Chapter 5

Countability and Separation Axioms

5.1 First and Second Countable Topological Spaces

Definition 5.1.1. A topological space (X, J ) is said to have a countable local

basis (or countable basis) at a point x ∈ X if there exists a countable collection say
Bx of open sets containing x such that for each open set U containing x there exists
V ∈ Bx with V ⊆ U .
Definition 5.1.2. A topological space (X, J ) is said to be first countable or said
to satisfy the first countability axiom if for each x ∈ X there exists a countable local
base at x.
Examples 5.1.3. (i) Let (X, d) be a metric space then for each x ∈ X,
Bx = {B(x, n1 ) : n ∈ N} is a countable local basis at x. Hence (X, Jd ) is a first
countable space. So, we say that every metric space (X, d) is a first countable space.
(ii) Let X = N and J = {φ, X, {1}, {1, 2}, . . . , {1, 2, . . . , n}, . . . , } then obviously
(X, J ) is a first countable topological space.

Note that this is not an interesting example of a first countable topological

space. Once the topology J is a countable collection then (X, J ) is a first countable
space.
Example 5.1.4. Let X = R and Jl be the lower limit topology on R generated by
{[a, b) : a, b ∈ R, a < b}. For each x ∈ X, Bx = {[x, x + n1 ) : n ∈ N} is a countable

111
local base at x. Hence (R, Jl ) = Rl is a first countable topological space. Now let us
see a stronger version of first countable topological space.
Definition 5.1.5. If a topological space (X, J ) has a countable basis B then we
say that (X, J ) is a second countable topological space or it satisfies the second
countability axiom.
Exercise 5.1.6. Though it is trivial from the definition, prove that every second
countable topological space (X, J ) is a first countable topological space. 2

What about the converse?

Let X be any uncountable set and JD be the discrete topology on X. Then

(X, JD ) is first countable, but it is not second countable. In fact, for each x ∈ X,
Bx = {{x}} is a countable local base at x. Take any open set U containing x then
there exists V = {x} ∈ Bx such that x ∈ V ⊆ U . Hence Bx is a local base at x.

How to prove that (X, J ) is not a second countable topological space ? Well
we use the method of proof by contradiction.

Suppose there exists a countable basis say B = {B1 , B2 , . . . , } for (X, J ). Let
us assume that each Bk 6= φ. For each k ∈ N, let xk ∈ Bk . Since X is an uncountable
set we can select an x ∈ X such that x 6= xk for all k ∈ N. Now {x}, the singleton set
containing x, is an open set and B is a basis for (X, J ) implies there exists k ∈ N
such that x ∈ Bk ⊆ {x} this implies Bk = {x}. But xk ∈ Bk implies x = xk , a
contradiction to our assumption that x ∈ X such that x 6= xk for all k ∈ N. Hence if
X is an uncountable set then the discrete topological space (X, JD ) is first countable
but not second countable.

Also we have seen that the lower limit topological space Rl is first countable.
Now let us prove that Rl = (R, Jl ) is not a second countable topological space. That

112
is we will have to prove that if B is a basis for (R, Jl ) then B is not a countable
collection. So, fix a basis say B for (R, Jl ). For each x ∈ R, [x, x + 1) ∈ Jl . Hence
B is a basis for (R, Jl ) implies there exists Bx ∈ B such that x ∈ Bx ⊆ [x, x + 1).

x y x +1 y +1

Figure 5.1

For x, y ∈ R, x 6= y we have [x, x + 1) 6= [y, y + 1). Also Bx ⊆ [x, x + 1) implies

inf Bx ≥ inf[x, x + 1) = x. Also x ∈ Bx implies x ≥ inf Bx . Hence x = inf Bx . Now
define f : R → B as f (x) = Bx . Then x 6= y implies Bx 6= By . (Bx = By implies
inf Bx = inf By ) That is f (x) 6= f (y). Hence f is an one-one function. This implies
that f : R → f (R) ⊆ B is a bijective function. Therefore f (R) is an uncountable
set and hence B is an uncountable set. We have proved that if B is a basis for
(R, Jl ) then B is an uncountable set. Hence (R, Jl ) cannot have a countable basis
and therefore (R, Jl ) is not a second countable topological space.

It is a simple exercise to check Q = R in (R, Jl ). That is Q is a countable

dense subset of R with respect to (R, Jl ). Such a topological space is known as a
separable topological space.
Definition 5.1.7. A topological space (X, J ) is said to be a separable topological
space if there exists a countable subset say A of X such that A = X.
Definition 5.1.8. A topological space (X, J ) is said to be a Lindelöf space if for any
collection A of open sets such that X = ∪ A, there exists a countable subcollection
A∈A

113
say B ⊆ A such that X = ∪ B. That is, a topological space (X, J ) is said to be a
B∈B
Lindelöf space if and only if every open cover of X has a countable subcover for X.

By definition every compact topological space (X, J ) is a Lindelöf space. But

the converse need not be true. It is easy to prove that R (R with usual topology) is
a Lindelöf space. But R is not compact space.

Now let us prove that every second countable topological space is a Lindelöf
space.

Theorem 5.1.9. If (X, J ) is a second countable topological space then (X, J ) is a

Lindelöf space.

Proof. Let B = {B1 , B2 , B3 , . . .} be a countable basis for (X, J ) and A be an open

cover for X. Let us assume that, X 6= φ, A 6= φ for each A ∈ A and B 6= φ, for each
B ∈ B. Fix A ∈ A and x ∈ A. Now x ∈ A, A is an open set implies there exists
B ∈ B such that
x ∈ B ⊆ A. (5.1)

For each n ∈ N, let Fn = {A ∈ A : Bn ⊆ A}. Here it is possible that Fn = φ, for

some n ∈ N. At the same time note that, from Eq. (5.1), {n ∈ N: Fn 6= φ} is a
nonempty set. Let {n ∈ N: Fn 6= φ} = {n1 , n2 , . . . , nk . . . , } (it may be a finite set)
and for each such k take Ank ∈ Fnk . This will give us Bnk ⊆ Ank ∈ A.
∞
Let us prove that ∪ Ank = X. So, let x ∈ X. Now A is an open cover for X
k=1
implies x ∈ A for some A. Now x ∈ A, B is a basis for (X, J ) implies there exists
k ∈ N such that x ∈ Bnk ⊆ A. This implies that A ∈ Fnk . Also Ank ∈ Fnk . Hence by
our definition of Fnk , Bnk ⊆ Ank . Hence x ∈ X implies x ∈ Ank , for some k ∈ N.

114
 ∞
This implies that X ⊆ ∪ Ank . That is {Ank }∞
k=1 is a countable subcover for
k=1
A. Therefore every open cover A of X has a countable subcover. Hence (X, J ) is a
Lindelöf space. 

∞
Note. Recall that, for 1 ≤ p < ∞, lp = {x = (xn )∞ |xn |p < ∞} is a second
P
n=1 :
n=1
countable metric space, where for x = (xn ) ∈ lp , y = (yn ) ∈ lp ,

∞
X 1
p
dp ((xn ), (yn )) = dp (x, y) = |xn − yn |p .
n=1

Also note that Rn = {x = (x1 , x2 , . . . , xn ) : xi ∈ R, i = 1, 2, . . . , n} is a second

countable metric space with respect to any of the metric given by dp (x, y) =
 ∞ 1
p
|xn − yn |p for x = (x1 , x2 , . . . , xn ) ∈ Rn , y = (y1 , y2 , . . . , yn ) ∈ Rn , 1 ≤ p < ∞
P
n=1
or d∞ (x, y) = max{|xk − yk | : k = 1, 2, . . . , n}. So, all the above mentioned metric
spaces are all Lindelöf spaces. But none of these metric spaces is a compact space. ?
Now let us prove that a second countable topological space is a separable space.
Theorem 5.1.10. Every second countable topological space (X, J ) is a separable
space.

