TOPOLOGY - MATH 6620

DAVID TWEEDLE

These are course notes for Math 6620. These notes use the following sources [Mun00].

Contents
1. Definition of topological space and basis for a topological space 1
2. Product topology and subspace topology 4
3. Closure, interior, and limit points 5
4. Continuous functions 8
5. Arbitrary products 9
6. Metric spaces 10
7. Quotient spaces 13
8. Connectedness 14
9. Compactness 18
10. Countability axioms 23
11. Separation axioms 24
12. The Tychonoff Theorem 29
13. Paracompactness 30
14. Complete Metric Spaces 34
15. The Fundamental Group 37
16. Applications of the Fundamental Group 44
17. Covering Spaces 49
References 53

1. Definition of topological space and basis for a topological space

Definition 1.1. A topology on a set X is a collection T of subsets of X having the
following properties
a) ∅ ∈ T , X ∈ T S
b) if ∅ =
̸ U ⊂ T , then U ∈U U ∈ T
Tn
c) if n > 0 is an integer and U1 , U2 , . . . , Un ∈ T then i=1 Ui ∈ T .
A topological space is an ordered pair (X, T ) where T is a topology on X.
Definition 1.2. Let (X, T ) be a topological space. An element U ∈ T is called an open
set.
Definition 1.3. Let (X, T ) be a topological space. If F ⊆ X and F c = {x ∈ X | x ∈ / F}
is the complement of F , then F is called closed if F c is open. In other words, F is closed
if F c ∈ T .
Date: March 7, 2022.
1
Proposition 1.4. Let (X, T ) be a topological space.
a) ∅, X are closed T
b) if F is a collection of closed sets then F ∈F F is also closed
Sn
c) if F1 , F2 , . . . , Fn are closed then so is i=1 Fi is closed.
Proof. Apply De Morgan’s laws to the axioms in Definition 1.1. □

Definition 1.5. Let X be a set and let T and T ′ be two topologies on X. Then we say
that T is finer than T ′ if T ′ ⊂ T .
If T ′ ⊊ T then we say that T is strictly finer than T ′ .
There is a nice analogy in [Mun00] about finer topologies and gravel.
Maybe it might also help to think about the markings on a ruler. A ruler with finer
markings has more markings. A finer topology has more open sets.
Definition 1.6. Let (X be a set. A collection B of subsets of X is called a basis for a
topology on X if
S
a) For each x ∈ X, there exists B ∈ B such that x ∈ B. In other words, B = X
b) If x ∈ B1 ∩ B2 for some B1 , B2 ∈ B, then there exists B3 ∈ B such that x ∈ B3 ⊆
B1 ∩ B2 .
Lemma 1.7. Let B be a basis for a topology on X. Define a collection T by the following
rule, U ∈ T if for each x ∈ U , there exists B ∈ B such that x ∈ B ⊆ U .
Then (X, T ) is a topological space. We say that T is the topology generated by the basis
B.
S
Proof. Since B∈B B S= X we have X ∈ T . Also ∅ ∈ T (vacuously). Now suppose
{Uα } ⊂ T . Let U = Uα . Let x ∈ U . Then x ∈ Uα for some α. Since Uα ∈ T , there
exists B ∈ B with x ∈ B ⊆ Uα ⊆ U . This provesTn U ∈ T .
Now, suppose U1 , . . . , Un ∈ T . Let x ∈ i=1 Ui . Then there exists B1 , B2 , . . . , Bn ∈ B
with
x ∈ Bi ⊆ Ui
for i = 1, 2, 3, . . . , n. By the definition of basis Definition 1.6 and using induction, there
exists B ∈ B with x ∈ B ⊆ B1 ∩ B2 ∩ · · · ∩ Bn . This proves that U1 ∩ · · · ∩ Un ∈ T .
So T is a topology. □
Lemma 1.8. Let X be a set and B a basis for a topology on X, which generates T . Then
T equals the collection of all unions of elements of B.
In other words, ( )
[
T = B |A⊆B .
B∈A

Proof. Note that a union of elements of B is clearly in T . Let U ∈ T . We claim that

[
U= B.
B∈B
B⊆U

Clearly,
[
B ⊆ U.
B∈B
B⊆U
2
On the other hand, let x ∈ U . Then there exists B ∈ B with x ∈ B ⊆ U by Definition 1.1.
This shows the reverse inclusion, concluding the lemma. □

Lemma 1.9. Let X be a set. Let B be a basis for a topology T and B ′ be a basis for a
topology T ′ . The following are equivalent:
a) T ′ is finer than T
b) For each x ∈ X and each B ∈ B with x ∈ B, there exists B ′ ∈ B ′ such that
x ∈ B ′ ⊆ B.

Proof. Assume that T ′ is finer than T . Let x ∈ X and B ∈ B such that x ∈ B. Then
B ∈ T . Since T ′ is finer than T , B ∈ T ′ . So B is the union of the elements of B ′ contained
in B by Lemma 1.8. One of those B ′ ∈ B ′ with B ′ ⊆ B must contain x as x ∈ B. This B ′
satisfies the second item in the Lemma.
Suppose now that the second condition holds. Let B ∈ B. Then the second condition
implies that we can write B as the union of elements of B ′ contained in B. This implies
that B ∈ T ′ . This implies that T ⊆ T ′ . □

Lemma 1.10. Let (X, T ) be a topological space. Suppose that C ⊂ T is such that for each
open U ∈ T , and for each x ∈ U , there exists C ∈ C such that

x ∈ C ⊆ U.

Then C is a basis for T .

Proof. By hypothesis, we see that for each U ∈ T can be written

[
U= C.
C∈C
C⊆U

Therefore, if we can show that C is a basis, it will be a basis of T by Lemma 1.7 and
Lemma 1.9.
Let C1 , C2 ∈ C, and let x ∈ C1 ∩ C2 . Then C1 ∩ C2 ∈ T , so there exists C3 ∈ C with

x ∈ C3 ⊆ C1 ∩ C2 .
S
Also, if x ∈ X, then there exists C ∈ C such that x ∈ C as X = C∈C C since X is open
and by the equation above. So C is a basis, so it must be a basis for the topology T . □

Definition 1.11. Let X = R. Let B be the collection of all open intervals

[a, b] = {x ∈ R | a < x < b}
for all real numbers a < b in R.
Definition 1.12. Let X be a set. Let S be a collection of subsets of X. Then S is called
a subbasis for a topology on X if [
U = X.
U ∈S
The basis generated by S is the set of all finite intersections of elements of S and the
topology generated by S is the set of all arbitrary unions of finite intersections of elements
of S.
3
 2. Product topology and subspace topology
Definition 2.1. Let X and Y be topological spaces. Define
B = {U × V | U is an open set of X, V is an open set of Y }.
Then B is a basis for a topology of X × Y .
The topology that B generates is called the product topology for X × Y .
Proposition 2.2. Suppose X and Y are topological spaces. Then B = {U × V } where U
runs over all open sets in X and V runs over all open sets in Y is a basis.
Proof. Let x ∈ X × Y . Then X and Y are both open and so X × Y is in B and x ∈ X × Y .
So B satisfies the first condition of Definition 1.6.
Suppose U1 × V1 , U2 × V2 ∈ B (so that U1 , U2 , V1 , V2 are open sets. Suppose also that
x ∈ U1 × V1 ∩ U2 × V2 .
Then U1 ∩U2 and V1 ∩V2 are both open by Definition 1.1. But x ∈ (U1 ∩U2 )×(V1 ∩V2 ) =
(U1 × V1 ) ∩ (U2 × V2 ), and (U1 ∩ U2 ) × (V1 ∩ V2 ) ∈ B. So B is a basis. □
Theorem 2.3. Suppose B is a basis for the topology of X and C is a basis for the topology
of Y . Then D = {B × C | B ∈ B, C ∈ C} is a basis for the product topology on X × Y .
Proof. It is clear that D is a subset of the basis which defines the product topology Def-
inition 2.1. So the topology generated by D is coarser than the product topology, by
applying Lemma 1.7. Now, we want to show that D is finer than the product topology.
Let (x, y) ∈ U × V where U and V are open. Then x ∈ U and y ∈ V so there are B ∈ B
and C ∈ C such that x ∈ B ⊆ U and y ∈ C ⊆ V since B and C are bases. But then
(x, y) ∈ B × C ⊆ U × V . Now, by Lemma 1.9, we see that D generates a finer topology
than the one given in Definition 2.1. So it gives the same topology. □

Definition 2.4. Let X and Y be topological spaces. Then define pX : X × Y → X by

pX (x, y) = x and pY : X × Y → Y by pY (x, y) = y.
Theorem 2.5. Let
[
S = {p−1
X (U ) | U open in X} {p−1
Y (V ) | V open in Y }.

Then S is a subbasis for the product topology.

Proof. Let B = {U × V } be the basis for X × Y given by Definition 2.1. Then let B =
U × V ∈ B. Then B = p−1 −1
X (U ) ∩ pY (V ). So every element of B is an intersection of at
most two elements of S. In general a finite intersection of elements of S will be of the form
U × V since
p−1 −1 −1
X (U1 ) ∩ pX (U2 ) = pX (U1 ∩ U2 )
and similarly for pY . So S generates the basis given by Definition 2.1. □
Remark 2.6. We will come back to this after introducing the concept of continuous functions
(section 4). We will see that the product topology is the coarsest topology such that each
projection X × Y → X and X × Y → Y is continuous. Remember that the finer we make
the topology on X × Y , the easier it is for each projection to be continuous. So the product
topology is only as fine as we need it so that each projection is continuous.
Q
Q product i∈I Xi , we will choose
To define the product topology of an arbitrary cartesian
the coarsest possible one that makes each projection i∈I Xi → Xj continuous.
4
Definition 2.7. Let (X, T ) be a topological space. If Y is a subset of X, define
TY = {Y ∩ U | U ∈ T }.
This is a topology on Y , called the subspace topology. The topological space (Y, TY ) is
called a subspace of X (or (X, T )).
Proposition 2.8. The collection TY defines a topology on Y .
Proof. Notice ∅ = ∅ ∩ Y and Y = Y ∩ X.
Also [  [
Uα ∩ Y = (Uα ∩ Y )
and !
n
\ n
\
Ui ∩Y = (Ui ∩ Y ).
i=1 i=1
This is enough to see that Definition 1.1 is satisfied. □

Lemma 2.9. Let (X, T ) be a topological space with basis B. Let Y ⊆ X. Then BY =
{B ∩ Y | B ∈ B} is a basis for the subspace topology on Y .
Proof. Apply Lemma 1.10. Let V = U ∩ Y where U is open in X and suppose x ∈ V .
Then x ∈ U and so there exists B ∈ B so that x ∈ B ⊆ U . But then x ∈ B ∩ Y ⊆ U ∩ Y
and B ∩ Y ∈ BY . So BY is a basis for Y by Lemma 1.10. □

Lemma 2.10. Let Y be a subspace of X. If U is open in Y (that is, U ∈ TY and Y is

open in X (that is, Y ∈ T ), then U is open in X (that is, U ∈ T ).
Proof. Since U is open in Y , U = Y ∩ V for some open V in X, and since Y, V are open
in X so is U = Y ∩ V . □

Theorem 2.11. Let (X, T ) and (Y, U) be topological spaces with subspaces A and B re-
spectively. Then we can view A × B as a subspace of X × Y , and we can view A × B as a
product of the subspaces A and B. These give the same topology on A × B.
Proof. A subbasis for A × B when viewed as a subspace of X × Y would be {(A × B) ∩ (U ×
Y )} ∪ {(A × B) ∩ (X × V )}. This is equal to the subbasis {(A ∩ U ) × B} ∪ {A × (V ∩ B)}
which is the subbasis you get when you look at the product topology on A × B. □

3. Closure, interior, and limit points

Theorem 3.1. Let (X, T ) be a topological space and Y be a subspace of X. Then if A is
a subset of Y , A is closed in Y if and only if A = Y ∩ F for some closed set F of X.
Proof. If A = F ∩ Y , for F a closed subset of X, then Y − A = Y ∩ (X − F ) which is the
intersection of Y and an open set, so is open in Y . So A is closed. Now, if Y − A is an
open set, then Y − A = Y ∩ U for some open set U . Then notice that A = Y ∩ (X − U )
and X − U is closed. □

Theorem 3.2. Let (X, T ) be a topological space and Y be a subspace of X. If A is closed

in Y and Y is closed in X then A is closed in X.
Proof. Follows because the intersection of closed sets is closed Proposition 1.4. □
5
Definition 3.3. Let (X, T ) be a topological space. Let A ⊆ X be a subset. Define the
interior of A, denoted Int(A), by
[
Int(A) = U.
U ∈T ,
U ⊆A

Definition 3.4. Let (X, T ) be a topological space. Let A ⊆ X be a subset. Define the
closure of A, denoted Cl(A), by the rule
\
Cl(A) = F.
F closed in X,
A⊆F

Proposition 3.5. Let A be a set. Then

Int(A) ⊆ A ⊆ Cl(A).
We have that x ∈ Int(A) if and only if there exists open U such that x ∈ U ⊆ A if and
only if for all closed sets F with F ∩ A = ∅, we have x ∈/ F . We have that x ∈ Cl(A) if
and only if for all closed F with A ⊆ F , we have x ∈ F if and only if for all open U with
x ∈ U , U ∩ A ̸= ∅. We have
Int(A) ∪ Int(B) ⊂ Int(A ∪ B)
Int(A ∩ B) = Int(A) ∩ Int(B)
Cl(A ∩ B) ⊂ Cl(A) ∩ Cl(B)
Cl(A ∪ B) = Cl(A) ∪ Cl(B).
Finally,
Int(Ac ) = (Cl(A))c
Cl(Ac ) = (Int(A))c .
Proof. Omitted, except let us prove that x ∈ Cl(A) if and only if for all open U with x ∈ U ,
U ∩ A ̸= ∅. Suppose x ∈ Cl(A). Let U be open with x ∈ U . Suppose U ∩ A = ∅. Then let
F = X − U . It follows that A ⊆ F . Since x ∈ Cl(A) ⊆ F , we must have x ∈ F . So x ∈ / U.
Therefore, A ∩ U ̸= ∅.
Now, suppose that for all open U with x ∈ U , we have U ∩ A ̸= ∅. Let F be closed and
contain A. Then if x ∈
/ F then x ∈ X − F = U which implies that U ∩ A ̸= ∅ which implies
that F does not contain A. This is a contradiction so x ∈ F . So x ∈ Cl(A). □
Theorem 3.6. Let (X, T ) be a topological space and Y a subspace of X. Let A be a subset
of Y . Let Cl(A) be the closure of A in X.
Then the closure of A in Y is equal to Cl(A) ∩ Y .
Proof. By Theorem 3.1, if F is closed in Y then F = Y ∩G for some closed G in X. So each
closed set of Y containing A may be written Y ∩ G where G is closed in X and contains A.
Therefore, ClY (A) = ∩A⊆G G ∩ Y = (∩A⊆G G) ∩ Y = Cl(A) ∩ Y . □
Theorem 3.7. Let (X, T ) be a topological space and let A ⊆ X.
a) x ∈ Cl(A) if and only if U ∩ A ̸= ∅ for all open U with x ∈ U
b) x ∈ Cl(A) if and only if B ∩ A ̸= ∅ for all B ∈ B with x ∈ B, where B is a basis
for T .
6
Proof. The first part was proved in Proposition 3.5.
The second part follows from the first part. □
Definition 3.8. Let (X, T ) be a topological space and let A ⊆ X. Let x ∈ X. Then x is
called a limit point of A if for all open U with x ∈ U , there exists y ̸= x with y ∈ U ∩ A.
Theorem 3.9. Let (X, T ) be a topological space and let A ⊆ X. Let Lim(A) be the set of
limit points of A. Then Cl(A) = A ∪ Lim(A).
Proof. Let x ∈ Cl(A) and x ∈ / A. Then by Theorem 3.7, U ∩ A ̸= ∅ for all open U with
x ∈ U . Let U be open and x ∈ U . As U ∩ A ̸= ∅, y ∈ U ∩ A for some y. As x ∈
/ A, y ̸= x.
Clearly A ⊆ Cl(A). Suppose x ∈ Lim(A). Let U be open with x ∈ U . Then U ∩ A ̸= ∅
by Definition 3.8. So x ∈ Cl(A) by Theorem 3.7. □
Definition 3.10. Let (X, T ) be a topological space and let A ⊆ X. The boundary of A
is defined to be Cl(A) − Int(A).
Corollary 3.11. Let (X, T ) be a topological space and A ⊆ X. Then A is closed if and
only if Lim(A) ⊆ A.
Proof. Note that A is closed if and only if A = Cl(A). Since Cl(A) = A ∪ Lim(A), the
result follows. □
Definition 3.12. Let (X, T ) be a topological space. Then X is called a Hausdorff space if
for each x, y ∈ X with x ̸= y, there exists U, V ∈ T such that x ∈ U, y ∈ V and U ∩ V = ∅.
Theorem 3.13. Let (X, T ) be a Hausdorff topological space. Then if A ⊆ X and A is a
finite set, then A is closed.
Proof. Let x ∈ X. Then we claim that Lim({x}) is empty. Suppose y ∈ X with y ̸= x.
Since X is Hausdorff, find open U and V such that U ∩ V = ∅ and x ∈ U and y ∈ V . Now,
V ∩ {x} = ∅ so there is no z ∈ V ∩ {x} with z ̸= y. Therefore, y ∈
/ Lim{x}. So Lim{x} = ∅.
So {x} is closed for any x ∈ X. Since closed sets are closed under finite unions, any finite
set is closed. □
Theorem 3.14. Let (X, T ) be a Hausdorff topological space. Let A ⊆ X. The point x ∈ X
is a limit point of A if and only if for every open U with x ∈ U , U contains infinitely many
points of A.
Proof. Suppose that x is a limit point of A. Let U be an open set with x ∈ U . Since x is a
limit point of A there exists y0 ∈ A ∩ U with y0 ̸= x. Now, suppose that we have a nested
sequence of open sets
U = U0 ⊃ U1 ⊃ U2 ⊃ · · · ⊃ Un
and points
y0 ∈ U0 ∩ A, y1 ∈ U1 ∩ A, . . . , yn ∈ Un ∩ A.
but constructed in such a manner so that yi ∈ / Ui+1 for i = 0, 1, 2, . . . , n − 1 and x ∈ Un .
Since X is Hausdorff, we can find open sets V, W such that yn ∈ V and x ∈ W such
that V ∩ W = ∅. Let Un+1 = W ∩ Un .
Then x ∈ Un+1 since x ∈ W and x ∈ Un . Also, Un+1 ⊆ Un . But yn ∈ / Un+1 as yn ∈/ W.
Since Un+1 is open and x ∈ Un+1 there exists yn+1 ∈ Un+1 ∩ A where yn+1 ̸= x.
By induction now, we have a sequence y0 , y1 , . . . , yn , . . ., and notice that each yn is
distinct from the previous terms in the sequence. In other words, we have infinitely many
points of A in U . □
7
 4. Continuous functions
Definition 4.1. Let X and Y be topological spaces. Let f : X → Y be a function. We
say that f is continuous if for all open V of Y
f −1 (V ) = {x ∈ X f (x) ∈ V }
is open in X.
Remark 4.2. Let X and Y be topological spaces and let B be a basis for Y . To check that
f : X → Y is continuous, we need only check that f −1 (B) is open in X for each B ∈ B.
Theorem 4.3. Let X and Y be topological spaces. Let f : X → Y be a function. Then
the following are equivalent:
a) f is continuous
b) f (Cl(A)) ⊆ Cl(f (A)) for all A ⊆ X
c) for all closed B ⊆ Y , f −1 (B) is closed in X.
Proof. Recall that
f −1 (Y − U ) = X − f −1 (U ).
So if f is continuous, then the inverse image of any closed set is a closed set. The same
argument works in reverse.
Now, suppose f −1 (B) is closed for all closed sets B. Then write
\
Cl(A) = B,
A⊆B
B closed
Now recall that
f (∩Vα ) ⊆ ∩f (Vα ),
so that \ \
f (Cl(A)) ⊆ f (G) ⊆ F = Cl(f (A)).
G closed F closed
A⊆G f (A)⊆F
On the other hand, suppose f (Cl(A)) ⊆ Cl(f (A)) for all A. Let B be closed, and let
A = f −1 (B). Then B = f (A) is closed, so
f (Cl(A)) ⊆ B.
−1
So Cl(A) ⊆ f (B) = A so A is closed. □
Definition 4.4. Let X and Y be topological spaces. Let f : X → Y be a function. We
say that f is a homeomorphism if f is continuous and has a continuous inverse.
Definition 4.5. Let X and Y be topological spaces. Suppose that f : X → Y is a function.
We say that f is an imbedding if f is a homeomorphism from X to Z = f (X).
Theorem 4.6. Let X, Y and Z be topological spaces.
a) If f : X → Y is defined by f (x) = y0 for all x ∈ X, where y0 ∈ Y , then f is
continuous.
b) If A is a subspace of X, the inclusion function j : A → X is continuous.
c) If f : X → Y and g : Y → Z are continuous then so is g ◦ f : X → Z defined by
(g ◦ f )(x) = g(f (x)) for all x ∈ X.
d) If f : X → Y is continuous and A is a subspace of X then the restriction of f to
A, denoted f |A : A → Y , is also continuous.
8
 e) Iff : X → Y is continuous and if Z is a subspace of Y which contains f (X) then
g : X → Z defined by g(x) = f (x) for all x ∈ X is continuous. If Y is a subspace of
Z then the function h : X → Z defined by h(x) = f (x) for all x ∈ X is continuous.
f) The function
S f : X → Y is continuous if there exists open sets {Uα }α∈I such that
X = α∈I Uα and each f |Uα : Uα → Y is continuous.
g) The function f : X → Y is continuous if for each x ∈ X and each open V of Y
with f (x) ∈ V , there exists open U of X such that f (U ) ⊂ V .

