myweb.ttu.

edu/bban
lev.ban@ttu.edu

Real Analysis
Byeong Ho Ban
Mathematics and Statistics
Texas Tech University

Chapter 4. Point Set Topology(Last update : March 2, 2018)

1. If card(X)≥ 2, there is a topology on X that is T0 but not T1 .

Proof. (Jan 14th 2018)

Let x ∈ X be given and let (X, T ) be a topological space with the topology given as below.
T = {E ⊂ X : E = X or x 6∈ E}
Then it is topology since ∅, X ∈ T , and for any subcollection S ⊂ T ,
[ [
U = X ∈ T if X ∈ S or x∈
6 U ∈T if X 6∈ S
U ∈S U ∈S

\ \
U =X if S = {X} otherwise x 6∈ U ∈T
U ∈S U ∈S

If y 6= z and y, z ∈ X \ {x}, then U = X \ {x, y} is the open set containing y but not containing z.
Thus, (X, T ) is T0 space. However, if we choose y 6= x, there is no open set containing x but not containing
y since the open set containing x can only be X. 

2. If X is an infinite set, the cofinite topology on X is T1 but not T2 , and is first countable iff X is countable.
Proof. (Jan 15th 2018)

Let the cofinite topology be T . Note that, for any x ∈ X, {x}c is open, so {x} is closed in T . Thus,
(X, T ) is T1 . However, if x, y ∈ X and x 6= y, for any open set U containing x but not y, since U c should
be finite, any set containing y and disjoint with U should be finite. Thus, there is no open set containing
y and disjoint with U . Therefore, (X, T ) is not Hausdorff(T2 ) space.

Suppose that (X, T ) is first countable and let x ∈ X. Then there is a countable neighborhood base
{Nn }∞
n=1 at x. Since (X, T ) is T1 , if y ∈ {x}c , {y} isSan open neighborhood of x. So there is n ∈ N
such that Nn ⊂ {y}c , so y ∈ n∈N Nnc . Thus, {x}c ⊂ n∈N Nnc . Since each Nnc is finite, {x}c should be
S
countable, so X = {x} ∪ {x}c is countable.

Suppose that X is countable and let x ∈ X. And let X \ {x} = {xi }i∈N N be defined as below.
N = {X \ {x1 } , X \ {x1 , x2 } , X \ {x1 , x2 , x3 } , X \ {x1 , x2 , x3 , x4 } , · · · }
Clearly, x ∈ E for any E ∈ N by construction. Also, if x ∈ U and U ∈ T , then U c should be finite. Thus,
∃k ∈ N such that U c ⊂ {xi }ki=1 , so X \ {xi }ki=1 ⊂ U , so N is the neighborhood base for T at x. Also, it is
clear that N is countable. Therefore, (X, T ) is first countable. 
1
2

3. Every metric space is normal. (If A,B are closed sets in the metric space (X, ρ)) consider the sets of
points x where ρ(x, A) < ρ(x, B) or ρ(x, A) > ρ(x, B)).
Proof. (Jan 14th 2018)

Let
U = {x ∈ X : ρ(x, A) < ρ(x, B)} Q = {x ∈ X : ρ(x, A) > ρ(x, B)}
Note that A ⊂ U and B ⊂ Q by the definition of metric, and also note that U and Q are disjoint because
of the below.
U ⊂ {x ∈ X : ρ(x, A) ≤ ρ(x, B)} = Qc
We need to prove that U and Q are open sets. In order to prove that, it suffice to prove that ρ(x, E) is con-
tinuous in x for any closed set E since, if f (x) = ρ(x, A)−ρ(x, B), f is continuous, and so U = f −1 ((−∞, 0))
and Q = f −1 (0, ∞) are open.

For any x, y ∈ X and  > 0, without loss of generality, assuming that ρ(x, E) − ρ(y, E) ≥ 0, observe
the below.
ρ(x, E) − ρ(y, E) = inf {ρ(x, z) : x ∈ E} − inf {ρ(y, z) : z ∈ E}
= inf {ρ(x, z) : x ∈ E} + sup {−ρ(y, z) : z ∈ E}
≤ inf {ρ(x, z) : x ∈ E} − ρ(y, z) +  (∃z ∈ E)
≤ ρ(x, z) − ρ(y, z) + 
≤ ρ(x, y) + 
Since  is arbitrary,
|ρ(x, E) − ρ(y, E)| ≤ ρ(x, y)
Therefore, ρ(x, E) is Lipschitz function, so it is continuous. 

4. Let X = R, and let T be the family of all subsets of R of the form U ∪ (V ∩ Q) where U,V are open in
the usual sense. Then T is a topology that is Hausdorff but not regular.
(In veiw of Exercise 3, this shows that a topology stronger than a normal topology need not be normal or
even regular)
Proof. (Jan 15th 2018)(Regular part is Incomplete)

Note that R and ∅ are open set in usual sense. So R = R ∪ (∅ ∩ Q) and ∅ = ∅ ∪ (∅ ∩ Q) are in T .
Also, if S = {Qi }i∈A ⊂ T where A is any index set, any element Qi in S is Ui ∪ (Vi ∩ Q) where each Ui
and Vi are open in usual sense. Then,
! ! !
[ [ [ [
Qi = Ui ∪ (Vi ∩ Q) = Ui ∪ Vi ∩ Q ∈ T
i∈A i∈A i∈A i∈A
S S
since i∈A Ui and i∈A Vi are open in usual sense.
Also, observe that, for any finite subcollection {Qi }ni=1 ⊂ T ,
n n
! n
! !
\ \ \
Qi = Ui ∪ Vi ∩ Q ∈ T
i=1 i=1 i=1

since ( i=1 Ui ) and ( ni=1 Vi ) are open in usual sense. Thus, T is a Topology on R
Tn T
 3

If x, y ∈ R and x 6= y, since every metric space is normal, so Hausdorff space, there are two disjoint
open sets(in usual sense) Q and U such that x ∈ Q and y ∈ U . And observe the below.
Q = Q ∪ (∅ ∩ Q) U = U ∪ (∅ ∩ Q)
Thus, Q, U ∈ T , so (X, T ) is Hausdorff space.

However, (X, T ) is not a regular space because of following counter example.

Note that Qc = (∅ ∪ (R ∩ Q))c is a closed set and let Q = U ∪ (V ∩ Q) be any open set containing Qc .
Then Qc ⊂ U 

5. Every separable metric space is second countable.

Proof. (Jan 15th 2018)

Let X be a separable metric space and let {xn }∞

n=1 be the countable dense subset of X. Let B(xn )
be defined as below.
B(xn ) = {B(xn , r) : r ∈ Q}
where B(xn , r) is an open ball centered at xn and having radius r. Note that B(xn ) is countable.
Now, let B be defined as below.
[∞
B= B(xn )
n=1

Also, note that B is countable since it is countable union of countable collections.

Now, we will show that B is a base for X.

Let x ∈ X be given. If x ∈ {xn }∞
n=1 , B(x) is clearly the neighborhood base at x. Otherwise, for any r ∈ Q,
there exists n ∈ N such that x ∈ B(xn , r). Thus,
B(x) = {B(xn , r) : r ∈ Q, n ∈ N, x ∈ B(xn , r)}
is the neighborhood base at x, since x ∈ E ∀E ∈ B by the construction . Let x ∈ U and U be open in the
metric space. Since U is open, for any xi ∈ {xn }n∈N there is r ∈ Q such that B(xi , r) ⊂ U . Since x ∈ U
and {xn }∞n=1 is dense in X, there is n ∈ N such that x ∈ B(xn , r) ⊂ U . Thus, B is the countable base for
X and so it is second countable. 

