MTH 304: General Topology

Semester 2, 2017-2018

Dr. Prahlad Vaidyanathan

Contents
I. Continuous Functions 3
1. First Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Open Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Continuity by Open Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

II. Topological Spaces 8

1. Definition and Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2. Metric Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3. Basis for a topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4. The Product Topology on X Q× Y . . . . . . . . . . . . . . . . . . . . . . 18
5. The Product Topology on Xα . . . . . . . . . . . . . . . . . . . . . . . 20
6. Closed Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
7. Continuous Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8. The Quotient Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

III. Properties of Topological Spaces 36

1. The Hausdorff property . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2. Connectedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3. Path Connectedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4. Local Connectedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5. Compactness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6. Compact Subsets of Rn . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
7. Continuous Functions on Compact Sets . . . . . . . . . . . . . . . . . . . 52
8. Compactness in Metric Spaces . . . . . . . . . . . . . . . . . . . . . . . . 56
9. Local Compactness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

IV.Separation Axioms 62
1. Regular Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2. Normal Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
3. Tietze’s extension Theorem . . . . . . . . . . . . . . . . . . . . . . . . . 67
4. Urysohn Metrization Theorem . . . . . . . . . . . . . . . . . . . . . . . . 71
5. Imbedding of Manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

V. Instructor Notes 80

2
I. Continuous Functions
1. First Definitions
Following [Crossley, Section 2.1,2.2]
Let S ⊂ R. A function in this section will be a real-valued function whose domain is S.
1.1. Remark: Consider two graphs (one continuous and other discontinuous at x = 1).
Continuity means that we can draw the graph of f without lifting our pencil. ie.
If we approach a point on the x axis from either direction, the value of f (x) should
be ‘predicted’ by the values of f (y) where y is near x. Furthermore, Continuity is
a local property.
1.2. Definition: A function f : S → R is said to be sequentially continuous at a ∈ S if,
for any sequence (xn ) ⊂ S such that xn → a, we have f (xn ) → f (a).
1.3. Example: f (x) = x/|x| for x 6= 0 and f (0) = 1
(i) If we choose a = 0 and xn = 1/n, then f (a) = lim f (xn )
(ii) However, if we choose xn = −1/n, then f (a) 6= lim f (xn ).
So f is not sequentially continuous.
1.4. Definition: A function f : S → R is said to be continuous at a if, for every
 > 0, ∃δ > 0 such that
|x − a| < δ ⇒ |f (x) − f (a)| <  (∗)

1.5. Example:
(i) f (x) = x2 is continuous at 2
(a) If a = 0,  = 1, we want δ > 0 such that (∗) holds. ie. We want
|x| < δ ⇒ |x2 | < 1
Since |x2 | = |x|2 , we may choose δ = 1.
(b) If a = 2,  = 1, we want δ > 0 such that (∗) holds. ie. We want
|x − 2| < δ ⇒ |x2 − 22 | < 1
Notice that δ = 1 does not work, because if x = 2.9 then x2 ≈ 9.
However,
|x2 − 22 | = |x − 2||x + 2|
So ∃δ > 0 that works.

3
 (ii) f (x) = x2 if x 6= 0 and f (0) = 0.5 is discontinuous at 1.
(a) If  = 1, then δ = 0.5 works because if

|x| < 0.5 ⇒ |x2 | < 0.25 < 1, and |f (0)| = 0.5 < 1

(b) However, if  = 0.2, then no δ > 0 works because if |x| < δ, then we may
choose small enough x so that |x| < 0.5, so that |x2 | < 0.25 and hence

|x2 − 0.5| > 0.25

So f is discontinuous at 0.
1.6. Theorem: f is continuous at a if and only if it is sequentially continuous at a
Proof. (i) Suppose f is continuous at a and (xn ) ⊂ S is a sequence such that
xn → a. WTS: f (xn ) → f (a), so choose  > 0, then ∃δ > 0 such that

|x − a| < δ ⇒ |f (x) − f (a)| < 

For this δ > 0, ∃N ∈ N such that |xn − a| < δ for all n ≥ N . Hence,

|f (xn ) − f (a)| <  ∀n ≥ N

This is true for any  > 0 so f (xn ) → f (a)

(ii) Suppose f is sequentially continuous at a, but it is not continuous at a, then
∃ > 0 for which no δ works. Hence, δ = 1/n does not work, so ∃xn ∈ S such
that
|xn − a| < 1/n, but |f (xn ) − f (a)| ≥ 
Clearly, xn → a, but f (xn ) does not converge to f (a). Hence, f is not
sequentially continuous - a contradiction.

