 . Then f (y) = cy − (2 + 2c).
Since y = 0 and f (0) = 0, c = −1. Then f (y) = −y, ∀y ∈ R, which can be verified as a
solution to the functional equation. This completes the problem.
2010 Mock 3
Time: 4.5 Hours
1. For any n ≥ 2, let N (n) be the maximal number of triples (ai , bi , ci ), i = 1, · · · , N (n) consisting
of nonnegative integers ai , bi , ci such that the following two conditions are satisfied:
• ai + bi + ci = n for all i = 1, · · · , N (n)
• If i 6= j, then ai 6= aj , bi 6= bj , ci 6= cj .
Determine N (n) for n ≥ 2.
Solution: The answer is N (n) = b2n/3c + 1. For brevity, let N = N (n).
N
P N
P N
P
Note that ai , bi , ci ≥ N (N − 1)/2. Therefore,
i=1 i=1 i=1
n
3N (N − 1) X
≤ (ai + bi + ci ) = N n.
2
i=1
Therefore, N ≤ b2n/3c + 1.
It remains to construct such triples for each n. We will split this into 3 cases; n ≡ 0
(mod 3), n ≡ 1 (mod 3), n ≡ 2 (mod 3).
If n ≡ 0 (mod 3), let n = 3k for some positive integer k. Then N (n) = 2k + 1. The following
is a construction of triples satisfying the conditions given in the problem.
ai bi ci
2k 0 k
2k − 1 2 k−1
.. .. ..
. . .
k 2k 0
k−1 1 2k
k−2 3 2k − 1
.. .. ..
. . .
0 2k − 1 k+1
If n ≡ 1 (mod 3), let n = 3k + 1 for some positive integer k. Then N (n) = 2k + 1. The
following is a construction of triples satisfying the conditions given in the problem.
ai bi ci
2k 0 k+1
2k − 1 2 k
.. .. ..
. . .
k 2k 1
k−1 1 2k + 1
k−2 3 2k
.. .. ..
. . .
0 2k − 1 k+2
If n ≡ 2 (mod 3), let n = 3k + 2 for some positive integer k. Then N (n) = 2k + 2. The
following is a construction of triples satisfying the conditions given in the problem.
ai bi ci
2k + 1 0 k+1
2k 2 k
.. .. ..
. . .
k 2k + 2 0
k−1 1 2k + 2
k−2 3 2k + 1
.. .. ..
. . .
0 2k − 1 k + 3
This completes the problem.
2. Let f : R → R be any function. Prove that there exist x, y ∈ R such that
f (x − f (y)) > yf (x) + x.
Solution: Suppose no such x, y ∈ R exist. Then
f (x − f (y)) ≤ yf (x) + x, ∀x, y ∈ R. (12)
Substituting x ← x + f (0) and y = 0 into (12) yields
f (x) ≤ x + f (0). (13)
Substituting x = f (y) into (12) and using (13) yields f (0) ≤ yf (f (y)) + f (y) ≤ yf (f (y)) +
y + f (0). Therefore,
f (f (y)) ≥ −1, ∀y > 0. (14)

Therefore, for all x, y ∈ R such that x > f (y), we have by (14), (13) and (12),
−1 ≤ f (f (x − f (y))) ≤ f (x − f (y)) + f (0) ≤ yf (x) + x + f (0). (15)
Suppose there exists x ∈ R such that f (x) > 0. By (13), we have that f (y) approaches −∞
as y approaches −∞. Then we can choose a sufficiently small y such that f (y) < x and
yf (x) + x + f (0) < −1. This contradicts (15).
Therefore, f (x) ≤ 0 for all x ∈ R. Suppose there exists a positive real x such that f (x) < 0.
Then for all y ∈ R, f (y) ≤ 0 < x. Then by (15), we have −1 ≤ yf (x) + x + f (0). This is not
possible if y is sufficiently large. Therefore, f (x) = 0 for all positive real x.
Finally, choose y = 1 and any x < 0. Since f (y) = f (1) = 0, by (12) f (x) ≤ f (x) + x. Then
x ≥ 0. This contradicts the choice of x. This solves the problem.
3. Let P be a polygon that is convex and symmetric about some point O (i.e. if a point is on the
polygon, then its reflection in O is also on the polygon). Prove that for some parallelogram
R satisfying P ⊂ R we have
|R| √
≤ 2
|P |
where |R| and |P | denote the areas of R and P , respectively.
Solution: Let A, B be two vertices of P such that [OAB] is maximum, where [XY Z] denotes
the area of XY Z. Let A0 , B 0 be the reflections of A, B across O, respectively. Then A0 , B 0
are vertices of P , since P is symmetric about O. Let l1 , l3 be the lines passing through A, A0
respectively parallel to OB and l2 , l4 the lines passing through B, B 0 respectively parallel to
OA. Let P1 be the parallelogram formed by l1 , l2 , l3 , l4 . Clearly, P ⊆ P1 , since [OAB] has
maximum area. Note that A, B, A0 , B 0 are midpoints of the sides of P1 . Therefore, ABA0 B 0
is a parallelogram. Let P2 be this parallelogram. Then
|P1 | = 2|P2 |.
Note that P2 ⊆ P . Let P3 be the smallest parallelogram whose sides are parallel to that of
P2 such that P ⊆ P3 . Note that each side of P3 contains a vertex of P . Let X be a point
on the side of P3 “closest” to AB which is on P . Let X 0 be the image of the reflection of X
about O. Note that X 0 is on the side of P3 closest to A0 B 0 which is on P . Let Y be a point on
the side of P3 closest to BA0 which is on P . Let Y 0 be the image of the reflection of Y about O.
Let x be the ratio of the distance from X to AB to the distance from O to AB. Let y be the
ratio of the distance from Y to BA0 to the distance from O to BA0 . Note that the polygon
AXBY A0 X 0 B 0 Y 0 is contained in P , since P is convex. By symmetry, we have
|P2 | |P2 |
[AXBY A0 X 0 B 0 Y 0 ] = 2[OAXB] + 2[OBY A0 ] = 2(1 + x) + 2(1 + y)
4 4
|P2 | |P2 | |P2 |
= (1 + x) + (1 + y) = (2 + x + y).
2 2 2
Hence,
|P2 |
(2 + x + y) ≤ |P |.
2
Also, note that
2(1 + x)(1 + y)
|P3 | = (1 + x)(1 + y)|P2 | ≤ |P |
2+x+y
Therefore,
|P2 | 2 |P1 | 4
≤ ⇒ ≤
|P | 2+x+y |P | 2+x+y
√ √
If 2 + x + y ≥ 2 2, then |P 1| 4
|P | ≤ 2+x+y ≤ 2. Therefore, P1 satisfies the desired conditions.
√
Otherwise, we may assume that 2 + x + y < 2 2. Let a = 1 + x, b = 1 + y. Therefore,
|P3 | 2ab (a+b)2 a+b
√
|P | ≤ a+b ≤ 2(a+b) = 2 < 2. Thus, P3 satisfies the desired conditions. We are done.
2010 Mock 4
Time: 4.5 Hours
1. Find the largest possible integer k, such that the following statement is true:
Let 2009 arbitrary non-degenerated triangles be given. In every triangle the three sides are
colored, such that one is blue, one is red and one is white. Now, for every color separately,
let us sort the lengths of the sides. We obtain
b1 ≤ b2 ≤ · · · ≤ b2009 the lengths of the blue sides

r1 ≤ r2 ≤ · · · ≤ r2009 the lengths of the red sides
and w1 ≤ w2 ≤ · · · ≤ w2009 the lengths of the white sides.
Then there exist k indices j such that we can form a non-degenerated triangle with side
lengths bj , rj , wj .
Solution: The answer is k = 1.
We first prove that k > 0. Suppose k = 0. Then WLOG, suppose b2009 ≤ r2009 ≤ w2009 .
Since k = 0, these three numbers do not form the sides of a non-dengerate triangle. Therefore,
b2009 + r2009 ≤ w2009 . Since bi ≤ b2009 and rj ≤ r2009 for all i, j ∈ {1, · · · , 2009}, bi + rj ≤ w2009
for all i, j ∈ {1, · · · , 2009}. Therefore, w2009 cannot be the side length of any of the 2009 tri-
angles. This is a contradiction.
Therefore, k ≥ 1. The following is an example for which k = 1. For i ∈ {1, · · · , 2008}, let
bi = i, ri = 2i, wi = 3i.
Let
b2009 = 2009, r2009 = 2008 · 3, w2009 = 2008 · 3.
Then for all i ∈ {2, · · · , 2009}, it is easy to verify that (bi , ri−1 , wi−1 ) form the sides of a trian-
gle, since bi = i, ri−1 = 2i − 2 and wi−1 = 3i − 3 and i + (2i − 2) > 3i − 3, and (b1 , r2009 , w2009 )
form the sides of a triangle, since 1, a, a are the sides of a triangle for any positive integer a.

2. Given a cyclic quadrilateral ABCD, let the diagonals AC and BD meet at E and the lines
AD and BC meet at F . The midpoints of AB and CD are G and H, respectively. Show that
EF is tangent at E to the circle through the points E, G and H.
Solution: WLOG, suppose A, B are closer to F than D, C, respectively. By tangency prop-

erties, it suffices to show that ∠F EG = ∠GHE.
Note that ∆F CD ∼ ∆F AB. Let r be the ratio of similarity of ∆F CD to ∆F AB. Let ϕ

map the plane to itself by reflecting a point about the angle bisector of ∠AF B and dilating
the result by a ratio of r. We note the following properties of ϕ:
• Since reflection and dilation preserves the collinearity of any three points, ϕ preserves the
collinearity of any three collinear points.
• ϕ(A) = C, ϕ(B) = D and ϕ(G) = H by the similarity of ∆F AB and ∆F CD.
Let C 0 = ϕ(C) and D0 = ϕ(D). Then by the similarity, C 0 lies on line AD and C 0 C||BD.
Similarly, D0 lies on BC and DD0 ||AC. Let E 0 = CC 0 ∩ DD0 . Since ϕ preserves collinear-
ity and (A, E, C) is a triple of collinear points, (ϕ(A), ϕ(E), ϕ(C)) = (C, ϕ(E), C 0 ) is a
triple of collinear points. Similarly, (D, ϕ(E), D0 ) is a triple of collinear points. Hence,
ϕ(E) = CC 0 ∩ DD0 = E 0 . Note that EDE 0 C is a parallelogram. Since H is the midpoint of
CD, EH = HE 0 and E, H, E 0 are collinear. Since ϕ preserves angles, ∠F EG = ∠F E 0 H. We
want to show this angle is equal to ∠GHE to solve the problem; i.e. GH||F E 0 .
Let E 00 = ϕ(E 0 ) and H 0 = ϕ(H). Then by similarity, H 0 is the midpoint of C 0 D0 . Using

