Homework 4

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Homework 4

Exercise 4.1. Let X = (X , d ) be a metric space, and A ⊂ X any subset. A limit point of A is a point x ∈ X
such that every ε > 0, we have (Bd (x, ε) r {x }) ∩ A , ∅. In words, every ε–ball around x intersects A in a
point other than x.
(i) Give, with proof, the set of all limit points of the subset A = {−1} ∪ (2, 4) ∪ {1/n : n ∈ N} ⊂ R.
(ii) Let y ∈ X be a limit point of a set A. Show that there exists a sequence of points {x n }n ∈N , with each
x n ∈ A, converging to y ∈ X . In particular, there is an inclusion

{limit points of A in X } ⊂ {limits in X of convergent sequences in A}.

Is this inclusion always an equality? That is, for any X and A, is every limit in X of a convergent
sequence in A also a limit point in the sense described above?
(iii) Prove that, for any A, X , the set A = A ∪ {limit points of A} is always a closed subset of X . The set A is
often called the closure of A in X .
Solution.
(i) The set of limit points is [2, 4] ∪ {0}: First, we check that every x ∈ [2, 4] ∪ {0} is indeed a limit point of
A. There are several cases:
(i) x ∈ (2, 4): Let ε > 0 and write r = min{ε, 1 − |x − 3|}. Then (x − r, x + r ) ⊂ B(x, ε ) ∩ (2, 4).
Consequently, for example x + r /2 ∈ (B(x, ε) r {x }) ∩ A.
(ii) x = 2: If ε > 0 and r = min{ε, 2}, then 2 + r /2 ∈ B(2, ε ) r {2} and 2 + r /2 ∈ A.
(iii) x = 4: Similarly, if ε > 0 and r = min{ε, 2}, then 4 − r /2 ∈ B(4, ε) r {4} and 4 − r /2 ∈ A.
(iv) x = 0: Suppose ε > 0. Let N ∈ N be an integer large enough that N > 1/ε. Then we have 1/N ∈ A
and 0 < 1/N < ε, i. e. 1/N ∈ (B(0, ε) r {0}) ∩ A.
Next, we need to verify that every limit point of A is in [2, 4] ∪ {0}. For this suppose x ∈ R r ([2, 4] ∪ {0}).
We again distinguish several cases:
(i) x < −1: If ε < |x + 1|, observe that B(x, ε ) ∩ A = ∅.
(ii) x = −1: We have (B(−1, 1/2) r {−1}) ∩ A = ∅.
(iii) −1 < x < 0: If ε < min{|x + 1|, |x |}, then B(x, ε ) ∩ A = ∅.
(iv) 0 < x < 1 and x = n1 ∈ A: If ε < n+1 1
, then B(x, ε) ∩ A = {x } because ( n+1
1
, x ) ∪ (x, n−1
1
) 1 A.
(v) 0 < x < 1 and x < A: In this case, there is some n ∈ N such that n+1 < x < n ; in fact n = b1/xc.
1 1

Then, for any ε < min{|x − n+1 1


|, |x − n1 |}, we have B(x, ε) ∩ A = ∅.
(vi) x = 1: We have B(1, 1/3) ∩ A = {1}.
(vii) 1 < x < 2: If ε < min{|x − 1|, |x − 2|}, then B(x, ε ) ∩ A = ∅.
(viii) 4 < x: For every ε < |x − 4| we have B(x, ε ) ∩ A = ∅.
So, no x < [2, 4] ∪ {0} can be a limit point of A.
(ii) Suppose y ∈ X is a limit point of A. Then for all n ∈ N the set Bd (y, n1 ) ∩ A is nonempty; so pick some
x n ∈ Bd (y, n1 ) ∩ A. Then {x n }n converges to y: Indeed, suppose ε > 0 and N ∈ N with N > ε1 . Then
d (y, x n ) < n1 ≤ N1 < ε whenever n ≥ N .
This shows that any limit point of A is a limit in X of a convergent sequence in A. The converse is false.
For example, −1 is not a limit point of {−1} ⊂ R, but the constant sequence {−1}n converges to −1 in R.
(iii) Suppose x < A. Then there must be some ε > 0 such that (Bd (x, ε) ∩ A) r {x } = ∅. Since in particular
x < A it must be the case that Bd (x, ε ) ∩ A = ∅. Now, suppose for contraction that some limit point
y ∈ X of A were to lie in Bd (x, ε ). Let δ > 0 be small enough such that Bd (y, δ ) ⊂ Bd (x, ε); such a
number δ exists because Bd (x, ε) is an open subset of X . But then Bd (y, δ ) ∩ A ⊂ Bd (x, ε) ∩ A = ∅
contradicting y being a limit point.
Hence, we conclude that Bd (x, ε ) ∩ A = ∅ or in other words Bd (x, ε ) ⊂ X r A. Since x ∈ X r A was
chosen arbitrarily this implies that X r A is open, that is, A is closed.