Proof. Omitted, except we will show that the inclusion function j : A → X is continuous
with respect to the subspace topology on A. Let U be open in X. Then j −1 (U ) = U ∩ A
which is open in A by definition of the subspace topology. So j is continuous. In fact, we
see that the definition of subspace topology (Definition 2.7) is made exactly so that the
inclusion function is continuous. □

Theorem 4.7. Let X = A ∪ B where A and B are both closed in X. Let f : A → Y and
g : B → Y be continuous. Suppose f (x) = g(x) for all x ∈ A ∩ B. Then the function
h : X → Y defined by

f (x) if x ∈ A
h(x) =
g(x) if x ∈ B
is well-defined and continuous.

Proof. The assumption that f (x) = g(x) for all x ∈ A ∩ B assures that h is well-defined.
Let F be closed in Y . Write h−1 (F ) = h−1 (F ) ∩ A ∪ h−1 (F ) ∩ B = f −1 (F ) ∩ g −1 (F ) which
is closed as both f, g are continuous and F closed. □

Theorem 4.8. Let f : A → X × Y be defined by f (a) = (f1 (a), f2 (a)) where f1 : A → X

and f2 : A → Y . Then f is continuous if and only if the functions f1 and f2 are continuous.

Proof. First, note that pX : X × Y → X and pY : X × Y → Y are both continuous.

Suppose that f : A → X × Y is continuous. Then f1 = pX ◦ f is the composition of
continuous maps, so by Theorem 4.3, f1 is continuous. Similarly, f2 is continuous.
Now, suppose that f1 and f2 are continuous. Consider U × V ⊆ X × Y where U, V are
open in X and Y , respectively. Then f −1 (U × V ) = (f1−1 (U )) ∩ (f2−1 (V )) which is open
as f1 and f2 are continuous.
So the inverse image of U × V is open, and the sets of the form U × V form a basis for
X ×Y.
Now, write any open set of X × Y as the union of basis elements and use the formula

f −1 (∪Uα ) = ∪f −1 (Uα ). □

5. Arbitrary products
Remark 5.1. Notice that the product topology on X × Y is the coarsest topology for which
the projections X × Y → X andQ X × Y → Y are both continuous.
For an arbitrary product,
Q α∈I Xα we can define a topology on the product which
makes each projection α∈I Xα → Xβ continuous for each β ∈ I.
9
Definition 5.2. Suppose {Xα }α∈I is a collection of topological spaces. Let pβ :
Q
α∈I Xα → Xβ be the projection onto the Xβ factor. That is, pβ ((xα )α∈I ) = xβ . Define
Sβ = {p−1
β (U ) | U is open inSXβ } for each β.
Define the subbasis S = β∈I Sβ . The product topology is defined to be the topology
Qthe subbasis S. By construction it is the coarsest topology for which all the
generated by
projections Xα → Xβ are continuous.

Q Suppose each factor Xα has a basis Bα . Then a basis for the product
Theorem 5.3.
topology on Xα is given by the collection of all sets of the form
Y
Uα
α∈I

where Uα ∈ Bα for finitely many indices α ∈ I and Uα = Xα for the remaining indices.
Proof. Notice that finite intersections of elements of S give the basis listed above. □
Q
Theorem Q 5.4. Suppose Aα is a subspace of Xα for each α ∈ I. Then α∈I Aα is a sub-
space of Xα . It has the same Q topology whether you consider it as a product of subspaces
or as a subspace of the product Xα .
Proof. Similar to Theorem 2.11. □
Q
Theorem 5.5. If each Xα is Hausdorff then so is Xα .
Q
Proof. Let (xα ), (yα ) ∈ Xα with (xα ) ̸= (yα ). Then there exists some index β such that
xβ ̸= yβ . There exists open Uβ , Vβ such that xβ ∈ Uβ , yβ ∈ Vβ and Uβ ∩ Vβ = ∅ and Uβ , Vβ
open in Xβ , since Xβ is Hausdorff. Define U to be the product of Xα when α ̸= β and Uβ ,
and similarly for V . Then U ∩ V = ∅ and (xα ) ∈ U , (yα ) ∈ V and U and V are open. □
Q
Theorem 5.6. Suppose f : A → Xα is given by the formula f (a) = (fα (a))α∈I where
fα : A → Xα . Then f is continuous if and only if each fα is continuous.
Proof. Proof is similar to Theorem 4.8. □

6. Metric spaces
Definition 6.1. Let X be a set. A metric on X is a function
d:X ×X →R
such that
a) d(x, y) ≥ 0 for all x, y ∈ X, with equality if and only if x = y
b) d(x, y) = d(y, x) for all x, y ∈ X
c) d(x, z) ≤ d(x, y) + d(y, z) for all x, y, z ∈ X.
If d is a metric on X then we say that the ordered pair (X, d) is a metric space.
Definition 6.2. Let X be a metric space. Let ϵ > 0 be a real number. Define the ϵ-ball
centred at x ∈ X, denoted Bd (x, ϵ), by
Bd (x, ϵ) = {y ∈ X | d(x, y) < ϵ}.
Definition 6.3. Let (X, d) be a metric space. Define Bd = {Bd (x, ϵ)} where x ranges over
all x ∈ X and ϵ ranges over all real ϵ > 0. Then Bd is a basis for a topology on X called
the metric topology.
10
Definition 6.4. Let (X, T ) be a topological space. Then X is said to be metrizable if
there is a metric d whose metric topology is equal to T .
Definition 6.5. For x, y ∈ Rn define
p
d(x, y) = (x1 − y1 )2 + · · · (xn − yn )2 ,
we call d the Euclidean metric on Rn .
For x, y ∈ Rn define
ρ(x, y) = max {|xi − yi |}.
i=1,2,...,n
Then ρ is called the square-metric.
Lemma 6.6. Let d and d′ be two metrics on the set X with corresponding topologies T
and T ′ . Then T ′ is finer than T if and only if for each x ∈ X and ϵ > 0 there exists δ > 0
such that
Bd′ (x, δ) ⊆ Bd (x, ϵ).
Proof. We apply Lemma 1.9, and note that the open sets Bd′ form the basis for T ′ and
Bd form the basis for T .
fix this
□
Lemma 6.7. Let ρ and d be the square and Euclidean distances on R, respectively. For
all x, y ∈ Rn ,
√
ρ(x, y) ≤ d(x, y) ≤ nρ(x, y).
Proof. Note that
p
|xi − yi | ≤ (xi − yi )2
≤ d(x, y)
p
≤ n max{|xi − yi |2 }
√
= nρ(x, y).
As ρ(x, y) = max{|xi − yi |} we are done. □
Theorem 6.8. The topologies induced by the metrics d and ρ on Rn are both equivalent
to the product topology on Rn .
Proof. Apply the inequality from Lemma 6.7 to show that
√
Bρ (x, ϵ/ n) ⊆ Bd (x, ϵ) ⊆ Bρ (x, ϵ)
and then apply Lemma 6.6.
also argue that square-metric gives the product topology
□
Definition 6.9. Let X = Rω - the space of sequences of real numbers - in other words the
space of functions a : N → R.
Define a metric D on X by the rule
 
min{|xi − yi |, 1}
D(x, y) = sup .
i
11
Theorem 6.10. The function D is a metric on Rω and the topology induced by D is the
product topology.
Proof. A basis for Rω is given by
I1 × I2 × · · · × In × R × R × · · ·
where Ij are open intervals. Let xi ∈ Ii for each i = 1, 2, . . . , n. Then choose ϵi < 1 so that
(xi − ϵi , xi + ϵi ) ⊆ Ii for each i = 1, 2, 3, . . . , n. Then choose ϵ = min{ϵ1 , ϵ2 /2, . . . , ϵn /n}.
Then check that BD (x, ϵ) ⊆ I1 × I2 × · · · × In × R × · · ·. By Lemma 1.9, the topology given
by D is finer than the product topology.
Let x and ϵ > 0 be given. Suppose that D(x, y) < ϵ. Then for i ≤ 1/ϵ, we have that
|xi −yi |
i ≤ D(x, y). For i > 1/ϵ, we have that min{|xi − yi |, 1}/i ≤ 1/i < ϵ. So we see that
if Ii = (xi − iϵ, xi + iϵ) then
I1 × I2 × · · · × In × R × R × · · · ⊆ D(x, y).
This proves that the product topology is finer than the metric topology coming from D.

fix minor errors

□
Theorem 6.11. Let X and Y be metric spaces with metrics dX and dY , respectively. Let
f : X → Y be a function. Then f is continuous if and only if for all x ∈ X and ϵ > 0,
there exists δ > 0 such that if dX (x, y) < δ then dY (f (x), f (y)) < ϵ.
Proof. Omitted, but it is standard. □
Definition 6.12. A sequence in X is a function a : N → X. We may use the notation
(xn ) or (x1 , x2 , . . . , xn , . . .) to denote a sequence in X.
We say that a sequence (xn ) in X converges to x ∈ X if for every open U with x ∈ U ,
there exists an integer N such that if n ≥ N , xn ∈ U .
If (xn ) converges to x we write xn → x as n → ∞.
Lemma 6.13. Let (X, T ) be a topological space. Let A ⊆ X. If there is a sequence of
points of A converging to x, then x ∈ Cl A. The converse holds if X is metrizable.
Proof. Recall Theorem 3.7. Suppose that (xn ) converges to x and xn ∈ A for all n ≥ 1.
Then if U is a given open set with x ∈ U , there exists N such that if n ≥ N then xn ∈ U .
Thus U ∩ A ̸= ∅ for all open U with x ∈ U . So x ∈ Cl(A).
Suppose X is metrizable, that is, that the topology of X comes from a metric d. Suppose
x ∈ Cl(A). Then for each n ≥ 1 define Un = Bd (x, 1/n). Since x ∈ Cl(A), there exists
xn ∈ Un ∩ A for all n ≥ 1. Let U be open with x ∈ U . Then there exists ϵ > 0 such
that Bd (x, ϵ) ⊆ U . Choose N so that 1/N ≤ ϵ. Then Un ⊆ U and in particular xn ∈ U if
n ≥ N . So xn → x as n → ∞. □
Theorem 6.14. Let f : X → Y where X is a metric space and Y is a topological space.
The function f is continuous if and only if for every convergent sequence xn → x we have
f (xn ) → f (x) as n → ∞.
Proof. Let us apply Theorem 4.3. Suppose that for every convergent sequence xn → x,
we have f (xn ) → f (x) as n → ∞. Suppose that A ⊆ X. We will check that f (Cl(A)) ⊆
Cl(f (A)). Let x ∈ Cl(A). By Lemma 6.13, there exists xn → x. By hypothesis f (xn ) →
f (x) as n → ∞. So f (x) ∈ Cl(f (A)) by Lemma 6.13.
12
 Now, suppose that f is continuous and let xn → x. Let V be an open set of Y such
that f (x) ∈ V . Then x ∈ f −1 (V ) so there exists N such that if n ≥ N then xn ∈ f −1 (V ).
Then f (xn ) ∈ V . So by Definition 6.12, f (xn ) → f (x) as n → ∞. □

Lemma 6.15. The addition, subtraction and multiplication operations are continuous
functions R × R → R. The quotient operation is a continuous function R × (R − {0}) → R.
Proof. Omitted, it may be a good exercise to try to remember them from calculus. □

Theorem 6.16. If X is a topological space and f, g : X → R are continuous then f +g, f −g

and f g are continuous. If g(x) ̸= 0 for all x then f /g is continuous.
Proof. Combine Lemma 6.15 and Theorem 4.6. □

Definition 6.17. Let fn : X → Y be a sequence of functions from X to the metric space

Y . Let d be the metric for Y . We say that (fn ) converges uniformly to f : X → Y if given
ϵ > 0 there exists N such that
d(fn (x), f (x)) < ϵ
for all n ≥ N and x ∈ X.
Theorem 6.18. Let fn : X → Y be a sequence of continuous functions from X → Y
where X is a topological space and Y is a metric space. If (fn ) converges uniformly to f ,
then f is continuous.
Proof. Let ϵ > 0 and x ∈ X. There exists N such that for n ≥ N and z ∈ X we have
d(fn (z), f (z)) < ϵ/3.
Choose n = N and since fn is a continuous function there exists δ > 0 such that if
d(x, y) < δ then d(fn (x), fn (y)) < ϵ/3. So suppose d(x, y) < δ. Then by the triangle
inequality

d(f (x), f (y))

≤ d(f (x), fn (x)) + d(fn (x), fn (y)) + d(fn (y), f (y))
< ϵ/3 + ϵ/3 + ϵ/3 = ϵ
proving that f is continuous by Theorem 6.11. □

7. Quotient spaces
Definition 7.1. Let X and Y be topological spaces. Let p : X → Y be a surjective
function. The map p is said to be a quotient map if for all subsets U of Y , U is open in Y
if and only if p−1 (U ) is open in X.
Definition 7.2. Let X be a topological space and A a set. Let p : X → A be a surjective
function. Then there exists a unique topology on A for which p is a quotient map. This
topology is called the quotient topology induced by p.
Definition 7.3. Let X be a topological space and ∼ an equivalence relation on X. Let
X/ ∼ be the set of equivalence classes. Then define the quotient topology of X/ ∼ to be
the quotient topology induced by the canonical map X → X/ ∼.
13
Theorem 7.4. Let p : X → Y be a quotient map. Let Z be a topological space. Suppose
that g : X → Z is a continuous function such that is constant on each set p−1 ({y}) for all
y ∈ Z. In other words, if p(x) = p(x′ ) then g(x) = g(x′ ). Then there exists a continuous
map f : Y → Z such that f ◦ p = g.

Proof. Let y ∈ Y , suppose p(x) = y (since p is onto). Then define f (y) = g(x). This is
well-defined by our hypothesis that g is constant on each set p−1 ({y}). Certainly f ◦ p = g
by construction. It remains to check that f is continuous. Let U be open in Z. Let
V = f −1 (U ). Then V is open in Y if and only if p−1 (V ) is open in X. But p−1 (V ) = g −1 (U )
which is open since g is continuous. So f is continuous. □

Theorem 7.5. Let g : X → Z be a surjective continuous map. Let ∼ be the equivalence

relation x ∼ x′ if and only if g(x) = g(x′ ). Give X/ ∼ the quotient topology.
a) If Z is Hausdorff, so is X/ ∼.
b) There is a bijective continuous map f : X/ ∼→ Z such that g = f ◦ p where
p : X → X/ ∼ is the canonical map. The map f is a homeomorphism if and only
if g is a quotient map.

Proof. Define p : X → X/ ∼ to be the quotient map. Since g : X → Z is constant on the

sets p−1 ({z}), there exists continuous f : X/ ∼→ Z such that f ◦ p = g. By construction
f is bijective as well.
Now, suppose that Z is Hausdorff. Let [x1 ], [x2 ] ∈ X/ ∼ be such that [x1 ] ̸= [x2 ], where
[x] denotes the equivalence class of x modulo ∼.
Suppose f ([x1 ]) = y1 and f ([x2 ]) = y2 . Then y1 ̸= y2 and there exists U, V open in
Z such that y1 ∈ U and y2 ∈ V and U ∩ V = ∅. But since f is continuous f −1 (U ) and
f −1 (V ) are open in X/ ∼ and they separate [x1 ] and [x2 ] (since f is bijective f −1 (U ) and
f −1 (V ) are disjoint). So X/ ∼ is Hausdorff.
Suppose that f : X/ ∼→ Z is a homeomorphism. Then U is open in Z if and only if
f −1 (U ) is open in X/ ∼, if and only if p−1 (f −1 (U )) = g −1 (U ) is open in X, by Defini-
tion 7.2. In other words, g is a quotient map.
Now, suppose that g is a quotient map and let h be the inverse of f . Let U be a subset
of X/ ∼. Then U is open if and only if p−1 (U ) is open in X as p is a quotient map. But
p−1 (U ) = g −1 (f (U )) is open if and only if f (U ) is open as g is a quotient map. So U
is open if and only if f (U ) is open, and since f is bijective, we can conclude that f is a
homeomorphism. □

8. Connectedness
Definition 8.1. Let X be a topological space. A separation of X is a pair U, V of non-
empty open sets of X such that
a) U ∩ V = ∅
b) U ∪ V = X.
The topological space X is said to be disconnected if there exists a separation of X, and
connected if there is no separation of X.