Proof. Given that (X, J ) is a second countable topological space. Hence there
exists a countable basis say B = {B1 , B2 , . . .} for (X, J ). When we write
B = {B1 , B2 , . . . , }, it does not mean that B is a countably infinite set. It
means that either for some n ∈ N, B = {B1 , B2 , . . . , Bn } or B = φ or B is a
countably infinite set. If X 6= φ then B 6= φ. If for some k ∈ N, Bk = φ, then
B 0 = {B1 , B2 , . . . , Bk−1 , Bk+1 , . . . , } is also a basis for (X, J ).

So, let us assume that each Bn 6= φ for all n. Since Bn 6= φ, for each n ∈ N, let
∞
xn ∈ Bn (note that by axiom of choice there exists a function f : N → ∪ Bn such
n=1

115
that xn = f (n) ∈ Bn ) and A = {x1 , x2 , x3 . . . , }. Here also it is quite possible that
A is a finite set. Now let us prove that A = X. So, take an x ∈ X and an open set
U containing x. Now B is a basis for (X, J ), U is an open set containing x implies
there exists Bn ∈ B such that x ∈ Bn and Bn ⊆ U . Also xn ∈ Bn . Hence xn ∈ U ∩ A.
This gives that U ∩ A 6= φ. That is we have proved that U ∩ A 6= φ for each open set
U containing x. Hence x ∈ A. That is x ∈ X and hence x ∈ A and hence A = X.
Therefore (X, J ) has a countable dense subset and therefore (X, J ) is a separable
space. 

Now let us prove that subspace of a separable metric space is separable.

Theorem 5.1.11. Let (X, d) be a separable metric space and Y be a subspace of X
(that is Y ⊆ X, and for x, y ∈ Y , dY (x, y) = d(x, y)). Then (Y, dY ) is a separable
space.

Proof. (X, d) is a separable metric space implies there exists a countable subset
say A = {x1 , x2 , x3 . . . , } of X such that A = X (here A denotes the closure of A
with respect to (X, d)). We will have to find a countable subset say B of Y such
that B Y = Y (here B Y = B ∩ Y , the closure of B with respect to the subspace
(Y, dY )). For n ∈ N, let An,k = B(xn , k1 ) ∩ Y . Here we do not know whether An,k = φ
or An,k 6= φ. If An,k 6= φ (n, k ∈ N) let an,k ∈ An,k be a fixed element. Then
B = {an,k : an,k ∈ An,k whenever An,k 6= φ} is a countable subset of Y . Now let us
prove that B Y = B ∩ Y = Y. So let x ∈ Y and U be an open set in X containing
x. Hence there exists k ∈ N such that B(x, k1 ) ⊆ U. Again x ∈ A = X implies
1 1
B(x, 2k ) ∩ A 6= φ. Then there exists xn ∈ A such that xn ∈ B(x, 2k ). Therefore
1
x ∈ B(xn , 2k ) ∩ Y = An,2k . Hence An,2k 6= φ. Now An,2k 6= φ implies an,2k ∈ B.
1 1 1
Further an,2k ∈ B(xn , 2k ). Now d(x, an,2k ) ≤ d(x, xn ) + d(xn , an,2k ) < 2k
+ 2k
= k1 .

116
Hence an,2k ∈ B(x, k1 ) ∩ B ⊆ U ∩ B. That is U ∩ B 6= φ for each open set U containing
x. This implies x ∈ B. Also x ∈ Y . Therefore x ∈ B ∩ Y = B Y . This implies
Y ⊆ B Y ⊆ Y and hence B Y = Y . That is B is a countable dense subset of Y. This
proves that the subspace (Y, dY ) is a separable metric space. 

Subspace of a separable topological space need not be separable.

We give an example to show that subspace of a separable topological space

need not be separable.

Let X = {(x, y) : x ∈ R, y ≥ 0}. Basic open sets are of the type:

(i) for (x, y) ∈ R2 , x ∈ R, y > 0 basic open sets containing (x, y) are of the form
B((x, y), r), 0 < r < y, and (ii) for (x, 0) ∈ R2 , x ∈ R, basic open sets are of the form
(B(x, 0), r) ∩ X)K{(y, 0) : 0 < |y − x| < r}, r > 0. Here B((x, y), r) = {(a, b) ∈ R2
p
: d((x, y), (a, b)) = (x − a)2 + (y − b)2 < r}, the open ball centered at (x, y) and
radius r with respect to the Euclidean metric d on R2 .

It is easy to see that the above collection of sets will form a basis for a topology
on X.

Let J be the topology on X induced by the collection of basic open sets

described as above and if A = {(x, y) : x, y ∈ Q, y ≥ 0} then A is a countable
subcollection of X such that A = X. That is A is a countable dense subset of X, and
hence (X, J ) is a separable topological space.

Now for Y = {(x, 0) : x ∈ R} (an uncountable set), JY = P(Y ). That is the

subspace (Y, JY ) of (X, J ) is the discrete topological space. That is every subset U
of Y is both open and closed in (Y, JY ). Therefore if B is a countable subset of Y
then B = B 6= Y here (B = B JY , the is closure of B in (Y, JY )). This proves that
(Y, JY ) is not a separable subspace, though (X, J ) is a separable topological space.

117
 Note that we have already proved that every second countable topological
space is separable.
Exercise 5.1.12. Prove that every separable metric space is second countable. 2
Exercise 5.1.13. Prove that Rn = {(x1 , x2 , . . . , xn ) : xi ∈ R, i = 1, 2, . . . , n} is
 n  p1
p
P
a separable metric space for 1 ≤ p < ∞, dp (x, y) = |xi − yi | ,
i=1 
x = (x1 , x2 , . . . , xn ) ∈ Rn , y = (y1 , y2 , . . . , yn ) ∈ Rn and d∞ (x, y) = max |xi − yi | :
2

1≤i≤n .

It is easy to prove that if J2 is the topology induced by d2 then Jp = J2 ,

∀ p ≥ 1. That is all these metrics dp , 1 ≤ p ≤ ∞ will induce the same topology on
Rn . So if we want to prove that (Rn , dp ) is a separable metric space, it is enough to
prove that (Rn , J1 ) (or say (Rn , J2 )) is separable.
∞
|xn |p < ∞}. If we define,
P
For 1 ≤ p < ∞, let lp = {x = (xn ) : xn ∈ R for all n and
n=1
∞  p1
p
P
dp (x, y) = |xn − yn | , then dp is a metric on lp . Now let us see how to prove
n=1
that (lp , dp ) is a separable metric space.
Step 1: For each n ∈ N, let An = {(r1 , r2 , . . . , rn , . . . , 0, 0, . . .) : ri ∈ Q,
i = 1, 2, . . . , n}.

If we define f : Q × Q × · · · × Q (n times) = Qn → An as f (r1 , r2 , . . . , rn ) =

(r1 , r2 , . . . , rn , . . . , 0, 0, . . .). Then f is a bijective function. Now Q is a countable set
implies Q × Q × · · · × Q (finite product of countable sets is countable) is countable.
Hence there is a bijection between Qn and An implies An is a countable set. Now
∞
each An is a countable set implies ∪ An is also countable.
n=1
∞ ∞
We leave it as an exercise to prove that ∪ An = lp . That is ∪ An is a
n=1 n=1
countable dense subset of lp . Hence (lp , dp ) is a separable metric space. These spaces

118
are important examples of Banach spaces. If l∞ = {x = (xn ) : (xn ) is a bounded
real sequence } and d∞ (x, y) = sup{|xn − yn | : n ≥ 1}, then (l∞ , d∞ ) is also a metric
space. Let X = {x = (xn ) : xn = 0 or xn = 1}. For x, y ∈ X, x 6= y, d(x, y) = 1.
Hence (X, d∞ ) (that is d∞ is restricted to the subspace X of l∞ ) is a metric space.
Now the topology J on X induced by the metric d∞ is the discrete topology on X. In
this topological space (X, J ) every subset A of X is both open and closed. Therefore,
if A is a countable subset of X, then A = A 6= X, (note that X is an uncountable
subset of X). Now the subspace X of l∞ is not a separable space implies l∞ is not a
separable space.