6. Let E = {(a, b] : −∞ < a < b < ∞}

a.E is a base for a topology T on R in which the members of E are both open and closed.

b.T is first countable but not second countable.

c.Q is dense in R with respect to T .(Thus the converse of Proposition 4.5 is false.)
Proof. (Jan 15th 2018)

(a) Let x ∈ R be given. Then note that x ∈ (x − 1, x] ∈ E. Also, if (a, b], (c, d] ∈ E and ∃x ∈ (a, b] ∩ (c, d] =
(c, d] (without loss of generality let c < b), then x ∈ (c, d] ∈ E. Thus, E is a base for a topology.
Now, recall that every non-empty member of T is a union of the members of E. Thus, since any member
in E is a union of itself, every member in E is open. Also, for any (a, b] ∈ E,observe the below.
[∞
c
(a, b] = (b + n − 1, b + n] ∪ (a − n, a − n + 1] ∈ T
n=1
4

Therefore, every member of E is both open and closed.

(b) Let x ∈ R be given. Let B(x) be given as below.

 
1
B(x) = (x − , x] : n ∈ N
n
Then, for all n ∈ N, x ∈ (x − n1 , x]. Now observe that if x ∈ E ∈ T , there is a subcollection S ⊂ T such
that E = (a,b]∈S (a, b] and x ∈ (a, b] for some (a, b] ∈ S. Let x − a = d then there is n ∈ N such that n1 < d
S

so (x − n1 , x] ⊂ (a, b]. Lastly, note that B(x) is countable. Thus, B(x) is a countable neighborhood base at
x, so T is first countable.
Let B be a base for T and let x ∈ R be given. Note that there exists Ux ∈ B such that x ∈ Ux ⊂ (x − 1, x].
Note that if x 6= y ∈ R, Ux 6= Uy since sup Ux = x 6= y = sup Uy . Therefore the map x 7→ Ux is injective,
thus B should be uncountable. Therefore, T is not a second countable.

(c) Let x ∈ R be given and let N ∈ T be a neighborhood of x. Then there is a subcollection S of E

such that
[
x∈N = E
E∈S

Thus, x ∈ E for some E ∈ S. Let E = (a, b], then Q ∩ (a, b] 6= ∅. Thus, N ∩ Q 6= ∅. It means that Q is
dense in R with respect to T . 

7. If X is a topological space, a point x ∈ X is called a cluster point of the sequence {xj } if for every
neighborhood U of x, xj ∈ U for infinitely many j. If X is first countable, x is a cluster point of {xj } iff
some subsequence of {xj } converges to x.
Proof. (Jan 15th 2018)

Suppose that X is first countable.

Let x be a cluster point of the sequence {xi }∞i=1 . Since X is first countable, there is a sequence of neigh-
borhood base {Un }∞ n=1 at x such that Un+1 ⊂ Un for all n ∈ N. Let’s pick n1 such that xn1 ∈ U1 .(It is
possible since x is a cluster point of the sequence). And generally, if we have picked nk , pick nk+1 such
that nk ≤ nk+1 and xnk+1 ∈ Uk+1 . Now, observe that if Q is a neighborhood of x, there is Un for some
n ∈ N such that x ∈ Un ⊂ Q, so xk ∈ Q for all k ≥ n. Thus, {xnk }∞ k=1 is a subsequence of the sequence
and xnk → x.

Conversely, suppose that {xnk }∞

k=1 is the subsequence of {xj } converges to x. Then, for any neighborhood
U of x, there exists K ∈ N such that xnk ∈ U for all k ≥ K. Since {xnk } ⊂ {xj }, ∞ = card(U ∩ {xnk }) ⊂
card(U ∩ {xj }). Thus, xj ∈ U for infinitely many j ∈ N, so x is a cluster point of {xj }. 

8. If X is an infinite set with the cofinite topology and {xj } is a sequence of distinct points in X, then
xj → x for every x ∈ X.
Proof. (Jan 15th 2018)

Let x ∈ X be given and let U be a neighborhood of x. Since x ∈ U o , so (U o )c is finite. Thus, since

each xj is distinct, there exists N ∈ N such that xj 6∈ U o ∀j ≥ N . In other words, xj ∈ U for all j ≥ N .
Therefore, xj → x. 
 5

9. If X is a linearly ordered set, the topology T generated by the sets {x : x < a} and {x : x > a} (a ∈ X)
is called the order topology.

a. If a, b ∈ X and a < b, there exist U, V ∈ T with a ∈ U , b ∈ V , and x < y for all x ∈ U and
y ∈ V. The order topology is the weakest topology with this property.

b. If Y ⊂ X, the order topology on Y is never stronger than, but may be weaker than, the relative topology
on Y induced by the order topology on X.

c. The order topology on R is the usual topology.

Proof. (Jan 15th 2018)

(a) Let a, b ∈ X with a < b be given. If there is c such that a < c < b, let U = {x : x < c} and
V = {x : x > c}, then a ∈ U and b ∈ V and x < y for all x ∈ U and y ∈ V .
If for any c ∈ X a ≥ c or c ≥ b, let U = {x : x < b} and V = {x : x > a}. Then a ∈ U and b ∈ V and
x < y for all x ∈ U and y ∈ V .

(b) Let
CX = {{x ∈ X : x < a} , {x ∈ X : x :> a} : a ∈ X}
CY = {{x ∈ Y : x < a} , {x ∈ Y : x :> a} : a ∈ Y }
Then the order topology on Y is T (CY ) and the relative topology on Y induced by the order topology
T (CX ) is TY = {E ∩ Y : E ∈ T (CX )}. Observe the below.
CY = {C ∩ Y : C ∈ CX } ⊂ TY
Thus, T (CY ) ⊂ TY .

(c) The order topology on R is generated by the the collection C below.

C = {(a, ∞), (−∞, a) : a ∈ X}
And C is the subbase for usual topology on R. 

10. A topological space X is called disconnected if there exist nonempty open sets U,V such that U ∩V = ∅
and U ∪ V = X; otherwise X is connected. When we speak of connected or disconnected subsets of X,
we refer to the relative topology on them.

a. X is connected iff ∅ and X are the only subsets of X that are both open and closed.
T S
b. If {Eα }α∈A is a collection of connected subsets of X such that α∈A Eα 6= ∅, then α∈A Eα is con-
nected.

c. If A ⊂ X is connected, then A is connected.

d. Every point x ∈ X is contained in a unique maximal connected subset of X, and this subset is closed.
(It is called the connected component of x)
Proof. (Jan 15th 2018)

(a) Suppose that there is a nonempty proper subset U ⊂ X such that U is both open and closed. Then
Q = U c is also open and closed, Q ∪ U = X, and Q ∩ U = ∅. Thus, X is disconnected.
Conversely, if X is disconnected, then there exist open sets U, V ⊂ X such that U ∩ V = ∅ and U ∪ V = X.
6

Then U = V c is closed set. Thus, U is both open and closed set.

(b) If A =let α∈A Eα T

S
is disconnected, then there are two open subsets U, V of A such that U ∪ V = A
and U ∩ V = ∅. Since α∈A Eα 6= ∅, there exists β ∈ A such that Eβ ∩ U 6= ∅ and Eβ ∩ V 6= ∅. Note that
the two intersections Eβ ∩ U and Eβ ∩ V are disjoint open sets such that the union is Eβ . Therefore, Eβ
is disconnected and it is a contradiction.

(c) If A is disconnected, there exist open sets U, V ⊂ A such that U ∩ V = ∅ and U ∪ V = A. Then
U ∩ A and V ∩ A are open in A, (U ∩ A) ∪ (V ∩ A) = A, and (U ∩ A) ∩ (V ∩ A) = ∅. Therefore, A is
disconnected.

(d) Let x ∈ X be given and let a collection C be defined as below.

C = {C : C is connected and x ∈ C} =let {Cα }α∈A
Then, by (b), x ∈ α∈A Cα =let H is connected and it is the connected component. If there is an connected
S
set D containing H, then D ∈ C, so H = D.
If H is not closed, there is a point x ∈ acc(H) such that x 6∈ H and there is a neighborhood N of x such
that N ∩ H 6= ∅. Since the neighborhood is connected, N ∪ H is another connected component which
implies contradiction. 