Lemma 8.2. If Y is a subspace of X and A, B ⊆ Y such that A ∩ B = ∅, A ∪ B = Y and

A, B are non-empty, then A, B is a separation of Y if and only if A does not contain any
limit point of B, and B does not contain any limit point of A.
14
Proof. Suppose that A and B form a separation of Y . Then A and B are both closed
relative to Y . So A = A ∪ Lim(A) and B = B ∪ Lim(B) (the limit points being those
relative to Y ). Since A and B are disjoint, B contains no limit point of A and vice verse.
Suppose now that A does not contain any limit point of B, and B contains no limit
point of A (relative to Y ). If we show that A and B are closed, then they are both open,
and so form a separation. Now, Lim(A) ⊆ Y = A ∪ B but Lim(A) has no point in B so
Lim(A) ⊆ A. Similarly, Lim(B) ⊆ B. By Theorem 3.9, B and A are closed, and so A and
B are open, so form a separation of Y . □
Lemma 8.3. Suppose C, D is a separation of X and Y is a connected subspace of X.
Then either Y ⊆ C or Y ⊆ D.
Proof. Let U = C ∩ Y and V = D ∩ Y . Then U ∩ V = ∅, U ∪ V = Y and U and V are
open. Since Y is connected, either U = ∅ or V = ∅. This means either Y ⊆ C or Y ⊆ D
(since C ∪ D = X). □
Theorem 8.4. Suppose {Uα }α∈I is a collection of subsets of X such that
a) there exists x ∈ X such that x ∈ Uα for all α ∈ I
b) Uα is connected for all α ∈ I.
Then [
Uα
α∈I
is connected.
Proof. Suppose V, W are open sets in X such that ∪Uα ⊆ V ∪ W , V ∩ W is disjoint from
∪Uα . We need to show that either V , W are disjoint from ∪Uα .
Let α be fixed. Since Uα ⊆ V ∪ W and Uα is connected, either Uα ⊆ V or Uα ⊆ W .
WLOG say Uα ⊆ V . Let β be any other index. Then by a similar argument, Uβ ⊆ V or
U . If Uβ ⊆ U then it follows that Uβ ∩ Uα = ∅. This is a contradiction as Uα and Uβ
contain a common point. So all Uβ ⊆ V . SO ∪Uβ ⊆ V . This establishes that U ∩ ∪Uβ = ∅
as required. □
Theorem 8.5. Suppose that A is a connected subset of X. Suppose that A ⊆ B ⊆ Cl(A).
Then B is connected.
Proof. Suppose that B ⊆ U ∪ V , where U, V are open and U ∩ V is disjoint from B and
B ∩ U and B ∩ V are non-empty. Then A ⊆ U , without loss of generality, since A is
connected (otherwise, U, V will give a separation of A, as well). Now, let x ∈ Lim(A) ∩ B.
If x ∈ V , then a limit point of U ∩ B is contained in V ∩ B, so U ∩ B, V ∩ B is not a
separation of B, by Lemma 8.2. This is a contradiction. So Lim(A) ∩ B ⊆ U . In other
words Cl(A) ∩ B ⊆ U as B ⊆ Cl(A) = A ∪ Lim(A), by Theorem 3.9. □
Theorem 8.6. Let f : X → Y be a continuous function and suppose that X is connected.
Then f (X) is connected.
Proof. Suppose f (X) ⊆ U ∪ V and U ∩ V is disjoint from f (X). Then f −1 (U ) and f −1 (V )
is possibly a separation of X (both are open, both are disjoint, their union is X), except
that one may be empty. Since X is connected either f −1 (U ) or f −1 (V ) is empty. In other
words f (X) ⊆ V or f (x) ⊆ U . This proves that f (X) is connected. □
Q
Theorem 8.7. Suppose Xα is connected for all α ∈ I. Then α∈I Xα is also connected.
15
Proof. First, we show that X × Y is connected if X and Y are. If x ∈ X and y ∈ Y then
X × {y} and {x} × Y are both connected, being the images of the connected spaces X
and Y , by Theorem 8.6. So X × {y} ∪ {x} × Y is connected being the union of connected
spaces with the point (x, y) in common. Call this set Tx,y . Now, look at ∪y∈Y Tx,y . This
is the union of connected spaces each containing (x, y) so it is connected. But this union
is the entire space X × Y .
By induction, we may now conclude that any finite cartesian product of connected spaces
is connected. Q
Now, let X = Xα be a product of connected spaces. Choose x ∈ X.
For a finite subset J ⊆ I, define X(J) to be those y ∈ X such that yα = xα for all
α ∈ I − J. Q
Notice that X(J) is homeomorphic to the product α∈J Xα which is connected as it is
a finite product of connected spaces.
Let Y be the union of all the spaces X(J) as J ranges over all finite subsets of the index
set I. Then Y is connected by Theorem 8.4. Q
We now claim that Cl(Y ) = X. Let y ∈ X. Let Uα be a basis element of the product
topology for X (see Definition 5.2), so that Uα = Xα except for α ∈ J where J is finite.
Then construct z by putting zα = yα for α ∈ J and zα = xα for α ∈ / J. So z ∈ Y by
construction. So every basis element which contains y contains a point of Y . So y ∈ Cl(Y ).
So Cl(Y ) = X. By Theorem 8.5, X is connected. □

Theorem 8.8. Let a < b where a, b ∈ R. Let I = (a, b) = {x ∈ R | a < x < b}. Then I is
connected.
Proof. Let U and V be a separation of I. We may assume that a ∈ U . Let U ′ = {x | (a, x) ⊆
U }. Let x0 be the supremum of U ′ , as U ′ ⊆ (a, b) x0 exists and we have x0 ≤ b. Notice that
/ U . Since if x0 ∈ U , then some interval (x0 − ϵ, x0 + ϵ) ⊆ U which means x0 + ϵ ∈ U ′ ,
x0 ∈
a contradiction. So x0 ∈ / U . So x0 ∈ V OR x0 = b If x0 ∈ V then (x0 − ϵ, x0 + ϵ) ⊆ V
as V is open. This means that a, x0 − ϵ/2) is not contained in U so contradicting the
definition of x0 . So x0 ∈
/ V and so x0 = b. This proves that (a, b) ⊆ U so V is empty, a
contradiction. □

Theorem 8.9. Let f : X → R be a continuous function. Suppose that X is connected. If

a, b ∈ X and r ∈ R is such that f (a) < r < f (b) (or alternatively, f (b) < r < f (a)), then
there exists c ∈ X such that f (c) = r.
Proof. Suppose for a contradiction that f (x) ̸= r for all x ∈ X. Let U = f −1 ((−∞, r))
and V = f −1 ((r, ∞)). Then clearly, U and V are disjoint open sets whose union is X.
Furthermore, a ∈ U and b ∈ V so both are non-empty. This implies that U and V form a
separation of X, which is a contradiction. So r = f (x) for some x ∈ X. □

Definition 8.10. Let X be a topological space. Let x, y ∈ X. Then a path from x to y

is a continuous function f : [a, b] → X such that f (a) = x and f (b) = y, where [a, b] is a
closed interval in R.
We say that X is path-connected if for all x, y ∈ X there exists a path from x to y.
Definition 8.11. Let X be a topological space. Define an equivalence relation ∼ on X
by the rule that x ∼ y if and only if x, y ∈ C for some connected subspace C of X.
An equivalence class under ∼ is called a connected component of X.
16
Remark 8.12. Note that ∼ is an equivalence relation. Symmetry and reflexivity are clear.
The proof that ∼ is transitive uses Lemma 8.3.
Theorem 8.13. Let X be a topological space. The union of the components of X is X.
The components of X are pairwise disjoint. If A is a connected subspace of X then A is
contained in exactly one of the components of X.
Proof. Use Remark 8.12 to deduce the first two parts, and Lemma 8.3 to prove that any
connected subspace of X must be contained in exactly one connected component. □
Definition 8.14. Let X be a topological space. Define an equivalence relation ∼ on X
by saying x ∼ y if there is a path from x to y.
The equivalence classes of ∼ are called the path components of X.
Remark 8.15. That ∼ is reflexive and symmetric is clear. To prove that ∼ is transitive,
suppose f : [a, b] → X is a path from x to y and g : [b, c] → X is a path from y to z (we
may assume that g is presented this way), then define h : [a, c] → X by h(u) = f (u) if
a ≤ u ≤ b and h(u) = g(u) for b ≤ u ≤ c. Then h is continuous by the pasting lemma
Theorem 4.7.
Theorem 8.16. Let X be a topological space. The union of the path-components of X is
X. The path-components of X are pairwise disjoint. If A is a path-connected subspace of
X then A is contained in exactly one of the path-components of X.
Proof. These conclusions follow since path-connected components are defined by an equiv-
alence relation. For the third conclusion, suppose A is a path-connected subspace of X.
Then let a ∈ A. Then a must be in one of the path-connected components of X, say C.
Then any element of A must also be contained in C by transitivity. So A ⊆ C. □
Definition 8.17. Let X be a topological space. Then X is said to be locally connected
at x if for every open set U with x ∈ U , there exists an open set V such that x ∈ V ⊆ U
and V is connected.
We say that X is locally connected if it is locally connected at x for all x ∈ X.
Remark 8.18. We can see that X is locally connected if and only if there exists a basis for
X consisting of connected sets.
Definition 8.19. Let X be a topological space. Then X is said to be locally path-
connected at x if for every open set U with x ∈ U , there exists an open set V such that
x ∈ V ⊆ U and V is path-connected.
We say that X is locally path-connected if it is locally path-connected at x for all x ∈ X.
Theorem 8.20. Let (X, T ) be a topological space. Then X is locally connected if and only
if for every open U of X, each component of U is open in X.
Proof. Suppose X is locally connected, let U be an open set. Let C be a connected
component of U . We claim that C is open. Let x ∈ C. There exists a connected open
set V with x ∈ V ⊆ U . Since V is connected, we have V ⊆ C. Therefore, x ∈ Int(C) by
Proposition 3.5. So Int(C) = C and C is open.
Now, suppose that whenever U is open in X, each component of U is open. This directly
implies that X is locally-connected. □
Theorem 8.21. The topological space X is locally path connected if and only if for every
open set U of X, each path component of U is open in X.
17
Proof. Suppose X is locally path connected and let U be an open set. Let P be a path-
component of U . Let x ∈ P . Then there exists open, path-connected V with x ∈ V ⊆ U ,
and since V is path-connected, V ⊆ P by Theorem 8.16. This implies that x ∈ Int(P ) by
Proposition 3.5. So P is open as Int(P ) = P . □

Theorem 8.22. If X is a topological space, each path component of X lies in a component

of X. If X is locally path connected, then the components and the path components are the
same.
Proof. Note that any path-connected set is also connected. (This follows since any interval
is connected Theorem 8.8).
Let C be a connected component. Then C is the disjoint union of path-components
contained in C by Theorem 8.16. Each path-component of C is open since X is locally
path-connected by Theorem 8.21. So C must have exactly one path-component (else the
path-components would form a separation of C). So C must be path-connected. □

9. Compactness
Definition 9.1. Let X be a topological space.
S An open covering of X is a collection
{Uα }α∈I of open subsets of X such that X = α∈I Uα .
Definition 9.2. Let X be a topological space. We say that the open cover {Uα } admits Sn a
finite subcover if there exists a finite set {U1 , U2 , . . . , Un } ⊆ {Uα } such that X = i=1 Ui .
We say that X is compact if every open cover admits a finite subcover.
Lemma 9.3. Let Y be a subspace of X. ThenSY is compact if and only if for every
collection {Uα }Sof open sets (of X) such that Y ⊆ Uα admits a finite subset {U1 , . . . , Un }
n
such that Y ⊆ i=1 Ui .
Proof. Combine the definition of the subspace topology Definition 2.7 and Definition 9.2.
□

Theorem 9.4. Let X be a topological space and A ⊆ X be a closed subset. If X is compact

then A is compact.
Proof. Let {Uα } be an open cover of A. Let U be the complement of A in X, and note
that U is open in X.
Now, the collection {Uα } ∪ {U }} is an open cover of X. Since X is compact, the cover
admits a finite subcover
U1 , U2 , . . . , Un
and with the possible inclusion of U . Since A ⊆ X = U1 ∪ · · · ∪ Un ∪ U and A ∪ U = ∅, it
must be that A ⊆ U1 ∪ · · · ∪ Un , as required. □

Theorem 9.5. Let X be a Hausdorff topological space. Let A ⊆ X be a compact subset.

Then A is closed.
Proof. Let U be the complement of A. Let x ∈ U . For each y ∈ A, there exists two open
sets Vy and Wy with
a) x ∈ Vy
b) y ∈ Wy
c) Vy ∩ Wy = ∅
18
The collection {Wy } forms an open cover of A. There exists y1 , . . . , yn ∈ A such that
A ⊆ Wy1 , . . . , Wyn . Let V = Vy1 ∩ · · · ∩ Vyn and notice that V is an open set that is
completely disjoint from A. In other words, x ∈ V ⊆ U . So U is an open set. So A is
closed. □
Lemma 9.6. Let X be a Hausdorff topological space. Let A ⊆ X be a compact set and
x ∈ X − A. Then there exists open sets U, V with A ⊆ U , x ∈ V and U ∩ V = ∅.
Proof. Proved in the course of the proof of Theorem 9.5. □
Theorem 9.7. Let f : X → Y be a continuous function and suppose X is compact. Then
f (X) is compact.
Proof. Let {Uα } be an open cover of f (X). Then {f −1 (Uα )} is an open cover of X. So
it admits a finite subcover, X ⊆ f −1 (U1 ) ∪ · · · ∪ f −1 (Un ). So f (X) ⊆ U1 ∪ · · · ∪ Un as
required. □
Theorem 9.8. Suppose X is a compact topological space, and Y is a Hausdorff topological
space. Suppose f : X → Y is a continuous bijection. Then f is a homeomorphism.
Proof. Let F be a closed subset of X. Then F is compact by Theorem 9.4. Then f (F )
is a compact subset of Y by Theorem 9.7. Then f (F ) is a closed subset of Y since Y is
Hausdorff by Theorem 9.5. Since f is a bijection, we have (f −1 )−1 (F ) = f (F ), so that
f −1 is continuous (the inverse image of closed sets are closed Theorem 4.3). □
Lemma 9.9. Suppose X is a topological space and Y is a compact topological space. If N
is an open set of X × Y containing the slice x0 × Y of X × Y , then N contains some tube
W × Y where W is an open set in X containing x0 .
Proof. Since N is open, by definition of the product topology, we may write
[
N= Uα × Vα ,
where Uα are open sets in X, Vα are open sets in Y . Not only that, but the slice x0 × Y
is compact as Y is compact (Theorem 9.7). So we have
x0 × Y ⊆ U1 × V1 ∪ · · · ∪ Un × Vn .
Let W = U1 ∩ U2 ∩ · · · ∩ Un and notice that W × Y ⊆ N . □
Theorem 9.10. Let X1 , . . . , Xn be compact topological spaces. Then X1 × X2 × · · · × Xn
is compact.
Proof. By induction, we may assume that X and Y are compact (and we must show X × Y
is compact). So let {Uα } be an open cover of X × Y . Let x ∈ X. Since x × Y is compact
(by Theorem 9.7), we may cover x × Y by U1 , . . . , Un . But now let N = U1 ∪ · · · ∪ Un and
apply Lemma 9.9, to see that there exists open W such that W × Y ⊆ N . Label Wx = W .
Now, the collection {Wx } forms an open cover of X, so we may find x1 , . . . , xm such that
X = Wx1 ∪ · · · ∪ Wxm .
So we have
X × Y = ∪m i=1 Wxi × Y
and each Wxi × Y is covered by finitely many Uα , and so X × Y is covered by finitely many
Uα . So X × Y is compact. □
19
Definition 9.11. Let X be a topological space. Let C be a collection of subsets of X. The
collection C satisfies the finite intersection property (FIP) if for every finite subcollection
{C1 , C2 , . . . , Cn } ⊆ C
the intersection
n
\
Ci
i=1
is non-empty.

