00

elem ents of number theory

translated from the fifth revised edition by saul kravetz
 ELEMENTS OF
NUMBER THEORY
 ELEMENTS OF
NUMBER THEORY

BY I. M. VINOGRADOV

T ra n s la te d from the Fifth R evised Edition by

Saul Kravetz

DOVER PUBLICATIONS, INC.

 C o pyright 1954 by

D over P u b l i c a t i o n s , Inc.

F i r s t E n g li s h t r a n s l a t i o n of the
F if th R u s s i a n ed itio n o f 1949.

M an u factu red in th e U n ited S t a t e s of Am

 CONTENTS

P reface vii

C hapter I

DIVISIBILITY THEORY
§ 1. B asic C oncepts and Theorem s ( 1). § 2 . The G reatest
Common D ivisor (2). § 3 . The L e a s t Common Multiple (7).
§ 4 . The Relation of E u c lid ’s Algorithm to Continued F ractions
(8). § 5 . Prime Numbers (14). § 6 . T he U nicity of Prime De­
composition (15). Problem s for C hapter I (17). Numerical
E x e r c is e s for Chapter I (20).

C hapter II

IMPORTANT NUMBER-THEORETICAL FUNCTIONS

§1. The F u n ctio n s { x}»x (21). § 2. Su ms Extended over the
Divisors of a Number (.22). §3. The Mobius Function (24).
§4. The E uler Function (26). Problems for C hapter II (28).
Numerical E x e r c is e s for Chapter II (40).

Chapter III .

CONGRUENCES
§ 1. B asic C oncepts (41). § 2. P ro p erties of Congruences
Similar to those of Equation® (42). §3. Further P ro p erties of
Congruences (44). §4. Complete Systems of R esid u e s (45).
§5. Reduced S ystem s of R esid u e s (47). § 6. The Theorem s of
ill
Euler and Ferm at (48). Problem s for Chapter III (49). Numeri­
cal E x e r c is e s for Chapter III (58).

C hapter IV

CONGRUENCES IN ONE UNKNOWN

§ 1. B a s ic C oncepts (59). § 2. C ongruences of the F irs t
Degree (60). §3. System s of Congruences of the F ir s t Degree
(63). §4. C ongruences of Arbitrary Degree with Prime Modu­
lus (65). § 5. Cong ruences of Arbitrary Degree with Com­
posite Modulus (66). Problem s for C hapter IV (71). Numerical
E x e r c is e s for C hapter IV (77).

C hapter V

CONGRUENCES OF SECOND DEGREE

§1. General T heorem s (79). § 2. The Legendre Symbol (81).
§3. The Jacobi Symbol (87). § 4 . The C a s e of Composite
Moduli (91). Problem s for C hapter V (95). Numerical Exer­
c i s e s for Chapter V (103).

C hapter VI

PRIMITIVE ROOTS AND INDICES

§ 1. General Theorem s (105). §2. Prim itive Roots Modulo
p a and 2pa (106). §3. E valuation of Prim itive Roots for the
Moduli p a and 2pa (108). § 4. Indices for the Moduli p a and
2pa (110). § 5 . C on seq u en ce s of the P receding Theory (113).
§ 6 . Indices Modulo 2a (116). §7. Indices for Arbitrary Com­
posite Modulus (119). Problem s for Chapter VI (121). Numeri­
cal E x e r c is e s for Chapter VI ( 130).

SOLUTIONS OF THE PROBLEMS

Solutions for Chapter I (133). Solutions for Chapter II (139).
Solutions for C hapter III (161). Solutions for Chapter IV (178).
Solutions for Chapter V (187). Solutions for Chapter VI (202).
IV
 ANSWERS TO THE NUMERICAL EXERCISES
Answers for Chapter I (217). A nsw ers for C hapter II (217).
A nsw ers for C hapter III (218). A nsw ers for C hapter IV (218).
A nsw ers for C hapter V (218). Answers for C hapter VI (219).

TABLES O F IN D IC E S ..................................................................... 220

TA BLES OF PRIMES <4000 AND THEIR LEAST PRIMITIVE

R O O T S ............................................................................................... 226
 P R IN T E R ’S NOTE

Throughout th is book the a which ta k es the following form

in the formula (a) assu m es the following form (a) whenever it
is used in the s m a lle s t type s iz e , as a su b sc rip t or a super­
script. Both forms are meant to be the same.
 PREFACE

A s e r ie s of R ussian m athem aticians— Chebyshev, Korkin,

Zolotaryov, Markov, Voronoi and others— have worked on the
theory of numbers. One can become acquainted with the con­
tent of the c l a s s ic a l work of th e se notable m athem aticians in
B. N. D elone’s book " T h e P ete rsb u rg School of the Theory of
N um bers" ( “ P eterburgskaya shkola teorii c h i s e l , ” in R u ssian ,
1947).
Soviet m athem aticians, working in the field of number theory,
have continued the great tradition of their p red ec esso rs and
have created powerful new methods which have been used to
obtain a s e rie s of f irs t-c la s s re s u lts; in the number theory s e c ­
tion of the book “ M athematics in the USSR after 30 y e a r s "
("M atem atika v SSSR za 30 l e t , " in R u ssian , 1948) one can
find a report on the attainm ents of Soviet mathem aticians in the
field of number theory, and the corresponding bibliographical
references.
In my book I present a sy stem atic exposition of the funda­
m entals of number theory within the scope of a university
course. A large collection of problems introduces the reader
to some of the new id e a s in number theory.
T h is fifth edition of my book differs considerably from the
fourth. A s e r ie s of changes, allowing a simpler exposition,
have been made in all the chapters of the book. The most
important changes are the merging of the old chapters IV and
V into one chapter IV (reducing the number of chapters to six)
and the new, simpler proof of the e x isten ce of primitive roots.
T he problems at the end of each chapter have been e s s e n tia lly
revised. The order of the problems i s now in complete cor­
respondence with the order of the presentation of the theoreti-
vii
cal m aterial. Some new problems have been added; but the
number of numbered problems h a s been s u b sta n tia lly reduced.
T h is was accom plished by the unification, under the le tte rs
a, b, c, . . . , of previously s e p a ra te problems which were re­
lated by the method of solution or by content. All the solutions
of the problems have been reviewed; in many c a s e s these solu­
tions have been sim plified or replaced by better o n es. P articu ­
larly e s s e n tia l changes have been made in the solutions of the
problems relating to the distribution of n-th power resid u es and
non-residues, and primitive roots, as well as in the estim ations
of the corresponding trigonometric sums.
I. M. Vinogradov
 ELEMENTS OF
NUMBER THEORY
 CHAP TER 1

DIVISIBILITY THEORY
§1. Bas ic Concepts and Theorems

a. The theory of numbers is concerned with the study of the

properties of integers. By integers we mean not only the num­
bers of the natural number sequence 1, 2, 3, . . . (the positive
integers) but also zero and the negative integers: —1, —2,
—3, . . . .
A s a rule, in presenting the theoretical material, we will
use le tte rs only to denote in teg ers. In the c a s e s in which
le tte rs may denote non-integers, if th is is not c lear in itse lf,
we will mention it sp ecifically .
The sum, difference and product of two integers a and b are
also integers, but the quotient resulting from the division of a
by b (if b is different from zero) may be an integer or a non­
integer.
b. In the c a s e in which the quotient resulting from the di­
vision of a by b is an integer, denoting it by q, we have a = bq,
i.e. a is equal to the product o f b by an integer. We will then
say that a is d iv isib le by b or that b divides a. Here a is said
to be a m ultiple of b and b is said to be a divisor of the number
a. The fact that b divides a is written a s: b \ a .
We have the following two theorems.
1. I f a is a multiple o f m, and m is a multiple o f b, then a is
a multiple o f b.
Indeed, it follows from a = a,m, m = m tb that a = a ^ m ^ ,
where a 1ml is an integer. But th is proves the theorem.
1
 2 . I f we know that in an equation o f the form, h + I + . . . + n
= p + q + . . . + s , all terms exc ep t one are m ultiples o f b, then
this one term is also a m ultiple o f b.
Indeed, let the exceptional term be k. We have

I ' Zj b, . . . , n —n^ b, p ■
—p ^ Z), ^ —q ^ b ^ • • • , 5 —s ^ b ^
h = p + q + . . . + s — I — . . . —n
= (pi + qt + • • • + Si - l x - . . . - n f)b t
proving our theorem.
c. In the general c a s e , which includes the particular c a s e in
which o is d iv isib le by b, we have the theorem:
E very integer a is uniquely representable in terms o f the
p o sitive integer b in the form

a = bq + r, 0 <r<b

Indeed, we obtain one such representation of a by taking bq

to be equal to the largest multiple of b which does not exceed
a. Assuming th a t we also have a = bqx + rlt 0 < r, < b, we find
that 0 = b(q - qf) + r — rlf from which it follows (2, b) that r - r1
is a multiple of b. But since |r —r, | < b, the la tte r is only
p o ssib le if r —rt = 0, i.e. if r = rlf from which it also follows
that q =
The number q is called the partial quotient and the number
r is called the remainder resulting from the division of a by b.
E x a m p le s . L et b = 14. We have

177 = 14 12 + 9, 0 < 9 < 14;

- 6 4 = 14 (-5) + 6 , 0 < 6 < 14;
1 5 4 = 14 11 + 0, 0 = 0 < 14.

§ 2. The G reatest Common D ivisor

a. In what follows we s h all consider only the po sitiv e di­
v iso rs of numbers. Every integer which divides all the integers
a, b, . . . , I is said to be a common d iviso r of them. The
la rg e st of th e s e common divisors is s a id to be their greatest
common divisor and is denoted by the symbol (o, b, . . . , I).
2
In view of the fin ite n e s s of the number of common divisors the
ex iste n c e of the g re a te st common divisor is evident. If (a,
b, . . . , / ) = 1, then a, b, . . . , I are said to be relatively prime.
If each of the numbers a, b, . . . , I is re la tiv e ly prime to any
other of them, then a, b, . . . , I are said to be pairwise prime.
It is evident that pairw ise prime numbers are also relatively
prime; in the case of two numbers the concepts of “ pairwise
prime” and “ relatively prime” coincide.
Exam ples. The numbers 6, 10, 15 are relatively prime since
(6 , 10, 15) = 1. T h e numbers 8 , 13, 21 are pairw ise prime
sin ce (8 , 13) = (8 , 21) = (13, 21) = 1.
b. We first consider the common divisors of two numbers.
1 . I f a is a multiple o f b, then the s e t o f common diidsors o f
the numbers a and b coincides with the s e t o f divisors o f b\ in
particular, (a, b) = b.
Indeed, every common divisor of the numbers a and b is a
divisor of b. C onversely, if a is a multiple of b, then (1, b, § 1)
every divisor of the number b is also a divisor of the number
a, i.e. it is a common divisor of the numbers a and b. Thus
the s e t of common divisors of the numbers a and b coincides
with the s e t of d iv iso rs of b, but since the g re a te st divisor of
the number b is b itse lf, we have (a, b) = b.
2. If
a = bq + c,

then the s e t o f common divisors o f the numbers a and b coin­

cides with the s e t o f common divisors o f the numbers b and c;
in particular, (a, b) = (b, c).
Indeed, the above equation shows that every common divisor
of the numbers a and b divides c (2 , b, § 1) and therefore is a
common divisor of the numbers b and c. Conversely, the same
equation shows that every common divisor of the numbers b
and c divides a and consequently is a common divisor of the
numbers a and b. Therefore the common divisors of the num­
bers a and b are ju s t those numbers which are also common
divisors of the numbers b and c; in particular, the g re a te st of
these divisors must also coincide, i.e. (a, b) = (b, c).
3
 c. In order to obtain the le a s t common divisor as well as to
deduce its most important p roperties, E u c lid ’s algorithm is ap­
plied. The la tte r c o n s is ts of the following p ro cess. L et a
and b be positive integers. By c, § 1, we find the sequence
of equations:

a = bq2 + r a, 0 < r2 < b,

b = r2q 3+ r3, 0 < r s < ra,
r2 = ryq4 + r4, 0 < r4 < r3,

r n —i — r n —i ? n + r n» 0 r n ^ r n —l

r n —i = r n 7 n + i»

which term inates when we obtain some r n+1 = 0 . T he la tter

must occur since the seq u en ce b, ra, r3, . . . as a d ecre asin g
sequence of integers cannot contain more than b positive
integers.
d. C onsidering the equations of ( 1), proceeding from the top
down, (b) shows that the common divisors of the numbers a
and b are identical with the common divisors of the numbers b
and ra, are moreover identical with the common divisors of the
numbers ra and r3, of the numbers r3 and r4, . . . , of the numbers
rn_ t and rn, and finally with the divisors of the number rn .
Along with th is, we have

(a, b) = (b, r2) = (ra, r3) = . . . = (rn_ lf rn) = rn.

We arrive at the following re s u lts.

1. The s e t o f common divisors o f the numbers a and b co­
incides with the s e t o f divisors o f their greatest common
divisor.
2 . This greatest common divisor is equal to rn, i.e. the la st
non-zero remainder in E u c lid ’s algorithm.
Exam ple. We apply E u c lid ’s algorithm to the evaluation of
(525, 231). We find (the auxiliary c a lcu la tio n s are given on
the left)

4
 525 525 = 231 • 2 + 63,
462 231 = 6 3 - 3 + 42,
231 63
63 = 4 2 - 1 + 21,
189 3
42 = 21 • 2.
63 42
42 1
42 21
42 2

Here the la s t positive remainder is r4 = 21. T h is means that

(525, 231) = 21.
e . l . I f m denotes any p o sitiv e integer, we have (am, bm)
= (a, b)m.
2 . I f S is any common divisor o f the numbers a and b, then
a b\ (a, b) / a b
s ’i ) = '• “ p a n i c “ l a' - ’ w b)
the quotients resulting from the division o f two numbers by
their greatest common d iviso r are rela tively prime numbers.
Indeed, multiply each of the terms of the equations (1) by
m. We obtain new equations, where a, b, ra, . . . , rn are re­
placed by am, bm, r^m, . . . » rnm. Therefore (am, bm) = rnm,
showing that proposition 1 is true.
Applying proposition 1, we find that

<0 > „ . ( ^ , A s ) = ( |,} ) s ;

and th is proves proposition 2 .

f . l . I f (a, b) = 1, then (ac, b) = (c, b).
Indeed, (ac, b) divides ac and be, which implies (1, d) that
it also divides (ac, be) which is equal to c by 1, e; but (ac, b)
also divides b and therefore also divides (c, b). Conversely,
(c, b) divides ac and b, and therefore also divides (ac, b).
T h u s (ac, b) and (c, b) divide each other and are therefore
equal to one another.
2. I f (a, b) = 1 and ac is d iv isib le by b, then c is d iv is ib le
by b.

5
 Indeed, since (a, A) = 1, we have (ac, b) = (c, b). But if ac
is a multiple of b , then (1, b) we have (ac, b) = A, which means
that (c, 6) = 6, i.e . c is a m ultiple of b.
3. I f each alf a 2, . . . , a m is rela tively prime to each bu A2,
. . . , An, tAera tAe product a i a 2. . . a m is rela tively prime to the
product bibj . . . bn.
Indeed (theorem 1), we have

bjf) — (fljflj • • • bjf)

= (ctj • • ‘ Omi Aj^) = . . . = (<zm, A^) = 1,

and moreover, s e ttin g a la J . . . a m = A , in the same way we find

(bib3b i . . . bn, A ) = (AjAj • • • An , /4)

= (As. . . An, / ! ) = . . . = (An, /I) = 1.

g. T he problem of finding the g re a te s t common divisor of

more than two numbers re d u c e s to the same problem for two
numbers. Indeed, in order to find the g re a te s t common divisor
of the numbers a 2, a 2, . . . , an we form the seq u en ce of numbers:

(a2, flj) = dj, ((fj, flj) = (^s, 04) ^ • • • » (<^n—1* ^n*

The number dn is also the g re a te s t common divisor of all the
given numbers.
Indeed ( l, d), the common divisors of the numbers a2 and a 2
coincide with the divisors of d2; therefore the common divisors
of the numbers Oi, a 2 and a} coincide with the common divisors
of the numbers dt and a 3, i.e. coincide with the divisors of d}.
Moreover, we can verify that the common divisors of the num­
bers a 2, a 2, a3, a, coincide with the divisors of rf4, and so forth,
and finally, that the common divisors of the numbers at, a 2,
. . . , a n coincide with the divisors of d n . But since the la rg e st
divisor of dn is d n itse lf, it is the g re a te s t Common divisor of
the numbers a 2, a 2, . . . » a n.
Considering the above proof, we can s e e that theorem 1, d
is true for more than two numbers also. Theorem s 1, e and
2, e are also true, b ec a u se m ultiplication by m or division by

6
8 of all the numbers a lt a3, . . . , an c a u s e s all the numbers
d i, d3, to be multiplied by m or to be divided by 8.

§3. The L e a s t Common Multiple

a. Any integer which is a multiple of each of a s e t of given
numbers, is said to be their common m ultiple. The sm allest
p o sitiv e common multiple is called the le a s t common multiple.
b. We first consider the le a s t common multiple of two num­
bers. L e t M be any common multiple of the integers a and b.
Since it is a multiple of a, M = a k , where k is an integer. But
M is also a multiple of b, and hence
ak
~b

must also be an integer which, settin g (a, b) = d, a = aid,

a k
b = bid, can be represented in the form — —, where (alt b,) = 1
bi
(2 , e, § 2). Therefore (2 , f, § 2) k must be divisible by bu
b
k = bxt = — t, where t is an integer. Hence
d
ab
M = — t.
d
C onversely, it is evident that every M of this form is a mul­
tiple of a as well as b, and therefore, th is form gives all the
common multiples of the numbers a and b.
The s m a lle s t positive one of th e se m ultiples, i.e. the le a s t
common multiple, is obtained for t = 1. It is

ab
m — ----- .
d '
Introducing m, we can rewrite the formula we have obtained
for M as:
M = mt.

7
 T he la s t and the next to the la s t equations lead to the
theorems:
1. The common m ultiples o f two numbers are identical with
the m ultiples o f their le a st common m ultiple.
2 . The le a s t common m ultiple o f two numbers is equal to
their product divided by their greatest common divisor.
c. Assume that we are now required to find the le a s t common
multiple of more than two numbers a2, a2, . . . , an . Letting the
symbol m(a, b) denote the le a s t common multiple of the num­
bers a and b, we form the sequence of numbers:

flj) = rn2, mfni2, flj) — •••, —mn .

The mn obtained in this way will be the le a s t common multiple
of all the given numbers.
Indeed ( l, b), the common m ultiples of the numbers a2 and
a2 coincide with the m ultiples of m2, and hence the common
multiples of the numbers a lt a2 and a 3 coincide with the common
multiples of m 2 and a3, i.e. they coincide with the m ultiples of
m3. It is then c lear that the common multiples of the numbers
a 2, a2, a3, a4 coincide with the m ultiples of m4, and so forth,
and finally, that the common m ultiples of the numbers a lt a2,
. . . , a n coincide with the m ultiples of mn, and s in c e the small­
e s t positive multiple of mn is mn itse lf, it is also the le a s t
common multiple of the numbers alt a2, . . . , a n.
Considering the proof given above, we s e e that theorem
1, b is also true for more than two numbers. Moreover, we
have shown the validity of the following theorem:
The le a s t common multiple o f pairwise prime numbers is
equal to their product.

§ 4 . The R elation o f E u c lid ’s Algorithm to Continued Fractions

a. L e t a be an arbitrary real number. L e t q be the largest
integer which does not exceed a .
For a non-integer a , we have

1
a = , a 2 > 1,

8
Similarly, for non-integers a 2, . . . , a s _t we have
1
= + a 3 > 1;

a . - i - Rs-i 7 Ctg > 1,

a.

from which we obtain the following developm ent o f a in a con­

tinued fraction:
(1) a = q 1 + ----------------------

<7j +
<73 +

+
<7s-i + a .

If a is irrational, then it is evident that there can be no

in teg ers in the sequence a , a 2, . . . , and the above pro cess
can be continued indefinitely.
If a is rational, then, as we shall s e e la te r (b), there will
eventually be an integer in the sequence a, a 2, . . • , and the
above p ro c e s s will be terminated.
b. If a is an irreducible rational fraction, then the develop­
ment of a in a continued fraction is clo sely connected with
E u c lid ’s algorithm. Indeed, we have
a = bq2+ r2; — = <7i + —
b b

b = r2<74+ r3; = <72 +

'2 ’4
— = <73 + — *
r, r.

9
 2 fn
r n —2 = r n —l ? n —1 + r n? = 7 n —l +
rn - 1 rn_t

^n—1
^"n—l ~ *"n^r ?ni

from which we find

a 1
T = 7i + -----------

7i +
7a +

1
+ ---- .
7n

c. The numbers q lt q2, . . . , which occur in the expansion of

the number a in a continued fraction, are called the partial
quotients (for the c a s e of rational a th e se are, by b, the par­
tial quotients of the s u c c e s s iv e d iv isio n s of the E uclidean
algorithm), and the fractions

1 1
§1 = 7l> ^2 = 71 + > ^3 = 7l + » •••
72 1
+ —
7a

are called the convergents.

d. The very simple rule for the formation of the convergents
is e a s ily obtained by noting that 6s (s > 1) is obtained from
1
by replacing q B_t in the expression for 8S_, by q s-1 + ------•
7«
Indeed, s e ttin g P 0 = 1, Q0 = 0, for the sak e of uniformity,
we can rep resen t the convergents recursively in the following

10
way (when the equation — is written here, it means

that A is denoted by the symbol P s , and B by the symbol Qa):

<7. _Pi ^ q' + Ti _ Wx + 1 q2Pt + Po P2

1 = <?. 1 " 9,'1 + ° “ + Qo = Qi

<hPi + p i =
5, =
1 ? 3& + Qx Q3
<72 + Qi + Qo

e tc ., and in general

<7s P « -i + P . - ! Ps
< 7 « (? « -1 + <?«

T h u s the numerators and the denominators of the convergents

can be recursively calcu la ted by means of the formulae
t -
Pa — (JaPa—l + Pa—2 »
( 2)
Qa = tfaQs—l + Qa—2 *

T h e s e c a lcu la tio n s can e a s ily be carried out by means of the

following schema:

9a <7i <72 <7« <7n

Pa 1 <7i Pi P .-2 P a—l Pn—i a

Qa 0 1 Qi Qa-l e. b

11
 105
Exam ple. Develop the nu m b e r------ in a continuous fraction.
38
Here

105 1
2 +
~38 1
1 +
1
3 +
1
4 + —
2

Therefore the aforementioned schem a gives:

<7s 2 1 3 4 2

Ps 1 2 3 11 47 105

Qs 0 1 1 4 17 38

e. We now consider the difference 8S — 8s_ l of s u c c e s s iv e

convergents. For s > 1, we find

Ps Pa—1
6S — 5S ! ------------ = -----------
Qs Qs- l Qs Qs- l

where hg = Ps Qa- i — QaPa-1 5 replacing Ps and Qs by their ex­

p ressio n s in (2) and making the evident sim plifications, we
find that = —hs_ 1 . The latter, in conjunction with
Ai = <7, .0 - 1 * 1 = - 1 , gives ^ = (-1)®. Thus

(3) Ps Q s - l ~ QsPa-L = (-I)* (S > 0),

12
 Exam ple. In the table of the example given in d, we have

105 • 17 - 38 ‘ 47 = (-1 )5 = - 1 .

f. It follows from (3) that (P a , Qa) divides (-1)® = i l

Pa
(2, b, $1). Hence (Pa, Q„) = 1, i.e . the convergent& ---- are
Qs
irreducible.
g. We now in v e stig ate the sign of the difference 8S — a for
6S which are not equal to a (i.e. we exclude the c a s e in which
6S is the la s t convergent for rational a ) . It is evident that 8S
is obtained by replacing a s by q s in the expression (1) for a.
But, a s is evident from a, as a re su lt of th is replacement

as i s decreased,
a .-i is increased,
^ 3—2 is d ecreased ,

is d ecreased for odd s,

is in c re ased for even s .

Therefore 8S — a < 0 for odd s and 8S — a > 0 for even s ,

and consequently, the sign of 8S — a coincides with the sign
of (-1)®.
h. We have
1
Ia “ bs _ , | <;
QbQ s - 1

Indeed, for 8S = a this a sse rtio n follows (with the equality

sign) from (4). F or 8a unequal to a , it follows (with the in­
equality sign) from (4) and from the fact that, 8S — a and
8 ,_ 1 — a have different sig n s, beca u se of g.
13
 §5. P r im e Numbers

a. T h e number 1 h as only one p o sitiv e divisor, namely 1.

In th is re s p e c t the number 1 s ta n d s alone in the sequence of
natural numbers.
Every integer, greater than 1, h a s no fewer than two divisors,
namely 1 and itse lf; if th e s e divisors exhaust all the positive
divisors of an integer, then it is s a id to be prime. An integer
> 1 which h as po sitiv e divisors other than 1 and itse lf, is said
to be co m p o site .
b. The s m a lle s t divisor, different from one, o f an integer
greater than one, is a prime number.
Indeed, let q be the s m a lle s t divisor, different from one, of
the integer a > 1. If were composite, then it would have
some divisor q l such th a t 1 < <7i < <7; but the number a, being
divisible by q, would also be divisible by q x (1, b, $l), and
th is contradicts our h y p o th e sis concerning the number q.
c. The s m a lle s t divisor, different from 1, o f a com posite
number a (by b, it w ill be prime) does not e x c e e d Va~.
Indeed, let q be th is divisor; then a = qat, at > q from
which, multiplying by q, we obtain a > r f , q < \Ta .
d. The number o f primes is infinite.
The validity of th is theorem follows from the fact that no
matter what different primes p w p a , . . . , pk are considered, we
can obtain a new prime which is not among them. Such a
prime is any prime divisor of the 6um p 2p2 . . . p* + 1 which,
dividing the whole sum, cannot be equal to any of the primes
Pi* Pi* • • • > Pk (2, b, $l).
e. T here is a simple method, called the s ie v e o f Eratos­
thenes, for the formation of a table of the primes not exceed­
ing a given N. It c o n s is ts of the following.
We write down the numbers

(1) 1, 2...........N.

The first number of th is seq u en ce greater than one is 2; it

is only divisible by 1 and itse lf, and hence it is a prime.

14
 We delete from the sequence ( 1) (since they are composite
numbers) all the numbers which are m ultiples of 2, except 2 it­
self. The first number following 2 which is not deleted is 3;
it is not divisible by 2 (otherwise it would have been deleted),
and hence 3 is divisible only by 1 and itself, and hence it is
also prime.
D elete from the seq u en ce ( 1) all the numbers which are
m ultiples of 3, except 3 itse lf. The first number following 3
which is not d eleted is 5; it is not divisible by either 2 or 3
(otherw ise it would have been deleted). Therefore 5 is divis­
ible only by 1 and itse lf, and therefore it is also prime.
And so forth.
When th is p ro c e s s h as deleted all the numbers which are
m ultiples of primes le s s than the prime p, then all the numbers
remaining which are l e s s than p2 are primes. Indeed, every
com posite number a which is le s s than p2 h a s already been
deleted s in c e it is a multiple of its s m a lle s t prime divisor
which is <C \fa < p. T h is implies:
1. In the p rocess o f d eletin g the m ultiples o f the prime p,
this s e t o f d e le te d numbers m ust start with p2.
2 . The formation o f the table o f primes ^ N is com pleted
once we have d e le te d all the com posite m ultiples o f primes
not e xc eed in g 'JW .

§ 6 . The Unicity o f Prime D ecomposition

a. E very integer a is either rela tiv e ly prime to a given prime

p, or is d iv is ib le by p.
Indeed, (a, p), being a divisor of p, is either 1 or p. In the
first c a s e , a is rela tiv e ly prime to p, and in the second, a is
divisible by p.
b. I f the product o f se v e r a l factors is div isib le by p, then
at le a st one o f the factors is d iv is ib le by p.
Indeed (a), every factor is either divisible by p or is rela­
tively prime to p. If all the factors were rela tiv e ly prime to p,
then their product (3, f, §2) would be relatively prime to p;
therefore at le a s t one factor is divisible by p.

15
 C. Every integer greater than one can be decom posed into
the product o f prime factors and uniquely, i f we disregard the
order o f the factors.
Indeed, let a be an integer greater than unity; if pt is its
s m a lle s t prime divisor, then a = p la l . If a, > 1, then if p2 is
its s m a lle s t prime divisor, we have a j = p2a 2. If a 2 > 1, th en,
in exactly the same way, we find a2 = p3a3, etc. until we come
to some an equal to one. Then a n - l = pn . Multiplying all
these equations together, and simplifying, we obtain the fol­
lowing decomposition of a into prime factors:

a ~ P 1P2 • • • P n •

Assume that there e x is ts a second decomposition of the

same a into prime factors a = q xq2 . . . q s > Then

P1P2 • • • Pn = • qs‘

T he right side of this equation is divisible by q x. There­

fore (b), at le a s t one of the factors of the left side must be
divisible by q x. For example, let p 2 be divisible by qx (in
the order of enumeration in our arrangement) then p 3 = <7, (p,
is divisible only by p t except for 1). Dividing both sid e s of
the equation by p, = qx, we have p2p3 . . . pn = q2q3 . . . q B.
R epeating the preceding argumentation applied to this equa­
tion, we find p , . . . p n = q3 . . , q B, e tc ., until we finally find
that all the factors on one side, sa y the left side, are divided
out. But all the factors on the right side must be cancelled
sim ultaneously since the equation 1 = ^n+1. . . ^ s for ^ n +1,
. . . » 9s greater than 1, is im possible.
Therefore the second decomposition into prime factors is
identical with the first.
d. In the decomposition of the number a into prime factors,
several of them may be repeated. L e ttin g p lt p2, . . . , pjt be
the different primes and a t, a 2, . . . , a k be the m ultiplicity of
their occurrence in a, we obtain the so-called canonical de­
composition o f a into factors:

16
 CL* Qm CLj _
a = p ^ p p . . . pkk .

Example. The canonical decomposition of the number

588 000 is: 588 000 = 2s • 3 • 53 • I 1.
e. L e t a = p ^ p ? 3. . - p k k be the canonical decomposition of
the number a. Then all the divisors of a are ju s t all the num­
bers of the form

d = . . . p£*;
(l)
o < Pi < 0Cj, 0 < ^ a„ ..., 0 ^ < a*.

Indeed, let d divide a. Then (b, §l) a = dq, and therefore

all the prime divisors of d enter into the canonical decomposi­
tion of a with indices no sm aller than those with which they
enter into the canonical decomposition of d. Therefore d is of
the form (1).
C onversely, every d of the form (1) evidently divides a.
Example. All the d iv iso rs of the number 720 = 2 4 • 3 J • 5
can be obtained if we let (3t , (3a, (33 in 2 ^ , 3 ^ J5^1 run inde­
pendently through the values (33 = 0, 1, 2, 3, 4; (3a = 0, 1, 2;
(3j = 0, 1. Therefore th e se divisors are: 1, 2, 4, 8, 16, 3, 6,
12, 24, 48, 9, 18, 36, 72, 144, 5, 10, 20, 40, 80, 15, 30, 60,
120, 240, 45, 90, 180, 360, 720.

Problems for Chapter I

1. L e t a and b be integers which are not both zero, and let

d = ax0 + by0 be the s m a lle s t positive number of the form
ax + by lx and y integers). Prove that d = (a, b). From th is
deduce theorem 1, d, $2 and the theorem s of e, $2. Generalize
these r e s u lts by considering numbers of the form ax + by +
+ . . . + fu.
2. Prove that, of all the rational numbers with denominators

P» rep resen ts the number a most

< Q a, the convergent 8a = ----
ft
exactly.
17
 3. L et the real number a be developed in a continued frac­
tion; le t /V be a positive integer, let k be the number of decimal
digits in it, and let n be the la rg e st integer such that Q„ N.
Prove that n ^ 5k + 1. In order to prove this, compare the
ex p ressio n s for Q2, Qi7 ()4, . . . , Qn with those which would
occur if all the q s were equal to 1, and compare the latter with
the numbers 1, t;, %2, . . . , E,n~2 where E, is the positive root of
the equation E,1 = ^ + 1.
4. L e t t 1. The sequence of irreducible rational fractions
with positive denominators not exceeding r, arranged in in­
crea sin g order, is called the Farey se r ie s corresponding to r.
a. Prove that the part of the F arey s e r i e s corresponding to
r, containing fractions a such that 0 <C a ^ 1, can be ob­
tained in the following way: we write down the fractions
0 1 . f . 0 + 1 1
— , — . If 2 < r, then we in sert the fraction ---------- = —
1 1 1 + 1 2
0
between these fractions, and then in the resu ltin g sequence — ,

1 1 < a, c,
— , — between every two neighboring fractions — and —
2 1 b, d2
a. + c
with b. + d. ^ r we insert the fraction — , and so
bi +
forth as long a s th is is p o s sib le . F ir s t prove that for any two
a c
pairs of neighboring fractions — and — of the sequence, ob-
b d

tained in the above manner, we have ad — be = —1.

b. C onsidering the F arey s e r ie s , prove the theorem: let
r > 1, then every real number a can be represented in the
form

a — + — ; 0 < Q ^ r, (P , Q) = l , | d\ < 1.
Q Qt
18
 c. Prove the theorem of problem b using h, 44.
5. a. Prove that there are an infinite number of primes of
the form 4m + 3.
b. Prove there are Em infinite number of primes of the form
6m + 5.
6 . Prove that there e x is t an infinite number of primes by
counting the number of integers, not exceeding N, whose
canonical decomposition does not contain prime numbers dif­
ferent from p lt p2, . . . , p*.
7. L e t K be a po sitiv e integer. Prove that the sequence of
natural numbers contains an infinite s e t of s e q u e n c e s M, M +
+ 1, . . . , M + K — 1, not containing primes.
8 . Prove that there are an infinite number of composite
numbers among the numbers represented by the polynomial
a0x n + + . . . + an , where n > 0, a 0, a 1? . . . , a n are
integers and a 0 > 0 .
9. a. Prove that the indeterminate equation (1) x 2 + y 2 = z 2,
x > 0 , y > 0 , z > 0, (x, y, z) = 1 is s a tis f ie d by those, and
only those, s y ste m s x, y, z for which one of the numbers x
and y is of the form 2uv, the other of the form u2 — V2, and
finally z is of the form u2 + v2; here u > v > 0, (u, v) = 1,
uv i s even.
b. Using the theorem of problem a, prove that the equation
x* + y* = z* cannot be solved in positive integers x, y, z.
10. Prove the theorem: if the equation x n + a lx n~l + . . . +
+ an = 0, where n > 0 and a lt a2, . . . , an are integers, h a s a
rational root then th is root is an integer.

11, a. L e t S = — + — + . . . + — ; n > 1. Prove that S

2 3 n

is not an integer.

1 1
b. L e t S = — + + + ------------; n > 0. Prove that S
3 5 2n + 1

is not an integer.

19
 12. L e t n be an integer, n > 0. Prove that all the coef­
ficien ts of the expansion of the Newtonian binomial (a + b)n
are odd if and only if n is of the form 2k — 1.

Numerical E xercises for Chapter I

1, a. Applying the E u clid ean algorithm, find (6188, 4709).

b. F ind (81 719, 52 003, 33 649, 30 107).
125
2, a. Expanding a = ^ in a continuous fraction and form­

ing the table of convergents (d, $4), find: a) 6«; (3) the repre­
sentation of a in the form considered in problem 4, b, with
x = 20.
5391
b. Expanding a = --------- in a continuous fraction and form-
3976
ing the table of convergents, find: a ) S6; j8 ) the representation
of a in the form considered in problem 4, b, with r = 1000.
3, Form the F arey s e r ie s (problem 4) from 0 to 1, excluding
1, with denominators not exceeding 8 .
4, Form the ta b le of primes l e s s than 100.
5, a. Find the canonical decomposition of the number
82 798 848.
b. F in d the canonical decomposition of the number
81 057 226 635 000.

20
 CHAP TER II
IMPORTANT NUMBER-
THEORETICAL FUNCTIONS
§ 1. The Functions \_x\, fx:}

a. T he function [x\ plays an important role in number theory;

it is defined for all real numbers x and is the la rg e st integer
not exceeding x. T h is function is called the integral part o f x.
Examples.

[7] = 7; [2.6] = 2; [-4.75] = - 5 .

The function \x\ = x — [x] is also considered sometimes. T h is

function is called the fractional part o f x.
E x a m p le s .

17! = 0; (2.6! = 0.6; (-4.75! = 0.25.

b. In order to show the u s e fu ln e s s of the functions we have

introduced, we prove the theorem:
The power with which a given prime p enters into the product
n! is equal to

~n~ ‘ n~ “ n~
— 4- + 3

w |_P J IP J

Indeed, the number of factors of the product n\ which are

21
m ultiples of p is of th e se the number of m ultiples of p1

is of the la tte r the number of m ultiples of p* is

etc. T he sum of the la tte r numbers gives the required power

s in ce each factor of the product n! which is a multiple of the
maximal pm i s counted m tim es by the above p ro cess, a s a
multiple of p, pJ , ps, . . . , and finally, pm .
Example. T he power to which the number 3 enters into the
product 40! is

r40"| r401 T401

kJ +hd +hd- u+4+1- 18-
§2. Sums Ex t en d ed over the Divisors o f a Number

a. M ultiplicative functions play an important role in number

theory. A function 0(a) is s a id to be multiplicative if the
following conditions are s atisfied :
1. The function 0(a) is defined for all po sitive integers a
and is not equal to zero except p o s si b ly for at most one such a.
2 . For any two relatively prime po sitive integers a r and a a,
we have
Bi.ala1) == d{af) di.a3).