Sn Sn
11. If E1 , . . . , En are subsets of a topological space, the closure of 1 Ej is 1 Ej .
Proof. Observe that
n
[ n
[
Ej ⊂ Ej ∵ Ej ⊂ Ej ∀j ∈ {1, 2, . . . , n}
1 1
Sn
Since the closure of 1 Ej is the smallest closed set containing itself,
n
[ n
[
Ej ⊂ Ej
1 1
Conversely, note the below.
n
[
Ej ⊂ Ej ∀j ∈ {1, 2, . . . , n}
1
Therefore,
n
[ n
[
Ej ⊂ Ej
1 1
So,
n
[ n
[
Ej = Ej
1 1


12. Let X be a set. A Kuratowski closure operator on X is a map A 7→ A∗ from P(X) to itself
satisfying (i) ∅∗ = ∅, (ii) A ⊂ A∗ for all A, (iii)(A∗ )∗ = A∗ for all A, and (iv)(A ∪ B)∗ = A∗ ∪ B ∗ for all
A,B.

a. If X is a topological space, the map A 7→ A is a Kuratowski closure operator.(Use Exercise 11.)

 7

b. Conversely, given a Kuratowski closure operator, let F = {A ⊂ X : A = A∗ } and T = {U ⊂ X : U c ∈ F}.

Then T is a topology, and for any set A ⊂ X, A∗ is its closure with respect to T .
Proof. (Jan 15th 2018)(Incomplete at closure under union of (b) )

(a)
(i) ∅ = ∅
(ii) A ⊂ A
(iii) A = A
(iv) (A ∪ B) = A ∪ B by Exercise 11.
Therefore, the given map is Kuratowski closure operator

(b) Note that ∅ ∈ T . Also, from X ⊂ X ∗ , X ∗ = X, so X ∈ T . Let {Uα }α∈A is any subcollection of
T . Then observe the below.
!c !∗
[ \ \
Uα = Uαc Uαc
α∈A α∈A α∈A

And for any finite collection {Qi }ki=1 , observe the below.
k
!c k k
!∗ k k
\ [ [ [ [
Qi = Qci ∈ F since Qci = (Qci )∗ = Qci by property (iv)
i=1 i=1 i=1 i=1 i=1

Therefore, T is a topology.
Let A ⊂ X. By the property (ii), A ⊂ A∗ . Also, since (A∗ )∗ = A∗ , (A∗ )c ∈ T , A∗ is closed set containing
A. Let B be a closed set containing A, then B = B ∗ = (A ∪ B)∗ = A∗ ∪ B ∗ = A∗ ∪ B, thus A∗ ⊂ B. It
implies that A∗ is the smallest closed set containing A, so A∗ = A with respect to T . 

13. If X is a topological space, U is open in X, and A is dense in X, then U = U ∩ A.

Proof. (Jan 15th 2018)

It is clear that U ∩ A ⊂ U since U ∩ A ⊂ U and U ∩ A is the smallest closed set containing U ∩ A.

Let x 6∈ U ∩ A, then there is a neighborhood N of x such that N ∩ (U ∩ A) = ∅. Actually, N is also disjoint
from U . If not, ∃y ∈ N ∩ U and N ∩ U is an open set. Since A is dense, N ∩ (U ∩ A) = A ∩ (U ∩ N ) 6= ∅.
It is contradiction. Therefore, x 6∈ U , so (U ∩ A)c ⊂ (U )c and so U ⊂ U ∩ A.
Thus, U = U ∩ A. 

14. If X and Y are topological spaces, f : X → Y is continuous iff f (A) ⊂ f (A) for all A ⊂ X iff
f −1 (B) ⊂ f −1 (B) for all B ⊂ Y
Proof. (Jan 16th 2018)
 
Suppose that f is continuous and let A ⊂ X be given. Then f −1 f (A) is a closed set containing
 
A. Since A is the smallest set containing A, A ⊂ f −1 f (A) , so f (A) ⊂ f (A).

Now suppose that f (A) ⊂ f (A) for all A ⊂ X and let B ⊂ Y be given. Let A = f −1 (B), then A ⊂ X.
Then observe the below.
   
−1 −1
−1
f (B) = A ⊂ f f (A) = f f (f (B)) = f −1 (B)
−1
8

Lastly, suppose that f −1 (B) ⊂ f −1 (B) for all B ⊂ Y and let a closed set C ⊂ X be given. Then
observe the below.
f −1 (C) ⊂ f −1 (C) = f −1 (C) ⊂ f −1 (C)
Therefore, f −1 (C) = f −1 (C), so f −1 (C) is a closed set. Thus, f is continuous. 

15. If X is a topological space, A ⊂ X is closed, and g ∈ C(A) satisfies g = 0 on ∂A, then the extension
of g to X defined by g(x) = 0 for x ∈ Ac is continuous.
Proof. (Jan 16th 2018)

Let G be the extension and let C ⊂ X be any closed set. If 0 6∈ C, observe the below.
G−1 (C) = g −1 (C)
Since g is continuous in A, g −1 (C) is closed relative to A. Since A is closed in X, g −1 (C) is closed in X.
If 0 ∈ C, observe the below.
G−1 (C) = G−1 (C) ∩ A ∪ G−1 (C) ∩ Ac

= (g −1 (C) ∩ A) ∪ (G−1 ({0}) ∩ Ac )

= (g −1 (C) ∩ A) ∪ (Ac ∩ Ac )
By previous argument, since g −1 (C) is closed in X, and since A is closed, G−1 (C) is closed. 

16. Let X be a topological space, Y a Hausdorff space, and f,g continuous maps from X to Y.

a. {x : f (x) = g(x)} is closed.

b. If f = g on a dense subset of X, then f = g on all of X.

Proof. (Jan 16th 2018)

(a) Observe the below.

{x : f (x) = g(x)} = {x : (f − g)(x) = 0} = (f − g)−1 ({0})
Note that {0} is closed since Y is Hausdorff space, so T1 space. Since f − g is continuous, (f − g)−1 ({0})
is closed, so {x : f (x) = g(x)} is closed.

(b) Suppose that f = g on a dense subset A of X. Then A ⊂ {x : f (x) = g(x)} . Since {x : f (x) = g(x)}
is closed from (a), X = A ⊂ {x : f (x) = g(x)}. Therefore, f = g on all of X. 

17. If X is a set, F a collection of real-valued functions on X, and T the weak topology generated by F,
then T is Hausdorff iff for every x, y ∈ X with x 6= y there exists f ∈ F with f (x) 6= f (y).
Proof. (Jan 16th 2018)

Suppose that T be Hausdorff. Let x, y ∈ X with x 6= y be given. Then there exist disjoint open
sets U, V ⊂ X such that x ∈ U and y ∈ V . Note that the collection of the sets below
k
\
fj−1 (Uj )
j=1
 9

for any k ∈ N and any fj ∈ F with open sets Uk ⊂ R is a neighborhood base for T . 

18. If X and Y are topological spaces and y0 ∈ Y , then X is homeomorphic to X × {y0 } where the latter
has the relative topology as a subset of X × Y .
Proof. (Jan 17th 2018)

Let f : X × {y0 } 7→ X be defined by (x, y0 ) 7→ x and let πX and πY be coordinate maps such that
πX : X × Y 7→ X and πY : X × Y 7→ Y . Clearly, f is bijective. Note that, for any open set U in X,
−1
f −1 (U ) = U × {y0 } = (U × Y ) ∩ (X × {y0 }) = πX (U ) ∩ (X × {y0 }) is open.
On the other hands, note that any open set in X × {y0 } is the form of Q ∩ X × {y0 } such that Q is open
−1
in X × Y , so Q is any union of finite intersections of πX (U ) for any open set U ⊂ X or πY−1 (V ) for any
open set V ⊂ Y . Since f is bijection, it preserves any union and intersections. Thus, it suffice to prove
−1
that f (πX (U ) ∩ (X × {y0 })) and f (πY−1 (V ) ∩ (X × {y0 })) are open ∀U, V : open. Observe the below.
−1
f (πX (U ) ∩ (X × {y0 })) = f (U × {y0 }) = U
f (πY−1 (V ) ∩ (X × {y0 })) = f (X × {y0 }) = X if y0 ∈ V
f (πY−1 (V ) ∩ (X × {y0 })) = f (∅) = ∅ if y0 6∈ V
Since U, X, and ∅ are all open sets, and any union of finite intersections of open sets is open, f (R) is open
for any open sets in X × {y0 }. Thus, both of f and f −1 are continuous, so f is homeomorphism. It implies
that X is homeomorphic to X × {y0 }. 