the same argument as before, we have E 0 H 0 = H 0 E 00 and E 0 , H 0 , E 00 are collinear. Therefore,
EE 00 ||HH 0 . Note that ϕ(ϕ()) is a dilation about F . Therefore, F, E, E 00 are collinear. There-
fore, F E||HH 0 . Since ϕ preserves parallel lines, F E 0 ||H 0 ϕ(H 0 ). But since ϕ(ϕ()) is a dilation,
GH||ϕ(ϕ(G))ϕ(ϕ(H)) = H 0 ϕ(H 0 ). Hence, F E 0 ||GH, as desired.
3. There exists a sequence of positive integers a1 , a2 , · · · , an satisfying
a2k + 1
ak+1 = −1
ak−1 + 1
for every k with 2 ≤ k ≤ n − 1. Prove that n ≤ 4.
Solution: Rewrite the equation as
(ak+1 + 1)(ak−1 + 1) = a2k + 1.
Since n ≥ 5,
(1) (a3 + 1)(a1 + 1) = a22 + 1.
(2) (a4 + 1)(a2 + 1) = a23 + 1.
(3) (a5 + 1)(a3 + 1) = a24 + 1.
We claim that each of a1 , a2 , a3 , a4 , a5 are even. If a1 is odd, then by (1), a2 is odd. Then by
(2), a3 is odd. Since a1 and a3 are both odd, in (1), a22 + 1 is divisible by 4. This is impossible
since no integer of the form a2 + 1 is divisible by 4. Therefore, a1 is even. If a2 is odd, then
similarly, a3 and a4 are odd. Sine a2 and a4 are both odd, in (2), a23 + 1 is divisible by 4. This
is again impossible. Therefore, a2 is even. If a3 is odd, then by (1), a2 is odd and by (3), a4
is odd. Again, a23 + 1 is divisible by 4, which is a contradiction. If a4 is odd, then by (2), a3 is
odd. By (1), a2 is odd. Again, a2 , a4 cannot both be odd. If a5 is odd, then a4 is odd, whose
case has already been handled. Therefore, each of a1 , · · · , a5 are even.
By (1) and (2), a3 + 1|a22 + 1 and a2 + 1|a23 + 1. It suffices to show that there are no pairs of
even positive integers x, y such that x + 1|y 2 + 1, y + 1|x2 + 1. Suppoe on the contrary that
there exist a pair of even numbers (x, y) such that x + 1|y 2 + 1 and y + 1|x2 + 1. Without
loss of generality, suppose x, y are chosen minimally and x ≥ y. If x = y, then x + 1|x2 + 1,
which implies that x + 1|2 ⇒ x = 1. This contradicts x being even. Therefore, x > y. The
conditions on x, y implies that x + 1|(y 2 + 1) + (x2 − 1) = x2 + y 2 . Similarly, y + 1|x2 + y 2 .
I claim that gcd(x + 1, y + 1) = 1. Let d = gcd(x + 1, y + 1). Then x, y ≡ −1 (mod d).
Hence, x2 + y 2 ≡ 2 (mod d). Since d|x + 1, d|x2 + y 2 . Hence, x2 + y 2 ≡ 0 (mod d). This
implies that 0 ≡ 2 (mod d). Therefore, d = 1 or d = 2. But d 6= 2 since x is even and
d|x + 1. Therefore, d = 1. Since x + 1|x2 + y 2 and y + 1|x2 + y 2 and gcd(x + 1, y + 1) = 1,
(x + 1)(y + 1)|x2 + y 2 . Let
x2 + y 2
= k. (16)
(x + 1)(y + 1)
for some positive integer k. This can be rewritten as x2 − k(y + 1)x + (y 2 − k(y + 1)) = 0.
Let x0 be the second root of t2 − k(y + 1)t + (y 2 − k(y + 1)) = 0, other than x. Note that x0
is an integer, since x is an integer and x + x0 = k(y + 1). Note that
x2 + y 2 y2 − x
k(y + 1) = =x+ .
x+1 x+1
2−x
If x < y 2 , then since x + x0 = k(y + 1) = x + yx+1 , x0 > 0. Since xx0 = (y 2 − k(y + 1)) and
x > y, x < y < x. Hence, x < x. We conclude that (x0 , y) is a smaller solution to (16),
0 0
contradicting the minimality of (x, y). If x > y 2 , then since x + 1|y 2 + 1, x + 1 ≤ y 2 + 1. Hence,
y 2 + 1 < x + 1 ≤ y 2 + 1. This is a contradiction. Finally, if x = y 2 , then since y + 1|x2 + 1,
y + 1|y 4 + 1. Since y + 1|y 4 − 1, y + 1|2. This implies that y = 1. This contradicts that y
is even. Therefore, there are no even numbers x, y such that x + 1|y 2 + 1 and y + 1|x2 + 1.
Hence, n 6≥ 5.
2010 Mock 5
Time: 4.5 Hours
1. Let O be the centre of the excircle of ∆ABC opposite A. Let M be the midpoint of AC, and
P the intersection of M O and BC. Prove that if ∠BAC = 2∠ACB, then AB = BP .
Solution 1: Let ∠ACB = θ. Then ∠BAC = 2θ and ∠ABC = 180 − 3θ. Note that O
lies on the angle bisector of ∠BAC. Therefore, ∠BAO = θ. Let D be the intersection
of AO and BC. To prove that BA = BP , it suffices to show that ∠BAP = 3θ/2, i.e.
∠DAP = ∠P AC = θ/2 ⇔ AP bisects ∠DAC ⇔ AD/AC = DP/P C. By Sine Law, we have
AD sin θ
= .
AC sin 2θ
Since M, P, O are collinear, by Menelaos Theorem on ∆ADC,
AO DP CM
· · = 1.
OD P C M A
Since CM = M A,
DP OD
= .
PC OA
Let r be the radius of the excircle of ∆ABC opposite A. Note ∠CDO = ∠BDA = 180 −
∠ABC − ∠BAO = 2θ. Let X, Y be the foot of the perpendicular from O on BC, AC,
r
respectively. Since OX = r, OD = sin ∠CDO = sinr2θ . Furthermore, OA = sin ∠CAO
r
= sinr θ .
Hence,
DP OA sin θ
= = .
PC OD sin 2θ
By looking at triangle ADC, by Sine Law,
AD sin ∠ACD sin ∠ACD sin θ DP
= = = = ,
AC sin ∠ADC sin(180 − ADC) sin 2θ PC
as desired.
Solution 2: Let AO and BC intersect at D. Since AO bisects ∠BAC and ∠BAC = 2∠ACB,
∠BAD = ∠CAD = ∠ACD. Therefore, AD = DC. Consider the triangles OAC and ODC.
Since they have equal altitudes from O and the same altitude from C,
[OAC] AC AO
= = .
[ODC] DC DO
Since M is the midpoint of AC, [OAM ] = [OCM ] and [P AM ] = [P CM ]. Therefore, [OAP ] =
[OCP ]. Then,
AC AC AO [OAP ] [OCP ] CP
= = = = = .
AD DC DO [ODP ] [ODP ] DP
Therefore, AP is the bisector of ∠DAC. It follows that ∠BAP = ∠BAD + ∠DAP =
∠ACP + ∠P AC = ∠AP B. Therefore, AB = BP , as desired.
2. Let f be a non-constant function from the set of positive integers into the set of positive
integers such that a − b divides f (a) − f (b) for all distinct positive integers a, b. Prove that
there exist infinitely many primes p such that p divides f (c) for some positive integer c.
Solution: Suppose the conclusion of the problem statement is false. Then there exist a finitely
number of primes p1 , · · · , pt such that the prime divisors of f (n) is a subset of {p1 , · · · , pt } for
every n ∈ N
First, I claim that there exists a positive integer N such that m ≥ N implies f (m) 6= f (1).
Suppose the contrary; i.e. there are infinitely many positive integers m such that f (m) = f (1).
Let k be the smallest positive integer such that f (k) 6= f (1). Such a positive integer exist
since f is a non-constant function. Let d = |f (k) − f (1)|. If f (m) = f (1) for some m > k + d.
Then m − k | |f (m) − f (k)| = |f (1) − f (k)| = d 6= 0 and m − k > d. This is a contradiction.
Let f (1) = pa11 · · · pat t for some non-negative integers a1 , · · · , at . Note that pm m m
1 p2 · · · pt di-
m m
vides f (p1 · · · pt + 1) − f (1). Consider this statement when m > max{a1 , · · · , at }. Then
pai i ||f (pm m m m a1 at
1 · · · pt +1) for all such m. This implies that f (p1 · · · pt +1) = p1 · · · pt = f (1), since
the only possible prime divisors of f (pm m m m
1 · · · pt + 1). But by our claim, f (p1 · · · pt + 1) > f (1)
when m gets sufficiently large. This is a contradiction.
Hence, there are indeed infinitely many primes p such that p divides f (c) for some positive
integer c.
3. A multiple choice contest with n questions was written by K students. The jury assigns the
difficulty to each question - a positive integer which is awarded to each student who solves the
question. If a student does not solve the question, the student gets 0 points. The student’s
score is the sum of the scores received for each question. It turns out that when the answer
sheets are submitted, the jury can assign the difficulty of each question in a way so that the
ranks of the students are in any pre-determined order. What is the maximum possible value
of K?
Solution: The maximum value of K is n.
Label the students S1 , · · · , Sn . An example of n students where the rank of the students can
be pre-determined is by having each student solve a unique question Qi . If the ranking order
is S1 , · · · , Sn , then assign problem Qi a score of n + 1 − i for each i ∈ {1, · · · , n}.
Assume the result holds for K ≥ n + 1. Take n + 1 different students, and clone each of them
infinitely many times so that we get n + 1 types of students. Let us show that if we can find
two non-identical teams with finitely many students but with the same results, i.e. for every
question, the number of students on each team who solved the question is the same, we arrive
at a contradiction.
We can assume every type of students is present on at most one team (otherwise keep remov-
ing 1 student of the same type from both teams until this is no longer possible). Without
loss of generality, the first team has at least as many students as the second. The sum of the
scores of each of the two teams are the same. Then it is impossible to assign the difficulty
of questions so that each student on the first team is ranked higher than each student on the
second team.
It remains to find the two teams. We will suppose that it is possible and construct a system
of n linear equations on n + 1 variables. The non-zero solution of this system will correspond
to the assignment of the teams.
For j = 1, 2, · · · , n + 1, if the contestants that solved problem j are on team 1, then let xj be
the number of such contestants. If the contestants that solved problem j are on team 2, then
let xj be the negative number of such contestants.
For i = 1, 2, · · · , n let equation i state that the difference between the number of contestants
on the two teams who solved question i is 0; this is an equation involving only variables
x1 , x2 , · · · , xn+1 . We get a homogenous system of n equations and n + 1 unknowns. It has
a non-zero solution; it must be rational since all the coefficients in the equation are rational
numbers (namely, 0 or 1). Multiplying it by an appropriate constant still gives a solution
which contains only integers. This gives us the two teams.
Winter Camp 2011 Warmup Solutions
1 Problems
1.1 Algebra
A1. Prove that for any positive numbers a, b, c the following inequality holds:
p p p p
ab(a + b) + bc(b + c) + ca(c + a) > (a + b)(b + c)(c + a).
A2. Find all pairs of positive real numbers (x, y) that satisfy both
2 +y 2
2x + 2y +x = 8, and
√ √
x + y = 2.
A3. Let x1 , x2 , . . . , xn be real numbers. Prove that

n
X n
X
|xi + xj | ≥ n |xi |.
i,j=1 i=1
A4. Find all functions f mapping rational numbers to real numbers, and satisfying
f (a + b + c) + f (a) + f (b) + f (c) = f (a + b) + f (b + c) + f (c + a) + f (0)
for rational numbers a, b, c.
1.2 Combinatorics
C1. There are eight rooks on a chessboard, no two attacking each other. Prove that some two of
the pairwise distances between the rooks are equal. (The distance between two rooks is the
distance between the centers of their cells.)
C2. Every vertex of the unit squares on an n × m grid is coloured either blue, green, or red, such
that all the vertices on the boundary of the board are coloured red. We say that a unit square
on the board is properly coloured if exactly one pair of adjacent vertices of the square are the
same colour. Show that the number of properly coloured squares is even.
C3. 3n points are marked on a circle, dividing it into 3n arcs, n of which have length 1, n others
have length 2, and n have length 3. Prove that it is possible to find two marked points
diametrically opposite to each other.
C4. In a round robin chess tournament each player faces every other player exactly once. For
every game, 1 point is awarded for a win, 0.5 points for a draw, and 0 points for a loss. Given
a positive integer m, a tournament is m−special if among every set S of m players, there is
one player who won all her games against the other m − 1 players in S and one player who
lost all her games against the other m − 1 players in S. For a given integer m ≥ 4, determine
the minimum value of n (as a function of m) such that in every m−special round robin chess
tournament with n players, the final scores of the n players are all distinct.
1 of 12
1.3 Geometry
G1. Let ABCD be a fixed parallelogram with AB < BC, and let P be a variable point on side
CD. Let Q be the point on side BC so that P C = CQ. Prove that, as P moves, the
circumcircle of 4AP Q passes through a fixed point in addition to A.
G2. Let ABCD be a cyclic quadrilateral. The perpendiculars to AD and BC at A and C respec-
tively meet at M , and the perpendiculars to AD and BC at D and B meet at N . If the lines
AD and BC meet at E, prove that ∠DEN = ∠CEM .
G3. Let ABC be an acute-angled triangle with altitudes AD, BE, CF . Let the line tangent to the
circumcircle of 4ABC at A and the line tangent to the circumcircle of 4DEF at D intersect
at a point X. Define Y and Z analogously. Prove that X, Y, and Z are collinear.
G4. In an acute triangle ABC, the angle bisector of ∠A meets side BC at D. Let E and F be the
feet of the perpendiculars from D to AC and AB respectively. Lines BE and CF intersect
at H, and the circumcircle of 4AF H meets BE at H and G. Show that the triangle with
side lengths BG, GE, BF is right-angled.
1.4 Number Theory

N1. Can 1 be written in the form:
1 1 1
+ + ... + ,
3a1 − 1 3a2 − 1 3a2011 − 1
where each ai is a positive integer?
N2. A sequence (ai ) of natural numbers has the property that for all i 6= j, gcd(ai , aj ) = gcd(i, j).
Show that ai = i for all i.
x2 −1
N3. Let’s say a natural number is Yorky if it can be expressed in the form y 2 −1
for integers
x, y > 1. Is is true that all but finitely many natural numbers are Yorky?
N4. Let a1 , a2 , a3 , . . . be a sequence of positive integers such that a1 > 1 and iteratively, for all
n ≥ 2, an is the smallest positive integer such that
an 6∈ {a1 , a2 , . . . , an−1 } and gcd(an−1 , an ) > 1.
Prove that the sequence contains every positive integer greater than one.
2 of 12
2 Solutions
2.1 Algebra
A1.
p p p 2 X Xp
ab(a + b) + bc(b + c) + ca(c + a) = ab(a + b) + 2 ab2 c(a + b)(b + c)
cyc cyc
X
> ab(a + b) + 6abc
cyc
X
> ab(a + b) + 2abc
cyc
= (a + b)(b + c)(c + a).
Comment: Most Olympiad inequalities are tighter approximations, and steps like replacing
6abc with 2abc will not be good enough. Expanding can help though, and the cyclic sum
notation used here is a good way to write things down when you do expand.
Source: Russia 1990.
√ √ 1 √ √ 2
A2. Suppose x + y = 2. By the power-mean inequality, we have x + y ≥ 2 · x+ y ≥ 2,
and x2 + y 2 ≥ 12 · (x + y)2 = 2. Now, by the AM-GM inequality:
2 +y 2 +x
p
2x + 2y ≥ 2· 2x2 +y · 2y2 +x
x2 +y 2
+ x+y
= 21+ 2 2
≥ 8.
Equality holds if and only if x = y, so the only possible solution to the given pair of equations
is x = y = 1. Substituting this pair in, we see that it is indeed a valid solution.
Comment: Solving a system of equations like this requires doing two things: (a) proving
that nothing other than x = y = 1 works, and (b) proving that x = y = 1 does work. It is
always a good idea to check (b) separately. Your argument for (a) might take care of (b),
but it might not, and it’s better to be safe than sorry.
Source: Olymon 1997-1998.
A3. Let a = maxi∈{1,2,...,n} |xi |, let S0 denote the set of indices i such that xi = a, let S1 denote
the set of indices i such that xi = −a, and let S2 denote the remaining indices. Finally, let
b = maxi∈S2 |xi |. (If S2 is empty, then define b = 0.)
Consider replacing each xi in S0 with b and each xi in S1 with −b. This will have the following
effect on each |xi + xj | term:
• If i, j are both in S0 or both in S1 , then |xi + xj | will decrease from 2a to 2b. There are
exactly |S0 |2 + |S1 |2 ordered pairs (i, j) for which this case applies.
• If one of i, j is in S0 and the other is in S1 , then |xi + xj | will be 0 both before and after
the replacement, and therefore will be unchanged.
3 of 12
• If one of i, j is in S0 ∪ S1 and the other is in S2 , then |xi + xj | will decrease by exactly

a − b. (Assume without loss of generality that xi = a, and let c = xj ; then |xi + xj | will
decrease from a + c to b + c. There will be no sign change since |c| ≤ b ≤ a.) There are
exactly 2 · (|S0 | + |S1 |) · |S2 | ordered pairs (i, j) for which this case applies.
• If i, j are both in S2 , then |xi + xj | will not change.
Therefore, ni,j=1 |xi + xj | will decrease by exactly
P
(a − b) · 2 · (|S0 | + |S1 |) · |S2 | + 2 · |S0 |2 + 2 · |S1 |2

≥ (a − b) · (|S0 | + |S1 |) · |S2 | + (|S0 | + |S1 |)2

= (a − b) · (|S0 | + |S1 |) · n.
On the other hand, n · ni=1 |xi | will decrease by exactly (a − b) · (|S0 | + |S1 |) · n. Therefore,
P
this replacement will always decrease the left-hand side by at least as much as it will decrease
the right-hand side by.
Performing this replacement repeatedly, we will eventually reach the configuration where
every xi is equal to 0, at which point it is obvious that the left-hand side is equal to the
right-hand side. Therefore, we must originally have had the left-hand side being greater than
or equal to the right-hand side, as required.
Solution #2: If {x1 , x2 , . . . , xn } contains only non-negative numbers or only non-positive
numbers, the claim is obvious.
Otherwise, assume without loss of generality that x1 , x2 , . . . , xm ≥ 0 and xm+1 , xm+2 , . . . , xn <
0. Let
x1 + . . . + xm xm+1 + . . . + xn
a= , b=− .
m n−m
Clearly, a, b ≥ 0. Returning to the original inequality, we can now write the left-hand side
minus the right-hand side in the following form:
m
X n
X m
X n
X m
X n
X
|xi + xj | + |xi + xj | + 2 |xi + xj | − n |xi | − n |xi |.
i,j=1 i,j=m+1 i=1 j=m+1 i=1 j=m+1
This simplifies further to

m
X n
X
2 2
LHS-RHS = 2m a + 2(n − m) b + 2 |xi + xj | − nma − n(n − m)b.
i=1 j=m+1
Now assume without loss of generality that a ≥ b. Since |x + c| is a convex function, Jensen’s
inequality implies LHS-RHS is greater than or equal to
n
2m2 a + 2(n − m)2 b + 2m
P
|a + xj | − nma − n(n − m)b
j=m+1
≥ 2m2 a + 2(n − m)2 b + 2m(n− m)(a − b) − nma − n(n − m)b
a 2m2 + 2m(n − m) − nm + b 2(n − m)2 − 2m(n − m) − n(n − m)

=
= amn + b(n2 − 5mn + 4m2 )
≥ bmn + b(n2 − 5mn + 4m2 )
= b(n2 − 4mn + 4n2 )
= b(n − 2m)2 ≥ 0.
4 of 12
Comments: The technique of changing one or more variables, and seeing what happens to
the inequality is called “smoothing” or “mixing variables”. Usually, you make two variables
equal, or reduce one of them to zero. On some problems, smoothing is too messy to be useful,
but when the algebra is not too complicated, it is a very powerful technique.
Source: Iran 2006.
A4. Let us say a function f is “good” if it satisfies the given equation. Note that:
• If f1 and f2 are good, then so is f1 + f2 .