Example. It i s not difficult to s e e that the function 0 (a ) =

= a " , where s is any real or complex number, is m ultiplicative.
b. From the aforementioned properties of the function 0 (a)
it follows in particular that 0(1) = 1. Indeed, le t 0 ( a o) be
different from zero, then 0 (a0) * 0( 1 • a 0) * 0 ( l ) 0 (ao), i.e.
0(1) = 1. Moreover we have the following important property:
if 0,(a) and 04(a) are m ultiplicative functions, then 0o(a) =
= 0,(a)0j(a) is also a m ultiplicative function. Indeed, we find
that
0,(1) = 0 1(1)0J(1) = 1.
22
Moreover, for (a,, a2) = 1, we find

d0i a xa 2) = = d ^ i a fi d ^ i a ^ d ^ a ^ d ^ a f) =

= d ^ a J d j i a J d ^ a J d t i a J = dQiaf)d6ia2).

c. L e t 6 (a) be a multiplicative function and let a =

= p f t p f t • • • p t k be the canonical decomposition o f the number
a. T h e n , denoting by the symbol E t the sum extended over
d\a
all the divisors d o f the integer a, we have

2 ] did) = ( i + e ( Pl) + ei p\) + . . . + d i p “ 0 ) . . .

d \a
. . . (1 + 0 (p/c) + 0 (pjc) + ••• + 0 (pfc*))

i i f a = 1 the right side is considered to be equal to 1).

In order to prove th is identity, we multiply-out the right
s id e . Then we obtain a sum of terms of the form

d ip f t ) d i p f 2) . . . d ip f t ) = d i p f t p f t • • • p f t ) ;

0 < Pi ^ « i , 0 ^ ft2 ^ a 2, . . . , 0 ^ /3k < a k,

where no terms are lacking and there are no repeated terms,

and th is is ex actly the situation on the left (e, §6 , ch. I).
d. F o r dia) = a® the identity of c takes on the form

(i) E d * = ( i + P? + P\* + . . . + Pf t * ) . . .
d \e
• • *(1 + Pk + P*S + • • • + Pk*")-

In particular, for s = 1, the left side of (1) rep resen ts the

sum o f the divisors S(a) of the number a. Simplifying the right
s id e we find

Pla ‘ +1 - 1 p?’+l - 1 p ^ +1 - 1
Sia) = --------------- ■ . . . -----------------.
Pi - l P2 ~ 1 Pk ~ 1
23
 Example.

S(720) = S ( 2 4 • 3 2 * 5) =

2*+1 - 1 31+I - 1 51+l - 1

= --------------- • • = 2418.
2 -1 3 -1 5 -1

F or s = 0 , the left sid e of (1) re p re s e n ts the number of

divisors r(a) of the number a and we find

t (a) = (<xt + 1) ( a 2 + l ) . . . ( a k + 1).

Example.

t (720) = (4 + 1) (2 + 1) (1 + 1) = 30.

§3. The Mobius Function

a. T he Mobius function p(a ) is defined for all positive

integers a. It is given by the equations: p(a) = 0 , if a is
d iv isib le by a square different from unity; p (o) = (—l ) k if a
is not divisible by a square different from unity, where k de­
notes the number of prime divisors of the number a; in particu­
lar, for a = 1, we let k = 0, and hence we take p (1) = 1 .
E x a m p le s .

p(D = 1, g (5 ) = -1, p(9) = 0,

g (2 ) = -1, p(6 ) = 1, g(10) = 1,
g(3) = -1, g(7) = -1, pill) = - 1,
g(4) = 0, g (8 ) = 0, ft (12) = 0.

b. L e t 6(a) be a multiplicative function and let

a. a, aL
a = Pi'Pi1---Pkk
be the canonical decomposition o f the number a. Then

p(d)6(d) = (1 - 0( p , ) ) ( l - 0(p2)) . . . (1 - 0(pk)).

d\a
24
(If a = 1 the right side is taken to be equal to 1.)
Indeed it is evident that the function fi(a) is m ultiplicative.
Therefore the function 0,(a) = p (a) 6 {a) is also m ultiplicative.
Applying the identity of c, 52 to the la tte r, and noting that
0,(p) = —0(p); 6t (ps ) = 0 for s > 1, we have proved the va­
lidity of our theorem.
c. In particular, s e ttin g 6(a) = 1, we obtain from b,

^ = 0, if a > 1,
£ pW )
d \a
= 1, if a = 1.
1
Setting 6 (d) — , we find
d

1
1 --------j , if a > 1 ,
E p(d) Pi - - Pk /
------- <
d \a d
1, if a = 1 .

d. L e t the real or complex f = f lt f it . . . » f n correspond to

the pos itive integers 8 = 5W S2, , 8n .Th en , letting S ' be
the sum o f the values o f f corresponding to the values o f 8
equal to 1, and letting S j be the sum o f the values o f f cor­
responding to the values o f 8 which are multiples o f d, we
have

S ' = 2> W )S d,

where d runs through all the positive integers dividing at lea st

one value o f 8.
Indeed, in view of c we have

s '= A £ p ( d ) + ft E pW ) + . . . + fn £ pW)-
A s, A s, A sn
25
Gathering those terms with the same value of d and bracketing
the coefficient of this p(d), the bracket contains those and
only th o se f whose corresponding S are m ultiples of d, and
this is ju st S d .

§4. The Euler Function

a. Euler's function cp(a) is defined for all positive integers

a and represents the number of numbers of the sequence

(1) 0, 1, a — 1

which are rela tiv e ly prime to a.

E x a m p le s .

9(1) = 1, cp(4) = 2,
9(2) = 1, 9(5) = 4,
9(3) = 2 , 9 (6) = 2.

b. L e t

( 2)
a, a>
a = px lp p Pk'

be the canonical decomposition o f the number a. Then

^ Pi y v Pi I ^ Pk

or also

(4) 9 (a) = (p? 1 - P i 1~1) i p P - p?3~l ) • . . (pk‘k P k k •<);

in particular,

(5) 9 (pa) = p a - p a~l , 9(p) = p - 1.

Indeed we apply the theorem of d, $ 3 . Here the numbers

8 and the numbers f are defined a s follows: le t x run through
26
the numbers of the seq u en ce (1); to each value of x let the
number 5 = (x , a) and the number f = 1 correspond.
Then S ''b e c o m e s the number of v alu es S = (x , a) equal to
1, i.e . becomes cp(a). Moreover S d becom es the number of
values S = (x, a) which are m ultiples of d. But (x, a) can be
a multiple of d only if d is a divisor of the number a. On the
strength of th e s e conditions S d red u ce s to the number of values
a
of x which are m ultiples of d, i.e. to — . Thus we find
d

cf>(a) = J 2 V-(d) —
d \a d

from which formula (3) follows in view of c, 43, and formula

(4) follows from (3) in view of (2).
E x a m p le s .

cp(60) = 60 (1 ------

cp(81) = 81 - 27 = 54;

q>(5) = 5 - 1 = 4.

c. The function cp(a) is multiplicative function.

Indeed, for (a w a a) = 1, it follows evidently from b that

cp(a,aj) = cpCaJ cp(a2).

Example, cp(405) = cp(81) cp(5) = 54 • 4 = 216.

d. £ cp(d) = a.
d \a
In order to prove the validity of this formula we apply the
identity of c, 42, which for 0(a) = cp(a) gives

cp(d) = (1 + cp(Pl) + cp(p|) + . . . + cp(p^)) . . .

. . . ( 1 + cp(pk ) + cp(pi) + . . . + cp(pfcfc) ),

27
In view of (5), the right side can be rewritten as

(1 + (pt - 1) + (pj - Pj) + . . . + ( p ^ - p?1 ‘ I ) . . .

. . . (1 + (p k — 1) + (p^ - pk) + . . . + (p kk - p k k )),

which turns out to be equal to p ^ p ^ 1 ■. • pk k - ° after gather­

ing sim ilar terms in each large p are n th e sis.
Example. Setting a = 12, we find

c p ( l) + cp(2) + cp(3) + cp(4) + cp(6) + c p (1 2 ) =

= l+ l + 2 + 2 + 2 + 4 = 12.

Problem s for C hapter II

1, a. L et the function fix) be continuous and non-negative

in the interval Q ^ x ^ R. Prove that the sum

y . [fu)]
Q < x^.R

is equal to the number of la ttic e points (points with integer

coordinates) in the plane region: Q < x ^ R, 0 < y <: fix).
b. L et P and Q be positive odd relatively prime integers.
Prove that

c. L e t r > 0 and le t T be the number of la ttice points in

the region x 2 + y 2 < P . Prove that

T = 1 + 4[r] + 8 ^ [VP - ** ] -
V2
o<x<—=
V 2

28
 d. L et n > 0 and le t T be the number of la ttic e points of
the region x > 0, y > 0, x y ^ n. Prove that

n
T = 2 E x

2. L e t n > 0, m an integer, m > 1, and let x run through

the p o sitiv e integers which are not divisible by the m-th power
of an integer exceeding 1. Prove that

= M .

3. L e t the positive numbers a and j8 be such that

Caxj, x = 1, 2, • • •, [^ y ]j y = 1, 2, • • •

form, taken together, all the natural numbers without repeti­

tions. Prove that this occurs if and only if a is irrational and

1 1
— + — = 1.
a 0

4, a. L e t t ^ 1, t = [r], and le t x lt x 2, . . . , x t be the num­

bers 1, 2, . . . , t in some order so that the numbers

0, { a x j , { a x j, . . . , {ax, I, 1

are non-decreasing. Prove the theorem of problem 4, b, ch. I,

by considering the differences of neighboring numbers of the
la tte r sequence.
b. L et X , y , . . . , Z be real numbers, each of which is not
l e s s than 1; le t a , /3, . . . , y be real numbers. Prove that there
ex ist integers x , y, . . . , z, not all zero, and an integer u,
satisfy in g the conditions:
29
 I*I ^ *, Iy I ^ y, •••, UI ^ z ,
1
(x, y, . . . , z) = 1, | cxx + (3y + . . . + y z - u \ <

5. L et a be a real number, c an integer, c > 0. Prove that

"[a]' a
c_ _c

6, a. L e t a , /S, . . . , A be real numbers. Prove that

[ a + j8 + . . . + A] £ [ a ] + [/3] + . . . + [A].

b. L e t a , & , . . . , Z be positive in teg ers, and le t a + b + . . .

. . + Z = n. Applying b, $1, prove that

a!

a \ b \ . . . Z!

is an integer.
7. L e t A be a positive integer, p a prime and

p s+1 - 1
u a =* --------------
p - 1

R epresenting h in the form h = pmum + pm_1nm_ 1 + . . .

. . . + pl ul + p0, where um is the la rg e st u a not exceeding h f
pmum is the la rg e st multiple of u m which does not exceed h ,
Pm-ium-i is the la rg e s t multiple of which does not ex­
ceed h - pmu m, p m_jHm_2 is the la rg e s t multiple of um_2
which does not exceed h — p mu m — p m_ ium-i e t c *» prove that
numbers a such that the number p enters into the canonical
representation of a! with th e power h, exist if and only if all
the p m, pm_lt . . . , p lf p0 are l e s s than p, while, if th is occurs,
the numbers a are ju s t all the numbers of the form
30
 ° = PmPm+‘ + Pm—IP”1 + ■•• + PiPJ + PoP + p ' ,

where p ' has the values 0, 1, . . . , p — 1.

8, a. L e t the function f(x) have a continuous second deriva­
tive in the interval Q ^ x ^ R. Setting

p(x) = — - \x ], a( x) = f p(z)dz,
2

prove (Sonin’s formula)

_____ r\

Z /(* > “ f fU )d x + p{H)f{R) - p(Q)f{Q) -

Q <x<R J

- o ( R ) f ' ( R ) + a ( 0 / '( < ? ) + J a ( x ) ( "( x ) d x .

Q

b. L e t the conditions of problem a be s a tis f ie d for arbi-

00

trarily large R , while J \ f " { x ) | dx converges. Prove that

R
R

Z fix) = C + f f(x)dx + p(R)f(R) -

Q<x^R J

- o ( R ) f' ( R ) - J
R
o ( x ) f" ( x ) d x ,

where C does not depend on R.

|A
c. If B ta k e s on only p o sitiv e values and the ratio ----
B
is bounded above, then we write A = 0(B).
L e t n be an integer, n > 1. Prove that

ln(n!) = n In n — n + 0 (ln n)
31
 9, a. L e t n > 2, 0 ( z , z 0) =
Z0<p^2
E
I np, where p runs through

the primes. Moreover, let 0 ( z ) = 0 ( z , 0) and for x > 0,

ib{x) = 0(x) + 0(Vx~) + ®{\fx ) + . . .

Prove that

a) In ([n]!) = ipin) + ip f — ) + ip ( — ] + . . .

/3) ip{n) < 2n

( n \ n n \ n n\

"*T + e 7 'T + 0 7 ' J + ' ” =

= n In 2 + 0{\fn ).

b. F or n > 2, prove that

_ ln p
2^ ------- = In n + 0(1),
p^n P

where p runs through the primes.

c. L e t e be an arbitrary positive constant. Prove that the
sequence of natural numbers contains an infinite number of
pairs pn , pn+l of prime numbers such that

Pn+l ^ P n ( l f)*

d. L e t n > 2. Prove that

1
7~1 — = C + I n i n n + 0 9

p^n P I nn

where p runs through the primes and C does not depend on n.

32
 e. L e t n > 2. Prove that

1 - 1 + 0
P(^7n Inn Jn n ,

where p runs through the primes and C0 does not depend on n.

10. a. L et 0(a) be a m ultiplicative function. Prove that
9t(a) = £ 0(d) is a lso a m ultiplicative function.
d \a
b. L e t the function 0(a) be defined for all positive integers
a and let the function ip(a) = 2!] 0(a) be m ultiplicative.
d \a
Prove that the function 0(a) is also m ultiplicative.
11. F o r m > 0, le t rm(a) denote the number of solutions of
the indeterminate equation x tx 2 . . , x m = a (x2, x2, . . . , x m run
through the positive integers independently of one-another);
in particular, it is evident that r^a) = 1, r4(a) = r(a). Prove
that
a. Tm(a) is a m ultiplicative function.
b. If the canonical decomposition of the number a is of the
fot in a = P1P2P3 • « «P k t then r^(a) = tti .
C. If e is an arbitrary positive constant, then

rl i m ------—
rm(a) = n
0.
a -* CD CL

d. £ rm(a) i s equal to the number of solutions of the in-

O^a^n
equality x 2x2 . . . x m ^ a in positive integers x 2, . . . , x m.
12. L e t R(s) be the real part of the complex number s . For

R(s) > 1, we s e t £ ( s ) = £ ——. L e t m be a positive integer.

nm1
Prove that

(£(s))m = £
n_ 1 na
33
 13, a. For R{s) > 1, prove that
1
t ( s ) = / 7 ----------
1 --------
PS
where p runs through all the primes.
b. Prove that there e x ist an infinite number of primes, s ta rt­
ing from the fact th a t the harmonic s e r ie s diverges.
c. Prove th a t there e x ist an infinite number of prim es, start-
77 2
ing from the fact that £(2) = ---- is an irrational number.
6
14. L e t A (a) = In p for a = p 1, where p is a prime and I
is a positive integer; and le t A (a) = 0 for all other positive
integers a. For R{s) > 1, prove that

£ '( s ) 00 A (n)

15, L e t R{s) > 1. Prove that

CD
p (n )
= £

where p runs through all the primes.

16, a. L et a > 1. Applying d, $3, prove that

n
1 = Z] fiW)
O^cJ^n 7
b. L e t Miz, z 0) = /*(<*); Mix) = Mix, 0). Prove that
z 0<a^z

a ) Min) + M + . = 1 , 71 ^ 1.
( t) * < T>
34
 (' n \ (n n\ (n n\
~2)+ \T ' t ) + l l ’l ) +•••=* ^ 2.

C. L e t n > 1, let I be an integer, I > 1, and le t T lt„ be the

number of in teg ers x , such that 0 < x ^ n, which are not di­
v is ib le by the Z-th power of an integer exceeding 1. Applying
d, $3, prove that

T l,n

17, a. L e t a be a p o sitiv e integer and let the function f{x)

be uniquely defined for the integers x lt x 2, . . . , x n . Prove

s'= E ^ ) s dt
d \a

where S ' is the sum of the v a lu e s of f{x) extended over th o se

v alu es of x which are rela tiv e ly prime to a, and S d is the sum
of the valu es of f[x) extended over those v alu es of x which
are m ultiples of d.
b. L e t k > 1 and consider the sy stem s

/ / / // // // _ . (n) (n>
i • * • i x k> x i » xt » • • • » x k > • • • » i X 2 * 4 n>,

each of which c o n s is ts of in teg ers, not all zero. Moreover,

le t the function f[xlt x 2t . . . , x k ) be uniquely defined for these
sy stem s. Prove that

s ' = y j i{ d ) s d,

where S ' is the sum of the v a lu e s of fbclt x 2t . . .', x k) extended

over s y ste m s of rela tiv e prime numbers, and S d is the sum of
the values of /(*,, x 2, . . . , x k) extended over s y ste m s of num­
bers which are all m ultiples of d. Here d runs through positive
integers.
35
 C. L et a be a po sitiv e integer, and le t F{8) be uniquely de­
fined for the divisors 8 of the number a. Setting

G(8) = £ F{d),
d \S

prove (the inversion law for number-theoretic functions)

F(a) = T fM(d)G
d \e

d. A sso ciate with the positive integers

8lt 8a, . . . . 8n

arbitrary real or complex numbers

fit fit • • • f fn

different from zero. Prove that

„ M(c()
r ' - n p d

where P ' is the product of the valu es f a s s o c ia te d with valu es

of 5 equal to one, and P d is the product of the v alues f a s ­
sociated with values of § which are m ultiples of d, where d
runs through all the po sitiv e integers which divide at le a s t
one 8.
18. L e t a be an integer, a > 1, a min) = l m + 2m + . . . +
+ nm; le t iJjm{a) be the sum of the m-th powers of the numbers
of the seq u en ce 1, 2, . . . , a which are relatively prime to a;
let p,, pJf . . . , pk be all the prime divisors of the number a.
a. Applying the theorem of problem 17, a, prove that

= £ ^ ) d mo m ( — \ .

36
 b. Prove that

a
<A,(a) — <?(a).
2

c. Prove that

a
ip2(a) 2 + — ; — P i P i ' - P k1j y iw ,.
(" 1)k
3

19. L e t z > 1, le t a be a positive integer; let T z be the

number of numbers x such that 0 < x ^ z, (x, a) = 1; le t (
be an arbitrary positive constant.
a. Prove that

z
Tz = E p(d)
d \a 7
b. Prove that

T z = — cp(a) + 0 { a €) .
a

c. L e t z > 1; le t n(z) be the number of prime numbers not

exceeding z; le t a be the product of the primes not exceeding
\fz . Prove that

n(z) = * ( V 7 ) - 1 + 2Z pU)
d \a

20. L e t R(s) > 1 and let a be a positive integer. Prove that

37
where, on the left s id e , n runs through the positive integers
relatively prime to a, while, on the right sid e, p runs through
all the prime divisors of the number a.
21, a. The probability P that k po sitiv e integers x1, x 2, . . . ,
x k are rela tiv e ly prime i s defined as the lim it, as N —> oo, of
the probability P N that the k numbers x lt x 2, . . . , x k are rela­
tively prime, when these k numbers take on the v alu es 1, 2,
. . . , N independently and with equal probability. Applying the
theorem of problem 17, b, prove that P = (£(/c))- 1 .
b. Defining the probability of the irreducibility of the frac-
x 6
tion — as in problem a for k = 2, prove that P = — .
y 77*
22, a. L e t r > 2 and le t T be the number of la ttic e points
(x, y) with rela tiv e ly prime coordinates in the region ** + y 2 <
^ r. Prove that

T = — r 2 + O (rln r).
77
b. L e t r > 2 and let T be the number of lattice points {x, y,
z) with relatively prime coordinates lying in the region x 2 +
+ y 2 + z 1 ^ r 2. Prove that

4>77
T = --------- r J + 0 ( r 2)

23, a. Prove the first theorem of C, $3, by considering the

divisors of the number a which are not divisible by the square
of an integer exceeding 1, and having 1, 2, . . . prime divisors.
b. L e t a be an integer, a > 1, and le t d run through the di­
visors of the number a having no more than m prime divisors;
Prove th > 0 for m even, a n d pid) < 0 for m o dd.
C. Under the conditions of the theorem of d, $3, assuming
all the f to be non-negative and lettin g d run only through the
numbers having no more than m prime d iv iso rs, prove that

S ' <i£p(rr)sd . S ' :>2>(rf)s„

38
according as m is even or odd.
d. Prove the validity of the same in e q u alities as in problem
c, under the conditions of problem 17, a, assum ing all the
v alu es of f(x) are non-negative, as well a s under the conditions
of 17, b, assum ing all the values of f ( x lt x 2, . . . , x k) are non­
negative.
1
24. L e t ( be an arbitrary co n stan t such that 0 < ( < — ; let
6
N be an integer, r = ln/V, 0 < q < N l~ e, 0 ^ I < q, (q, I) = 1;
let n(N, q, I) be the number of primes such that p < N , r =
- qt + I, where t is an integer. Prove that

N (qr) e
n(N, q, I) = 0(A); A
qr

In order to prove this, s e ttin g h = r 1_e, the primes satisfy ­

ing the above condition can be considered to be among all
numbers s a tisfy in g th e se conditions relatively prime to a,
where a is the product of all primes which do not exceed e h
and do not divide q. Ue can then apply the theorem of problem
23, d (under the conditions of problem 17, a) with the above a
and m - 2[2 In r + 1].
25. L e t k be a positive even number, le t the canonical de­
composition of the number a be of the form a = p tp2 . . . p k and
let d run through the divisors of the number a such that 0 <
< d < Va~. Prove that

J^p(d) = 0.
d

26. L e t H e a positive integer, le t d run through the posi­

tive integers such that cp(J) = k. Prove that

£ p(d) = 0.

27. Using the expression for cp(a), prove that there exist an
infinite number of primes.
39
 28. a. Prove the theorem of d, §4 by showing that the num­
ber of integers of the seq u en ce 1, 2, . . . , a which have the

same g re a te st common divisor 5 with o, is equal to cp

b. Deduce e x p re ssio n s for cp(a):

a) using the theorem of problem 10, b;

j8) using the theorem of problem 17, c.

29. L e t R(s) > 2. Prove that

® yU) £(s - 1)
„?1

30. L e t n be an integer, n > 2. Prove that

n 3
cp(m) = — + 0 ( n In »).
m-1 n

Numerical E xercises for Chapter II

1. a. Find the exact power with which 5 enters into the
canonical decomposition of 5258! (problem 5).
b. Find the canonical decomposition of the number 125!
2. a. Find r(2 800) and S(2 800).
b. F in d r( 232 848) and S(23 2 848).
3. Form the table of v alu es of the function g(o) for all
a = 1, 2, . . . , 100.
4. F ind a ) cp(5040); /3) cp( 1 294 700).
5. Form the table of v alu es of the function cp(a) for all
a = 1, 2, . . . , 50, using only formula (5), §4, and theorem
c, §4.

40
 CIIA PTE 1! Ill

CONGRUENCES
§1. Bas ic Concepts

a. We will consider integers in relation to the remainders

resulting from their division by a given positive integer m
which we call the modulus.
To each integer corresponds a unique remainder resulting
from its division by m (c, 41, ch. 1); if the same remainder r
corresponds to two integers a and b, then they are said to be
congruent modulo m.
b. The congruence of the numbers a and b modulo m is
written as

a = Mmod m),

which is read: a is congruent to b modulo m.

c. The congruence o f the numbers a and b modulo m is
equivalent to:
1. The po s si b il i ty o f representing a in the form a = b + mt,
where t is an integer.
2. The di vis ibility o f a - b by m.
Indeed, it follows from a = Mmod m) that

a = mq + r, b = mqt + r; 0 ^ r < m.

and hence

a - b = m(q — qx), a - b + mt, t - (q — qf).

41
Conversely, from a = b + mt, representing b in the form

b = mqx + r, 0 ^ r < m,

we deduce

a = mq + r; q = q 2 + t,

i.e.

a = Mmod m)

proving assertio n 1.
A ssertion 2 follows immediately from assertion 1.

§2. Properties o f Congruences similar to those o f Equations

a. Two numbers which are congruent to a third are congruent

to each other.
T h is follows from a, $1.
b. Congruences can be added termwise.

Indeed, let

(1) o t = 6,(mod m), a2 = &2(mod m), . . . , a k - fefc(mod m)

Then (1, c, § 1)

(2) a^ - b2 + mt^j a2 = b2 + mt2f . . . , a k — b k + mtkt

and hence

a 2 + flj "I- . . . 4- a k - + b2 4- • • • + bk + 4- t2 -I- • • • 4- tff),

or (1, c, 51)

a i + a2 + . . . + a k = + b2 + . . . + 6fc(mod m)
42
 A summand on either side o f a congruence can be put on the
other side by changing its sign.
Indeed, adding the congruence a + b = c(mod m) to the evi­
dent congruence —b = —6(mod m), we find a = c — b(mod m).
Any number which is a multiple o f the modulus can be added
to {or subtracted from) any s id e o f a congruence.
Indeed, adding the congruence a = b(mod m) to the evident
congruence mk s 0(mod m), we obtain a + mk = b(mod m).
C. Congruences can be multiplied termwise.
Indeed, we again consider the congruences (1) and deduce
from them the equations (2). Multiplying equations (2) together
term wise we find

a^a2 • • • ®k b|b2 • • • ^k mN,

where N is an integer. C onsequently (1, c, §1),

a , a 2 . . . a k = blb2 . . . fek(mod m).

Both s id e s o f a congruence can be raised to the same power.

T h is follows from the preceding theorem.
Both sid es o f a congruence can be multiplied by the same
integer.
Indeed, mutliplying the congruence a = 6(mod m) by the
evident congruence k = /c(mod m), we find ak = &A;(mod m).
d. P ro p erties b and c (addition and multiplication of con­
gruences) can be generalized to the following theorem.
I f we replace A, x lt x 2, . . . , x k in the expression o f an
integral rational function S = J ^ A x ^ x ^ 1 . . . x k k with integral

c o e f f i c i e n t s , by the numbers B , y ,, y2, . . . , y k which are con­

gruent to the preceding ones modulo m, then the new expres­
sion S will be congruent to the old one modulo m.
Indeed, from

A = Z?(mod m), x x = y,(mod m),

x 2 h ya(mod m), . . . , x k = y k(mod m)

43
we find (c)

A = fl(mod m), x ^ = y,ai(mod m)

xf* = y^2(mod m), . . ., x £ k = y£k(mod m),

A x * lX j 2 . . .x%k = B y ^ y f 1 • • • y£k(niod m)

from which, summing, we find

2^ Ax^- xf 1 . . . x £ k = Y '.B y ^ y ^ .. . y£k(mod m).

If

a = 6(mod m), a, s 6,(mod m), . . . , a n s £>„(mod m),

x = x,(mod m),

then

axn + a ,* " -1 + . . . + an = bx" + btx"~l + . . . + fen(mod m).

T h is result is a s p e c ia l c a s e of the preceding one.

e. Both sides o f a congruence can be divided by one o f their
common divisors i f it is relatively prime to the modulus.
Indeed, it follows from a = f»(niod m), a = avd, b = b tdt
(id, m) = 1 that the difference a - b, which is equal to (a, -
- bt )d, is divisible by m. Therefore (2, f, $ 2, ch. I) al — bi
is divisible by m, i.e . a t = ^ (m o d m).

§ 3 . Further Properties o f Congruences

a. Both s i d e s o f a congruence and the modulus can be

multiplied by the same integer.
Indeed, it follows from a = 6(mod m) that

a = b + mt, ak = bk + mkt

and hence, ak = 6£(mod mk).

44
 b. Both sid es o f a congruence and the modulus can be
divided by any one o f their common d iv i s o r s .
Indeed, let

a -■ 6 (mod m), a = a 2d, b = b2d, m = m2d.

We have

a = b + m t , a2d = b2d + m2dt, a 2 = b2 + m2t

and hence a, = M(mod m,).

c. I f the congruence a = b holds for severa l moduli, then it
also holds for the modulus equal to the lea st common multiple
o f the se moduli.
Indeed, it follows from a - Mmod m2), a = Mmod m2), . . . ,
a s Mmod mk ) that the difference a — b is divisible by all the
moduli mlt m2, . . . , mk . Therefore (c, 43, ch. I) it must be
d iv isib le by the l e a s t common multiple m of these moduli,
i.e . a = b (mod m).
d. I f a congruence holds modulo m, then it also holds
modulo d, which is equal to any divisor o f the number m.
Indeed, it follows from a h i(mod m) that the difference
a — b must be divisible by m; therefore (1, b, 41, ch. I) it
must be d iv isib le by any divisor d of the number m, i.e.
a = b(mod d).
e. I f one side o f a congruence and the modulus are divisible
by some number then the other si d e o f the congruence must
also be divis ibl e by the same number.
Indeed, it follows from a - Mmod m) that a — b + mt, and
if a and m are multiples of d, then (2, b, 41, ch. I) b must also
be a multiple of d, a s was to be proven.
I. I f a = Mmod m), then (a, m) = (b, m).
Indeed, in view of 2, b, 42, ch. I this equation follows im­
m ediately from a = b + mt.

§ 4. Complete S ys te m s o f Resid ues

a. Numbers which are congruent modulo m form an

equivalence class modulo m.
45
 It follows from this definition that all the numbers of an
equivalence c la s s have the same rem ainder r, and we obtain
all the numbers of an equivalence c l a s s if we let q in the form
mq + t run through all the integers.
Corresponding to the m different values of r we have m
equivalence c l a s s e s of numbers modulo m.
b. Any number of an equivalence c l a s s is said to be a resi­
due modulo m with re sp e c t to all the numbers of the equiva­
lence c la s s . T he residue obtained for q = 0 is equal to the
remainder r itse lf, and is called the least non-negative residue.
T he residue p of s m a lle s t ab so lu te value is called the
absolutely le as t residue.
m m
It is evident that we have p = r for r < — ; for r > — we
K 2 2
m
have p = r — m; finally, if m is even and r = — , then we can

m m m
take for p e ith er of the two numbers — and — — m = —— .
2 2 2
T aking one residue from each equivalence c la s s , we obtain
a complete s y s te m o f residues modulo m. F requently, a s a
complete system of re sid u e s we use the le a s t non-negative
re sid u e s 0, 1, . . . , m — 1 or the absolutely le a s t resid u es; the
la tter, as follows from our above d is c u s s io n , is represented in
the case of odd m by the sequence

m —1 1
~ * • • • t ~1» 0, 1,

and in the c a s e of even m by e ith er of the two seq u en ce s

m m
1, . . , - 1 , 0, 1, .
2 + ~2’
m m
.., 1 , 0 , 1, • — + 1.
2 ’ 2
46
 c. Any m numbers which are pairwise incongruent modulo m
form a complete s y s t e m o f residues modulo m.
Indeed, being incongruent, these numbers must belong to
different equivalence c l a s s e s , and s in ce there are m of them,
i.e. as many as there are c l a s s e s , it follows that one number
falls into each c l a s s .
d. I f (a, m) - 1 and x runs over a complete s y s t e m o f resi­
dues modulo m, then ax + b, where b is any integer also runs
over a complete s y s te m o f residues modulo m.
Indeed, there are a s many numbers ax + b as there are
numbers x, i.e. m. Accordingly, it only remains to prove that
any two numbers a x j + b and ax2 + b corresponding to incon­
gruent x t and x 2 will also be incongruent modulo m.
But, assum ing that a x 2 + b = ax2 + Mmod m), we arrive at
the congruence a x 2 = ax2 (mod m), from which we obtain
x x = x 2 (mod m) as a consequence of (a, m) = 1, and this
contradicts the assum ption of the incongruence of the num­
bers x t and x 2.

§5. R ed uc ed Sys tem s o f R es i d u es

a. By f, $3, the numbers of an equivalence c la s s modulo m

all have the same g re a te st common divisor relative to the
modulus. P articu larly important are the equivalence c la s s e s
for which this divisor is equal to unity, i.e. the c l a s s e s con­
taining numbers relatively prime to the modulus.
Taking one residue from each such c l a s s we obtain a re­
duced s y s t e m o f residues modulo m. A reduced system of
resid u es therefore c o n s is ts of the numbers of a complete s y s ­
tem which are relatively prime to the modulus. A reduced
system of resid u es is usually chosen from among the numbers
of the system of l e a s t non-negative resid u es 0, 1, . . . , m — 1.
Since the number of these numbers which are relatively prime
to m is cp(m), the number of numbers of a reduced system ,
which is equal to the number of equivalence c la s s e s contain­
ing numbers relatively prime to the modulus, is cp(m).
47
 Example. A reduced system of resid u es modulo 42 is

1, 5, 11, 13, 17, 19, 23, 25, 29, 31, 37, 41.

b. Any cp(m) numbers which are pairwise incongruent modulo

m and relatively prime to the modulus form a reduced sy ste m
o f residues modulo m.
Indeed, being incongruent and relatively prime to the
modulus, th e s e numbers belong to different equivalence c la s ­
s e s which contain numbers relatively prime to the modulus,
and s in c e there are cp(m) of them, i.e . a s many as there are
c la s s e s of the above kind, it follows that there is one number
in each c la s s .
c. I f (a, m) = 1 and x runs through a reduced s y s t e m o f
residues modulo m, then ax also runs through a reduced s y s ­
tem o f residues modulo m.
Indeed, there are a s many numbers ax a s there are numbers
x, i.e . cp(m). By b, it only remains to prove that the numbers
ax are incongruent modulo m and are relatively prime to the
modulus. But the first was proved in d, $4 for the numbers of
the more general form ax + b, and the second follows from
(a, m) = 1, (x, m) = 1.

§ 6 . The Theorems o f Euler and Fermat

a. For m > 1 and (a, m) = 1, we have (Euler's theorem)".

i (mod m).

Indeed, if x runs through a reduced system of resid u es

* = rt , r,, . . . , r c; c = cp(m),

which c o n s is ts of the le a s t non-negative re s id u e s , then the

le a s t non-negative resid u es p ,, p , , . . . , p c of the numbers
ax will run through the same system , but, generally speaking,
in a different order (c, (• 5).
48
 Multiplying the congruences

ari = Pi (m°d ft), a r2 — Pi (mod m), . . . , arc = p c (mod m)

together term w ise, we find

a cr lr , . . . r c = p lpt , . . p e (mod m),

from which we find

a c = 1 (mod m)

by dividing both s id e s by the product rtr2 . . . r c = p 2p2 . . . p c .

b. I f p is a prime and a is not div isi bl e by p, then we have
{Fermat's theorem):

(1) a p“ 1 s 1 (mod p).

T h is theorem is a consequence of theorem a for m = p. The

la tte r theorem can be put in better form. Indeed, multiplying
both s i d e s of the congruence (1) by a , we obtain the congruence

ap = a (mod p),

which is valid for all integers a, s in c e it is valid for integers

a which are multiples of p.

Problems for Chapter III

1, a. R epresenting an integer in the ordinary decimal s y s ­

tem, deduce criteria for divisibility by 3, 9, 11.
b. R epresenting an integer in the calculational system to
the b a s e 100, deduce a criterion for divisibility by 101.
C. R epresenting an integer in the calculational system to
the base 1000, deduce criteria for divisibility by 37, 7, 11, 13.
2, a. L et m > 0, (a, m) > 1, le t b be an integer, let x run
through a complete, while £ runs through a reduced, system of
49
re sid u e s modulo m. Prove that

1 ),

1
— cp{m).
o

b. L e t m > 0, (a, m) = 1; let b, N, t be integers, t > 0; let

ax + b
f{x) = ----------- , f(N) > 0, /(/V +mt) > 0. Prove, for the
m
trapezoid bounded by the lines x = /V, x = N + mt, y = 0,
y = fix), that

(1) 5 =

where S is the area of the trapezoid, while the sum on the

right is extended over all the la ttic e points of the trapezoid
1
where S = 1 for the interior points, S = — for the v ertices,
4
1
5 = — for the remaining points of the contour.

1
c. L etting, in contradistinction to problem b, S = —
6
for the v e rtic e s , prove formula (1) for a triangle with la ttice
point v e rtic e s .
3, a. L e t m > 0, (a, m) = 1, h > 0, le t c be a real number,
let

ax + ipix)
s = £ tn

where ifjix) takes on v alues such that c ^ ipix) ^ c + h for

the values of x considered in the sum. Prove that
50
 1
S ——m ^ h + .
2 2

b. Let M be an integer, m > 0, (a, m) = 1, let A and B be

real numbers, le t

a X M+ m -l
A = — + — , S = £ M , + 5}
m
" m
" x-Af

Prove that

1
S —— m < IM + -■
2

c. L e t M be an integer, m > 0, (a, m) = 1,

Af + m—l
s - £
x-M

where the function f(x) has continuous derivatives f ' ( x ) and

f " { x ) in the interval M $ x ^ M + m - 1, while

f'{M) = — + (a, m) = 1; j 0 | < 1 ;-^ - | f"(x)\ <

m m 2 A A

wh ere

1 ^ m ^ t, t = /I5, /4 ^ 2, A > 1.