Theorem 9.12. Let X be a topological space. Then X is compact if and only if for T every
collection C of closed sets satisfying the finite intersection property, the intersection C∈C C
is non-empty.
Proof. Suppose X is compact. Let {Fα } be a collection of closed sets satisfying
T the FIP.
Let Uα = X − Fα for each α. If, for a contradiction, the intersection Fα = ∅, then
{Uα } is an open cover of X. But X is compact so X = U1 ∪ U2 ∪ · · · ∪ Un . This implies
F1 ∩ F2 ∩ · · · ∩ Fn = ∅, a contradiction to FIP of {Fα }. T
Now, suppose that every collection {Fα } of closed sets satisfying T
FIP has Fα ̸= ∅.
Let Uα be an open cover of X. Let Fα = X − Uα for each α. So Fα = ∅, so {Fα }
cannot satisfy FIP by hypothesis. So there is some finite subcollection F1 , . . . , Fn such
that F1 ∩ F2 ∩ · · · ∩ Fn = ∅.
So X = U1 ∪ · · · ∪ Un as required. □

Corollary 9.13. Let X be a topological space. Then X is compact if and only if for every
collection A satisfying the finite intersection property, the intersection
\
Cl(A)
A∈A

is non-empty.
Proof. Omitted. □

Theorem 9.14. Let a < b be two real numbers and let I = [a, b]. Then I is compact.
Proof. Let {Uα } be an open cover of [a, b]. For a contradiction assume that no finite
subcollection of {Uα } covers [a, b]. It follows that
a) no finite subcollection of {Uα } covers [a, (a + b)/2], or
b) no finite subcollection of {Uα } covers [(a + b)/2, b]
because if we can find finite subcovers of both halves of the interval, then we can combine
them to get a finite subcover of the entire interval.
Now, inductively construct a sequence of intervals
[a0 , b0 ] ⊃ [a1 , b1 ] ⊃ · · · ⊃ [an , bn ] ⊃ · · ·
such that
a) a0 = a, b0 = b
b) [ai , bi ] is not covered by any finite subcollection of {Uα } for all i = 0, 1, 2, . . .
c) ai+1 equals either ai or (ai + bi )/2, and bi+1 = (ai + bi )/2 or bi+1 = bi
d) bn − an = (b − a)/2n
20
 Now, let a∗ be the supremum of {an }. Choose U ∈ {Uα } such that a∗ ∈ U . Since U is
open and a∗ ∈ U , there exists ϵ > 0 such that
(a∗ − ϵ, a∗ + ϵ) ⊆ U.
Choose n to be large enough so that (b − a)/2n < ϵ. Now, an ≤ a∗ ≤ bn and bn − an =
(b − a)/2n . Therefore, [an , bn ] ⊆ (a∗ − ϵ, a∗ + ϵ) ⊆ U . This is a contradiction as [an , bn ] is
to have no finite subcover. □
Theorem 9.15. Let A ⊆ Rn . Then A is compact if and only if it is closed and bounded
(relative to either the Euclidean metric or square-metric).
Proof. Suppose A is compact. Then A is closed as Rn is Hausdorff and by Theorem 9.5.
Also, the balls B(x, N ) for N = 1, 2, 3, . . . form an open cover of A. Since A is compact
there is a finite subcover. This implies that A is bounded.
Suppose now that A is bounded (say by the square-metric). Then there exists an interval
[a, b] such that
A ⊆ [a, b]n .
Also, [a, b]n is compact since [a, b] is compact by Theorem 9.10. Therefore, A is compact
by Theorem 9.4. □
Theorem 9.16. Let f : X → R be a continuous function and suppose X is compact. Then
there exists c, d ∈ X such that f (c) ≤ f (x) ≤ f (d) for all x ∈ X.
Proof. Note that f (X) is compact by Theorem 9.7. So f (X) is closed and bounded by
Theorem 9.15. Let y be the supremum of f (X). Then y ∈ f (X) as f (X) is closed (let
ϵ > 0, then there exists y ′ = f (x′ ) such that y − ϵ < y ′ < y by definition of supremum).
Similarly, the infimum of f (X) is in f (X). □
Theorem 9.17. Let X be a non-empty compact Hausdorff space. If every point of X is a
limit point, then X is uncountable.
Proof. Suppose X = {x1 , . . . , xn , . . .} is countable and Hausdorff and in which every point
is a limit point. If X is finite, then no point of X is a limit point (the topology is the
discrete topology since X is Hausdorff).
First, construct open U1 so that x1 ∈ / Cl(U1 ). Find U, W with x1 ∈ W and x2 ∈ U with
U, W open and disjoint. Then x1 ∈ / Cl(U ). So take U1 = U .
Now, suppose we have non-empty open sets
U1 ⊃ U2 ⊃ · · · ⊃ Un
Construct subsequence xij and open non-empty Uj with xij ∈ / Cl(Uj )
Given xi1 , . . . , xin and U1 , . . . , Un . Now, Un is non-empty. Let xin+1 ∈ Un . Since xin+1
is a limit point, there exists y ̸= xin+1 in Un . Find U, W with xin+1 ∈ U and y ∈ W , U, W
are disjoint open sets.
Let Un+1 = W ∩ Un and y ∈ Un+1 but xin+1 ∈ / Cl(Un+1 ).
Construct U1 with x1 ∈ / Cl(U1 ). Construct U2 with x1 , x2 ∈ / Cl(U2 ).
Suppose x1 , . . . , xn ∈/ Cl(Un ) and Un ̸= ∅. If xn+1 ∈ / Cl(Un ) we are done. As xn+1 is
a limit point there exists xm ∈ Un with xm ̸= xn+1 . Let V, W be disjoint open sets such
that xm ∈ W and xn+1 ∈ V . Let Un+1 = W ∩ Un . Then xm ∈ Un+1 so Un+1 is non-empty.
But also xn+1 ∈ / Cl(Un+1 ). Now, notice that we proceed to construct a sequence of nested,
non-empty closed sets whose intersection is empty, so X is not compact.
21
tidy up
□
Definition 9.18. Let X be a topological space. Then X is called limit-point compact if
every infinite subset of X has a limit point. That is, if A is an infinite subset of X then
Lim(A) ̸= ∅.
Definition 9.19. Let X be a topological space. Then X is called sequentially compact if
every sequence a : N → X has a convergent subsequence.
Theorem 9.20. Let X be a compact topological space. Then X is limit-point compact.
Proof. Let A be a subset of X. We will show that if Lim(A) = ∅ then A is finite. If
Lim(A) = ∅ then A is closed and hence compact. If Lim(A) = ∅, then for all a ∈ A there
are open Ua such that Ua ∩ A = {a}. As {Ua } forms an open cover of A, there is a finite
subcover, say Ua1 , . . . , Uan . But then A = {a1 , . . . , an } as required. □
Theorem 9.21. Let X be a metric space that is limit-point compact. Then X is sequen-
tially compact.
Proof. Suppose that X is limit-point compact and let {xn } be a sequence in X. If {xn } is
finite as a set, then it has a convergent subsequence by the pigeonhole principle. Otherwise
{xn } is infinite, and so Lim({xn }) ̸= ∅ by the definition of limit-point compactness. Let
x0 ∈ Lim{xn }. Let Un = B(x0 , 1/n) for n = 1, 2, 3, . . .. Since x0 ∈ Lim({xn }) there exists
i1 such that xi1 ∈ B(x0 , 1/n) and xi1 ̸= x0 . Choose i1 as small as possible. Now choose
n so that 1/n < d(x0 , xi1 ). Then there exists i2 so that xi2 ∈ B(x0 , 1/n). And since
d(x0 , xi2 ) < d(x0 , xi1 ) it must be that i2 > i1 .
Continue in this fashion to construct xi1 , xi2 , . . . with i1 < i2 < i3 < · · · with d(xij , x0 ) →
0.
tidy proof
□
Lemma 9.22. Let X be a metric space that is sequentially compact. Let A be an open
cover. Then there exists a δ > 0 such that for each subset of B of X with diam(B) < δ,
there exists A ∈ A such that B ⊆ A.
Recall that
diam(B) = sup{d(x, y) | x, y ∈ B}.
Proof. Let {Uα } be an open cover of X. Suppose the conclusion of the lemma is false.
So for all δ > 0, there exists a subset B of X with diam(B) < δ and B ⊆ Uα for no α.
For n = 1, 2, 3, . . . consider δ = 1/n and choose Cn with diam(Cn ) < 1/n and Cn is not
contained in any Uα . Now, let us choose xn ∈ Cn for each n = 1, 2, 3, . . . (if Cn = ∅ then
certainly Cn ⊆ Uα ). We claim that {xn } has no convergent subsequence.
Let x ∈ X and let {xij } be a subsequence of {xn } and suppose xij → x as j → ∞.
There exists Uα with x ∈ Uα . Find ϵ with x ∈ B(x, ϵ) ⊆ Uα . Then there exists N such
that j ≥ N implies xij ∈ B(x, ϵ). Now, notice that Cn ⊆ B(x, ϵ) as long as 1/n < ϵ/2 by
the triangle inequality:
d(x, y) ≤ d(x, xij ) + d(xij , y) < ϵ + 1/n

22
tidy proof
□
Theorem 9.23. Let f : X → Y be a continuous function, where X is a compact metrix
space, and Y is a metric space. Then f is uniformly continuous: for every ϵ > 0 there
exists δ > 0 such that for all x, y ∈ X, if dX (x, y) < δ, then dY (f (x), f (y)) < ϵ.
Proof. Let ϵ be given. Then f −1 (B(y, ϵ/2)) is an open cover of X. Choose δ according to
Lemma 9.22 with the open cover {f −1 (B(y, ϵ/2))}.
Suppose that d(x, y) < δ. Then there exists yi such that x, y ∈ f −1 (B(yi , ϵ/2)). There-
fore, d(f (x), f (y)) ≤ d(f (x), yi ) + d(yi , f (y)) ≤ ϵ/2 + ϵ/2 = ϵ as required. □
Theorem 9.24. Let X be a metric space. The following are equivalent:
a) X is compact
b) X is limit-point compact
c) X is sequentially compact
Proof. We have shown that the first condition implies the second, and the second implies
the third for metric spaces. We have shown that the third implies the Lebesgue number
lemma Lemma 9.22. So suppose that X is sequentially compact and let Uα be an open
cover of X. Then there exists δ such that if diam(C) < δ then there exists α such that
C ⊆ Uα . First, we claim that {B(x, ϵ)} has a finite subcover of X, where ϵ > 0.
Suppose not, then given x1 choose x2 ∈ / B(x1 , ϵ), and given x1 , . . . , xn with xi ∈
/
B(x1 , ϵ) ∪ · · · B(xi−1 , ϵ) for i = 1, 2, . . . , n choose xn+1 ∈/ B(x1 , ϵ) ∪ · · · ∪ B(xn , ϵ).
We claim that xn has no convergent subsequence.
Once this is done X = B(x1 , ϵ) ∪ · · · ∪ B(xn ), ϵ). Now, choose ϵ so that B(xi , ϵ) has
diameter < δ.
Then B(xi , ϵ) ⊆ Ui for some Ui in the collection and so X = U1 ∪ · · · ∪ Un .
tidy proof
□

10. Countability axioms

Definition 10.1. Let X be a topological space. Then X is said to have a countable basis
at x ∈ X, if there exists a countable collection B of open sets such that for every open U
with x ∈ X, there exists B ∈ B with x ∈ B ⊆ U .
If X has a countable basis at x for every x ∈ X, then we say that X satisfies the first
countability axiom, or we say that X is first countable.
Theorem 10.2. Suppose that X is a first countable topological space.
a) If x ∈ X, then x ∈ Cl(A) if and only if there is a sequence (an ) of points of A
converging to x.
b) The function f : X → Y is continuous if and only if for every convergent sequence
xn → x, f (xn ) → f (x) as n → ∞.
Proof. Suppose that X is first countable. Let x ∈ X and suppose x ∈ Cl(A). If x ∈ A
then choose an = x for all n ≥ 1. Otherwise, there exists a countable basis {Un } at x. We
may assume that U1 ⊃ U2 ⊃ U3 ⊃ · · ·. Choose an ∈ Un for each n ≥ 1. Let U be open
with x ∈ U . Then x ∈ Un ⊆ U for some n ≥ 1. By construction xn , xn+1 , . . . ∈ Un ⊆ U so
that xn → x as n → ∞.
23
 Conversely, suppose that an → x as n → ∞ and an ∈ A for all n ≥ 1. Let U be open
and suppose x ∈ U . Then there exists n such that xj ∈ U for all j ≥ n. In particular,
U ∩ A ̸= ∅ for any open U with x ∈ U .
Now, f : X → Y is continuous if and only if f (Cl(A)) ⊆ Cl(f (A)). Now, suppose
an → x, then f (x) ∈ Cl({f (xn }) which implies that f (xn ) → f (x) as n → ∞.
On the other hand suppose if xn → x then f (xn ) → f (x). Let x ∈ Cl(A). Then an → x
with {an } ⊆ A. Then f (xn ) → f (x) by hypothesis. So f (x) ∈ Cl(f (A)) as required.
Tidy proof
□
Definition 10.3. Let X be a topological space. We say that X is second countable if
there exists a countable basis B for X.
Theorem 10.4.
a) A subspace of a first countable space is first countable,
b) a countable product of first countable spaces is first countable,
c) A subspace of a second countable space is second countable,
d) a countable product of second countable spaces is second countable.
Proof. Omitted. □

Definition 10.5. Let X be a topological space. Then a subset A of X is said to be dense

if Cl(A) = X.
Theorem 10.6. Suppose that X is second countable. Then
a) Every open covering of X admits a countable subcover.
b) There exists a countable subset of X which is dense in X.
Proof. Let B = {Bn } be a countable basis for X. Let {Uα } be an open cover. For each
n = 1, 2, . . . choose Ui in the cover such that Bi ⊆ Ui (unless you can’t in which case take
Ui = ∅, say). We claim that U1 , U2 , . . . , Un , . . . is an open cover of X. Let x ∈ X. Then
x ∈ Uα for some α. There exists Bi with x ∈ Bi ⊆ Uα as {Bi } is a basis. But then we
have x ∈ Bi ⊆ Ui . □

11. Separation axioms

We’ve already seen one separation axiom - see Hausdorff Definition 3.12.
Definition 11.1. Suppose that X is a topological space. Suppose that each singleton {x}
is closed for all x ∈ X. Then X is called regular if for each pair (x, B) where x ∈ X, B is
a closed set, and x ∈/ B, there exists disjoint open U, V such that x ∈ U and B ⊆ V .
Definition 11.2. Suppose that each singleton {x} is closed for all x ∈ X. The space X
is normal if for each pair A, B of closed disjoint sets of X, there exists disjoint open U, V
such that A ⊆ U and B ⊆ V .
Lemma 11.3. Suppose that each singleton {x} is closed for all x ∈ X.
a) X is regular if and only if for all x ∈ X and open U with x ∈ U , there exists open
V with x ∈ V ⊆ Cl(V ) ⊆ U
b) X is normal if and only if for all closed sets A and open U with A ⊆ U , there
exists an open set V containing A such that Cl(V ) ⊆ U .
24
Proof. Suppose X is regular. Suppose x ∈ U with U open. Then let F be the complement
of U . Then x ∈/ F and F closed, So there exists disjoint open V, W such that x ∈ V and
F ⊆ W . Let G be the complement of W . Then we have G ⊆ U . Furthermore since V is
disjoint from W , V ⊆ G. So V ⊆ Cl(V ) ⊆ G ⊆ U and x ∈ V .
Now, let x ∈ X and B is closed and x ∈/ B. Let U be B complement. We have x ∈ U
and U open. So there exists open V with x ∈ V ⊆ Cl(V ) ⊆ U . Let W be the complement
of Cl(V ). Then B ⊆ W and x ∈ V and V, W are disjoint open.
Suppose X is normal. Suppose A is closed and U is open with A ⊆ U . Let B be the
complement of U . Then since X is normal and A and B are disjoint there exists disjoint
open sets W, V with A ⊆ W and B ⊆ V . Let F be the complement of V . Then we have
A ⊆ W ⊆ F ⊆ U,
and since F is closed A ⊆ W ⊆ Cl(W ) ⊆ U .
Let A and B be disjoint closed sets. Let U be the complement of B so that A ⊆ U .
Then there exists open V with A ⊆ V ⊆ Cl(V ) ⊆ U . Let W be the complement of Cl(V )
and then notice that B ⊆ W and A ⊆ V and V and W are disjoint. □

Theorem 11.4.
a) A subspace of a Hausdorff space is Hausdorff
b) A product of Hausdorff spaces is Hausdorff
c) A subspace of a regular space is regular
d) A product of regular spaces is regular
e) A subspace of a normal space is not necessarily normal
f) A product of normal spaces is not necessarily normal
Proof. Omitted.
come back later
□

Theorem 11.5. Let X be a metric space. Then X is normal.

Proof. Note that one point sets in X are closed. Let x ∈ X and U = X − {x}. Let y ∈ U
and let ϵ = d(x, y)/2. Then B(y, ϵ) ⊆ U so that U is open.
Now, let A, B be disjoint closed sets. For each a ∈ A, there exists ϵa such that B(a, ϵa )
is disjoint from B (as B is closed). Similarly for each b ∈ B such that ϵb such that B(b, ϵb )
is disjoint from A. Let U = ∪a∈A B(a, ϵa /2) and V = ∪b∈B B(b, ϵb /2). Certainly A ⊆ U
and B ⊆ V and U and V are open. We also claim that they are disjoint.
Suppose z ∈ X such that z ∈ B(a, ϵa /2) and y ∈ X such that y ∈ B(b, ϵb /2). Then
d(z, b) > ϵ and d(y, a) > ϵ so
d(a, b) ≤ d(a, z) + d(z, y) + d(y, b) =< ϵ + d(z, y).
So if z = y then d(a, b) < ϵ so that B(a, ϵ) is not disjoint from B. So U and V are disjoint.

tidy up proof
□

Theorem 11.6. Let X be a compact Hausdorff space. Then X is normal.