• For all rational numbers A, B, C, we have f (x) = Ax2 is good, f (x) = Bx is good, and
f (x) = C is good. Therefore, all quadratical polynomials are good.
We claim this is the complete set of good functions.
Fact1 : Fix real numbers a, b, c, t with t 6= 0. Then there exists a quadratic polynomial
P (x) such that P (−t) = a, P (0) = b, and P (t) = c.
Lemma: Fix a positive integer

n. If f is good, then there exists a quadratic polynomial
m m

P (x) such that f n = P n for all integers m.
Proof of lemma: Let P (x) be the quadratic polynomial that equals f (x) for x = −1 1
n , 0, n ,
−1 1
and let g(x) = f (x) − P (x). Then g is a good function satisfying g n = g(0) = g n = 0.
m
We need to show g n = 0 for all m.
Taking a = b = n1 and c = −1 1 −1 2 2

n , we have 3g n + g n = g n + 3g(0), and hence g n = 0.
Now, for any m, take a = b = n1 and c = m−2 n . Then

m 1 m−2 m−1 2
g + 2g +g = 2g +g + g(0),
n n n n n
and hence g m m−1

− g m−2 m m−2
and g m−1

n = 2g n n . Therefore, g n is 0 if both g n n
are 0. It follows immediately that g mn is 0 for all positive integers m. Similarly, g
m−2
n is
m−1 m m

0 if both g n and g n are 0, and hence g n is also 0 for all negative integers m.
Now pick an arbitrary good function f . By the lemma, there exists a quadratic polynomial P
t
suchthat f (m) = P (m) for all integers m. Suppose there exists a rational number n so that
t t
f n 6= P n . Applying the lemma again, there exists a quadratic polynomial P2 such that
f m m
n = P2 n for all m. But then P − P2 is a quadratic polynomial that has roots at -1,
0, and 1, but is non-zero at nt . However, this is impossible: a non-zero quadratic polynomial
can have at most 2 roots. Therefore it must be that f (x) ≡ P (x), and the problem is solved.
Comments: This problem generalizes Cauchy’s equation: f (a) + f (b) = f (a + b). Like
Cauchy’s equation, there can be all sorts of weird solutions if you let the domain of f be the
set of all real numbers. It depends on whether you assume the Axiom of Choice, and is far
beyond the scope of the IMO!
1
You can quote the following more general theorem: Given any k + 1 pairs (x1 , y1 ), (x2 , y2 ), . . . , (xk+1 , yk+1 ) with
xi 6= xj , there exists a unique degree-k polynomial P satisfying P (xi ) = yi for all i. In our case, the polynomial is
P (x) = a+c−2b
2t2
· x2 + c−a
2t
· x + b.
5 of 12
2.2 Combinatorics
C1. Suppose that 8 rooks are placed on a chessboard with no two attacking each other. Then
there must be one rook in each row and one in each column. Consider a pair of rooks that
are
√ either √in adjacent√ rows or in √ adjacent columns. The distance between them is one of
12 + 12 , 12 + 22 , 12 + 32 , . . . , 12 + 72 . Note that there are exactly 7 distinct distances
of this form.
If all rook pairs are a different distance apart, it follows that there are at most 7 rook pairs
in adjacent rows or columns. However, there are exactly 7 pairs in adjacent rows (the rook
in row 1 paired with the rook in row 2, the rook in row 2 paired with the rook in row 3, etc.),
and exactly 7 pairs in adjacent columns. If these combine to give only 7 pairs altogether, it
must be that any time two rooks are in adjacent rows, they are also in adjacent √ columns. But
in this case, all 7 of these rook pairs are separated by a distance of exactly 2, and we have
a contradiction.
Solution #2: If there are 8 rooks on a board, this gives 82 = 28 pairs of rooks. The

p
distance between any pair must be of the form i2 + j 2 for i, j ∈ {1, 2, . . . , 7}. There are
7

exactly 2 +7 = 28 such distances. Therefore, if every pair of rooks is separated by a different
distance, each of these 28 distances must be used.
√
In particular, there are two rooks separated by a distance of 72 + 72 , meaning there are two
rooks in opposite corners. But then no more rooks can be placed on the boundary√without
attacking one of these two. Hence, no two rooks can be separated by a distance of 72 + 62 ,
and we have a contradiction.
Comments: With Pigeonhole Principle arguments like this, it is always a good idea to count
the number of possible distances and the number of rook pairs. In this
case, the total number
of rook pairs and the total number of possible distances are both 82 , which is not enough for
a contradiction. You need to tweak the argument a little bit, like we did here.
C2. We will call an edge of a square good if both its endpoints are of the same colour, and bad
otherwise. Any square can have either no good edges (call such squares type A), one good
edge (type B), two good edges (type C), or four good edges (type D). Note that it cannot
have three good edges.
Let the number of squares of type A, B, C, D be a, b, c, d respectively. For each square, count
the number of good edges, and add the numbers. We must get an even number, since each
edge not on the boundary is counted twice, and all 2(m + n) edges on the boundary are good.
On the other hand, this number is b + 2c + 4d. Hence b is even and the result follows.
Comments: We are counting in two different ways the number of (square, good edge) pairs.
This kind of argument is very important in combinatorics problems!
Source: India 2001.
C3. Assume it is not possible to find the two required points.

Starting at any marked point, divide the circumference into 6n arcs of length 1. Colour the
endpoints of these arcs blue if they are marked, and red otherwise. By assumption, each blue
point is diametrically opposite a red point. Since the total number of red points and the total
6 of 12
number of blue points are both equal to 3n, all red points must be used in this way, and
therefore each red point is also diametrically opposite a blue point.
Let us call any arc between two adjacent blue points “elementary”. Consider an elementary
arc AC of length 2. Its endpoints are blue, and its midpoint B 0 is red. As argued above, the
point B diametrically opposite to B 0 must be blue.
Now suppose arc AB contains a1 elementary arcs of length 1, a2 elementary arcs of length
2, and a3 elementary arcs of length 3. Each length-1 elementary arc in AB is diametrically
opposite a length-3 elementary arc in BC: the two blue endpoints of the length-1 arc must
be opposite red points and no elementary arc can have length greater than 3. Conversely,
each length-3 elementary arc in BC is diametrically opposite a length-1 elementary arc in
AB. Therefore, the total number of length-3 elementary arcs in BC is exactly a1 . Since the
total number of length-3 elementary arcs on the whole circle is k, we have a1 = k − a3 .
Finally, arc AB is of length 3k − 1, so we have 3k − 1 = a1 + 2a2 + 3a3 = 2a2 + 2a2 + k.
Therefore, 2k − 1 = 2a1 + 2a2 , but this is impossible because the left-hand side is odd, and
the right-hand side is even.
C4. The answer is 2m − 3.

For n ≤ 2m − 4, consider a tournament with players P1 , P2 , . . . , Pn where Pi beats Pj for all
i > j except Pm−1 draws against Pm−2 . Consider any set of m players, and let i, j denote
the highest and lowest index respectively of any player in this set. Clearly, i ≥ m and since
n ≤ 2m − 4, we also have j ≤ m − 3. Since i, j are not m − 1 and m − 2, we know that Pi
beat all other players in the set, and Pj lost to all other players in the set. Therefore, the
tournament is m-special. Since players Pm−2 and Pm−1 have identical scores, the answer to
the problem must be at least 2m − 3.
Conversely, consider an arbitrary m-special tournament. First suppose two players P1 and
P2 drew against each other. Let A be the set of players who beat or tied P1 , and let B be
the set of players who lost to P1 . If |A| ≥ m − 2, then consider a set of m − 2 players from
A, along with P1 and P2 . Each player in A beat or drew against P1 , P1 drew against P2 ,
and P2 drew against P1 . Therefore, no player lost to all other players in the set, and so the
tournament cannot be m-special, which is a contradiction. Therefore, |A| ≤ m − 3. Similarly,
|B| ≤ m − 3. Since A and B together comprise all players other than P1 and P2 , we must
have n ≤ 2 + 2(m − 3) = 2m − 4.
Next suppose there exists a “3-cycle”: three players P1 , P2 , P3 such that P1 beat P2 , P2 beat
P3 , and P3 beat P1 . Let A be the set of players who beat or tied P1 , and let B be the set of
players who lost to P1 . Looking at size-m sets containing all of P1 , P2 , P3 as above, we find
|A|, |B| ≤ m − 4, and hence n ≤ 2m − 5.
Finally, consider an arbitrary tournament with no draws and with no 3-cycles. Let P be the
player with the largest number of wins. Suppose some player Q is not beaten by P . Since
there are no ties, it must be that Q beat P . Since there are no 3-cycles, it must also be
that Q beat every player beaten by P , and hence Q has more wins than P . This contradicts
the definition of P , so the only possibility is that P beat all other players. Label P as Pn
and now restrict our attention to the remaining n − 1 players who together form a smaller
7 of 12
tournament with no draws and with no 3-cycles. By the same argument, this tournament also
has a player Pn−1 who beats all remaining players. Proceeding recursively, we can label the
players P1 , P2 , . . . , Pn so that player Pi beats Pj whenever i > j. Notice that no two players
can have the same score for this kind of tournament.
Therefore, if two players have the same score, then the tournament must have a tie or a
3-cycle. In the case of m-special tournaments, we showed that this means n ≤ 2m − 4, and
the proof is complete.
Comments: In the final case, we proved one player had beaten all other players by starting
with the player who had the most wins and analyzing what properties she must have. This
is called an extremal argument, and it is very useful.
Source: China 2007.
2.3 Geometry
G1. Let E be the reflection of A through the angle bisector of ∠BCD. Since P is the reflection of
Q through the same line, it follows that P Q and AE are both perpendicular to the bisector,
and that AQ = EP . In particular, AQP E is an isosceles trapezoid, and is therefore cyclic.
It follows that the circumcircle of 4AP Q passes through E no matter where P is.
B Q C
A D
Comments: For many problems where you need a show a shape always passes through a
fixed point, the key is to first figure out where that point is, and then go from there. Always
have a straight-edge and compass ready to help with this kind of guessing!
G2. Assume without loss of generality that E is closer to AB than to CD. Since ∠EBN =
∠EDN = 90◦ , the points E, B, D, N lie on a circle. (We do not know what order they lie in
however.) Similarly, the points E, A, C, M lie on a circle.
Now, suppose segments BN and DE cross. Then ∠EBD > ∠EBN = 90◦ , and ∠EAC =
180◦ − ∠CEA − ∠ECA = 180◦ − ∠DEB − ∠EDB = ∠EBD > 90◦ . It follows that segments
AM and CE also cross, as shown in the left diagram.
In this configuration, we have ∠DEN = 180◦ − ∠BED − ∠BDN = 90◦ − ∠BED − ∠BDA =
90◦ − ∠AEC − ∠ACB = 180◦ − ∠AEC − ∠ACM = ∠CEM .
8 of 12
If segments AM and CE cross, then segments BN and DE also cross as argued above, and
we are in the same case as before. Otherwise, it must be that AM does not cross CE, and
BN does not cross DE, as shown in the right diagram.
In this configuration, we have ∠DEN = ∠DBN = ∠DBC −90◦ = ∠CAD −90◦ = ∠CAM =
∠CEM . Thus, ∠DEN = ∠CEM in all possible configurations.
N
D
A
M C
E
B
B
C E
A
N
M D
Solution #2: We know ∠ACE = ∠ACB = ∠BDA = ∠BDE and ∠CEA = ∠DEB, so
4EAC must be similar to 4EBD.
Also, E, A, C, M lie on a circle with diameter EM since ∠EAM = ∠ECM = 90◦ . In
particular, this means AM is the diameter of the circumcircle of 4EAC. Similarly, BN is
the diameter of the circumcircle of 4EBD.
Now consider the similarity transformation (rotation, scaling, translation, and/or reflection)
taking 4EAC to 4EBD. This transformation will also take the point on the circumcircle
of 4EAC opposite A to the point on the circumcircle of 4EBD opposite B; i.e., it will take
M to N . Thus, the entire figure EACM is similar to EBDN , and hence ∠DEN = ∠CEM .
Comments: Many students will just assume one configuration and ignore the possibility
of the other one, even when the argument needs to be tweaked. This will usually cost you
a point. In this problem, it would cost you multiple points. Always, always, always pay
attention to all possible configurations!
Source: Iran 2004.
G3. Let 4ABC have angles a, b, c, and assume without loss of generality that X lies on the same
side of AD as C. Then ∠XAD = ∠XAC + ∠CAD = b + (90◦ − c).
Now, ∠AF C = ∠ADC = 90◦ so AF DC is a cyclic quadrilateral. Similar arguments apply
for BDEA and CEF B. Therefore,
∠XDA = ∠XDE + ∠EDA = ∠DF E + ∠EDA = ∠DF C + ∠CF E + ∠EDA
= ∠DAC + ∠CBE + ∠EBA = (90◦ − c) + (90◦ − c) + (90◦ − a)
= 90◦ + b − c = ∠XAD.
Therefore, 4XAD is isosceles and XA = XD. Since XA and XD are both tangents, it
follows that X has equal power with respect to the circumcircles of 4ABC and 4DEF . In
9 of 12
particular, X is on the radical axis of these circles. The same argument can be applied to Y
and Z, and so all three points lie on the one line as required.
C
X
D
A F B
Comments: As with G1, the trick here is to guess the line in advance. Once you know the
goal is to prove X, Y, Z are on the radical axis of the two circles, it is straightforward. If you
don’t know what power of a point and radical axes are, you should look them up. They are
important!
G4. First note that 4AF D ∼

= 4AED by angle-angle-side. Now extend AH to meet BC at K.
Since AF = AE, Ceva’s theorem guarantees that:
BK BF
=
CK CE
BD · cos B
=
CD · cos C
BA · cos B
= by the angle bisector theorem.
CA · cos C
BK 0
Now let K 0 be the base of the altitude from A to BC. Then CK 0 = BA·cos B
CA·cos C = BK
CK , so
K = K 0 , and AK is perpendicular to BC.
Since ∠AF D = ∠AKD = 90◦ , it follows that AF DK is a cyclic quadrilateral. By power of
a point on circles AF DK and AF GH, we have BK · BD = BF · BA = BH · BG. Therefore,
KDGH is also a cyclic quadrilateral. Since ∠DKH = 90◦ , it has diameter DH and hence
∠DGH = 90◦ as well. In other words, DG is perpendicular to BE.
Therefore, BG2 − GE 2 = BD2 − DE 2 = BD2 − DF 2 = BF 2 as required.
E
K
D
H
G
A F B
10 of 12
Comments: The key observations are that AK is perpendicular to BC and DG is perpen-

dicular to BE, both of which you would probably see if you made a good straight-edge and
compass diagram!
Source: China 2003.
2.4 Number Theory

Q2011 Q
N1. The answer is no. Let X = i=1 (3ai − 1), and let xi = j6=i (3aj − 1). Note that
2011 2011
X 1 1 X
= · xi .
3ai − 1 X
i=1 i=1
3 however, we have X ≡ 22011 ≡ 2, and 2011

P P2011 2010
Working moduloP i=1 xi ≡ i=1 2 ≡ 2011·1 ≡ 1.
Therefore, X 6= 2011
i=1 x i , and hence the original sum cannot be equal to 1.
Comments: If you look at the equation mod 3, the argument is intuitively clear, but a
nuisance to write up clearly. Think about how you would do it.
Source: Ukraine 2005.
N2. For any integer n, we have n = gcd(n, 2n) = gcd(an , a2n ), which divides an . Therefore, n
divides an for all n.
Now suppose there exists n for which an 6= n. Then an = nm for some m > 1. Since nm
also divides anm , we have nm| gcd(an , anm ). However, nm does not divide n = gcd(n, nm),
so this is a contradiction.
It follows that an = n for all n.
N3. Let p be an odd prime. We claim that p2 is not Yorky, which implies there are infinitely many
non-Yorky numbers.
x2 −1
Indeed, suppose y 2 −1
= p2 . Then:
x2 − 1 = y 2 p2 − p2
⇒ p2 − 1 = y 2 p2 − x2
⇒ p2 − 1 = (yp − x)(yp + x).
If yp − x = 1, then yp + x = p2 − 1 and 2yp = p2 , contradicting the fact that p is odd.