(End of Day 1)

2. Open Sets
[Crossley, Section 2.3]

2.1. Remark:  − δ definition says that f is continuous at a if and only if, for any
 > 0, ∃δ > 0 such that

x ∈ (a − δ, a + δ) ⇒ f (x) ∈ (f (a) − , f (a) + )

2.2. Definition:
(i) Open interval

4
 (ii) Open set: A set U ⊂ R is open if and only if it can be written as a union of
open intervals. (Note: We are not restricting ourselves to finite unions. ie.
We are referring to ‘arbitrary’ unions)
2.3. Theorem: A set U ⊂ R is open iff for all x ∈ U, ∃δx > 0 such that (x−δx , x+δx ) ⊂ U
Note: The value of δx depends on x.
Proof. (i) Suppose that, for any x ∈ U, ∃δx > 0 such that (x − δx , x + δx ) ⊂ U ,
then [
U= (x − δx , x + δx )
x∈U

so U is open.
S
(ii) Conversely, if U is open, then write U = α∈J Iα , where each Iα is an open
interval. If x ∈ U , then ∃α ∈ J such that x ∈ Iα . Write Iα = (a, b), then
a < x < b, so
δx = min{|x − a|/2, |b − x|/2}
works.

2.4. Examples:
(i) (a, b)
(ii) A closed interval (or even a half-open interval) is not open.
(iii) {0} is not open. A finite set is not open.
2.5. Proposition:
(i) An arbitrary union of open sets is open.
(ii) A finite intersection of open sets is open.
Proof. (i) is obvious, so we prove (ii): By induction, it suffices to consider the
case of two sets, U1 , U2 say. WTS: U1 ∩ U2 is open, so fix x ∈ U1 ∩ U2 , then
∃δ1 , δ2 > 0 such that (x − δi , x + δi ) ⊂ Ui , i = 1, 2. Then if δ = min{δ1 , δ2 }, then
(x − δ, x + δ) ⊂ U1 ∩ U2 , which verifies Theorem 2.3.
2.6. Example: A countable intersection of open sets may not be open. Un = (−1/n, 1/n)
2.7. Definition: A set F ⊂ R is closed if F c is open.
2.8. Examples:
(i) Closed interval
(ii) [2, ∞) is closed.
(iii) Arbitrary intersection of closed sets is closed.
(iv) Finite union of closed sets is closed.
(v) [1, 2) is neither open nor closed.

5
3. Continuity by Open Sets
[Crossley, Section 2.4]

3.1. Definition: Let f : X → Y be a function between two sets and A ⊂ Y , then

f −1 (A) = {x ∈ X : f (x) ∈ A}

Note: This definition does not imply that f −1 exists as a function. It is simply
notation.
3.2. Example: f (x) = x2 − x = x(x − 1)
(i) f −1 (R) = R
(ii) f −1 (∅) = ∅
(iii) f −1 [−1, ∞) = R
(iv) f −1 [0, ∞) = R \ (0, 1)
(v) f −1 ({0}) = {0, 1}
3.3. Proposition: Let f : X → Y and {Aα : α ∈ J} be a collection of subset of Y , then
(i) f −1 (∅) = ∅
(ii) f −1 (Y ) = X
(iii) f −1 ( Aα ) = f −1 (Aα )
T T

(iv) f −1 ( Aα ) = f −1 (Aα )
S S

[HW]
3.4. Theorem: Let f : R → R, then f is continuous if and only if f −1 (U ) is open
whenever U is open.
Proof. (i) Suppose f is continuous and U is open in R. WTS: f −1 (U ) is open,
so fix x ∈ f −1 (U ). So that f (x) ∈ U , so ∃ > 0 such that

(f (x) − , f (x) + ) ⊂ U

By definition of continuity, ∃δ > 0 such that

|y − x| < δ ⇒ |f (y) − f (x)| < 

So if y ∈ (x − δ, x + δ), then f (y) ∈ (f (x) − , f (x) + ) ⊂ U . Hence,

(x − δ, x + δ) ⊂ f −1 (U )

This is true for any x ∈ f −1 (U ). By Theorem 2.3, f −1 (U ) is open.