25
Proof. We first show that X is regular. Let x ∈ X and B be a closed set with x ∈ / B.
For each b ∈ B find disjoint open Ub and Vb with x ∈ Vb and B ⊆ Ub . Since B is closed,
B is compact. So there exists b1 , . . . , bn such that B ⊆ Ub1 ∪ · · · ∪ Ubn and, and taking
V = Vb1 ∩ · · · ∩ Vbn , we have x ∈ V , and with U = Ub1 ∪ · · · ∪ Ubn U, V are disjoint with
x ∈ V and B ⊆ U .
Now, use the regularity of X and the same strategy to separate closed disjoint sets A
and B. □
Theorem 11.7. Let X be a second-countable regular topological space. Then X is normal.
Proof. Let A and B be disjoint closed sets. For each a ∈ A, there exists disjoint open sets
Ua , Va such that a ∈ Ua and B ⊆ Va . So Ua does not intersect B for each a ∈ A. The
collection {Ua } is an open cover of A, and X is second-countable, so we may cover A by a
countable collection {Un }. Similarly, we may cover B with a countable collection {Vm } of
open sets which are disjoint from A.
The problem is that we have ∪Un may not be disjoint from any particular Vm . Let
Wm = Um − (∪m m
n=1 Cl(Vn )) and Zm = Vm − (∪n=1 Cl(Un )). Then notice that Wm is
disjoint from V1 , . . . , Vm and since Zn ⊆ Vn , Wm is disjoint from Z1 , . . . , Zm .
Similarly, Zm is disjoint from W1 , . . . , Wm .
But then Zm and Wm′ for any m, m′ are disjoint as either m ≤ m′ or m′ ≤ m.
But by construction A ⊆ W = ∪Wm and B ⊆ Z = ∪Zm and Z and W are disjoint and
open. □
Theorem 11.8. Let X be a normal space. Let A, B be disjoint closed subsets of X. Let
[a, b] be a closed interval in R. Then there exists a continuous function f : X → [a, b] such
that f (x) = a for all x ∈ A and f (x) = b for all x ∈ B.
Proof. We may assume a = 0, b = 1. Let P = Q ∩ [0, 1] and write
P = {x1 , x2 , . . .}
where x1 = 0, x2 = 1.
Our goal is to construct open sets Up (one for each p ∈ P ), such that
Cl(Up ) ⊆ Uq
if and only if p < q, with the restriction that A ⊆ U0 and U1 = X.
Suppose we have done so and let us construct the required function f : X → [0, 1].
For each x ∈ X, define
f (x) = sup{p ∈ P | x ∈ Up }.
We claim that f is continuous. Let p, q ∈ P with p < q. Then f −1 (p, q) = {x | p < f (x) <
q} = Uq − Cl(Up ) which is open.
Now, (p, q) with p < q, p, q ∈ P gives a basis for the open sets in [0, 1]. So f is continuous.
Now, let us construct the collection {Up }.
Since X is normal, there exists disjoint open U0 and V0 such that A ⊆ U0 and B ⊆ V0 .
Define U1 = X.
Now, suppose we have defined Up for p ∈ {x1 , x2 , . . . , xn } with the property that
Cl(Us ) ⊆ Ur for all s < r with s, r ∈ {x1 , . . . , xn }.
Then look at p = xn+1 .
As any subset of P is linearly ordered and {x1 , . . . , xn+1 } is finite, there are r, s ∈
{x1 , . . . , xn } such that r < xn+1 < s (and no other element of {x1 , . . . , xn+1 } is between r
and s.
26
 Since X is normal there exists open Up and Vp with Cl(Ur ) ⊆ Up and Vp contains the
complement of Us . In other words Cl(Ur ) ⊆ Up and Cl(Up ) ⊆ Us .
So by induction we have constructed the collection {Up }.
tidy up proof
□
Definition 11.9. If A, B ⊆ X and there exists continuous f : X → [0, 1] such that
f (x) = 0 for all x ∈ A and f (x) = 1 for all x ∈ B, then we say that A and B can be
separated by a continuous function.
Theorem 11.10. Let X be a normal topological space. Let A be a closed subset of X.
a) Let f : A → [a, b] be a continuous function. Then there exists g : X → [a, b] which
is continuous and g(x) = f (x) for all x ∈ A
b) Let f : A → R be a continuous function. Then there exists g : X → R which is
continuous and g(x) = f (x) for all x ∈ A.
Proof. We claim that if f : A → [−r, r] then there exists g : X → [−r, r] such that
a) |g(x)| ≤ 3r for all x ∈ X
b) |g(x) − f (x)| ≤ 2r3 for all x ∈ A.
Let A = f −1 ([−r, −r/3] and B = f −1 ([r/3, r]. There exists g : X → [−r/3, r/3] such that
g(x) = −r/3 for x ∈ A and g(x) = r/3 for x ∈ B by Theorem 11.8. Notice that g satisfies
our claim.
Now, suppose f : A → [−1, 1] (the general case follows from this). Find g1 : X →
[−1/3, 1/3] with
|g1 (x) − f (x)| ≤ 2/3 for all x ∈ A,
by our claim above.
Now, find g2 : X → [−2/9, 2/9] such that
|f (x) − g1 (x) − g2 (x)| ≤ (2/3)2
by applying the claim with r = 1/3 to f (x) − g1 (x).
After repeating, we get a function
gn (x) : X → [−(1/2)(2/3)n , (1/2)(2/3)n ]
such that
|f (x) − g1 (x) − g2 (x) − · · · − gn (x)| ≤ (2/3)n
for all x ∈ A. Let
∞
X
g(x) = gn (x)
n=1
for all x ∈ X. Notice that since |gn (x)| ≤ (1/2)(2/3)n , the infinite sum is well-defined. Not
only that but g : X → [−1, 1].
Check that g(a) = f (a) for all a ∈ A:
∞
X
|g(a) − f (a)| ≤ |gi (a)| + |f (a) − g1 (a) − · · · − gN (a)|
i=N

and both terms tend to zero as N → ∞.

Now, let sN (x) = g1 (x) + · · · + gN (x). We must check that sN → g uniformly.
27
 Notice that
∞
X (1/2)(2/3)N
|sN (x) − g(x)| ≤ (1/2)(2/3)i = = (2/3)N −1
1 − (2/3)
i=N

independently of x so sN → g uniformly.
So g is the uniform limit of continuous functions so g is continuous.
For the second part of the theorem, we may assume f : A → (−1, 1). We may extend f
so that g : X → [−1, 1] by the first part of the theorem.
Let D = g −1 ({1, −1}). This is a closed subset of X. Furthermore, D is disjoint from A
as f (A) ⊆ (−1, 1). So find h : X → [0, 1] such that h(D) = {0} and h(A) = {1}.
Let H : X → (−1, 1) be defined by H(x) = h(x)g(x). Notice that H is continuous and
H : X → (−1, 1) and H extends f .
tidy up proof
□

Theorem 11.11. Let X be a regular second-countable topological space. Then X is metriz-

able (there exists d : X × X → R which makes X into a metric space).
Proof. Let B be a countable basis for X. For each U ∈ B, we claim that there exists a
continuous function f : X → [0, 1] whose support is exactly U . That is,
U = {x ∈ X | f (x) > 0}.
To prove this is an exercise in [Mun00]:
a) Prove that if U is open in X then U is a countable union of closed sets.
b) Then prove that there exists f : X → [0, 1] such that U = {x | f (x) > 0}.
For the first part: let A be closed and for each x ∈ X − A find disjoint open Ux , Vx such
that x ∈ Vx and A ⊆ Ux . Then A = ∩Ux .
For the second part, given U = ∪Ai where {Ai } are closed sets contained in U . Mimic
the proof of Theorem 11.8,
Now write B = {Bn } and for each n find fn : X → [0, 1] such that
Bn = {x | fn (x) > 0}.
ω
Define F : X → R by the rule
F (x) = (f1 (x), f2 (x), . . .).
Then F is continuous, as each fi is continuous (see Theorem 5.6). Also, if x ̸= y then
there exists disjoint open U, V with x ∈ U and y ∈ V and so there exists n for which
fn (x) > 0 and fn (y) = 0.
We also have to prove that X is homeomorphic to its image F (X). That means if U is an
open set in X we must prove that F (U ) is open in F (X), in other words, F (U ) = F (X)∩V
where V is open in Rω .
Let z ∈ F (U ) and so z = f (x) for some x ∈ U . Then there exists N so that x ∈ BN ⊆ U ,
−1
and so fN (x) > 0 while fN (0) contains the complement of U .
Let V = {(xi ) | xN > 0}. This is an element of the basis of the product topology for
Rω .
But V ∩ F (X) = {z = f (x) | fN (x) > 0} ⊆ F (U ) so that F (U ) is open. □
28
 12. The Tychonoff Theorem
Theorem 12.1. Suppose that {Xα } is a collection of compact topological spaces. Then
Q
Xα is compact.
Q
Proof. Let X = Q α∈I Xα . We use Theorem 9.12 and Zorn’s lemma. Let F be a collection
of closed sets of Xα satisfying Q the finite intersection property. Let A be the collection of
G of collections of closed sets of Xα such that
a) G contains F, and
b) G satisfies the finite intersection property.
To apply Zorn’s lemma we need to check that if C ⊆ A is a chain, then C has an upper
bound in A.
Given C define C to be [
C∗ = C,
C∈C
in other words G ∈ C ∗ if and only if G ∈ C for some CQ∈ C.
It is clear that C ∗ is a collection of closed sets of Xα containing F, but we need to
show that C ∗ satisfies the finite intersection property.
So let F1 , . . . , Fn ∈ C ∗ . Then there exists C ∈ C such that F1 , F2 , . . . , Fn ∈ C (this is
because C is a chain).
Since C satisfies the finite intersection property, F1 ∩ · · · ∩ Fn ̸= ∅, as required.
Now, let G be a maximal element of A.
We claim that \
G ̸= ∅,
G∈G
and since this intersection is contained in ∩F ∈F F , we will be done.
For each G ∈ G and projection πα : X → Xα let Gα = πα (G). Let α be fixed, then
the collection {Gα } where G runs over all elements of G satisfies the FIP so there exists
xα ∈ ∩Gα .
We now claim thatQx = (xα ) ∈ G for all G ∈ G. This is not obvious, as we cannot
necessarily write G = Gα .
For a closed set Fα ⊆ Xα let Fα′ = πα−1 (Fα ).
First, suppose that G ⊆ Fα′ . We claim that Fα′ ∈ G. This is because G ∪ {Fα′ } satisfies
the FIP.
But G is maximal w.r.t. being a collection of closed sets satisfying the FIP. So Fα′ ∈ G.
Now, suppose Fα′ 1 , Fα′ 2 are two such sets such that G ⊆ Fα′ 1 ∪ Fα′ 2 . We claim that
Fα1 ∪ Fα′ 2 ∈ G. Again, we see that G ∪ {Fα′ 1 ∪ Fα′ 2 } is a collection of closed sets satisfying
′

the FIP, so by maximality of G, we have

Fα′ 1 ∪ Fα′ 2 ∈ G.
We may extend this to finite unions by induction.
Now, let J ⊆ I be a finite subset of indices, and suppose that
[
G⊆ Fj′ .
j∈J

Then it follows that j∈J Fj′ ∈ G, and in particular, (xα ) ∈ j∈J Fj′ .
S S

Now, G is an intersection of sets of the form j∈J Fj′ , and so (xα ) ∈ G, as required. □
S
29
Remark 12.2. For the above proof to work, we need to know that every closed set of the
product X is an arbitrary intersection of finite unions of sets of the form
Fj′ = πj−1 (Fj ).
To establish this, it is enough to see that the complement of Fj′ is just
Uj′ = πJ−1 (Uj )
where Uj is the complement of Fj in Xj .
Then the description we need, is obtained because every open set is an arbitrary union
of a finite intersection of open sets of the form Uj′ , and so every closed set is an arbitrary
intersection of finite union of closed sets of the form Fj′ .

13. Paracompactness
Tidy up all proofs in this section
Definition 13.1. Let X be a topological space. Let A be a collection of subsets of X.
Then A is called locally finite if for every x ∈ X, there exists open U such that x ∈ U and
U ∩ A ̸= ∅ for only finitely many A ∈ A.
Lemma 13.2. Let A be a locally finite collection of subsets of X. Then:
a) Any subcollection A′ of A is locally finite
b) The collection B = {Cl(A) | A ∈ A} is locally finite
c) !
[ [
Cl = Cl(A).
A∈A A∈A

Proof. Parts 1 and 2 are an exercise.

come back and prove later
S

Let x ∈ Cl A∈A . We must show that x ∈ Cl(A) for all A ∈ A. There exists open
U that only intersects finitely many A ∈ A, say A1 , . . . , An , with x ∈ U . We claim that
x ∈ Cl(Ai ) for some i. If not there exists open Vi with x ∈ Vi such that Ai ∩ Vi = ∅ for
i = 1, 2, . . . , n. Consider U ′ = U ∩ V1 ∩ · · · ∩ Vn . Then x ∈ U ′ but U ′ ∩ ∪A = ∅ which
implies that x ∈ / Cl(∪A), a contradiction. □
Definition 13.3. A collection of subsets of X B is called countably locally finite if there
exists locally finite collections B1 , B2 , . . . , Bn , . . . such that
[∞
B= Bi .
i=1

Definition 13.4. Let A be a collection of subsets of X. A collection B is called a refinement

of A if for each B ∈ B, there exists A ∈ A with B ⊆ A.
If the elements of B are open we call B an open refinement, similarly, B is a closed
refinement if the elements of B are closed.
Lemma 13.5. Let X be a metric space. If A is an open cover of X then there exists a
collection D of subsets of X such that
a) D is an open covering of X
b) D is a refinement of A
30
 c) D is countably locally finite.
Proof. Let d be a metric on X. There exists a well-ordering of A (assuming the axiom of
choice). For each U ∈ A and integer n ≥ 1, define
Sn (U ) = {x ∈ X | B(x, 1/n) ⊆ U }.
Now we define sets Sn (U )′ (using a similar idea as in Theorem 11.7).
!
[
Sn (U )′ = Sn (U ) − V .
V <U

First, let us establish that if V < U and x ∈ Sn (V )′ and y ∈ Sn (U )′ then d(x, y) ≥ 1/n.
Suppose d(x, y) < 1/n. Then y ∈ B(x, 1/n) ⊆ V , a contradiction to the definition of
Sn (U )′
Then define [
En (U ) = B(x, 1/(3n)).
x∈Sn (U )′
Let x ∈ En (U ) and y ∈ En (V ) We claim that d(x, y) ≥ 1/3n. Suppose d(x, y) < 1/(3n).
There exists x′ ∈ Sn (U )′ with d(x, x′ ) < 1/(3n) and similarly there is y ′ ∈ Sn (V )′ with
d(y, y ′ ) < 1/(3n). Now, x′ ∈ Sn (U ) so B(x′ , 1/n) ⊆ U . But
d(x′ , y ′ ) ≤ d(x′ , x) + d(x, y) + d(y, y ′ ) < 1/n,
which is a contradiction to the fact that d(x′ , y ′ ) ≥ 1/n for x′ ∈ Sn (U )′ and y ′ ∈ Sn (V )′
as established above.
So En (U ) and En (V ) are disjoint if U < V . Furthermore, En (U ) ⊆ U for each U ∈ A,
so that {En (U )} is an open refinement of A.
We now claim that {En (U )} is locally finite (here n is fixed and U ranges over U ∈ A).
Let x ∈ X and suppose that d(x, y) < 1/(6n). Then y ∈ En (U ) for at most one U ∈ A.
Otherwise, there exists z, w with z ∈ En (Z) and w ∈ En (W ) such that d(z, w) ≤ d(z, y) +
d(y, w) ≤ 1/3n, a contradiction.
So B(x, 1/(6n)) intersects only one element of {En (U )} so {En (U )} is locally finite.
Now, we finally claim that [ [
(En (U )),
n≥1 U ∈A
is a countably locally finite, refinement of A which covers X.
The only thing left to do is to show that this covers X. Let x ∈ X. Find U ∈ A such
that x ∈ U and x ∈ / V for any V < U . Then there exists n ≥ 1 such that B(x, 1/n) ⊆ U
and so x ∈ Sn (U )′ ⊆ En (U ). □
Theorem 13.6. Let X be a metric space. Then X has a basis that is countably locally
finite.
Proof. Fix m ≥ 1. Let Am = {B(x, 1/m) | x ∈ X}. Then Am has a countably locally
finite refinement Dm which covers X. Now let
D = ∪m≥1 Dm ,
which is a countable union of a countable union of locally finite collections (so it is also
countably locally finite).
We now claim that D is a basis for X. Let U be open in X, and let x ∈ U . Then
B(x, ϵ) ⊆ U for some ϵ > 0. Now, choose n so that 1/n < ϵ/2. There exists B ∈ Dn such
31
that x ∈ B. Since Dn is a refinement of An , we have B ⊆ B(y, 1/n). Now, let z ∈ B.
Then
d(x, z) ≤ d(x, y) + d(y, z) ≤ 1/n + 1/n < ϵ
by the triangle inequality and since B ⊆ B(y, 1/n).
In particular, B ⊆ B(x, ϵ). This proves that D is a basis for X by Lemma 1.10. □
Definition 13.7. Let X be a Hausdorff topological space. Then X is called paracompact
if every open covering A of X has a locally finite open refinement B that covers X.
Theorem 13.8. Let X be paracompact. Then X is normal.
Proof. That X is Hausdorff is included in Definition 13.7.
Let a ∈ X and B ⊆ X be a closed set with a ∈ / B.
Our goal is to show that there exists disjoint open U and V with a ∈ U and B ⊆ V
(this will show that X is regular). Since X is Hausdorff, for each b ∈ B find disjoint open
Ub , Vb with a ∈ Ub and b ∈ Vb
Consider the open cover {Ub }b∈B ∪ {X − B}.
Since X is paracompact, there exists an open refinement A which is locally finite. Let
B = {A ∈ A | A ∩ B ̸= ∅}. Then B is a locally finite as well.
Now, let [
V = U.
U ∈B
Now, we claim that a ∈ / Cl(V ).
Since B is locally finite we have (by Lemma 13.2)
[
Cl(V ) = Cl(U ).
U ∈B

Let U ∈ B. We claim that a ∈ / Cl(U ). But since U ∩ B ̸= ∅ we must have U ⊆ Ub for

some b ∈ B (as B is a refinment of A). And Cl(Ub ) does not contain a, so too for Cl(U ).
therefore, a ∈
/ ∪U ∈B Cl(U ) = Cl(V ) as required.
Now to show that X is normal. Start with disjoint closed sets A and B. Using the
regularity of X instead of Hausdorff-ness, and replacing all occurences of a with A, repeat
the above argument. □
Theorem 13.9.
a) Every closed subspace of a paracompact space is paracompact.
b) A subspace of a paracompact space is not necessarily paracompact
c) A product of paracompact spaces is not necessarily paracompact.
Proof. Let X be paracompact. Let Y be a closed subset of X. Let A be an open cover
of Y by open sets in Y . We can write A = A′ ∩ Y where A′ is open in X. Let A′ =
{A′ | A′ ∩ Y ∈ A} ∪ {X − Y }. And note that A′ is an open cover of X. There exists a
locally finite refinement B of A′ . Take C = {U ∩ Y | U ∈ B and U ∩ Y ̸= ∅}. Notice that
C is locally finite since B, it is certainly a refinement of A, and it covers Y . □
Theorem 13.10. Let X be a metric space. Then X is paracompact.
Proof. We already know that if A is an open cover of X, then it has a countably locally
finite refinement (Lemma 13.5).
We also know that any metric space is normal Theorem 11.5, and so regular. The
following lemma will complete the proof. □
32
Lemma 13.11. Let X be regular. Then the following are equivalent: Every open covering
of X has a refinement that is
a) an open covering of X and countably locally finite
b) a covering of X and locally finite
c) a closed covering of X and locally finite
d) an open covering of X and locally finite.
Proof. Let A be an open covering of X.
Let B be a open refinement of A that is countably locally finite. Then
B = ∪∞
n=1 Bn .