Prove th a t

k + 3
S —— m
2

4. L et all the partial quotients in the continued fraction

development of the irrational number A be bounded, let M be
51
an integer, le t m be a positive integer, and le t B be a real
number. Prove that

M+m-l 1
2^ \Ax + B j = — m + 0 ( ln m).
x-M ^

5, a. L e t A > 2, k > 1 and le t the function f{x) have a

continuous second derivative s a tisfy in g the condition

A
— « Ir w I
A A

on the interval Q < x R. Prove that

L - <?> + M < i.
Q < x^.R 2

A = (2T O - Q) In A + 8k A ) A - v>.

b. L e t Q and R be integers, and le t 0 < a ^ 1. Under the

assum ptions of problem a, prove that the number 0 (a) of frac­
tions \f(x)]\ x = Q + 1, . . . , / ? such that 0 ^ Ifip)] < a is
given by the formula

0 ( a ) = o(R - Q) + O' * 2A; \ 0 ' \ < 1.

6, a. L e t T be the number of la ttic e points (x , y) of the re­

gion ac2 + y2 ^ r2 (r ^ 2). Prove that

T = TO2 + 0 ( r /j In r).

b. L et n be an integer, n > 2, and le t £ be E u le r’s constant.

Prove that

r(l) + r(2) + . . . + r(n) = n(ln n + 2E — 1) + (In n)1).

52
 7. A system of n positive integers, each of which i s repre­
sented to the b a s e 2, is s a id to be proper if for every non­
negative integer s , the number of in teg ers in whose representa­
tion 2* occurs, is even, and is said to be improper if this
number is odd for at le a s t one s .
Prove that an improper system can be made proper by de­
creasing or completely deleting some one of its members,
while a proper system can be made improper by d ecreasing or
completely deleting any one of it s members.
8. a. Prove that the form

3nx n + 3n—Ix n_| + . . . + 3*, + x 0,

where x n, . . . , x lf x 0 run through the v alu es —1, 0, 1

independently of one another, rep resen ts the numbers

3 n+l - 1
-H, . . . , - 1 , 0, 1, H = -----------
3 -1

and rep resen ts each of them uniquely.

b. L e t mj, m2, . . . , mk be positive integers which are rela­
tively prime in pairs. U sing c , $4, prove that we obtain a
complete residue system modulo mlm2 . . . m n, by inserting in
the form

“t 1 [71 2 4" jfH g X j "f" • • • “t- Wj • * • X

the numbers x 2, . . . , x k which run through complete residue

s y ste m s modulo mlf mlt . . . , mk .
9. L et mlt m2, . . . , mk be integers which are relatively
prime in p airs, and let

mlm2 . . • 77z k — tt\^\f ^ = m2M2 — . . . — mkMk »

a. Applying c, 44, prove that we obtain a complete system

modulo mlm1 . . , m k by inserting in the form

Mlx l + M2x 2 + . . . + MifXk

53
the numbers x lt x 2f . . . , x k which run through a complete s y s ­
tem of re sid u e s modulo mlt m2, . . . , mk .
b. Applying c, $4, ch. II and b, 55, prove that we obtain
a reduced system of re s id u e s modulo mim2 . . . mk by inserting
in the form

M2x x + M2x 2 + . . . + Mkx k

the numbers x lt x 2l . . . , x k which run through a reduced residue

system modulo mlt m2, . . . , mk .
c. Prove the theorem of problem b independently of theorem
c, 54, ch. II, and then deduce the la tte r theorem from the
former one.
d. F in d an expression for cp(pa ) by an elementary method,
and using the equation in c, 54, ch. II, deduce an expression
for cp(o).
10. L e t m „ m2, . . . , mk be in teg ers greater than 1, which
are relatively prime in pairs, and le t m =* mlm1 . . . mk,
msMa = m.
a. L e t x u x 2, . . . , x k , x run through complete residue s y s ­
tems, while . . . , £k , £ run through reduced resid u e s y s ­
tems modulo mlt m2, . . . , mk , m. Prove that the fractions

Xl X2 Xk
■----- + ----- + . . . + ----- ■
Tfl^ 7712 m

coincide with the fractions • , while the fractions

m
’ £ 6 £k
---- + ------ + coincide with the fractions
m, m, mk m
b. C onsider k entire rational functions with integral coef­
ficien ts of the r v aria b les x , . . . , w( r ^ 1):

f a(x.w) = £ c a, \ . . ,s x<1‘ • - wS; s = ^ ' • - »

G»f • • •9S

54
and le t

fix. w) = ,8 w
.8

s-i
x a, • • ■, w s run through complete residue s y ste m s , while
£ af . . . , &ja run through reduced residue sy stem s modulo ms ;
x, . . . , w run through complete residue sy stem s, while
£, . . . , co run through reduced residue system s modulo m.
Prove that the fractions

A(*i. • ■• » f k ix k> W,

m. m.

' f{x, . . . , w )
coincide with the fractions • , while the fractions
m

• • • > ^ i) f k i ^ k t • • • » 03k)
m. TTl f

fi£, • • •» <y)
coincide with the fractions (a generalization
m
of the theorem of problem a).
11, a. L e t m be a positive integer, le t a be an integer, and
let x run through a complete residue system modulo m. Prove
that

m, if a is a multiple of m

0, otherw ise.

b. L et a be a real number, and let M and P be integers with

P > 0. L e ttin g (a) denote the numerical value of the differ­
ence between a and the integer c lo s e s t to a (the distance

55
from a to the n e a re s t integer), prove that

2 alw ays
Af+P-
e 27Tia. x
L < min ; h >
1
3, for ( a ) ^ — .
6

c. L e t m be an integer, m > 1, and le t the functions M(a)

and P(a) take on integral v alu es such th a t P{a) > 0 for the
values a = 1, 2, . . . , m — 1. Prove that

in (« y p (flh i a
277J-5TX
e
L L
x»Af (a )

\.m In m — m, for m ^ 60.

12, a. L e t m be a po sitiv e integer, and le t £ run through a

reduced residue system modulo m. Prove that

fi(m) = J 2 e 2nl m
€

b. Using the theorem of problem a, prove the first of the

theorems of c , $3, ch. II (cf. solution of problem 28, a, ch. II).
c. Deduce the theorem of problem a, using the theorem of
problem 17, a, ch. II.
d. L e t

CL, . . . , S

be an entire rational function with integral coefficients of the

r v a ria b le s x, . . . , w{r > 1) and let a, m be integers with
m > 0; x, . . . , w run through complete residue sy stem s, while
£, . . . , a) run through reduced residue sy stem s modulo m. We
introduce the symbols
56
 a / ( x , . . . , w)
2 TTi -
S..» - E - - - E

S a,m = Z! • * • Z ] eXP(flA£» • » " ) / m)

Moreover, let m = . . . m k , where m „ . . . , mk are integers

exceeding 1 which are relatively prime in pairs, and let
msMs = m. Prove that

<? <? _ c
Kj6 l f m l , ’ ’ ‘J a k , m k KjM i a l + . . . + M k a k , m f

S'J a' l , m | ‘ ‘S
^ a' k , m k - ,JM
S' la l + .. . + Mka k , m •

e. Using the notation of problem d we se t

A(m) ™ ' Z ] Sa,m * A '(™) mt

a

where a runs through a reduced residue system modulo m.

Prove that

A (m,) . . . A (mk) = A{m), A ' ( m j . . . A '( m k) = A ' ( m ) .

13, a. Prove that

277 i *
cp(a) = £

where p runs through the prime divisors of the number a.

b. Deduce the well-known expression for cp(a) from the
identity of problem a.
14. Prove that

r(a) = lim 2f £ £ £ " (l+C)exp(277iW*) + 5

€—0
*“ 1
57
where 5 = 1 or 5 = 0, according as a is or is not the square
of an integer.
15, a. L e t p be a prime and le t h lf h3, . . . , h e be integers.
Prove that

(hl + h2 + . . . + h g)p = hp + h3 + . . . + h p (mod p).

b. Deduce F erm at’s theorem from the theorem of problem a.

c. Deduce E u le r’s theorem from F erm at’s theorem.

Numerical Exercises for Chapter III.

1, a. F ind the remainder resu ltin g from the division of

(12 371s6 + 34)28 by 111.
b. Is the number 21093 - 2 divisible by 1 093J ?
2, a. Applying the divisibility criteria of problem 1, find
the canonical decom position of the number 244 943 325.
b. F ind the canonical decomposition of the number
282 321 246 671 737.

58
 CHAPTER IV

CONGRUENCES IN ONE
UNKNO W N
§ 1 . B a s ic Concepts

Our immediate problem is the study of congruences of the

general form:

(1) fix) = 0(mod m); fix) = axn + a 1x n~i + . . . + an .

If a is not d iv isib le by m, then n is said to be the degree

o f the congruence.
Solving a congruence means finding all the v alu es of x
which s a tis f y it. Two congruences which are s a tis f ie d by the
same values of x are s a id to be equivalent.
If the congruence (1) is s a tisfie d by some x = x lf then
(d, $2, ch. Ill) this congruence will also be s a tis f ie d by all
numbers which are congruent to x t modulo m: x = x l (mod m).
T h is whole c l a s s of numbers is considered to be one solution.
In accordance with this convention, congruence (1) has as
many solutions as residues o f a complete s y s te m s at isf yi ng it.
Example. The congruence

x s + x + 1 s 0(mod 7)

is s a tisfie d by two numbers x = 2 and x = 4 among the num­

bers 0, 1, 2, 3, 4, 5, 6 of a complete residue system modulo 7.
59
Therefore the above congruence h a s the two solutions:

x = 2(mod 7), x = 4(mod 7).

§ 2. Congruences o f the First Degree

a. A congruence of the first degree whose constant term h as

been placed on the right side (with opposite sign) can be put
in the form

(1) ax - Mmod m).

b. Turning to the investigation of the number of so lu tio n s,

we first re s tric t the congruence by the condition (a, m) = 1.
According to §1, our congruence has a s many solutions a s
resid u es of a complete system s a tisfy it. But when x runs
through a complete system of re sid u e s modulo m, ax also runs
through a complete residue system (d, 4, ch. III). Therefore,
in particular, ax will be congruent to b for one and only one
value of x taken from the complete residue system . Therefore
congruence (1) h as one solution for (a, m) = 1.
c. Now le t (a, m) = d > 1. Then, in order that the con­
gruence (1) have a solution it is n e c e s s a ry (e, $3, ch. Ill)
that b be divisible by d, for otherw ise the congruence (1) is
im possible for all integers x. Assuming then that b is a
multiple of d , we s e t a = a xd, b = bxd, m = mxd. Then the
congruence (1) is equivalent to the following one (obtained
by dividing through by d): a xx s fe^mod mf), in which (ct,
mx) = 1, and therefore it will have one solution modulo ml .
L e t x x be the le a s t non-negative residue of this solution
modulo ml , then all the numbers x which are solutions of this
equation are found to be of the form

(2) x = aCjfmod m j .

But modulo m the numbers of (2) do not form one solution,

but many so lu tio n s, and indeed as many solutions as there are
60
numbers of (2) in the sequence 0, 1, 2, . . . , m ~ 1 of le a st
non-negative re s id u e s modulo m. But th e se co n s is t of the
following numbers of (2):

x lt x t + mir x l + 2mlt . . . , x l + (d - l)m ,,

i.e . d numbers of the form (2), and hence the congruence (1)
h as d solutions.
d. Gathering together our r e s u lts , we obtain the following
theorem:
L e t (a, m) = d. The congruence ax - fe(mod m) is impossi­
ble i f b is not divis ibl e by d. For b a multiple o f d , the con­
gruence has d s o l u t i o n s .
e. Turning to the finding of solutions of the congruence (1),
we shall only consider a method which is based on the theory
of continued fractions, where it is sufficient to re s tric t our­
s e lv e s to the ca s e in which (a, m) = 1.
Developing the fraction m /a in a continued fraction,

m 1
~ = <h +
a 1
+ -------
?! + 1
+
<7n

and considering the l a s t two convergents:

P n -l Pn _

Qn —l Qn a

by the properties of continued fractions (e, §4, ch. I) we have

m Q n-i ~ a P n -i = (-!)">

aPn-i s ( - l ) n-1 (mod m),

a • ( - l ) n-1P n_1£> s 6(mod m).

61
 H ence, our congruence has the solution

x = ( - l ) n 1/ , n_, t(m od m),

for whose calculation it is sufficient to calcu la te P n_, by

the method described in d, §4, ch. I.
Example. We solve the congruence

(3) 111* = 75(mod 321).

Here (111, 321) = 3, while 75 is a multiple of 3. Therefore

the congruence h as three solutions.
Dividing both s id e s of the congruence and the modulus by
3, we obtain the congruence

(4) 37x = 25(mod 107),

which we m ust first solve. We have

107 II
74 2
37 33
33 1
33
32
4 _1_
_4 4
»»

7 2 1 8 4

Ps 1 2 3 26 107

Hence n = 4, P n-1 = 26, b = 25, and we have the solution

of congruence (4) in the form

x = - 2 6 • 25 s 99(mod 107).
62
 From th is the so lu tio n s of congruence (3) can be represented
in the form:

* = 99, 99 + 107, 99 + 2 • 107(mod 321),

i.e.

* = 99, 206, 313(mod 321).

§ 3. S ys te m s o f Congruences o f the First Degree

a. We shall only consider the sim p le st system of congruences

(1) x - (mod r a j , x : b2 (mod m2) .......... x - bk (mod mk)

in one unknown, but with different and pairw ise prime moduli.
b. It is p o s sib le to solve the system (1), i.e . find all values
of x satisfy in g it, by applying the following theorem:
Let the numbers Ms and Ma be defined by the conditions

m2m2 . . . ma = Mam a, MaMa = 1 (mod m a)

and let

x0 = + M2M2'b 2 + . . . + MkMkbk .

Then the s e t o f values o f x s a ti s fy in g the s y s te m (1) are de­

fined by the congruence

(2) x = x 0 (mod mlm2 . . • ™k)

Indeed, in view of the fact that all the Mj which are different
from Ma are divisible by m„ , for any s = 1, 2, . . . , k, we have

x 0 = MaMab s s ba (mod ma),

and therefore system (1) is s a tis fie d by x = x 0. It follows

immediately from this, that the system (1) is equivalent to
63
the system

(3) x s x 0 (mod m,), x = x 0 (mod m2), . . . , x = x 0 (mod mk)

(i.e. the sy stem s (1) and (3) are s a tis f ie d by the same values
of x). But the system (3), in view of the theorems of c, $3,
ch. Ill and d, $3, ch. Ill, is s a tis f ie d by those and only those
valu es of x which s a tis f y the congruence (2).
c. I f b2, b2, b k independently run through complete
residue s y s t e m s modulo mit m2, . . . , mk , then x 0 runs through
a complete residue s y s t e m modulo m1mI .. , m k.
Indeed, x 0 runs through m2m2 . . , m k values which are incon-
gruent modulo m1m1 . . . m k , in view of d, §3, ch. III.
d. Example. We solve the system

x = 6j(mod 4), x = b2 (mod 5), x = 63(m od7).

Here 4 • 5 * 7 = 4* 35 = 5* 28 = 7* 20, while

35 * 3 = 1 (mod 4), 28 • 2 s 1 (mod 5), 20 • 6 b 1 (mod 7).

Therefore

x = 35 • U t + 28 • 2b2 + 20 • 6b2 = 105b2 + 566, + 120b 2

and hence the se t of values of x satisfying the system , can be

represented in the form

X = 1056, + 5662 + 1206, (mod 140).

T h u s, for example, the s e t of v alues satisfying the system

x b 1 (mod 4), x = 3 (mod 5), x = 2(mod 7),

is

X 3 105 * 1 + 56 • 3 + 120 • 2 = 93 (mod 140)

64
while the s e t of v a lu e s of x satisfy in g the system

x = 3 (mod 4), x = 2 (mod 5), x = 6 (mod 7),

is

X ee 105 • 3 + 56 • 2 + 120 • 6 ee 27 (mod 140).

§4. Congruences o f Arbitrary Degree with Prime Modulus

a. L e t p be a prime. We shall prove general theorems re­

lating to congruences of the form

(1) f(x) = 0 (mod p); f{x) = axn + a ,x n“ ' + . . . + a n .

b. A congruence o f the form (1) is equivalent to a con­

gruence o f degree not higher than p — 1.
Indeed, dividing fix) by x p — x , we have

fix) = (xp - x)Q(x) + /?(*),

where the degree of R{x) is not higher than p — 1. But

xp - x = 0(mod p) im plies that fix) = Rix) (mod p), from
which our theorem follows.
c. I f the congruence (1) has more than n s o lu ti o n s , then all
the coefficients o f fix) are multiples o f p.
Indeed, le t the congruence (1) have at le a s t n + 1 solutions.
L etting x,, Xj, . . . , xn , x n+l be the re sid u e s of th e se solutions,
we can represent fix) in the form

(2) fix) = a(x - x ,) ( x - x2) . . ,(x - x n_a)(x - xn_ ,) ( x - x n) +

+ bix - x,) (x - x2) . . . (x - x n_2) (x - x n_,) +
+ c(x - X,) (x - Xj) . . . (x - xn_2) +
+ .................................................................+
+ ki x — x,) (x - x2) +
+ li x ~ x,) +
+ m.
65
 To this end, develop the summands on the right sid e into
polynomials, and then choose b so that the sum of the coef­
ficients of x n~l in the first two polynomials coincide with a,;
knowing b, we choose c so that the sum of the coefficients of
x n~1 in the first three polynomials coincides with a 2, etc.
P u ttin g x = x lf x 2, . . . , x n , x n+l s u c c e s s iv e ly in (2), we
find that all the numbers m, I, k, c, b, a are m ultiples of
p. T h is means that all the coefficients a, al . . . , an are
m ultiples of p (since they are sums of numbers which are
multiples of p).
d. For prime p, we have the congruence {Wilson's theorem)

(3) 1 • 2 . . . (p — 1) + 1 = 0 (mod p).

Indeed, if p = 2, then the theorem is evident. If p > 2,

then we consider the congruence

(x - 1) {x - 2 ) . . . (x - (p - 1)) - (xp~l - 1) = 0 (mod p);

its degree is not higher than p — 2 and it h a s p - 1 solu­

tions, indeed solutions with re sid u e s 1 , 2 , . . . , p — 1. There­
fore, by theorem c, all its coefficients are m ultiples of p; in
particular the constant term is also divisible by p and the
constant term is ju s t equal to the left s id e of the congruence
(3).
Example. We have 1 • 2 • 3 • 4 * 5 • 6 + 1 = 721 = 0 (mod 7).

§5. Congruences o f Arbitrary Degree with Composite Modulus

a. I f mlt m2, . . . , mk are pairwise prime, then the congruence

(1) f{x) = 0(mod mlm1 . . . mk)

is equivalent to the s y s t e m

fix) = 0(mod m,),

f{x) = 0 (m o d m ,), . . . , fix) = 0 ( m o d m k).

66
L e tt in g Tlt Tlf . . . , T k be the numbers o f solutions o f the
individual congruences o f this s y s te m with respect to the
corresponding moduli, and letting T be the number o f solutions
o f the congruence (1), we have

T = T J 2. . . T k.

Indeed, the first part of the theorem follows from c and

d, $3, ch. III. T he second part of the theorem follows from
the fact that each congruence

(2) f{x) = 0 (mod ms)

is s a tis f ie d if and only if one of the T s congruences of the

form

x = bs (mod m s ),

where b B runs through the re sid u e s of the so lu tio n s of the

congruence (2), is s a tis f ie d , while all T 2T2 . . . T k different
combinations of the form

i s t , (mod m,), x = b2 (mod m2), . . . , x = b k (mod mk),

are p o s sib le , which le a d s (c, $3) to different c l a s s e s modulo

• • • TTb•
Example. The congruence

(3) fix) = 0(mod 35), f{x) = x* + 2x* + 8x + 9

is equivalent to the system

f(x) = 0 ( m o d 5 ) , fix) - 0 (m o d 7 ).

It is easy ($1) to verify that the first congruence of this s y s ­

tem h as two solutions: x = 1; 4 (mod 5), the second con­
gruence h as three solu tio n s: x s 3; 5; 6 (mod 7). Hence the
67
congruence (3) h a s 2 * 3 = 6 solutions. In order to find these
six s o lu tio n s, we must solve six sy stem s of the form

(4) x s bx (mod 5), x = b2 (mod 7),

which we obtain by letting bx run through the values bx = 1; 4,

while b2 runs through the v alu es b2 = 3; 5; 6. But since

35 = 5 * 7 = 7* 5, 7 * 3 s 1 (mod 5), 5 *3 = 1 (mod 7),

the s e t of v a lu e s of x satisfying the system (4) can be repre­

sented in the form (b, §3)

x = 21 bx + 156a (mod 35).

Therefore the solutions of congruence (3) are

x s 31; 26; 6; 24; 19; 34 (mod 35).

b. In view of theorem a the investigation and solution of

congruences of the form

fix) = 0 (mod p f ‘p2a j . . . p k k)

reduces to the investigation and solution of congruences of

the form

(5) fix) = 0 ( m o d p a);

th is la s t congruence reduces in general, as we s h all soon s e e ,

to the congruence

(6) fix) = 0 (mod p)

Indeed, every x satisfying the congruence (5) must n e c e s ­

sarily s a tisfy the congruence (6). L et

x = x x (mod p)
68
and hence the la tte r equation h a s one solution:

t2 = t f (mod p),

t2 — t2 + p t3.

The expression for x ta k e s on the form

x = x2 + p*t2 + p3t3 = x 3 + p3t3;

and so forth. In this way, given a solution of the congruence

(6) we can find a solution of the congruence (5) which is con­
gruent to it. Hence, i f f ' ( x f ) is not divis ibl e by p, each so­
lution x = x, (mod p) o f the congruence (6) gives a solution
o f the congruence (5):

x = xa + p ata5
x = x a (mod p a).

Example. We so lv e the congruence

” f(x) = 0(mod 27);

(7)
f{x) = X* + l x + 4.

The congruence f(x) = 0(mod 3) h a s one solution x = 1 (mod

3); here f ' ( 1) = 2(mod 3), and hence, is not divisible by 3.
We find
x = 1 + 3t,,

f( 1) + 3 t , n i ) = 0 (mod 9), 3 + 3t, • 2 = 0(mod 9),

2t, + 1 = 0(mod 3), t, = l ( m o d 3 ) , t, = 1 + 312,

x = 4 + 9t2,

f{4) + 9t2f ' ( 4 ) = 0 (mod 27), 18 + 9t2 • 2 = 0 (mod 27),

2t2 + 2 s 0(mod 3), t2 2 (mod 3), t = 2 + 3t3,

x = 22 + 2713.
70
Therefore, the congruence (7) has one solution:

x = 22 (mod 27).

P roblem s for C hapter IV

1, a. L e t m be a positive integer and le t fix, . . . , w) be an

entire rational function with integral coefficients of the r
v ariables x, . . . , w(r ^ 1). If the system x = x 0, . . . , w = w 0
s a ti s f i e s the congruence

(1) f{x, . . . , w) = 0 (mod m),

then (generalizing the definition of 41) the system of c l a s s e s

of integers modulo m:

x - x 0 (mod m), . . . » w = w 0 (mod m)

will be considered to be one solution of the congruence (1).

L e t T be the number of s o lu tio n s of the congruence (1).
Prove that

m -rl m —l m —1 al(x, . . . , w)
e
2TTI m
T m - Z Z - Z .
am 0 J T -0 * V -0

b. U sing the notation of problem a and problem 12, e, ch.

ID, prove that

Tm = mr Z ^ (m0).
m0\m

C. Apply the equation of problem a to the proof of the

theorem on the number of solutions of a congruence of the
first degree.
d. L e t m be a positive integer; le t a, . . . , / , g be
r + I (r > 0) integers; d = (a, . . . , /, m); le t T be the
71
number of solutions of the congruence

ax + . . . + fw + g = 0(mod m).

U sing the equation of problem a, prove that

mr~l d , if g is a multiple of d,

H I 0, otherw ise.

e. Prove the theorem of problem d, starting from the theorem

on the number of solutions of the congruence ax = 6 (mod m).
2, a. L e t m > 1, (a, m) = 1. Prove that the congruence
ax = b (mod m) has the solution x = ba^>^m ^~1 (mod m).
b. L e t p be a prime, 0 < a < p. Prove that the congruence
ax = fc(mod p) h as the solution

1/
x = o ( - l)
(P— !Hp - - a + D (mod
, J p).,
1*2... a

C, a ) F ind the sim p lest p o ssib le method of solving a con­

gruence of the form

2 kx = b (mod m); (2, m) = 1.

/3) Find the sim p lest p o ssib le method of solving a con­

gruence of the form

3kx = b (mod m); (3, m) = 1.

y) L e t (a, m) = 1, 1 < a < m. Applying the methods used

in problems a ) and /3), prove that finding the so lu tio n s of the
congruence ax = b (mod m) can be reduced to finding the solu­
tions of a congruence of the form b + mt = 0(mod p) where p
is a prime divisor of the number a.
3. L e t m be an integer, m > 1, 1 < r < m, (a, m) = 1.
U sing the theory of congruences prove the ex iste n c e of
72
integers x and y such that

m
ax = y (m o d m ), 0 < * < r, 0 < | y | < —.

4, a. F or (a, m) = 1, we will co n sid er the symbolic frac-

b
tion — modulo m, which denotes any residue of a solution of
a
the congruence ax = b (mod m). Prove that (the congruences
are taken modulo m)
b b,
a ) F or a a lf b = 6, we have — = — .

/3) The numerator b of the symbolic fraction — can be re-

a
placed by a congruent b0 which is a multiple of a. Then the
b
symbolic fraction — is congruent to the ordinary fraction
a
K
— , where the congruence is taken with ordinary integers.
a
b d be + ad
y) ~ + ~ = ----------
a c ac
b d bd
S ) --------- --- -----
a c ac

b f a ) L e t p be a prime, p > 2, and let a be an integer,

0 < a < p — 1. Prove that

^ ' h (-1)° (mod p).

j3) L et p be a prime, p > 2. Prove that

2P - 2 1 1 1
----------- = 1 _ — + ------ . . . -------------- (mod p).
p 2 3 p - 1

73
 5, a. L e t d be a divisor of the number a which is not di­
v isib le by primes sm aller than n, and le t k be the number of
different divisors of the number d. Prove that the number of
m ultiples of d in the sequence

(1) 1 * 2 . . . n, 2 * 3 . . . (n + 1), , a(a + 1 ) . . . (a + ra — 1)

n K„a
is ------- .
d
b. L e t p t, p2f . . . , p k be the different prime divisors of the
number a which are not sm aller than n. Prove that the number
of in teg ers of the seq u en ce (1) re la tiv e ly prime to a is

n
a [1 - — 1 n 1
V Pi Pi Pk

6. L e t mliai . . . if be the l e a s t common multiple of the num­

bers , n 2, . . . , tnk •
a. L e t d = (m„ m2). Prove that the system

x = b1 (mod m,), x = b2 (mod m,)

is solvable if and only if b2 — 6, is a multiple of d, and if the

system is so lvable, the s e t of valu es of x satisfying this s y s ­
tem is determined by a congruence of the form

x = x 1^ (mod ml>2).

b. Prove that, if the system

x = (mod wij), x = b2 (mod m2), . . . , x = bk (mod mk)

is so lvable, the s e t of v alues of x satisfy in g it is determined

by a congruence of the form

X - x 1»2, • (mod m.

74
 7. Let m be an integer, m > 1, let a and b be integers, and
let

ax + b x '
( — ) = £ exP
m

where x runs through a reduced residue system modulo m,

1
while x ' = — (mod m) (in the s e n s e of problem 4, a). Prove

the following properties of the symbol

/ a, b \
a ) ( ------- ] is real.
\ m )

f a, b h \ ah, b
y ) F or (h, m) = 1, we have
\' m 1/ \ m
S ) For m „ mlt . . . , mk relatively prime in pairs, setting
mxmt . . . mk = m, m = m„il/s , we have

/ °1. 1 \ / g2» 1 ak, 1

\ ml ) \ m2 ffl »

M]al + M\a2 + . . . + M*kak , 1

m

8. L e t the congruence

+ ayXn~' + . . . + an = 0(mod p)

have the n solutions

x = x lt x 2, . . . , x n (mod p).
75
 Prove that

Oj s ~a0S , (mod p),

aj s aJS2 (mod p),

o 3 = —(Zq5j (mod p),

a n s (~ l)na 0Sn (mod p),

where St is the sum of a ll the x s , S2 is the sum of the products

of pairs of the x s , S3 is the sum of the products of triples of
the x a, etc.
9, a. Prove Wilson’s theorem by considering pairs x, x ' of
numbers of the sequence 2, 3, p — 2, satisfy in g the con­
dition x x ' s 1 (mod p).
b. L e t P be an integer, P > 1, 1 * 2 . . . (P - 1) + I s
s 0(mod P ). Prove that P is a prime.
10, a. L e t (a0, m) = 1. Find a congruence of degree
n(n > 0) with leading coefficient 1, equivalent to the
congruence

Oqx" + axx n 1 + . . . + an = 0(mod m).

b. Prove that a n e c e ss a ry and sufficient condition in order

that the congruence f(x) = 0(mod p); f{x) = x n + a2x n~l +
+ . . . + an; n ^ p; h as n so lu tio n s, is the divisibility by p of
all the coefficients of the remainder after the division of
x p - x by f[x).
c. L e t n be a divisor of p — 1; n > 1; {A, p) = 1. Prove
that a n e c e ss a ry and s u fficie n t condition for the solvability
p-i
of the congruence x n = A (mod p) is A n s i (mod p), while
if the congruence is solvable, it h a s n solutions.
11. L et n be a positive integer, {A, m) = 1, we assum e that
we know a solution x = x 0(mod m) of the congruence
76
x n = A (mod m). Prove that all the solutions of this con­
gruence can be represented as the product of x 0 and a residue
of a solution of the congruence y n = l ( mo d m) .

Numerical E xercises for Chapter IV

1, a. Solve the congruence 256* = 179 (mod 337).

b. Solve the congruence 1215*: = 560 (mod 2755).
2, a. Solve the congruences of e x e rc is e s 1, a and 1, b by
the method of problem 2, c.
b. Solve the congruence 1296x = 1105 (mod 2413) by the
method of problem 2, C.
3, Find all pairs x , y satisfy in g the indeterminate equation
1245* - 1603y = 999.
4, a. Find a general solution of the system

x = 6, (mod 13), x = bj (mod 17).

U sing this general solution, find three numbers whose divi­

sion by 13 and 17 g ives the resp ectiv e remainders 1 and 12,
6 and 8, 11 and 4.
b. Find a general solution for the system

x = 6, (mod 25), x = b2 (mod 27), x = f>3 (mod 59).

5, a. Solve the system of congruences

x = 3 (mod 8), x = 11 (mod 20), x = 1 (mod 15).

b. Solve the system of congruences

x = 1 (mod 3), x = 4 (mod 5), x = 2(mod 7),

x s 9 (mod 11), x = 3 (mod 13).

6, Solve the system of congruences

x + 4y — 29 — 0(mod 143), 2x — 9y + 84 s 0(mod 143).

77
 7, a. What congruence of degree sm aller than 5 is equivalent
to the congruence

3a:14 + 4a:13 + 3a;12 + 2xu + x9 + 2x * + 4a:7 + x 6 +

+ 3a:4 + x i + 4a:2 + 2x = 0(mod 5)?

b. What congruence of degree sm aller than 7 is equivalent

to the congruence

2a:17 + 6*16 + a:14 + 5*12 + 3a:11 + 2a;10 + a* + 5a;5 +

+ 2a:7 + 3a:5 + 4a;4 + 6a^ + 4a:2 + x + 4 = 0(mod 7)?

8, What congruence with leading coefficient 1 is equivalent

to the congruence (problem 10, a)

7<V + 78a:5 + 25a;4 + 68a:3 + 52a^ + 4a: + 3 = 0(mod 101)?

9, a. Solve the congruence

fix) = 0(mod 27), fix) = l x 4 + 19a: + 25,

by first finding all the solutions of the congruence

fix) = 0 (mod 3)

by trial.
b. Solve the congruence 9a;2 + 29a: + 62 = 0(mod 64).
10, a. Solve the congruence a^ + 2a: + 2 = 0 (mod 125).
b. Solve the congruence x 4 + 4x* + 2a^ + 2x + 12 =
= 0(mod 625).
11, a. Solve the congruence 6a^ + 27a^ + 17a: + 20 =
- 0 (mod 30).
b. Solve the congruence 31a:4 + 57a^ + 96a; + 191 =
= 0 (mod 225).

78
 CHAPTER V

CONGRUENCES OF THE
SECOND DEGREE
§ 1 . General Th eorems

a. We shall only consider the sim p le st of the congruences

of degree n > 1, i.e . the two-term congruences:

(1) x n s a (mod m); (a, m) = 1

If the congruence (1) has solutions, then a is s a id to be an

n-th power resid ue , otherw ise a is said to be an n-th power
non-residue. In particular, for n = 2 the re s id u e s or non­
re s id u e s are s a id to be quadratic, for n = 3 cubic, for n = 4
biquadratic.
In th is chapter we shall consider the ca s e n = 2 in detail
and we first consider the two-term congruences of the second
degree for odd prime modulus p:

(2) x? h a (mod p); (a, p) = 1.

c. I f a is a quadratic residue modulo p, then the congruence

(2) has two s o l u t i o n s .
Indeed, if a is a quadratic residue, then the congruence
(2) h a s at le a s t one solution x = x x (mod p). But since
(—x lY = x\, the same congruence also h as the second solution
X = —x x (mod p). T h is seco n d solution is different from the
79
first s in ce x t = - x t (mod p) would imply 2xx = 0(mod p),
which is im possible since (2, p) = (xlf p) = 1.
T h e s e two so lu tio n s e x h au st all the solutions of the con­
gruence (2) sin ce the latter, being a congruence of the second
degree, cannot have more than two solutions (c, $4, ch. IV).
p - 1
d. A reduced residue s y s te m modulo p c o n s is ts o f ---------
2
quadratic residues which are congruent to the numbers
2
P ~ 1
(3) 1J
i , ^2J , . . . ,
2

P —1
and — - — quadratic non-r es idu es .

Indeed, among the re sid u e s of a reduced system modulo p,

the quadratic re s id u e s are those and only those which are
squares of the numbers (a reduced system of residues)

p - 1 P “ 1
(4)— , . . . , - 2 , - 1 , 1, 2, . . . , — —

i.e . with the numbers of (3). Here the numbers of (3) are in-
p —1
congruent modulo p, s in c e Id = I1 (mod p), 0 < k < 1 ^ ---------
2
it would follow that the congruence xd = P (mod p) is s a tis fie d
by four numbers: x = ~k, k, I among the numbers of (4),
contradicting c.
e. I f a is a quadratic residue modulo p, then

p -i
(5) a 2 = 1 (mod p);

i f a is a quadratic non-residue modulo p, then

p -i
(6) a 2 = - 1 (mod p).

80
 Indeed, by F erm at’s theorem,

p - i P-1
ap - i
1 (mod p); a 2
1 a 1 + 1 0(mod p).

One and only one of the factors of the left side of the la tter
congruence is divisible by p (both factors cannot be divisible
by p, for if they were, then 2 would be divisible by p). T here­
fore one and only one of the congruences (5) and (6) can hold.
But every quadratic residue a s a ti s f i e s the congruence

(7) a = x2 (mod p)

for some x, and therefore also s a tis f ie s the congruence (5),

which can be obtained by raising each sid e of (7) to the power
p - 1
----------. Here the quadratic re s id u e s exhaust all the solutions
2
P ~ 1
of the congruence (5), s in ce it cannot have more than
2
p — 1
solutions beca u se it is a congruence of degree ---------- .
2
Therefore the quadratic non-residues s a tisfy the congruence
( 6 ).

§ 2 . The Legendre Symbol

a. We now consider L eg en d re's sym bol [ — ] (read as:

\ P I
the symbol of a with re sp e c t to p). T h is symbol is defined
for all a which are not divisible by p; it is equal to 1 if a is
a quadratic resid u e, and equal to —1 if a is a quadratic non­
resid u e. The number a is s a id to be the numerator, the num­
ber p the denominator, of the symbol.
b. In view of e, $ 1, it is evident that we have
p-i

81
 c. Here we deduce the most important properties of the
Legendre symbol and in the next paragraph, the properties of
the generalization of th is symbol— J a c o b i’s symbol, which is
useful for the rapid calculation of this symbol, and hence
s o lv es the problem of the p o ssib ility of the congruence

x2 = a (mod p).

d. I f a = at (mod p), then ( — ] = ( —- ] .

\ P ) \ P }
T his property follows from the fact that the numbers of an
equivalence c la s s are all either quadratic re s id u e s or
non-residues.

G* I p ' j = lm
Indeed, 1 = l 2 and hence 1 is a quadratic residue.