5.2 Properties of First Countable Topological Spaces

Theorem 5.2.1. If (X, J ) is a first countable topological space then for each x ∈ X
there exists a countable local base say {Vn (x)}∞
n=1 such that Vn+1 (x) ⊆ Vn (x).

Proof. Fix x ∈ X. Now (X, J ) is a first countable topological space implies there
exists a countable local base say {Un }∞
n=1 at x. Let Vn (x) = U1 ∩ U2 ∩ · · · ∩ Un then

{Vn (x)}∞
n=1 is a collection of open sets such that Vn+1 (x) ⊆ Vn (x) for all n ∈ N. So,

it is enough to prove that {Vn (x)}∞

n=1 is a local base at x. So start with an open set

V containing x. Now {Un }∞

n=1 is a local base at x and V is an open set containing x

implies there exists n0 ∈ N such that Un0 ⊆ V . By the definition of Vn (x)0 s we have
Vn0 (x) ⊆ Un0 . Hence we have the following: for each open set V containing x there
exists n0 ∈ N such that Vn0 (x) ⊆ V . This implies that {Vn (x)} is a local base at x
satisfying Vn+1 (x) ⊆ Vn (x) for all n ∈ N. 

Let us use the above characterization of a first countable base to show that,
in some sense, first countable topological spaces behave like metric spaces.

119
Theorem 5.2.2. Let (X, J ) be a first countable topological space and A be a
nonempty subset of X. Then for each x ∈ X, x ∈ A if and only if there exists a
sequence {xn }∞
n=1 in A such that xn → x as n → ∞.

Proof. First let us assume that x ∈ A. Now (X, J ) is a first countable topological
space implies there exists a countable local base say B = {Vn }∞
n=1 such that Vn+1 ⊆ Vn ,

for all n ∈ N. Hence x ∈ A implies A ∩ Vn 6= φ, for each n ∈ N. Let xn ∈ A ∩ Vn .

Claim: xn → x as n → ∞.
So start with an open set U containing x (enough to start with Vn ) then there exists
n0 ∈ N such that x ∈ Vn0 ⊆ U . Hence xn ∈ Vn ⊆ Vn0 ⊆ U for all n ≥ n0 . That is
xn ∈ U for all n ≥ n0 . This means xn → x as n → ∞.

Conversely, suppose there exists a sequence {xn }∞

n=1 in A such that xn → x.

Then for each open set U containing x there exists a positive integer n0 such that
xn ∈ U for all n ≥ n0 . In particular xn0 ∈ U ∩ A. Hence for each open set U
containing x, U ∩ A 6= φ and this implies x ∈ A. 

Theorem 5.2.3. Let X and Y be topological spaces and further suppose X is a first
countable topological space. Then a function f : X → Y is continuous at a point
x ∈ X if and only if for every sequence {xn }∞
n=1 in X, xn → x as n → ∞, then the

sequence {f (xn )}∞

n=1 converges to f (x) in Y.

Proof. Assume that f : X → Y is continuous at a point x ∈ X. Also assume that

{xn }∞
n=1 is a sequence in X such that xn → x as n → ∞.

To prove: f (xn ) → f (x) in Y .

So start with an open set V in Y containing f (x). Since f is continuous at x,
U = f −1 (V ) is an open set in X. Now f (x) ∈ V implies x ∈ f −1 (V ) = U . That is

120
U is an open set containing x. Hence xn → x implies there exists n0 ∈ N such that
xn ∈ U for all n ≥ n0 . This implies f (xn ) ∈ V for all n ≥ n0 . That is, whenever
xn → x as n → ∞ then f (xn ) → f (x) as n → ∞.

Conversely, suppose that {xn } is a sequence in X, xn → x as n → ∞ implies

f (xn ) → f (x). Now we will have to prove that f is continuous at x. It is to be
noted that to prove this converse part we will make use of the fact that X is a first
countable space. Now X is a first countable space implies there exists a local base
{Vn }∞
n=1 at x such that Vn+1 ⊆ Vn for all n ∈ N. We will use the method of proof

by contradiction. If f is not continuous at x then there should exist an open set W

containing f (x) such that f (U ) * W for any open set U containing x. In particular
for such an open set W , f (Vn ) * W for all n = 1, 2, 3, . . .. Hence there exists xn ∈ Vn
such that f (xn ) ∈
/ W.
Claim: xn → x as → ∞. So start with an open set V in X containing x.

Now {Vn }∞
n=1 is a local base at x implies there exists n0 ∈ N such that Vn0 ⊆ V .

Hence xn ∈ Vn ⊆ Vn0 ⊆ V for all n ≥ n0 . That is for each open set V containing x
there exists n0 ∈ N such that xn ∈ V for all n ≥ n0 . Hence xn → x as n → ∞. But
this sequence {xn } in X is such that f (xn ) ∈
/ W , where W is an open set containing
f (x). So we have arrived at a contradiction to our assumption namely xn ∈ X,
xn → x ∈ X implies f (xn ) → f (x). We arrived at this contradiction by assuming f
is not continuous at x. Therefore our assumption is wrong and hence f is continuous
at x. 

Example 5.2.4. Let Jc = {A ⊆ R : Ac is countable or Ac = R}, the co-countable

topology on R, and X = (R, Jc ), Y = (R, Js ), where Js is the standard topology
on R. Define f : X → Y as f (x) = x for all x ∈ X. Suppose {xn } is a sequence

121
in X such that {xn } converges to x ∈ X = R. Then it is easy to prove that there
exists n0 ∈ N such that xn = x for all n ≥ n0 . (If this statement is not true then
there exists a subsequence {xnk }∞ ∞
k=1 of {xn }n=1 such that xnk 6= x for all k ∈ N. Then

U = RK{xnk : k ∈ N} is an open set in X containing x. Hence {xn } converges to x

in X implies there exists n0 ∈ N such that xn ∈ U for all n ≥ n0 . In particular for
k ≥ n0 , nk ≥ k ≥ n0 and this implies xnk ∈ U .) So we have the following: xn → x in
X implies f (xn ) → f (x) in Y. But the given function f : X → Y is not a continuous
function (note: f −1 (0, 1) = (0, 1) is not an open set in (R, Jc )). This example does
not give any contradiction to theorem 5.2.3. From this example we conclude that
X = (R, Jc ) is not a first countable topological space.

5.3 Regular and Normal Topological Spaces

Definition 5.3.1. A topological space (X, J ) is called a T1 space if for each x ∈ X,

the singleton set {x} is a closed set in (X, J ).
Definition 5.3.2. A T1 -topological space (X, J ) is called a regular space if for
each x ∈ X and for each closed subset A of X with x ∈
/ A, there exist open sets U ,
V in X satisfying the following:
(i) x ∈ U, A ⊆ V , (ii) U ∩ V = φ.
Result 5.3.3. Every regular topological space (X, J ) is a Hausdorff space.