Let
Vn = ∪U ∈Bn U.
Define
Cn = {B − (V1 ∪ · · · ∪ Vn ) | B ∈ Bn }.
Note that C = ∪∞ C
n=1 n is a refinement of A that is locally finite.
Suppose A is an open covering of X. Start by constructing
B = {U open∥ Cl(U ) ⊆ A for some A ∈ A}.
This is an open cover of X: let x ∈ X, and A ∈ A, then there exists disjoint open U, V
with x ∈ U and Ac ⊆ V . In other words, U ⊆ Cl(U ) ⊆ V c ⊆ A.
Now, let C be a refinement of B that covers X and is locally finite. Then let D =
{Cl(C) | C ∈ C}. Then D is a refinement of A and is locally finite.
To find an open covering that is locally finite given that every open cover has a closed
refinement that is locally finite is the trickiest part.
First let A be an open covering of X. Let B be a refinement of A that is locally finite.
The elements of B may not be open sets. To make them open we need to make them bigger
(they may have no interior, for example). Even if they all had non-empty interior, if we
made them smaller, then they may no longer cover X.
But the problem with enlarging the sets is that if we make them too much larger, the
collection may no longer be locally finite, or it may no longer be a refinement of A.
Since B is locally finite, for each x ∈ X, there exists an open Ux with x ∈ Ux and Ux
only intersects finitely many B ∈ B. So {Ux }x∈X is an open cover of X, there exists a
closed refinement of {Ux } that is locally finite, call it C.
Note that for each C ∈ C, C only intersects finitely many elements of B. For B ∈ B,
define
C(B) = {C ∈ C | C ∩ B = ∅}.
Then define
E(B) = {x ∈ X | x ∈ / C for all C ∈ C(B)}.
Notice, we have
B ⊆ E(B).
Also E(B) is open since it is the complement of all C ∈ C(B) which is closed by
Lemma 13.2.
So {E(B)} is an open cover of X. We can prove that it is locally finite, but it may be
too large to be a refinement of A. For each B ∈ B choose F (B) ∈ A such that B ⊆ F (B).
Now, define D = {E(B) ∩ F (B) | B ∈ B}.
It is certainly a refinement of A by construction. Is it locally finite?
33
 Given x ∈ X, find an open U such that U intersects only C1 , C2 , . . . , Cn ∈ C. Then
U ⊆ C1 ∪ C2 ∪ · · · ∪ Cn (if not, C cannot cover X).
We claim that each Ci can only intersect finitely many E(B) ∩ F (B). Then it would
follow that D = {E(B) ∩ F (B)} would be locally finite.
Suppose Ci intersects E(B) ∩ F (B). Then by definition of E(B), Ci must intersect B
(or else, it would be excluded by definition of E(B)).
Now, Ci only intersects finitely many B ∈ B (this is how we constructed C, using
regularity of X). So Ci only intersects finitely many E(B) ∩ F (B). So U only intersects
at most finitely many E(B) ∩ F (B). SO {E(B) ∩ F (B)} is a locally finite open refinement
of A. □

14. Complete Metric Spaces

Definition 14.1. Let X be a metric space. A sequence (xn ) of points of X is said to be
Cauchy if for each ϵ > 0 there exists N > 0 such that if n, m ≥ N then d(xn , xm ) < ϵ.
If every Cauchy sequence in X converges, then we say that X is complete.
Lemma 14.2. Let X be a metric space. Then X is complete if and only if every Cauchy
sequence has a convergent subsequence.
Proof. Certainly if X is complete then every Cauchy sequence has a convergent subse-
quence.
Now, suppose that every Cauchy sequence has a convergent subsequence.
Let (xn ) be a Cauchy sequence.
Let (xnj ) be a convergent subsequence xnj → x0 as j → ∞ (remember, 1 ≤ n1 < n2 <
· · · < nj < · · ·).
Let ϵ be given.
Find N large enough so that if n, m ≥ N then d(xn , xm ) < ϵ/2. Then find j large
enough so that
a) nm ≥ N for all m ≥ j
b) d(xnm , x0 ) < ϵ/2 for all m ≥ j
Let M = max(N, nj ) and let m ≥ M . Then
d(xm , x0 ) ≤ d(xm , xnj ) + d(xnj , x0 ) < ϵ/2 + ϵ/2 = ϵ
which proves that xm → x0 as m → ∞. □
Theorem 14.3. The metric space Rn , with either the Euclidean metric or square metric,
is complete.
Proof. Let (xn ) be a Cauchy sequence, let ρ be the square metric. Note that
ρ(xn , 0) ≤ ρ(xn , xN ) + ρ(xN , 0)
and so choose N so that if n, m ≥ N then d(xn , xm ) < 1 and notice that
ρ(xn , 0) ≤ ρ(xN , 0) + 1
for all n ≥ N . In other words, if M = maxm≤N {ρ(xm , ))} then xj ∈ [−M, M ]n for all j.
But [−M, M ]n is a closed and bounded subset of Rn so it is compact by Theorem 9.15.
Now since a set is compact if and only if it is sequentially compact (for metric spaces, see
Theorem 9.24), (xn ) has a convergent subsequence. Now every Cauchy sequence has a
convergent subsequence, so by Lemma 14.2, Rn is complete.
34
 Now, xn converges to x0 in the square metric if and only if it converges in the Euclidean
metric. Similarly, xn is a Cauchy sequence in the square metric if and only if it is Cauchy
in the Euclidean metric. So Rn is complete also in the Euclidean metric. □
Definition 14.4. Let Y be a metric space. Let J be an index set. Then define
d(x, y) = min{d(x, y), 1},
for all x, y ∈ Y .
Note that d and d give the same topology on Y .
Now for f, g ∈ Y J = {f : J → Y } define
ρ(f, g) = sup{d(f (j), g(j)) | j ∈ J}.
We call ρ the uniform metric, and d the standard bounded metric. WARNING: the
topology given by ρ does not give the product topology on Y J .
Theorem 14.5. If Y is a complete metric space then Y J is complete with respect to the
uniform metric ρ.
Proof. Suppose (xn ) is a Cauchy sequence of elements in Y J (with respect to the uniform
metric).
Write
xn = (xj,n )j∈J .
Now, find N such that if n, m ≥ N then
ρ(xn , xm ) < 1.
Then it follows that
d(xn,j , xm,j ) = d(xn,j , xm,j ) < ρ(xn , xm )
for all j ∈ J, and so xn,j is Cauchy for all j ∈ J. Since Y is complete, let xn,j → x0,j .
Let ϵ > 0 and assume that ϵ < 1. Now, let N be such that n, m ≥ N implies that
ρ(xn , xm ) < ϵ/2.
So for all j ∈ J, we have
d(xn,j , xm,j ) < ϵ/2.
Letting m → ∞ (as n and j are fixed) we have
d(xn,j , x0,j ) ≤ ϵ/2.
So
ρ(xn , x0 ) ≤ ϵ/2 < ϵ,
as required. □
Theorem 14.6. Let X be a topological space and Y be a metric space. Let C(X, Y ) be
the set of all continuous functions from X to Y . Then C(X, Y ) is closed in Y X under the
uniform metric. Therefore, if Y is complete then C(X, Y ) is complete with respect to the
uniform metric.
Proof. Suppose {fn } is a sequence of continuous functions converging (under ρ) to a func-
tion f : X → Y .
First, we show that fn → f uniformly (Definition 6.17).
Let ϵ > 0. Then there exists N such that if n ≥ N then
ρ(fn , f ) < ϵ.
35
Now, assuming ϵ < 1, we have that ρ(fn , f ) = sup d(fn (x), f (x)). In other words,
d(fn (x), f (x)) ≤ ρ(fn , f ) < ϵ
for all x ∈ X so that fn → f uniformly. Now, apply Theorem 9.23 to see that f : X → Y
is continuous.
Now, if Y is complete then so is Y X under ρ, the reason being that if a sequence
in C(X, Y ) is Cauchy, it converges to f ∈ Y X , and so converges uniformly to f and so
f ∈ C(X, Y ) (in other words, closed subsets of complete spaces are complete). □

Definition 14.7. Let Y be a metric space. Let F ⊂ Y X where X is a set. Assume that
F is bounded: for every f, g ∈ F the set {d(f (x), g(x)) | x ∈ X} is bounded.
Then define the sup-metric, denoted ρ, by the rule
ρ(f, g) = sup{d(f (x), g(x)) | x ∈ X}
for all f, g ∈ F.
Lemma 14.8. Let X be a topological space. Let B(X, R) be the set of all bounded functions
f : X → R. Then B(X, R) is complete under the sup-metric.
Proof. Let (fn ) be a Cauchy sequence in B(X, R). There exists N such that ρ(fn , fm ) < 1
for n, m ≥ N . In this case, ρ(fn , fm ) = ρ(fn , fm ), and so (fn ) is also Cauchy with respect
to ρ. So fn → f with f : X → R (convergence with respect to ρ. We claim that f is
bounded. It will then follow that fn → f in the sup metric. Now,
|f (x)| ≤ |f (x) − fn (x)| + |fn (x)|
and choose n so that ρ(f, fn ) < 1 and then since fn ∈ B(X, R) also f ∈ B(X, R). □
Theorem 14.9. Let X be a metric space. Then there is a isometric imbedding of X into
a complete metric space.
Proof. Fix x0 ∈ X. For a ∈ X define ϕa : X → R by the rule
ϕa (x) = d(x, a) − d(x, x0 ).
We claim that ϕa ∈ B(X, R.
d(x, a) ≤ d(x, x0 ) + d(x0 , a)
in other words
ϕa (x) ≤ d(x0 , a)
and similarly,
d(x, x0 ) ≤ d(x, a) + d(a, x0 )
so that
−d(a, x0 ) ≤ ϕa (x).
So |ϕa (x)| ≤ d(a, x0 ) for all x ∈ X. So ϕa ∈ B(X, R).
Define Φ : X → B(X, R) by the rule Φ(a) = ϕa . We now prove that d(a, b) = ρ(ϕa , ϕb ).
ϕa (x) − ϕb (x) = d(x, a) − d(x, b) ≤ d(a, b),
and
ϕb (x) − ϕa (x) = d(x, b) − d(x, a) ≤ d(a, b)
so ρ(ϕa , ϕb ) ≤ d(a, b). But ϕa (a) − ϕb (a) = d(a, b) so ρ(ϕa , ϕb ) = d(a, b). □
36
Definition 14.10. Let X be a metric space and assume h : X → Y is an isometric imbed-
ding of X into Y . Then Cl(h(X)) is a complete metric space, it is called the completion
of X.

15. The Fundamental Group

Definition 15.1. Let X and Y be topological spaces. Let f, f ′ : X → Y be continuous
functions. We say that f is homotopic to f ′ if there exists a continuous function
F : X × I → Y,
such that
F (x, 0) = f (x) for all x ∈ X
and
F (x, 1) = f ′ (x) for all x ∈ X.
We say that F is a homotopy from f to f ′ and we write f ∼ = f ′.
Definition 15.2. Let f, f ′ be two paths in the topological space X from a to b. So
f, f ′ : [0, 1] → X are continuous functions such that f (0) = a = f ′ (0) and f (1) = b = f ′ (1).
Then f and f ′ are said to be path homotopic if there exists a continuous function
F : [0, 1] × [0, 1] → X
such that
F (x, 0) = f (x) and F (x, 1) = f ′ (x)
F (0, y) = a and F (1, y) = b.
for all x, y ∈ [0, 1].
We call F a path homotopy from f to f ′ . If f is path homotopic to f ′ , we write f ∼
=p f ′ .

Lemma 15.3. Path homotopy and homotopy are equivalence relations (on the set of paths
in X from a to b and on the set of continuous functions from X to Y , respectively).

Proof. Suppose f : X → Y is a continuous function. Then F : X × [0, 1] → Y defined by

F (x, t) = f (x) is also continuous as the inverse image of U ⊆ Y is f −1 (U ) × [0, 1], i.e.,
open in the product topology. So f ∼ = f (homotopy is reflexive).
Now, suppose f ∼ = f ′ . Then there exists F : X × [0, 1] → Y and just define

F (x, s) = F (x, 1 − s)

and notice that F is continuous and satisfies the definition for f ′ ∼

= f . So homotopy is
symmetric.
Now, suppose f ∼ = g and g ∼ = h, with corresponding continuous functions F1 , F2 :
X × [0, 1] → Y .
Then define G : X × [0, 1] → Y by

G(x, 2s) if s ≤ 1/2
G(x, t) =
G(x, 2s − 1) if s ≥ 1/2

and note that G is continuous by Theorem 4.7.

Note also that f ∼=p f , and if F is a path homotopy then so is F , and if F1 , F2 are
path-homotopies, then so is G. □
37
Definition 15.4. Let f be a path in X from a to b, and g be a path from b to c in X.
Define a new path h : [0, 1] → X from a to c by the rule

f (2s) for s ∈ [0, 1/2]
h(s) =
g(2s − 1) for x ∈ [1/2, 1]
Then h is called the composition of f and g, denoted f ∗ g.
Theorem 15.5. Suppose that f ∼ =p f ′ and g ∼
=p g ′ , where f, f ′ are paths from a to b and
′
g, g are paths from b to c. Then
f ∗g ∼=p f ′ ∗ g ′ .
Let [f ] denote the path homotopy class of f , and define [f ] ∗ [g] = [f ∗ g].
a) If [f ] ∗ ([g] ∗ [h]) is defined, then so is ([f ] ∗ [g]) ∗ [h] and they are equal
b) Suppose f is a path from a to b. Let ea denote the constant path ea (s) = a for all
s ∈ [0, 1], and similarly for eb . Then [ea ] ∗ [f ] = [f ] = [f ] ∗ [eb ]
c) If f is a path from a to b, let f be the path defined by
f (s) = f (1 − s)
so that f is a path from b to a. Then [f ] ∗ [f ] = [eb ] and [f ] ∗ [f ] = [ea ].
Proof. First, check that if f ∼ =p g ′ then f ∗ g ∼
=p f ′ and g ∼ =p f ′ ∗ g ′ .
So let F1 : [0, 1] → X be a path-homotopy from f to f ′ and F2 a path-homotopy from
2

g to g ′ . Define H : [0, 1]2 → X by the rule


F1 (2s, t) if s ≤ 1/2
H(s, t) =
F2 (2s − 1, t) if s ≥ 1/2
and notice that H is continuous by Theorem 4.7, and is a path-homotopy from f ∗ g to
f ′ ∗ g′ .
So we may define [f ] ◦ [g] = [f ∗ g] (in other words, multiplication of two path-homotopy
classes is well-defined).
Now, suppose f is a path from a to b and g is a path from b to c. Then let

f (2s) if s ≤ 1/2
h1 = f ∗ g =
g(2s − 1) if x ≥ 1/2
To prove that ∗ is associative, we prove that there is a path homotopy from h1 = f ∗ g to

f (3s) if s ≤ 1/3
h2 (s) =
g((3s − 2)/2) if s ≥ 1/3
Write
(1 − t)/2 + t/3 = 1 − t/6 = (6 − t)/6.
So define
F : [0, 1]2 → X
by the rule
(
f (6s/(3 −t)) if s ≤ t/3 + (1 − t)/2 = (3 − t)/6
F (s, t) =

6s+t−3
g 3+t if s ≥ t/3 + (1 − t)/2
Now, we can show that each half of the definition of F is continuous and they agree on the
intersection of their domain, so F is continuous. For example, let R = {(s, t) ∈ [0, 1]2 | s ≤
t/3 + (1 − t)/2} and let us show that
F1 (s, t) : R → X
38
is continuous:
F1 (s, t) = f ◦ T
where T : R → [0, 1] is defined by
6s
T (s, t) =
3−t
But T is the quotient of the two continuous functions (s, t) 7→ 6s and (s, t) → 7 3 − t and
3 − t ̸= 0 for (s, t) ∈ R, so T is continuous, and therefore F1 is continuous.
Now, this shows that f ∗ (g ∗ h) ∼ =p f ′ where

 f (3s) if s ≤ 1/3
f ′ (s) = g(3s − 1) if 1/3 ≤ s ≤ 2/3
h(3s − 2) if 2/3 ≤ s ≤ 1

′ ∼
And a similar argument shows that f =p (f ∗ g) ∗ h. So ∗ is associative.
Now we show that
f ∗ eb ∼
=p f.
Let 
f (2s/(2 − t)) if s ≤ (1 − t) + t/2 = 1 − t/2
F (s, t) =
b if s ≥ 1 − t/2
The function F is path homotopy from f ∗ eb to f . Similarly, f ∼ = ea ∗ f .
Let us now show that
f ∗f ∼=p e a .
Define 

 a if 0 ≤ s ≤ t/2
f (2s − t) if t/2 ≤ s ≤ 1/2

F (s, t) =

 f (1 + t − 2s) if 1/2 ≤ s ≤ 1 − t/2
a if 1 − t/2 ≤ s ≤ 1


If this function is continuous then f ∗ f ∼ =p ea . Replacing f with f and using the fact that
∼
f = f , we also obtain that f ∗ f =p eb . □
Definition 15.6. Let X be a topological space. Let x ∈ X. A loop based at x is a path
f : [0, 1] → X such that f (0) = f (1) = x. The set of path homotopy classes of loops based
at x with the operation ∗ is called the fundamental group of X relative to x. It is denoted
by π1 (X, x).
Definition 15.7. Let x, y ∈ X. Let α be a path from x to y. Define
α̂ : π1 (X, x) → π1 (X, y)
by the rule
α̂([f ]) = [α] ∗ [f ] ∗ [α].
Theorem 15.8. The map α̂ is a group isomorphism.
Proof. Check that
α̂([f ] ∗ [g]) = [α] ∗ [f ] ∗ [g] ∗ [α]
= [α] ∗ [f ] ∗ [α] ∗ [α] ∗ [g] ∗ [α]
= α̂([f ]) ∗ α̂([g])
using Theorem 15.5.
39
 Now, check that α̂−1 = α̂. □
Corollary 15.9. If X is path-connected and x, y ∈ X then π1 (X, x) ∼
= π1 (X, y).
Definition 15.10. Let X be a topological space. Then X is called simply connected if X
is path-connected and Π1 (X, x) is trivial for some x ∈ X.

Lemma 15.11. Let X be a simply connected space. Then if f and f ′ are two paths from
a to b then [f ] = [f ′ ], in other words f is path homotopic to f ′ .
Proof. Now, f ∗ f ′ ∼
=p ea since f ∗ f ′ is a loop based at a and π1 (X, a) is trivial.
Compute
(f ∗ f ′ ) ∗ f ′ ∼
=p e a ∗ f ′ ∼
=p f ′
∼p f ∗ (f ′ ∗ f ′ ) ∼
(f ∗ f ′ ) ∗ f ′ = =p f
In other words, f ∼
=p f ′ as required. □

Definition 15.12. Suppose that h : X → Y is a continuous function such that f (x) = y.

We may use the notation
h : (X, x) → (Y, y)
to denote this. Define
h∗ : π1 (X, x) → π1 (Y, y)
by the rule
h∗ ([f ] = [h ◦ f ].
Then h∗ is called the homomorphism induced by h. We can also call it the push forward
of h.
Proposition 15.13. Let
h : (X, x0 ) → (Y, y0 )
be a continuous function. Let
h∗ : π1 (X, x0 ) → π1 (Y, y0 )
be defined by
h∗ ([f ] = [h ◦ f ].
Then h∗ is well-defined and is a group homomorphism.
Proof. Suppose f, f ′ are paths from x0 to x such that f ∼
=p f ′ . We must show that
h◦f = ∼ h ◦ f ′.