■■ (?) - < - » -
T h is property follows from b for a = —1.
p — 1
Since ---------- is even for p of the form 4 m + 1 and odd for
2
p of the form 4m + 3, it follows that - 1 is a quadratic residue
of primes of the form 4m + 1 and a quadratic non-residue of
primes of the form 4m + 3.
/ ab . . . Z\ / a \ / b \ /I
g. ------------- =
7 )(7 ) ~ ( 7 )
Indeed, we have

(mod p),
from which it follows that our assertio n is true. A consequence
of our result is

fab2 a

i.e. we can delete any square factor from the numerator of a

symbol.
h. In order to deduce further properties of L eg e n d re’s
symbol, we first give it another interpretation. Setting
p - 1
Pi = ■, we consider the congruences

a ‘ 1 = (1rl (mod p)

a ‘ 2 = t2r2 (mod p)
( 1)

p - i
a ‘ Pi = < p /Pl (mod p); Pi =

where ( xrx is the absolutely l e a s t residue of ax and rx is its

modulus so that ( x = ±1.
T he numbers a • 1, —a * 1, a * 2, —a • 2, . . . , a • p,, - a ' p,
form a reduced residue system modulo p (c, $5, ch. Ill); their
ab so lu tely le a s t re s id u e s are ju s t e,r,, —t ,r,, (2r2, —e2r2, . . . ,
fPirPi» “ f Pir Pi- T hose which are positive i.e. r „ r2, . . . , rpj,
must coincide with the numbers 1, 2, . . . , p, (b, 54, ch. III).
Multiplying together the congruences (1) and dividing
through by

1 - 2 . . . P , = rxr2 . . . r p^

p-i
we find a 2 s ( x(2 . . . (p (mod p), from which (b) we have

a
( 2)

83
 i. The expression for L eg e n d re’s symbol which we have
found can be put in a more concise form. We have

2 ax ~ax~ ax ~ax " ax ^

— 2 + 2 ■ = 2 + 2- f
L p J L pJ [ P J LpJ I P JJ

which is even or odd according as the le a s t positive residue

of the number ax is le s s or greater than -yp, i.e. according as
( x = 1 or ( x = —1. It is evident from this that

and therefore we find from (2) that

j. Assuming a to be odd, we transform the la tte r equation.

We have (a + p is even)

The formula (3) allow s us to deduce two very important

properties of the Legendre symbol.
84
 T h is follows from formula (3) for a = 1.
Moreover, since

(8m ± l ) 2 - 1
8m2 + 2m, even
8
while

(8m ± 3)2 - 1
----------------------- = 8m2 ± 6m + 1, odd,

it follows that 2 is a quadratic residue of primes of the form

8m ± 1 (8m + 1, 8m + 7) and a quadratic non-residue of
primes of the form 8m ± 3 (8m + 3, 8m + 5).
1. I f p and q are odd prim es, then (the quadratic reciprocity
law )
p—1 q—1
7 P
= (-lP “
P 7

P ~ 1 7 ~ 1
Since is odd only in the ca se in which
2 2
both numbers p and q are of the form 4m + 3 and even if one
of th e se numbers is of the form 4m + 1, the above property
can be formulated as follows:
If both the numbers p and q are of the form 4m + 3, then

R_ P_
P

if one of them is of the form 4m + 1, then

/ 7\ P_ \
/ P
P 7

85
 In order to prove our re s u lts , we note that, in view of k,
formula (3) ta k es on the form

(4)

Setting ---------- = qlf we consider p xq x pairs of numbers

2
which are obtained when the numbers a and y in the expres­
sio n s qx, py run through the system s of values

x — 1, 2, Pi t y — 1, 2, q Xt

independently.
We can never have qx - py, b ec a u se it would follow from
this equation that py is a multiple of q which is im possible
b eca u se (p, q) = (y, q) = 1 (since 0 < y < q). Therefore we
can s e t p ^ j = St + S2, where S, is the number of pairs with
qx < p y and S2 is the number of pairs with py < qx.
It is evident that is also the number of pairs with
P
x < — y. F o r given y we can take x = 1, 2, —y .
9 9
p p p P
(Since — y ^ — q2 < — we have — y ^ Pi*)*
9 9 2 .9
Consequently,

p
— y .
s‘ - L
9

Analogously, we can prove that

9
—x .
X- 1 P
86
But then equation (4) gives

t 7

and hence

\ / \ / = (“ 1)Sl+Sl = (~ 1)Pl91

from which the required property follows.

§ 3 . The Jacobi Symbol

a. In order to evaluate L eg e n d re’s symbol most quickly, we

consider the more general Jacobi sym bol. L e t P be an odd
number greater than unity, and let P = p tp2 . . . p r be its de­
composition into prime factors (some of which may be equal).
Moreover, le t (a, P) = 1. T hen J a c o b i’s symbol is defined
by the equation

a a a

Pi Pi Pr

T he well-known properties of the Legendre symbol allow us

to es ta b lis h analogous properties for the Jacobi symbol.

b. I f a = o, (mod P ), then

Indeed,

a.

Pi Pi
87
so that a, being congruent to a, modulo P , is also congruent to
a1 modulo p t , p2t . . . , p r, which are the divisors of P .

In order to es ta b lis h this, we note that

but

P ~ 1 _ P1P2 - P r ~ 1

2 2

Pr ~ 1
1 + 2A J i l ) i + 2^ ± .1 + 2

Pi “ 1 P i- 1 Pr ~ 1 nM
---------- + ---------- + . . . + ---------- + 2 N

and hence from formula (1) we deduce

2
- » r

88
 ab . . . I
e.

and multiplying the symbols with the sam e numerators, we ob­

tain the required property. From this we obtain the corollary:

8 8 8
89
and hence we deduce from formula (2)

g. I f P and Q are p o sitiv e relatively prime odd numbers, then

Indeed, let Q = q tq2 . . . q s be the decomposition of Q into

prime factors (some of them may be equal). We have

P ar1 q/3-1
y y 2 2 r s
= ( - l ) a“l ^ 77 77
a - i /3«i

r
s 2 P
= (- 1)
a -i /3-i
\.
QJ

But, as in d, we find

f
Q - i £ ^ 3 — L + 2/V
L + 2/V,
2 2

and hence

p -1
2 2
(- 1)

90
 Exam ple, As an example of the calculation of the Legendre
symbol (we will consider it to be a particular c a s e of the
Jacobi symbol) we in v e stig ate the solutions of the congruence

x 2 : 219 (mod 383).

We have (applying in sequence the properties g, b, the corol­

lary of e, g, b, e , f, g, b, d):

and hence the congruence under consideration h a s two

solutions.

§4. The Case o f Composite Moduli

a . Congruences of the second degree with composite moduli

are investigated and solved in accordance with the general
methods of $5. ch. IV.
b. We s ta rt with a congruence of the form

(1) x 2 = o(m od p a); a > 0, (a, p) = 1,

where p is an odd prime.

Setting f{x) = x 2 - a, we have f ' ( x ) = 2x, and if x = x,
(mod p) is a solution of the congruence

( 2) x 2 = a (mod p)
91
then s in c e (a, p) = 1 we also have (x lf p) = 1, and since p is
odd, (2*lt p) = 1, i.e. f ' { x x) is not divisible by p. Therefore
to find the solutions of the congruence (1) we can apply the
argument of b, §5, ch. IV, while each solution of the con­
gruence (2) g iv e s one solution of the congruence (1). It
follows from th is that
The congruence (1) has two solutions or none according as
a is a quadratic residue or a quadratic non-residue modulo p.
c. We now consider the congruence

(3) x 2 = a (mod 2 a); a > 0, (a, 2) = 1.

Here f ' i x j - 2xt is divisible by 2, and hence the argument

of b, 95, ch. IV is inapplicable; it can be changed in the
following way:
d. If the congruence (3) is solvable, then, since (a, 2) = 1,
we have (x, 2) = 1, i.e . x = 1 + 21, where t is an integer.
T he congruence (2) ta k e s on the form

1 + 4t(t + 1) = a (mod 2 a ).

But one of the numbers t, t + 1 is even and hence 4 t(t + 1) is

a multiple of 8. Therefore, for the solvability of the la tter
congruence, and along with it a ls o the congruence (3), it is
n e c e ss a ry that

(4) a = 1 (mod 4) for a = 2; a = 1 (mod 8) for a > 3.

e. In the c a s e s in which condition (4) is s a tisfie d , we con­

s id e r the question of finding so lu tio n s and the number of
solutions.
F or a ^ 3, all the odd numbers s a tis f y the congruence in
view of d. Therefore the congruence x 1 s a (mod 2) h as one
solution: x = 1 (mod 2), the congruence x 1 = a (mod 4) has two
solutions: x = 1; 3 (mod 4), the congruence X1 = a (mod 8) has
four solutions: x = 1, 3, 5, 7 (mod 8).
92
 In order to consider the c a s e s a = 4, 5, . . . all the odd
numbers are put in the two arithmetic progressions:

(5) x = ±(1 + 413)

(1 + 4 13 h 1 (mod 4); - 1 - 4 1, = - 1 = 3 (mod 4))

We now decide which of the la tte r numbers satisfy the con­

gruence x 2 = a (mod 16). We find

a - 1
(1 + 4t,)2 s a (mod 16), t 3 = ----------(mod 2),
8
t, = t 3 + 2t4, x = ±(1 + 4 13 + 8£4) = ± (x4 + 8t4).

We now decide which of the la tte r numbers satisfy the con­

gruence x 1 = a ( m o d 3 2 ) . We find

(x4 + 8 £4)2 s a (mod 32), t4 = t4 + 2t s, x = ±(rcs + 16f5),

etc. In th is way we find that the valu es of x satisfying the

congruence (3) for a > 3, are representable in the form

x = ± { x a + 2a~l t a ).

T h e s e values of x form four different solutions of the con­

gruence (3)

x = x a ; x a + 2a_1; - x a ; - x a - 2a" ‘ (mod 2a )

(modulo 4 the first two are congruent to 1 while the second

two are congruent to —1).
Exam ple. The congruence

(6) x 1 = 57 (mod 64)

has four solutions sin ce 57 s l(m o d 8). R epresenting x in

the form x - ±(1 + 4£,), we find
93
 (1 + 4*3f = 57 (mod 16), 8*3 h 56 (mod 16),

<3 = 1 (mod 2), *3 = 1 + 2*«, x = ±(5 + 8*J,

(5 + 8*4)a s 57 (mod 32), 5 • 16<4 = 32 (mod 32),

t4 = 0 ( m o d 2 ) , *„ = 2*s, x = ±(5 + 16*s),

(5 + 16fs)J = 57 (mod 64), 5 * 32*, = 32 (mod 64),

*5 = l(m o d 2), * j = l + 2*6, x = ±(21 + 32*t ).

Therefore the solutions of the congruence (6) are:

x = ±21; ±53 (mod 64).

f. It follows from c, d, and e that:

The n e c e ssa ry conditions for the so lv a b ility o f the
congruence

x? = a (mod 2 a ); (a, 2) = 1

are: a = l(m o d 4) for a = 2, a = l(m o d 8) for a ^ 3. I f

th e se conditions are s a tis fi e d , then the number o f solutions
is: 1 for a = 1; 2 for a = 2; 4 for a > 3.
g. It follows from b, f and a, §5, ch. IV that:
N ece ssa ry conditions for the so lv a b ility o f congruences o f
the form

x 2 = a (mod m); m = 2 ap f 1p^ 1 . . . p**; (a, m) = 1

are:

a = l(m od 4) for a = 2, a = 1 (mod 8) for a ^ 3,

I f all o f these conditions are s a t i s f i e d , the number o f solutions

is: 2 k for a = 0 and a = 1; 2/t+l for a = 2; 2k+2 for a ^ 3 ,
94
P roblem s for C hapter V

Here p will always denote an odd prime.

1. Prove that finding the solutions of a congruence of the
form

ax2 + bx + c = O(modm), (2a, m) = 1

reduces to finding the solutions of a congruence of the form

x2 = q (mod m).
2, a. Using e, §1, find the s o lu tio n s of the congruence
(when they exist)

x2 = a (mod p); p = 4m + 3.

b. Using b and k, 42, obtain a method of finding the solu­

tions of the congruence

x 2 = a (mod p); p = 8m + 5.

C. Find the sim plest p o s sib le method of finding the solu­

tions of a congruence of the form

x2 = a (mod p); p = 8m + 1

when we know some quadratic non-residue N modulo p.

d. Using Wilson’s theorem, prove th a t the solutions of the
congruence

x 2 + 1 = 0 (mod p); p = 4m + 1

are

x = ± 1 • 2 . . . 2m (mod p).

3, a. Prove that the congruence

(1) x2 + 1 = 0 (mod p)
95
is solvable if and only if p is of the form 4m + 1; the
congruence

(2) x2 + 2 = 0(mod p)

is solvable if and only if p i s of the form 8m + 1 or 8m + 3;

the congruence
X1 + 3 h 0 (mod p)

is solvable if and only if p is of the form 6m + 1.

b. Prove that there are an infinite number of primes of the
form 4m + 1.
c. Prove that there are an infinite number of primes of the
form 6m + 1.
4. Dividing the numbers 1, 2, . . . , p - 1 into two s e ts , the
second of which contains a t le a s t one number, we assum e that
the product of two numbers of the same se t are congruent to a
number of the first se t modulo p, while the product of two
elem ents of different s e t s is congruent to a number of the
second s e t modulo p. Prove that th is occurs if and only if
the first s e t c o n s is ts of quadratic r e s id u e s , while the second
s e t c o n s is ts of quadratic non-residues modulo p.
5. a. Deduce the theory of congruences of the form

x2 s a(m od p a ); (a, p) = 1,

by representing a and x in the calculational system to the

base p .
b. Deduce the theory of congruences of the form

x2 = a (mod 2 a ); (a, 2) = 1,

by representing a and x in the calculational system to the

base 2.
6. Prove that the s o lu tio n s of the congruence

x2 - a (mod p a ); (a, p) = 1
96
are x = ± P Q ' (mod p a ), where

(z + \ f a ) a + (z - V c i ) a (2 + Va~)a - (2 - Va^)a
P = --------------------------------------- , Q = -------------------- --- ---------------
2 2vT

22 = a ( m o d p ) , = 1 (mo'd p a ).

7. F in d a method of solving the congruence x 2 = 1 (mod m)

based on the fact that this congruence is equivalent to the
congruence (x — l ) ( x + 1) = 0 (mod m).

8 . L et ^ j = 0 for (a, p) = p.

a. F o r (k , p) = 1, prove that

p - ‘ ( x(x + h)
L
5=o \
-------
P
-1.

b. L e t each of the numbers e and 7/ have one of the values

±1, let T be the number of pairs x , x + 1, where x = 1, 2,

. . . , p — 2, such that ( — ^ = f, ( --------- -] = 77.

\ P j \ P I
Prove that

T 77-07

c. Let (k, p) = 1, and let

5 =

where x and y run through increasing s eq u en ce s consisting,

resp ectiv ely , of X and Y re s id u e s of a complete system modulo
97
p* Prove that

Is | < Vm>
In the proof use the inequality

2
'xy + k
£ X
L L
P

d. L et Q be an integer, 1 < Q < p.

p- i /x + z
S = £ SJX; S x = £

a) Prove that S = (p — Q)Q.

j8) L et A be a co nstant, 0 < A < 1 . Prove that the number

T of integers x = 0, 1, . . . , p - 1 for which the condition
S x < Q°‘s+0-5'^ is not s a tis f ie d , s a tis f ie s the condition
T i PC - \

y) L e t p > 25, and le t M be an integer. Prove that the

sequence

M, M + 1, . . . , M + 3[Vp"] - 1

contains a quadratic non-residue modulo p.

9, a. Prove that the number of rep resen tatio n s of an integer

m > 1 in the form

(1) m = x? + y2, (x, y) = 1, x > 0, y > 0

is equal to the number of solutions of the congruence

(2) z2 + 1 = 0 (mod m).

98
 In proving th is, s e t r = Vm~ and use the representation of
z
a = — given in the theorem of problem 4, b, ch. I, and then
m
consider the congruence obtained by multiplying (2) termwise
by Q* •
b. L e t a be one of the numbers 2, 3. Prove th a t the number
of rep resen tatio n s of a prime p > a in the form

(3) p = x2 + ay2, x > 0, y > 0

is equal to half the number of so lu tio n s of the congruence

(4) z 2 + a = 0(mod p).

c. L e t p be of the form 4m + 1, (k , p) = 1,

P-- »1 /:xix2
P + k)
S (« - £ -
X«0 \

Prove that (D. S. Gorshkov)

a ) S(k) is an even number.

/3) S(kt) = ( - ~ j S(k).

y) For ( — ] = 1, ( — | = - 1 , we have (cf. problem a.)

PJ \ P }

P = isw )\ (isw)\
10. L e t D be a positive integer which is not the square of
an integer. Prove that:
a. If two pairs x = x x, y = y, and x = x lt y = ya of
integers satisfy the equation

x 2 - Dy2 = k
99
for a given integer k, then the equation

X 2 - D Y 2 = k?

is s a tisfie d by integers X , Y defined by the equation (the ±

sign can be taken arbitrarily)

X + YVZ)~ = (z, + y l V T T )(xi ± y i 'JTT).

b. The equation ( P e l l ’s equation)

(1) x2 - Dy2 = 1

is solvable in po sitiv e integers x , y.

c. If x 0, y 0 i s a pair of positive x , y with minimal x (or,
equivalently, with minimal x + yVO~) satisfy in g equation (1),
then all p airs of positive x , y satisfy in g th is equation are
defined by the equations

(2) x + yVD~ = (x 0 + y 0\ f D ) r; r = 1, 2, . . .

11, a. L e t a be an integer. L et

a) F or (a, p) = 1, prove that | £/a, p | = Vp~ .

In proving this, multiply UB>p by its conjugate, which is
obtained by replacing i by —i. L etting the le tte rs x x and x
be the summation varia b les of the original and conjugate sum s,
we then gather together the terms of the product such that

x t = x t (mod p),

or

x x = x + t (mod p)
100
for fixed t.

j3) Prove that

a \ Ua, p

P / V>,P

b. L e t m > 2, (a, m) = 1,

1771 '
= L
x-0

a ) Prove that S a p = Ua p (problem a).

/9) It follows from the theorems of problems a ) and a, a)
that S a>p = VP*- Pro ve the following more general result:

l S a , m | = Vm", if m = 1 (mod 2),

l^a, m| = 0, if m = 2 (mod 4),
l ^ a, ml = , if m = 0 (m o d 4 ).

y ) L e t m > 1, {2A, m) = 1, and let a be an arbitrary

integer. Prove that

n^ 1 / A x2 + ax
^ exp --------------- = Vm-

12, a. L et m be an integer exceeding 1, let M and Q be

integers such that 0 ^ M < M + Q ^ m, and let denote

a sum extended over the z in a given s e t of integers, while £

r
denotes a sum extended over the z in th is s e t which are con­
gruent modulo m to the numbers
101
 M, M + 1........... M + Q - 1.

Moreover, let the function $ (z) be such that, for some A and
any a = 1, 2, . . . , m - 1, we have

/ az\
£ 0 (z) exp 2/rt---- ^ A.
V m ,

Prove that

£ $ ( 2 ) = — £«&(*) + 0A(ln 771 S ),

9- ^ 5!

where | d\ < 1, S > 0 alw ays, 5 > — for m > 12,

2
S > 1 for m > 60.
b. L e t M and Q be in teg ers such that 0 < M < M + Q ^ p.
a ) Prove that

M+Q-i / x
< Vp~ In p.
L 7
x-M \ P

/3) Let R be the number of quadratic residues and let N be

the number of quadratic non-residues in the sequence M,
M + 1, . . . , M + Q ~ 1. Prove that

1 d r - 1 Q . .
R = ~2® + ~ 2 ^ ln p ’ N = ~2® ~ In P; ^ <

y ) Deduce the formulae of problem /3), using the theorem of

problem 11, b, ^3) and the theorem of problem a.
5) L e t m > 2, ( 2 A , m) =■ 1, and le t M0 and Q0 be integers
such that 0 < M0 < M0 + Q0 ^ m. Prove that
102
 Mo+ Qo-A Ax2
2n i ----- < In m.
L cxp . m
x-A fo

t ) L e t p > 2, (A, p) = 1, le t M0 and Q0 be integers such

that 0 < M0 < M0 + Q0 ^ p and let T be the number of integers
of the sequence A x 2; x = M0, M0 + 1, . .. , M0 + Q0 - 1,
which are congruent modulo p to the numbers of the sequence
M, M + 1, . . . , M + Q - 1. Prove that

T = — + Q \ f ^ (In p)2.
P

c. Deduce the formulae of problem b, £}) by considering the

sum

p-i p-1 Af+Q-i M+0-» / a \ / a^x ~ a y)

EEL
a - 0 dml x-M
E
y-M
- “ p i\ * * —
\ P /
—
r

Numerical E xercises for Chapter V

1, a. Find the quadratic re sid u e s in a reduced residue s y s ­

tem modulo 23.
b. Find the quadratic non-residues in a reduced residue s y s ­
tem modulo 37.
2, a. Applying e, §1, find the number of so lu tio n s of the
congruences:

a ) x 2 = 3 (mod 31); /3) x 2 s 2 (m o d 3 1 ).

b. Find the number of solutions of the congruences:

a ) x2 = 5 (mod 73); /3) x 2 = 3 (mod 73).

3, a. Using the Jacobi symbol, find the number of solutions

of the congruences:
103
 a) y? = 226 (mod 563); ft) x 2 = 429 (mod 563).

b. Find the number of so lu tio n s of the congruences:

a) x 2 = 3766 (mod 5987); ft) x2 = 3149 (mod 5987).

4, a. Applying the methods of problems 2, a; 2, b; 2, c,

solve the congruences:

a) x 2 = 5 (mod 19); ft) x 2 = 5 (mod 29); y ) x2 = 2 (mod 97).

b. Solve the congruences:

a ) x 2 = 2 (mod 311); ft) x 2 - 3 (mod 277);

y ) x 2 = 11 (mod 353).

5, a. Solve the congruence x 2 = 59 (mod 125) by the methods

of:

a) b, 4 4 ; ft) problem 5, a; y ) problem 6.

b. Solve the congruence ac2 s 91 (mod 243).

6, a. Solve the congruence x 2 s 41 (mod 64) by the methods
of:

a) e, 44; ft) problem 5, b.

b. Solve the congruence x 2 s 145 (mod 256).

104
 CHAPTER VI

PRIMITIVE ROOTS AND

INDICES
§ 1 . General Theorems

a. F or (a, m) = 1 there ex ist positive y such th a t a7 = 1

(mod m), for example (by E u le r ’s theorem) y = cp(m). The
s m a lle s t of these is called: the exponent to which a belongs
modulo m.
b. I f a belongs to the exponent 8 modulo m, then the num­
bers 1 = a0, a 1, a^-1 are incongruent modulo m.
Indeed, it would follow from a 1 = a* (mod m), 0 ^ k < I < 8
that a '~ k = 1 (mod m), 0 < I — k < 8, which contradicts the
definition of 5.
c. I f a belongs to the exponent 8 modulo m , then a7 = a ^
(mod m) i f and only i f y = y* (mod 5); in particular (for y ' = 0),
a y = 1 (mod m) i f and only i f y is d iv isib le by 8.
Indeed, le t r and r t be the le a s t non-negative resid u es of
the numbers y and y ' modulo 5; then for some q and q t we
have y = 8q + r, y ' — 8qt + r. From th is and from a = 1
(mod m) it follows that

a ? = (a^)qa r = o ' (mod m),

ay = (a ^)Qla ri o '1 (mod m).

Therefore a ? = a ? 1 (mod m) if and only if ar = a r ‘ (mod m),

i.e. (b), when r = rl .
105
 d. It follows from 0 ^ ”^ = 1 (mod m) and from c { y ' = 0)
that cp(m) is divisible by 8. Thus the exponents to which
numbers belong modulo m are ju s t the divisors o f cp(m). The
la rg e st of th e se divisors is cp(m). The numbers belonging to
the exponent cp(m) (if such ex ist) are called the primitive
roots modulo m.

§2. Primitive R oots Modulo p a and 2 p a

a. L e t p be an odd prime and le t a > 1. We s h all prove

the existence of primitive roots modulo p a and 2pa .
b. I f x belongs to the exponent ab modulo m, then x a be­
longs to the exponent b.
Indeed, let x B belong to the exponent 8. Then x a ^ = 1
(mod m), and hence (c, §1) a8 is divisible by ab, i.e. 8 is
divisible by b. On the other hand, (x a)b = 1 (mod m) implies
(c, § l ) that b is divisible by 8. Hence 8 = b.
c. I f x belongs to the exponent a, and y belongs to the e x ­
ponent b modulo m, where (a, b) = 1, then x y belongs to the
exponent ab.
Indeed, le t x y belong to the exponent 8. Then (xy) = 1
(mod m). Hence x bSy bS = 1 (mod m) and (c, 51) x bS = 1
(mod m). Hence (c, §1) b8 is d iv isib le by o, and since
(b, a) = 1, 8 is divisible by o. In the same way we find
that 8 is divisible by b. Since (a, b) = 1, being divisible by
a and b, 8 is also divisible by ab. On the other hand,
(x y )eb = 1 (mod m) im plies (c, 51) that ab is divisible by 8.
Hence 8 = ab.
d. There e x i s t primitive roots modulo p.
Indeed, le t r be the le a s t common multiple of all those
exponents

(1) Si, Sa, . . . , 8„

to each of which belongs at le a s t one number of the sequence

1, 2, . . . , p — 1 modulo p, and le t r = • • • 9 kk the
canonical decomposition of the number r. Then for each s ,
106
among the numbers (1) there e x is ts some 8 which is divisible
by <7a« and is therefore representable in the form 8 = aqgs .
If x is a number belonging to the exponent 5, then, by b,
x„ = x a belongs to the exponent q t s • T h is holds for
s = 1, 2, . . . , k; by c, the number g = x tx 2 . . . x k belongs
to the exponent q ^ q ^ 1 . . . q%k = r.
But since the exponents (1) are ju s t the divisors of the
number r, all the numbers 1, 2, . . . , p - 1 s a tisfy (c, §1)
the congruence x T = 1 (mod p). T h is means (c, $4, ch. IV)
that p — 1 <C t . But t is a divisor of p - 1. Hence
t = p — 1, i.e . g is a primitive root.

e. L e t g be a primitive root modulo p. We can find a t

such that u, which is defined by the equation (g + pt)p~* =
= 1 + pu, is not d iv isib le by p. The corresponding g + p t
is a primitive root modulo p a for any a > 1.
Indeed, we have

gp“ ‘ = 1 + PT,
(2)
(g + pt)p~l = 1 + p ( T 0 - gp“ 1t +. pT) = 1 + pu,

where, along with t, u runs through a complete residue system

modulo p. Therefore, we can find a t such that u is not di­
v isib le by p. For th is I, we deduce from (2) the equations

(g + p t ) p(p-1) = (1 + p u ) p = 1 + p 2u a ,

(3) < (g + p f ) p2(p" l > = (1 + P2 u 2) p = 1 + pJu „

where u2, u 3, . . . are not divisible by p.

L e t g + pt belong to the exponent 8 modulo p a . Then

(4) (g + p t ) S s 1 (mod p a ).

Hence (g + pt)° e 1 (mod p); and consequently 5 is a

multiple of p — 1, and s in c e 8 divides cp(pa ) = p°~l (p — D,
107
it follows th a t 8 = p^-1 (p — 1), where r is one of the numbers
1 , 2 , . . . , a . R eplacing the left sid e of the congruence (4)
by its expression in the appropriate equation of (2) or (3), we
find (u = uf)

1 + p ru r = l ( m o d p a ), p r = 0 ( m o d p a ), r = a , 8 = cp(pa ),

i.e. g + pi is a primitive root modulo p a .

f. L e t a ^ 1 and let g be a primitive root modulo p a .
Whichever o f the numbers g and g + p a is odd, is a primitive
root modulo 2 p a .
Indeed, every odd x which s a ti s f i e s one of the congruences
x y = 1 (mod p a ) and x ^ = 1 (mod 2 p a ) obviously s a tis f ie s
the other a lso . H ence, s in c e cp(pa ) = cp(2pa ) for all odd x,
a primitive root for one of the moduli p a and 2 p a , is also a
primitive root for the other. But, of the two primitive roots
g and g + p a modulo p a, at le a s t one is odd; and conse­
quently, it will be a primitive root modulo 2 p a .

§3. Evaluation o f the Prim itive Roots

for the Moduli p a and 2 p a

The primitive roots for the moduli p a and 2 p a where p is

an odd prime and a ^ 1, can be found by using the following
general theorem:
L e t c = cp(m) and let q lt q2, . . . , qk be the different prime
divisors o f the number c. In order that a number g , which is
relatively prime to m, be a primitive root modulo m, it is
n e c e ssa ry and s u ffic ie n t that this g s a t i s f y none o f the
congruences

c c
g Qi = 1 (mod m), g q* = 1 (mod m),

(1)
. . . , g qk S 1 (mod m).

108
 Indeed, if g is a primitive root, then a fortiori it belongs to
the exponent c and hence none of the congruences o f (1) can
be s a tisfie d .
C onversely, we now assum e that g s a tis f ie s none of the
congruences of (1). If the exponent 8 to which g belongs,
turns out to be l e s s than c, then, letting q be one of the prime
c c c —
divisors of — , we would have — = qu, — = 8u, gq= 1
8 8 q
(mod p), which contradicts our assum ption. Hence 8 = c
and g is a primitive root.
E xam ple 1. L e t m = 41. We have cp(41) = 40 = 21 *5,
40 40
---- = 8, ----- = 20. Therefore, in order that the number g,
5 2
not divisible by 41, be a primitive root modulo 41, it is
n e c e ss a ry and sufficient that this g s a tisfy neither of the
congruences

(2) g' = 1 (mod 41), g10 = 1 (mod 41).

But going through the numbers 2, 3, 4, . . . we find (modulo 41)

2s = 10, 3* = 1, 4‘ = 18, 5' = 18, 6s h 10,

2“ = 1, 410 s 1, 5J0 = 1, 6” = 40.

From th is we s e e that the numbers 2, 3, 4, 5 are not primitive

roots since each of them s a ti s f i e s at le a s t one of the con­
gruences (2). The number 6 is a primitive root s in c e it s a t i s ­
fies neither of the congruences of (2).
Exam ple 2. L e t m = 1681 = 4 12. A primitive root can also
be obtained here by using the general theorem. But we can
find it more simply by applying theorem e, 42. Knowing
(example 1) that 6 is a primitive root modulo 41, we find

640 = 1 + 41(3 + 41/)

(6 + 41t)40 = 1 + 41(3 + 41/ - 6Mt + 41T) = 1 + 41u.

109
In order that u be non-divisible by 41, it is sufficient to take
t = 0. We can therefore take the number 6 + 41 • 0 = 6 as a
primitive root modulo 1681.
Example 3. L e t m = 3362 = 2 1681. The primitive root
can also be-obtained here by using the general theorem. But
we can find it more simply by applying theorem f, 42. Know­
ing (example 2) that 6 is a primitive root modulo 1681, we can
take as a primitive root modulo 3362 the odd number in the
pair 6, 6 + 1681, i.e . the number 1687.

§ 4. Indices for the Moduli p a and 2 p a

a. L e t p be an odd prime, a > 1 ; le t m be one of the num­

bers p a and 2 p a ; c = cp(m), and le t g be a primitive root
modulo m.
b. I f y runs through the le a st non-negative residues
y = 0, 1, . . . , c — 1 modulo c, then g * runs through a
reduced residue s y s te m modulo m.
Indeed, g * runs through c numbers which are relatively
prime to m, and by b, 41, incongruent modulo m.
c. F or numbers a, which are relatively prime to m, we
introduce the concept of index, which is analogous to the
concept of logarithm; here, a primitive root plays a role
analogous to the role of the base of a logarithm:
If
a = g* (mod m)

(we assum e that y ^ 0), then y is said to be the index o f the

number a modulo m to the base g and is denoted by the symbol
y = ind a (more precisely: y = inds a).
In view of b, every a, relatively prime to m, h as some
unique index y ' among the numbers of the sequence

y = 0, 1, . . . , c — 1.

Knowing y ' , we can find all the indices of the number a;

by c, 41, these are all the non-negative numbers of the c la ss
no
 y = y ' (mod c).

It follows immediately from the definition of the index which

we have given here that the numbers with a given index y form
an equivalence c la s s of numbers modulo m.

d. We have

ind ab . . . I = ind a + ind b + . . . + ind I (mod c)

and in particular.

ind an = n ind a (mod c).

Indeed,

a = g lnd a (mod m), b = g lnd 6 (mod m),

. . . , I s g lnd 1 (mod m),

and multiplying the la tte r together, we find

ab . . . I = g lnd a +lnd b+ • ‘ *+lnd ' (mod m).

T herefore, ind a + ind b + . . . + ind I is one of the

indices of the product ab . . . I.

e. In view of the practical u se of in d ices, for each prime

modulus p (which is not too large) ta b le s of indices have
been constructed. There are two ta b le s : one for finding the
index from the number, and the other for finding the number
from the index. The ta b le s contain the le a s t non-negative
residues of the numbers (a reduced residue system ) and their
s m a lle s t indices (a complete system) corresponding to a
modulus p and c = cp(p) = p — 1.
Exam ple. We construct the preceding table for the modulus
p = 41. It was shown above (example 1, $3) that g = 6 is a
primitive root modulo 41; we take it ns the b a sis of the
111
in d ic e s. We find (congruences are taken modulo 41):

6° = 1 6s = 10 616 = 18 624 = 16 632 = 37

6l = 6 6* = 19 617 = 26 62S = 14 633 17
62 = 36 610 = 32 618 = 33 62# = 2 6*4 20
63 = 11 6U = 28 619 = 34 627 = 12 635 - 38
64 = 25 612 = 4 620 = 40 62# = 31 63# = 23
6s = 27 613 = 24 621 = 35 629 = 22 637 = 15
66 = 39 6M = 21 622 = 5 630 = 9 63s 8
67 = 29 6,s = 3 623 - 30 631 = 13 639 = 7

and hence our tables are:

N 0 1 2 3 4 5 6 7 8 9
0 0 26 15 12 22 1 39 38 30
1 8 3 27 31 25 37 24 33 16 9
2 34 14 29 36 13 4 17 5 11 7
3 23 28 10 18 19 21 2 32 35 6
4 20

I 0 '1 2 3 4 5 6 7 8 9
0 1 6 36 11 25 27 39 29 10 19
1 32 28 4 24 21 3 18 26 33 34
2 40 35 5 30 16 14 2 12 31 22
3 9 13 37 17 20 38 23 15 8 7

Here the row number is the first digit and the column number
is the second digit of the number (index). At the place common
to the given row and given column we place the corresponding
index (number).
F or example, we find the ind 25 at the place in the first
table common to the 2-nd row and the 5-th column, i.e.
ind 25 = 4. The number whose index is 33 is found in the
place in the second table common to the 3-rd row and the
3-rd column, i.e. 33 = ind 17.
112
 §5. Consequences o f the Preceding Theory

a. L e t p be an odd prime; a ^ 1, le t m be one of the

numbers p a , 2 p a , and finally, le t c = cp(m).
b. Let (n , c) = d; f/terc:
1. TAe congruence

(1) x n = a (mod m)

is solvable ( and hence a is an n-th power residue modulo m)

i f and only i f ind a is a multiple o f d.
In the case o f solva bi lit y the congruence has d solutions.
2. The number o f n-th power residues in a reduced residue
c
s y s te m modulo m is — .

Indeed, the congruence (1) is equivalent to the congruence

(2) n ind x s ind a (mod c)

which is solvable if and only if ind a is a multiple of d

(d, §2, ch. IV).
If the congruence (2) is solvable, we find d values of
ind x which are incongruent modulo c; corresponding to them
we find d values of x which are incongruent modulo m, proving
a ssertio n 1.
Among the numbers 0, 1, . . . , c — 1, which are the sm allest
indices of a reduced residue system modulo m, there are
c
— which are multiples of d, proving assertio n 2.
d
Example 1. For the congruence

(3) x* = 23 (mod 41)

we have (8, 40) = 8, while ind 23 = 36 is not divisible by 8.

Therefore the congruence (3) is unsolvable.
113
 Example 2. For the congruence

(4) xi2 = 37 (mod 41)

we have (12, 40) = 4, while ind 37 = 32 is divisible by 4.

Therefore the congruence (4) is solvable and has 4 solutions.
T h ese solutions are obtained in the following way:
The congruence (4) is equivalent to the following ones:

12 ind x = 32 (mod 40), ind x = 6 (mod 10).

Hence we find 4 v a lu e s of ind x which are incongruent

modulo 40:

ind x = 6, 16, 26, 36,

from which we obtain the 4 solutions of the congruence (4)

x = 39, 18, 2, 23 (mod 41).