Q
19. If {Xα } is a family of topological spaces, X = α Xα (with the product topology) is uniquely determined
up to homeomorphism by the following property : There exists continuous maps πα : X → Xα such that
if Y is any topological space and fα ∈ C(Y, Xα ) for each α, there is a unique F ∈ C(Y, X) such that
fα = πα ◦ F . (Thus, X is the category-theoretic product of the Xa lpha’s in the category of topological
spaces.)
Proof. 

Q
20. If A is a countable set and Xα is a first (resp. second) countable space for each α ∈ A, then α∈A Xα
is first (resp.second) countable.
Proof. 

21. If X is an infinite set with the cofinite topology, then every f ∈ C(X) is constant.
Proof. (Jan 17th 2018)

(We don’t know which topology is defined on C. Let’s just assume that the topology defined on C is
Hausdorff.)
If f is not constant, there exist x, y ∈ X with x 6= y such that f (x) 6= f (y). Since C is Hausdorff, there
exist disjoint open sets U, V ⊂ C such that f (x) ∈ U and f (y) ∈ V , so x ∈ f −1 (U ) and y ∈ f −1 (V ). Since
c
f is continuous, f −1 (U ) and f −1 (V ) are disjoint open sets. Since X is cofinite topological space, {f −1 (U )}
is finite and contains an infinite set f −1 (V ). And It is the contradiction. 

22. Let X be a topological space, (Y, ρ) a complete metric space, and {fn } a sequence in Y X such that
supx∈X ρ(fn (x), fm (x)) → 0 as m, n → ∞. There is a unique f ∈ Y X such that supx∈X ρ(fn (x), f (x)) → 0
as n → ∞. If each fn is continuous, so it f .
10

Proof. 

23. Give an elementary proof of the Tietze extension theorem for the case X = R
Proof. 

24. A Hausdorff space X is normal iff X satisfies the conclusion of Urysohn’s lemma iff X satisfies the
conclusion of the Tietze extension theorem.
Proof. (Jan 17th 2018)

Suppose that X is Hausdorff space.(Do we need Hausdorff?)

If X is normal, by Urysohn’s lemma, X satisfies the conclusion of the theorem.

If the Hausdorff space X satisfies the conclusion of Urysohn’s lemma, by Tietze extension theorem, X
satisfies the conclusion of the theorem.

Now, suppose that X satisfies the conclusion of Tietze extension theorem and let the closed subsets
A, B ⊂ X be given. And, for any given a, b ∈ R with a < b, define a function as below.
f (x) = a if x ∈ A f (x) = b if x ∈ B.
Note that f is a continuous function on A ∪ B since for any closed set C ⊂ [a, b],
f −1 (C) = ∅ if a, b 6∈ C f −1 (C) = A if a ∈ C but b 6∈ C
f −1 (C) = B if b ∈ C but a 6∈ C f −1 (C) = A ∪ B if a, b ∈ C.
and A and B are closed in A ∪ B. Note that A ∪ B is closed set in X. Therefore, by Tietze Extension
Theorem, there exists F ∈ C(X, [a, b]) such that F |A∪B = f . Let Q = ( a+b 4
, b] and V = [a, a+b
4
), then Q
and V are disjoint open sets in [a, b]. Since f is continuous, f −1 (Q) and f −1 (V ) are disjoint open sets and
B ⊂ f −1 (Q) and A ⊂ f −1 (V ). Therefore, X is normal. 

25. If (X, T ) is completely regular, then T is the weak topology generated by C(X).
Proof. 

26. Let X and Y be topological spaces.

a. If X is connected,(see Exercise 10) and f ∈ C(X, Y ), then f (X) is connected.

b. X is called arcwise connected if for all x0 , x1 ∈ X there exists f ∈ C([0, 1, X]) with f (0) = x0
and f (1) = x1 . Every arcwise connected space is connected.

c. Let X = {(s, t) ∈ R2 : t = sin(s−1 )} ∪ {(0, 0)} with the relative topology induced from R2 . Then X is
connected but not arcwise connected.
Proof. 

Q
27. If Xα is connected for each α ∈ A(see Exercise 10), then X = α∈A Xα is connected.(Fix x ∈ X and let
Y be the connected component of x in X. Show that Y includes {y ∈ X : πα (y) = πα (x) for all but finitely many α}
and that the latter set is dense in X. Use Exercise 10 and 18.)
 11

Proof. 

28.
Proof. 

29.
Proof. 

30. If A is a directed set, a subset B of A is called cofinal in A if for each α ∈ A there exists β ∈ B such
that β & α.

a. If B is cofinal in A and hxα iα∈A is a net, the inclusion map B → A makes hxβ iβ∈B a subnet of
hxα iα∈A .

b. If hxα iα∈A is a net in a topological space, then hxα i converges to x iff for every cofinal B ⊂ A there is
a cofinal C ⊂ B such that hxγ iγ∈C converges to x.
Proof. (Jan 18th 2018)

(a) First, note that xβ = xαβ . Now, let α ∈ A be given. By definition of cofinal, there exists β0 ∈ B
such that β0 & α. Then αβ = β & β0 & α whenever β & β0 . Therefore, hxβ iβ∈B is a subnet of hxα iα∈A .

(b) Let X be the topological space and U ⊂ X be a neighborhood of x. Then there exists α0 ∈ A such
that xα ∈ U whenever α & α0 . Let the cofinal B ⊂ A be given, then there is β0 ∈ B such that β0 & α0 .
Then, again, there is γ0 ∈ C such that γ0 & β0 & α0 . Then for any γ & γ0 , γ & α0 , thus, xγ ∈ U . Thus,
xγ → x. 

31. Let hxn in∈N be a sequence.

a. If k → nk is a map from N to itself,

H then hxnk ik∈N is a subnet of hxn i iff nk → ∞ as k → ∞,
and it is a subsequence (as defined in 0.1) iff nk is strictly increasing in k.

b. There is a natural one-to-one correspondence between the subsequences of hxn i and the subnet of hxn i
defined by cofinal sets as in Exercise 30.
Proof. (Jan 18th 2018)

(a) Suppose that hxnk i is the subnet. Let k ∈ N be given with nk . Then, by definition of subnet, there
is k0 ∈ N such that nk1 & nk whenever k1 & k0 . It means that nk → ∞ as k → ∞.
Conversely, let n ∈ N be given. Since nk → ∞ as k → ∞, there exists k0 ∈ N such that nk & n whenever
k & k0 . Thus, hxnk i is a subnet of the net.

(b) 
12

32. A topological space X is Hausdorff iff every net in X converges to at most one point. (If X is not
Hausdorff, let x and y be distinct points with no disjoint neighborhoods, and consider the directed set
Nx × Ny where Nx , Ny are the families of neighborhoods of x,y.)
Proof. (Jan 18th 2018)

Suppose that there is a net hxα iα∈A which converges to more than one point. Let x and y be two of
them. Since X is a Hausdorff space, there are two disjoint open sets U and V such that x ∈ U and y ∈ V .
Since the xα → x, there is α0 such that xα ∈ U for every α & α0 . Then for any β0 ∈ A there exists β such
that xβ 6∈ V even if β & β0 . It means that the net does not converges to y and it is a contradiction.