Since h is continuous, these are both paths that go from y0 to f (x) = f ′ (x). So suppose
that F : [0, 1]2 → X is a path-homotopy from f to f ′ . Let us try G = h ◦ F : [0, 1]2 → Y .
Certainly, G = h ◦ F is continuous. Then check that G is a path-homotopy from h ◦ f to
h ◦ f ′.
Now, suppose f is a path from a to b and g is a path from b to c. We must check that
(h ◦ f ) ∗ (h ◦ g) ∼
=p h ◦ (f ∗ g).
Check that 
h ◦ f (2s) if s ≤ 1/2
(h ◦ f ) ∗ (h ◦ g)(s) =
h ◦ g(2s − 1) if s ≥ 1/2
40
while

f (2s) if s ≤ 1/2
h ◦ (f ∗ g) = h ◦
g(2s − 1) if s ≥ 1/2

h ◦ f (2s) if s ≤ 1/2
=
g(2s − 1) if s ≥ 1/2
= (h ◦ f ) ∗ (h ◦ g)(s)

So (h ◦ f ) ∗ (h ◦ g) = h ◦ (f ∗ g) and so certainly they are homotopic, so h∗ defines a group

homomorphism. □

Theorem 15.14. Let h : (X, x0 ) → (Y, y0 ) and k : (Y, y0 ) → (Z, z0 ). Then (k ◦ h)∗ =
k∗ ◦ h∗ . If i : (X, x0 ) → (X, x0 ) is the identity map then i∗ is the identity homomorphism.
Proof. Let f be a path from a to b. Note that
(k ◦ h) ◦ f = k ◦ (h ◦ f )
which tells us that (k ◦ h)∗ = k∗ ◦ h∗ .
Notice that i ◦ f = f so that i∗ is the identity. □

Corollary 15.15. Suppose that h : (X, x0 ) → (Y, y0 ) is a homeomorphism. Then h∗ :

π1 (X, x0 ) → π1 (Y, y0 ) is a group isomorphism.
Proof. Use Theorem 15.14. □

Definition 15.16. Let p : E → B be a continuous surjective function. The open set U of

B is said to be evenly covered by p if the inverse image p−1 (U ) can be written as the union
of disjoint open sets Vα such that for each α the restriction of p to Vα is a homeomorphism
Vα → U . The collection {Vα } will be called a partition of p−1 (U ) into slices.
Definition 15.17. Let p : E → B be continuous and surjective. If for every point b ∈ B,
there exists open U that is evenly covered by p, then p is called a covering map, and E is
said to be a covering space.

Theorem 15.18. Define p : R → S 1 by the rule p(x) = (cos(2πx), sin(2πx)) is a covering

map.
Proof. Note that p is continuous and surjective. Let U = S 1 − {(1, 0)} and V = S 1 −
{(−1, 0)}. Then you can prove that U and V are both evenly covered by p. □

Definition 15.19. Let p : E → B be a function. Let f : X → B be a continuous function.

A lifting of f is a function f˜ : X → E such that p ◦ f˜ = f . That is, we have the following
diagram
E
f˜
p
f
X B

Lemma 15.20. Let p : E → B be a covering map. Let p(e0 ) = b0 . Any path f : [0, 1] → B
beginning at b0 has a unique lifting to a path f˜ in E beginning at e0 .
41
Proof. Let {Ui } be a covering of B such that each Ui is evenly covered by p.
Suppose that f (s) ∈ Ui for all s ∈ [0, 1] and some i.
Then there exists V ⊆ p−1 (Ui ) such that
p|V : V → Ui
is a homeomorphism and such that e0 ∈ V .
Since p|V : V → Ui is a homeomorphism, there exists a lifting of f to V .
Furthermore, if f˜ is any lifting of f , then f˜(s) ∈ V . This is because if the image of f˜
were spread across, say, two disjoint opens, the image would be disconnected, but this is
impossible since f˜ is continuous.
But now, the lifting of f to V must be unique.
Now note that one can reduce to the case where the image of f is contained in a single
Ui by using Lemma 9.22. We find δ such that given any f −1 (Ui ) there exists y ∈ [0, 1]
such that [0, 1] ∩ (y − δ, y + δ) ⊆ f −1 (Ui ). So there exists s0 < s1 < . . . < sn with s0 = 0,
and sn = 1 such that each f ([si , si+1 ]) ⊆ Uj for some j. Then we may argue as in the case
that f ([0, 1]) ⊆ Ui . □
Lemma 15.21. Let p : E → B be a covering map. Let p(e0 ) = b0 . Let F : [0, 1]×[0, 1] → B
be continuous and F (0, 0) = b0 . Then there is a lifting of F to a continuous map
F̂ : [0, 1] × [0, 1] → E
such that F̂ (0, 0) = e0 . If F is a path homotopy then so is F̂ .
Proof. We proceed as in Lemma 15.20 and we assume that there exists V ⊆ E such that
p|V : V → U ⊆ B is such that p|V is a homeomorphism and F ([0, 1]2 ) ⊆ U . And we may
also assume that we have lifted F to F̃ on the set
{0} × [0, 1] ∪ [0, 1] × {0}.
2
Since [0, 1] is connected, it follows that F̃ (s, t) ∈ V (just as in Lemma 15.20). Since p|V is
a homeomorphism, so define F̃ (s, t) = p|−1 V (F (s, t)). This tells us that the lifting F̃ exists
and is uniquely defined. □
Theorem 15.22. Let p : E → B be a covering map and let p(e0 ) = b0 . Let f, g be two
paths in B from b0 to b1 . Let f˜ and g̃ be their liftings to paths in E beginning at e0 . If f
and g are path homotopic then f˜ and g̃ end at the same point of E and are path homotopic.
Proof. Let F : [0, 1]2 → B be a path-homotopy from f to g. By Lemma 15.21, there exists
F̃ : [0, 1]2 → E which is a lifting of F , and since F is a path-homotopy, so is F̃ . Now,
that means F̃ (0, s) must define a path from e0 to some e1 with p(e1 ) = b1 , and same with
F̃ (1, s). Since F̃ is a path-homotopy these two paths end at the same point. By uniqueness
of liftings of paths Lemma 15.20, F̃ must be a path homotopy from f˜ to g̃. □
Theorem 15.23. The fundamental group of S 1 is an infinite cyclic group.
Proof. Let p : R → S 1 be the covering map from Theorem 15.18. Let x0 = (1, 0) ∈ S 1 ,
say. For a loop f : [0, 1] → S 1 , starting and ending at x0 , let f˜ be the lift of f to R. Since
f (1) = (1, 0), we have that f˜(1) = n for some n ∈ Z. Define ϕ : π1 (X, x0 ) → Z by the rule
ϕ(f ) = n.
We claim that ϕ is well-defined. If F is a path homotopy from f to f ′ , say, then there
is a lifting F̃ from f˜ to f˜′ . So f˜ and f˜′ end at the same point, this proves that ϕ is
well-defined.
42
 Now, suppose that f and g are both loops at x0 . Let f˜ and g̃ be their lifts to R, and
suppose f˜ ends at n and g̃ ends at m with m, n ∈ Z. Then define ĝ by the rule
ĝ(s) = g̃(s) + n.
Then notice that f˜ ∗ ĝ is a lifting for f ˜∗ g. In other words, ϕ(f ∗ g) = m + n.
Now, suppose ϕ(f ) = 0. Then f˜ is a path in R beginning and ending at 0. Since R is
simply connected, f˜ is homotopic to the constant map. That implies that f is homotopic
to the constant map as well. □

Theorem 15.24. Let p : (E, e0 ) → (B, b0 ) be a covering map. If E is path connected then
there is a surjection
ϕ : π1 (B, b0 ) → p−1 (b0 ).
If E is simply connected, ϕ is a bijection.
Proof. Omitted. □

Definition 15.25. If E is simply connected and p : E → B is a covering map then we say

that E is a universal covering space of B.

Theorem 15.26. Let x0 ∈ S 1 . The inclusion map

j : (S 1 , x0 ) → (R2 − {0}, x0 )
induces an isomorphism of fundamental groups.
Proof. Define r : R2 − {0} to S 1 by the rule
x
r(x) = .
∥x∥
Notice that r ◦ j is the identity on S 1 , so that r∗ j∗ is the identity on π1 (S 1 , x0 ) by Theo-
rem 15.14.
Now, let f be a loop in R2 − {0} based at x0 . Let f ′ = j∗ r∗ (f ). We claim that f ′ ∼ =p f
in R2 − {0}.
Define F (s, t) = t ∥ff (s)∥
(s)
+ (1 − t)f (s). Note that ∥·∥ : R2 → R is continuous, so the only
way that F is not continuous is if f (s) = 0 for some s. But f (s) ̸= 0 since f is a path in
R2 − {0}. Also, F (s, t) ∈ R2 − {0} since
 
t
F (s, t) = f (s) +1−t
∥f (s)∥
which is a product of two non-zero numbers.
This shows that f ∼=p f ′ so that j∗ r∗ is the identity on π1 (R2 − {0}, x0 ), as required. □

Theorem 15.27. Let x0 ∈ S n−1 . The inclusion map

j : (S n−1 , x0 ) → (Rn − {0}, x0 )
induces an isomorphism of fundamental groups.
Proof. Use the same proof as Theorem 15.26. □
43
Definition 15.28. Let A be a subspace of X. Then A is said to be a strong deformation
retract of X if there is a continuous map H : X × [0, 1] → X such that
H(x, 0) = x for all x ∈ X,
H(x, 1) ∈ A for all x ∈ X,
H(a, t) = a for all a ∈ A and t ∈ [0, 1]

Theorem 15.29. Let A be a strong deformation retract of X. Let a0 ∈ A. Then the

inclusion map
j : (A, a0 ) → (X, a0 )
induces an isomorphism of fundamental groups.
Proof. Omitted. □

16. Applications of the Fundamental Group

Definition 16.1. Let h : X → Y where X and Y are topological spaces. The map h is
called inessential if h is homotopic to a constant function. Otherwise, h is called essential.

Lemma 16.2. Let h : S 1 → Y . Then the following are equivalent:

a) h is inessential.
b) h can be extended to a continuous map g : B 2 → Y .
Recall that B 2 is the closed unit ball in R2 . That is
B 2 = {x ∈ R2 | ∥x∥ ≤ 1}.
Proof. Let F : S 1 × [0, 1] → Y be a homotopy from h to a constant function. Define
π : S 1 × [0, 1] → B 2 by the rule
π(x, t) = (1 − t)x.
Then π(x, 1) carries S 1 × {1} to {0} and π when restricted to S 1 × [0, 1) is a bijection from
S 1 × [0, 1) with B 2 − {0}. Now, π is a closed map (π maps closed sets to closed sets). This
is because π is continuous, so it maps compact sets to compact sets (Theorem 9.7. Also,
any compact subset of a Hausdorff space is closed by Theorem 9.5. Now, any closed subset
of a compact set is compact by Theorem 9.4. Finally, B 2 is Hausdorff and S 1 × [0, 1] is
compact.
Now, H is constant on S 1 ×{1}, so there exists g : B 2 → Y such that g(π(x, t)) = H(x, t).
We can define g by putting g(0) = H(x, 1) = y0 , and then g(v) = g(π(x, t)) = H(x, t) where
we choose x, t such that π(x, t) = v. The function g is continuous since π is a closed map
and H is continuous.
Now, suppose g : B 2 → Y is such that g(x) = h(x) for all x ∈ S 1 . Then define
H : S 1 × [0, 1] → Y by H(x, t) = g((1 − t)x). Then H is a homotopy of h with a constant
function. □

Theorem 16.3. Let h : X → Y be a continuous function. If h is inessential, then h∗ is

the zero homomorphism.
Proof. First, suppose X = S 1 . Suppose h : S 1 → Y is inessential. Let g : B 2 → Y be the
extension of h from Lemma 16.2. Let j : S 1 → B 2 be the inclusion map. Then we have
44
the diagram
h
S1 Y
j g

B2
Then by Theorem 15.14, we have
h∗
π1 (S 1 , x0 ) π1 (Y, y0 )
j∗ g∗

π1 (B 2 , x0 )

and since π1 (B 2 , x0 ) trivial, g∗ is zero, and so h∗ is zero.

Now, suppose that h : X → Y is inessential and let f : I → X be a loop in X at x0 . We
want to show that h ◦ f is path homotopic to a constant so that h∗ ([f ]) = 0. Then there
exists continuous k : S 1 → X defined by
k((cos(2πx), sin(2πx)) = f (x).
1
If we define ϕ : I → S such that ϕ(x) = (cos(2πx), sin(2πx)), then we have the following
diagram
f h
I X Y
ϕ k

S1
We now claim that h ◦ k is homotopic to a constant. Let H : X × I → Y and define
H1 : S 1 × I → Y
by
H1 (x, t) = H(k(x), t)
and H1 is a homotopy of h ◦ k to a constant. So by the first part of the proof, (h ◦ k)∗ = 0.
So
0 = (h ◦ k)∗ ([ϕ]) = [h ◦ k ◦ ϕ] = [h ◦ f ] = h∗ ([f ])
as required. □

Corollary 16.4. Let T be a closed triangular region in R2 . Let Bd(T ) be the boundary of
T (the union of the edges of T ). There is no continuous map f : T → Bd(T ) that maps
each edge of T into itself.
Proof. Suppose f : T → Bd(T ) is continuous and suppose f maps each edge of T into
itself. Let g : Bd(T ) → Bd(T ) be the restriction of f and note that g is homotopic to a
constant map. To see how: (x, s) 7→ (1 − s)g(x) + sx this makes sense because g(x) and x
always lie on a common line segment.
But Bd(T ) is homeomorphic to S 1 . The g∗ cannot be zero, or else the identity map on
1
S would induce the zero map.
So g is not homotopic to a constant. On the other hand, there exists an extension of g,
f : T → Bd(T ). This implies that f is homotopic to a constant. □
45
Theorem 16.5. A polynomial equation
xn + an−1 + · · · + a1 x + a0 = 0
of degree n > 0 with real or complex coefficients has at least one (real or complex) root.
Proof. We may reduce to the case where
|an−1 | + |an−2 | + · · · + |a1 | + |a0 | < 1.
Let c ̸= 0 consider
1
g(x) = f (cx).
cn
Then g(x) has a root if and only if f does and g satisfies the hypothesis above for c large
enough.
So now assume |an−1 | + · · · + |a0 | < 1. Let g be the restriction of f (x) to S 1 . Supposing
now that g has no root in B 2 = {z ∈ C | |z| ≤ 1}. Then g : S 1 → C − {0}.
We now claim that g is homotopic to h(x) : S 1 → C − {0} defined by h(x) = xn for all
x ∈ S1.
To see this, define H(x, t) = xn + t(an−1 xn−1 + · · · + a1 x + a0 ) for t ∈ [0, 1] and x ∈ S 1 .
Certainly H is continuous and is a homotopy from h to g, the only possible problem is
that maybe H(x, t) = 0 for some x ∈ S 1 and some t ∈ [0, 1].
But
t|an−1 xn−1 + · · · + a1 x + a0 | ≤ t(|an−1 | + · · · + |a1 | + |a0 |) < t ≤ 1
and
|xn | ≥ 1
for t ∈ [0, 1] and x ∈ S . This implies that |H(x, t)| > 0 for all x ∈ S 1 and t ∈ [0, 1].
1

So H is a homotopy from x 7→ xn and g(x) as maps S 1 → C − {0}.

But g can be extended to a continuous function f : B 2 → C − {0}, which implies that
g is homotopic to a constant by Lemma 16.2.
But the map x 7→ xn is certainly not homotopic to a constant, it generates an infinite
subgroup of π1 (S 1 , x0 ). This is a contradiction. □

Definition 16.6. Let v : B 2 → R2 be a continuous function. Then we call v a vector field

on B 2 . Each element x ∈ B 2 gets assigned (x, v(x)) which we think of as an arrow starting
at x and pointing in the direction v(x).

Theorem 16.7. Suppose that v is a nonvanishing vector field on B 2 . There exists x ∈ S 1

where the vector field points directly inward, and a point of S 1 where it points directly
outward.
Proof. Let u be the restriction of S 1 and so u : S 1 → R2 − {0}. Suppose that there is no
point x ∈ S 1 such that u(x) is a negative multiple of x. This means the same as saying
that the vector field does not point directly inward at any point x ∈ S 1 .
We claim that u is homotopic to the inclusion map j : S 1 → R2 − {0}. Define H(x, t) =
tx + (1 − t)u(x). Certainly H is continuous and starts at u and ends up at the identity.
The only problem may be that H(x, t) = 0 for some x ∈ S 1 and t ∈ [0, 1]. Immediately we
see that if H(x, t) = 0 then t ̸= 0, 1 as both u and the identity do not vanish on S 1 .
If H(x, t) = 0 for some 0 < t < 1 we have
tx + (1 − t)u(x) = 0
46
or in other words,
−t
u(x) = x,
1−t
so that u(x) is a negative multiple of x, a contradiction. So H is a homotopy from the u
to the identity. Therefore, u is an essential map S 1 → R2 − {0}. But on the other hand
u has an extension v : B 2 → R2 − {0} which implies that u is inessential (Lemma 16.2).
This is a contradiction.
Now, suppose that v does not point directly outward at any point of S 1 . Then −v(x) is
a nonvanishing vector field that does not point directly inward, which is impossible by our
previous argument. □
Theorem 16.8. If f : B 2 → B 2 is continuous, then there exists x ∈ B 2 such that f (x) = x.
Proof. Suppose that f (x) ̸= x for all x ∈ B 2 . Let v(x) = f (x) − x which is a nonvanishing
vector field B 2 → R2 . Therefore, there exists some x ∈ S 1 such that v(x) = ax for some
a > 0 by Theorem 16.7.
So f (x) = (1 + t)x for some x ∈ B 2 and t > 0. But then f (x) ∈ / B 2 , a contradiction. □
Corollary 16.9. Let A be a 3-by-3 matrix of positive real numbers. Then A has a positive
real eigenvalue.
Proof. Let
B = {x = (x1 , x2 , x3 ) ∈ R3 | ∥x∥ = 1, x1 , x2 , x3 ≥ 0}.
Then B is homeomorphic to B 2 .
Define T : B → B by
Ax
T (x) = .
∥Ax∥
Then T is well-defined since A has positive entries. So T has a fixed point by Theorem 16.8,
say x0 . Then
Ax0 = ∥Ax0 ∥x0 ,
which implies that ∥Ax0 ∥ is a positive eigenvalue for A. □
Theorem 16.10. There is no nonvanishing tangent vector field on S 2 .
Proof. Suppose that v : S 2 → R3 − {0} is a non-vanishing vector field on S 2 . Let p =
(0, 0, 1) ∈ S 2 We may assume wlog that v(p) is parallel to the y-axis. Let U be a small
enough open ball centred at p so that for q ∈ U , v(q) is almost parallel to the y-axis.
Let F : S 2 − {p} → R2 be stereographic projection (this is a homeomorphism). Not
only that, there is a way to (continuously) map each tangent vector (x, v(x)) to a (y, w)
where y = F (x), and the mapping (x, v(x)) → (y, w) is continuous.
To see how, in S 2 − {p} find a curve C intersecting x, whose tangent vector at x is the
point v(x). Then map C to F (C) ⊆ R2 and find the tangent vector to F (C) at y = F (x),
call this vector w.
Check that if v is non-vanishing then the vector field w : R2 → R2 is also non-vanishing.
Now, look at the image B = F (S 2 −U ). This is a large open ball. Look at what happens
to the vector field v on the boundary of U when mapped to B. On the boundary of U the
vectors v(q) point in almost the same direction. When we do the stereographic projection,
the vectors will now rotate twice around the origin. Let h : S → R2 − {0} be the restriction
of w to the boundary of B. It follows that h is not homotopic to the constant map.
47
 But now w : B → R2 − {0} is non-vanishing, and an extension of h to B, so that h must
be homotopic to the constant map (Lemma 16.2. □
Theorem 16.11. Let h, k : X → Y . Let h(x0 ) = y0 and k(x0 ) = y1 . If h and k are
homotopic, then there is a path α in Y from y0 to y1 such that k∗ = α̂ ◦ h∗ . If y0 = y1 ,
and if the base point x0 remains fixed during the homotopy, then k∗ = h∗ . We have the
following diagram

h∗
π1 (X, x0 ) π1 (Y, y0 )
k∗
α̂

π1 (Y, y1 )