Example 3. The numbers

(5) 1, 4, 10, 16, 18, 23, 25, 31, 37, 40

whose in d ic e s are m ultiples of 4, are ju s t all the biquadratic

re sid u e s (or the re s id u e s of any power n = 12, 28, 36, . . . ,
where (n , 40) = 4), among the le a s t positive re sid u e s modulo
40
41. T he number of integers in the sequence (5) is 10 = —— .

c. Along with assertio n b, 1, we shall also find the follow­

ing one useful:
The number a is an n-th power residue modulo m i f and
only i f

c
~3~
6
( ) a = 1 (mod m).
114
 Indeed, the condition ind a = 0 (mod <f) is equivalent to'the
c
condition: — ind a = 0 (mod c). T h e la tte r is equivalent to
d
condition (6).
Example. By the theorem of $3, the im possibility of the
C
congruence g q s i (mod m) is equivalent to the statem ent
that g is a ^-th power non-residue modulo m. In particular,
c
the im possibility of the congruence g 2 = 1 (mod m) is
equivalent to the s tatem en t that g is a quadratic non-residue
modulo m (cf. e , $1, ch. V).
d, 1. The exponent 8 to which a belongs modulo m is de-
c
fined by the equation find a, c) = — ; in particular, the fact
8
that a belongs to a number o f primitive roots modulo m is
equivalent to the equation (ind a, c) = 1.
2. In a reduced residue s y s t e m modulo m, the number o f
numbers belonging to the exponent 8 is cp(S); in particular,
the number o f primitive roots is cp(c).
Indeed, 8 is the s m a lle s t divisor of c such that o S s 1
(mod m). T h is condition is equivalent to

8 ind a = 0 (mod c),

or

ind a = 0

This means that 8 is the sm allest divisor of c for which

c ^
— divides ind a, from which it follows that — is the largest
8 8

divisor of c which divides ind a, i.e. — (ind a, c), proving

8
a sse rtio n 1.
115
 Among the numbers 0, 1, . . . , c - 1, which are the s m a lle s t
indices of a reduced residue system modulo m, the m ultiples
c c
of — are the numbers of the form — y , where y = 0, 1,
8 8
c
. . . , 8 — 1. The condition c — is equivalent to
5
the condition (y, 8) = 1; and the la tte r is s a tisfie d by cp(5)
values of y, proving as s e rtio n 2.
Example 1. In a reduced resid u e system modulo 41, the
numbers belonging to the exponent 10 are the numbers a such
40 = 4, i.e . the numbers
that (ind a, 40) = -----

4, 23, 25, 31.

The number of these numbers is 4 = cp(10).

Example 2. In a reduced residue system modulo 41, the
primitive roots are the numbers a such that (ind o, 40) = 1,
i.e. the numbers

6, 7, 11, 12, 13, 15, 17, 19, 22, 24, 26, 28, 29, 30, 34, 35.

The number of th e se primitive roots is 16 = cp(40).

§ 6 . Indices Modulo 2a

a. The preceding theory is replaced, for the modulus 2a , by

a somewhat more com plicated one.
b. L et a = 1. T hen 2 a = 2. We have cp(2) = 1. A primi­
tive root modulo 2 is , for example, 1 = —1 (mod 2). The
number 1° = (—1)° = 1 forms a reduced residue system modulo
2.
c. L e t a = 2. Then 2s = 4. We have cp(4) = 2. The
number 3 = - 1 (mod 4) is a primitive root modulo 4. The
numbers (—1)° = 1, (—l ) 1 = 3 (mod 4) form a reduced residue
system modulo 4.
116
 d. L e t a ^ 3. Then 2 a ^ 8. We have c p ( 2 a ) = 2 a _ I . It is
easy to s e e that there are no primitive roots in th is case; more
precisely: the exponent to which the odd number x belongs
1
modulo 2a does not exceed 2 ar~2 cp(2a ). Indeed, we have
2

x2 = 1 + 8tlf

x* = 1 + 16t2,

C
L—2
X2 = 1 + 2at a_, = 1 (mod 2 a ).

Therefore, there exist numbers belonging to the exponent

2ar~2. F o r example, 5 would be such a number. Indeed,

5 = 1 + 4,

52 = 1 + 8 + 16,

54 = 1 + 16 + 32uj,

52a ""> = 1 + 2 ^ 2 + 2 au a_3,

from which we s e e that none of the powers 5‘ , 52, 54, . . . , 52

is congruent to 1 modulo 2a .
It is not difficult to s e e that the numbers of the following
two rows:

form a reduced residue system modulo 2 a . Indeed, the number

of th e se numbers is 2 * 2a-2 = cp(2a ); the numbers of each
individual row are incongruent among them selves modulo 2 a
(b, $1); finally, the numbers of the upper row are incongruent

117
to the numbers of the lower row since the former are congruent
to 1, while the la tte r are congruent to -1 modulo 4.
e. F or convenience in la te r in v e stig a tio n s , we ex p ress the
resu lts of b, c, d in more unified form, which is also applicable
in the ca se a = 0.
Let

c = 1; c0 = 1, i f a = 0, or a = 1;

c = 2; c0 = 2 a~2, i f a ^ 2

(therefore c c 0 = cp(2a )) and let y and y 0 run independently

through the lea st non-negative residues

y : 0, •« •, c 1, y0 = 0, •« •, c 0 1

modulo c and c 0. Then runs through a reduced

residue s y s te m modulo 2 a .
f. The congruence

(1) ( - l ) r 5 y °' = ( - l ) y V ° (mod 2a )

holds i f and only i f

y = y ' (mod c), y0 = y ' (mod c 0).

Indeed, the theorem is evident for a = 0. We therefore

assum e that a > 0. L e t the le a s t non-negative re sid u e s of
the numbers y and y0 be r and r0, and of the numbers y ' a n d
yg be r ' and r£ modulo c and c0. In view of c, $1 (—1 belongs
to the exponent c, while 5 belongs to the exponent c0), the
congruence (1) holds if and only if ( l ) r5 r° = (—l ) r 5 r° (mod
2 a ), i.e. if and only if r = r ' t r0 = r£ (in view of e).
e . If

a = (- 1 ) ^ 5 ^ ° (mod 2a ),
118
then the system y, y0 is called an index s y s t e m o f the number
a modulo 2a .
In view of e, every a relatively prime to 2 a (i.e. every odd
a) has a unique index system y ' i y 0' in the c c 0 = cp(2a ) pairs
of values y, y0 considered in e.
Knowing a system y ' , y 0', we can also find all index s y s ­
tems of the number a; according to f, th e se will be all pairs
y, y0 co n s is tin g of the non-negative numbers of the equiva­
lence c l a s s e s

Y = y ' (mod c), y0 = y 0' (mod c 0).

It foil ows immediately from the definition we have given of

index sy stem s that the numbers with a given index system
y, y0 forms an equivalence c l a s s of numbers modulo 2 a .
h. The indices o f a product are congruent modulo c and c0
with the sums o f the indices o f the factors.
Indeed, let y(a), y0(a); . . . ; y(l), y 0(l) be index system s for
the numbers a , We have

a , , , l = ( - l ) r ( a ) + • • *+ y ( 0 5 7 o(a)+ • •. + 7o(0

Therefore y(a) + . . . + y(l), y0(a) + . . . + yo(0 are the

indices of the product a . . . I.

§7. Indices for Arbitrary Composite Modulus

a. L e t m = 2 a p ^ 1p f J . . . p£k be the canonical decomposi­

tion of the number m. Moreover le t c and c 0 have the values
considered in e, $6; c s = ^(ps®); g s be the s m allest
primitive root modulo p f s .
b. If

a = ( - l ) r 5^<> (mod 2a ),
(1)
a = g f 1 (mod p,a»), . . . , a s g]fk (mod p j fc),

119
then y, y0, y lt . . . , y k is called an index s y s te m o f the number
a modulo m.
It follows from this definition that y, y 0 is an index system
of the number a modulo 2 a , w hile y lt .. . , y k are indices of the
number a for the moduli p “l, . . . , pk k . Hence (g, $6; c, $4)
every a which is relatively prime to m (and hence also rela­
tively prime to all the numbers 2 a , p^1, . . . , p kk ), has a unique
index system y y 0', y f , . . . , y'k in the cc fic l . . . c k = cp(m)
sy stem s which are obtained by letting y, y0, y w . . . , y k run
independently through the l e a s t non-negative resid u es for the
moduli c, c0, c lt . . . , c k , while all the index sy stem s of the
number a are ju s t all the system s y, y0, y lf . . . , y k consisting
of the non-negative numbers of the equivalence c l a s s e s

y = y ' (mod c ), y0 = y 0' (mod c0),

y 1 = y [ (mod c j , . . • , y k = y'k (mod c k).

The numbers a with a given index system y, y0, y lf , yk

can be found by solving the system (1), and hence they form
an equivalence c l a s s of numbers modulo m (b, $3, ch. IV).
c. Since the in d ices y, y 0, yt , . • . , y* of the number a
modulo m are the in d ices for the resp ectiv e moduli 2 a , p^1,
. . . , p k k , we have the theorem:
The indices o f a product are congruent modulo c, c0, . . . , c k
to the sums o f the indices o f the factors.

d. L e t r = cp(2a ) for a < 2 and r = — cp(2a ) for a > 2

2
and le t h be the le a s t common multiple of the numbers r, clt
.. . , c k. For every a which is relatively prime to m the con­
gruence a h = 1 holds for all the moduli 2 a , p ^ 1, . . . , p kk ,
which means that this congruence a lso holds for the modulus
m. Hence a cannot be a primitive root modulo m in those
c a s e s in which h < cp(m). But the la tte r holds for a > 2,
for k > 1, and for a = 2, k = 1. Hence for m > 1, primitive
roots can only e x is t if m = 2, 4, p f l , 2 p B u t the ex istence
of primitive roots in th e se c a s e s was proven above (§6, $2).
120
Hence
A l l the c a s e s in which primitive roots modulo m, exceeding
I , e x i s t are j u s t the c as es in which

m = 2, 4, p a , 2 p a .

Problems for Chapter VI

The le tte r p alw ays denotes an odd prime, except in problem

I I , b where we a lso allow the value 2.
1, a. L e t a be an integer, a > 1. Prove the odd prime
divisors of the number op — 1 divide a ~ 1 or are of the
form 2 p x + 1.
b. L e t a be an integer, a > 1. Prove that the odd prime
divisors of the number ap + 1 divide a + 1 or are of the form
2 p x + 1.
c. Prove that there are an infinite number of primes of the
form 2 p x + 1.
d. L e t n be a positive integer. Prove that the prime divisors
of the number 21 + 1 are of the form 2n+1x + 1.
2, L e t a be an integer, a > 1, and le t n be a positive
integer. Prove that cp(an — 1) is a multiple of n.
3, a. L e t n be an integer, n > 1. Starting from the sequence
1 , 2 , . . . , n we form, for odd n , the permutations

1, 3, 5, . . . , n ~ 2, n, n 1, n 3, . . . , 4, 2,

1, 5, 9, . . . , 7, 3

e tc ., while for even n we form the permutations

1, 3, 3, . . . , n 1, n , n “ 2, . . . , 4, 2,

1, 5, 9, . . . , 7, 3,

etc. Prove that the /c-th operation gives the original sequence
if and only if 2k = ±1 (mod 2n - 1).
b. L e t n be an integer, n > 1, and le t m be an integer,
m > 1. We consider the numbers 1, 2, . . . . n in direct order
121
from 1 to n, then in reverse order from n to 2, then in direct
order from 1 to n , then in reverse order from n to 2, etc. From
this sequence we take the 1 -st, (m + l) - s t , (2m + l) - s t , etc.,
until we obtain n numbers. We repeat the same operation with
this new sequence of n numbers, etc. Prove that the A-th
operation gives the original sequence if and only if

mk = +1 (mod 2n — l)

4. Prove that there e x ist cp(5) numbers belonging to the

index 5, by considering the congruence x = 1 (mod p)
(problem 10, c, ch. IV) and applying d, $3, ch. II.
5, a. Prove that 3 is a primitive root of any prime of the
form 2" + 1, n > 1.
b. Prove that 2 is a primitive root of any prime of the form
2p + 1 if p is of the form 4zi + 1, while - 2 is a primitive root
of any prime of the form 2p + 1 if p is of the form 4n + 3.
c. Prove that 2 is a primitive root of any prime of the form
4p + 1.
d. Prove that 3 is a primitive root of any prime of the form

2np + 1 for n > 1 and p >

6, a, a ) L e t n be a positive integer and let S = l n + 2n +

+ . . . + (p - l ) n . Prove that

S = - 1 (mod p), if n is a multiple of p — 1,

S = 0 (mod p), otherw ise.

/9) U sing the notation of problem 9, c , ch. V, prove that

 b. Prove Wilson’s theorem by applying b, $4.
7. L e t g and g, be primitive roots modulo p, and let
a indg g. = 1 (mod p - 1).
a. L et (a, p) = 1. Prove that

i n d ^ a = a indg a (mod p — 1).

b. L e t re be a divisor of p - 1, 1 < re < p - 1. The num­

bers relatively prime to p can be divided into re s e t s by putting
those numbers such that ind a = s (mod re) in the s-th s et
(s = 0, 1, . . . , re — 1). Prove that that the s-th s e t for the
b a s e g is identical with the s t-th s e t for the b a s e glt where
s, = a s (mod re).
8. F in d the sim plest p o s sib le method of solving the con­
gruence x n = a (mod p) (convenient for (re, p — 1) not too
large) when we know some primitive root g modulo p.
9. L e t m, a, c, c 0, c lt . c k , y , y 0, ................... have the
v alu es considered in $7. C onsidering any roots R , R ot R lf
. . . , R k of the equations

R c = 1, R 0c o - 1, R f i = 1, R%* - 1,

we s e t

X (a) - R y R?<>R'?' . . . R y *.

If (a, m) > 1, then we se t y (a ) * 0.

A function defined in this way for all integers a is said to
be a character. If R = R 0 = R x = . . . = R k = 1, then we say
that the character is principal; it h as the value 1 for (a, m) =
= 1, and the value 0 for (a, m) > 1.
a. Prove that we obtain cp(m) different characters in this
way (two characters are s a id to be different if they are not
equal for at le a s t one value of a).
b. Deduce the following properties of characters:

a ) y ( l ) = 1,
123
 /3) x ( a iaj) = x ( ai)x K )»

y) x K ) = a i - ° j (m0(i m)*

C. Prove that

cp(m), for the principal character,

z><«>
a-0 0, for other ch ara cters.

d. Prove that, for given a , summing all cp(m) characters,

we find

cp(m), if a = 1 (mod m)

X 0, otherw ise.

e . By considering the sum

X(a)
H - L L
'A(a)
X a

where a runs through a reduced residue system modulo m,

prove that a function ift(a) defined for all integers a and
satisfy in g the conditions

tfj{a) = 0, if (a, m) = 1,

is not identically equal to 0,

^ ( a , a 2) =

^ ( a j) * i/>(a2), if Oj = a2 (mod m).

is a character.
f. Prove the following theorems.
a ) If Xi(°) an(i Xa^a ) are chara cters, then X i ^ X a ^ 0) i s
also a character.
124
 /9) If x ^ a ) is a ch ara cter and \ ( a ) runs through all the
ch ara cters, then X i(o )y (a) a^so runs through all the characters.
y ) F or (I , m) = 1, we have

cp(m), if a = I (mod m)

0, otherw ise.

10, a. L et n be a divisor of p' - 1, 1 < n ^ p — 1, and

let I be an integer which is not divisible by n. The number
f
Rl = e " is a root of the equation R " = 1, and hence the
t lnd x
power e n , which is assum ed to be equal to 0 for x
a m ultiple of p, is a character modulo p.
a) F or (h, p) = 1, prove that

p-1 / I ind {x + k) — I ind x \

S , exp [ 2 n i — ^— ) ■ - 1-

ft) L et Q be an integer, 1 < Q < p, and let

p-i , 2 C -1 / I ind (x + z) \
S- £ |S,,n„l ; S ,,„ „ - £ « p N ------- --------- )
X -0 z-0 ' '

Prove that S = (p — Q)Q.

y ) L e t M be an integer, p > 4n1, n > 2. Prove that the
sequence M, M + 1, . . . , M + 2[nVp~] - 1 c o n tain s a number
of the s-th s e t of problem 7, b.

/ p - I \ 1
b. L e t p > 4 I ---------------- 22k, le t k be the number of
p U ( p - D /
different prime divisors of p — 1, and let M be an integer.
125
Prove that the sequence M, M + 1, . . . , M +
1
+ 2 2k\ U - 1 contains a primitive root modulo
cp(p - 1)

11, a. L et a be an integer, let n be a divisor of p — 1,

1 < n ^ p — 1, and le t A: be an integer which is not divisible
by n.

p- 1 / k ind x \ ( a x
JJa p = Y 1 exp 277-t----------- exp 2 n i-----
’ hi \ n \ P

a) For (a, p) = 1, prove that K . p I - VF-

/3) Prove that

—k ind a u a, p
exp 2ni-
"I,

y ) L e t p be of the form 4m + 1, and let

p-j / ind (x1 + x )'

S = e x P (277-r-------------------

Prove that (cf. problems 9, a and 9, c, ch. V) p = /42 + fi2,

where A and B are integers defined by the equation
S = A + Bi.
b. L et n be an integer, n > 2, m > 1, (o, m) = 1,

Sa , m - T m exP “ £ exP ( 2™“ ~ )

where x runs through a complete residue system , while £

runs through a reduced resid u e system modulo m (cf. problem
12, d, ch. ID and problem 11, b, ch. V).
a ) L e t 8 = (n, p - 1). Prove that

| S a > p | < (5 - 1 ) V F .
126
 j8) L e t (n , p) = 1 and le t s be an integer, 1 < s ^ n. Prove
that

s» I P 6 =P
«-i
, S a,
' p = 0.

y ) L e t s be an integer, s > n. Prove that

s a, p a pn-
■
lS a
*Pl
S'
p
= 0.

S ) Prove that

I S a , m I < Cm1 - " ,

where C only depends on n.

12. L e t M and Q be integers such that 0 < M < M + Q ^ . p .
a. L e t n be a divisor ofp — 1, 1 < n < p — 1, and le t k be
an integer which is not divisible by n. Prove that

A f+ < ?-i / Ji i n d x
exp I27ri----------- < \fp In p.

b. L e t T be the number of integers of the s-th s e t of

problem 7, b, contained in the s e t of numbers M, M + 1, ,
. . . , M + Q — 1. Prove that

T =* — + 0V jT In p; |(?| < 1 .
n

c. L e t k be the number of prime divisors of p — 1, and let

// be the number of primitive roots modulo p in the s e t of
numbers M, M + 1, . . . , M + Q — 1. Prove that

cp(p - 1)
// = Q + 92k \Tp In p; \ 0 \ < 1.
P ~ 1

127
 d. L e t Mx and (?,be integers, 0 ^ Mx < Mx + Qx ^ p ~ 1,
and le t / be the number of integers of the sequence ind M,
ind (M + 1), . . . , ind (M + Q - 1) in the sequence Mit Mx + 1,
. . . , Mx + Q - 1. Prove that

J = + 0VJT ( l n p ) a ; \ d \ <1.
P “ 1

13. Prove that there e x is ts a constant p0 such that: if

p > p0, n is a divisor of p — 1, 1 < n < p — 1, then the
s m a lle s t of the positive non-residues of degree n modulo p is

— 2 / 1 \
< h; h = p c (In p) ; c = 2exp I 1 - — 1 .

14, a. L e t m > 1, (a, m) = 1,

m -i m -1 / a ry \
s = r E v(x)p(y)exp ( 2^ t ---------- ) ;
x - 0 y - 0 \ m I

m —1 m ^i
£ kwl - X, £ Ip<y) r - r.
x - 0 y - 0

Prove th a t | S | ^ V X Y m .
b, a ) L e t m > 1, (a, m) = 1, le t n be a positive integer,
le t K be the number of solutions of the congruence x n = 1
(mod m), rind let
n
1 7 ,-1 1| . flX

S= £ *(*)« m *
X-l

Prove th a t | S | ^ KVtn .
/3) L e t f be an arbitrary positive constant. F or constant n
prove that K = 0(m c) where X is the number considered in
problem a).
128
 15, a. L e t (a, p) = (6, p) = 1 and let n be an integer,
I « | = n lt 0 < n l < p.

p -i / a x n + bx
S = 2^ exP ( 2m
Xml \

Prove that

b. L et 04, p) = 1, let n be an integer, | n | = n lf 0 < n t < p,

and let M0 and Q0 be integers such that 0 < M0 < M0 + Q0 < p.
a) L et

Mo+Qo—1 / Axn
S = 2^ ex P ------
x-M0 \ P

3 JL —
Prove that | S | < — n \ p * In P-
2
fi ) L e t M and Q be in teg ers such that 0 < M < M + Q JC p,
and le t T be the number of in teg ers of the sequence A x n ;
x = M0, Ma + 1, . . . , M0 + Qc - 1, congruent to numbers of
the sequence M, M + 1, M + — 1 modulo p.
Prove that

Q.Q 3 i 1 . ..
T = — + 0 — n \ p * (In p)1; \ o\ < 1.
p 2

c. L e t b and c be integers, (a, p) = 1, (62 - 4 a c , p) = 1.

a ) L e t y be an integer,

ax2 + bx + c \ j yx
--------------------- exp 2m'----
s = L
P / \ P
129
 3 1
Prove that I S I < — p 4 .
2
/3) Let M and Q be integers such that 0 < M < M + Q < p ,
and let

M+Q-i / ax* + bx + c
s L ---------------------
x-M \ P

Prove that | S | P-

Numerical E xe rc is es for Chapter VI

1, a. Find (in the sim p lest p o ssib le way) the exponent to

which 7 belongs modulo 43.
b. F ind the exponent to which 5 belongs modulo 108.
2, a. Find the primitive roots modulo 17, 289, 578.
b. Find the primitive roots modulo 41, 1681, 3362.
c. Find the sm a lle st primitive roots modulo:

a) 1682; /S) 3362.

3, a. Form the table of indices modulo 17.

b. Form the table of indices modulo 41.
4, a. F ind a primitive root modulo 71, using the method of
the example of c, $5.
b. F in d a primitive root modulo 191.
5, a. U sing the table of in d ices find the number of solutions
of the congruences:

a ) x60 ee 79 (mod 97); /3) x5S = 17 (mod 97);

y) x15 s 46 (mod 97).

b. F in d the number of solutions of the congruences:

130
 a ) 3x“ = 31 (mod 41); ft) 7x7 = 11 (mod 41);

y) 5X30 = 37 (mod 41).

6, a. U sing the table of in d ic e s, solve the congruences:

a ) x2 = 59 (mod 67); ft) x3s = 17 (mod 67);

y) x10 h 14 (mod 67).

b. Solve the congruences:

a ) 23xs = 15 (mod 73); /S) 37a:4 = 69 (mod 73);

y) 44a:21 - 53 (mod 73).

7, a. U sing the theorem of c, §5, determine the number of

solutions of the congruences:

a) x3 = 2 (mod 37); ft) x14 = 10 (mod 37).

b. Determine the number of solutions of the congruences:

a ) x s = 3 (mod 71); ft) a^1 = 5 (mod 71).

8, a. Applying the methods of problem 8, solve the con­

gruences (in the solution of the second congruence use the
table of primitive roots at the end of the book):

a) x7 37 (mod 101); ft) x? = 44 (mod 101).

b. Solve the congruence

x3 = 23 (mod 109).

9, a. U sing the table of in d ices, in a reduced residue s y s ­

tem modulo 19 find: a ) the quadratic re sid u e s; ft) the cubic
resid u es.
131
 b. In a reduced resid u e system modulo 37, find: a ) the
resid u es of degree 15; /3) the re sid u e s of degree 8.
10, a. In a reduced residue system modulo 43, find: a) the
numbers belonging to the exponent 6; y3) the primitive roots.
b. In a reduced residue system modulo 61, find: a) the
numbers belonging to the exponent 10; /3) the primitive roots.

132
 SOLUTIONS OF THE
PROBLEMS
Solutions o f the Problems for Chapter /.

1. The remainder resulting from the division of ax + by

by d , being of the form a x ' + b y ' and l e s s than d, must be
equal to zero. Therefore d is a divisor of all numbers of the
form ax + by, and in particular is a common divisor of the
numbers a ■ 1 + b * 0 = a and a • 0 + £> • 1 = 6. On the
other hand, the expression for d shows that every common
divisor of the numbers a and b divides d. Therefore
d = (a, b), and hence theorem 1, d, $2 is valid. The
theorems of e, $2 are deduced a s follows: the sm allest
positive number of the form amx + bmy is amx0 + bmy0m ,
a b
the s m a lle s t positive number of the form — x + — y is
8 8

-*o+ -7o.

T he generalization of these resu lts is trivial.

2. We first note that the difference of two unequal rational
k m 1
fractions — and — (l > 0, n > 0) is numerically > ----- .
I n In

We r e s tr ic t o u rs e lv e s by the assum ption 8 S < S*+i. L et —

b
be an irreducible fraction, which is not equal to 8 e , such that
133
 a
0 < b < Q a . We cannot have 5 S < — < S8+1 ; otherw ise
b
we would have
a 1

a 1
S s+l
T bQ *+1

1
^s+l 5. >

a a
Therefore — < 5 S or S B+1 < — . I n both c a s e s 8 m is closer
b b
a
to a than — .
b
3. F o r n ^ 6 the theorem is evident; we therefore assume
n > 6. We have

„ 1 + V5 ,
£ = ----- = 1.618 . . . ; lo g 10£ * 0.2 . . . ;

& > 1 =* gi = 1

<?, > & + 1 > §2 = 22 > £ ,

Qa > (?, + & > ga = & + Si > € + 1 = f* .

<?„ > < ? „ - ! + C?n —

2> g n - 1 =» gn- 2 + g n -, > T " ’ + f •

Hence

^QgicA
N > £ n“ 2; n < + 2 < 5A; + 2; n ^ 5& + 1.
^ g io ^
134
 0 1
4, a. F or the fractions — and — we have 0 • 1 - 1 • 1 =
1 1
A C
= —1. Between the fractions — and — with AD ~ BC = —1,
B D
A + C
we in se rt the fraction ------------ , and hence A(B + D) ~
B + D
~ B(A + C) = (A + C)D ~ (B + D)C = - 1 . Therefore the
assertio n at the end of the problem is true. T he existen ce of
k a k c
a fraction — such that — < — < — , Z < r is im p o ssib le.
I b i d
Otherwise we would have
k a l c k l c a b + d ^
7 ~ ~b ' lb ’7 ~T * Id ’ d ~ ~b ' Ibd > ~bd
b. It is evident that it is su fficie n t to consider the case in
a c a c
which 0 < a < 1. L e t — < a < — , where — and — are
b d b d
neighboring fractions of the F arey s e rie s corresponding to r .
There are two p o s sib le c a s e s :

a a + c a +c
— ^ a < -----------; ---------- < a < —
b b + d b +d d

We therefore have one of the two in e q u alities

a 1 c
a —— < --------------; a ——
b b(b + d) d d(b + d)

from which the required theorem follows because b + d > t .

c. F o r a irrational, the theorem follows from h, 44, if we
P P „
take for — the convergent — > where QB-x ^ r < Qa .
Q <?.->
135
In the c a s e of rational a = — , the above argument is only
b
valid for b > r . But the theorem is true for b ^ r , s in c e we
a P
can then take the fraction — its e lf for — , s e ttin g 0 = 0 .
b Q 6
5, a. The remainder resu ltin g from the division of an odd
prime by 4 is either 1 or 3. The product of numbers of the
form 4 m + 1 is of the form 4m + 1. Therefore the number
4pj . . . pit — 1, where the p lf . . . , pk are primes of the form
4m + 3, h a s a prime divisor q of the form 4m + 3. Moreover
q is different from the primes pt, . . . , p k .
b. The primes greater than 3 are of the form 6m + 1 or
6m + 5. The number 6pt . . . p k — 1, where the p lt . . . , pk
are primes of the form 6m + 5, h a s a prime divisor q of the
form 6m + 5. Moreover, q is different from the numbers
Pl> • • • » P k ‘
6. L e t p,, . . . , pic be any k primes, and let N be an integer
such that 2 < N, (3 In N) k < N. The number of integers a of
the sequence 1, 2, ... , N, whose canonical decomposition is
of the form a = p * 1 . . . p k k , is

/ In /V
^ ------- + 1 < (3 In N) k < N
\ In 2 J

s in ce a . < -------- . Therefore there are numbers in the

In 2
sequence 1, 2, . . . , /V whose canonical decomposition con­
tains primes different from p lf . . . , p k .
7. F o r example, we obtain such s e q u e n c e s for

M = 2 • 3 • • • (K + l ) t + 2; t = 1, 2, . . .

8. Taking an integer x 0 such that f(x) > 1 and f ' ( x ) > 0

for x ^ x 0, we s e t f( x 0) = X. All the numbers f(x + Xt)\
t = 1 , 2 , . . . are composite (multiples of A).
136
 9, a. If (1) holds, one of the numbers x, y, say x , is even;
it follows from

x z + y 2- y
2

z + y z - y
where, clearly, = 1, that there exist
. 2 2 /
positive integers u and v such that

x z + y
— = uv, ---------- i»2
2 2

T h is im plies the n e c e s s ity of the condition considered in the

problem.
The sufficiency of th e se conditions is evident.
b. In the solution of this problem all le tters denote positive
integers. A ssum e the ex iste n c e of s y ste m s x , y, z such that
x4 + y 4 = z 1, x > 0, y > 0, z > 0, (x, y, z) = 1, and choose
the system with s m a lle s t z. Assuming x to be even we find
x 2 = 2uv, y3 = u1 - v 1, u > v ' ^ 1, (a, v) = 1, where v is
even (for even u we would have y* = 4N + 1, u* = 4N lt
v1 = 4N 2 + 1, 4/V + 1 =■ 4/Vj - 4/Va - 1, which is im possible).
Hence u = z \, v = 2w2, y1 + 4u / = z a, 2u/ = 2ulv l , Uj =
y, = y a, Xj + y\ = z a, which is im possible since z t < z.
It follows from the non-solvability of the equation
x 4 + y 4 = z 1 that the equation x 4 + y 4 = t4 is not solvable
in positive integers x, y, t.

10. Setting x = (k, I) = 1, we find

k n + aikn~l l + . . . + anl n = 0.

Therefore k n is a multiple of I and hence / = 1.

137
 11, a. Let k be the la rg e s t integer such that 2k ^ n and
let P be the product of all the odd numbers which do not
exceed n. T he number 2k~' PS is a sum, all of whose terms,

except 2k~l P—r , are integers.

b. L e t k be the la rg e st integer such that 3k ^ 2n + 1 and

let P be the product of all the integers relatively prime to 6
which do not exceed 2n + 1. The number 3^ PS is a sum,

all of whose terms, except , are integers.

12. F o r n < 8, the theorem is immediately verifiable. It is

therefore s u fficie n t to assum e that the theorem is true for the
binomials a + b, (a + b¥ , . . . , (a + 6)n_1 for n > 8, and
prove that the theorem holds for (a + b)n . But the coefficients
of this binomial, except for the extreme o n es, which are equal
to 1, are ju s t the numbers

n n(n — 1) n(n — 1) . . . 2
1 ’ 1-2 ’ " ' ’ 1 • 2 . .. U - 1)

A n e c e ss a ry and s u fficie n t condition in order that all these

numbers be odd is that the extreme numbers, both equal to n,
be odd, and the numbers obtained by deleting the odd factors
from the numerators and denominators of the remaining numbers
be odd.
But, s e ttin g n = 2nl + 1, th e se numbers can be represented
by the terms of the sequence

nl n l(nl — 1) — 1) . . . 2
l ’ 1*2 1 • 2 . . . (nt - 1) ‘

Since n t < n, the la tte r are all odd if and only if n l is of the
form 2k — 1, i.e . if and only if n is of the form 2(2* — 1) +
+ 1 = 2*+1 - 1.
138
Solutions o f the Problems for Chapter II

1, a. On the ordinate of the point of the curve y = f[x)

with a b s c i s s a x there are [f{x)] la ttic e points of our region.
b. The required equation follows from T t + T 2 = T,
where T lf T it T denote the number of la ttic e points of the
regions

Q P
0 < x < 0 < y < — x,

P Q
0 < y < — , o < x < — y.

Q p
0 < x < — , 0 < y < — .
2 2

c. T he required equation follows from

T = 1 + 4 (7 , + 7 a + T a - TJ,

wh ere T lf T tt 7 „ 7< denote the number of la ttice points of

the regions

x = 0, 0 < y r\

0 < x < -------, 0 < y ^ VrJ - x 1 ;

\f2

0 < y $ — , 0 < * < VrJ - y 3 ;

\f2

0 < x ^ , 0 < y $
\/2 V2 '
139
 d. The required equation follows from T = T l + T 2 — T
where T lf T2, T%denote the number of la ttic e points of the
regions

0 < x ^ Vri , 0 < y ^ — ;

x

r— n
0 < y <L \ n , 0 < x ^ — :
y
0 < x ^ Vrc , 0 < y ^ Vri .

2. The number of positive integers which do not exceed n

is equal to [n]. Each of them is uniquely representable in the
form x k m, where A: is a positive integer; moreover, to a given

S'! n
x there correspond numbers of this form.

3. We prove the n e c e s s ity of our conditions. L e t N be an

integer, /V > 1. The number of values x such that [ax] * N
N
can be represented in the form — + A; 0 < A ^ C, while the
a
number of v alu es y such that [fty] < /V can be represented in
N
the form — + At ; 0 < At ^ Clt where C and C 2 do not de-

N N
pend on N. Dividing — + A + — + At = N by /V, and
a ft
1 1
letting N —* oo, we find — + — = 1. The la tte r equation
a ft
a
for rational a = — (a > b > 0) would give [ab] = [ft(a — 6)].
b

L e t our conditions be s a tisfie d . L e t c be a po sitiv e integer,

c c
and le t x 0 = — + £ and y 0 = — + 77 be the s m a lle s t
a ft
140
 L. L. v
integers such that x 0 > — , y 0 ^ — . Evidently, [ax] < c
a jS
for x < *o and t/3y] < c for y ^ y 0, 0 < £ < 1, 0 < 77 < 1,
a £ and /377 are irrational. Since x0 4- y0 = c + 77 + we
a£ /3?7
have £ 4- 77 = 1, ------ 4- =1;therefore one and only one
a /3
of the numbers a £ and fir] is l e s s than 1. Therefore, one
and only one of the numbers [ a x 0] and [/3y0] is equal to c.
4, a. Our differences are equal to

{ a x j , fot(xj - x t) ] ........... | a ( x f - x t^ ) \ , | - a x t |,

they are non-negative, their sum is equal to 1, there are t + 1

of them; therefore at le a s t one of these differences does not
1 1
exceed --------- < — , and hence there ex ists a number smaller
t 4- 1 t

than — of the form | ± a ( ) ] , where 0 < Q ^ r . Prom

T

± a Q = [±a(?] 4- {±a<?|, s e ttin g ±[±a(}] = P, we find that

o-Q - P | < a - <

Qt
b. Setting X 0 = [A'], T0 = [T], . . . , Z 0 = [Z], we consider
the sequence formed by the numbers {ax 4- fi y 4- . . . + y z ]
and the number 1 arranged in non-decreasing order, assuming
that x, y, . . . , z run through the values:

x = 0, 1, . . . , X 0; y = 0, 1, . . . , T0; 2 = 0, 1, . . . , Z 0.

We obtain (A0 4- 1 ) ( T 0 4- 1) . . . (Z0 + 1) 4- 1 numbers, from

which we obtain (A0 4- 1) ( 7 0 + 1) ••• (^0 + 1) differences.
At le a s t one of th e se differences does not exceed
1 1
<
ix0 4- i)(y0 4- 1) ... (z0 + 1) Ay ... z
141
It is easy to obtain the required theorem from this.
5. We have a = cq + r + f a | ; 0 ^ r < q,

"[a]" r a r + faP
_ _
? + — = R, R +
c c c c
L_ _

6, a. We have [ a + /3 + . . . + A] = [a ] + [/3l + . . . +
+ [A] = [ f a | + |/3 | + . . . + |A |].
b. The prime p divides n\ , a!, . . . , / ! to the exact powers

— —i i— _ _ _ _ _
n n a a
— + + *• • f — +
L p_ Lp j P |_p’J
PI +
r n
___ 1
„
Q.
_P_

Moreover '
n a r n
+ • •• +
|_puJ Lp “J Lp 'J

7. Assuming that there e x ists a number a with the required

properties, we rep re se n t it in the form

Jr 4*1 If 3 /
a = RkP + R k-iP + . . . + R tP 1 + RoP + R 5

^ ^ R k ^ P t ® ^ : R k —l P t ••• *

0 « S 9 i < p , 0 ^ q0 < p, 0 ^ r ' < p

By b, *1,

^ = Rku k + R k-lu k -l + ••• + R lU l + R 0U0 .

142
Moreover, for any s = 1, 2, . . . , m, we have

+ 7 « -2 “ s- j + • • • + <7i“ i + <7o“ o < “ a •

Therefore our expression for h must coincide completely with

the one considered in the problem.
8, a. L e ttin g x t be an integer, Q ^ a < /3 ^ R,
x t < a < ft < x 1 + 1, and integrating by parts, we find

P P
f{x)dx = j p'{x)f{x)dx = p(F})f{{3) - p (a )/(a ) -
a cl

- o{fi)f'((3) + a ( a ) f ' ( a ) + j o(x)f"(x)dx.

In particular, for Q ^ x lt x, + 1 ^ R, passing to the limit,

we have

*i+‘ *i+‘
- J f t o d x = - — f( x l + 1) - + j o{x)f"(x)dx.
* i * i

We can then obtain the required formula easily,

b. Rewriting the formula of problem a in the form

R Q
£
Q<x^R
fix) =
J
f f(x)dx -
J
f f(x)dx + p(R)f{R) - p(Q)f(Q) -

0
0 C
O

- o{ R)f'(R) + a{Q)f'(Q) + j o(x)f"(x)dx - J a(x)f"{x)dx,

we obtain the required formula.