Otherwise, yp − x > 1. Taking the equation modulo p, we have x2 ≡ 1 mod p, and hence
yp − x ≡ −x ≡ ±1 mod p. Since yp − x is positive and greater than 1, it follows that
yp − x ≥ p − 1. Also, x must be greater than 1, so yp + x > yp − x + 2 ≥ p + 1. But then
(yp − x)(yp + x) > (p − 1)(p + 1), and we have a contradiction.
Comments: Unless you are 100% completely sure you know whether there are infinitely
many non-Yorky numbers, do NOT put all your eggs in one basket. Spend time both trying
to prove there are infinitely many, and trying to prove there are not. When trying to construct
non-Yorky numbers, it is natural to look at perfect squares because they let you factor the
11 of 12
expression. The advantage of trying to construct examples is you can restrict your attention
to whatever is easy to work with!
N4. Lemma 1: Let n > 1 be a positive integer. If the sequence contains infinitely many multiples
of n, then it contains all multiples of n.
Proof of lemma 1: Let kn be an arbitrary multiple of n. Consider the first kn + 1 mul-

tiples of n in the sequence. Each of these terms shares a non-trivial divisor with n, and
each is followed by a distinct number. Eventually these numbers will become larger than kn,
at which point kn has to be used. Therefore, kn will appear in the sequence at some point.
Lemma 2: Let n > 1 be a positive integer. If the sequence contains infinitely many multiples
of n, then it contains all positive integers greater than 1.
Proof of lemma 2: Let m > 1 be an arbitrary positive integer. By Lemma 1, the se-
quence contains all multiples of n, including all multiples of nm. Therefore, the sequence
contains infinitely many multiples of m, which means it contains all multiples of m, including
m itself. Since m was arbitrary, the proof is now complete.
We now show that the sequence must contain infinitely many prime numbers. Suppose
by way of contradiction that it contains only the primes p1 , p2 , . . . , pk . By Lemma 2, it can
contain only finitely many multiples of each of these primes. Therefore, it must eventually
reach a number m not divisible by any of them. Let q be the smallest prime factor of m.
Then q is the smallest number sharing a non-trivial divisor with m, so m will be immediately
followed by q if q has not already appeared in the sequence. Either way, q appears at some
point, and we have a contradiction.
Therefore, the sequence does indeed contain infinitely many prime numbers. For any prime p,
the smallest number sharing a non-trivial divisor with p is 2p. If 2p has not already appeared
in the sequence, it will therefore immediately follow p. Thus, having infinitely many primes
in the sequence guarantees that there will be infinitely many even numbers, and we are now
done by Lemma 2.
Comments: The devil here is in the details. Take a look at how the proof is made rigorous,
and make sure you could do the same thing!
Source: China 2010.
12 of 12
Winter Camp 2011 Buffet Problems
1 Problems
1.1 Algebra
A1. Positive real numbers x, y, z are given such that the difference between any two of them is
less than 2. Prove that:
p p √
xy + 1 + yz + 1 + zx + 1 > x + y + z
A2. Let M be a set of n ≥ 4 points in the plane, no three of which are collinear, and not all lying
on a circle. Suppose that f is a function assigning a real number to each point in M such
that for any circle C passing through at least three points of M ,
X
f (P ) = 0.
P ∈M ∩C
Prove that f (P ) = 0 for all points P .
A3. A sequence of non-negative real numbers a1 , a2 , . . . , an is given. For k = 1, 2 . . . , n, let mk be

equal to
ak−l+1 + ak−l+2 + . . . + ak
max .
1≤i≤k l
Prove that for any x > 0, the number of integers k such that mk ≥ x, is less than or equal to
a1 +a2 +...+an
x .
1.2 Combinaotircs
C1. All entries of an 8 × 8 matrix are positive integers. One may repeatedly transform the entries
of the matrix according to the following rules:
(a) Multiply all entries in the same row by 2.

(b) Subtract 1 from all entries in the same column.
Prove that it is possible to transform the given matrix into the zero matrix.
C2. Let n, k be positive integers and I1 , I2 , . . . , In be n closed intervals on a line such that among
any k of the intervals, there are two with non-empty intersection. Prove that one can choose
k − 1 points on the line such that any of the intervals contains at least one of the chosen
points.
C3. In a certain country, every town is connected by a road to exactly three other towns. A
tourist traveling by roads visited each town exactly once and returned to the initial town.
Next year he comes back for a round trip different from the last years trip (not the original
path in either regular or reverse order), again visiting each town exactly once. Prove that he
can always do so.
1 of 8
1.3 Geometry
G1. Let ABC be a triangle. Let P be the point on line BC such that B is between P and C, and
BP = BA. Similarly, let Q be the point on line BC such that C is between Q and B, and
CQ = CA. If R is the second intersection of the circumcircles of 4ACP and 4ABQ, prove
that 4P QR is isosceles.
G2. Points K, L, M, N are respectively the midpoints of sides AB, BC, CD, DA in a convex quadri-
lateral. Line KM meets diagonals AC and BD at points P and Q, respectively. Line LN
meets diagonals AC and BD at points R and S, respectively. Prove that if AP ·P C = BQ·QD,
then AR · RC = BS · SD.
G3. Let ABCD be a convex quadrilateral whose opposite sides are not parallel. Let E, F be the
intersections of the opposite sides of ABCD, P the intersection of AC and BD, and J the
foot of the perpendicular from P to EF . Prove that ∠AJD = ∠BJC.
1.4 Number Theory

N1. Show that the equation 3y 2 = x4 + x has no solutions in positive integers.
N2. Prove that for any positive integer n greater than 10000, there is a positive integer m that
√
can be written as a sum of two squares, such that 0 < m − n < 3 4 n.
N3. k is a given natural number. Find all functions f mapping natural numbers to natural
numbers, such that for all pairs of natural numbers m, n, we have
f (m) + f (n)|(m + n)k .
2 of 8
2 Solutions
2.1 Algebra
A1.
x+y 2
|x − y| < 2 ⇒ (x − y)2 < 4 ⇒ x2 + 2xy + y 2 < 4 + 4xy ⇒ ( ) < xy + 1
2
x+y p y+z p z+x √
Hence < xy + 1. Similarly < yz + 1 and < zx + 1. Adding these
2 2 2
three inequalities we get the result.

n
A2. Label all possible pairs of points in M from 1 to K = . Look at an arbitrary pair i of
2
points Ai , Bi in M and consider the set Ci of all circles passing through the two points in the
pair and through at least one other point in M . The number of circles in Ci is mi ≥ 2, since
not all points lie on a circle. Label the circles from ω1i to ωmi .
i
X
Let Si = f (Ai ) + f (Bi ) and S = f (P ). We have:
P ∈M
mi
X X S
S=[ f (P )] − (mi − 1)Si = (1 − mi )Si ⇒ Si = (1)
1 − mi
j=1 P ∈ωm
i
j
K K
X Si 1
X
Assume S 6= 0. Equation (1) holds for all i hence n−1 = = . But 1−mi < 0
S 1 − mi
i=1 i=1
for all i which gives n − 1 < 0, a contradiction. Hence S = 0, and equation (1) implies Si = 0
for all i, hence f (P ) = 0 for all points P .
Source: Romania 1998.
A3. Let bi = a1 + ... + ai . Then b1 ≤ b2 ≤ ... ≤ bn . For any positive integers l < k with m = k − l:
al+1 + al+2 + . . . + ak bk − bl
=
m k−l
bk − bl
Consider points B0 (0, 0), B1 (1, b1 ), B2 (2, b2 )..., Bn (n, bn ) in the coordinate plane. Then
k−l
is equal to the tangent of the angle formed by line Bk Bl and the x−axis. The condition
mk > x is equivalent to the condition that the line lk passing through Bk with the slope angle
of tan−1 (x) lies above at least one point Bl for l < k. We will call such a point Bk good.
a1 + a2 + . . . + an bn
Also = , which is the distance between the point (n, 0) and the point of
x x
intersection of line ln with the x−axis.
Let us prove by induction on n that this distance is greater than the number of good points.
The base case is clear. If point Bn is not good, remove it; then the number of good points
does not change and the distance decreases, since bn−1 ≤ bn . If Bn is good, let k be the
largest integer such that Bk lies below ln . Remove points Bk+1 through Bn ; the number of
good points will then decrease by n − k, and the distance will decrease by more than n − k,
which finishes the induction step.
3 of 8
2.2 Combinatorics
C1. Repeat the following procedure on the left-most column. If all integers in the left-most column
are greater than 1, subtract 1 from all entries in the column. Otherwise, multiply by 2 all
entries in every row containing a 1 in the column. Repeat this procedure until all entries in
the left-most column are 1. The process will eventually stop since the difference between the
smallest and the greatest entries in the column is non-decreasing, and it cannot stay constant
forever. Once all entries in the left-most column are 1, subtract 1 from every entry in the
column, so that the left-most column contains all zeroes. Once all entries in the left-most
column are 0, perform the same procedure on the next column, and so on.
Source: Netherlands 1999.
C2. Denote by P1 the left-most right endpoint of an interval. Throw away all intervals containing
P1 , and let P2 be the left-most right endpoint of the remaining intervals. Define P3 , P4 . . . in
the same way until all intervals have been thrown away. Let Pm be the last point defined. The
intervals with the right endpoints P1 , P2 , . . . , Pm cannot intersect, hence m ≤ k − 1. Every
interval that has been thrown away contains at least one of the chosen points, hence every
interval in the collection contains one of the chosen points.
C3. Consider a graph where the vertices are the towns, and edges are the roads. We call a directed
path P1 , P2 , . . . , Pn Hamiltonian if the path contains every vertex in the graph exactly once.
We call such a Hamiltonian path adjacent to another Hamiltonian path, if the latter path can
be obtained from the first one by inserting edge Pn Pi and deleting edge Pi Pi+1 for i > 1, so
that the latter path is P1 , P2 , . . . , Pi , Pn , Pn−1 , . . . , Pi+1 . Since Pn is adjacent to exactly 3 other
vertices, every Hamiltonian path is adjacent to two paths or to one path, the latter case hap-
pening only if Pn is adjacent to P1 . (1)
Consider a graph G where vertices are the paths, and edges join adjacent paths. Look at the
cycle along which the tourist traveled the first time. Remove one edge in the cycle to get
a Hamiltonian path X = S1 , S2 , . . . , Sn . Look at the subgraph of G which contains path X
and all paths that can be reached from X by following the edges of G. This subgraph must
be itself a path because of (1). Let Y = T1 , T2 , . . . , Tn be the endpoint of this path different
from X. Then Tn is adjacent to T1 , so path Y can be extended to a Hamiltonian cycle. By
definition of adjacency of Hamiltonian paths, T1 = S1 , T2 = S2 , hence the new cycle is not
the reverse of the old one. Since X 6= Y , the new cycle is different from the old one, hence it
satisfies the required conditions.
Source: Japan 2004
Comment: The result also holds if we only assume that all vertices in the graph have odd
degree. Think about how the above solution needs to be modified for this more general
problem.
4 of 8
2.3 Geometry
G1. Solution 1: Let S be the intersection of AR with the circumcircle of 4P AQ. Then:
1 1
∠P SA = ∠AQP = ∠ACP = ∠ARP ⇒ ∠P SA = ∠SP R ⇒ RP = RS
2 2
Similarly RQ = RS. Hence RP = RQ.
P Q
B C
R
Solution 2: Since P ACR and QABR are cyclic, it follows that:

∠ABC ∠BCA
∠ARC = ∠AP C = ; ∠ARB = ∠AQB =
2 2
∠ABC
∠P RB = 180◦ − ∠P AC − ∠CRB = 180◦ − −
2
∠BCA ∠ABC ∠BCA
−∠CAB − − = = ∠BRA
2 2 2
Consider 4P AR. By Ceva’s Theorem in Sine Form applied to concurrent cevians P B, AB, RB,
sin(∠RP B) sin(∠P AB) sin(∠ARB)

· · =1
sin(∠AP B) sin(∠RAB) sin(∠P RB)
Since ∠P AB = ∠BP A, ∠P RB = ∠BRA, it follows that sin(∠BP R) = sin(∠BAR). We

cannot have ∠BP R + ∠BAR = 180◦ , hence ∠BP R = ∠BAR ⇒ ∠P AR = ∠AP R and
P R = AR. Similarly RQ = AR, so RP = RQ.
G2. Let P Q and AD intersect at T . By Menelaus Theorem in 4ABD applied to points T, K, Q:

AK BQ DT BQ DT
· · = −1 ⇒ · = −1 (2)
KB QD T A QD T A
By Menelaus Theorem in 4ACD applied to collinear points T, P, M :
AP CM DT AP DT
· · = −1 ⇒ · = −1 (3)
PC MD TA PC TA
BQ AP
(We use directed lengths). From equations (3), (4) it follows that QD = PC . Similarly it
AR
follows that RC = DS
SB .
5 of 8
D
M
C
Q
P
N
R S L
A K B
BQ AP
Since QD = P C , AP · P C = BQ · QD, we have |AP | = |BQ|, |P C| = |QD|, and |AC| = |BD|.
BQ AP
Since |AC| = |BD|, QD = PC we have |AR| = |DS|, |RC| = |SB|. Then AR · RC = BS · SD,
as required.
Source: Mathlinks.
G3. Lemma: Points A, C, B, D lie on a line in this order. P is a point not on on this line. Then
any two of the following conditions imply the third:
CA DB
1. · = −1.
CB DA
2. P B is the angle bisector of ∠CP D.
3. AP ⊥ P B.
Proof: Assume 2 and 3 hold. Then P A, P B are the external and internal angle bisectors of
∠CP D, hence:
|CA| |CP | |CB| CA DB
= = ⇒ · = −1
|DA| |DP | |DB| CB DA
Assume 1 and 2 hold. Let A0 be a point on line CD such that A0 C < A0 D and A0 P ⊥ P B.
0
Then CA DB 0
CB · DA0 = −1. Since 1 holds, it follows that A ≡ A and AP ⊥ P B.
Assume 1 and 3 hold. Let C 0 be on line segment AB such that P B is the angle bisector of
C 0 A DB
∠C 0 P D. Since 3 holds, it follows that C 0 B · DA = −1. Since 1 holds, it follows that C ≡ C
0
and P B is the angle bisector of ∠CP D. The lemma is proved.