Conversely, suppose that X is not Hausdorff. Then we can choose x and y which are distinct points with
no disjoint neighborhoods. And let Nx and Ny are the families of neighborhoods of x and y respectively.
And define a binary relation . on Nx × Ny as below.
(U, V ) . (Q, G) iff Q ⊂ U and G ⊂ V
Then it is clear that Nx ×Ny is a directed set with the binary relation .(think of reverse inclusion). And
let’s pick x(U,V ) ∈ U ∩ V since each U, V are not disjoint each other. And let N and M be neighborhoods
 Then x(U,V ) ∈ N for any (U, V ) & (N, V ) and x(U,V ) ∈ M for nay (U, V ) & (U, M ).
of x and y respectively.

Thus, the net x(U,V ) (U,V )∈Nx ×Ny converges to two distinct points x and y. 

33. Let hxα iα∈A be a net in a topological space, and for each α ∈ A let Eα = {xβ : β & α}. Then x is a
T
cluster point of hxα i iff x ∈ α∈A Eα .
Proof. (Jan 18th 2018)

Let α ∈ A be given and let U be a neighborhood of x. Since x is a cluster point of hxα i, thus, there is
β & α such that xβ ∈ U . Since xβ ∈ Eα , (U \ {x})T∩ Eα 6= ∅. Therefore, x is an accumulation point of Eα ,
so x ∈ E α . Since α was arbitrarily selected, x ∈ α∈A E α .
T
Let U be a neighborhood of x. Note that, since x ∈ α∈A E α , Eα ∩ U 6= ∅ ∀α ∈ A. Let xα ∈ U , then,
since there is β & α and Eβ ∩ U 6= ∅, there exists γ & β such that xγ ∈ U . Since for any xα ∈ U there
exits xβ ∈ U with β & α, xα ∈ U infinitely many α ∈ A. Thus, hxα i is frequently in U , so x is a cluster
point of the net. 

34. If X has the weak topology generated by a family F of functions,

Q then hxα i converges to x ∈ X iff
hf (xα )i converges to f (x) for all f ∈ F.(In particular, if X = i∈I Xi , then xα → x iff πi (xα ) → πi (x)
for all i ∈ I.)
Proof. (Jan 19th 2018)(Nearly complete)

If hxα i converges to x, then by the theorem in the book, f (hxα i) converges to f (x) whenever f is con-
tinuous.

Conversely, suppose that there exists a net hxα i such that f (xα ) → f (x) but xα 6→ x. Let U be a
neighborhood of x, then xα is eventually in U . So, for any α there exists β ∈ A such that
xα ∈ U but xβ 6∈ U
Note that f −1 (Q) : Q ∈ Nf (x) is a neighborhood base at x ,where Nf (x) is a collection of neighborhoods


of f (x), since f is continuous. Thus, there exists V ∈ Nf (x) such that f −1 (V ) ⊂ U . Since there exists
α0 ∈ A such that f (xα ) ∈ V for any α & α0 , xα ∈ f −1 (V ) ⊂ U for any α & α0 . And it is contradiction
... 
 13

35. Let X be a set and A the collection of allPfinite subsets of X, directed by inclusion. Let f : X → R be an
arbitrary function, and for A ∈ A,
P let zA = x∈A f (x). Then the net hzP A i converges in R iff {x : f (x) 6= 0}
is a countable set {xn }n∈N and ∞ 1 |f (x n )| < ∞, in which case zA → ∞
1 f (xn ). (Cf. Proposition 0.20)

Proof. 

36. Let X be the set of Lebesgue measurable complex-valued functions on [0, 1]. There is no topology T
on X such that a sequence hfn i converges to f with respect to T iff fn → f a.e.(Use Corollary 2.32 and
Exercises 30b and 31b.)
Proof. 

37. Let 00 denote a point that is not an element of (−1, 1) , and let X = (−1, 1) ∪ {00 }. Let T be the
topology on X generated by the sets (−1, a), (a, 1), [(−1, b) \ {0}] ∪ {00 }, and [(c, 1) \ {0}] ∪ {00 } where
−1 < a < 1, 0 < b < 1, and −1 < c < 0. (One should picture X as (−1, 1) with the point 0 split in two.)

a. Define f, g : (−1, 1) → X by f (x) = x for all x, g(x) = x for x 6= 0, and g(0) = 00 . Then f and g are
homeomorphisms onto their ranges.

b. X is T1 but not Hausdorff, although each point of X has a neighborhood that is homeomorphic to
(−1, 1) (and hence is Hausdorff ).

c. The sets − 12 , 12 and − 12 , 12 \ {0} ∪ {00 } are compact but not closed in X, and their intersection is
   

not compact.
Proof. (Jan 20th 2018)

(a) With the a, b and c given by the problem, observe the below.
f −1 ((−1, a)) = (−1, a) : open f −1 ((a, 1)) = (a, 1) : open
f −1 ([(−1, b) \ {0}] ∪ {00 }) = (−1, 0) ∪ (0, b) : open f −1 ([(c, 1) \ {0}] ∪ {00 }) = (c, 0) ∪ (0, 1) : open
g −1 ((−1, a)) = (−1, a) or (−1, 0) ∪ (0, a) : open g −1 ((a, 1)) = (a, 1) or (a, 0) ∪ (0, 1) : open
g −1 ([(−1, b) \ {0}] ∪ {00 }) = (−1, b) : open g −1 ([(c, 1) \ {0}] ∪ {00 }) = (c, 1) : open
Thus, f and g are continuous.
Also, for all a ∈ R observe below.
f ((a, ∞)) = (a, 1) or ∅ : open g((a, ∞)) = (a, 1), (a, 0) ∪ (0, 1) or ∅ : open
f ((−∞, a)) = (−1, a) or ∅ : open g((−∞, a)) = (−1, a), (−1, 0) ∪ (0, a) or ∅ : open
Thus, f −1 and g −1 are also continuous.

Therefore, f and g are Homeomorphisms onto each of their ranges.

(b) If x, y ∈ (−1, 1) with (without loss of generality) x < y, trivially, y ∈ ( x+y

2
, 1) and x ∈ (−1, x+y2
)
0 0
are the disjoint open sets. If, without loss of generality, x = 0 and y ∈ (−1, 1), then [(−1, y) \ {0}] ∪ {0 }
contains x but not y, and for any b ∈ (y, 1), (−1, b) contains y but not x. Therefore, X is T1 .

However, it is not Hausdorff space. Think of when x = 0 and y = 00 . Note that any open set con-
taining x is A = (a, b) for some a, b ∈ (−1, 1). Further note that any open set containing y is a form of
14

B = (c, 0) ∪ (0, d) ∪ {00 } for some c ∈ (−1, 0) and d ∈ (0, 1). And lastly note that A ∩ B 6= ∅. Thus, X is
not Hausdorff space.

(c) (Incomplete) 

38. Suppose that (X, T ) is a compact Hausdorff space and T 0 is another topology on X. If T 0 is strictly
stronger than T , then (X, T 0 ) is Hausdorff but not compact. If T 0 is strictly weaker than T , then (X, T 0 )
is compact but not Hausdorff.
Proof. (Jan 20th 2018)

Suppose that T ( T 0 and let x, y ∈ X with x 6= y. Then there are two disjoint open sets U, V ∈ T ⊂ T 0
such that x ∈ U and y ∈ V . Thus, (X, T 0 ) is Hausdorff. However, (X, T 0 ) is not compact. In fact, if it
is compact, T = T 0 by following argument. If U ∈ T 0 , U c is closed in (X, T 0 ), so compact. Note that an
identity map from (X, T 0 ) to (X, T ) is continuous map. So, U c is compact in (X, T ) so closed since (X, T )
is Hausdorff. Therefore, U is open so U ∈ T . Thus, T = T 0 .