143
 c. Applying the r e s u lt of problem b, we find

l n l + l n 2 + . . . + l nra = C + r c l n r c - r a +

1 f n v
+ — In n + I -------dx = n In n — n + 0(ln n).
2 J x1
n

9, a. a) We have (b, §1 and problem 5)

n n
(1) ln([n]!) = £ — + In
p<r _P_

T he right side represents the sum of the valu es of the function

In p, extended over the la ttic e points (p, s , u) with prime p of
n
the region p > 0 , s > 0 , 0 < u ^ ----- . The part of this sum

corresponding to given s and u is equal to 0 ( y — ] ; the

u

part corresponding to given u is equal to ip ^ — j .

fi) Applying the resu lt of problem a) for n ^ 2, we have

In ( [ n ] !) - 2 In

0( b) - j j - 0 f — j + . . . > 0 ( « ) “ 0 f— j

Setting m, we then find ([n] = 2 m, [n] = 2 m 4- 1)

144
 (2m + 1)1
i//(n) - i]t ^ In
(ml)1
5 . . . (2m + 1)
^ In $ In (2'"3n') < n,
1 •2 ... m

i/'(«) - i!>(n) - \jj + 0 +

2 2

n n
+ 0 + ..• n -t- ~ ~ ■ ™ 2 n.
4 ' i 2 4

y ) Wc hnvc (by the solution of problem ft) and the re su lt of

problem 8, c)

i/j(n) - < /'(“ ') + 0 {— ) - 0 (— I + In

[nil
n
2

[n ] ln [ „ ] ~ ( n ] ~ 2 [ j j In + 0(ln n) ■

- n In 2 + 0(1 n n).

Moreover, for s ^ 2 wc find (problem f t ) )

0(V^T ) - 0 +

r
< 2 always

+ 0
- 0 for .<! > r; r
In 2_
145
Therefore

< 2 VirT + 2 VrT + 2Vn" +

+ . .. + 2Vre~ < 2 (\fn + t \fn ) * 0 ( Vr c " ) .

b. The re su lt follows from equation (1), the inequality of

problem a, /3) and the equation of problem 8, c.
c. The equation of problem b for sufficiently large m gives

■c-i In p In m 4
2_i = In m + 0(1) ^ -------- , 2_i— > 1.
P ^ m<Cp^m* P

If pn+l > pn( l + f ) for all p airs pn, pn+1 such that
m < p n < pn+l ^ m* then we would have

00 4

r ------------— > 1
r^o m (l + ()r

which is im possible for sufficiently large m.

d. It is evidently su fficie n t to consider the case in which
n is an integer.
In r
Setting y(r) = ------- forr prime and y(r) = 0 for r = 1, and
r
for r com posite, we have (problem b)

y ( l) + y(2) + . . . + y(r) * In r + a(r); | a (r) | < C „

146
where C2 is a co nstant. Hence, for r > 1 (we consider
oc(l) - 1)

y(r) * In r - In (r - 1) + a(r) - a ( r - 1),

I n r — 1n (r - 1)
— =* T l + T 2\ 7 \ =*
0<p^n P K r^n In

a(r) - a(r - 1)
T
1 2 L
K r^n In n

We have (8, b)

1 1
£ 1 + £ / + "I + . . .I
l<r<n r ln r l< ^ n V2'* ln r ,n

C2 + In In n + 0
i In n

where C2 is a co n stan t. Moreover we find

T = a (2)
In 2 In 3

1 \ a (n)
+ . . . + a (n - 1) +
In (n — 1) ln n In n

But, for an integer m > 1, we have

1
1 \In m In (m + 1)

1 1
+ Ci I + . . . —
In (m + 1) In (m + 2) In m
147
Therefore the s e r ie s

1 \ ( 1
a(2) + a(3)
In 2 In 3 In 3 In 4

converges; therefore, if C3 is its sum, then

T. = C, + 0
In n ,

e. We have

in 77 fi - P P
(-p r +
\
j t
3?
+■■
p^n ) p<in p^n

= C ' — In In n + 0
In n

where C ' is a constant. Setting C ' = In C0 in the latter

equation, we obtain the required equation.
10, a. T h is re su lt follows from c, 52.
b. Since 0(1) = 0(1) = 1, the function 9(a) s a ti s f i e s condi­
tion 1, a, 52. L e t a = a ia z be one of the decom positions of
a into two rela tiv e ly prime factors. We have

£ QUidj) = \p(a) = 0 ( a t) 0 ( a z) =
di \ a i

(1 )
= L L W W -
^ i\ai

If condition 2, a, 52 is s a tis fie d for all products sm aller than

a, then, for d 3d3 < a we have 9(did2) = 9(dt )9(d2), and equation
(1) gives 9(ata3) = 9(al)9(a}), i.e. condition 2, a, 52 is also

148
s a tis fie d for all products a va2 equal to a. But condition
2, a, $2 is s a tis f ie d for the product 1 • 1 which is equal to 1.
Therefore, it is s a tis f ie d for all products.
11, a. L et m > 1; for each given x m dividing a , the in­

determinate equation x x . . . x m^ l x m = a has rm_,

so lu tio n s. Therefore

Tm (o) = y * rm—\
xm\a

but when x m runs through all the divisors of the number a, the
a
numbers d = ------ run through all th e se same divisors in re-

verse order. Therefore

rm(a) = 2 ]
tf\a

Hence (problem 10, a), if the theorem is true for the function
rm_i(a), then it is also true for the function rm(a). But the
theorem is true for the function rj(a) *= 1, and hence it is
alw ays true.
b. If m > 1 and the theorem is true for the function rm-1(a),
then

'■m(a) = rm(p,) . . . rm(pk) =

= (^m—l( l ) "I" ••• (rm—1^1) + ^m—1(pfc) ) =

= (1 + m - 1)* - mk .

But the theorem is true fcr the function r^ a ) , and hence it is

alw ays true.
c. L e t r * mfj, f2 = 27/, and le t a = p?1 . . . p k k be the
canonical decomposition of the number a, where p,, . . . , p k
149
are arranged in in c re asin g order. For the function
r,(a) = r(a) we have

T(a) a.l + 1 a2 + 1 ak + 1
2 ai V 3 aJ^ (h + l ) a *77 *

Each of the factors of the product on the right is sm aller than

— ; the factors ——— - — such that r > 2 ^ i s sm aller than

aM + 1 , „ / 1 \ ■”
-------------- ^ 1. Therefore, settin g C * ( — I , we find

r( a ) r(a ) C
—— < C, lim ----— < lim —— = 0.
a ,V a - oo a cj 5-co a 7?

It is evident that rm(a) < ( r ( a ) ) m for m > 2. Therefore

rw(a) ( r(a)
lim --------- — < lim I — —
a-.oo a _oo ^ 0 e a

d. We divide the sy stem s of values x lr . . . , xm s a tisfy in g

our inequality into [a] s e ts with su b sc rip ts 1, 2, . . . , [a]. The
sy stem s such that x t . . . xm = a are put in the s e t with
subscript a; the number of th e se sy stem s is rm(a).
12. The s e r ie s defining £ (s) converges absolutely for
R(s) > 1. Therefore

00 00
(C is)r 8

while, for given positive n, the number of system s n lf . . . , n m

such that n2 . . . nm = n is equal to Tm(n).
150
 1
13, a. The product P = j j ------------- converges absolutely
p i - —
P“

1 1 .
for R{s) > 1. Since ------------- =■= + + . . . for
1 _ J _ Pa P2S
PS
N > 2, we have

1
77
P^N 1 n
1
P
where the second sum on the right is extended over those
numbers n which are not divisible by primes larger than /V.
As /V —* oo, the left side tends to P, the first sum of the
right s id e tends to £ ( s ) , while the second sum on the right
tends to zero.
b. L e t N > 2. A ssum ing that there are no primes other
than p,, . . . , pkt we find that (cf. the solution of problem a)

k 1
L - •
n
'- 1 —
1 - —r
P 0< n< JV

Pi
T h is inequality is im possible for sufficiently large N because
1 1
the harmonic s e r ie s 1 + — + — + . . . diverges.
2 3
c. Assuming that there are no primes other than p „ . . . , Pk
we find (problem a)
k i

i -1
n i
= £ 2).

i —

151
T his equation is im possible in view of the irrationality of
TT2
C© - — •

14. The infinite product for C,(s) of problem 13, a converges

absolutely for R(s) > 1. Therefore

in c (s) = r (-— + ----- + ---------- + . .

p \P S 2 p 2s 3 p 3s
/

where p runs through all the primes. D ifferentiating, we find

C'k) = y I ln P In p ln “ A(n)
2s Is ~~ 2_l
M “ H P* n -1 Tl

15. L e t N > 2. Applying theorem b, §3, we have

/
1 £ p (n) p (n)
77 1 E
P^N P oKn^N n ns

where the seco n d sum on the right is extended over those

numbers n larger than N which are not divisible by primes
exceeding A. Taking the limit a s N —* °°, we obtain the
required identity.
16, a. We apply d, §3 to the c a s e in which

5 = 1, 2, . . . , [n], f = 1, 1, . . . , 1.

It is then evident th a t S ' = 1. Moreover S d is the number of

n
values 5 which are m ultiples of d, i.e .
~d

b, a ) The right side of the equation of problem a is the sum

of the v alues of the function p{d), extended over the la ttice
152
points (d , u) of the region fZ > 0, 0 < u ^ — . T he part of
d
this sum corresponding to a given value of a, is equal to

ft) The required equation is obtained by termwise subtrac­

tion of the equations

M(n) + M ( y j + M I j j + AH — ) + . . . = 1,

™ It' + 2/W( t ) + ••• - 2-

c. L e t n 1 = [n]; let 8 ,, 8lt . . . , S„ be defined by the con­

dition: 8 e is the la rg e st integer whose Z-th power divides
s , f a = 1. Then S ' = T t f n , S d is equal to the number of

multiples of d1 not exceeding n, i.e. S d =* |^— J ' ^ rom this

we obtain the required expression for

In particular, sin ce n(2) = ----- , we have

6

T2 = — n + )
n

for the number 7'2j„ of integers not exceeding n and not

divisible by the square of an integer exceeding 1.
17, a. We obtain the required equation from d, $3, if we set

5 , — (^"s» fa x ft^a)'
153
 b. We obtain the required equation from d, 53, if we s et

ss = (*<•>, 4 a)), f. = /(48), .... 4 S)).

c. Applying d, 53 to the ca se

where we have written down all the divisors of a in the first

row, we have

£ F (— j -c(f) .

d. T h e required equation follows from

S M(d) E M(d) E M(d)

p ' x fd\8i -d\Sj j . d \ & jn
'1 '2 n

18, a. We apply the theorem of problem 17, a, le ttin g ac run

through the numbers 1, 2 a and taking f(x) = x m. Then

S ' =. ifrm(a)t S d = dm + 2mdm + . . . + (—! dm

^ 7 .

b. We have

E ( a1 a a
11 (d) + — — cp(a).
d\a 1 2d 2 2
154
 We can obtain the same re su lt more simply. We first write
down the numbers of the sequence 1, . . . , a relatively prime
to a in in c re asin g order, and then in d ecreasing order. The
sum of the terms of the two s e q u e n c e s equally d is ta n t from the
initial terms, is equal to a; the number of terms in each s e ­
quence is equal to cp(a).
c. We have

iAj(a) £ /*W)
d \a

O a
= — cp(a) + — (1 - Pi) . . . (1 P
3 6

19, a. We apply the theorem of problem 17, a, le ttin g x run

through the numbers 1, 2, . . . , [z] and taking fix) = 1. Then
S ' = T z , S d is equal to the number of multiples of d which

do not exceed z, i.e. S d =

b. We have

T z ~ 22 fi(d) — + 0 ( r ( a ) ) - — cp(a) + 0 ( a £ ).
d \a d a

c. T h is follows from the equation of problem 19, a.

20. We apply the theorem of problem 17, a, letting x run
through the numbers 1, 2, . . . , /V, where N > a, and taking

f{x) = —— . We then find

1 Hid)
- L nid) 22 L z r •
d \a d ° x e - Ld \e N x
0<x< —
0<x^

T aking the limit as N oo, we obtain the required identity.

155
 21, a. We apply the theorem of problem 17, b, considering
the s y ste m s of valu es x,, x 2t . . . , x k considered in the defini­
tion of the pro b ab ilities P , and talcing /(x, x 2, . . . , x k) = 1.

s'
Then P „ = —— , Sd “ — » and we obtain
" Nk L d]

N 1
PN
°iz: N d k~l
c f-1

Therefore

PN = + 0(A ); A = for > 2,

A

In N
A * -------- for h = 2.
N

77
b. We have £ (2) * ---- .
6
22, a. Elementary arguments show th a t the number of la ttic e
points (u, v) of the region u2 + v2 p2; p > 0, not counting
the point (0, 0), is equal to np2 + Oip). We apply the theorem
of problem 17, b, considering the coordinates x, y of the la t­
tice points of the region x2 + y2 ^ r2, different from (0, 0),
and s e ttin g f ( x , y) = 1. Then T = S ' + 1, S d is equal to the

number of la ttic e points of the region u2 + v2 ^ ^— not

considering the point (0, 0). Therefore

156
 [r] r2 /M r'
* 0 £ - r2 + 0 ( r In r).
dal

b. Arguing in analogy to the above, we find

M 4 r> /M r2 \ 4^r*
Z / W j » - + o l £ - \ - acoT + Ofr2).

23, a. The number of divisors d of the numbers

a = p ^ 1 . . . p “ * which are not divisible by the square of an
integer exceeding 1, and having x. prime divisors, is equal to

( k
; moreover fi(d) = ( - I ) * . Therefore
\ H .

£ p(<fl = £ \ (-1)" - (l - Dk = o.
d \a H -o \* /

b. L e t a be of the same form as in problem a. It is sufficient

to consider the c a s e m < k. For the sum under consideration
we have two e x p re ssio n s

If m is even, then for m ^ — , the first expression < 0 , and

k
for m > — the second expression is ^ 0. If m is odd, then

157
 k k
for m < — , the first expression <0. and for m > — the
2 2
second expression ^ 0 .
c. T he proof is alm ost the same as in d, $3, except that
the re s u lt of problem b must be taken into account.
d. T h e proof is almost the same as in problems 17, a and
17, b.
24. L e t d run through the divisors of the number a, let
(!(</) be the number of prime divisors of the number d , and
let Q(a) = s. Follow ing the process given in the problem,
we have

(/V, q, 1) < Z] /x(d) (— + 0d ] =

\9 I

- T +T0- Tj K | < 1,

|r | < E i. T , - Ir.l - E 4 -
Q(dy^jn 9 d ® Q(cf )>m 9

Moreover

s qr N
s m < enm < exp( 5 ^ “ 6 In r ) - ---------- 0(A ),
L n N qr

1
N
To 0(A).
9
n i

F inally, le ttin g C „ CJt C s denote co n s ta n ts , we find

158
 . N a _ 1
r.l < - E L 7 «
*7 n-m +1 Q(d)-n “

/V • / 3 \ /V - 4 l n - , x
< C2— 7 " — < C,— r 3 = 0(A ).
q « k» U q

25. To every divisor d, of the number a such that d, < Va~

there corresponds a divisor d2 such that d2 > ^ , d,d2 = a.
Here p(dt) = p(d2). Therefore

2 ^ p { d t) = 7 ] M W + 2 ] = Zj “ 0.
dj dt dj d \a

26. We consider p airs of numbers d which are not divisible

by the square of an integer exceeding 1, and satisfying the
condition cp(d) = k, such that each pair co n sists of some odd
number d, and the even number 2d t . We have p(dt) + p(2d,) = 0.
27. L e t p,, . . . , p k are distin ct prime numbers. Setting
a * p t . . . p k, we have

cp(a) = (p, - 1) . . . (pk - 1).

If there were no primes other than p lt . . . , p k , we would have

cp (a ) * 1 .
159
 28, a. Our numbers are among the numbers s8", s = 1, 2,
a a
, — . But (s8, a) =* 8 if and only if s, — 1
8 \ 8
(e, 52, ch. I). Therefore the assertio n in the problem is true
and we have

a = H cp [ — j = 2Z cp(oO.
d \a J d \a

b, a) L e t a « p*1 . . . p%n be the canonical decomposition

of the number a. By a, the function cp(a) is m ultiplicative,
while

P s° * Z! P s a~ l = 2Z 'K A P s s “ P®S_1 * ? ( ? « “)•

d \p B d \p s~l

f!i) F o r a po sitiv e integer m, we have

m = 2Z W)-
d\m

Therefore

cp(a) = 2 ] .
d \a d

29. We have (p runs through all the primes)

00 cp(re) cp(p) <p(p2)

n i + + •«
p \ p

r - —
p8 C(s - l)
77 £(s)
l - „ s —1

160
 30. We have

cp(l) + cp(2) + . . . + cp(rc) =

p (<D p(d)
z E — + + « E
d\i ~ d ~ d\t d d\n d

p(d) [ 1 + 2 + . +
d-l \

Z] TT + 0(n In n)
dml 2 d2

n2 w p(d)
— 2^ — ~ + 0(n In n) = ——n1 + 0(n In n).
2 d-i d 77

Solutions o f the Problems for Chapter III

1, a. It follows from

P = a n 10n" 1 + a n_j 10n" J + + a i.

that

P = an + a n-l + . . . + a f (mod 9)

since 10 = 1 (mod 9). Therefore P is a multiple of 3 if and

only if the sum of its digits is a multiple of 3; it is a multiple
of 9 if and only if this sum is a multiple of 9.
Noting that 10 = —1 (mod 11), we have

P s (a, + a 3 + . ..) - (a4 + a 4 + . . .) (mod 11).

161
Therefore P is a multiple of 3 if and only if the sum of its
digits in the odd p la c e s minus the sum of its digits in the
even p la c e s is a multiple of 11.
b. It foil ows from

P = bn lOO"-1 + bn_, 100n" 2 + . . . + b2

that

P = (bi + h, + . . . ) - (b2 + bA + . . .) (mod 101)

since 100 = —1 (mod 101). Therefore P is a multiple of 101

if and only if (bt + bi + . . . ) — (b2 + + . . .) is a multiple
of 101.
c. It follows from

P = c n 1000n“l + c n_, 1000"“* + . . . + c x

that

P = c„ + cn_, + . . . + Cj (mod 37)

s in c e 1000 = 1 (mod 37). Therefore P is a multiple of 37 if

and only if c n + c n-1 + . . . + c, is a multiple of 37.
Since 1000 = —1 (mod 7 * 11 * 13), we have

P = (cj + c 3 + . . . ) — (c2 + c« + . . .) (mod 7 • 11 * 13).

Therefore P is a multiple of one of the numbers 7, 11, 13 if

and only if (c, + c 3 + . . . ) - (c2 + c4 + . . . ) is a multiple of
that number.
2, a. a ) When x runs through a complete system of residues
modulo m, then ax + b also runs through a complete residue
system ; the s m a lle s t non-negative resid u e r of the numbers
ax + b also runs through the numbers 0 , 1 , . . . , m — 1.
162
 /9) Applying the resu lt of problem 18, b, ch. II, we find

ifjyim) 1
— cp(m).
m 2

b. F or t =* 1, we have [f(N + m)] - [/(/V)] = a,

N+m I 1
E [ / W ] -------W + m)] + - [ / ( A ) ] - - + — m =«
x-w+i 2 2 2 2

/V+m W+m 1 1
= E ^ ~ Z] IfM l - + — (rn - 1) =» S;
x - N +l X‘ f f + 1 2 2

and the case in which t > 1 also reduces to this case trivially.
c. L e t A, A/, P t , P 4 be integers, A/ > 0, P t > 0, P 2 > 0.
The trapezoid with v ertices (A, 0), (A, P t), (A + A/, 0),
(A + A/, Pj) is a s p ecia l c a s e of the one considered in problem
b. Therefore equation (l) is also valid for it. Equation (1)
can also be obtained easily for such a trapezoid by consider­
ing the rectangle with v ertices (A, 0), (A, P + P ), (A + A/, 0),
(A + A/, P + P), which is equal to two such trapezoids. For
this rectangle, the equation

E s - s ',

analogous to equation ( l) , is evident. Since = 2 l>

this implies S ' = 2S , so that we obtain equation (1).
163
 T he analogous formula for the triangle considered in the
problem follows trivially from this resu lt. But it is of some
in terest to consider the following derivation: our triangle can
be obtained from a certain parallelogram with integral vertices
by dividing it into two equal triangles. L et 5 be the area of

the parallelogram and let T = where the sum is extended

over all the la ttic e points of the parallelogram and 5 is de­
fined a s in problem b. We will have proven the property of the
triangle that in te re s ts u s, if we prove that S = T. We con­
sider a square whose side A in c re a s e s to infinity. The whole
plane can be divided into an infinite number of parallelogram s
of the above type. L e t k be the number of parallelogram s com­
pletely within the interior of the sq u are, and le t R be the
number of la ttice points in the square. As A —►°°, we find

kS A2 R
lim ---- = 1, lim ---- = 1, lim ----- = 1.
A1 R kT

Multiplying these ex p ressio n s termwise, we find

s
lim — = 1, 5 =■= T.
T

3, a. L e t r be the s m a lle s t positive residue of the number

ax + [c] modulo m. We have

r + 4>(r) 1
m \

where f <1 D(r) $ t + A; f - [ c | . T he theorem is evident for

m ^ 2h + 1. We therefore only consider the ca s e m > 2h + 1.
Setting

r + <D(r)
— = S(r),
m m

164
 h + f
we have - 1 + — <C 5(r) < --------- for r = m — [h + (},
m m
e ^ + f
. . . , m - 1; in the other c a s e s — <; 5(r) ^ ---------- .
m m
Therefore

m - 1 1
—[A + c] + €. ^ S — ^ A + £. S — —m ^ A+
2 2

b. We have

m_l f az + iA(z) 1 „ X
S = £ j ------------ — r ; i W = m<<AM + B) + — 2.
2-0 I m m

We apply the theorem of problem a, s e ttin g A = | \ | . Then

we obtain the required result,
c. We find

az f " \ M + z„) '

s = £ m + — + 9
m

0 < z 0 < m - 1.

W'e apply the theorem of problem a, s e ttin g A = 1 + — . We

2
then obtain the required result.
4. W'e develop A in a continued fraction. L e t Qn = Q '
be the la rg e s t of the denominators of the convergents which
does not exceed m, and note (h, 44, ch. I)

P' Q'
, ( P ' t Q ' ) = 1, \ 0 ' \ < 1.
~Q' + ~Q^n
165
It follows from m < (^n+1 < (<<rn+l + l)<?n ^ CQm where C is
a constant which is not larger than all the q a + 1, that, for
the la rg e st integer H ' such that H ' Q ' < m, we have H ' < C.
Applying the theorem of problem 3, b, we find

m +h 'q '- i 1 3
£ U x + fij - — H'Q' — C.
XmM ^ 2

L et = m - H ' Q ' . If ml > 0, then, choosing Q " and H "

depending on m as we chose Q ' and H ' depending on m, we
find

3
£ \Ax + B \ -------H " Q — C.
2 2

L et m2 = ml - H " Q " . If m2 > 0, then, as above, we find

m 2+h " ' Q " ' - i l 3

E \Ax + B } - - H ' " Q ' "
x-M, 2

etc., until we find some mk = 0. We then have { H ' Q ' +

+ H " Q " + . . . + //<*><?<*> = m)

M + m —i 1 3
E \Ax + b ] - —m
2
< — Ck.
2

The numbers Q ' , Q " , . . , , Q^k ^ s a tisfy the conditions

m ^ Q ' > m2 ^ Q " > m2 ^ . . . > mk_l Q^k ^ ^ 1.

Therefore (problem 3, ch. I) k = 0(ln m), and hence the re­

quired formula is true.
166
 5, a. Let the sum on the left be denoted by S. L et r = A 3 .
The theorem is evident for r ^ 40. We therefore assume that
t > 40. T aking Ml = « [ ( ) + 1], we can find numbers a „ mt, 0
such that

/'(iW,) = —- + — —; 0 < m ^ r , (ait m2) = 1, | 0 11 < 1.

mj 77i,r

Taking M2 = + mlt we find the numbers a2, m2, 0 2

analogously; taking 1/, = M2 +• m2, we find the numbers
° 3* m 3 t O31 e tc ., until we come to Ms+l = Ms + m a such that
0 ^ [/i] — M8+i < [r]. Applying the theorem of problem 3, C,
we find

1
S - — (m, + m2 + + + [/?] — 1) <

k + 3 1 /r ,
< s — ~— + “ M° +i)»

k + 3 t + 1
S -------(ft - Q ) < s ---------- + ----------
2

a 1
The length of the interval for which — - ----- ^ f ' ( x ) ^
m mt
a 1 2A
<C — + ------ does not exceed . Therefore there are
m TTLT TUT
2A
< ------- + 1 numbers mlf m2, . . . . m a a s s o c ia te d with the
m2r
u
fraction — . L et a , and a 2 be the s m allest and largest values
m
of a a s s o c ia te d with a given m.
167
 We have

2 H R - Q)
m -----
m t ^ ------ A~A-------

H R - Q)m
a 2 - a, + 1 < ----------------- + 1.05.

Therefore, there are

2A \ I HR - Q)m
\< — + 1 — + 1.051
m1T

H R ~ 0) 2 m 2A
r + V — + 1H.05
m2r

numbers m2, . . . , mk a s s o c ia te d with a given m. Summing

the la tte r expression over all m = 1, 2, . . . , [r], we find

HR ~ Q ) r* + r \ 10/4
S < ---------- — 2 In r + 2 + 1.05 <
2 rJ 3T
HR-Q) , A 7 / 4
< --------------- In A + --------- ,
r 2 r

H{R - Q) A
S~ - (R ~ Q ) < 2 ------------— In A + 8k —
T T

b. We have

£ \f{x) + 1 - a] - ~ ( R - (?) < A,

C><X<R 2

£ ~ - (R - Q) < A,
Q<x<R 2
168
from which, se ttin g 5(x) =• ff(x) + 1 — a \ — \f (x)l, we find

H 8(x) < 2A.

Q < x^.R

But, for \f(x)\ < a we have S(x) = 1 ~ o, while for

\f(x) I 'Z o we have 5(r) = - a, and hence | (1 — a)xfj{a) —
- a(R ~ Q ~ \Jj{o) ) | < 2A , from which we obtain the required
formula.
6, a. We apply the formula of problem 1, c, ch. II. Setting
f(x) =* V r1 - x 2 , we have

—r
f'ix) - - f"(x)
( r 1 - x 2)%

1
1/
r

in the interval 0 ^ x ^ Therefore (problem 8, a, ch. II,

V2
problem 5, a)

x 2 dx + 8 p —— - 8 p (0)r -
V2

r
- 4 + 0(r3 In r) -
V2
2

* n r 2 + 0 (rJ In r).

b. We have (problems 11, d and 1, d, ch. II)

r ( l ) + r(2) + . . . + r(n) * 2 E — w z r .
0<x^/rT x

169
It is sufficient to consider the ca se n > 64. We divide the
interval X < x ^ Vn" , where X * 2 n J , into 0(ln n) intervals
of the form M < x ^ M ' t where M ' ^ 2 M. Setting f{x) -
n
=- — , we have
x
2n n 8n
rw - - 4X » / " ( * > ^ r w ^
4 M* 4

in the interval M < x ^ M '. Therefore (problem 5, a)

Moreover (problem 8, b, ch. II)

T. — « En + — n In n + p(Vn* )Vn" + 0(1).

o<x^/TT x 2

Therefore

r(l) + r(2) + . . . + r(n) *

» 2 En + n In n + 2 p{\fn )Vn" _ _ n +

+ 2 V/T } +0 ( n 5 (In n)2) *

=•= n(ln n + 2 E — 1) + 0(n* (In n)2).

170
 7. L e t the system be improper and le t s be the largest
integer such that 2 s enters into an odd number of numbers of
the sy stem . We replace one of the la tte r numbers by the
sm a lle st number containing only those powers 2s which enter
into an odd number of integers of the r e s t of the system .
L e t the system be proper. A number sm aller than one of
the numbers T of this sy stem , differs from T in at le a s t one
digit in its representation to the base 2.
8, a. Adding the number H » 3n 4- 3n-rl 4- . . + 3 4- 1 to
each of the numbers of the system represented in the afore­
mentioned manner, we obtain numbers which we can obtain by
letting x n , acn_t , . . . , x l r x 0 in the same form, run through the
values 0, 1, 2, i.e. we obtain all the v alues 0, 1, . . . , 2 H.
b. In this way we obtain mlm2 . . . mk numbers which are
incongruent to one-another modulo mxm2 . . . m k , since

Xj "Hm xx 2 4* m xnx2x x • <* 41n x xn x 2 • • • flXk***\x k =

= x[ 4- m xx 2 + m xm 2x 3' + . . . + m lm 2 ... m k-\x k

(mod a t ... ttxjj)

implies in sequence:

xx = (mod m,), x x = x ' ; m xx 2 = m xx ' (mod rnxm2)f - x' ;

tti ^//ij Xt^ -- ttxXm2x j (mod ttxxttt2/ti , x x ^ Xy ,

etc.
9, a. In th is way we obtain m xm 2 . . . mk numbers which are
incongruent modulo m xm 2 . .. m k , since

,1 /,* ! + M2x2 + • • • + f^kXk = Mxx { + M2x ' + . . . + Mkx k

(mod ... tti

171
would imply (every Mj, different from M g, is a multiple of m g)

Mgx g = Mgx g (mod m), x g = x ' (mod ms ), x g =* x g .

b. In this way we obtain cp(m,)cfi(m2) . . . cp(mk) =

=* cp(m1mJ . . . mk) numbers which Eire incongruent modulo
m,m2 . . . mk by the theorem of problem a, and are relatively
prime to mim2 . . . mk s in c e (Mix l + M2x2 + . . . +
+ MkXk , mg) - (Mgx e , m g) = 1.
c. By the theorem of problem a, the number Mlx l + M2x 2 +
+ . . . + Mkx k runs through a complete residue system modulo
m,m2 . . . mk when x lt x 2, . . . , x k run through complete residue
system s modulo mlt //Ij y • • • y T 7 1 T h is number is relatively
prime to m1m2 . . . mk if and only if ( xlt m t) =* (x 2, m2) >=
= . . . = (x k, mk) * 1. Therefore y ( m lm2 . . . mk) =*
* cp(m,)cp(m2) . . . cp(mk).
d. To obtain the numbers of the sequence 1 , 2 , . . . , p a
relatively prime to p a we delete the numbers of this sequence
which are m ultiples of p , i.e. the numbers p , 2 p , . . . , par~1p-
Therefore cp(pa ) =* p a ~ p a_1. The expression for cp(a)
follows from the la tte r and theorem c, $ 4 , ch. II.
10, a. The first a ssertio n follows from

Xj x k "I I Mlx l + . . . + Mkx k

• — + • • • + ------ \ “ I ------------------------------
m l m k J L m

the second as s e rtio n follows from

Jt ^1^1 + • • • + Mk£k
+ +
mk m

b. The fractions

f S x i* w t) fk(x k f Wi
m. m.

172
coincide with the fractions

f M U x i + ••• + • ••» + . . . + Mkw k )

+ • •• +
m.

f k(Mlx l + . . . + M k x kr ...» M lW l + . . . + Mkw k)

m,

i.e. with the fractions

\ fib, ...» fkb t ...»

•j--------------------- + . . . + ---------------------
I m l W fc

The first assertio n follows trivially from th is. T he second

a ssertio n is proved analogously.
11, a. If a is a multiple of m, we have

I axi
£ exp l2rri----
\ m i E 1-"-
If a is not a multiple of m, we have

/ am 1
exp I 2 n i----- 1 - 1
ax
2ri ■ 0.
m
exp I 2774— 1 - 1

b. F or non-integral a , the left sid e is equal to

exp(2nia(M + P ) ) - exp(2rria.M) 1 1
—---------- < -------- .
exp(27ria) — 1 sin n-(a) h ( a)

173
 c. By the theorem of problem b, the left s id e does not ex­
ceed T m, where

Tm - E

F o r odd m.

2a + 1
Tm < m ^ In —-------— * m In 77i,
2a - 1
K a<—

and for even m,

m 2a + 1 77i 2a+l
T < —
2
V I n ----------- + —
2a — 1 2
y I n ------------- < 77i In 77i.
2a — 1
(Ka^ 2 °<a< “T

1 1 1
F or 77i > 6, s in ce — — — the bound tti In tti can be
2 3 6
d ecreased to

77i 2a + 1 771 771

2— y In-------- — In 2
6 m 2a — 1 3 7
0<a< —
N »

The la tte r expression is > — for m ^ 12 and > m for

77i ;> 6 0 .
IE, a. L et tti * pj*1 . . . p®* be the canonical decomposition
of the number m. Setting p ^ 1 =* tti, , . . . , p£* » 77ik, and using
the notation of problem 10, a, we have
F o r a 8 > 1, s e ttin g m a - p am ' , we find

£ exP

m -—I u
y 1 exp l 2 n i ----- I - £ e x P I 2tt»
U "0
n\.

m ~ l X
b. L e t m be an integer, m > 1. We have /"] exp 2ni—
x -o m
» 0. By the theorem of problem a, the sum of the terms on the
left side of th is equation such that (x, m) =» d, is equal to

c. We find

H e x p ll n i — \ =* £ p W S dr
£ \ m ) A n.

where, settin g m =• m0d, we have

mo-l / u
S d ** XX exP | 2tt£----
u-o I mo

The la tte r is equal to 0 for d < m and equal to 1 for d = m.

From th is we obtain the theorem of problem a.
175
 d. T h is equation follows from problem 10, b.
e. We have

/4(mj) . . . A(m k) =« m £ . .. ^ OTi . . . $ a kt m k ,

ai ak

where a lt . . . , a k run through reduced residue sy stem s modulo

mlt • • •» Hence (problem d) the first equation of the prob­
lem follows immediately.
We also prove the second re su lt analogously.
13, a. We have

p-1 / nx\ p, if n is a multiple of p,

£ exp f 2™ —
*■0 0, otherw ise.

b. Expanding the product corresponding to a given n, we

find

' nx
2jri----
£ —r L cxp d
d \a “ x-o

H ence, summing over all the n =* 0, 1, . . . , a — 1, we obtain

the expression for cp(a).
14. The part of the expression on the right corresponding
to x dividing a, is equal to

1
lim 2r Y '
e-*o
^■1 I 7 7

K ak
Setting $(K) * 2^
r ex P (2n-i---- , the part corresponding to
kmi
176
x, not d iv isib le by a, can be represented in the form

<H(1) $ (2 ) - <D(1) <D(3) - 4>(2)

lim 2 f
e—o 1 + 2 1+e + 3 1+e

lim 2 f ( $>(1) 1 -
— ♦ *<2> 2 l + e 1+ €

The factor to the right of 2 e , is < x in absolute value since

I I < x; here lim 2 ex * 0. Therefore the right side of
€-* 0
the equation considered in the problem is equal to twice the
number of divisors of the number a which are sm aller than '/ a ,
m ultiplied by 8, i.e . equal to r(a).
15, a. We have

(A, + h 2)p =*

= A? + ^ j hrlht + ... + l^pP_^ M r1+ A? =

= hp + /if (mod p);

(Aj + h2 + hi )p = {ht + h2)p + h f s + h% + h% (mod p),

etc.
b. Setting h x = h2 =* . . . * Aa = 1, the theorem of problem
a g iv e s F erm at’s theorem.
C. L e t (a, p) = 1. F o r certain integers /V,, /V2, . . . , N we
have

a p- ‘ = 1 + Nip, ap ^ - l) = (1 + N lP)p = 1 + N 2p2,

a P J (p - i ) = x + /V 3p \ . . . , a P ^ C P - O = l + /V a p a ,

Qcp(p ) s j (mod p a).

177
L et m = p?1 . . . p £ k be the canonical decomposition of the
number m. We have

a <P(Pi*) = \ (mod p f 1), = 1 (mod p®*),

a 9 ( m) = 1 (mod p ^ 1), . . . , a cP(m^ s 1 (mod p®*),

a cp(m) s j (mod m).

Solutions o f the Problems for Chapter IV

1, a. The theorem follows immediately from the theorem of

problem 11, a, ch. III.
b. L e t d be a divisor of the number m, m - m0d, and le t Hd
denote the sum of the terms such that (a, m) = d in the expres­
sion for Tm in problem a. We find

a0f[xt . . . , w)
£ £ exP 2ni rnn
9

where a0 runs through a reduced residue system modulo m0.

From this we deduce
mQ—1 aof(x o wn
= m rA(m0).
d’Z ze0 *o” °
£ ex p 2ni
F7ln

c. Let m > 0, (c l p ni ) = c/j cz Qq(I} ft z //Zq(/y and let T be

the number of solutions of the congruence ax = b (mod m).
We have

m—1 m—1 a (ax -

Tm = £ £ ex p \ 2ni
C t* 0 X -0

in-1 m—l a a„ 6a \
£ 2 ] eXP 2?7l------ x 2774----- I
0^0 X*0 ' m /

d_1 / 6a t mrf, if 6 is a multiple of

= m y ' exp I -277t------
aj-o ' ^ 0, otherwise

178
 d. S etting (a, m) = dlt (b, d t) ~ d29 . . ., (/, dM ) = rff
/7i ~ d ^ m ^ f ^ • • •> = d f f r i f f we find {/ *-

m—l m—1 m—1 m—1 a (ax + 6y + . . . + fw + g)

Tm=0*0
T, E E • • • tv-0
jr-0 y -0
E exP 12 n i m

% 1 m~l m~l / . . a i(*y + . . . + fw + g)

= m a1*o
E Ey—o • • • Ew-0 c x p \
2m- - - - - -

a M (/u> + g ) ’
E E exp 2771

dr- i a rg
= mr 2^ exp 2tt7~
ar-o

e. We apply the method of induction. Using the notation of

problem d, assum e that the theorem is true for r varia b les. W'e
consider the congruence

(2) Iv + ax + . . . + fw + g = 0 (mod m).