Now back to the problem. It suffices to show that JP is the angle bisector of ∠AJC and
∠BJD. Let Q = AP ∩ EF, Q0 = BD ∩ EF . Using the lemma, it suffices to prove
AQ AP BP BQ0
= , = 0
QC PC PD QD
By Menelaus’ Theorem on ∆ACF via the three collinear points B, P, D, we have

AP CB F D
· · = 1.
P C BF DA
By Ceva’s Theorem on ∆ACF via the three concurrent lines AB, CD, F Q, we have
AQ CB F D
· · = 1.
QC BF DA
6 of 8
Q’
F
J
Q
D
C
P
A B E
AP AQ BP BQ0
Hence, PC = QC . Similarly, PD = Q0 D . This completes the problem.
Source: Mathlinks.
Comment: Consider four points A, C, B, D, occurring on a line in this order. These points
CA DB
are called harmonic iff (A, B; C, D) = · = −1. Let P be a point not collinear with
CB DA
A, B, C, D; we define the pencil P (A, B, C, D) to be made up of 4 lines P A, P B, P C, P D.
There are a few useful results involving harmonic points that make them a powerful geometry
tool.
Fact: A pencil P (A, B, C, D) is given. The lines P A, P B, P C, P D intersect a line l at A0 , B 0 ,
C 0 , D0 respectively. Then (A0 , B 0 ; C 0 , D0 ) = (A, B; C, D).
Fact: In 4ABC, points D, E, F are on sides BC, CA, AB. Let F E intersect BC at G. Then
(B, C; D, G) is harmonic iff AD, BE, CF are concurrent.
Fact: The lemma used in the solution. Condition 1 is equivalent to (A, B; C, D) = −1.
2.4 Number Theory

N1. Factor the right side: 3y 2 = x(x+1)(x2 −x+1). Then every two of the terms x, x+1, x2 −x+1
are either coprime, or share a factor of 3. Hence each of the terms is a perfect square or 3
times a perfect square. Since x, x + 1 cannot both be perfect squares, it follows that exactly
one of these two terms is 3 times a perfect square, and the other term is not divisible by 3.
Hence x2 − x + 1 must be a perfect square. But (x − 1)2 < x2 − x + 1 ≤ x2 , with equality
only when x = 1. However, x = 1 does not yield a positive integer solution for y.
Source: South Korea 2004.
√ √
N2. Let a = b nc, b = b n − a2 c + 1. We claim that m = a2 + b2 works.
√
a2 ≤ n < (a + 1)2 ⇒ n − a2 ≤ 2a ≤ 2 n
√
p q
2 2 2
(b − 1) ≤ n − a ⇒ b − 1 ≤ n − a ≤ 2 n
√ √
(b − 1)2 ≤ n − a2 < b2 ⇒ m − n = b2 − (n − a2 ) ≤ 2b − 1 ≤ 2 2 4 n + 1
√ √ √
It is clear that 2 2 4 n + 1 < 3 4 n for n > 10000, and we are done.
7 of 8
N3. We first prove that f is injective. Assume f (a) = f (b) for a 6= b, then for every positive
integer n, we have f (a) + f (n)|(a + n)k , f (b) + f (n)|(b + n)k . Since f (a) + f (n) = f (b) + f (n),
it follows that f (a) + f (n) is a common divisor of (a + n)k and (b + n)k . Since gcd(a + n, b +
n) = gcd(a + n, b − a), we can take n such that a + n is a prime greater than b − a, then
gcd(a + n, b + n) = 1, and f (a) + f (n) cannot be a common divisor of (a + n)k and (b + n)k .
Fix a positive integer m. For every n, f (n) + f (m)|(n + m)k , f (n) + f (m + 1)|(n + m + 1)k .
Since gcd(m + n, m + n + 1) = 1, it follows that
gcd(f (n) + f (m), f (n) + f (m + 1)) = gcd(f (n) + f (m), f (m + 1) − f (m)) = 1 (4)
Assume there is a prime p dividing f (m + 1) − f (m). Let a be a positive integer such that
pa > m. Let n = pa − m, then f (n) + f (m)|(n + m)k = pak , hence p|f (n) + f (m). But then
p| gcd(f (n) + f (m), f (m + 1) − f (m)), contradicting (4). Hence f (m + 1) − f (m) = ±1. Since
f is injective, f (m + 1) − f (m) is always 1 or always -1 for all m. Since f only takes positive
values, f (m + 1) − f (m) = 1 for all positive integers m.
Therefore for some non-negative integer c, f (n) = n + c, for all n. If c > 0, let p be a prime
greater than 2c. Then f (1) + f (p − 1)|pk , hence p|f (1) + f (p − 1) = 2c + p, which is impossible
since 0 < 2c < p. Hence c = 0 and the only solution is f (n) = n for all positive integers n.
Source: Iran 2008.
8 of 8
Winter Camp 2011 Mock Olympiad Solutions
1 Problems
1. Sequences (an ), (bn ), (cn ), and (dn ) satisfy the following conditions:
an+1 = an + bn , bn+1 = bn + cn ,
cn+1 = cn + dn , dn+1 = dn + an ,
for n = 1, 2, . . .. If the sequences are periodic1 , prove that a2 = b2 = c2 = d2 = 0.
2. Let m, n be positive integers. Show that the number of positive integers x such that
n j k
X x
=x−m
2i
i=1
is at most 2n .
3. In a cyclic quadrilateral ABCD with AB = AD, points M, N lie on the sides BC and CD
respectively so that BM + N D = M N . Lines AM and AN meet the circumcircle of ABCD
again at points P and Q respectively. Prove that the orthocenter of the triangle AP Q lies on
the segment M N .
4. Let n ≥ 3 be an odd integer. Amy has coloured the squares in an n × n grid white and black.
We will call a sequence of squares S1 , S2 , . . . , Sm a “path” if all these squares are the same
colour, if Si and Si+1 share an edge for all i ∈ {1, 2, . . . , m − 1}, and if no other squares in
the sequence share an edge. Prove that if both the white squares and black squares form a
single path, then one of these paths must begin or end at the center of the grid.
For example, the grid on the left is a valid colouring, but the other two are invalid. In the
middle grid, the white squares loop back and touch themselves, and in the right grid, the
black squares are disconnected.
1
A sequence (tn ) is said to be “periodic” if there exists a positive integer p for which tn+p = tp for all n.
1 of 5
2 Solutions
1. If a sequences (tn ) is periodic, note that |tn | must be upper-bounded by some constant2 .
Specifically, if the sequence has period p, then an ∈ {a1 , a2 , . . . , ap } for all n, and hence
|an | ≤ max(a1 , a2 , . . . , ap ).
In our case, this means the sequence (an + bn + cn + dn ) is bounded (since (an ), (bn ), (cn ), and
(dn ) are all bounded). Substituting the given recurrence in for n ≥ 1, we have an+1 + bn+1 +
cn+1 + dn+1 = 2(an + bn + cn + dn ). To prevent |an + bn + cn + dn | from getting arbitrarily
large, an + bn + cn + dn must be 0 for all n.
If n ≥ 2, we then have:
an+2 = an+1 + bn+1
= an + 2bn + cn
= an−1 + 3bn−1 + 3cn−1 + dn−1
= 2bn−1 + 2cn−1
= 2bn .
Similarly bn+2 = 2cn , cn+2 = dn , and dn+2 = 2an . Therefore, an+8 = 2bn+6 = 4cn+4 =
8dn+2 = 16an , and so |an | will become arbitrarily large unless a2 = 0. Thus, a2 = 0. The
same reasoning also implies b2 = c2 = d2 = 0.
Source: Yugoslavia 1992.
2. Let f (x) = x − ni=1 2xi .
P
Any integer x can be written uniquely in the form a · 2n + b for non-negative integers a, b with
b < 2n , Using this notation, we have:
n
a · 2n + b
X
n n
f (x) = f (a · 2 + b) = a · 2 + b −
2i
i=1
n n
n
X
n−i
X b
= a·2 +b− a·2 −
2i
i=1 i=1
n
X
= a · 2n + f (b) − a · 2n−i
i=1
= a + f (b).
Now fix a positive integer m. For any b, there is at most one a such that f (x) = a + f (b) = m.
Since there are only 2n legal values of b, there are at most 2n values of x for which f (x) = m.
Comment: You can think of 2xi as stripping off the last i binary digits from x. Given this,

it is natural to consider a binary representation of x. The main insight is to group the last n
digits, and handle everything else separately.
Source: Adapted from Iran 2008.
3. Let H be the point on N M such that N H = N D and M H = M B.
2
This is not true for most aperiodic sequences. For example, if tn = n for all n, there is no such thing as maxn |tn |.
For any constant C you choose, I can find an n such that |tn | > C.
2 of 5
M’
D
A
N
Q
H B
M
C P
Lemma: H is the reflection of D across line AQ.

−−→
Proof of Lemma: Let M 0 be the point on ray N D such that DM 0 = BM . Then ∠ADM 0 =
∠ABM since ABCD is cyclic, and we are given that AB = AD. Therefore, 4ADM 0 ∼ =
4ABM by side-angle-side.
It follows that AM 0 = AM . Furthermore, M 0 N = DM 0 + N D = BM + N D = M N , and
so 4M 0 AN ∼ = 4M AN by side-side-side. In particular, ∠AN D = ∠AN M 0 = ∠AN M =
∠AN H. The lemma now follows from the fact that N H = N D.
1
By the same argument, H is the reflection of B across AP , and so ∠P AQ = 2 · ∠BAD.
Finally, the Lemma implies ∠AQH = ∠AQD = · ∠DCB = · 1
2
1
2 (180◦
− ∠DAB) = 90◦ −
∠QAP , and so P H is perpendicular to AQ. Similarly, QH is perpendicular to AP . Therefore,
H is the orthocenter of 4AP Q, as required.
Solution #2: Define H as in the previous solution. Let BH and DH intersect the cir-
cumcircle of ABCD again at points Q0 and P 0 . Also let α = ∠HDN = ∠DHN and let
β = ∠M BH = ∠HBM .
Then ∠N HQ0 = β and ∠Q0 DC = ∠Q0 BC = ∠HBM = β. Therefore, ∠HDQ0 = ∠DHQ0 =
α + β, and so Q0 D = Q0 H. Since N D = N H as well, Q0 N is perpendicular to DH, and also
Q0 N bisects ∠DQ0 H = ∠DQ0 B.
This means that Q0 N must meet the circumcircle at the midpoint of arc BD, but this midpoint
is exactly A since AB = AD. Therefore, A, N, Q0 are collinear and hence Q0 = Q. Similarly
P 0 = P . We already showed line P HD is perpendicular to line AN Q, and similarly line QHB
is perpendicular to line AM P , which means H is the orthocenter of 4AP Q.
Comment: When you see one length is the sum of two others, it is often useful to divide
into two lengths like we did here. A good straightedge and compass diagram should also tell
you exactly where H is.
Source: International Zhautykov Olympiad 2010.
4. Let us begin by making some definitions:
• Call the four corner vertices of the grid corner points.
• Call any point on the boundary of the grid where we switch between white and black
squares a transition point.
• Each path must have two endpoint squares where it begins and ends. Call the centers
of these four squares origin points.
3 of 5
Lemma: Let A be an origin point. Then there exist two line segments `1 and `2 connecting
A to either a boundary point or a corner point, and satisfying the following conditions: (1)
`1 and `2 form a 45◦ angle with the gridlines, (2) `1 and `2 are perpendicular to each other,
and (3) `1 and `2 do not pass through any other origin points.
Proof of Lemma: First note that no 2 × 2 square can form a checkerboard pattern with two
opposite squares being white, and the other two being black. If this happened, then the two
white squares would have to be connected by a single path, but this path would disconnect
the two black squares, which is not allowed.
Now consider the endpoint A of a single path. Assume without loss of generality that A is
white, and let B be the one white square adjacent to A:
X
@
@
4@ 3 3 4
3 @2 1 2 3
1 @A 1
B
In the picture above, the squares labeled 1 must all be black since A is assumed to be an
endpoint. The squares labeled 2 must also be black to avoid a 2 × 2 checkerboard. But now
the squares labeled 2 already have two black neighbours, so the squares labeled 3 must be
white. And then the squares labeled 4 must be white to avoid a 2 × 2 checkerboard. This
same argument can be extended until we reach squares on the boundary, say X and Y .
Since X and Y already have two neighbours of the same colour, they must either be at a
corner of the grid, or the other square they are adjacent to must be of the opposite colour.
Thus, the lines through the center of A and the centers of X, Y must end at either a corner
point or a transition point. Note that this is true even if A itself is on the boundary.
It is clear that these two lines are perpendicular and form a 45◦ angle with the gridlines.
Finally, they cannot pass through another origin point because each square between A and
X, Y must have two neighbours of the same colour.
Next, let us consider the boundary of the grid. Suppose it has two disjoint white components
W1 and W2 . Then there must also be two disjoint black components B1 and B2 . W1 and W2
must be part of the same path, but this path would have to disconnect B1 and B2 , which is
not allowed. Therefore, the boundary contains only one white component and only one black
component, and hence there are exactly two transition points.
Consider the following eight 45◦ diagonals: one emerging onto the grid from each corner point
(call these corner diagonals), and two emerging onto the grid from each transition point (call
these transition diagonals). By the Lemma, each origin point must be at the intersection of
two perpendicular diagonals. Since there are only eight diagonals altogether, each diagonal
can be used by only one origin point, and there are four origin points, each diagonal must be
used. If two transition diagonals are used by the same origin point, then another origin point
must use two corner diagonals to compensate, and hence be in the center of the grid. The
problem is solved in that case.
4 of 5
Otherwise, every transition diagonal must intersect a corner diagonal at an origin point. Let
A, B, C, D be the four corners of the grid, and let P be a transition point. Without loss of
generality, we can assume that P lies on side AB. If each grid square has side length 1, then
AB has odd length, so either AP or P B is even. Assume further without loss of generality
that AP is even. But then the diagonal through P parallel to BD never intersects BD, and it
intersects AC only on the corner of a square. Therefore, it cannot possibly intersect a corner
diagonal at an origin point, and we have the desired contradiction.
Source: Adapted from Google Code Jam Finals 2010.
5 of 5
Winter Camp 2014 Pre-Camp Problem Set Solutions
Pre-Camp Problem Set Solutions
1 Algebra
1. Find all real numbers x, y and z which satisfy the simultaneous equations x2 − 4y + 7 = 0,
y 2 − 6z + 14 = 0 and z 2 − 2x − 7 = 0.
2. Let a1 , a2 , . . . , an and b1 , b2 , . . . , bn be positive integers such that a1 < a2 < . . . < an , b1 >
b2 > . . . > bn , and (a1 , a2 , . . . , an , b1 , b2 , . . . , bn ) is a permutation of (1, 2, . . . , 2n). Prove that
|a1 − b1 | + |a2 − b2 | + . . . + |an − bn |
3. Prove that if a, b, c, d are non-negative real numbers, no two of which are zero, then
ab ac ad bc bd cd
+ 2 + 2 + 2 + 2 + 2 ≥ 3.
c2 +d 2 b +d 2 b +c 2 a +d 2 a +c 2 a + b2
4. Prove there exists no function f : R → R such that f (f (x)) = x2 − 2 for all x.
2 Combinatorics
1. 100 queens are placed on a 100 × 100 chessboard so that no two attack each other. Prove
that each of the four 50 × 50 corners of the board contains at least one queen.
2. Consider a planar region of area 1 which is the union of circular disks. Prove that from these
disks we can select some that are mutually disjoint and have total area of at least 91 .
3. On each of 12 points around a circle we place a disk with one white side and one black side.
We may perform the following move: select a black disk, and reverse its two neighbors. Find
all initial configurations from which some sequence of such moves leads to a position where
all disks but one are white.
4. Fix a positive integer n. Let S1 , S2 , . . . , Sk be subsets of {1, 2, . . . , n} such that no subset

completely contains another. Prove that
k
X 1
n
≤ 1.
i=1 |Si |
1 of 9
3 Geometry
1. Let A, B, C and D be points on a line in that order, and let P be a point not on the line such
that AB = BP and P C = CD. If the circumcircles of ACP and BDP intersect at P and Q,
prove that Q is equidistant from A and D.
2. A trapezoid ABCD with AB||CD and AD < CD in inscribed in a circle c. Let DP be a

chord parallel to AC. The tangent to c at D meets the line AB at E, and the lines P B and
DC meet at Q. Prove that EQ = AC.
3. Five points are given on a circle. A perpendicular is drawn through the centroid of the
triangle formed by any three of them to the chord connecting the remaining two. Such a
perpendicular is drawn for each triplet of points. Prove that the ten lines obtained in this
way have a common point.
4. An acute-angled triangle ABC is inscribed in a circle ω. A point P is chosen inside the

triangle. Line AP intersects ω at the point A1 . Line BP intersects ω at the point B1 . A line
` is drawn through P and intersects BC and AC at the points A2 and B2 . Prove that the
circumcircles of triangles A1 A2 C and B1 B2 C intersect again on line `.
4 Number Theory
1. Natural numbers a and b are such that a+1 b +
b+1
a is an integer. If d is the greatest common
2
divisor of a and b, prove that d ≤ a + b.
2. For a positive integer n ≥ 2, consider the n − 1 fractions