Proof. Let x, y ∈ X, x 6= y. By definition every regular space is a T1 -space. Hence

{y} is a closed set. Also x 6= y implies x ∈
/ A = {y}. Now {y} is a closed set which
does not contain x. Since (X, J ) is a regular space, there exist open sets U, V in X
satisfying the following:
(i) x ∈ U, A = {y} ⊆ V ,

122
(ii) U ∩ V = φ that is U, V are open sets in X such that x ∈ U, y ∈ V and U ∩ V = φ.
Hence (X, J ) is a Hausdorff topological space. 

Exercise 5.3.4. Prove that every Hausdorff space is a T1 -space. 2

Example 5.3.5. Let X be an infinite set and Jf be the cofinite topology on X.
Then (X, Jf ) is a T1 -space, but (X, Jf ) is not a Hausdorff space. For each x ∈ X,
U = XK{x} is an open set. Hence U c = XKU = {x} is a closed set in X. That is
for each x ∈ X, the singleton set {x} is a closed set. Therefore (X, J ) is a T1 -
space. Take any x, y ∈ X, x 6= y. Suppose there exist open sets U , V in X such that
x ∈ U, y ∈ V and U ∩ V = φ. Now U, V are nonempty open subsets of the cofinite
topological space (X, Jf ) implies U c , V c are finite sets. Hence X = φc = (U ∩ V )c =
U c ∪ V c is a finite set. Therefore there cannot exist any open sets U, V in (X, Jf )
satisfying x ∈ U, y ∈ V and U ∩ V = φ. This means (X, Jf ) is not a Hausdorff space.

Now let us give an example of a topological space which is Hausdorff but not
regular. Take X = R and BK = {(a, b), (a, b)KK : a, b ∈ R, a < b}, where
K = {1, 21 , 31 , . . .}. Now it is easy to prove that (left as an exercise) BK is a basis for
a topology on R. Let JK be the topology on R generated by BK . If J is the usual
topology on R then we know that J is generated by B = {(a, b) : a, b ∈ R, a < b}.
Since we have B ⊆ BK and this implies that J = JB ⊆ JBK = JK .

From this, it is clear that (R, JK ) is a Hausdorff space. For x, y ∈ R, x 6= y,

(R, J ) is a Hausdorff space implies there exist open sets U and V in (R, J ) such
that x ∈ U, y ∈ V and U ∩ V = φ. But J ⊆ JK . Hence U, V ∈ JK are such that
x ∈ U, y ∈ V and U ∩ V = φ and this shows that (X, JK ) is a Hausdorff topological
space.

123
 Is K = {1, 21 , 13 , . . .} a closed set? Here K is a subset of R and J , JK are two
different topologies on R, 0 ∈ K and 0 ∈
/ K with respect to (R, J ). Hence K is not
∞
a closed set in (R, J ). But RKK = ∪ An , where An = (−n, n)KK for each n ∈ N.
n=1

Each An is an open set in (R, JK ) implies RKK is an open set in (R, JK ). This
implies K is a closed set in (R, JK ). Also 0 ∈
/ K. What are the open sets containing
1
K? If V is an open set containing K, then for each n ∈ N, n
∈ V, there exists a basic
1
open set say (an , bn ) such that n
∈ (an , bn ) ⊆ V ( n1 ∈
/ (an , bn )KK) and 0 < an < bn
∞
implies K ⊆ ∪ (ak , bk ) ⊆ V.
k=1

Suppose U, V are open sets such that 0 ∈ U and K ⊆ V . Since 0 ∈ U, there

exists a basic open set B such that 0 ∈ B ⊆ U. If B is of the form (a, b) then
(a, b) ∩ K 6= φ. So U ∩ V 6= φ. If B is of the form (a, b)KK, choose n0 ∈ N such that
1 1 1
n0
< b. Since n0
∈ V, there exists an open interval (c, d) such that n0
∈ (c, d) ⊆ V.
1
Now since (a, b) ∩ (c, d) is not empty (it contains n0
), it is an interval and hence
uncountable. As K is countable, ((a, b) ∩ (c, d))KK 6= φ, i.e, ((a, b)KK) ∩ (c, d) 6= φ.
Therefore U ∩ V 6= φ.

So we have proved that there cannot exist open sets U, V in (R, JK ) with
0 ∈ U , K ⊆ V and U ∩ V = φ. This shows that (R, JK ) is not a regular space.
Definition 5.3.6. A topological space (X, J ) is said to be a normal space if and
only if it satisfies:
(i) (X, J ) is a T1 -space,
(ii) A, B closed sets in X, A ∩ B = φ implies there exist open sets U, V in X such
that A ⊆ U, B ⊆ V and U ∩ V = φ.
Remark 5.3.7. It is to be noted that every normal space is a regular space. 

124
Theorem 5.3.8. Every metric space (X, d) is a normal space, That is if Jd is the
topology induced by the metric then the topological space (X, Jd ) is a normal space.

Proof. Let A, B be disjoint closed subsets of X. Then for each a ∈ A, a ∈

/B=B
implies d(a, B) = inf{d(a, b) : b ∈ B} > 0. If ra = d(a, B) > 0 then B(a, ra ) ∩ B = φ
(if there exists b0 ∈ B such that d(b0 , a) < ra , then ra = d(a, B) ≤ d(a, b0 ) < ra a
contradiction). Similarly for each b ∈ B there exists rb > 0 such that B(b, rb )∩A = φ.
Let U = ∪ B(a, r3a ), V = ∪ B(a, r3b ). Now it is easy to prove that U ∩ V = φ.
a∈A b∈B
Hence if A, B are disjoint closed subsets of X then there exist open sets U, V in X
such that A ⊆ U, B ⊆ V and U ∩ V = φ. This implies (X, Jd ) is a normal space. 

Theorem 5.3.9. A T1 -topological space (X, J ) is regular if and only if whenever x

is a point of X and U is an open set containing x then there exists an open set V
containing x such that V ⊆ U.

Proof. Assume that (X, J ) is a regular topological space, x ∈ X and U is an open

/ A = U c = XKU, the complement of the
set containing x. Now x ∈ U implies x ∈
open set U. Now A is a closed set and x ∈
/ A. Hence X is a regular space implies there
exist open sets V and W of X such that x ∈ V , A = U c ⊆ W and V ∩ W = φ. Now
V ∩ W = φ implies V ⊆ W c ⊆ U (we have U c ⊆ W ), V ⊆ W c implies V ⊆ W c = W c
(W is an open set implies W c is a closed set) implies V ⊆ U . Hence for x ∈ X and
for each open set U containing x, there exists an open set V containing x such that
V ⊆ U.

Now let us assume that the above statement is satisfied. Our aim here is to
prove that (X, J ) is a regular space. So take a closed set A of X and a point x ∈ XKA.
Now A is a closed subset of X implies U = XKA is an open set containing x. Hence

125
by our assumption there exists an open set V containing x such that V ⊆ U = Ac .
Now V ⊆ Ac implies A ⊆ (V )c = XKV . So V and (V )c = W are open sets satisfying
x ∈ V, A ⊆ W and V ∩ W = V ∩ (V )c ⊆ V ∩ V c = φ. (V ⊆ V implies (V )c ⊆ V c .)
Hence by definition (X, J ) is a regular space. 

In a similar way we prove the following theorem.

Theorem 5.3.10. A T1 -topological space is a normal space if and only if whenever A
is a closed subset of X and U is an open set containing A, then there exists an open
set V containing A such that V ⊆ U.