1
Exercise 4.2. Let X = {a, b, c, d}, and let T = {∅, X , {a, b, d }, {b, c, d}, {b, d}, {d}}.
(i) Prove that X = (X , T) is a topological space; that is, that T is a topology.
(ii) Prove that X is not metrizable. That is, prove there does not exist a metric d : X × X [0, ∞) such
that T = Td .
Solution.
(i) From the definition of T we immediately have ∅, X ∈ T.
Suppose we have a potentially infinite collection of sets Ui ∈ T, i ∈ I . Since there are only finitely
many elements of T, the union i ∈I Ui can actually be written as a union of finitely many elements of
S
T. This means we only have to prove that the union of any two elements of T is again in T. First,
A ∪ ∅ = A and A ∪ X = X for any subset A ⊂ X . Furthermore,

{a, b, d} ∪ {b, c, d} = X ∈ T
{a, b, d } ∪ {b, d} = {a, b, d} ∈ T
{a, b, d } ∪ {d} = {a, b, d} ∈ T
{b, c, d } ∪ {b, d} = {b, c, d} ∈ T
{b, c, d } ∪ {d} = {b, c, d} ∈ T
and
{b, d } ∪ {d} = {b, d} ∈ T.

For intersections of finitely many sets in T, we compute

{a, b, d } ∩ {b, c, d} = {b, d} ∈ T


{a, b, d} ∩ {b, d} = {b, d} ∈ T
{a, b, d } ∩ {d} = {d} ∈ T
{b, c, d} ∩ {b, d} = {b, d} ∈ T
{b, c, d } ∩ {d} = {d} ∈ T
and
{b, d } ∩ {d} = {d} ∈ T.

Also, for any subset A ⊂ X we have ∅ ∩ A = ∅ and X ∩ A = A. All of this put together shows that T is
a topology on X .
(ii) Suppose d is a metric on X . Then the set {d } ⊂ X would be closed because singletons are closed in any
metric space. Consequently, the complement X r {d } = {a, b, c} would have to be open in the topology
induced by {d}. But {a, b, c} < T. Hence, the topology induced by d cannot be T and therefore T can’t
be metrizable.

Exercise 4.3. Let X = (X , T) be a topological space, and A ⊂ X any subset. Suppose that for every x ∈ A,
there exists an open set U ⊂ X containing x which is contained in A. Then, prove that A is open in X .
Solution. For any x ∈ A pick an open set Ux ⊂ A with x ∈ Ux . Then x ∈A Ux = A: Since any Ux is a subset
S
of A, there union x ∈A Ux is still a subset of A. Conversely, if y ∈ A, then y ∈ Uy ⊂ x ∈A Ux . But then
S S
A = x ∈A Ux is a union of open subsets of X and therefore open itself.
S

Proposition. Let (Y , d ) be any metric space, and {x n }n ∈N any sequence of points in Y . Then the following
statements are equivalent:
(i) The sequence {x n } converges to y ∈ Y .
(ii) For any ε > 0, the ball Bd (y, ε ) contains all but finitely many of the x n .
(iii) Any open set U containing y also contains all but finitely many of the x n .

2
Proof. First, suppose that {x n } converges to y ∈ Y and let ε > 0. Then there is some natural number N ∈ N
such that d (x n , y) < ε for all n > N . This means that we have an inclusion {x n : n > N } ⊂ Bd (y, ε) of
sets and therefore {n ∈ N : x n < Bd (y, ε)} ⊂ {n ∈ N : n ≤ N }. But the latter set is finite. So the set
{n ∈ N : x n < Bd (y, ε)} must also be finite; in other words, Bd (y, ε) contains all but finitely many of the x n .
Next, suppose (ii) holds and let U be an open set containing y. Because U is an open subset of the metric
space Y , there is some ε > 0 such that Bd (y, ε) ⊂ U . Then, by (ii), Bd (y, ε) contains all but finitely many of th
x n . But because Bd (y, ε) ⊂ U this implies that U contains all but finitely many of the x n as well.
Finally, suppose (iii) holds and ε > 0. Then Bd (y, ε) is an open set ontaining y and therefore the set
{n ∈ N : x n < Bd (y, ε)} is finite. Consequently, it has a maximum element N . So, if n > N then we must have
x n ∈ Bd (y, ε). Since ε > 0 was arbitary this means that {x n } converges to y ∈ Y . 