Proof. Let H : X × [0, 1] → Y be the homotopy from h to k. Let α(s) = H(s, x0 ). Then
α is a path in Y from y0 to y1 . We need to show
k∗ = α̂ ◦ h∗ .
That means for all loops f at x0 ,
k∗ ([f ]) = α̂ ◦ h∗ ([f ]),
and once again rewriting
[k ◦ f ] = [α] ∗ [h ◦ f ] ∗ [α],
We can rewrite again as
[α] ∗ [k ◦ f ] = [h ◦ f ] ∗ [α].
So we need to construct a homotopy between
α ∗ (k ◦ f )
and
(h ◦ f ) ∗ α.
Define a path-homotopy G such that

 α(2s) s ≤ t/2
G(s, t) = H(f (2s − t), t) t/2 ≤ s ≤ (t + 1)/2
α(2s − 1) (t + 1)/2 ≤ s ≤ 1


One can check that G is continuous. Furthermore, G is a path homotopy from α ∗ (k ◦ f )

to (h ◦ f ) ∗ α.
So
k∗ = α̂h∗ .
If in addition y0 = y1 and the base point x0 remains fixed in the homotopy, then it
follows that α(s) = H(s, x0 ) is a constant path, and so k∗ = h∗ . □
Corollary 16.12. Let h, k : X → Y and let h(x0 ) = y0 and k(x0 ) = y1 . Suppose that h
and k are homotopic. If k∗ is injective (or surjective, or zero) then so is h∗ . In particular,
if h is homotopic to a constant map, then h∗ is zero.
Definition 16.13. A continuous map f : X → Y is called a homotopy equivalence if there
is a continuous map g : Y → X such that g ◦ f is homotopic to the identity iX : X → X
and f ◦ g is homotopic to the identity iY : Y → Y . The map g is said to be a homotopy
inverse for f .
48
Theorem 16.14. Let f : X → Y be continuous. Let f (x0 ) = y0 . If f is a homotopy
equivalence then
f∗ : π1 (X, x0 ) → π1 (Y, y0 )
is an isomorphism.
Proof. Let g : Y → X be a homotopy inverse for X. Let x1 = g(y0 ), and y1 = f (x1 ). We
have
f∗,0
π1 (X, x0 ) π1 (Y, y0 )

g∗
f∗,1
π1 (X, x1 ) π1 (Y, y1 )
Now, g ◦ f : (X, x0 ) → (X, x1 ) and since g ◦ f is homotopic to the identity map X → X,
we have
(g ◦ f )∗ = α̂ ◦ (iX )∗ = α̂.
So g ◦ f )∗ = g∗ f∗,0 is an isomorphism.
Similarly, f ◦ g)∗ = f∗,1 ◦ g∗ is an isomorphism π1 (Y, y0 ) → π1 (Y, y1 ).
This implies that g∗ is an isomorphism. Since α̂ is an isomorphism and g∗ is an isomor-
phism so is f∗,0 . □

17. Covering Spaces

Definition 17.1. Suppose p : E → B and p′ : E ′ → B are two covering spaces for B. We
say that (p, E) is equivalent to (p′ , E ′ ) if there exists a homeomorphism h : E ′ → E such
that the following diagram is commutative
h
E′ E
′
p
p
B

Lemma 17.2. Let p : E → B be a covering map, where E and B are path-connected. Let
b0 ∈ B. As e ranges over p−1 (b0 ), the group p∗ (π1 (E, e)) ranges precisely over a conjugacy
class of subgroups of π1 (B, b0 ).
Proof. Let e0 , e1 ∈ p−1 (b0 ). Let
H0 = p∗ (π1 (E, e0 ))
and
H1 = p∗ (π1 (E, e1 ))
and we have to show that H1 and H0 are conjugate in π1 (B, b0 ). Since E is path-connected,
let γ be a path from e0 to e1 . Let α = p ◦ γ, which is a loop at b0 . Let [h] ∈ H1 so that
[h] = [p ◦ h̃] for some loop h̃ in E at e1 . Then consider
[α] ∗ [h] ∗ [α] = [p ◦ γ] ∗ [p ◦ h̃] ∗ [p ◦ γ]
= p∗ ([γ ∗ h̃ ∗ γ]) ∈ H0
since γ ∗ h̃ ∗ γ is a loop based at e0 .
So [α]H1 [α]−1 ⊆ H0 and the same argument shows
[α]H0 [α]−1 ⊆ H1
49
so that H0 = [α]H1 [α]−1 , as required.
Now, let H = [α]H0 [α]−1 for some loop α at b0 .
We claim that there exists e1 such that p(e1 ) = b0 and for which H = p∗ π1 (E, e1 ).
There exists a lifting of α, γ, by Lemma 15.20. Let e1 = γ(1) and notice that
p∗ (π1 (E, e1 )) = [α]H0 [α]−1 = H,
as required. □
Lemma 17.3. Let p : E → B be a covering map. Let p(e0 ) = b0 . Let f : Y → B be a
continuous map, with f (y0 ) = b0 . Suppose Y is path connected and locally path connected.
The map f can be lifted to a map f˜ : Y → E such that f˜(y0 ) = e0 if and only if
f∗ (π1 (Y, y0 )) ⊂ p∗ (π1 (E, e0 )).
Furthermore, if such a lifting exists, it is unique.
Proof. Suppose that f can be lifted to f˜ : Y → E such that f˜(y0 ) = e0 . Then
f∗ (π1 (Y, y0 )) = p∗ (f˜∗ (π1 (Y, y0 ))
⊆ p∗ (π1 (E, e0 ))
Now, we must show that if such f˜ exists then it is unique. Let y1 ∈ Y and choose a
path α from y0 to y1 . Then f˜◦ α is a lifting to E of the path f ◦ α in B. But Lemma 15.20
says that a lifting of a path will be unique, in other words f˜(y1 ) is determined by the fact
that it comes from the unique lifting of f ◦ α.
Now, for each y1 ∈ Y , choose a path α from y0 to y1 , lift the path f ◦ α to E to get a
path γ (Lemma 15.20) and define f˜(y0 ) = γ(1). We first claim that this choice does not
depend on the choice of path α, so let β be another path from y0 to y1 , and let γ and δ
be liftings of the paths f ◦ α and f ◦ β respectively. We want to show that γ and δ end at
the same point.
Let ϵ be a lifting of f ◦ β to a path of E starting at γ(1). Our goal will be to show that ϵ
starts at γ(1) and ends at e0 . Then ϵ will start at e0 and be a lifting of f ◦ β. We therefore
would have δ = ϵ which would imply that δ(1) = ϵ(1) = γ(1) which would prove that f˜ is
well-defined.
Consider γ ∗ ϵ - a lifting of the path (f ◦ α) ∗ (f ◦ β) = f ◦ (α ∗ β) = f∗ (α ∗ β). We have
that f∗ (α ∗ β) ∈ p∗ (π1 (E, e0 )) by hypothesis. So
[(f ◦ α) ∗ (f ◦ β)] = [p ◦ ϕ]
for some loop ϕ at e0 .
Now, γ ∗ ϵ and ϕ are both liftings of the same homotopy class. By Theorem 15.22, ϕ
and γ ∗ ϵ end at the same point of E, that is, ϵ(1) = e0 .
So f˜ is well-defined. Now, let U be an open set in E such that f˜(y1 ) ∈ U . Choose an
open set V of B for which f (y1 ) ∈ V and V is evenly covered by p. There exists open V0 of
E with f˜(y1 ) ∈ V0 such that p|V0 : V0 → V is a homeomorphism since V is evenly covered
by p. We may assume that V0 ⊆ U .
Now, let W be open in B, path-connected and with W ⊆ f −1 (V ). We claim that
˜
f (W ) ⊆ V0 . Let y ∈ W and let β be a path from y1 to y. Then f ◦ β is a path in B
from f (y1 ) to f (y). This is a path in V which we may lift to a path in V0 by using the
homeomorphism p|V0 : V0 → V . Call this path in V0 , δ. Now, γ ∗ δ is a path that starts at
e0 , it is a lifting of f ◦ (α ∗ β), and so f˜(y) is the end point of γ ∗ δ, which lies in V0 . So
W ⊆ f −1 (V0 ) ⊆ f −1 (U ) as required. □
50
Theorem 17.4. Let B be path connected and locally path connected. Let p : E → B and
p′ : E ′ → B be path-connected covering spaces of B. Let p(e0 ) = p′ (e′0 ) = b0 . Then p and p′
are equivalent covering maps if and only if p∗ (π1 (E, e0 )) and p′∗ (π1 (E ′ , e′0 )) are conjugate
subgroups of π1 (B, b0 ).
Proof. Suppose that h : E ′ → E is a homeomorphism as in Definition 17.1. Then, by
Theorem 15.14, we have
h∗
π1 (E ′ , e′0 ) π1 (E, e1 )
where e1 = h(e′0 ). Therefore,
p′∗ (π1 (E ′ , e′0 )) = p∗ (π1 (E, e1 ))
and by Lemma 17.2, p′∗ (π1 (E ′ , e′0 )) is conjugate to p∗ (π1 (E, e0 )).
Now, supposing that p∗ (π1 (E, e0 )) and p′∗ (π1 (E ′ , e′0 )) are conjugate, by Lemma 17.2, we
may assume that the two are equal by changing the base point e′0 .
Consider the diagram
E′
p̃
p′
p
E B
and
E
p̃′
p

′ p′
E B
where the dotted arrows are provided by Lemma 17.3 (and E and E ′ are locally path
connected because B is locally path connected and E and E ′ are locally homeomorphic to
B).
Now we claim that p̃′ ◦ p̃ is the identity on E. This is because there is a unique lifting
of the map p : E → B to a map E → E, that being the identity map (Lemma 17.3). As
p̃′ ◦ p̃ is another such map p̃′ ◦ p̃ is the identity. Similarly, p̃ ◦ p̃′ is the identity on E ′ , as
required. □
Definition 17.5. Let B be a topological space. Then B is called semilocally simply
connected if for every point b ∈ B there exists open V with b ∈ V and the homomorphism
induced by the inclusion map i : V → B, i∗ : π1 (V, b) → π1 (B, b) is zero.
Theorem 17.6. Let B be path connected, locally path connected, and semilocally simply
connected. Let b0 ∈ B. Given H ⊆ π1 (B, b0 ) there exists a path connected covering space
p : E → B and a point e0 ∈ p−1 (b0 ) such that
p∗ (π1 (E, e0 )) = H.
Proof. Let P be the set of all paths in B starting at b0 . For α, β ∈ P, define α ∼ β if and
only if α and β end at the same point in B and [α ∗ β] ∈ H.
Note that ∼ is an equivalence relation on P and put E = P/ ∼. For α ∈ P, let α# be
its equivalence class in E. Define p : E → B by the rule p(α# ) = α(1). Immediately, p is
surjective, and we want E to be path connected, and for p to be a covering map.
Note that
a) if α ∼
=p β, then α# = β #
51
 b) if α# = β # , then (α ∗ δ)# = (β ∗ δ)# for any path δ beginning at α(1) = β(1).
The first point is because then α∗β is homotopic to a constant loop, so that [α∗β] = e ∈ H.
The second point is because
[α ∗ δ ∗ δ ∗ β] = [α ∗ β] ∈ H.
Now, let α ∈ P and U be an open set of B. Define B(U, α) to be the collection of those
(α ∗ δ)# such that δ is a path in U starting at α(1).
First, we show that if β # ∈ B(U, α) then B(U, α) = B(U, β). If β # ∈ B(U, α) then we
can write
β # = (α ∗ δ)#
for some path δ starting at α(1) in U . A generic element of B(U, β) will then be of the
form
(β ∗ γ)# = ((α ∗ δ) ∗ γ)# = (α ∗ (δ ∗ γ))# ∈ B(U, α).
But also α# = (β ∗ δ)# , which implies that B(U, α) ⊆ B(U, β). So B(U, α) = B(U, β).
Clearly {B(U, α)} covers E. Suppose γ # ∈ B(U1 , α) ∩ B(U2 , β). Let V be a path
connected open subset of U1 ∩ U2 such that γ(1) ∈ V . Then B(V, γ) ⊆ B(U1 , α) and
B(V, γ) ⊆ B(U2 , β).
So we have a basis of a topology. Under this topology for E, p is continuous and open.
Now, we prove that p is a covering map. Let b1 ∈ B and let U be a path-connected
open with b1 ∈ U such that
π1 (U, b1 ) → π1 (B, b1 )
is the trivial map.
We claim that p−1 (U ) is the union of all B(U, α) where α runs over paths in B from b0
to b1 .
Certainly if b ∈ U , then there is a path δ from b1 to b in U and we have
(α ∗ δ)# ∈ B(U, α)
and also δ(1) = b, so that p((α ∗ δ)# ) = b.
Now, we claim that if B(U, α) ̸= B(U, β) then the two are disjoint. If not, then γ # ∈
B(U, α) ∩ B(U, β) which implies that B(U, α) = B(U, γ) = B(U, β), a contradiction.
Now, we show that the restriction of p, B(U, α) → U is bijective. Suppose p((α ∗ δ)# ) =
p((α ∗ γ)# ) for paths γ and δ in U starting at α(1). Then δ(1) = γ(1) by definition of p. So
δ and γ are paths in U which both start and end at the same point. So δ ∗ γ is homotopic
to the trivial loop since π1 (U, b1 ) is trivial.
This implies that α ∗ δ ∼=p α ∗ γ which implies
(α ∗ δ)# = (α ∗ γ)# .
So p restricted to B(U, α) is 1-1 and onto U . Also, p is continuous and open, so it is a
homeomorphism B(U, α) → U .
So p is a covering map.
It remains to show that E path connected and to show that p∗ (π1 (E, e0 )) = H for some
e0 with p(e0 ) = b0 .
Let α be a path in B starting at b0 . Let αt be the path in B starting at b0 and ending
at α(t) (tracing along the path of α). Then define α̃(t) = (αt )# for 0 ≤ t ≤ 1.
It is clear that p(α̃(t)) = α(t) for all 0 ≤ t ≤ 1.
We claim that α̃ is continuous. Let t0 be given and let B(U, αt0 ) be a given neighborhood
of α̃(t0 ) = (αt0 )# . It is our job to find ϵ such that if |t1 − t0 | < ϵ then (αt1 )# ∈ B(U, αt0 ).
52
But α is continuous, so choose ϵ so that if |t1 − t0 | < ϵ then α(t1 ) ∈ U . Now, we claim that
if |t1 − t0 | < ϵ then there exists a path δ in U such that
αt ∼
1 =p αt ∗ δ.
0

This will establish that α̃(t1 ) ∈ B(U, αt0 ) so that α̃ is continuous.

To find δ, we assume that t0 < t1 (the case that t1 < t0 ) is similar). Let δ be the path
(following the path of α) going from α(t0 ) to α(t1 ). That is
δ(t) = α(t · t1 + (1 − t)t0 ).
∼
We claim that αt0 ∗ δ =p αt1 . The path homotopy is:
(  
1 t)2s
α (t0 (1−t)+t
1+t if 0 ≤ s ≤ (t + 1)/2
F (s, t) =
α(t0 (1 − (2s − 1)) + t1 (2s − 1)) if (t + 1)/2 ≤ s ≤ 1
Which proves that α̃ is continuous.
Immediately we see that E is path-connected. Let e0 be (eb0 )# where eb0 is the constant
path at b0 . Then if (α)# ∈ E then let α̃ be the lifting of the path α to E and notice that
α̃ starts at e0 and ends at α# .
The last thing is to show that p∗ (π1 (E, e0 )) = H. Let [α] ∈ H. Let α̃ be the lifting of α
to a path in E. We claim that α̃ is a loop at e0 , whence [α] = p∗ (α̃). Now, α̃(1) = (α)# .
But (α)# = (eb0 )# where eb0 is the constant loop at b0 . This is because [α ∗ eb0 ] = [α] ∈ H
so α ∼ eb0 so α# = (eb0 )# . So α̃ is a loop at e0 , as required.
Now, suppose that [α̃] ∈ π1 (E, e0 ). Let α = p ◦ α̃, so that α is a loop at b0 . But there
is a unique lifting of α to E, so it must be that α̃(t) = (αt )# . But then e0 = α̃(1) = (α)# ,
which implies that
[α ∗ eb0 ] ∈ H
so that [α] ∈ H as required. □
Corollary 17.7. Let B be path connected and locally path connected.
a) If B has a universal covering space, that covering space is uniquely determined up
to equivalence.
b) If B is semilocally simply connected, then B has a universal covering space.
Proof. To get the first part, apply Theorem 17.4 because if E is a universal covering space
then p∗ : π1 (E, e0 ) → π1 (B, b0 ) is the trivial map.
For the second part, use Theorem 17.6 to construct path connected (E, e0 ) so that
p∗ (π1 (E, e0 )) = 0. Now prove that p∗ is injective. Suppose p∗ [α̃] = [eb0 ]. Let α = p ◦ α̃.
Since α ∼ =p eb0 it follows that αt ∼ =p eb0 for all 0 ≤ t ≤ 1. Therefore, α̃(t) = (eb0 )# for all
t, so α̃ is the constant loop. □

References
[Mun00] James R. Munkres. Topology. Prentice Hall, Inc., Upper Saddle River, NJ, 2000,
pp. xvi+537.

53