If T 0 ( T , then (X, T 0 ) is compact. Let O = {U } ⊂ T 0 be an open cover

S of X. Since O ⊂ T 0 and (X,0 T )
is compact, there exists an finite open subcover U ⊂ O such that X ⊂ Q∈U Q. Since U ⊂ T , (X, T ) is
also compact. However, it is not Hausdorff. Let’s assume that (X, T 0 ) is Hausdorff. Note that an identity
map from (X, T ) to (X, T 0 ) is continuous map. Suppose that U ∈ T , then U c is closed set, so compact.
Thus, f (U c ) = U c is compact in (X, T 0 ), so closed. It implies that U ∈ T 0 . Therefore, T = T 0 and it is a
contradiction. 

39. Every sequentially compact space is countably compact.

Proof. (Jan 21st 2018)

Let {Un }∞
n=1 be given countable open cover Sof X. If X is not countably compact, we can construct a
sequence {xn }∞ n c
n=1 by picking each x n from ( k=1 Uk ) for all n ∈ N. Since X is sequentially compact,
{xn }n=1 has convergent subsequence xnk converging to x ∈ X. Since {Un }∞
∞
n=1 is a covering of X, there
exists m ∈ N such that x ∈ Um . Thus, xn should be frequently in Um . However, from the construction,
xi ∈ Um only for i < m, which is finite. It is a contradiction. 

40. If X is countably compact, then every sequence in X has a cluster point. If X is also first countable,
then X is sequentially compact.
Proof. (Jan 20 2018)

If not, for any open set U in X, there exists N ∈ N such that xn 6∈ U for any n ≥ N . Let {Un }∞
n=1 be a
countable open cover of X. Then there exists finite subcover {Uj }m
j=1 . Let Nj be the natural number such
that xn 6∈ Uj for any n ≥ Nj . Let M = max1≤j≤m (Nj ). Then xn 6∈ X for any n ≥ M . It contradict the
fact that {xn }∞
n=1 ⊂ X.

Let {xn }∞
n=1 be a given sequence. Now ,we know that the sequence has a cluster point. Let the cluster
point be x. If X is first countable, there is a neighborhood base at x which can be described as a sequence
of neighborhoods {Un }∞ n=1 of x such that Un+1 ⊂ Un for all n ∈ N. Let’s pick nk such that xnk ∈ Uk . Then,
 15

when V is a given neighborhood of x, there exists N ∈ N such that UN ⊂ V . Then note that xn ∈ V for
any n ≥ N . Thus, {xnk }∞
n=1 converges to x, therefore, X is sequentially compact. 

41. A T1 space X is countably compact iff every infinite subset of X has an accumulation point.
Proof. (Jan 21st 2018)

If there is an infinite subset B ⊂ X which does not have any accumulation point and A ⊂ B is a
countably infinite subset, then for each x ∈ X, there exists a neighborhood Ux ⊂ X of x such that
(Ux \ {x}) ∩ A = ∅. Since acc(A) = ∅, A is closed, so Ac is open. Thus, {Uxo }x∈A ∪ {Ac } is a countable
open cover of X. However, we cannot find any finite subcover. Therefore, X is not countably compact.

Let S =let {xn }∞n=1 ⊂ X be a sequence of distinct points. Since S is an infinite set, it has an accumulation
point x. Let’s pick n1 such that x1 = xn1 . Having chosen {nk }j−1
k=1 , since X is T1 , weT
can find a neighborhood
Uj−1 such that Uj−1 contains x but not xnj−1 . Let’s pick nj such that xnj ∈ j−1 i=1 Ui . Then, for any
neighborhood N of x, there exists M ∈ N such that UM ∩ A ⊂ N ∩ A ∈ N . Note that xnk ∈ UM for
any k > M . Therefore, xnk ∈ N whenever k > M , so {xnk }∞ k=1 is a convergent subsequence. Thus, X is
sequentially compact, so X is countably compact from EX39.


H
42. The set of countable ordinals ( 0.4) with the order topology (Exercise 9) is sequentially compact and
first countable but not compact. (To prove sequential compactness, use Proposition 0.19)
Proof. 

43.
Proof. 

44. If X is countably compact and f : X → Y is continuous, then f(X) is countably compact.

Proof. (Jan 21st 2018)
∞
Let {Un }∞
n=1 be a countable open cover of f (X). Then, since f is continuous, {f
−1
(Un )}n=1 is a countable
−1 m m
open cover of X, so there is a finite subcover {f (Unk )}k=1 . Then {Unk }k=1 is a finite subcover of X. 

45. If X is normal, then X is countably compact iff C(X) = BC(X). (Use Exercise 40 and 44. If hxn i is
a sequence in X with no cluster point, then {xn : n ∈ N} is closed, and Corollary 4.17 applies.)
Proof. (Jan 21st 2018)

Clearly, BC(X) ⊂ C(X). Let f ∈ C(X) be given. If X is countably compact, by Ex 44, f (X) is also
countably compact. Since C is first countable, f (X) is sequentially compact by Ex 40. Since C is a metric
space, f (X) is bounded. Therefore, f ∈ BC(X).

If X is not countably compact, it is not sequentially compact. Thus, there exists a sequence hxn i with
no cluster point. Let A = {xn : n ∈ N}, then it is closed. Define a function f as below.
f (x) = max {k ∈ N : xk = x}
16

Let C ⊂ C be a closed. Let f ∈ C(A), then by Tietze extension theorem, there is F ∈ C(X) such that
F |A = f . 

46. Prove Theorem 4.34

Tietze Extension Theorem, Locally Compact Verision
Suppose that X is an LCH space and K ⊂ X is compact. If f ∈ C(K), there exists F ∈ C(X) such that
F |K = f . Moreover, F may be taken to vanish outside a compact set.
Proof. (Jan 23rd 2018)

Note that there exists a precompact subset V such that K ⊂ V ⊂ V ⊂ X. Further note that, by
Urysohn’s Lemma(LCH version), there exists g ∈ C(X, [0, 1]) ⊂ C(X) such that g = 1 on K and g = 0 on
c
V . Also, since Compact Hausdorff space is normal, by the original version of Tietze Extension Theorem,
there exists FV ∈ C(V ). Now, let’s define a function F as below.
F (x) = FV (x)g(x) If x ∈ V
c
F (x) = 0 If x ∈ V
Then by a part of proof of Proposition 4.32, F ∈ C(X) 

47. Prove Proposition 4.36.

If X, X ∗ , and T are as above, then (X ∗ , T ) is a compact Hausdorff space, and the inclusion map i : X →
X ∗ is an embedding. Moreover, if f ∈ C(X), then f extends continuously to X ∗ iff f = g + c where
g ∈ C0 (X) and c is a constant, in which case the continuous extension is given by f (∞) = c.
Also, show that if X is Hausdorff but not locally compact, Proposition 4.36 remains valid except that X ∗
is not Hausdorff.
Proof. (Jan 29th 2018)

N T is a topology on X ∗
(→) Since ∅ is open in previous topology, ∅ ∈ T . Also, since ∅ = X c is compact in X, X ∗ = X ∪{∞} ∈ T .
Let {Uα }α∈A for some index set A be a family of open sets. Observe the below.
[
Uα ∈ T if ∞ 6∈ Uα ∀α ∈ A (Since the union is open in X)
α∈A

If the union contains ∞,

!c
[ \
Uα = Uαc ⊂ Uβc for some β ∈ A
α∈A α∈A

where Uβ = K c ∪ {∞} and K is compact in X. Thus, Uβc = K which is compact. Since the intersection
above is closed subset of a compact set K, the intersection is compact in X. Therefore,
!
[ [
Uα = Uα \ {∞} ∪ {∞} ∈ T
α∈A α∈A

Let {Uk }nk=1

is a finite family of open sets in X ∗ . If ∞ ∈ Uk for all k ∈ {1, 2, . . . , n}, Ukc is compact, so
n n
!c n n
\ \ [ \
c
∞∈ Uk and Uk = Uk is compact =⇒ Uk ∈ T
k=1 k=1 k=1 k=1
 17

If ∞ 6∈ Ui for some i, then, since Uk \ {∞} is open in X for all k, observe the below.
\n n
\
Uk = (Uk \ {∞}) ∈ T
k=1 k=1
∗
Therefore, T is a topology on X .