Proof. Assume that (X, J ) is a normal topological space. Now take a closed set
A and an open set U in X such that A ⊆ U . Now A ⊆ U implies U c ⊆ Ac . Here
A, U c = B are closed sets such that A ∩ B = A ∩ U c ⊆ U ∩ U c = φ. That is A, B are
disjoint closed subsets of the normal space (X, J ). Hence there exist open sets U, W
in X such that A ⊆ V, B = U c ⊆ W and V ∩W = φ. Further V ⊆ W c (note: V ⊆ W c
implies V ⊆ W c = W c ). Now V ⊆ W c ⊆ U . Hence whenever A is a closed set and U
is an open set containing A then there exists an open set V such that A ⊆ V, V ⊆ U.
Now let us assume that the above statement is satisfied. So our aim is to prove that
(X, J ) is a normal space. So start with disjoint closed subsets say A, B of X. Now
A ∩ B = φ implies A ⊆ B c = U. That is U is an open set containing the closed set A.
Hence by our assumption there exists an open set V such that A ⊆ V, V ⊆ U . Now
V ⊆ U implies U c ⊆ (V )c implies B ⊆ (V )c . Further V ∩ (V )c ⊆ V ∩ V c = φ. That is
whenever A, B are closed subsets of X, then there exist open sets V and (V )c = W
such that A ⊆ V, B ⊆ W and V ∩ W = φ. Therefore by definition (X, J ) is a normal
space. 

126
Example of a topological space which is regular but not normal.
Let Jl be a lower limit topology on R. That is Rl = (R, Jl ). Now let us prove
that the product space Rl × Rl is a regular space. (If X, Y are regular topological
spaces then the product space X × Y is a regular space. Hence it is enough to prove
that Rl is a regular space.) For (x, y) ∈ R2 , each basic open set U of the form
U = [x, a) × [y, b) is both open and closed. Hence for each basic neighbourhood U of
(x, y) in Rl × Rl there exists a neighbourhood V = U of (x, y) such that V = U ⊆ U.
Now if U 0 is any open set containing (x, y) then there exists a basic open set U as
given above such that (x, y) ∈ U = [x, a) × [y, b) ⊆ U 0 . Therefore V = U is an open
set containing (x, y) and V = U = U ⊆ U 0 . Also Rl × Rl is a Hausdorff space. Hence
Rl × Rl is a regular space. Now let us take Y = {(x, y) ∈ R2 : y = −x} then for each
(x, y) ∈ Y there exists a, b ∈ R, x < a, y < b such that ([x, a) × [y, b)) ∩ Y = {(x, y)}.
Hence each singleton {(x, y)} is open in the subspace Y of Rl × Rl . This proves that
the subspace Y of Rl × Rl is discrete. Also Y is a closed subset of Rl × Rl . Let
A = {(x, y) ∈ R2 : y = −x ∈ Q}, B = {(x, y) ∈ R2 : y = −x ∈ Qc }. Now A, B are
closed sets in Y and Y is a closed set in Rl ×Rl implies A, B are closed in Rl ×Rl . Also
A ∩ B = φ. Suppose there exist open sets U, V in Rl × Rl satisfying A ⊆ U, B ⊆ V.
Then we can observe that U ∩ V 6= φ. Therefore the product space Rl × Rl is not a
normal space.
Remark 5.3.11. We can prove that (R, Jl ) = Rl is a normal space. So, Rl × Rl is a
regular space but it is not a normal space. 

We have already proved that every compact subset of Hausdorff topological

space is closed. Essentially we use the same proof technique used there to prove the
following theorem:

127
Theorem 5.3.12. Every compact Hausdorff topological space (X, J ) is a regular
space.

Proof. Let A be a closed subset of X and x ∈ XKA, then for each y ∈ A, x 6= y.

Hence X is a Hausdorff space implies that there exist open sets Uy , Vy in X satisfying
x ∈ Uy , y ∈ Vy and Uy ∩ Vy = φ. We know that closed subset of a compact space is
compact. Here A ⊆ ∪ Vy . That is {Vy : y ∈ A} is an open cover for the compact
y∈A
n
space A. Therefore there exists n ∈ N and y1 , y2 , . . . , yn ∈ A such that A ⊆ ∪ Vyi .
i=1
n n
Let U = ∩ Uyi and V = ∪ Vyi . Then U, V are open sets in X satisfying x ∈ U, A ⊆ V
i=1 i=1
and U ∩ V ⊆ U ∩ (Vy1 ∪ Vy2 ∪ · · · ∪ Vyn ) = (U ∩ Vy1 ) ∪ (U ∩ Vy2 ) ∪ · · · ∪ (U ∩ Vyn )
⊆ (Uy1 ∩ Vy1 ) ∪ (Uy2 ∩ Vy2 ) ∪ · · · ∪ (Uyn ∩ Vyn ) = φ. Hence by definition (X, J ) is a
regular space. 

Now let us prove that every compact Hausdorff space is a normal space.
Theorem 5.3.13. Every compact Hausdorff space (X, J ) is a normal space.

Proof. Let A, B be disjoint closed sets in X. Then for each x ∈ A, x ∈

/ B. Now
(X, J ) is a regular space implies there exist open sets Ux , Vx satisfying: x ∈ Ux ;
B ⊆ Vx and Ux ∩ Vx = φ. Now {Ux : x ∈ A} is an open cover for A implies there
n
exists n ∈ N, x1 , x2 , . . . , xn ∈ A such that A ⊆ ∪ Uxi . Let U = Ux1 ∪ Ux2 ∪ · · · ∪ Uxn
i=1
and V = Vx1 ∩ Vx2 ∩ · · · ∩ Vxn . Then U, V are open sets in X satisfying A ⊆ U, B ⊆ V
and U ∩ V = φ. Hence by definition (X, J ) is a normal space. 

Theorem 5.3.14. Closed subspace of a normal topological space (X, J ) is normal.

Proof. Let Y be a closed subspace of (X, J ). That is Y is a closed subset of (X, J )

and JY = {A ∩ Y : A ∈ J } is a topology on Y . So we will have to prove that (Y, JY )
is a normal space. To prove this, take a closed set A ⊆ Y and an open set U in (Y, JY )

128
such that A ⊆ U. Now U is an open set in (Y, JY ) implies there exists V ∈ J such
that U = V ∩ Y . Also A is a closed set in the subspace implies A = AY = A ∩ Y (here
AY denotes the closure of A in (Y, JY ) and A denotes the closure of A in (X, J )).
Now A, Y are closed sets in X implies A ∩ Y is also a closed set in X. Hence A is
a closed set in (X, J ) and V is an open set in (X, J ) containing A and (X, J ) is
a normal topological space implies there exists an open set W in (X, J ) such that
A ⊆ W and W ⊆ V. Now W ∩ Y is an open set in (Y, JY ) and A ⊆ W ∩ Y and
W ∩ Y ⊆ W ∩ Y ⊆ V ∩ Y ⊆ U . We started with a closed set A in (Y, JY ) and an
open set U in (Y, JY ) such that A ⊆ U. Now we have proved that there exists an
open set W ∩ Y in (Y, JY ) satisfying A ⊆ W ∩ Y and (W ∩ Y )Y = W ∩ Y ∩ Y =
W ∩ Y ⊆ U . That is W ∩ Y is an open set in the subspace containing A and closure
of this open set with respect to the subspace (Y, JY ) is contained in U . Hence (Y, JY )
is a normal space. 

5.4 Urysohn Lemma

Now let us prove the following important theorem known as Urysohn lemma.
Theorem 5.4.1. Let (X, J ) be a normal space and A, B be disjoint nonempty closed
subsets of X. Then there exists a continuous function f : X → [0, 1] such that
f(x) = 0 for every x in A, and f(x) = 1 for every x in B.