6
(ii) Suppose f −1 (U ) is open whenever U is open. Fix a ∈ R,  > 0. Then

U = (f (a) − , f (a) + )

is open in R so f −1 (U ) is open. Since a ∈ f −1 (U ), ∃δ > 0 such that

(a − δ, a + δ) ⊂ f −1 (U )

Hence, if x ∈ R such that |x − a| < δ, then |f (x) − f (a)| < .

(End of Day 2)

7
II. Topological Spaces
1. Definition and Examples
1.1. Definition: Let X be a set. A collection τ of subsets of X is called a topology on
X if
(i) ∅, X ∈ τ
(ii) If U1 , U2 ∈ τ , then U1 ∩ U2 ∈ τ
S
(iii) If {Uα : α ∈ J} is an arbitrary collection of sets in τ , then α∈J Uα ∈ τ
The pair (X, τ ) is called a topological space, and members of τ are called open
sets in X.
1.2. Examples:
(i) X = R and τ = the collection of open sets in R (as defined in the previous
section) is a topological space. This is called the usual topology on R
(ii) Let X = R2 .
(a) Fix a := (a1 , a2 ) ∈ X, r > 0. An open disc in X centered at x of radius
r is the set
p
B(a, r) := {(x1 , x2 ) ∈ R2 : (x1 − a1 )2 + (x2 − a2 )2 < r}

(b) A set U ⊂ R2 is said to be open if it is a union of open discs. As in

Theorem 2.3, a set U ⊂ R2 is open if and only if, for any a ∈ U, ∃r > 0
such that B(a, r) ⊂ U .
(c) As in Proposition 2.5, an arbitrary union of open sets is open, and a finite
intersection of open sets is open. Hence, this collection of open sets forms
a topology on R2 . This is called the Euclidean topology on R2 .
(iii) Let X be any set and τ = {∅, X}. This is called the indiscrete topology on
X.
(iv) Let X be any set and τ = P(X). This is called the discrete topology on X.
1.3. Definition: Let (X, τX ) and (Y, τY ) be topological spaces. A function f : X → Y
is said to be continuous if f −1 (U ) ∈ τX whenever U ∈ τY . ie. The inverse image
of an open set is open.
Note: We think of continuity as a global property here, and don’t care if a function
is continuous at all but one point.

8
1.4. Example:
(i) Let f : R → R be f (x) = x2 is continuous, but f (x) = x/|x| if x 6= 0 and
f (0) = 1 is discontinuous.
(ii) Let (X, τd ) be a discrete topological space, and (Y, τY ) any topological space.
If f : X → Y is any function, then f is continuous.
(iii) Similarly, if (X, τX ) is any topological space and (Y, τi ) is an indiscrete topo-
logical space, then any function f : X → Y is continuous.
(iv) Let f : X → Y be a constant function, then f is continuous.
Proof. Suppose f (x) = y0 for all x ∈ X. Let U be an open set in Y , then
(
∅ : if y0 ∈
/U
f −1 (U ) =
X : if y0 ∈ U

In either case, f −1 (U ) is open.

(v) Let A : R2 → R be the addition map A(x, y) = x + y. Then A is continuous.
Proof. Let U ⊂ R be open. We WTS: A−1 (U ) is open. As mentioned above,
it suffices to show that, for any point (a, b) ∈ A−1 (U ), ∃r > 0 such that
B((a, b), r) ⊂ A−1 (U ). So fix (a, b) ∈ A−1 (U ). Then a + b ∈ U , so ∃ > 0
such that (a + b − , a + b + ) ⊂ U . Note that A−1 ((a + b − , a + b + ))
describes the region enclosed by (but not including) the two lines

x + y = a + b −  and x + y = a + b + 

and (a, b) lies in this region. Now the distance of a point (x0 , y0 ) from a line
of the form αx + βy + γ = 0 is given by

|αx0 + βy0 + γ|
d= p
α2 + β 2

In this case, we get

|a + b + (−a − b − )| 
d= √ =√
2 2
√
Hence, if (x, y) ∈ B((a,
√ b), / 2), then (x, y) ∈ A−1 ((a + b − , a + b + )),
and hence B((a, b), / 2) ⊂ A−1 (U ), and so A−1 (U ) is open. Hence, A is
continuous.
(vi) Similarly, the multiplication map M : R2 → R given by (x, y) 7→ xy is also
continuous [We will give a simpler proof later]
(vii) Let d : R → R2 be the diagonal map d(x) = (x, x). Then d is continuous.