2 3 n
, ,...,
1 2 n−1
The product of these fractions equals n, but if you reciprocate (i.e. turn upside down) some
of the fractions, the product will change. Can you make the product equal 1? Find all values
of n for which this is possible and prove that you found them all.
3. Let S0 be a finite set of positive integers. We define sets S1 , S2 , . . . of positive integers as

follows: Integer a is in Sn+1 if and only if exactly one of a − 1 or a is in Sn . Show that there
exists infinitely many integers N for which SN = S0 ∪ {N + a : a ∈ S0 }.
4. Prove that infinitely many primes divide at least one number of the form 1! + 2! + . . . + n!.
2 of 9
5 Algebra Solutions
1. If we add the three given equations, we get
x2 − 4y + 7 + y 2 − 6z + 14 + z 2 − 2x − 7 = 0
=⇒ (x − 1)2 + (y − 2)2 + (z − 3)2 = 0
Therefore, we can only have x = 1, y = 2, z = 3. Checking, we see these values do indeed
satisfy the given equations. (England 2013)
Comment: You have to check your answer for questions like this (even if you do not show
much work). We first showed that if (x, y, z) satisfy the three given equations, then they must
be (1, 2, 3). It could still be that there are no (x, y, z) which satisfy the given equations. For
example, think about what happens if we replace the first equation with x2 − 4y + 6 = 0 and
the second equation with y 2 − 6z + 15 = 0.
2. Suppose ai > bi for some i. Then ai > a1 , a2 , . . . , ai−1 , bi , bi+1 , . . . , bn . Therefore, ai is greater
than n different terms from {1, 2, . . . , 2n}, and so must be at least n + 1. Similarly if bi > ai ,
then bi > a1 , a2 , . . . , ai , bi+1 , bi+2 , . . . , bn , and so must also at least n + 1.
Therefore, {max(a1 , b1 ), max(a2 , b2 ), . . . , max(an , bn )} is a subset of {n + 1, n + 2, . . . , 2n}.
Since both sets are of size n, it follows that
{max(a1 , b1 ), max(a2 , b2 ), . . . , max(an , bn )} = {n + 1, n + 2, . . . , 2n}.
This leaves {min(a1 , b1 ), min(a2 , b2 ), . . . , min(ab , bn )} on the one side and {1, 2, . . . , n} on the
other side, which implies
{min(a1 , b1 ), min(a2 , b2 ), . . . , min(an , bn )} = {1, 2, . . . , n}.
Finally,
n
X n
X
|ai − bi | = max(ai , bi ) − min(ai , bi )
i=1 i=1
n
X n
X
= max(ai , bi ) − min(ai , bi )
i=1 i=1
= (n + 1 + n + 2 + . . . + 2n) − (1 + 2 + . . . + n)
= n2 .
(Mathlinks)
3. Let u = a2 b2 + c2 d2 , v = a2 c2 + b2 d2 , w = a2 d2 + b2 c2 . Combining opposite terms and applying
the AM-GM inequality1 , we have
ab cd a3 b + ab3 + c3 d + cd3
+ =
c2 + d2 a2 + b2 a2 c2 + b2 c2 + a2 d2 + b2 d2
2a2 b2 + 2c2 d2
≥
a2 c2 + b2 c2 + a2 d2 + b2 d2
2u
= .
v+w
1 x1 +x2 +...+xn √
For non-negative real numbers x1 , x2 , . . . , xn , the AM-GM inequality states that n
≥ n x1 x2 . . . xn .
3 of 9
Applying similar reasoning to the other pairs of terms, we see that:

ab ac ad bc bd cd 2u 2v 2w
+ 2 + 2 + 2 + 2 + 2 ≥ + + .
c2 +d 2 b +d 2 b +c 2 a +d 2 a +c 2 a +b 2 v+w w+u u+v
u v w
Finally, Nesbitt’s inequality states that v+w + w+u + u+v ≥ 32 , and so we are done. (If
you have not seen Nesbitt’s inequality before, you can prove it by multiplying both sides by
(u + v)(v + w)(w + u) and then applying AM-GM again.)
For equality to hold in the first step, we need a3 b = ab3 , or equivalently one of the following:
a = b, a = 0, b = 0. The same holds for every pair of variables, so it must be that all variables
are equal, or one of the variables is 0 and the other three are equal. It is easy to check that
equality does indeed hold in both cases.
4. Let f (n) (x) denote the result of applying f to x exactly n times. Also, let o(x) denote the
smallest n such that f (n) (x) = x(possibly infinity), and let On denote the set of x such that
o(x) = n.
Lemma: If On is non-empty, then it has at least n elements.
Proof: For arbitrary x, let Sx denote the set {x, f (x), f (2) (x), . . .}. Pick y ∈ Sx , so that
y = f (i) (x) for some i. If o(x) = m for some finite m, then f (m−i) (y) = f (m) (x) = x, so x ∈ Sy .
Furthermore, f (m) (y) = f (m+i) (x) = f (i) (f (m) (x)) = f (i) (x) = y, so o(y) ≤ o(x). Since
x ∈ Sy , it follows similarly that o(y) ≤ o(x), and hence o(y) in fact equals m. Therefore, Sx
is a subset of Om . Additionally, if f (i) (x) = f (j) (x) for 0 ≤ i < j < n, then o(f (i) (x)) ≤ j − i,
which we just showed is impossible. It follows that |Sx | = m. Taking x ∈ On completes the
proof of the lemma.
Returning to the problem at hand, any f satisfying the given condition would have f (2) (x) −
2 (4) 4 2
x= x − x√−2= (x − √ + 1). Additionally, f √(x) − x√= x − 4x − x + 2 = (x − 2)(x +
2)(x
1) x + 1+2 5 x + 1−2 5 . Therefore, O4 = { −1−2 5 , −1+2 5 }, which contradicts the lemma.
(Vietnam 1990)
Comment: The lemma can be strengthened slightly: If On is finite, then |On | is divisible by
n. This is a powerful observation in other contexts as well, especially group theory.
6 Combinatorics Solutions
1. Suppose the top-left quadrant has no queens. There are 100 queens that must fit in 100 rows,
so each row must contain a queen. Since the top-left quadrant is empty, this means there
must be at least 50 queens in the top-right quadrant. Similarly, by looking at columns, we
see there must be at least 50 queens in the bottom-left quadrant.
However, any square in the bottom-left or top-right quadrant lies on one of 99 different
diagonals going from bottom-left to top-right. Therefore, there must be two queens on the
same diagonal, which is impossible. (Tournament of the Towns 2008)
2. We prove by induction on n that if n disks cover an area of size k, we can choose a subset of
disjoint disks that cover an area of size at least k9 .
If n = 1, the claim is trivial. Now let’s prove the result for n = m, assuming the result for
n < m. Let d be the disk with maximal radius r, let S denote the set of disks intersecting d,
4 of 9
and let T denote the remaining set of disks. Consider a disk s ∈ S. By assumption, it has
radius at most r, and so the centers of c and s differ by distance at most 2r. It follows that S
is contained entirely inside the disk centered at c with radius 3r. This means that the total
area covered by all disks in T is at least k − 9πr2 .
By the inductive hypothesis, there exists a subset of disjoint disks in T that cover an area of
size at least k9 − πr2 . By assumptions, these disks are all also disjoint from c, so we can add
in c to get a total area of at least k9 , as required. (Putnam 1998)
3. Every move, we change the colour of two disks. Therefore, the parity of the number of black
disks never changes, so if there are an even number of black disks to start with, the task is
impossible.
Otherwise, we claim the task is possible. We first describe how to transform the circle to have
11 black disks. Since the total number of black disks is odd, there must exist some contiguous
sequence of black disks of odd length: D1 , D2 , . . . , D2n+1 (n ≤ 4). Apply the move to black
disks D1 , D3 , . . . , D2n+1 . When everything is done, D1 , D2 , . . . , D2n+1 will all still be black,
and so will the disks on either side. Thus, we can keep increasing the number of black disks
until all but one disk is black.
This leaves one contiguous sequence of black disks and it is of odd length: D1 , D2 , . . . , D2n+1 .
Apply the move to black disks D2 , D4 , . . . , D2n . When everything is done, D2 , D3 , . . . , D2n
will all still be black, but D1 and D2n+1 will be white. Thus, we can keep decreasing the
number of black disks until only one disk is black, as required. (Japan 1998)
4. Pick a permutation π of {1, 2, . . . , n} uniformly at random. Let us say π matches Si if the first
|Si | elements of π are the elements of Si in some order. Let Pi denote the probability that π
matches Si , and let P denote the probability that π matches at least one of {S1 , S2 , . . . , Sk }.
Note that π cannot simultaneously match both Si and Sj for distinct i, j, because this would
imply one of the two sets contained the others. Therefore, P = P1 +P2 +. . .+Pk . Additionally,
Pi = n1 since the first |Si | elements of π are equally likely to be any of the |Si |-element
(|Si |)
subsets of {1, 2, . . . , 2n}.
Therefore, ki=1 n1 = ki=1 Pi = P ≤ 1.
P P
(|Si |)
7 Geometry Solutions
1. Let O be the circumcircle of 4AP D. Then ∠P OA = 2∠P DA = ∠P CA, so O lies on the
circumcircle of 4ACP . Similarly, it lies on the circumcircle of 4BDP so O = Q and the
problem is solved.
Comment: A slicker solution is to extend P Q to hit the circumcircle of 4AP D at R, and
angle chase from there. However, the solution presented here is better motivated.
2. Since AB and CD are parallel, we have ∠DCA = ∠BAC, which implies DA = BC. Addi-
tionally, we have ∠EDA = ∠DCA = ∠P DC = ∠P BC = ∠QBC and ∠DAE = ∠BCD =
∠BCQ. Therefore 4EDA is congruent to 4QBC by angle-side-angle. It follows that
EA = QC, and hence EACQ is a parallelogram. This gives EQ = AC, as required. (Nordic
2002)
5 of 9
3. Let the 5 points on the circle be A, B, C, D, E. Set up vectors with the center of the circle
being at 0. We claim the lines meet at X = A+B+C+D+E
3 .
Let CABC denote the centroid of 4ABC. We know CABC lies at vector A+B+C 3 , so the line
through CABC and X travels along the vector D+E 3 . This is parallel to the line between
the center of the circle and the midpoint of segment DE, and that line is orthogonal to DE
because the center of a circle lies on the perpendicular bisector of any chord. It follows that
CABC X is also orthogonal to segment DE.
The same argument can be applied for any triple of points, so the lines concur at X, as
required.
Comment: The main challenge with a problem like this is to figure out where X has to
be in the first place. One approach is to start with just two lines. Let M be the midpoint
of AB, and let CABC and CABE denote the centroids of triangles ABC and ABE. Let `C
denote the line through CABC perpendicular to DE and let È denote the line through CABE
perpendicular to CD. If we perform a homothety about M with factor 3, then `C becomes
the altitude from C to DE, and È becomes the altitude from E to CD. Therefore, the point
X we are looking for is mapped to the orthocenter of 4CDE by a homothety with factor
3 about M . Now it is natural to introduce vectors since, if the vectors are centered at the
circumcenter of 4CDE, then the orthocenter of 4CDE is at precisely C + D + E. (Do you
see why this is true? It is a consequence of the Euler line.)
4. First suppose ` is parallel to AB. Then ∠A2 P A1 = ∠BAA1 = ∠BCA1 = ∠A2 CA1 , so
6 of 9
A2 P CA1 is cyclic. Similarly, B2 P CB1 is cyclic, so the two circles meet at P , which is on `
as required.
Otherwise, let X denote the intersection of AB and `. We will assume the configuration
shown in the diagram; the remaining configurations are similar2 .
Let Q be the point on ` so that (using directed lengths) P Q · P X = P B · P B1 = P A · P A1 .
Then XQBB1 and XQAA1 are cyclic by Power of a Point. Therefore, ∠B2 QB1 = ∠XQB1 =
∠XBB1 = 180◦ − ∠ABB1 = 180◦ − ∠ACB1 = 180◦ − ∠B2 CB1 . It follows that Q lies on
the circumcircle of 4B1 B2 C. A similar argument shows it also lies on the circumcircle of
4A1 A2 C, which solves the problem since Q is on `. (Mathlinks)
8 Number Theory Solutions

2 2
1. Expanding, we have a+1 b+1
b + a =
a +b +a+b
ab . We know that d2 |ab, so if the given quantity is
an integer, then d2 |a2 + b2 + a + b. Since d2 also divides a2 and b2 , it follows that d2 |a + b.
Therefore, a + b is a positive integer that is a multiple of d2 , and hence a + b ≥ d2 . (Russia
1994)
a1 a2 ak
2. Suppose we reciprocate the fractions a1 −1 , a2 −1 , . . . , ak −1 . Then the product becomes:
a1 − 1 2 a2 − 1 2 ak − 1 2

2 3 n
· · ... · · · · ... ·
1 2 n−1 a1 a2 ak
2
(a1 − 1)(a2 − 1) . . . (ak − 1)
= n· .
a1 a2 . . . ak
2
a1 a2 ...ak
If this equals 1, then n = (a1 −1)(a2 −1)...(ak −1) , and in particular n is a perfect square.
2
Saying other configurations are “similar” on a contest is not usually a good idea. You should either explicitly
go through every possible configuration or use something like directed angles. You can find more on directed angles
here: https://sites.google.com/site/imocanada/2011-winter-camp.
7 of 9
Conversely, suppose n = m2 . If we reciprocate 12 , 32 , . . . , m−1

m
, then we get:
m2 m2 − 1

m+1 m−1 m−2 1
2
· 2
· ... · · · · ... ·
m −1 m −2 m m m−1 2
1
= m·
m
= 1.
Therefore, the product can be made equal to 1 if and only if n is a perfect square. (Bay Area
2013)
3. For each pair of integers n, i, let sn,i denote the number that is 1 if i ∈ Sn and 0 otherwise.
For n ≥ 0, we are given that sn+1,i ≡ sn,i−1 + sn,i mod 2.
We are now going to work with polynomials mod 2. Specifically, given two polynomials Q and
R with integer coefficients, we will write Q ≡ R mod 2 if every coefficient of Q is congruent
to the corresponding coefficient of R mod 2. You can add, subtract, and multiply polynomials
mod 2, and they work just like integers3 .
Now, for each n, define a polynomial Pn (x) = i sn,i xi . The given conditions can be rewritten
P
to state that Pn+1 (x) ≡ Pn (x) · (x + 1) mod 2, and hence Pn (x) ≡ P0 (x) · (x + 1)n mod 2.
k k
Lemma: (x + 1)2 ≡ x2 + 1 mod 2 for all non-negative integers k.
k
Proof: Using the binomial theorem, it suffices to show that 2i is even for
all i satisfying
0 < i < 2k . Recall that the largest power of 2 dividing n! is exactly n2 + n4 +. . .. Therefore,
k
the largest power of 2 dividing 2i is precisely
k k
2k − i

X 2 i
− j −
2j 2 2j
j=0
k−1 k
2k − i

X 2 i
= 1+ − j −
2j 2 2j
j=0
k−1 k
i + (2k − i)

X 2
≥ 1+ −
2j 2j
j=0
= 1.
2k

This shows i is indeed even, which completes the proof of the lemma.
Finally, let R denote the difference between the largest and smallest elements of S0 , and choose
n to be a power of 2 larger than R. Then Pn (x) ≡ xn P (x) + P (x) mod 2 by our lemma. It
follows that Sn (x) consists of the integers that are in exactly one of S0 and {n + a : a ∈ S0 }.
Since n > R, these two sets are disjoint, and hence Sn (x) = S0 ∪ {n + a : a ∈ S0 }. (Putnam
2000)
3
What does this really mean? The key property you want is this: if Q ≡ Q0 mod 2 and R ≡ R0 mod 2, then
Q + R ≡ Q0 + R0 mod 2, Q − R ≡ Q0 − R0 mod 2, and Q · R ≡ Q0 · R0 mod 2. Abstract algebra is very interested
in these kinds of questions.
8 of 9
4. Suppose, by way of contradiction, that there are only k such primes p1 , p2 , . . . , pk .