Proof. A ∩ B = φ implies A ⊆ B c = XKB. Hence B c is an open set containing

the closed set A. Now X is a normal space implies there exists an open set U0 such
that A ⊆ U0 and U 0 ⊆ B c = U1 . Now [0, 1] ∩ Q is a countable set implies there
exists a bijective function say f : N → [0, 1] ∩ Q satisfying f (1) = 1, f (2) = 0 and
f (NK{1, 2}) = (0, 1) ∩ Q. That is [0, 1] ∩ Q = {r1 , r2 , r3 , . . .} such that r1 = 1, r2 = 0

129
and f (k) = rk for k ≥ 3.
Aim: To define a collection {Up }p∈[0,1]∩Q of open sets such that for p, q ∈ [0, 1] ∩ Q,
p < q implies U p ⊆ Uq .
Let Pn = {r1 , r2 , . . . , rn }. Assume that Up is defined for all p ∈ Pn , where n ≥ 2 and
this collection satisfies the property namely p, q ∈ [0, 1] ∩ Q, p < q implies U p ⊆ Uq .
Note that this result is true when n = 2. Now let us prove this result for Pn+1 . Here
Pn+1 = Pn ∪ {rn+1 }.

0 p rn+1 q 1

Figure 5.2

Let p, q ∈ Pn+1 be such that p = max{r ∈ Pn+1 : r < rn+1 } and q = min{r ∈ Pn+1 :
r > rn+1 }. Now p, q 6= rn+1 implies p, q ∈ Pn . By our assumption Up , Uq are known
and U p ⊆ Uq . Now Uq is an open set containing the closed set U p and X is a normal
space. Hence there exists an open set say Urn+1 such that U p ⊆ Urn+1 and U rn+1 ⊆ Uq .
If r, s ∈ Pn then we are through.

Suppose r ∈ Pn and s = rn+1 then r ≤ p or r ≥ q. If r ≤ p, U r ⊆ Up ⊆ U p ⊆

Us . If r ≥ q, U s ⊆ Uq ⊆ U q ⊆ Ur and therefore by induction Up is defined for all
p ∈ [0, 1] ∩ Q and p, q ∈ [0, 1] ∩ Q, p < q implies U p ⊆ Uq .

Now define Up = φ, if p ∈ Q, p < 0 and Up = X if p ∈ Q, p > 1. Then

p, q ∈ Q, p < q implies U p ⊆ Uq . Define f : X → [0, 1] as f (x) = inf{p ∈ Q :
x ∈ Up }. Now x ∈ A, then x ∈ U0 . Hence x ∈ Up for all p ≥ 0. In this case
{p ∈ Q : x ∈ Up } = [0, ∞) ∩ Q. Hence inf{p ∈ Q : x ∈ Up } = 0.

130
 That is x ∈ A implies f (x) = 0. Now suppose x ∈ B = U1c then x ∈
/ Up for all
p ≤ 1. Hence {p ∈ Q : x ∈ Up } = [1, ∞) ∩ Q implies f (x) = 1 for all x ∈ B.

Now let us prove that f is a continuous function. S = {[0, a), (a, 1] : 0 < a < 1}
is a subbase for [0, 1]. Hence it is enough to prove that for each a, 0 < a < 1, f −1 ([0, a))
and f −1 ((a, 1])) are open sets in X. For 0 < a < 1, let us prove that f −1 ([0, a)) =
{x ∈ X : 0 ≤ f (x) < a} = ∪ Up . Now x ∈ f −1 ([0, a)) implies f (x) < a implies there
p<a
exists a rational number p such that f (x) < p < a. By the definition of f (x), x ∈ Up .
Hence
f −1 ([0, a)) ⊆ ∪ Up . (5.2)
p<a

Now let x ∈ Up for p < a implies f (x) ≤ p implies x ∈ f −1 ([0, a)). Hence we have

∪ Up ⊆ f −1 ([0, a)). (5.3)

p<a

From Eqs. (5.2) and (5.3) we have f −1 ([0, a)) = ∪ Up . Now each Up is an open set
p<a

implies that ∪ Up is an open set in X. In a similar way we can prove that f −1 ((a, 1])
p<a
is also an open subset of X for each 0 < a < 1. Now f : X → [0, 1] such that
inverse image of each subbasic open set is an open set implies that f : X → [0, 1] is
a continuous function. 

Theorem 5.4.2. Let (X, J ) be a normal space and A, B be disjoint nonempty closed
subsets of X. Then for a, b ∈ R, a < b there exists a continuous function f : X → [a, b]
such that f(x) = a for every x in A, and f(x) = b for every x in B.

Proof. Define g : [0, 1] → [a, b] as g(t) = a + (b − a)t then g is continuous. Now

by theorem 5.4.1 there is a continuous function f1 : X → [0, 1] such that f1 (x) = 0,
for all x ∈ A and f1 (x) = 1 for all x ∈ B. The function f = g ◦ f1 : X → [a, b]

131
is a continuous function and further f (x) = g(f1 (x)) = g(0) = a for all x ∈ A and
f (x) = g(f1 (x)) = g(1) = b for all x ∈ B. 

Remark 5.4.3. Let A, B be nonempty disjoint closed subsets of a metric space

d(x, A)
(X, d). Define f : X → R as f (x) = . Observe that f is a continuous
d(x, A) + d(x, B)
function satisfying the condition that f (x) = 0 for all x ∈ A and f (x) = 1 for all
x ∈ B. It shows that the proof of Urysohn lemma is trivial (or say simple) if our
topological space is a metrizable topological space. 

5.5 Tietze Extension Theorem

Theorem 5.5.1. Tietze Extension Theorem. Let A be a nonempty closed subset

of a normal space X and let f: A → [−1, 1] be a continuous function. Then there
exists a continuous function g : X → [−1, 1] such that g(x) = f(x) for all x in A.

Proof. The sets −1, −1

  1 
3
, 3 , 1 are closed subsets of [−1, 1] and f : A → [−1, 1]
is a continuous function implies A1 = f −1 31 , 1 , B1 = f −1 −1, −1
 
 

3
are closed
subsets of the subspace A. (Here consider A as a subspace of X.) Now x ∈ A1 ∩ B1
implies f (x) ∈ −1, −1
  1 
3
∩ 3 , 1 a contradiction. Hence A1 ∩ B1 = φ. Now A1 , B1
are closed in A and A is closed in X implies A1 , B1 are closed in the normal space
X. Hence by Urysohn’s lemma there exists a continuous function f1 : X → −1 1
 
,
3 3
1
such that f1 (A1 ) = 3
and f1 (B1 ) = − 13 then |f (x) − f1 (x)| ≤ 32 for all x ∈ A.
Now consider the function f − f1 : A → −2 , 2 then A2 = (f − f1 )−1 29 , 23 and
   

3 3

B2 = (f − f )−1 −2 , −2
 

3 9
are disjoint closed subsets of X. By Urysohn lemma there
exists a continuous function f2 : X → − 92 , 29 such that f2 (A2 ) = 29 and f2 (B2 ) = − 29 .
 

4
Also |f (x) − (f1 (x) + f2 (x))| ≤ 9
for all x ∈ A. By proceeding as above by induction

132
  −2n−1 n−1
, 2 3n

for each n ∈ N there exists a continuous function fn : X → 3n
such that
n
2 n
 X 

f (x) − fi (x) ≤ for all x ∈ A. (5.4)
3
i=1

That is fn : X → [−1, 1] is a sequence of continuous functions such that |fn (x)| ≤

∞ ∞
2n−1
P P
3n = M n and Mn < ∞. By Weierstrass M-test, the series fn (x) converges
n=1 n=1
n
P
uniformly on X. That is, if sn (x) = fi (x), x ∈ X then sn (x) converges uniformly
i=1
on X. Also each sn : X → R is continuous. We know, from analysis, if a sequence
sn : X → R of continuous functions converges uniformly to a function g : X → R then
P∞
g is also a continuous function. Hence g : X → R be defined as g(x) = fn (x) is
  n=1
 n 
n
fi (x) ≤ 23
P
continuous. Now for each x ∈ A, f (x) − (from (5.4)). Therefore
  i=1  
 n  Pn 
n
fi (x) − f (x) = lim  fi (x) − f (x) ≤ lim 23 = 0. This
P
|g(x) − f (x)| =  lim
n→∞ i=1 n→∞ i=1 n→∞
implies g(x) = f (x) for all x ∈ A. 