9
 Proof. Once again, fix an open set U ⊂ R2 and a point x ∈ d−1 (U ). WTS:
∃δ > 0 such that (x − δ, x + δ) ⊂ d−1 (U ). Since (x, x) ∈ U and U is open,
∃ > 0 such that B((x, x), ) ⊂ U . Consider the part of the √line y = x inside
this disc, and project it onto the X-axis. Note that if δ = / 2, then for any
y ∈ (x − δ, x + δ), we have
p
(x − y)2 + (x − y)2 <  ⇒ (y, y) ∈ B((x, x), )

Hence, (x − δ, x + δ) ⊂ d−1 (U )
(End of Day 3)
1.5. Theorem: Let (X, τX ) be a topological space and Y ⊂ X. Define

τY := {U ∩ Y : U ∈ τX }

Then τY is a topology on Y , and is called the subspace topology on Y . [HW]

1.6. Examples:
(i) Z ⊂ R. We claim that every subset of Z is open in the subspace topology
(ie. Z with the subspace topology is discrete). It suffices to show that every
singleton is open. To do this, fix n ∈ N, then (n − 1/2, n + 1/2) is open in R
and
(n − 1/2, n + 1/2) ∩ Z = {n}

(ii) Q ⊂ R. Here the subspace topology is not discrete because if U is an open

set in R, then U ∩ Q contains infinitely many points. In particular, singelton
sets are not open in Q.
(iii) S 1 ⊂ R2 : An example of an open set is the intersection of any disc in R2
with S 1 . This will give arcs in S 1 . Hence, every arc in S 1 is an open set.
Furthermore, since every open set in R2 is a union of discs, every open set in
S 1 is a union of arcs.
(iv) [0, 1] ⊂ R: Here, [0, 1] is itself an open set since

[0, 1] = R ∩ [0, 1]

Furthermore, [0, 1/2) is also an open set in [0, 1].

(v) If Y = [0, 1] ∪ [2, 3] ⊂ R, then [0, 1] is an open set in Y because

[0, 1] = (1/2, 3/2) ∩ Y

Similarly, [2, 3] is also an open set. Hence, [0, 1] is both open and closed in
Y.

10
2. Metric Spaces
2.1. Definition: Let X be a set. A function d : X × X → R is called a metric on X if
(i) d(x, y) ≥ 0 for all (x, y) ∈ X × X
(ii) d(x, y) = 0 if and only if x = y
(iii) d(x, y) = d(y, x)
(iv) d(x, y) ≤ d(x, z) + d(z, y) for all x, y, z ∈ X (Triangle Inequality)
The pair (X, d) is called a metric space.
2.2. Examples:
(i) R with d(x, y) = |x − y|
(ii) Similarly, C with d(z, w) = |z − w|
(iii) Rn with v
u n
uX
d(x, y) = t (xi − yi )2
i=1

Proof. Clearly, the first three axoims are satisfied, so it suffices to prove the
triangle inequality. For this, note that
n
X
d(x, y)2 = (xi − yi )2
i=1
n
X
= (xi − zi + zi − yi )2
i=1
n
X
= (xi − zi )2 + (zi − yi )2 + 2(xi − zi )(zi − yi )
i=1

But by Cauchy-Schwartz inequality,

v v
Xn u n u n
uX uX
(xi − zi )(zi − yi ) ≤ t (xi − zi ) t (zi − yi )2 = d(x, z)d(z, y)
2

i=1 i=1 i=1

Hence,

d(x, y)2 ≤ d(x, z)2 + d(y, z)2 + 2d(x, z)d(z, y) = [d(x, z) + d(y, z)]2

which gives the triangle inequality.

(iv) Rn with
d(x, y) = max |xi − yi |
1≤i≤n

This is called the uniform or supremum metric on Rn , and the metric is

written as d∞ .

11
 (v) Rn with
n
X
d(x, y) = |xi − yi |
i=1

This is called the L1 metric on Rn , and is written as d1 .

(vi) Let X be any set. Define d : X × X → R by
(
0 :x=y
d(x, y) =
1 : x 6= y

This is called the discrete metric on X.