Let Sn = 1!+2!+. . .+n!. For each n, we can write Sn = ki=1 pei i . Let f (n) denote the integer
Q
i such that pei i is maximum. For any N , the pigeonhole principle guarantees there must exist
n, m ∈ {N − k, N − k + 1, . . . , N } such that n < m and f (n) = f (m). Let p = pf (n) = pf (m)
and let
√ e be the largest integer such that pe | gcd(Sn , Sm ). The definition of p guarantees that
k
pe ≥ n!.

On the other hand, gcd(Sn , Sm )|m! − n!, which equals n! · (n + 1)(n + 2) . . . m − 1 . The
largest exponent of p dividing n! is

n n
+ 2 + ...
p p
n n
≤ + + ...
p p2
≤ n.

Therefore, pe ≤ pn · (n + 1)(n + 2) . . . m − 1 < pn · mk . Combining this with the previous
√
observation gives us pn · mk > n n!, but we claim this is impossible for large n.
Indeed, for large n, we have
r n
√
r
2k n n2 n
n 4k n
> pn , and
2k
n! > = = 4k
2 2 2
s
n
√
r n
2k 2k n 2 2k n+k 4 n
n! > > ≥ m 8k ≥ mk .
2 2
Multiplying these together gives a contradiction.
9 of 9
Winter Camp 2014 Buffet Contest Solutions
Buffet Contest Solutions
1 Algebra
1. Find all functions f from the set of non-zero real numbers to itself satisfying
f (x) + f (y) = f (xyf (x + y))
for all x, y 6= 0 with x + y 6= 0.
2. Consider the sequence (an )n≥1 of real numbers where a1 ∈ (1, 2) and an+1 = an + ann for all
n ≥ 1. Prove that there is at most one pair of terms in this sequence with integer sum.
3. For a positive integer n ≥ 5, let ai , bi (i = 1, 2, . . . , n) be integers satisfying the following two

conditions:
• The pairs (ai , bi ) are distinct for i = 1, 2, . . . , n;

• |a1 b2 − a2 b1 | = |a2 b3 − a3 b2 | = . . . = |an b1 − a1 bn | = 1.
Prove that there exist indices i, j such that 1 < |i − j| < n − 1 and |ai bj − aj b1 | = 1.
2 Combinatorics
1. One or more powers of 2 are written on each of n sheets of paper. The sum of the numbers
on each sheet are the same. If each number appears at most 5 times among the n sheets,
what is the largest possible value for n?
2. The numbers 1 and -1 are written in the cells of a 2000 × 2000 grid. If the sum of the entries
in the grid is positive, show that one can select 1000 rows and 1000 columns such that the
sum of the numbers written in cells of their intersections is at least 1000.
3. In a country, every pair of cities is either joined by a two-way road or not joined by any road.
The country has the property that for any pair of cities there is a sequence of roads one can
take to get from the first city to the second city. However, if any sequence of an odd number
of distinct roads beginning and ending at the same city is closed down, the country no longer
has this property. Prove that the cities in the country can be partitioned into 4 districts such
that there is no road between two cities in the same district.
1 of 8
3 Geometry
1. ABC is an isosceles triangle with AC = BC. Let O be its circumcenter, let I be its incenter,
and let D be the point on BC such that lines OD and BI are perpendicular. Prove that lines
ID and AC are parallel.
2. Consider the parallelogram ABCD with obtuse angle A. Let H be the foot of the perpen-
dicular from A to the side BC. The median from C in triangle ABC meets the circumcircle
of triangle ABC at the point K. Prove that points K, H, C, D lie on the same circle.
3. Let M be the midpoint of the internal bisector AD of 4ABC. Circle ω1 with diameter AC
intersects segment BM at E and circle ω2 with diameter AB intersects segment CM at F .
Show that B, E, F, C lie on a circle.
4 Number Theory
1. A perfect number greater than 28 is divisible by 7. Prove that it is also divisible by 49.
(Note: A number n is “perfect” if the sum of the divisors of n is equal to 2n. For example,
28 is perfect since 1 + 2 + 4 + 7 + 14 + 28 = 2 · 28.)
2. Find (with proof) all monic polynomials P (x) with integer coefficients that satisfy the fol-
lowing two conditions.
(a) P (0) = 2014.

(b) If x is irrational, then P (x) is also irrational.
Note: A polynomial is monic if its highest degree term has coefficient 1.
3. Let an be the leftmost digit of 2n and let bn be the leftmost digit of 5n . Prove that the reverse
of any k consecutive terms in the sequence an appears as k consecutive terms in the sequence
bn .
2 of 8
5 Algebra Solutions
1. Suppose f (z) 6= z1 for some z 6= 0. Let x = f (z)
1
and y = z − x. Clearly x 6= 0; we also know
1
y 6= 0 since we assumed f (z) 6= z , and we know x + y = z 6= 0. Therefore,
f (x) + f (y) = f (xyf (x + y))

1 1
=⇒ f + f (z − x) = f · (z − x) · f (z)
f (z) f (z)

1
=⇒ f = 0,
f (z)
which is a contradiction. It follows that f (x) = x1 is the only possible solution. Checking, we
see it does indeed satisfy the conditions of the problem. (Nordic 2003)
2. Suppose an > n for some n. Then
n (an − n)(an − 1)
an+1 = an + = + n + 1 > n + 1.
an an
It is given that an > n for n = 1, so it follows from induction that an > n for all n. Therefore,
n
an+1 = an + < an + 1.
an
Define bn = an − n. We have just shown that the sequence (bn ) is always positive and
decreasing. Furthermore, ai + aj is an integer if and only if bi + bj is an integer, so it suffices
to prove there exists at most pair (i, j) such that bi + bj is an integer.
a2 +1
Now, a2 = 1a1 = (a1 −2)(a a1
1 −0.5)
+ 2.5 < 2.5, so b2 < 0.5. Therefore, if i, j > 1, we have
bi + bj > 0 and bi + bj ≤ 2b2 < 1, so bi + bj cannot possibly be an integer.
It remains to consider b1 + bi . In this case, we know b1 + bi > 0 and b1 + bi ≤ 2b1 < 2, so this
is an integer if and only if bi = 1 − b1 . However, (bn ) is decreasing, so there is at most one i
with this property. (Russia 2009)
3. Throughout this write-up, indices will always be taken mod n.

Define Pi to be the 2-dimensional vector (ai , bi ). Let |Pi | denote the length of Pi and let Ki,j
denote the area of the triangle with sides Pi and Pj . Recall that Ki,j = 12 · |ai bj − aj bi |. Thus,
we are given Ki,i+1 = 12 for all i and we need to show Ki,j = 21 for some i, j with |j − i| = 6 1.
First suppose Pi and Pj lie on the same line for some i, j. Assume without loss of generality
that |Pi | ≤ |Pj |. Then, using the base times height formula for the area of a triangle, we have
|Pi |
Ki,j+1 = Kj,j+1 · |P j|
. This quantity is positive so it is at least 12 . (Recall that the formula for
Ki,j is an integer multiple of 12 ). Furthermore, it is at most Kj,j+1 = 21 , so in fact Ki,j+1 = 12 ,
and similarly Ki,j−1 = 12 . Since n ≥ 5, we cannot have j adjacent to both i + 1 and i − 1.
Therefore, one of Ki,j−1 or Ki,j+1 solves the problem.
Otherwise, let i be such that |Pi | is maximal. Since Pi−1 and Pi are not collinear, we can
express Pi+1 as a linear combination of the vectors Pi−1 and Pi : Pi+1 = uPi−1 + vPi for some
3 of 8
real numbers u, v. Using the base times height formula for the area of a triangle again, we
have 12 = Ki,i+1 = |u| · Ki−1,i = |u|
2 , which implies u = ±1.
Now, gcd(ai , bi ) = 1 or else |ai bi+1 −ai+1 bi | could not possibly be 1. Since vPi = −Pi+1 ±Pi−1
has integer coordinates, it then follows that v must also be an integer. If v = 0, then Pi−1
and Pi+1 are collinear, which we already covered. If |v| ≥ 2, then |Pi+1 | = |uPi−1 + vPi | ≥
|v| · |Pi | − |u| · |Pi−1 | ≥ |Pi |. Equality holds only if Pi−1 , Pi are collinear, which is impossible.
Thus, |Pi+1 | > |Pi | in this case, and that contradicts the choice of i. The only remaining
possibility is v = ±1, but then Ki−1,i+1 = |v| · Ki−1,i = 21 and the problem is solved. (Korea
2001)
6 Combinatorics Solutions
1. Suppose that we have n sheets with the same sum S. There exists k such that 2k ≤ S < 2k+1 .
The sum of all numbers must then be at least n · 2k . On the other hand, we cannot have any
power of 2 larger than 2k or the sum on that sheet would be too large. Since each number
appears at most 5 times, the sum of all numbers must be at most 5 · (1 + 2 + . . . + 2k ) < 10 · 2k .
Therefore, n · 2k < 10 · 2k and hence n < 10.
Conversely, if n = 9, we can use the following arrangement:
(a) 2 sheets containing 1, 1, 2, 4.
(b) 1 sheet containing 2, 2, 4.
(c) 1 sheet containing 4, 4.
(d) 5 sheets containing 8.
Therefore, the maximum value of n is 9. (St. Petersburg 1998)
2. Lemma: The numbers 1 and -1 are written in the cells of a grid with n rows and m columns.
Let S denote the sum of the entries in the grid. If m is odd, it is possible to delete one row
and end up with sum at least min(S + 1, n − 1). If m is even, it is possible to delete one row
and end up with sum at least min(S, n − 1).
Proof: If every row has positive sum, then we can drop an arbitrary row, and there will
remain n − 1 rows, each with sum at least 1, and the claim is proven.
Otherwise, there exists some row with non-positive sum. If m is odd, then this row actually
has sum at most -1, and therefore if we drop it, the remaining sum will be at least S + 1. If
m is even, then we still know this row has sum at most 0, and therefore if we drop it, the
remaining sum will be at least S. This covers all cases, and the lemma is proven.
Note that the same argument can be applied to removing columns as well. Now we use the
lemma repeatedly to solve the problem:
• We start with a 2000 × 2000 grid with S ≥ 1.

• By the lemma, we can delete 1 column to get a 2000 × 1999 subgrid with S ≥ 1.
• By the lemma, we can delete 1000 rows to get a 1000 × 1999 subgrid with S ≥ 1000.
• By the lemma, we can delete 999 columns to get a 1000 × 1000 subgrid with S ≥ 1000,
completing the proof.
4 of 8
(Russia 1995)
Comment: Can you solve the problem if 2000 × 2000 is replaced with 1500 × 1500? The
same approach works.
3. Let G be the graph with vertices corresponding to cities and with edges corresponding to
roads between cities. In graph theoretic terms, we are given that G is connected (there is
sequence of edges joining any two vertices) and that removing any odd cycle will disconnect
G. We need to show it is 4-colourable (the vertices can be assigned colours 0, 1, 2, 3 such
that no two vertices with the same colour have an edge between them).
Suppose we remove edges from G one by one without disconnecting it, stopping only when
it is impossible to remove any more edges. Let G1 be the subgraph of G containing only the
edges we removed, and let G2 be the subgraph of G containing only the edges we did not
remove. Since we can remove all of the edges in G1 without disconnecting the graph, the
problem conditions guarantee that G1 has no odd cycles. Furthermore, G2 has no cycles at
all. If it did have a cycle, we could remove one of the edges in the cycle without disconnecting
the graph, which contradicts the definition of G2 .
We will now use the following very useful theorem from graph theory: if a graph has no odd
cycles, then it is 2-colourable.1
It follows that G1 and G2 are both 2-colourable. In particular, there are functions f1 , f2
mapping cities to {0, 1} such that f1 (u) = f1 (v) only if there is no edge between u and v in G1 ,
and f2 (u) = f2 (v) only if there is no edge between u and v in G2 . Let f (v) = 2 · f1 (v) + f2 (v).
This assigns every vertex a colour in {0, 1, 2, 3}, and if f (u) = f (v), then f1 (u) = f1 (v) and
f2 (u) = f2 (v), which means u and v are not connected by an edge in either G1 or G2 . It
follows that f is a valid 4-colouring of G, and the problem is solved. (Russia 2010)
7 Geometry Solutions
1. Since ABC is isosceles, I and O both lie on the perpendicular bisector of AB. Therefore,
1
You can prove this as follows. First assume the graph is connected. If it is not, we can apply the proof
independently to each connected component. Pick an arbitrary vertex u and colour it 0. Now, consider another
vertex v. If there is both an even-length path from u to v and an odd length path from u to v, we could combine
them to get an odd-length cycle. Therefore, we have either that all paths from u to v are even length, or all paths
are odd length. In the former case, we colour v with 0; in the latter case, we colour it with 1. Do you see why this is
a valid 2-colouring?
5 of 8
∠DOI = 90◦ − ∠BIO = ∠ABI = ∠DBI, from which it follows that DBOI is a cyclic
quadrilateral.
This means ∠BDI = 180◦ − ∠BOI = 180◦ − ∠BOC = 2∠BCI = ∠BCA, and hence ID and
AC are parallel. (Russia 1996)
2. Let M be the midpoint of AB and let E be the point such that AHBE is a rectangle. Note
that, since AHBE is a rectangle, H, M, E are collinear.
Now, ∠DEH = ∠AEH = ∠ABH = ∠ABC = 180◦ − ∠DCB = 180◦ − ∠DCH, so DCHE is
cyclic. Consider the circumcircles of DCHE, AHBE, and ABC. The radical axis of DCHE
and AHBE is HE, and the radical axis of ABC and AHBE is AB, so the radical center of
the three circles is M . In particular, M is on the radical axis of DCHE and ABC.
Since C is also on the radical axis, we know the radical axis is equal to line CM , and so K is
also on the radical axis. Since K is on circle ABC, it follows that K is also on circle DCHE,
and the problem is solved. (Russia 2012)
3. Let C 0 be the point on line AB such that C 0 C is parallel to AD, and let P be the midpoint
of C 0 C. The homothety about B that takes AD to C 0 C also takes M to P . This implies that
P lies on line BEM .
6 of 8
Now, ∠CC 0 A = ∠DAB = ∠DAC = ∠C 0 CA, so AC = AC 0 . Since P is the midpoint of C 0 C,

we conclude that ∠AP C = 90◦ and hence P lies on ω1 . Since C 0 C||AD, this also implies
∠P AD = 90◦
Using the above facts, we have ∠BEC = 180◦ − ∠P EC = 180◦ − ∠P AC = 90◦ + ∠DAC =
90◦ + ∠BAC
2 . The same argument applies to ∠BF C, so we have shown ∠BEC = ∠BF C, and
the problem is solved. (Ukraine 2013)
8 Number Theory Solutions

1. Let n be
Q aeiperfect number divisible by 7 but not 49. Then we can prime factorize n as
k
2 · 7 · pi for some (possibly empty) set of primes pi and some non-negative e +1 integer
k.
p i −1
Recall2 that the sum of the divisors of n is then given by (2k+1 − 1) · 8 ·
Q i
pi −1 .
Since n is perfect, we have

!
Y pei i +1 − 1 Y
(2k+1
− 1) · 8 · = 2 · 2k · 7 · pei i
pi − 1
8 2k+1
The right-hand side must be a multiple of 8, so k ≥ 2. Then 7 ≥ 2k+1 −1
, which implies
e +1
pi i −1
(2k+1 − 1) · 8 ≥ 2 · 2k · 7. Furthermore pi −1 > pei i for all i. This means that if k ≥ 2,
!
Y pei i +1 − 1 Y e
(2k+1 − 1) · 8 · ≥ 2 · 2k · 7 · pi i
pi − 1
with equality only if k = 2 and there are no primes pi . However, this occurs only when
n = 28, so the problem is solved. (Russia 2000)
2. Clearly P (x) = x + 2014 satisfies the two conditions.