Definition 5.5.2. A topological space (X, J ) is said to be completely regular

if (i) for each x ∈ X, singleton {x} is closed in (X, J ) (that is (X, J ) is a T1 -space),
(ii) for x ∈ X and any nonempty closed set A with x ∈
/ A there exists a continuous
function f : X → [0, 1] such that f (x) = 0 and f (y) = 1 for all y ∈ A.
Result 5.5.3. Every normal space (X, J ) is completely regular.

Proof. Let x ∈ X and A be a nonempty closed set with x ∈

/ A. Now {x}, A are
disjoint closed sets. Hence by Urysohn’s lemma there exists a continuous function
f : X → [0, 1] such that f (x) = 0 and f (y) = 1 for all y ∈ A. 

Result 5.5.4. If Y is a subspace of a completely regular space (X, J ) then (Y , JY )

is also a completely regular space.

133
Proof. Let y ∈ Y and A be a closed set in (Y , JY ) with y ∈
/ A. Since A is a closed
set in Y there exists a closed set F in (X, J ) such that A = F ∩Y, y ∈
/ F , F is a closed
set in the completely regular space (X, J ) implies there exists a continuous function
f : X → [0, 1] such that f (y) = 0 and f (a) = 1 for all a ∈ F . Now f : X → [0, 1] is a
continuous function implies f |Y = g : (Y , JY ) → [0, 1] (here g(x) = (f |Y )(x) = f (x)
for all x ∈ Y ) is a continuous function. Now g: (Y , JY ) → [0, 1] is a continuous
function such that g(y) = f (y) = 0 and g(a) = f (a) = 1 for all a ∈ A = F ∩ Y . Also
subspace of a T1 -space (do it as an exercise) is T1 -space. Hence the subspace (Y , JY )
is a completely regular space. 

5.6 Baire Category Theorem

Baire category theorem has many applications in topology and analysis. Our
aim here is to state and prove this theorem for locally compact Hausdorff topological
spaces.
Definition 5.6.1. A subset A of a topological space (X, J ) is said to be nowhere
dense in X if and only if (A)◦ = int(A) = φ.
Example 5.6.2. (i) N is nowhere dense in R (R with standard topology).
(ii) Q is dense in R, that is Q = R, and hence Q is not nowhere dense in R. Here we
have (Q)◦ = R◦ = R 6= φ.
Definition 5.6.3. A topological space (X, J ) is said to be of first category if and
only if there exists a countable collection {En }∞
n=1 of subsets of X satisfying:

◦ ◦
(i) for each n ∈ N, En = En = φ, and
∞
(ii) X = ∪ En .
n=1

134
Definition 5.6.4. A nonempty subset Y of a topological space (X, J ) is said to be
of first category in X if and only if there exists a countable collection {En }∞
n=1 of

subsets of X satisfying:

◦
(i) for each n ∈ N, E n = φ, and
∞
(ii) Y = ∪ En .
n=1

Remark 5.6.5. If Y is a nonempty subset of a topological space (X, J ) then (Y, JY )

(JY = {U ∩ Y : U ∈ J }) is also a topological space. It is possible that a subset Y of
a topological space (X, J ) is of first category in (X, J ) but the subspace (Y, JY ) is
not of first category.
Example 5.6.6. Let X = R and Js be the standard topology on R. Then Y = N,
the set of all natural numbers, is of first category in R, but the subspace (N, Js/N )
is not of first category. For each n ∈ N let En = {n}. As En contains only one
∞ ∞
element namely n, (En )◦ = {n}◦ = φ in R. Also N = ∪ {n} = {1, 2, . . .} = ∪ En .
n=1 n=1

Hence N is of first category in (R, Js ). But note that the subspace (N, Js/Y ) is the
discrete topological space on N. For n ∈ N, (n − 1, n + 1) is an open set in R and
hence (n − 1, n + 1) ∩ N = {n} is an open set in the subspace (N, Js/N ). Now it is
easy to see that there cannot exist any countable collection {An }∞
n=1 of subsets of R
∞
satisfying (An )◦ = φ and ∪ An = N. Note that for An ⊆ N, (An )◦ = An with respect
n=1
to (N, Js/N ) and hence (N, Js/N ) is not of first category.
Definition 5.6.7. If a topological space (X, J ) is not of first category then we
say that the topological space (X, J ) is of second category.

Note. We have seen that N is of first category in (R, Js ) but the topological space
(N, Js/Y ) is of second category.

135
Now our main aim is to prove that every locally compact Hausdorff topological space
is of second category.
Remark 5.6.8. First let us prove that every locally compact Hausdorff topological
space (X, J ) is a regular space. So let us take a closed set A in (X, J ) and a point
x ∈ XKA.

We have seen that every compact Hausdorff space is a normal space and hence
every compact Hausdorff space is a regular space. We know that the one point of
compactification (X ∗ , J ∗ ) of (X, J ) is a compact Hausdorff space and J ∗ |X = J .
That is (X, J ) is a subspace of the compact Hausdorff space (X ∗ , J ∗ ). Also it is
easy to prove that subspace of a regular space is regular (and it is to be noted that
subspace of a normal space need not be a normal space) and hence (X ∗ , J ∗ ) is a
regular space implies the subspace (X, J ) of (X ∗ , J ∗ ) is also a regular space.

Now we are in a position to state and prove the main theorem.

Theorem 5.6.9. Baire Category Theorem. Let (X, J ) be a locally compact
Hausdorff topological space and {En }∞
n=1 be a countable collection of open sets in
∞
(X, J ). Further suppose for each n ∈ N, En = X (En is dense in X) then ∩ En is
n=1
∞

also dense in X. That is ∩ En = X.
n=1
∞
Proof. We want to prove that ∩ En is dense in X.
n=1
So take x ∈ X and an open set U containing x. Now (X, J ) is a locally compact
Hausdorff space implies there exists an open set V containing x such that V is
compact. Now let U0 = U ∩ V. Then U0 is an open set containing x. Also U0 ⊆ V
implies U0 is a compact set (since closed subset of compact set is compact). Now our
∞

aim is to prove that U ∩ ∩ En 6= φ. For each n, En is open and En = X, that is
n=1
each En is open and dense in X. Start with n = 1, now x ∈ X = E1 and U0 is an open