2.3. Definition: Let (X, d) be a metric space.
(i) An open ball of radius r > 0 centered at a point a ∈ X is the set

B(a, r) := {x ∈ X : d(x, a) < r}

(ii) A set U ⊂ R is said to be open if it is a union of open balls. Equivalently, if,

for each a ∈ U, ∃δa > 0 such that B(a, δa ) ⊂ U
2.4. Theorem: Let (X, d) be a metric space, and τd be the collection of open sets as
defined above. Then τd is a topology on X. This is called the metric topology on
X induced by d.
Proof. (i) Clearly, ∅ ∈ τd and X ∈ τd
(ii) τd is closed under arbitrary union by definition.
(iii) If U1 , U2 ∈ τd , WTS: U1 ∩ U2 ∈ τd , so fix a ∈ U1 ∩ U2 . Then ∃δi > 0
such that B(a, δi ) ⊂ Ui . Let δ = min{δ1 , δ2 }, then if x ∈ B(a, δ), then
d(x, a) < δ ≤ δ1 ⇒ x ∈ B(a, δ1 ) ⊂ U1 . Similarly, x ∈ U2 , so B(a, δ) ⊂ U1 ∩U2 .

2.5. Definition: Let (X, d) be a metric space. We say that a sequence (xn ) ⊂ X
converges to a point a ∈ X if, for each  > 0, ∃N ∈ N such that d(xn , a) <  for
all n ≥ N . If this happens, we write xn → a.
(End of Day 4)
2.6. Theorem: Let (X, dX ) and (Y, dY ) be two metric space, f : X → Y a function.
Then TFAE:
(i) For any a ∈ X and any sequence (xn ) ⊂ X such that xn → a implies
f (xn ) → f (a)
(ii) For any a ∈ X and each  > 0, ∃δ > 0 such that

dX (x, a) < δ ⇒ dY (f (x), f (a)) < 

12
 (iii) f −1 (U ) is open in X whenever U is open in Y (with respect to the metric
topologies on each).
Proof.

(i) ⇒ (ii): Suppose (i) holds and a ∈ X is fixed and  > 0 given. Suppose no δ > 0
works, then for each n ∈ N, δ = 1/n does not work. So ∃xn ∈ X such that
dX (xn , a) < 1/n, but dY (f (xn ), f (a)) ≥ 
So xn → a and f (xn ) does not converge to f (a) contradicting (i).
(ii) ⇒ (iii): Suppose U is open in X. WTS: f −1 (U ) is open in Y , so choose a ∈ f −1 (U ).
Then f (a) ∈ U and U is open, so ∃ > 0 such that
BY (f (a), ) ⊂ U
Now by (ii), choose δ > 0 such that
dX (x, a) < δ ⇒ dY (f (x), f (a)) < 
Then clearly BX (a, δ) ⊂ f −1 (U ), so that f −1 (U ) is open.
(iii) ⇒ (i) Suppose a ∈ X and xn → a. WTS: f (xn ) → f (a). So fix  > 0, then
U = BY (f (a), ) is open so f −1 (U ) is an open set containing a. Hence,
∃δ > 0 such that BX (a, δ) ⊂ f −1 (U ). Since xn → a, ∃N ∈ N such that
dX (xn , a) < δ ∀n ≥ N
Hence, xn ∈ f −1 (U ) so that f (xn ) ∈ U , whence
dY (f (xn ), f (a)) <  ∀n ≥ N
Hence, f (xn ) → f (a).

2.7. Example:
(i) Let M : R2 → R be the multiplication map (x, y) 7→ xy. Then M is continu-
ous.
Proof. Choose a sequence (xn , yn ) → (a, b). Then
p
|xn − a| ≤ |xn − a|2 + |yn − b|2 = d((xn , yn ), (a, b)) → 0
So xn → a in R. Similarly, yn → b in R. Hence,
|xn yn − ab| ≤ |xn yn − ayn | + |ayn − ab| = |xn − a||yn | + |a||yn − b|
Since yn → b, (yn ) is bounded, so ∃M > 0 such that |yn | ≤ M for all n ∈ N.
Hence,
|xn yn − ab| ≤ M |xn − a| + |a||yn − b| → 0
Hence, M is sequentially continuous, so it is continuous by the previous the-
orem.