We now show that is the only solution. Suppose by way of contradiction that we have a
polynomial P (x) = xd + ad−1 xd−1 + . . . + a1 x + 2014 that satisfies the two conditions and
has degree d ≥ 2. Let M = max(|a1 |, |a2 |, . . . , |ad−1 |), and let p be a prime such that
p > max(P (1), M d).
Since P (1) p + 2014 for large x, there must exist some x > 1 such that
P (x) = p + 2014. We know this x must be rational, or else we would have found x such that
x is irrational and P (x) is rational. Thus, we have shown there is a rational number x that
satisfies the following:
xd + ad−1 xd−1 + . . . + a1 x − p = 0.
By the rational root theorem, it must be that x ∈ {−p, −1, 1, p}. Since we know x > 1, it
follows that x = p, and hence
pd + ad−1 pd−1 + . . . + a1 p − p = 0.
e +1

pi i −1
2
pei i , then the sum of the divisors of n is
Q Q
The general formula is that if n = pi −1
. To prove this, note
e +1
pi i −1
that pi −1
= 1 + pi + p2i + . . . + pei i , and see what monomials appear when you multiply everything out.
7 of 8
However,
pd + ad−1 pd−1 + . . . + a1 p − p ≥ pd − M pd−1 − M pd−2 − . . . − M p − p

≥ pd − M dpd−1 > 0,
which is a contradiction. (Bay Area 2004)
3. Lemma: For every real number > 0 and every integer A, there exist integers n and m with
n > A and 10m ≤ 5n < 10m · (1 + ).
1
Proof: Let B be a positive integer such that (1 + )B > 10. Then 10 B < 1 + .
Consider the numbers {0 · log10 5}, {1 · log10 5}, . . . , {B · log10 5}, where {x} denotes the frac-
tional part of x. There are B + 1 such numbers and all lie in the range [0, 1]. Furthermore,
they are all distinct: if {u · log10 5} = {v · log10 5}, then 5u−v is a power of 10, but that is
impossible for u 6= v. Therefore, there must exist non-negative integers u, v ≤ B such that
0 < {u · log10 5} − {v · log10 5} ≤ B1 .
Let x = {u · log10 5} − {v · log10 5}, and let C be a positive integer larger than A + Bx . Then
there must exist some non-negative integer k such that kx ≤ {C · log10 5} < (k + 1)x. Let
n = C + k(v − u). Note that n > A since |k(v − u)| ≤ x1 · B < C − A. Furthermore,
{n · log10 5} = {C · log10 5} + k · {v · log10 5} − k · {u · log10 5}

1
=⇒ {n · log10 5} < x ≤ .
B
Letting m = bn log10 5c, we have:
1
m ≤ n · log10 5 < m +
B
1
=⇒ 10m ≤ 5n < 10m · 10 B
=⇒ 10m ≤ 5n < 10m · (1 + ),
which completes the proof of the lemma.

Now consider a sequence au , au+1 , . . . , av . For each i ∈ {u, u + 1, . . . , v}, there exist integers
ei such that (ai + 1) · 10ei − 1 ≥ 2i ≥ ai · 10ei .
1
Let = 1 + 2v and A = v. Choose n and m according to the lemma.
Then, for i ∈ {u, u + 1, . . . , v}, we have
5n−i ≥ 10m · 5−i = 10m−i · 2i ≥ ai · 10m−i+ei ,
and similarly
5n−i < 10m−i · 2i · (1 + ) ≤ 10m−i · (2i + 1) ≤ (ai + 1) · 10m−i+ei .
Therefore, bn−i = ai , and the proof is complete.
8 of 8
Mock Olympiad Solutions
1 Problems
1. Let a1 be a natural number not divisible by 5. The sequence a1 , a2 , a3 , . . . is defined by
an+1 = an + bn , where bn is the last digit of an . Prove that the sequence contains infinitely
many powers of 2.
2. Let ABC be a triangle and let D and E be points on the sides AB and AC respectively
such that DE is parallel to BC. Let P be any point interior to triangle ADE and let F and
G be the intersections of DE with the lines BP and CP respectively. Let Q be the second
intersection point of the circumcircles of triangles P DG and P F E. Prove that the points
A, P, and Q lie on a straight line.
3. Find all functions f : R → R such that for all x ∈ R and y ∈ f (R)
f (x − y) = f (x) + xy + f (y).
Note: y is said to be in f (R) if there exists z ∈ R such that f (z) = y.
4. There are a number of markets in a city, joined by one-way streets. Every market has exactly
two one-way streets going out of it, and there is at most one street joining any two markets.
Prove that the city may be partitioned into 2014 districts such that:
(a) No street joins two markets in the same district.

(b) For every pair of districts, the streets joining the two districts all go in the same direction
(from the first district to the second, or from the second district to the first).
1 of 4
2 Solutions
1. Checking each possible last digit for a1 , we can see there must always exist some k such that
ak has last digit equal to 2.
Focusing on that k, we have ak = 10m + 2 for some non-negative integer m, and hence
• ak+1 = 10m + 4
• ak+2 = 10m + 8
• ak+3 = 10(m + 1) + 6
• ak+4 = 10(m + 2) + 2.
At this point, the process repeats.
If m is even, then (an ) includes every sufficiently large number that is 4 mod 20. In particular,
it includes 4 · 16t = 24t+2 for all sufficiently large t.
If m is odd, then (an ) includes every sufficiently large number that is 12 mod 20. In particular,
it includes 32 · 16t = 24t+5 for all sufficiently large t.
Therefore, (an ) includes infinitely many powers of 2 in either case. (Russia 1994)
2. Let P 0 be the image of P under the dilation about A that takes BC to DE. Then A lies on
line P P 0 .
Now,
∠DQE = ∠DQP + ∠EQP
= ∠DGP + ∠EF P since DQGP and EQF P are cyclic
= ∠DEP 0 + ∠EDP 0 since DP 0 ||F P and EP 0 ||GP by the definition of P 0
= 180◦ − ∠DP 0 E.
Therefore, DP 0 EQ is cyclic. It follows that ∠DQP 0 = ∠DEP 0 = ∠DGP = ∠DQP , and so
Q lies on line P P 0 .
We have now shown A and Q both lie on line P P 0 , which implies A, P, Q must be collinear.
(India 1995)
2 of 4
3. It is easy to check that f (x) = 0 for all x is a solution. We focus on the other case, in which
there exists some a ∈ f (R) with a 6= 0.
Setting x = y, we have f (0) = 2f (y) + y 2 for all y ∈ f (R). (*) Let z be an arbitrary real
number, and set x = z−fa(a) , y = a. Then we have

z − f (a) z − f (a)
f −a =f + z.
a a
Therefore, there exist real numbers u, v such that f (u)−f (v) = z. Setting x = f (u), y = f (v),
we have
f (z) = f (f (u) − f (v)) = f (f (u)) + f (u)f (v) + f (f (v))

f (u)2 f (v)2
= − + f (u)f (v) − + f (0) by (*)
2 2
1
= − · (f (u) − f (v))2 + f (0)
2
z2
= − + f (0).
2
2
We have thus shown there exists a constant C such that f (z) = z2 + C for all z ∈ R. Setting
2 2 2
x = y = C, we have C = − C2 + C + C 2 − C2 + C, so C = 0 and f (z) = − z2 for all z.
Checking, we see this does indeed satisfy the given equation. (Korea 2002 and IMO 1999)
4. Let G be the graph with vertices corresponding to markets and directed edges corresponding
to one-way streets. We are given that every vertex has out-degree equal to 2.
Lemma: Let us say two vertices are “separated” if there is no path from one to the other
of length 1 or 2. Then we can assign every vertex a “colour” in {1, 2, . . . , 13} such that if
vertices u, v have the same colour (possibly with u = v), then they are separated.
Proof: We prove this by induction on the number of vertices, relaxing the conditions to say
each vertex has out-degree at most two.
If there is only one vertex, the claim is trivial. Now suppose it has been shown for n − 1
vertices, and consider a graph with n vertices. From every vertex, there are 2 paths going
out of length 1 and 4 paths going out of length 2. Therefore, the total number of unordered
vertex pairs (u, v) that are not separated is at most 6n. By the pigeonhole principle, there
exists a vertex v that belongs to at most 12 of these pairs.
Using the inductive hypothesis, all vertices except v can be coloured in the required manner.
Then there are up to 12 vertices that v cannot be the same colour as, so we can choose a
valid colour for v as well, completing the inductive proof.
Using the colouring described above, associate to each vertex a triple (a, b, c) where a is the
colour of the vertex, and b ≤ c are the colours of the vertices that the vertex has an edge
leading out to. We create one district for each possible value of this triple, and use that to
assign vertices into districts. There are 13 possible values for a. Once a is fixed, (b, c) can be
any pair chosen from the remaining colours, so there are 12 + 12 2 = 78 possibilities for them,
giving a total of 13 · 78 < 2014 districts.
3 of 4
If two vertices are in the same district, then they have the same colour and the colouring
guarantees there is no edge between them.
Next suppose there is an edge from vertex u in district (a1 , b1 , c1 ) to vertex v in district
(a2 , b2 , c2 ). If b2 or c2 equals a1 , then we can follow a second edge from v to another vertex
w with colour a1 . However, u and w then have the same colour but are not separated, which
is a contradiction. Thus, b2 and c2 are both distinct from a1 , and so there is no edge from
district (a2 , b2 , c2 ) to district (a1 , b1 , c1 ).
Therefore, our district division satisfies the required conditions, and the problem is solved.
(Russia 2002)
4 of 4

A Corner of Mathematical Olympiad and Competition Book II: Phnom Penh, October 24, 2014 Prepare By: Keo Sodara

Uploaded by

Copyright:

Available Formats

A Corner of Mathematical Olympiad and Competition Book II: Phnom Penh, October 24, 2014 Prepare By: Keo Sodara

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Corner of Mathematical Olympiad and Competition Book II: Phnom Penh, October 24, 2014 Prepare By: Keo Sodara

Uploaded by

Copyright:

Available Formats

A Corner of Mathematical

Olympiad and Competition

Problems with Solutions

Phnom Penh, October 24, 2014

The mathematical olympiad in Estonia is held annually in three stages:

Below, after each problem number, the numbers in brackets indicate

Problem 6 originates from Iceland ’92; problem 4 is due to an oral

2. (10, II) The point of osculation of the hypotenuse of a right triangle

3. (10, III) Natural numbers m , n and k have the properties: mn

4. (10-11, III) For which natural numbers n is it possible to cover

10. (12, III) Let us call a natural number “beautiful” if it is a perfect

Hints and Solutions

1. There are no such triples. It is easy to see that abc 6= 0 , and

3. By the assumptions, (mn )k = (mk )n divides (nk )m = (nm )k which di-

4. Assign numbers to the unit squares as shown on the figure:

The sum of numbers covered by shape (B) is always equal to zero;

7. Condition b) implies f (n + 1) = f (n)f (1) for any natural number n ,

k = 1 or k n0 = 2n0 ; in the latter case we have k = 2 and n0 = 1 or

8. For k = 2 , take n1 = 6 , n2 = 102 . Now, use the identity

taking the greatest of the denominators as m , at each step.

10. Denote x = 1993 and y = 2 · 1993 , then the number

is “beautiful” (it is obtained by writing side by side the perfect

The Estonian Mathematical Olympiad is held annually in three stages:

Since 1993, we also have two “open” mathematical competitions each

Here we present a variety of problems from both of these competitions.

1. (II, 9) The sides of a convex quadrangle serve as diameters for four

4. (II, 12) For a point C on the diameter AB of a semicircle, let P

5. (II, 11–12) The “Top Twenty” of popular songs is selected weekly.

12. (III, 11) For any sequence A = { a1 , a2 , . . . } of real numbers denote

13. (III, 12) For a triangle ABC , let A1 , B1 ja C1 be points on its

16. (OPEN) The two sides BC and CD of an inscribed quadrangle

17. (OPEN) Consider the sequence of real numbers a0 , a1 , a2 , . . . :

Determine all values of x and y for which this sequence is bounded

18. (OPEN) A number of balls of radius R are

19. (OPEN) Consider a function f (x) such that

for any real value of x .

1. Find all positive integers n such that 6n is divisible by 6 + n .

Final (selectional) round

(a) f (x) = −f (−x) ;

1. Exactly one of the following statements is known to be true.

Problems from the final round

1. Find all pairs of integers (a, b) such that a2 + b = b1999 .

Answer: (0; −1), (0; 0) and (0; 1).

is twice of absolute value of the other root.

SAP IK − SCP I = SAKI + SAP I − SCP I = SALI + SCLI − SALI = SCLI .

Analoguously we find that

SBM IK − SCQI = SCM I .

1. Find all pairs of integers (m, n) such that

2. Find the value of the expression

 1   2  1999 2000 2000 2000

4. For which values of n it is possible to cover the side wall of staircase of

Answer: n = 3k or n = 3k + 2, except n = 3 and n = 5.

Consider the cases n = 2, n = 3, n = 5. Clearly, if n = 2 then the plating is possible.

n=2 n=3 n=5

In the cases n = 6 and n = 9 the wall can be covered as shown in figure 4.

5. On the squares a1, a2, . . . , a8 of a chessboard there are respectively 20 , 21 , . . . , 27 grains

The remainder of The remainder of

2. Find the value of the integral

Solution. Let f (x) = ln(x + 1 + x2 ). We always have 1 + x2 > x2 , so 1 + x2 > |x|,

1 2 1999 2000 2000 2000

a2 + 4b2+ 8c2 − 3ab − 4bc− 2ca =