136
set containing x implies U0 ∩E1 6= φ. So take an element say x1 ∈ U0 ∩E1 . Now U0 , E1
are open sets implies U0 ∩E1 is also an open set. Now U0 ∩E1 is an open set containing
x1 and (X, J ) is a regular space (every locally compact Hausdorff space is a regular
space) implies there exists an open set U1 in X satisfying x1 ∈ U1 , U1 ⊆ U0 ∩ E1 . Now
x1 ∈ E2 = X implies U1 ∩ E2 6= φ. Let x2 ∈ U1 ∩ E2 . Since X is a regular space implies
there exists an open set U2 in X satisfying x2 ∈ U2 , U2 ⊆ U1 ∩E2 . Again x2 ∈ E3 = X
and U2 is an open set containing x2 implies U2 ∩ E3 6= φ. Let x3 ∈ U2 ∩ E3 . Choose an
open set U3 such that x3 ∈ U3 , U 3 ⊆ U2 ∩ E3 . Continuing in this way (that is using
induction) we get a sequence {xn }∞ ∞
n=1 in X and a sequence of open sets {Un }n=1
n

satisfying xn ∈ Un , U n ⊆ Un−1 ∩ En for all n ∈ N. Note that U n ⊆ U ∩ ∩ Ek for all
k=1
n ∈ N. Then {U k }∞
k=1 is a sequence of nonempty closed subsets of X and hence of the
compact subspace U 0 . Further U k+1 ⊆ U k for any k ∈ N implies {U k }∞
k=1 has finite

intersection property. That is {U k }∞

k=1 is a family of closed subsets of the compact

topological space U 0 and further {U k }∞

k=1 has finite intersection property. Therefore
∞ ∞
∩ U k 6= φ. Let a ∈ ∩ U k . Then a ∈ U k for all k ∈ N and hence a ∈ U. Also a ∈ En
k=1 k=1
∞ ∞

for all n ∈ N. So a ∈ ∩ En . Thus a ∈ U ∩ ∩ En . That is for each x ∈ X and
n=1 n=1
∞
∞

for each open set U containing x, U ∩ ∩ En 6= φ. Hence x ∈ ∩ En . This gives
n=1 n=1
∞
∞
∞
that X ⊆ ∩ En and hence X = ∩ En , that is ∩ En is dense in X. 
n=1 n=1 n=1

Exercise 5.6.10. Prove that a subset E of a topological space (X, J ) is nowhere

dense in X (that is (E)◦ = φ if and only if (E)c is dense in X.
Remark 5.6.11. It is known that every complete metric space X is of second
category. The notion of completeness cannot be defined in a topological space. So we
give the following version of Baire Category theorem for a locally compact Hausdorff
topological space.

137
Theorem 5.6.12. Every nonempty locally compact Hausdorff topological space
(X, J ) is of second category.

Proof. Proof by contradiction.

Suppose (X, J ) is of first category. Then there exists a countable collection {En }∞
n=1
◦ ∞ ∞
of subsets of X satisfying En = φ and X = ∪ En = ∪ E n . Therefore X c =
n=1 n=1
∞ c ∞ c c
φ = ∩ En and hence φ = φ = ∩ En . But {E n }∞
n=1 is a countable collection
n=1 n=1
∞ c
of dense open sets, implies by theorem 5.6.9 ∩ En = X 6= φ. This contradicts
n=1
∞ c
φ = φ = ∩ En . Hence (X, J ) is of second category. 
n=1

Now we are in a position to prove Urysohn metrization theorem that gives

sufficient conditions under which a topological space is metrizable. Also it is
interesting to note that the well known Nagata-Smirnov metrization theorem gives a
set of necessary and sufficient conditions for metrizability of a topological space.

5.7 Urysohn Metrization Theorem

Theorem 5.7.1. Urysohn Metrization Theorem. Every normal space (X, J )

with a countable basis is metrizable.

Proof. Let B = {B1 , B2 , . . . , } be a countable basis for (X, J ). Suppose n, m ∈ N

c
are such that B n ⊆ Bm then Bn ∩ Bm = φ. Hence by Urysohn’s lemma there exists
a continuous function say gn,m : X → R such that

c
gn,m (x) = 0 for all x ∈ Bm , (5.5)

and

gn,m (x) = 1 for all x ∈ Bn . (5.6)

138
Now take x0 ∈ X and an open set U containing x0 . Since B is a basis for (X, J )
there exists Bm ∈ B such that x0 ∈ Bm ⊆ U. Now Bm is an open set containing x0
implies there exists an open set V containing x0 such that V ⊆ Bm . Hence there
exists a basic open set Bn containing x0 such that B n ⊆ V ⊆ Bm . Hence for such
pair (n, m) we have a continuous function gn,m : X → R satisfying Eq. (5.5).

So if x0 ∈ X and U is an open set containing x0 then there exists a continuous

function gn,m : X → R such that gn,m (x0 ) = 1 and gn,m (x) = 0 for all x ∈ U c ⊆ Bm
c
.
So we have proved that there exists a countable collection of continuous functions
fn : X → [0, 1] such that for x0 ∈ X and open set U containing x0 , there exists
n ∈ N such that fn (x0 ) = 1 > 0 and fn (x) = 0 for all x ∈ U c . It is to be noted that
{(n, m) : n, m ∈ N} is a countable set. We know that (refer chapter 2, and exercise
9 of chapter 5) Rw = R × R × R × · · · with product topology is metrizable. That is
there is a metric d on Rw such that Jd , the topology on Rw induced by d, coincides
with the product topology on Rw .

Now let us define a map T : X → Rw as T (x) = (f1 (x), f2 (x), . . . , ) and using
this map we define d1 (x, y) = d(T (x), T (y)) and conclude that Jd1 = J . This will
prove that (X, J ) is a metrizable topological space. Now let us prove that (X, J )
is homeomorphic to a subspace of Rw . Each fn : X → R is a continuous function
implies T (x) = (f1 (x), f2 (x), . . .) is a continuous function.
To prove T is injective (one-one).

Let x, y ∈ X be such that x 6= y. Then there exist open sets U, V ∈ X such

that x ∈ U, y ∈ V and U ∩ V = φ. Now U is an open set containing x implies there
exists n ∈ N such that fn (x) = 1 and fn (y) = 0 (note that y ∈ U c ). This implies
fn (x) 6= fn (y) for this particular n ∈ N and hence (f1 (x), f2 (x), . . . , fn (x), . . .) 6=

139
(f1 (y), f2 (y), . . . , fn (y), . . .). This means T x 6= T y. That is x, y ∈ X, x 6= y implies
T x 6= T y. This implies T is 1-1.

Now it is enough to prove that T maps open set A in X to an open set T (A)
in Y = T (X). Let A be an open set and y0 ∈ T (A). Now y0 ∈ T (A) implies there
exists x0 ∈ A such that T (x0 ) = y0 . Now x0 ∈ A, A is an open set implies there exists
n0 ∈ N such that fn0 (x0 ) = 1 and fn0 (x) = 0 for all x ∈ Ac . We know that for each
n ∈ N the projection map pn : Rw → R defined as pn ((xk )∞
k=1 ) = xn is a continuous

map. Hence (0, ∞) is an open set implies V = p−1 w

n0 ((0, ∞)) is an open subset of R .

This implies V ∩ Y is an open set in Y .

Now let us prove that y0 ∈ V ∩ Y and V ∩ Y ⊆ T (A). pn0 (y0 ) = (pn0 · T )(x0 ) =
fn0 (x0 ) = 1 > 0 implies y0 ∈ V . Also y0 ∈ Y . Hence y0 ∈ V ∩ Y . That is V ∩ Y is
an open set in Y containing the point y0 .

Now we claim that V ∩ Y ⊆ T (A). So, let y ∈ V ∩ Y . Then there exists x ∈ X

such that y = T x. This implies pn0 (y) ∈ (0, ∞) and pn0 (y) = pn0 (T (x)) = fn0 (x) ∈
(0, ∞). Hence x ∈ A (fn0 (x) = 0 for x ∈ Ac ). So we have proved that y = T x ∈ V ∩Y
implies y = T x ∈ T (A). Hence V ∩ Y is an open set in Y containing T x and this
set is contained in T (A). Therefore T (A) is open in Y . Hence we have proved that
onto
T : (X, J ) −−→ (Y, dY ) is a homeomorphism. (Here (Y, dY ) is a subspace of (Rw , d).)
Now d1 (x, y) = d(T x, T y) for all x, y ∈ X implies d1 is a metric on X. Also it is easy
to see that a subset A of X is open in (X, J ) if and only if A is open in (X, Jd1 ).
Therefore Jd1 = J . 

140