L et (/, m) = d0. Congruence (2) holds if and only if ax +

+ . . . + fw + g = 0 (mod d0). The la tte r congruence holds
if and only if g is a multiple of d ' , where d ' = (a, f, d0) =
=• (1, a, m), and it has d£~l d ' solutions. Therefore the
congruence (2) holds only if g is a multiple of d ' y and it then

has d r l d ' I— | d„ mTd ' so lu tio n s. Therefore the theorem

is a lso true for r + 1 v aria b les. But the theorem is true for
one variable, and hence is always true.

2, a. We have a c^ m^ = 1 (mod m), a • hacP(m>“ 1 = b (mod m):

179
 b. We have

4 (p - 1) . . . (p - a + 1)
1 • 2 . . . (a - 1) ab (-1)
1 • 2 . . . (a - 1)

= b * 1 • 2 . . . (a - 1) (mod p),

and dividing by 1 • 2 . . . (a — 1), we obtain the required

theorem.
c. It is evidently su fficie n t to co n sid er the c a s e (2, b) == 1.
F or an appropriate choice of sign, b ± m = 0 (mod 4). L et
2 5 be the la rg e st power of 2 dividing b ± m. For 5 > k, we
have

b ± m
x ; ----- (mod m).
2*

If 8 < k, then

, s b ± m
2k~ i x = ----- r— (mod m).
2°

We rep ea t the analogous operation with this congruence, etc.

8 ) We consider (3, b) = 1. For an appropriate choice of
sign, we have b ± m = 0 (mod 3). L et 3^ be the largest
power of 3 dividing b ± m. F or 5 ^ k, we have

b ± m
x = ----- ;---- (mod m).
2k

If 8 < k, then

b ± m
3 k~ lx (mod m).
35

We repeat the analogous operation with this congruence, etc.

180
 y) L e t p be a prime divisor of the number a. Determine t by
the condition b + mt - 0 (mod p). L et p^ be the la rg e st power
of p dividing (a, b + mt), and let a = a tp^. We have

b + mt
a tx = ------ g---- (mod m).
P

If | a l | > 1, then we repeat th is operation with the new con­

gruence, etc.
T h is method is convenient for the c a s e in which a has small
prime factors.
3. Setting t =* [r], we write the congruences

a •0 s 0 (mod m),

a • 1 = y l (mod m).

a • t = y t (mod m),

a • 0 = m (mod m).

Arranging th e se congruences so that their right s id e s are in

order of in c re a se (cf. problem 4, a, ch. Il) and multiplying
termwise each congruence (except the la s t one) by its suc­
c e s s o r, we obtain t + 1 congruences of the form az = u
m
(mod m)\ 0 < | z | ^ r. Here 0 < u < — in at le a s t one con-
T
gruence. Indeed, u h as t + 1 > r v alu es, these valu es are
positive, and their sum is equal to m.
4, a, a ) T h is follows from the definition of symbolic
fractions.
ft) Here we can s e t b0 =* b + mt, where t is defined by the
condition b + mt = 0 (mod a); then the congruence ax = b
has a s solution an integer which rep resen ts the ordinary
K
fraction
a
181
 y) We have (b0 is a multiple of a,, d0 is a multiple of c)

b d K d0 b0c + ad0 be + ad
— + —' = — + — = -------------- #
a c a c ac ac

8) We have

b d 8a dQ Mo bd
a c a c ac ac

b, oc) We have (the congruences are taken modulo p)

p - l\ (p - 1) (p - 2) . . . (p - a)
a I 1*2 ... a

(-l)a 1 • 2 . . . a / x
1 •2 . . . a

Now problem 2, b is solved more simply as follows:

b M-D'-Mp - l) ... (p - (g - D) (mod p).

a 1 • 2 . . . (a - l ) a

j8) We have

2P - 2 p - 1 (p - 1) (p - 2)
= 1 + +
p 1*2 1*2*3

(p - 1) (p - 2) ... (p - (p - 2))
+ (mod p).
1*2 ... (p — 1)

5, a. The numbers s , s + l , . . . , s + n - l have no di­

visors in common with d. T he products s( s + 1) . . . (s +
+ n — 1) can be put in n K s e t s in a number of ways equal to
the number of ways that d can be decomposed into n relatively
prime factors, where order of the factors is taken into account
182
(problem 11, b, ch. II). L e t d = u lu2 . . . un be one of these
decom positions. The number of products such th a t s =
— 0 (mod Uj), s + 1 = 0 (mod u2), . . . , s + n — 1 s 0 (mod
a
un) is equal to — . Therefore the required number is equal to

b. T h is number is equal to

H p (< /)S d ; Sd -
d \a d ’

where k is the number of different prime divisors of the number

d. But we have

n ka n n
L n(d) 1 •«
d \a d Pi

6, a. All the values of x s a tisfy in g the first congruence are

given by the equation x = b2 + where t is an integer. In
order to choose from them those values which also s a tisfy the
second congruence, it is only n e c e ss a ry to choose those
values of t which s a tis f y the congruence

m,I = b2 — b2 (mod m2).

But this congruence is solvable if and only if b2 — b2 is a

multiple of d. Moreover, when the congruence is solvable,
the s e t of valu es of t s a tisfy in g it is defined by an equation
Artj
of the form t = t0 + -----1 ' t where t ' is an integer; and hence
d
the s e t of valu es of x satisfy in g the system considered in the
problem is defined by the equation

m.
x £>i + m 2 (t0 + — t ' Xi,j + ; x 1>2 — bj + m2t0.

183
 b. If the system

x = bl (mod mt), x = b2 (mod m2)

is solvable, the s e t of values of x s a tisfy in g it is represent­

able in the form x = x lf2 (mod m1>2). If the system

x = x lf2 (mod mlf2) t x = b3 (mod m3)

is so lvable, the s e t of values of x s a tisfy in g it is represent­

able in the form x = x lj2>J (mod ml>2>3). If the system

x s Xi ,2,3 (m°d ^ 1, 2,3). x - bA (mod

is so lvable, the s e t of valu es of x s a tisfy in g it is rep resen t­

able in the form x = x 2 2>3>4 (mod m l 2>3 4), etc.
7, a ) If x is replaced by —x (and hence x ' is replaced by

I a, b \
—x ' ) the sum I ------- 1 is not changed.
\ m J
ft) When x runs through a reduced residue system modulo m,
x ' also runs through a reduced resid u e system modulo m.
y) Setting x = hz (mod m), we find

I a, b h \ ( ahz + b z ' \ I ah, b \

(— ) - ( — )•

S) We have

/ °2» ^
TJX* m.

a xm2x + a2m2y + m2x + m2y

= E E cxp
x y

184
Setting rn2x ' + m,y ' : ' i we have

(a2m2x + a2miy ) ( m 2x ' + mty ' ) = a 1m\ + a2m\ (mod m,m2),

a if M / a2> 1 m\a2 + m\a2t 1

m. m.

which proves our property for the c a s e of two factors. The

generalization to the ca se of more than two factors is trivial.
8, T h e congruence

a0x n + a 2x n~l + . . . + on - a0(x - x t)(x ~ x 2) . . . (x - x n) =

= 0 (mod p)

has n s o lu tio n s. Its degree is le s s than n. Therefore all its

coefficients are m ultiples of p, and this is also exp ressed in
the congruences considered in the problem.
9, a. Corresponding to x in the sequence 2, 3, . . . , p — 2
we find a number x ' f different from it, in the same sequence
such that x x ' = \ (mod p); indeed, it would follow from
x = x ' that (x — 1) (x + 1) = 0 (mod p), and hence x = 1 or
x = p — 1. Therefore

2 • 3 . . . (p — 2) s 1 (mod p); 1 • 2 . . . (p — 1) = —1 (mod p).

b. L e t P > 2. Assuming that P h as a divisor u such that

1 < u < P, we would have l * 2 . . . ( / , - l ) + l s l (mod u).
10, a. We find h such that a 0h = 1 (mod m). The given
congruence is equivalent to the following one:

x n + a lh x n~1 + . . . + anh = 0 (mod m).

b. L e t Q(x) be the quotient and let Ii(x) be the remainder

resu ltin g from the division of x p — x by f(x). All the coef­
ficients of Q(x) and R(x) are integers, the degree of Q(x) is
p — n, the degree of R(x) is l e s s than n ,

xp - X = f(x)Q(x) + R(x).
185
Let the congruence fix) = 0 (mod p) have n so lu tio n s. T h ese
solutions will also be solutions of the congruence Rix) =
= 0 (mod p); therefore all the coefficients of R(x) are multi­
ples of p.
Conversely, let all the coefficients of R{x) be multiples of
p. Then f(x)Q{x) is a multiple of p for those v alu es of x for
which x p — x is also a multiple of p; therefore the sum of the
numbers of so lu tio n s of the congruences

fix) = 0 (mod p), Q{x) = 0 (mod p)

is no sm aller than p. L et the first have a , and le t the second

have jS solutions. From

a % n, ^ p - n, p ~ n, p ^ c x + /3

we deduce a = n, /3 = p — n.
p ~ 1
c. R aisin g the given congruence to the power ---------- term-
n
w ise, we find that the given condition is n e c e ss a ry . L et this
condition be s a tis f ie d ; it follows from x p — x =
p - i p - i

= x{xp~l — A n + A n — 1) that the remainder resulting

p-i
from the division of x p — x by x n — A is (A n — 1)*, where
p-i
A n ~ 1 is a multiple of p.
11. It follows from x£ = A (mod m), y n = 1 (mod m) that
(.x 0y ) n = A (mod m); here the products x 0y, corresponding to
incongruent (modulo m)y, are incongruent. It follows from
Xg = A (mod m), x n = A (mod w) that x n = x " (mod m), while,
defining y by the condition x s y x 0 (mod m), we have

y n = 1 (mod m).

186
Solutions o f the Problems for Chapter V
1. T h is congruence is equivalent to the following one:
(2ax + b ) 1 = b 2 — 4ac (mod m). Corresponding to each solu­
tion z = z 0 (mod m) of the congruence z 1 = b 1 — 4ac (mod m),
from 2ax + b = z 0 (mod m) we find a solution of the con­
gruence under consideration.

2, a. F or ( — » 1 we have a 2m+1 : 1 (mod p),

(am+l)a = a(mod p), x = ±am+l (mod p).

b. F o r f — 1 we have a 4m+a : 1 (mod p), a Jm+l = ±1

2 \
(mod p), a 2 m + 2 +a (mod p). Since — > - 1 we also
P /
have 2 4m+J s —1 (mod p). Therefore, for a certain s , having
one of the v alu es 0, 1, we find

a lm+a2 (4m+J) = a (mod p), * = ± a m+J2 (am+l )s (mod p).

a
C. L et p 2kh + 1, where k ^ 3 and h is odd, 1.
P
We have

a2 h s 1 (mod p), a2 h = ±1 (mod p), /V2 * = —1 (mod p).

T herefore, for some non-negative integer s 2 we find

a2 h/ysji - I (mod p), a2 h/VSjJ = ±1 (mod p);

and hence for some non-negative integer s t we find

k -S k -2 , lc -« . k- » , ,^
aa h/VSjJ = 1 (mod p), a a h/V*jJ s=±l(modp),

187
e tc .; finally we find

h+l
a hN2sk = ^ (mod p), x - +a 2 N Sk (mod p).

d. We have

1 • 2 . . . 2m(p ~ 2m) . . . (p — 2) (p — 1) + 1 = 0 (mod p),

(1 • 2 . . . 2m)1 + 1 = 0 (mod p).

3, a. The conditions for the solvability of congruence (1)

and (2) are deduced trivially (f, §2 and k, §2). The con­
-3
gruence (3) is solvable if and only if = 1. But
P
-3 P
, while
P 3

1, if p is of the form 6m + 1,

—1, if p is of the form 6m + 5.

b. F or any d is tin c t primes p t, p2, . . . , p of the form

4m + 1, the sm a lle st prime divisor p of the number
(2p,p2 . . . p k)2 + 1 is different from p it p2, . . . , p k , and since
(2p,p2 . . . p fc)2 + 1 = 0 (mod p), it is of the form 4m + 1.
c. F o r any d istin ct primes p t, pIf . . . , p k of the form
6m + 1, the sm a lle st prime divisor p of the number
(2p,p2 . . . pk)2 + 3 is different from pit p2, . . . , p k, and since
(2ptp 2 . . . pk)2 + 3 = 0 (mod p), it is of the form 6m + 1.
4. There are numbers in the first s e t which are congruent
p —1 p —1
to 1 * 1, 2 * 2, . . . , ---------- ° -----------, i.e . all the quadratic
2 2
resid u es of a complete system ; a number in the seco n d s e t is
a quadratic non-residue, by definition. But the second s e t
188
contains with this non-residue, all the products of the non-
residue with resid u es, i.e . it contains all the quadratic
non-residues.
5, a. In the number system to the b a se p, let

a “ Oa-iPa" 1 + . . . + a,p + a„

and le t the required solution (the s m a lle s t non-negative r e s i­

due) be

x * x a-\ Pa_1 + . . . + *,p + x 0.

We form the table:

a a-i a4 a3 a2 ai “o

2Xq^-Or-2 2xgXj 2 x 0x 3 2 x q X^ 2x0x,

2x ,X(Xr~2 . ... 2x,x3 2x , x2 A
2x2x a_j *2

where the column under a s c o n s is ts of numbers whose sum is

the coefficient of p s in the decomposition of the square of the
right side of (1) in powers of p. W,e determine x 0 by the
condition

A = a (mod p).

A - a0
Setting p,, we determine x, by the condition
P

p, + 2x0x 1 = a , (mod p).

p, + 2x0x l - a,
Setting ------------------------ = p2, we determine x 2 by the

189
condition

p2 + 2x0x 2 + x\ = a2 (mod p),

etc. F o r given x 0, the numbers x lf x2, . . . , x ^ are uniquely

determined s in c e (x o. p) = 1-
b. Here

a = a a-i 2a_1 + * . • + a 32s + a 22J + a t 2 + a0.

x = + •.• + x 22? + x 22 2 + x 22 + *o>

and we have the following table:

0-1
.... °4 °3 a0
0 a 2 a i

* 0* 0 - 2 *0*2 *0*2 *0*1 *0

....
* l* (2 r-J *1*2 A
* 2 * 0 -4 A

We only consider the ca s e a ^ 3. Since (a, 2) = 1, it follows

that c 0 = 1. Therefore x g = 1. Moreover a 2 = 0, and since
x 0x l + x[ = x 2 + x\ = 0 (mod 2), we must have a2 ^ 0 . For
x 2 there are two po ssib le v alues: 0 and 1. The numbers
xJt x 3 t • • •» * 0-2 are uniquely determined, while for x a^ 2 , there
are two possible v alues: 0 and 1. Therefore, for a ^ 3, we
must have a = 1 (mod 8), and then the congruence under con­
sideration has 4 so lu tio n s.
6. It is evident that P and Q are integers, where Q is con­
gruent modulo p to a number which we obtain by replacing a
by z 2, for which it is s u fficie n t to replace \fa by z. T here­
fore Q = 2a~1z a~1 (mod p); therefore (Q, p) = 1 and Q ' is
determined by the congruence Q Q ' = 1 (mod p). We have

P 2 ~ aQ2 *= (z + \fa )a(z — Va" ) a = (z2 — a)a = 0 (mod pa),

190
from which it follows that

( P Q ' Y = a iQ Q 'Y ee a (mod p a).

7. L et m =*= 2 ap “ ‘ . . . p ^ k be the canonical decomposition

of the number m. Then m can be represented in the form
m =*= 2 aab, where (a, 6) = 1, in 2k ways.

L et a = 0. It follows from (x — 1) (x + 1) = 0 (mod m),

that for certain a and b

x s 1 (mod a); x = —1 (mod b).

Solving this system , we obtain x = x 0 (mod m). Therefore

the congruence under consideration has 2k solutions.
L e t a = 1. For certain a and b

x = 1 (mod 2a); x = —1 (mod 26).

Solving this system , we obtain x = x 0 (mod m). Hence this

congruence has 2k solutions.
L e t a = 2. F o r certain a and 6

x = 1 (mod 2a); x = —\ (mod 26).

Solving this system , we obtain x = %0^mod — j . Therefore

our congrunece has 2k+i solutions.

L et a ^ 3. For certain a and 6, one of the sy stem s

x = 1 (mod 2a); x = —1 (mod 2a-16)

x = \ (mod 2a - l a); x s - 1 (mod 26)

is s a tisfie d . Solving one of th e se sy stem s, we obtain

191
 m
X = xa mod — . Therefore our congruence has 2k+i

solutions.
8, a. Defi n i n g x ' by the congruence x x ' = 1 (mod p), we
have

p^‘ ( x{x + k ) \ p-X

= L
£ \ P / x-l

p-‘ f l + k x ' y
Z — —

It is evident that 1 + lex' runs through all the resid u es of a

complete system , except 1. The required theorem follows
from this.
b. T he required equation follows from

1 P -J / / x\ (x + 1 \ ( x ( x + 1))
= (1 +f (t) +71f----- ) +°7

c. We have

S i x t E Z y- + -
*-o yi. y \ P

The part of the expression of the right corresponding to the

ca se y t * y, does not exceed X p Y . We consider the part cor­
responding to a pair of unequal values y x and y, where we as­
sume that y > 0 for the sak e of d efin iten e ss. Setting

192
xy + h = z (mod p), we reduce this part to the form

z — Z + kl 1 - —
p-i
y \ y t

from which (problem a) we find that it is <X in absolute

value. Therefore S 2 < X p Y + X Y 2 $ 2pXY.
d, a ) We have

p-1 Q-1 o -l (x + z x) (x + z )
s- L E E
X-0 Zj-o Zm0

For z, = z, summation with re s p e c t to x gives p — 1. For

Zj / z, summation with re sp e c t to x (problem a) gives —1.
Therefore

S = (p - IN? - Q{Q - 1) = (p - Q)Q.

fi) By the theorem of problem a ) we have

r ( ( ? o.5+o.sX )a < p Q . T <

y) Setting [VpT ] = Q, we apply the theorem of problem a).

Assuming there are no quadratic non-residues in the sequence
under consideration, we find that | S x | ^ Q — 1 for
x - M, M + 1, . . . , M + 2Q - 1 and hence

m Q - DJ ^ (p - Q)Q, 2(Q - l)2 < (Q + l)2 - Q,

Q 2 - 5Q < 0,

which is im possible for Q > 5.

193
 9, a. If m is representable in the form (1), then the solution

(5) z = z 0 (mod m)

of the congruence x = z y (mod m) is also a solution of the

congruence (2). We will say that our representation is a s ­
so c ia te d with the solution (5) of the congruence (2).
With each solution (5) of the congruence (2) is a s s o c ia te d
not le s s than one representation (1). Indeed, taking r = \Fm ,
we have

z P Q
— = — + — — ; (P, Q) = 1, 0 < Q ^ V^T , | e\ < 1.
m Q Q\ m

Therefore z 0Q = mP + r, where | r | < 'Jrn . Moreover, it

follows from (2) that | r | 2 + Q 1 = 0 (mod m). From this and
from 0 < | r | + Q 1 < 2m, we find

(6) m * | r | 2 + Q 2.

Here ( | r | , Q) = 1, since

r1 + Q1 (z0Q - mP)z0Q — rmP + Q 1

1 rP (mod Q).
m m

If | r | = r, th en the representation (6) is a s s o c ia te d with the

solution (5) beca u se r s z0Q (mod m). If | r | = —r, then the
representation m = Q 1 + | r | 2 is a s s o c ia te d with the solution
(5) because z\Q = z 0r (mod m), Q = z 0 \ r| (mod m).
No more than one representation (1) is a s s o c ia te d with each
solution (5). Indeed, if there were two representations
m = x 2 + y 2 and m = x \ + y o ° f the number m in the form (1)
a s s o c ia te d with a single solution (5), then x = z 0y (mod m),
x, s z 0y l (mod m) would imply that x y x = X{y (mod m). There-
194
fore x y x = x { y , from which it follows that x = x lt y — y x
because (x, y) = (xlf y t) * 1.
b. If m is representable in the form (3), then the solution

(7) z = z 0 (mod p)

of the congruence x = z y (mod p) is also a solution of the

congruence (4). We will say that this representation is a s ­
s o c ia te d with the solution (7) of the congruence (4).
Knowing a solution (7) of the congruence (4), there is no
more than one representation (3). Indeed, taking r = Vp” ,
we have

£o P d
— + — — ; (P, Q) - 1, 0 < Q $ y f r , M < 1.
P Q Q\/7

Therefore z0Q = r (mod p), where | r | < p. Moreover, it fol­

lows from (4) that | r | J + a Q 2 = 0 (mod p). From this and
from 0 < \ r \ 2 + a Q 1 < (1 + a)p it follows that we must have
1r J2 + 2Q 2 = p or | r |2 + 2Q 2 * 2p for a = 2. In the latter
c a s e , | r | is even, | r | = 2rlt p = Q 1 + 2r*. For a = 3 we
must have | r |2 + 3 ^ 2 = p, or | r |2 + 3 ^ 2 = 2p, or
| r |2 + 3Q 1 * 3p. T h e second case is im possible: modulo 4
the left sid e is congruent to 0 while the right sid e is congruent
to 2. In the third c a s e , | r | is a multiple of 3, | r | = 3 rlf
p = Q 2 + 3 r ta .
Assuming that two representations p = x 2 + a y 2 and
p * x \ + a y 2 of the number p in the form (3) are a s s o c ia te d
with a sin g le solution of the congruence (4), we find x = x lt
y = y x. Assuming that th e se representations are a s s o c ia te d
with different solutions of the congruence (4), we find x s z y
(mod p), x x s - zy (mod p) and hence x y x + x^y = 0 (mod p),
which is im possible because

0 < (xy, + x xy)2 < (x2 + y2) ( x 2 + y\) < p.

195
 c, a ) The terms of the sum S(k) with x = x l and x = —x l
are equal.
/3) We have

p -1 j x t ( x 2t 2 + k t 2) \ I t \

W ) ■ S (— ;— ) ■ (7 ) m

y) Setting p — 1 = 2p lf we have

Pl(S(r))2 + Pl(S(n) )2. = £ (S(rt2) Y + £ (S(nt2) ) 2 =

f-i t-i

p-i p-} p-i p-i ( x y ( x 2 + k ) ( y 2 + k)

- E CS(*)J* - £ £ £ M ^ ------------
k-l x-1 y-1 k-l \ p

For y different from x and p — x, the resu lt of summation with

resp ect to k is —2 [——| ; for y = x and y — p — x it is

\ Pl
(p - 2) ’ Therefore

Pl(S(r))2 + Pl(S(n))2 = 4pplf p = ( — S(r)J +*

10, a. We have

- DY2 =

- (ac, + y l\ f W ) { x 2 ± y 2\ F D ) ( x l - y l\fD‘){x2 + y 2\flT ) ~ k 2.

b. T aking any t1 such that rt > 1, we find integers x lt y 1

such that | y ^ f l ) — x x | < — , 0 < y ^ r „ and multiplying

this termwise by y,VZT + x x < 2y,VZT + 1, we find

\ x \ — Dy i| < 2 \ /ZT + 1. T a k i n g s > r2 so that
196
I Y iV d - X l \ > — , we find new integers x2, y 2 such that
Ti
\ x \ - D y\ | < 2 VTT + 1, etc.
It is evident that there e x is ts an integer k, not equal to
zero, in the interval —2 VTX — 1 < k < 2 \ZZT + 1 such that
there is an infinite s e t of pairs x, y with x 2 — D y 2 — k among
the pairs x lt y 2; x 2, y 2; . . among these pairs there are two
pairs q 2 and £lt q2 such that = ^2 (mod | k | ),
q j = r]2 (mod | k | ). Defining the integers ^0, rj0 by means of
the equation <f0 + q0V W = (£j + jj2V1T) (£2 + q2"J~D), we
have (problem a)

€1 ~ Dt)20 = | k \ 2; £0 = £ \ - Dq\ = 0 (mod k );

7o = ~£irh + €i*h s 0 (mod U l ) .

Therefore = £ | k \, q0 = q | k | , where £ and 77 are integers

and £ 2 — Dq2 = 1.
c. The numbers x, y defined by the equation (2) sa tisfy
(problem a) the equation (1).
Assum ing that there ex ist pairs of integers x, y satisfying
equation (1), but different from the pairs determined by the
equation (2), we have

U„ 4- y 0\ f D ) r < x + y\pD < {x0 + y0V^D~)r+l

for certain r = 1, 2 , ___ Dividing this termwise by

(*0 + ) r, we find

(3) 1 < X + Y \[W < x 0 + yoVTT ,

where (problem a) A and Y are integers determined by the

equation

x + yVZT
A' + Y V D = (x + y V l T ) ( x — y yJ~D)T
(x0 + y0VD" )r
197
and satisfy in g the equation

(4) X 2 - D Y 2 = 1.

But from (4) we obtain the inequality 0 < | X \ — \ Y \/1T \ < 1,

which along with the first inequality of (3) shows that X nnd
Y arc p ositive. Therefore the second inequality of (3)
contradicts the definition of x 0 and y0.
11, a, a) We have

- p - l Qr.1 [ t \ I ax{t - 1)'

k a,pi" - ^a,p^a,p “ £ £ ( ~ j e*P -
---------- ---------

For t =■= 1, summation with re s p e c t to x gives p — 1; for

f > l i t gives - [ — | . Therefore

\P

p~i t
^ a , p |2 = P ~ 1 “ £ f— ) * P, I U,a ,p '/P »

or

i u — p - 1 pPT~1
PZf 11 / x + t\ Ix\ I at
I U.,pr - u.,pu.,p - E L i ------ — exp I 2ni----
f-0 X -0 \ P \P \ P

For t = 0 summation with r e s p e c t to x gives p - 1; for

/ at
t > 0 it gives - e x p \2 ni ---- Therefore
\ P

p - i
' at
P ~ 1 - £ gxp 2ni — P, I ^a,p I » Vp" •
i P,

198
 /3) The theorem is evident for (a, p) « p. For (a, p) « 1 it
follows from

a p ~1 ax \ / ax \ / a \
V,a , p
7 £ T exp r v = 7

b, a ) L e t r run through the quadratic resid u es and let n run

through the quadratic non-residues, in a complete system of
re s id u e s . We have

Sa ,p = 1 -I- 2 £ exp \2ni----

Subtracting

0 - 1 + 7"1, exp j + 2^ e x p ^ / r i ----- j

from the la tte r term wise, we obtain the required equation.

0 ) We have

m—1 m—1 a ( t 2 + 2tx)

sa , J = Z ] Z1
t-0 X -0
cxp m

at*
For given t, summation with resp ect to x gives m exp I2n-t------
m

or 0 according a s 21 is divisible by m or not. F o r odd m we

have

l, I a •0 \
Sa ,m I “ m eXP I2™- — I = m'

199
For even m = 2m, we have

a •02 a ‘m,
= m exp 2ti‘- + exp 277i-
m m

Here the right side is equal to zero for odd m, and equal to
2m for even m,.
y) For any integer b we have

A x 1 + 2Abx
E
X -0
cxp 2ni -
m

and choosing b such that 2Ab = a (mod m), we again obtain

the re su lt considered in problem ft).
12, a. We have

, m—
af+ 1
g-i a(x - z)
m E Y ' $ (z) exp 2ct-i
- L L m
Z S■M

The part of the sum on the right corresponding to a = 0 is

equal to Q $ (z); the part corresponding to the remaining

Z

valu es of a is numerically (problem 11, c, ch. Ill)

M +e-i /
<A E
Sm1 S - M
exP (277t
\
< A m(ln m — 5).

b, a ) T h is follows from the theorem of problem 11, a, a)

and the theorem of problem a.
ft) T he inequality of problem a) gives R — A' = d\/~p In p.
Moreover it is evident that R + N = Q.
y) It follows from the theorem of problem 11, b, ft) that the
conditions of the theorem of problem a are s a tis fie d if we take
m * p, $(z) = 1, while z runs through the values z = x 1;
x * 0, 1, . . . , p - 1, But, among the values of z there is
200
one which is congruent modulo p to 0 and two congruent
modulo p with each quadratic residue of a complete residue
system , and hence

22 $ ( z ) = 2R, 22 $ ( z ) = p,

and we obtain

Q H-
2R * — p + (?VfT In p.
P

S) T h is follows from the theorem of problem 11, b, y) and

the theorem of problem a.
f) It follows from the theorem of problem 5) that the condi­
tions of the theorem of problem a are s a tis fie d if we set
m * p, $ (z) * 1, w hile z runs through the valu es z =« A x 2;
x = Mot M0 + 1, . . . , M0 + Q0 — 1* Therefore

11 * ( z ) = T, 2 ] $ ( z ) = Qo,

from which the required formula follows.

/ a. \
c. The part of the sum containing the terms with I — 1 =* 1
P I \
is equal to p ( R 2 + /Va), the remaining part is equal to —2pRN.
Therefore the whole sum is equal to p(R — N)1.
The part of the sum containing the terms with a = 0 is
equal to 0. The remaining part is numerically sm aller than
(problem 11, c, ch. Ill)

m +Q- i ( ax A f+ Q -i / —a<xy
£ exp ( 2771 — exp 2iri-------- <
L L
a-1 y-M \ P

< p 2 (In p)2.

201
Therefore p{R - N)2 < p 1 (In p)2, R - N < '/p In p.

Solutions o f the Problems for Chapter VI

1, a. If q is an odd prime and a p = 1 (mod q), then a be­

longs to one of the exponents 8 =* 1, p modulo q. For 5 =* 1
we have a = 1 (mod q), for 8 » p we have q — 1 = 2px
where x is an integer.
b. If q is an odd prime and a p + 1 = 0 (mod q), then
a 2p = 1 (mod q). Therefore a belongs to one of the exponents
8 * 1, 2, p, 2p modulo q. The c a s e s 8 = 1, p are im possible.
F or 8 = 2 we have a 2 = 1 (mod q), a + 1 = 0 (mod q). F or
8 = 2p we have q — 1 = 2 px where x is an integer.
c. The prime divisors of 2P - 1 are primes of the form
2px + 1. L et p lt p2t p be any k primes of the form
2px + 1; the number (ptp2 . . . Pk)p ^ 1 bas a prime divisor
of the form 2px + 1 which is different from p,, p2, . . . , p fc.
d. If q is a prime and 2 2 + 1 = 0 (mod q), then 2 2 s
s 1 (mod q). Therefore 2 belongs to the exponent 2n+l modulo
q , and hence q — 1 = 2n+lx where x is an integer.
2, It is evident that a belongs to the exponent n modulo
a n - 1. Therefore n is a divisor of cp(an - 1).
3, a. Assume that we arrive at the original s e q u en ce after
k operations. It is evident that the k- th operation is equiva­
lent to the following one: consider the numbers in the sequence

If 2j . . . y n 1^ n^ /if n — 1^ . . . f 2, If 2f . . .

...fn ~ If n f nf n ~ If . .. f 2f If 2f ...

in p la ces 1 , 1 + 2k , 1 + 2 • 2 k, . . . . Therefore the number 2

is in the 1 + 2k place. Therefore the condition considered in
the problem is n e c e ss a ry . But it is a lso sufficient, s in ce it
implies that we have the following congruences modulo 2n — 1:

1 s 1, 1 + 2k = 0, 1 + 2 • 2 k = - 1 , . . .

202
or

1 = 1, 1 + 2k = 2, 1 + 2 • 2k = 3, . . .

b. The solution is analogous to the solution of problem a.

g
4. The solution of the congruence x ° = -1 (mod p) belongs
5
to an exponent of the form ---- where § ' is a divisor of S.
S'
I
Here S ' is a multiple of d if and only if x d = 1 (mod p). Con­
sidering the S values of S ' and taking f = 1, we find that
V- , . . 5
S ' - 2-i p(d)Sd, where S ' is the required number and S d = — .
d\ 8 d

g
5, a. Here (3; example c, 45) we must have
2" + 1
= - 1 . T h is condition is s a tis fie d for g = 3,
I S
b. Here we cannot have 9 " l > S 2 = 1 (mod
\ 2P + 1 /
2p + 1). T h is condition is s a tis f ie d for our values of g.

C. Here we cannot have S \

1, g 4 = 1 (mod
4p + 1
4p + 1). T h is condition is s a tis fie d for g = 2.
g
d. Here we cannot have = 1, g 2 = 1 (mod
2np + 1
2np + 1). T h is condition is s a tisfie d for g = 3.
6, a, a ) The theorem is evident if n is a multiple of p - 1.
Assume that n is not d iv isib le by p — 1. If we disregard the
order, the numbers 1, 2, . . . , p — 1 are congruent modulo p to
the numbers g, 2g, . . . , (p - 1 )g, where g is a primitive root
modulo p. Hence

S n = gnS n (mod p), S n H 0 (mod p).

203
 ft) We have

x j x 1 + 1) P_1 P -1 P -1

L = £ ] * 1 Ge1 + 1) 1 (mod p)
. p x -l

from which (problem a ) ) we obtain the required result.

b. F o r p > 2, we have

1 - 2 . . . (p - l) = gi +*+- - - +P- » == / * “ = - 1 (mod p).

7, a. We have g lndfi‘a = a (mod p), indg a ind^ g x =

= ind^ a (mod p — 1), indgx a = a indg a (mod p — 1).
b. It follows from indg a = s (mod re), indgx a = a ind^ a
(mod p — 1) that ind^ a = a s = s t (mod re)..
8. L e t (re, p — 1) = 1. Determining u by the condition
reu = 1 (mod p - 1) we find the solution x = a u (mod p).
L e t re be a prime, p - 1 » n at, where a is a positive
integer and (t , re) = 1. If the congruence is possible, then
CT-1 . # Or-1 .
an 1 = 1 (mod p); if a > 1, then, noting that x = g n r
(mod p), r = 0, 1, . . . , re — 1 are ju s t all the solutions of
the congruence x n = 1 (mod p); for some rx = 0, 1, . . . , re —
we have
a - 2. a - i.
a" gn = 1 (mod p);

if a > 2, then for certain r2 = 0, 1, . . . , re — 1 we have

a - 3 , a -J . . a - i,
an *g n tri+n == 1 (mod p).

e tc .; finally, for certain r a_1 =* 0, 1, . . . , re — 1 we have

t n f r i + n J frj+ . . . + n a 1tr
a g a 1 = 1 (mod p).
204
Determining u and v by the condition tu — nv = —1, we obtain
n solutions:

v u / ( r l + n r J + . . . + n a 2 r cl_ , ' y ¥ n <1 1 t r . . .

x = a g = (mod p);

r — Oj 1 ^ f n 1.