N (X ∗ , T ) is compact.
(→) Let O = {Uα }α∈A be an given open cover of X ∗ . Then there exists γ ∈ A such that ∞ ∈ Uγ . Note
that C =let {Uα }α∈A\{γ} is an open cover of Uγc and Uγc is compact in X, so there exists finite subcover
S ⊂ C. Therefore, S ∪ {Uγ } ⊂ O is the finite sub open cover of X ∗ . So X ∗ is compact.

N (X ∗ , T ) is a Hausdorff space.
(→) If x, y ∈ X with x = y, then clearly there exists disjoint open sets Ux (3 x), Uy (3 y) ∈ T since
X is a Hausdorff space. Thus, it suffice to prove that for any x ∈ X, there exists disjoint open sets
Ux (3 x), U∞ (3 ∞) ∈ T . Note that there exists a compact neighborhood K of x since X is LCH. Thus
N =let K c ∪ {∞} , K o ∈ T and K o ∩ N =. Therefore, the space is Hausdorff.

N The inclusion map i : X 7→ X ∗ is an embedding.

(→) Note the collection below is the topology of X.
T1 = {U ∩ i(X) : U ∈ T } = {U ∩ X : U ∈ T }
Therefore, i is embedding.

N If f ∈ C(X), then f extends continuously to X ∗ iff f = g + c where g ∈ C0 (X) and c is a constant,

in which case the continuous extension is given by f (∞) = c
(→)
Let f ∈ C(X). And let F (x) = f (x) if x ∈ X and F (x) = c if x = ∞ where c is a constant. Define
f − c = g. We will prove that g ∈ C0 (X).
Let  > 0 be given.
Observe the below.
{x ∈ X : |g(x)| ≥ } = {x ∈ X : g(x) ∈ B(0; )}
= {x ∈ X : f (x) ∈ B(c; )}
=f −1 (B(c; )c ) = f −1 (B(c; ))c

And note the below.

F −1 (B(c; )) = f −1 (B(0; )) ∪ {∞}
c
Since F is continuous, F −1 (B(c; )) is open, so (f −1 (B(0; ))) is compact. Therefore, {x ∈ X : |g(x)| ≥ }
is compact and g ∈ C0 (X).
(←)
Let f = g + c where g ∈ C0 (X) and c is a given constant as above. And let F be the defined as above.
Now observe the below.
F −1 (V ) = f −1 (V ) ∪ {∞} = {x ∈ X ∗ : g(x) + c ∈ V } ∪ {∞} = g −1 (V − c) ∪ {∞}
for any open set V ∈ C where c ∈ V and V − c = {x − c : x ∈ V }. Since open sets in C is translation
c
invariant, V − c is open. Since g ∈ C0 (X), g −1 (V − c) is open in X, so (g −1 (V − c)) = g −1 ((V − c)c ) is
closed in X. Note that 0 = c − c ∈ V − c, so there exists an open ball B(0; ) ⊂ V − c. Also, note that
B(0; ) 6⊂ (V − c)c . Thus, observe the below.

c
g −1 (V − c) = g −1 ((V − c)c ) ⊂ g −1 (C \ B(0; )) = {x ∈ X : |g(x)| ≥ }
c
Since g ∈ C0 (X), the rightmost set is a compact set, so (g −1 (V − c)) is compact set in X. Therefore,
F (V ) is open in X ∗ .
−1
18

In the case of when c 6∈ V , F −1 (V ) = f −1 (V ) is just open in X, so is open in X ∗ .

N If X is Hausdorff but not locally compact, Proposition 4.36 remains valid except that X ∗ is not
Hausdorff.
(→)
Assume that X ∗ is Hausdorff space. We want to prove that X is locally compact. Let x ∈ X be given.
Since X ∗ is Hausdorff, there exist disjoint open sets U (3 x) and V (3 ∞) in X ∗ . Note that V c is compact
in X and U is closed subset of V c , so is compact in X. Then U is a compact neighborhood of x in X.
Therefore, X is locally compact.


48. Complete the proof of Proposition 4.40.

Proof. 

49. Let X be a compact Hausdorff space and E ⊂ X.

a. If E is open, then E is locally compact in the relative topology.

b. If E is dense in X and locally compact in the relative topology, then E is open.(Use Exercise 13.)

c. E is locally compact in the relative topology iff E is relatively open in E.

Proof. (Jan 24th 2018)

(a) Let x ∈ E be given. Then there is open neighborhood N ⊂ E of x. Note that N is compact since
X is compact. Let {Uα }α∈A ⊂ P(E) be an open cover of N where for each α ∈ A, Uα is relatively open
with respect to E, so there is set Oα ⊂ X such that it is open relative to X and Oα ∩ E = Uα . Since N is
compact relative to X, there exists finite subcover {Ok }nk=1 of N . Then {Uk }nk=1 is also the finite subcover
of N . Therefore, N is compact relative to the space E. Thus, E is locally compact.

(b) Let x ∈ E be given. Then there exists a neighborhood K of x which is compact in the relative
topology. Observe that compactness is preserved by following argument.

Let {Uα }α∈A be open(In X) cover of K. Then {Uα ∩ E}α∈A is the open(relative to E) cove of K. Since
K is compact relative to E, there exists a finite open subcover {Uk ∩ E}nk=1 . Thus, {Uk }nk=1 is a finite
open(In X) subcover of K and K is compact in X.

So K is closed since X is Hausdorff. Since K is a neighborhood of x, x ∈ K o . Then there exists H ⊂ X

which is open in X such that K o = H ∩ E. Then, by Exercise 13, observe the below.
K = Ko = H ∩ E = H
Note that x ∈ H ⊂ K ⊂ E. Therefore, for any x ∈ E, there exists an open neighborhood H of x such
that H ⊂ E, so E is an open set.

(c) If E is locally compact in the relative topology, since E is dense in E, by (b), E is relatively open in
E.
If E is relatively open in E, since E is compact Hausdorff space, by (a), E is locally compact in the
relative topology. 
 19

50. Let U be an open subset of a compact Hausdorff space X and U ∗ its one-point compactification (See
Exercise 49a). If φ : X → U ∗ is defined by φ(x) = x if x ∈ U and φ(x) = ∞ if x ∈ U c , then φ is
continuous.
Proof. (Jan 25th 2018)

Let V ⊂ U ∗ be given open set.

If ∞ 6∈ V , φ−1 (V ) = V . Note that V is open relative to open set U , so V is open in X.
If ∞ ∈ V , then φ−1 (V ) = W ∪ U c and U ∩ W c is compact relative to U where W = V \ {∞}. Since U is
Hausdorff space, W is open relative to the open set U , so open in X. Let K = U \ W and note that it is
compact relative to U . Also, note that U c is compact in X. Thus, for any x ∈ K, there exist two disjoint
open sets Qx (3 x) and Vx (⊃ U c ). Note that {U ∩ Qx }x∈K is an open cover of K relative to U , so there
exists finite subcover {U ∩ Qxi }ni=1 such that
[n \n
K ⊂ (U ∩ Qxi ) =let G Uc ⊂ Vxi =let N
i=1 i=1
c
Note that N is open and N ∩ G = ∅, so N ⊂ U ∪ W . Thus, observe the below.
φ−1 (V ) = W ∪ U c = W ∪ U c ∪ N = W ∪ N
Therefore, φ−1 (V ) is open, so φ is continuous. 