Exercise 4.4. Let X = N ∪ {∞}, and consider the following collection of subsets of X :

T = {A : A ⊂ N} ∪ {B ∪ {∞} : B ⊂ N is the complement of a finite set}.

(i) Prove that T is a topology on X , and hence that X = (X , T) is a topological space.


(ii) Note that there is a natural inclusion i : N X . Prove that i is continuous, as a map from N with its
discrete topology and X with the topology above.
(iii) Let (Y , d ) be a metric space, which in particular induces a topological space (Y , Td ). Show that, for
each point y ∈ Y , there is a bijection between the following two sets:

A = {continuous maps f : X Y with f (∞) = y}


and
B = {convergent sequences {x n }n ∈N with limit y}.
Solution.
(i) First, ∅ ⊂ N and therefore ∅ ∈ T. Also, X = N ∪ {∞} and N r N = ∅ is finite, so X ∈ T as well. Now,
suppose {Ui }i ∈I is a family of sets Ui ∈ T. If ∞ < i ∈I Ui , then i ∈I Ui ⊂ N and so i ∈I Ui ∈ T.
S S S
Hence, suppose ∞ ∈ i ∈I Ui . Then there is some j ∈ I with ∞ ∈ U j and because U j ∈ T we conclude
S
that X r U j is finite. But then X r i ∈I Ui = i ∈I X r Ui ⊂ X r U j is also finite and we can conclude
S T
that i ∈I Ui ∈ T.
S
Finally, let U1 , . . . , Un be a finite collection of sets Ui ∈ T. Again, if ∞ < ni=1 Ui , then ni=1 Ui ⊂ N and
T T
Tn Tn
i=1 Ui ∈ T. So, suppose that ∞ ∈ i=1 Ui . Then ∞ T∈ Ui for all i = 1, . . . , n and because Ui ∈ T the
complements X r Ui are finite for all i. But then X r ni=1 Ui = ni=1 X r Ui is union of finitely many
S
Tn
finite sets and therefore finite itself. Hence, i=1 Ui ∈ T.
(ii) For any subset U ⊂ X the preimage i −1 (U ) ⊂ N is open in N because for the discrete topology any
subset is open. In particular, i −1 (U ) ⊂ N is open for U ⊂ X an open subset.
(iii) First, we define a map φ : A B. Take any f ∈ A, that is, f is a continuous map X Y with
f (∞) = y. Consider the sequence { f (n)}n ∈N . We claim that this sequence converges to y. For this, let
U ⊂ Y be an open set containing y. Because f is continuous the preimage f −1 (U ) is open and because
f (∞) = y ∈ U we also have ∞ ∈ f −1 (U ). Hence, by the definition of the topology on X there are
only finitely many n ∈ N with n < f −1 (U ), that is f (n) < U . So U contains all but finitely many of the
x n = f (n). Because U was arbitary, by the proposition we can conclude that {x n } converges to y and
therefore {x n } ∈ B. So, by setting φ( f ) = { f (n)}n ∈N we obtain a well-defined map φ : A B.
Conversely, we would like to define a map ψ : B A with ψ ({x n }n ∈N ) being the function n xn ,
∞ y. For this to make sense, we need to check that the function f : X Y with f (n) = x n and
f (∞) = y is in fact an element of A, that is, we need to check that this function is continuous. For this,
suppose U ⊂ Y is open. If y < U then ∞ < f −1 (U ) and f −1 (U ) is open in X . So, assume that y ∈ U .
Because {x n } converges to y, by the proposition, U contains all but finitely many of the x n . That is, the

3
set {n ∈ N : x n < U } is finite; but {n ∈ N : x n < U } = X r f −1 (U ) and this set being finite means that
f −1 (U ) is open in X . So f : X Y indeed is a continuous map.
In summary, we have two maps φ : A B and ψ : B A. To see that they are inverses of each
other, we compute

φ(ψ ({x n })) = {ψ ({x n })(n)}n ∈N = {x n }n ∈N


ψ (φ( f ))(n) = ψ ({ f (n)})(n) = f (n)

for f ∈ A and {x n } ∈ B. So, φ ◦ ψ = idB and ψ ◦ φ = idA and we can conclude in particular that
φ: A B is a bijection.

You might also like