L e t the prime n 2 divide (n , p — 1), n = n ^ , n2 > 1. Cor­

responding to each solution of the congruence y " 1 s a (mod p)
we obtain a solution of the congruence x n2 s y (mod p).
9, a. In this way we obtain cc0c 1. . . c fr = cp(m) characters,
b, a ) We have y ( l ) = R ° . . .R°k = 1.
ft) L et y i , y k\ y " , . . . , y ' ' be the index sy stem s of
the numbers a, and a,; then y ' + y " , . . . , y'k + y ” is an
index system for the number ajOa (c, §7).
y) For a l - a a (mod m), the indices of the numbers a a and
a a are congruent to one-another n.odulo c, . . . , ck respectively.
C. T h is property follows from

C|r-1
E E R7
r-°
E R7k ‘
y k.o

d. T h is property follows from

y(o) - E RJ ” ’ E R 7 k ‘
R Rk

e. L e t 0 ( a a) < 0. Then 0 ( a a) * 0 ( a t ) 0 (1 ). Therefore

0(1) - 1. Determining a ' by the condition a a ' = 1 (mod m),
we have 0 (a ) 0 ( a ' ) =* 1. Therefore 0 (a ) < 0 for (a, m) =*= 1.
F o r (alf m) = 1, we have

y (a ja ) y ( ° i)
y '
e 0 (a) 0 ( a ta) 0(at) 4- 0(«) ’
2 05
therefore, either Y"1 ------- * 0, or ijj{ax) = y ( a j) for all a,.
a V'(a)

But the first cannot hold for all y ; if it did, then we would
have H = 0, while II = cp{m) s in c e , summing over all char­
acters for given a, we have

cp(m), if a = 1 (mod m),

L
i/'(a) 0, otherw ise.

f, a) If R ' , . . . , R k and R " , . . . , R k' are v alues of

R, . . . , ft* corresponding to the ch aracters y ,(a ) and y 2(a),
then y i ( a ) y 2(a) is a character corresponding to the values
R ' R ' \ . . ' , R ' k R'k .
j8) When R, . . . , R k run through all the roots of the cor­
responding equations, then R ' R , . . . , R kR k run through the
same roots in some order.
y) Determining I ' by the condition I V = 1 (mod m), we have

X * (z) X

which is equal to cp(m) or 0, according as a = I (mod m) or

not.
10, a, a ) Defining x ' by the congruence x x ' = 1 (mod p),
we have

I ind (x + k) — I ind x
exp 27rr
L

p - i l ind (1 + k x ' )
- £ ex P 2n i----------------------- -1.
n

206
 ft) We have

e p~l *£;1 ^=,1 / I ind (x + zft - I ind (x + z)

S = £ £ £ ex P \ 2 m '
X-0 Z l “ 0 Z - 0

For z t = z, summing over x gives p - 1, and for z t unequal

to z, summation over x (problem a ) ) gives —1 . Therefore

S = ( p - 1 ) Q - Q ( Q - 1) = (p - Q ) Q .

y) L et Q be the number of integers of the sequence x + z;

z - 0, 1, 1 which are not divisible by p, while TnfX
is the number of integers of this s e q u en ce which are in the
s-th s e t . F in ally , let

P -1

u„ + s - £

We have

I n-1 Q-l I (ind (x + z) - s)

un £ £ exp 2ni
n ,_j

1 n“ ‘ / Zs
— £ eXP ( - 2 ^ — ] 5 J , n , x ,

^ n ,x < 4 (n “ 1} £ , S - — - I (p - <?)<?.
n /-I

Setting Q » [n V^" ], and assum ing that there are no numbers

of the s-th s e t in our sequence, we find that Un »*

Q ~ i

207
for x *= M, M 4- 1, . . . , M + Q + 1, and hence

Q- i n ~ 1
Q n
(p - Q)Q,

(n \ / p - 2)J < (raVp~ - Vp~ ) 2 ,

which is im possible.
b. L et p0 be the product of the different prime divisors of
the number p — 1, let Q x be the number of integers of the
sequence x + z; z = 0, 1, — 1 which are not divisible
by p, and le t Gx be the number of integers of the same
sequence which are primitive roots modulo p. F in ally , let

p(d) p - 1 1
P = w„ = — + GX 9
d \p 0 d cp(p - 1) P

P -1

«- z> X -0
2
x *

Talcing f( £) * 1 and letting £ run through the values

£ = ind (x + z); z = 0, 1, . . . , Q — 1, we obtain

S' * fi(d)Sd . Here S ' is the number of values of £ such

d \p o
that (£, p — 1) * 1 and hence S ' =■ Gx . Moreover, Sd is the
number of values of £ which are multiples of d and hence
Sd = T d t x (problem a, y) for s =« 0). Therefore

- ----- + 21 p ( d ) T dfX - £ p(d)U dfX t

P d \p 0 d\po

2k L udtxt a < 2k(P- Q)Q.

d \p 0

208
S etting Q * [P 2k Vp" ] and assum ing that there are no primi-

tive roots in our seq u en ce, we find that I| w x I\ ^ —

Q —
~ 1 for

x = Mr M + 1, . . . , M + Q ~ 1 and hence

Q — — < 2 - ( p - <?><?,

(P 2 k Vp - 2)’ < (/>2k V p - - - £ j = l .

which is im possible.
11, a, a ) . We have

p_l p - | I £ jn(J t \ / a (( _ I)*

^a,pl * £ Z! CXP I 2771---- ) eXP I 2771-------“

p -1 I k ind t \
p - 1 - exp 12/7-1----------- I - p.

j3) F o r (a, p) * p the theorem is evident. F or (a, p) * 1,

it follow s from

Ua,p =

~ k ind a\ p~1 ( k ind a x \ I ax \

(2n i----------------- j ^ ex P |2 n i------------ 1exp |2t7i-----j =

—k ind a
exp I 2rri - " i,P -

y) It is eviden t that A and B are in teg ers with |5 'f = A2 + B 1.

209
F or certain e, c% t " such th a t 11 | =« | e' | = ( " \ =» 1 we
have (problem ft ))

1
S=
e'J’p e Vp
p—1 ind zx + ind z zxx + z(x + 1)
■2ni- ) « P ( 2tri-
L L L * * * [ -
Z ^ 1 z-1 x -0

If z t + z is n o t equal to p, then summing over x gives zero.

T herefo re

<"Vp, i s r . P.
s - ‘t 7 “ p K

b, a ) F or given z , the congruence x n = z (mod p) is solv­

able only if ind z is d iv isib le by 5, and it then has 5 so lu tio n s.
T herefore, for S » 1 we have Sa>p => 0. If S > 1, then we have

p ~} | k ind z \ I az
S« )P = 1 + I ! E e x p — g— J ex P \ 2^ ~

F or k = 0, summation with re s p e c t to z gives —1; for k > 0

it gives a quantity whose m odulus is equal to \fp . The re­
quired re s u lt follow s from th is.
ft) Setting

x = u + p a~l v; u =•= 0, . . . , p®"1 - 1. v « 0, . . . , p - 1,

we have

ax
exp 27ti------ ] = exp(2nia(unp~s + nun~l p~i v ) .

For (u, p) = 1, summation with re s p e c t to v g iv es zero.

210
Therefore

S« ,p s * J2 exp(2niapn sx2) = p s_1 , Sa' pS * 0.

Xo*0

y) L e t p T be the la rg e st power of p dividing n. We have

s ^ r + 3. S ettin g

x * u + pa~l ~Tv ; u = 0, . . . , p s-1" T — 1, u * 0, p T+1 - 1,

we have

ax
exp 2771 = exp(277ia(u"p- s + nu"~1p~r~l v ) ) .

For (u, p) 1, summation with re s p e c t to v gives zero.

Therefore

P®-‘-1
„ n —1 c
S«, p ° exp P ^a.p® • Sa,P 0.
l
Xo“ 0

5) L e t m = p f l • • • Pkk be the canonical decom position of

the number m S etting

= m 1+VSa , m; v = — , m = pJM/j = Pk k

and defining a ,, . . . , a fc by the condition a s a 1rl/l + . . .

. . . + a kMk (mod m), we have (problem 12, d, ch. Ill)

Tei ’PJ4 • T
i<

F or 5 = 1 we have

l^a,p»l < p~l +vn ' f p < np

211
For 1 < s ^ n, in, p) = 1 we have

| r . lP. | - p - + * v ' « i .

F or 1 < s ^ n, (n, p) = p we have

| r . fP. | < p - s+^ p s < p $ n .

The c a s e s > n red u ces to the case s n sin ce

Ta,p* * p - s +s t/pn~lSa^pS—n » TBfPs- n. T herefore

| r . „ k c . ,

from which we obtain the required in eq u ality .

12, a. T h is follow s from the theorem of problem 11, a, a )
and the theorem o f problem 12, a, ch. V.
b. We have

M +g-i npi &(ind x ~ s)

Tn * l l , e x p 2771-------------------
x"M k m 0 \ n.

F or k = 0, summing with re s p e c t to x we obtain (); for

k > 0, we obtain a number w hose m odulus is <>/p~ In p. And
th is im plies the required formula.
c . T aking f{x) = 1 and le ttin g x run through the numbers
x = ind Mt ind {M + 1), . . . , ind (M + Q — 1), we find (prob­

lem 17, a, ch. II) S ' = p(d)Sd. Here S ' is the number of
d \p —1
x such that (x, p — 1) = 1; therefore S ' =* T. Moreover, S d is
the number of valu es of x which are m ultiples of d, i.e . the
number of re sid u e s of power d in the seq u en ce M, M + 1, ,
M + Q - 1. Then

Q
// = £ p(d)
d
+ Qd'fp In \ed\ < ! , < ? ! = o.
d \p —1
212
 d. It foil ow s from the theorem of problem a th at the condi­
tions of problem 12, a, ch. V are s a tis f ie d , if we s e t m = p — 1,
^ (z ) = 1, w hile we le t z run through the v alu es z = ind x;
x = M , M + 1, . . . , M + Q — 1. We then find (with Qx in
place of ())

£ ' 3>(z) = / , E *(*) = < ? , / = — Q + 0\[p (In p)1.

z z p ~ 1

13. A ssum e th a t there are no n on-residues not exceeding h.

The number of n-th power no n -resid u es among the numbers
1 , 2 , . . . , Q where

Q = Vp~ (In pY

can be estim ated by two m ethods: s ta rtin g from the formula of

problem 12, b and sta rtin g from the fact th at the non-residues
can only be numbers d iv isib le by prim es exceeding h. We find

i Y ‘" p + 2 l " l n p I i \
1 ---------< I n - -------------------------- + 0 ,
n 1 \ln p /
— In p + 2 In In p '
c

In In
1 + 4-
In p / I
0 < In p + o' In
ln In
1 + 2c-
In n

The im possibility of the la tte r ineq u ality for all su fficie n tly
large p proves the theorem.
14, a. We have
m—l m—1 a x ( y l - y)
sl < x 22 Z] Z] p W p W exP [2ni
m
x-0 yj-0 y- 0
213
F or given y, and y summation with re s p e c t to x gives
Xm | p(y) | or zero according as y, = y or not. Therefore

| s | J < XY m , |s U v ^ r .

b, a ) We have

1 „ _ / aunv n\
s= —cp(m)
— E„ ET x W y M exP I2\ ™- - - -m- - j

where u and v run through reduced resid u e sy ste m s modulo m.

Hence

1 m—i m-1 / axy

— — 2] E »^Wp(y)exP [2ir*—
x-o y-0 \ m

vU ) - 2! * ( “ )* p(y) - 2] * (« )•
u n i x ( m o d m ) v n = y (m o d m)

But we have (problem 11, ch. IV)

E I v(x) | ^ K cp(m), E I p(y) I ^ K cp(m).

T herefore (problem a)

| S | ^ —-— 'JK cp(m) K cp(m)m =» /C V^m .

cp(m)

/3) L et m =■* 2 ap “ ‘ . . . p “ k be the can o n ical decom position

of the number m. T he congruence x n = 1 (mod m) is equiva­
le n t to the system

x n = 1 (mod 2 a), x n s 1 (mod p®1), . . . , x n = 1 (mod p ^k).

214
L e t y(x) and y 0Gc) be the in d ices of the number x modulo 2a
(g, $6). The congruence x n = 1 (mod 2a ) is eq u iv alen t to the
system ny(x) = 0 (mod c), ny0{x) = 0 (mod c„). T he first
congruence of th is sy stem has at most 2 so lu tio n s, w hile the
seco n d has a t m ost n so lu tio n s. T herefore the congruence
x n = 1 (mod 2 a ) h as a t m ost 2n so lu tio n s. By b, §5, each of
the congruences x n = 1 (mod p ^ ) , . . . , x n = 1 (mod p£k ) have
no more than n so lu tio n s. T herefore

liLT Inn
K ^ 2n k+1 « 2 (2 k+1) n %2(r(m))rKT; K - 0 (m e ).

15, a. We have

p -i p-i a(tn - l) x n + bit - l)x

E L exp 277i-
f-l x-l

If tn = 1 (mod p), then summation with re s p e c t to x gives

p — 1 for f = 1 (mod p) and —1 in th e rem aining c a s e s .
O therw ise, taking z{t — l) -1 in place of x, we can rep resen t
the part of the double sum corresponding to given t in the form

/ bz \ a(tn - I) (t - l ) - nz n
/ ; exp I2t7i exp 2ni
* -1 \ P

and hence

p-1 p ~1 / auv
7^ r . p'(u) p(t>) exp 12/71----- >
u - l v-l \ P

where viu) is equal to the number of so lu tio n s of the con­

gruence (tn — 1) {t — l ) “ n s u (mod p), w hile | piv) | does
not exceed the number of so lu tio n s of the congruence
z n = v (mod p). T herefore viu) ^ 2nu |p ( v ) | < a .,

£ I viu) | ^ ip - 1)271!, £ |p(f)| ^ ip - l W

u-l v-l
215
Applying the theorem of problem 14, a, we find

|s | < p - 1 + V(p —l)2nj(p - lV^p < 2n1p 1 .

b, a) T h is follow s from problem a and the theorem of prob­

lem 12, a, ch. V.
ft) It follow s from the theorem of problem a ) th a t the con­
ditions of the theorem of problem 12, a, ell. V are s a tis fie d , if
we s e t m = p, 4>(z) = 1, where we le t z run through the v alu es
z = A x n; x = M0, M0 + 1, . . . , M0 + Q0 - 1. Then

£ ' < M z ) = T, = <?„,

from which we obtain the required formula.

c, a ) L et y = 4ay,(m od p). We have (problem 11, a, ch, V)

' a\ P~i / 4 a JxJ + 4abx + 4ac \ / 4ay.x

,7 s ■ £ (---------------; --------------- exp “ T

1 p-1 / z \ p ~} I z(4aJxJ + 4abx + 4ac + 4ay .x z 1

ul— E H
, p z>>1 \ P
Z>p
x-o
----------------------- ------

_ 1
p / —(b1 — 4 ac)z — 2hy1 — y x2z 1 \
= T"* exp \ 2 n i ----------------------------------------------
z-l \ P /

3 1
The la tte r sum (problem a) is num erically < — p* .
2
ft) T h is follow s from the theorem of problem a ) and the
theorem of problem 12, a , ch. V.

216
ANSWERS FOR THE NUMERICAL
EXERCISES
Answers for chapter I.

1 , a. 17.
b. 23.

2, a, a ) 84
15 19 e
5 ft) a
11 14 + 14 • 20 '

b, a ) S6
80 1002 e
739 + 739 • 1000 ‘
3. We obtain 22 fractio n s.
5, a. 2s *35 • l l 3.
b. 22 • 33 • 54 • 73 • l l 2 • 17 *23 *37.

Anstvers for chapter ll.

1. a. 1312.
b. 2ll# * 3s* • 531 • 71* • l l 12 • 13* • 17t • 19s • 23s • 29" x
x 314 • 373 • 413 • 432 * 472 • 532 • 59s • 612 • 67 • 71 x
x 73 • 79 • 83 • 89 • 97 • 101 • 103 • 107 • 109 • 113.
2. a. r (2800) = 30; S(2800) = 7688.
b. r (232 848) = 120; S(232 848) = 848 160.
3. The sum of all the v alu es is equal to 1.
4. a ) 1152; ft) 466 400.
5. T he sum of all the v alu es is equal to 774.
217
Answers for chapter III.

1, a. 70.
b. It is d iv isib le.
2, a. 33 • 52 • l l 2 • 2999.
b. 7 • 13 • 37 • 73 • 101 • 137 • 17 • 19 • 257.

Answers for chapter I V .

1, a. x = 81 (mod 337).
b. x s 200; 751; 1302; 1853; 2404 (mod 2755).
2, b. x = 1630 (mod 2413).
3, x = 94 + 111/; y - 39 + 471, w here t is any integer.
4, a. x = 1706, + 526, (mod 221); x = 131 (mod 221);
x = 110 (mod 221); x = 89 (mod 221).
b. x = 11 1516, + 11 8006, + 16 8756, (mod 39 825).
5, a. x = 91 (mod 120).
b. x = 8479 (mod 15 015).
6, x = 100 (mod 143); y = 111 (mod 143).
7, a. 3x4 + 2x3 + 3x* + 2x = 0 (mod 5).
b. x5 + 5x4 + 3xJ + 3x + 2 = 0 (mod 7).
8, x6 + 4x5 + 22x4 + 76x" + 70X2 + 52x + 39 = 0 (mod
101) .
9, a. x = 16 (mod 27).
b. x = 22; 53 (mod 64).
10, a. x = 113 (mod 125).
b. x = 43, 123, 168 , 248, 293 , 373 , 418, 498, 543 , 623
(mod 625).
11, a. x = 2, 5, 11, 17, 20, 26 (mod 30).
b. x = 76, 22, 176, 122 (mod 225).

Answers for chapter V.

1, a. 1, 2, 3, 4, 6, 8, 9, 12, 13, 16, 18.

b. 2 , 5, 6, 8, 13,14,15,17,18,19, 20, 22, 23, 24, 29,
31, 32, 35.
2, a. a) 0; 0 ) 2.
b. a ) 0; 0 ) 2.
218
 3, a. a ) 0; ft) 22.
b. a ) 0; ft) 2.
4, a. a ) x = 9 (mod 19); ft) x = 11 (mod 29); y) x s 14
(mod 97).
b. a ) x = 66 (mod 311); ft) x = 130 (mod 277);
y) x = 94 (mod 353).
5, a. x s 72 (mod 125).
b. x = 127 (mod 243).
6, a. x = 13, 19, 45, 51 (mod 64).
b. x s 41, 87, 169, 215 (mod 256).

Answers for chapter VI.

1, a. 6.
b. 18.
2, a. 3 , 3, 3.
b. 6, 6, 1687.
c . a ) 3; /3) 7.
5, a. a ) 0; ft) 1; y) 3.
b. a ) 0; ft) 1; y) 10.
6, a. a.) x = 40; 27 (mod 67), ft) x = 33 (mod 67),
y) x = 8, 36, 28, 59, 31, 39 (mod 67).
b. a ) x s 17 (mod 73); ft) x s 50, 12, 35, 23, 61, 38
(mod 73), y) x = 3 , 24 , 46 (mod 73).
7, a. a ) 0; /3) 4.
b. a ) 0; ft) 7.
8, a. a ) x e 54 (mod 101). ft) x = 53, 86, 90, 66, 8
(mod 101).
b, x = 59, 11, 39 (mod 109).
9, a. a ) 1, 4, 5, 6, 7, 9, 11, 16, 17; ft) 1, 7, 8, 11, 12, 18.
b. a ) 1, 6, 8, 10, 11, 14 , 23 , 26, 27 , 29, 31, 36; ft) 1, 7,
9, 10, 12, 16, 26, 33, 34.
10, a. a ) 7, 37; ft) 3 , 5, 12, 18, 19, 20, 26, 28, 29, 30,
33, 34.
b. a ) 3, 27, 41, 52; ft) 2, 6 , 7, ]0 , 17, 18, 23, 26, 30,
31, 35, 43, 44, 51, 54, 55, 59.

219
 TABLES OF INDICES
The Prime 3

N 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9

0 0 1 0 1 2

The Prime 5

A' 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9

0 0 1 3 2 0 1 2 4 3

The Prime 7

/V 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9

0 0 2 1 4 5 3 0 1 3 2 6 4 5

The Prime 11

A 0 1 2 3 4 5 6 7 8 9 1 0 1 2 3 4 5 6 7 8 9

0 0 1 8 2 4 9 7 3 6 D 1 2 4 8 5 10 9 7 3 6
1 5 1

The Prime 13

A’ 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9

0 0 1 4 2 9 5 11 3 8 0 1 2 4 8 3 6 12 11 9 5
1 10 7 6 1 10 7

220
 The Prime 17

A' 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9

0 0 14 1 12 5 15 11 10 2 0 1 3 9 10 13 5 15 11 16 14
1 3 7 13 4 9 6 8 1 8 7 4 12 2 6

The Prime 19

A’ 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9
0 0 1 13 2 16 14 6 3 8 0 1 2 4 8 16 13 7 14 9 18
1 17 12 15 5 7 11 4 10 9 1 17 15 11 3 6 12 5 10

The Prim e 23

A' 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9
0 0 2 16 4 1 18 19 6 10 0 1 5 2 10 4 20 8 17 16 11
1 3 9 20 14 21 17 8 7 12 15 1 9 22 18 21 13 19 3 15 6 7
2 5 13 11 2 12 14

T he Prim e 29

A’ 0 1 2 3 4 5 6 7 8 9 / 0 1 o4* 3 4 5 6 7 8 9

0 0 1 5 2 22 6 12 3 10 0 1 2 4 8 16 3 6 12 24 19
1 23 25 7 18 13 27 4 21 11 9 1 9 18 7 14 28 27 25 21 13 26
2 24 17 26 20 8 16 19 15 14 2 23 17 5 10 20 11 22 15

31

A 0 1 2 3 4 5 6 7 8 9 1 0 1 2 3 4 5 6 7 8 9
0 0 24 1 18 20 25 28 12 2 0 1 3 9 27 19 26 16 17 20 29
1 14 23 19 11 22 21 6 "(T 26 4 1 25 13 8 24 10 30 28 22 4 12
2 8 29 17 27 13 10 5 3 16 9 2 5 15 14 11 2 6 18 23 7 21
3 15

221
 The Prim e 37

N 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9
0 0 26 15 12 22 1 39 38 30 0 1 6 36 11 25 27 39 29 10 19
1 8 3 27 31 25 37 24 33 16 9 1 32 28 4 24 21 3 18 26 33 34
2 34 14 29 36 13 4 17 5 11 7 2 40 35 5 30 16 14 2 12 31 22
3 23 28 10 18 19 21 2 32 35 6 3 9 13 37 17 20 38 23 15 8 7
4 20

The 43

N 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9
0 0 27 1 12 25 28 35 39 2 0 1 3 9 27 38 28 41 37 25 32
1 10 30 13 32 20 26 24 38 29 19 1 10 30 4 12 36 22 23 26 35 19
2 37 36 15 16 40 8 17 3 5 41 2 14 42 40 34 16 5 15 2 6 18
3 11 34 9 31 23 18 14 7 4 33 3 11 33 13 39 31 7 21 20 17 8
4 22 6 21 4 24 29

The Prim e 47

N 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9
0 0 18 20 36 1 38 32 8 40 0 1 5 25 31 14 23 21 11 8 40
1 19 7 10 11 4 21 26 16 12 45 1 12 13 18 43 27 41 17 38 2 10
2 37 6 25 5 28 2 29 14 22 35 2 3 15 28 46 42 22 16 33 24 26
3 39 3 44 27 34 33 30 42 17 31 3 36 39 7 35 34 29 4 20 6 30
4 9 15 24 13 43 41 23 4 9 45 37 44 32 19

222
 The Prim e 53

A' 0 1 2 3 4 5 6 7 8 9 1 0 1 2 3 4 5 6 7 8 9
0 0 1 17 2 47 18 14 3 34 0 1 2 4 8 16 32 11 22 44 35
1 48 6 19 24 15 12 4 10 35 37 1 17 34 15 30 7 14 28 3 6 12
2 49 31 7 39 20 42 25 51 16 46 2 24 48 43 33 13 26 52 51 49 45
3 13 33 5 23 11 9 36 30 38 41 3 37 21 42 31 9 18 36 19 38 23
4 50 45 32 22 8 29 40 44 21 28 4 46 39 25 50 47 41 29 5 10 20
5 43 27 26 5 40 27

T he Prime 59

N 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9

0 0 1 50 2 6 51 18 3 42 0 1 2 4 8 16 32 5 10 20 40
1 7 25 52 45 19 56 4 40 43 38 1 21 42 25 50 41 23 46 33 7 14
2 8 10 26 15 53 12 46 34 20 28 2 28 56 53 47 35 11 22 44 29 58
3 57 49 5 17 41 24 44 55 39 37 3 57 55 51 43 27 54 49 39 19 38
4 9 14 11 33 27 48 16 23 54 36 4 17 34 9 18 36 13 26 52 45 31
5 13 32 47 22 35 31 21 30 29 5 3 6 12 24 48 37 15 30

The Prime 61

1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9
N \ °
0 0 1 6 2 22 7 49 3 12 0 1 2 4 8 16 32 3 6 12 24
1 23 15 8 40 50 28 4 47 13 26 1 48 35 9 18 36 11 22 44 27 54
2 24 55 16 57 9 44 41 18 51 35 2 47 33 5 10 20 40 19 38 15 30
3 29 59 5 21 48 11 14 39 27 46 3 60 59 57 53 45 29 58 55 49 37
4 25 54 56 43 17 34 58 20 10 38 4 13 26 52 43 25 50 39 17 34 7
5 45 53 42 33 19 37 52 32 36 31 5 14 28 56 51 41 21 42 23 46 31
6 30

The Prime 67

N 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9
0 0 1 39 2 15 40 23 3 12 0 1 2 4 8 16 32 64 61 55 43
1 16 59 41 19 24 54 4 64 13 10 1 19 38 9 18 36 5 10 20 40 13
2 17 62 60 28 42 30 20 51 25 44 2 26 52 37 7 14 28 56 45 23 46
3 55 47 5 32 65 38 14 22 11 58 3 25 50 33 66 65 63 59 51 35 3
4 18 53 63 9 61 27 29 50 43 46 4 6 12 24 48 29 58 49 31 62 57
5 31 37 21 57 52 8 26 49 45 36 5 47 27 54 41 15 30 60 53 39 11
6 56 7 48 35 6 34 33 6 22 44 21 42 17 34

223
 T he Prim e 71

N 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9
0 0 6 26 12 28 32 1 18 52 0 1 7 49 59 58 51 2 14 27 47
1 34 31 38 39 7 54 24 49 58 16 1 45 31 4 28 54 23 19 62 8 56
2 40 27 37 15 44 56 45 8 13 68 2 37 46 38 53 16 41 3 21 5 35
3 60 11 30 57 55 29 64 20 22 65 3 32 11 6 42 10 70 64 22 12 13
4 46 25 33 48 43 10 21 9 50 2 4 20 69 57 44 24 26 40 67 43 17
5 62 5 51 23 14 59 19 43 4 3 5 48 52 9 63 15 34 25 33 18 55
6 66 69 17 53 36 67 63 47 61 41 6 30 68 50 66 36 39 60 65 29 61
7 35

T he Prim e 73

N 0 1 2 3 4 5 6 7 8 9 1 0 1 2 3 4 5 6 7 8 9

0 0 8 6 16 1 14 33 24 12 0 1 5 25 52 41 59 3 15 2 10
1 9 55 22 59 41 7 32 21 20 62 1 50 31 9 45 6 30 4 20 27 62
2 17 39 63 46 30 2 67 18 49 35 2 18 17 12 60 8 40 54 51 36 34
3 15 11 40 61 29 34 28 64 70 65 3 24 47 16 7 35 29 72 68 48 21
4 25 4 47 51 71 13 54 31 38 66 4 32 14 70 58 71 63 23 42 64 28
5 10 27 3 53 26 56 57 68 43 5 5 67 43 69 53 46 11 55 56 61 13
6 23 58 19 45 48 60 69 50 37 52 6 65 33 19 22 37 39 49 26 57 66
7 42 44 36 7 38 44

The Prim e 79

N 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9

0 0 4 1 8 62 5 53 12 2 0 1 3 9 27 2 6 18 54 4 12
1 66 68 9 34 57 63 16 21 6 32 1 36 29 8 24 72 58 16 48 65 37
2 70 54 72 26 13 46 38 3 61 11 2 32 17 51 74 64 34 23 69 49 68
3 67 56 20 69 25 37 10 19 36 35 3 46 59 19 57 13 39 38 35 26 78
4 74 75 58 49 76 64 30 59 17 28 4 76 70 52 77 73 61 25 75 67 43
5 50 22 42 77 7 52 65 33 15 31 5 50 71 55 7 21 63 31 14 42 47
6 71 45 60 55 24 18 73 48 29 27 6 62 28 5 15 45 56 10 30 11 33
7 41 51 14 44 23 47 40 43 39 7 20 60 22 66 40 41 44 53

224
 The Prime 83

N 0 1 2 3 4 5 6 7 8 9 I 0 1 2 3 4 5 6 7 8 9

0 0 1 72 2 27 73 8 3 62 0 1 2 4 8 16 32 64 45 7 14
1 28 24 74 77 9 17 4 56 63 47 1 28 56 29 58 33 66 49 15 30 60
2 29 80 25 60 75 54 78 52 10 12 2 37 74 65 47 11 22 44 5 10 20
3 18 38 5 14 57 35 64 20 48 67 3 40 80 77 71 59 35 70 57 31 62
4 30 40 81 71 26 7 61 23 76 16 4 41 82 81 79 75 67 51 19 38 76
5 55 46 79 59 53 51 11 37 13 34 5 69 55 27 54 25 50 17 34 68 53
6 19 66 39 70 6 22 15 45 58 50 6 23 46 9 18 36 72 61 39 78 73
7 36 33 65 69 21 44 49 32 68 43 7 63 43 3 6 12 24 48 13 26 52
8 31 42 41 8 21 42

The Prime 89

N 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9

0 0 16 1 32 70 17 81 48 2 0 1 3 9 27 81 65 17 51 64 14
1 86 84 33 23 9 71 64 6 18 35 1 42 37 22 66 20 60 2 6 18 54
2 14 82 12 57 49. 52 39 3 25 59 2 73 41 34 13 39 28 84 74 44 43
3 87 31 80 85 22 63 34 11 51 24 3 40 31 4 12 36 19 57 82 68 26
4 30 21 10 29 28 72 73 54 65 74 4 78 56 79 59 88 86 80 62 8 24
5 68 7 55 78 19 66 41 36 75 43 5 72 38 25 75 47 52 67 23 69 29
6 15 69 47 83 8 5 13 56 38 58 6 87 83 71 35 16 48 55 76 50 61
7 79 62 50 20 27 53 67 77 40 42 7 5 15 45 46 49 58 85 77 53 70
8 46 4 37 61 26 76 45 60 44 8 32 7 21 63 llJ 33 10 30

The Prime 97

<V 0 1 2 3 4 5 6 7 8 9 / 0 1 2 3 4 5 6 7 8 9

0 0 34 70 68 1 8 31 6 44 0 1 5 25 28 43 21 8 40 6 30
1 35 6 42 25 65 71 40 89 78 81 1 53 71 64 29 48 46 36 83 27 38
2 69 5 24 77 76 2 59 18 3 13 2 93 77 94 82 22 13 65 34 73 74
3 9 46 74 60 27 32 16 91 19 95 3 79 7 35 78 2 10 50 56 86 42
4 7 85 39 4 58 45 15 84 14 62 4 16 80 12 60 9 45 31 58 96 92
5 36 63 93 10 52 87 37 55 47 67 5 72 69 54 76 89 57 91 67 44 26
6 43 64 80 75 12 26 94 57 61 51 6 33 68 49 51 61 14 70 59 4 20
7 66 11 50 28 29 72 53 21 33 30 7 3 15 75 84 32 63 24 23 18 90
8 41 88 23 17 73 90 38 83 92 54 8 62 19 95 87 47 41 11 55 81 17
9 79 56 49 20 22 82 48 9 85 37 88 52 66 39

225
 Table of primes < 4000 and their smelliest primitive roots

p g P g g g P g P g P g

2 1 179 2 419 2 661 2 947 2 1 229 2 1 523 2

3 2 181 2 421 2 673 5 953 3 1 231 3 1 531 2
5 2 191 19 431 7 677 2 967 5 1 237 2 1 543 5
7 3 193 5 433 5 683 5 971 6 1 249 7 1 549 2
11 2 197 2 439 15 691 3 977 3 1 259 2 1 553 3

13 2 199 3 443 2 701 2 983 5 1 277 2 1 559 19

17 3 211 2 449 3 709 2 991 6 1 279 3 1 567 3
19 2 223 3 457 13 719 11 997 7 1 283 2 1 571 2
23 5 227 2 461 2 727 5 1 009 11 1 289 6 1 579 3
29 2 229 6 463 3 733 6 1 013 3 1 291 2 1 583 5

31 3 233 3 467 2 739 3 1 019 2 1 297 10 1 597 11

37 2 239 7 479 13 743 5 1 021 10 1 301 2 1 601 3
41 6 241 7 487 9u 751 3 1 031 14 1 303 6 1 607 5
43 3 251 6 491 2 757 2 1 033 5 1 307 2 1 609 7
47 5 257 3 499 7 761 6 1 039 3 1 319 13 1 613 3

53 2 263 5 503 5 769 11 1 049 3 1 321 13 1 619 2

59 2 269 2 509 n4‘ 773 2 1 051 7 1 327 3 1 621 2
61 2 271 6 521 9 787 2 1 061 2 1 361 3 1 627 3
67 2 277 5 523 n 797 2 1 063 3 1 367 5 1 637 2
71 7 281 3 541 o 809 3 1 069 6 1 373 2 1 657 11

73 5 283 3 547 2 811 3 1 087 3 1 381 2 1 663 3

79 3 293 2 557 2 821 2 1 091 2 1 399 13 1 667 2
83 O
4. 307 5 563 2 823 3 1 093 5 1 409 3 1 669 2
89 3 311 17 569 3 827 2 1 097 3 1 423 3 1 693 2
97 5 313 10 571 3 829 2 1 103 5 1 427 2 1 697 3

101 2 317 2 577 5 839 11 1 109 2 1 429 6 1 699 3

103 5 331 3 587 O 853 2 1 117 2 1 433 3 1 709 3
107 2 337 10 593 9; 857 3 1 123 2 1 439 7 1 721 3
109 6 347 2 599 7 859 2 1 129 11 1 447 3 1 723 3
113 3 349 2 601 i 863 5 1 151 17 1 451 2 1 733 2

127 3 353 3 607 3 877 2 1 153 5 1 453 2 1 741 2

131 2 359 7 613 O 881 3 1 163 5 1 459 5 1 747 2
137 3 367 6 617 3 883 2 1 171 2 1 471 6 1 753 7
139 2 373 n 619 2 887 5 1 181 7 1 481 3 1 759 6
149 2 379 2 631 3 907 2 1 187 2 1 483 2 1 777 5

151 6 383 5 641 3 911 17 1 193 3 1 487 5 1 783 10

157 5 389 2 643 11 919 7 1 201 11 1 489 14 1 787 2
163 2 397 5 647 5 929 3 1 213 2 1 493 2 1 789 6
167 5 401 3 653 2 937 5 1 217 3 1 499 2 1 801 11
173 2 409 21 659 2 941 2 1 223 5 1 511 11 1 811 6

12*

226
 (c o n t i n u e d )

p g P g P g p g *> g P g P g

1 823 5 2 129 3 2 417 3 2 729 3 3 049 11 3 373 5 3 691 2

1 831 3 2 131 2 2 423 5 2 731 3 3 061 6 3 389 3 3 697 5
1 847 5 2 137 10 2 437 2 2 741 2 3 067 2 3 391 3 3 701 2
1 861 2 2 141 2 2 441 6 2 749 6 3 079 6 3 407 5 3 709 2
1 867 2 2 143 3 2 447 5 2 753 3 3 083 2 3 413 2 3 719 7

1 871 14 2 153 3 2 459 2 2 767 3 3 089 3 3 433 5 3 727 3

1 873 10 2 161 23 2 467 2 2 777 3 3 109 6 3 449 3 3 733 2
1 877 2 2 179 7 2 473 5 2 789 2 3 119 7 3 457 7 3 739 7
1 879 6 2 203 5 2 477 2 2 791 6 3 121 7 3 461 2 3 761 3
1 889 3 2 207 5 2 503 3 2 797 2 3 137 3 3 463 3 3 767 5

1 901 2 2 213 2 2 521 17 2 801 3 3 163 3 3 467 2 3 769 7

1 907 2 2 221 2 2 531 2 2 803 2 3 167 5 3 469 2 3 779 2
1 913 3 2 237 2 2 539 2 2 819 2 3 169 7 3 491 2 3 793 5
1 931 2 2 239 3 2 543 5 2 833 5 3 181 7 3 499 2 3 797 2
1 933 5 2 243 2 2 549 2 2 837 2 3 187 2 3 511 7 3 803 2

1 949 2 2 251 7 2 551 6 2 843 2 3 191 11 3 517 2 3 821 3

1 951 3 2 267 2 2 557 2 2 851 2 3 203 2 3 527 5 3 823 3
1 973 2 2 269 2 2 579 2 2 857 11 3 209 3 3 529 17 3 833 3
1 979 2 2 273 3 2 591 7 2 861 2 3 217 5 3 533 2 3 847 5
1 987 2 2 281 7 2 593 7 2 879 7 3 221 10 3 539 2 3 851 2

1 993 5 2 287 19 2 609 3 2 887 5 3 229 6 3 541 7 3 853 2

1 997 2 2 293 2 2 617 5 2 897 3 3 251 6 3 547 2 3 863 5
1 999 3 2 297 5 2 621 2 2 903 5 3 253 2 3 557 2 3 877 2
2 003 5 2 309 2 2 633 3 2 909 2 3 257 3 3 559 3 3 881 13
2 O il 3 2 311 3 2 647 3 2 917 5 3 259 3 3 571 2 3 889 11

2 017 5 2 333 2 2 657 3 2 927 5 3 271 3 3 581 2 3 907 2

2 027 2 2 339 2 2 659 2 2 939 2 3 299 2 3 583 3 3 911 13
2 029 2 2 341 7 2 663 5 2 953 13 3 301 6 3 593 3 3 917 2
2 039 7 2 347 3 2 671 7 2 957 2 3 307 2 3 607 5 3 919 3
2 053 2 2 351 13 2 677 2 2 963 2 3 313 10 3 613 2 3 923 2

2 063 5 2 357 2 2 683 2 2 969 3 3 319 6 3 617 3 3 929 3

2 069 2 2 371 2 2 687 5 2 971 10 3 323 2 3 623 5 3 931 2
2 081 3 2 377 5 2 689 19 2 999 17 3 329 3 3 631 21 3 943 3
2 083 2 2 381 3 2 693 2 3 001 14 3 331 3 3 637 2 3 947 2
2 087 5 2 383 5 2 699 2 3 O il 2 3 343 5 3 643 2 3 967 6

2 089 7 2 389 2 2 707 2 3 019 2 3 347 2 3 659 2 3 989 2

2 099 2 2 393 3 2 711 7 3 023 5 3 359 11 3 671 13
2 111 7 2 399 11 2 713 5 3 037 2 3 361 22 3 673 5
2 113 5 2 411 6 2 719 3 3 041 3 3 371 2 3 677 2

227