51.
Proof. 

52.
Proof. 

53.
Proof. 

54.
Proof. 

55.
Proof. 

56.
Proof. 

57.
Proof. 
20

58.
Proof. 

59.
Proof. 

60.
Proof. 

61.
Proof. 

62.
Proof. 

63.
Proof. 

64.
Proof. 

65.
Proof. 

66.
Proof. 

67. Prove Theorem 4.52

Let X be a noncompact LCH space. If A is a closed subalgebra of C0 (X, R)(= C0 (X) ∩ C(X, R)) that
separates points, then either A = C0 (X, R) or A = {f ∈ C0 (X, R) : f (x0 ) = 0} for some x0 ∈ X.
(If there exists x0 ∈ X such that f (x0 ) = 0 for all f ∈ A, let Y be the one-point compactification of
X \ {x0 }; otherwise let Y be the one-point compactification of X. Apply Proposition 4.36 and the Stone-
Weierstrass theorem to Y.)
 21

Proof. (Feb 4th 2018)

Firstly, suppose there exists x0 ∈ X such that f (x0 ) = 0 for all f ∈ A. Note that X \ {x0 } is LCH
since for any a ∈ X \ {x0 }, there exists a compact neighborhood a ∈ K ⊂ X and any compact set in X is
compact relative to X \ {x0 }.
And let Y = X \ {x0 } ∪ {x0 = ∞} be the one point compactification of X \ {x0 }. Now, let f ∈ C0 (X, R),
then f ∈ C0 (X \ {x0 }, R), thus, by Proposition 4.36, we can extend f continuously so that f ∈ C(Y, R).
Therefore, A is a closed subalgebra of C(Y, R). Then, by the Stone Weierstrass theorem, A = {f ∈
C(Y, R) : f (x0 ) = 0} = {f ∈ C0 (Y, R) : f (x0 ) = 0} = {f ∈ C0 (X, R) : f (x0 ) = 0} since Y is compact.

Otherwise, let Y = X ∪ {∞} be one-point compactification of X and let f (∞) = 0 for all f ∈ A. Since
f ∈ C0 (X, R), by Proposition 4.36, we can continuously extend f ∈ C0 (X, R) to f ∈ C(Y, R). By the
Stone-Weierstrass theorem, since A is also closed subalgebra of C(Y, R), A = {f ∈ C(Y, R) : f (∞) = 0} =
C0 (X, R).

Therefore, A = C0 (X, R) or A = {f ∈ C0 (X, R) : f (x0 ) = 0} for some x0 ∈ X. 

68. Let X and Y be compact Hausdorff spaces. The algebra generated by functions of the form f (x, y) =
g(x)h(y), where g ∈ C(X) and h ∈ C(Y ), is dense in C(X × Y ).

Proof. (Feb 24th 2018)

Let A be the algebra given by the problem. Note that C(X) and C(Y ) separate points and let
(x, y), (v, w) ∈ X × Y with (x, y) 6= (v, w). Then x 6= v or y 6= w.

If x 6= v, by Urysohn’s lemma, ∃g ∈ C(X) such that g(x) 6= g(v). If we set h ≡ c ∈ C(Y ) for some
nonzero constant c ∈ C, f (x, y) = cg(x) 6= cg(v) = f (v, w).

Similarly, if y 6= w, by Urysohn’s lemma ∃h ∈ C(Y ) such that h(y) 6= h(w). Then for some nonzero
constant function c ∈ C(X), f (x, y) = ch(y) 6= ch(w) = f (v, w).

If x 6= v and y 6= w, we can set any but only one of g ∈ C(X) and h ∈ C(Y ) be a nonzero constant and
do similar argument as above.

The arguments above implies that A ⊂ C(X × Y ) separates points.

Lastly, note that A contains a nonzero constant function since each C(X) and C(Y ) contain nonzero
constant functions.

Therefore, by the Stone-Weierstrass Theorem, A should be dense in C(X × Y ). 

69. Let A be a nonempty set, and let X = [0, 1]A . The algebra generated by the coordinate maps πα : X →
[0, 1](α ∈ A) and the constant function 1 is dense in C(X).

Proof. (Feb 3rd 2018)

N The addition is continuous.

Let  > 0 be given. And let δ = 2 . Then, for any (x, y) ∈ X × X , observe the below.
22


max(kx − vk, ky − wk) = k(x − v, y − w)k = k(x, y) − (v, w)k < δ =
2
 
=⇒ k(x + y) − (v + w)k = k(x − v) + (y − w)k ≤ kx − vk + ky − wk < + = 
2 2
Therefore, the addition operation is continuous.

N The Scalar multiplication is continuous.

 
Let  > 0 be given. Then, for any (α, x) ∈ K × X , let δ = min( 3kxk , 3|α| , 1, 3 ) and observe the below.

  
max(|α − β|, kx − yk) = k(α − β, x − y)k = k(α, x) − (β, y)k < δ = min( , , 1, )
3kxk 3|α| 3
=⇒ kαx − βyk ≤ kαx − βxk + kβx − βyk = |α − β|kxk + |β|kx − yk
 
≤ |α − β|kxk + |α|kx − yk + |α − β|kx − yk < + + kx − yk < 
3 3

N The norm operation is continuous.

Let  > 0 and x ∈ X be given. And let δ =  and, for any y ∈ X , by triangular inequality, note the
below.

kxk ≤ kx − yk + kyk =⇒ kxk − kyk ≤ kx − yk

kyk ≤ kx − yk + kxk =⇒ kyk − kxk ≤ kx − yk
Thus, |kxk − kyk| ≤ kx − yk.

Then observe the below.

kx − yk < δ =  =⇒ |kxk − kyk| ≤ kx − yk < 

Therefore, the mapping k · k : X → K is continuous.


70. Let X be a compact Hausdorff space. An ideal in C(X, R) is a subalgebra J of C(X, R) such that if
f ∈ J and g ∈ C(X, R) then f g ∈ J .

a. If J is an ideal in C(X, R), let h(J ) = {x ∈ X : f (x) = 0 for all f ∈ J }. Then h(J ) is a closed
subset of X, called hull of J .

b. If E ⊂ X, let k(E) = {f ∈ C(X, R) : f (x) = 0 for all x ∈ E}. Then k(E) is a closed ideal in
C(X, R), called the kernal of E.

c. If E ⊂ X, then h(k(E)) = E.

d. If J is an ideal in C(X, R), then k(h(J )) = J . (Hint : k(h(J )) may be identified with a subalgebra
of C0 (U, R) where U = X \ h(J ).)

e. The closed subsets of X are in one-to-one correspondence with the closed ideals of C(X, R).
 23

Proof. (Feb 25th 2018)

a.

Note the below

\
h(J ) = f −1 ({0})
f ∈J

Further note that f −1 ({0}) is closed in X for all f ∈ J since f is continuous and {0} is closed in R.
Since any intersection of closed sets is closed, h(J ) is closed in X.

b.

Let f, g ∈ k(E) and α, β ∈ R be given. First note that 0 ∈ k(E), and secondly observer the below.

αf (x) + βg(x) = α0 + β0 = 0 ∀x ∈ E
Thus, k(E) is a vector subspace of C(X, R).

Nextly, let f ∈ k(E) and g ∈ C(X, R) be given. Then observe the below.
f g(x) = f (x)g(x) = 0g(x) = 0 ∀x ∈ E
Thus, f g ∈ k(E), so k(E) is an ideal in C(X, R).

Lastly, suppose that f ∈ acc(k(E)), then there is a sequence {fn }∞

n=1 ⊂ k(E) such that

lim fn = f
n→∞

Observe that
f (x) = lim fn (x) = lim 0 = 0 ∀x ∈ E
n→∞ n→∞

Thus, f ∈ k(E) so k(E) is closed.

c.

Note the following representation.

h(k(E)) = {x ∈ X : f (x) = 0 for all f ∈ k(E)}
From the above, E ⊂ h(k(E)) and from a. and b, h(k(E)) is closed so E ⊂ h(k(E)).

Conversely, if x ∈ h(k(E)), f (x) = 0 for all f ∈ k(E).

d. 

71.
Proof. 
24

72.
Proof. 

73.
Proof. 

74.
Proof. 

75.
Proof. 

76.
Proof. 

77.
Proof.