13
 (ii) Let P : Rn → R be a polynomial function
X
P (x1 , x2 , . . . , xn ) = ai1 ,i2 ,...,in xi11 xi22 . . . xinn

Then P is continuous.
Proof. Similar to HW 1.4.
2.8. Theorem: Let (X, dX ) be a metric space and Y ⊂ X. Define dY : Y × Y → R by
dY (y1 , y2 ) = dX (y1 , y2 ). Then
(i) dY is a metric on Y , and
(ii) the metric topology induced on Y by dY coincides with the subspace topology
induced on Y from (X, τdX )
Proof. Part (i) is trivial. To check part (ii), let η denote the subspace topology on
Y and τ denote the metric topology on Y induced by dY .
(i) To show η ⊂ τ : So fix an open set V ∈ η, then ∃U open in (X, dX ) such that
V = U ∩ Y . To show that V ∈ τ , we fix a point a ∈ V . WTS: ∃δ > 0 such
that BY (a, δ) ⊂ V . Since U is open, ∃δ > 0 such that

BX (a, δ) ⊂ U

Then note that BY (a, δ) = BX (a, δ) ∩ Y ⊂ U ∩ Y = V .

(ii) To show τ ⊂ η: It suffices to show that every open ball BY (a, r) ∈ η. But
once again this follows from the fact that

BY (a, r) = BX (a, r) ∩ Y

2.9. Examples: Any subset of Rn inherits a metric topology from Rn , so is, in particular,
a metric space. For instance, this applies to
(i) (The circle) S 1 = {(x, y) ∈ R2 : x2 + y 2 = 1}
Pn+1
(ii) (The n-sphere) S n = {(x1 , x2 , . . . , xn+1 ) ∈ Rn+1 : i=1 x2i = 1}
(iii) (The cylinder) C = {(x, y, z) ∈ R3 : x2 + y 2 = 1, 0 ≤ z ≤ 1}
p
(iv) (The Torus) T = {(x, y, z) ∈ R3 : x2 + y 2 + z 2 − 4 x2 + y 2 + 3 = 0}
2.10. Theorem: Let f : X → Y be an injective function and dY is a metric on Y . Define
dX : X × X → R by
dX (x1 , x2 ) = dY (f (x1 ), f (x2 ))
Then dX is a metric on X, called the metric pulled back (or induced) by f . [HW]
Note that f is automatically continuous in this situation.

14
2.11. Lemma: Let f : X → Y be a bijective function and dY be a metric on Y . Let
dX be the metric on X induced by f . Then a function g : X → Z (some other
topological space) is continuous if and only if g ◦ f −1 : Y → Z is continuous.
Proof. Note that in the above situation, f −1 is automatically continuous from
Y → X. Hence, if g is continuous, so is g ◦f −1 . Conversely, if g ◦f −1 is continuous,
then
g = g ◦ f −1 ◦ f
is also continuous.
2.12. Example:
(i) Let Mn (R) denote the set of all n × n matrices with real entries. There is a
map
2
f : Mn (R) → Rn
that expands a matrix into a tuple. This map is clearly injective. Thus,
Mn (R) is a metric space with the metric induced by f . ie. we have
sX
d((ai,j ), (bi,j )) = (ai,j − bi,j )2
i,j

2
(ii) Consider the determinant map det : Mn (R) → R. Note that det ◦f −1 : Rn →
R is a polynomial map which is continuous. Hence, by the previous Lemma,
det is continuous.
(iii) Note that GLn (R), the set of invertible n × n matrices is the set
GLn (R) = det−1 (R \ {0})
Hence, GLn (R) is an open subset of Mn (R) and is a metric space in its own
right.
(End of Day 5)
2.13. Definition: Let X be a set and d1 , d2 be two metrics on X. We say that d1 and d2
are equivalent (In symbols, d1 ∼ d2 ) if ∃K, M > 0 such that
Kd1 (x, y) ≤ d2 (x, y) ≤ M d1 (x, y) ∀x, y ∈ X

2.14. Example: Let X = Rn and d1 , d2 be the uniform and Euclidean metrics respec-
tively. Then d1 ∼ d2
Proof.
d1 (x, y) = max{|xi − yi |} ≤ d2 (x, y)
v
u n
uX √
d2 (x, y) = t (xi − yi )2 ≤ nd1 (x, y)
i=1

15
2.15. Theorem: Let d1 and d2 be equivalent metrics on a set X, then τd1 = τd2
Proof. By symmetry, it suffices to show that τd1 ⊂ τd2 . So let K, M > 0 such that

Kd1 (x, y) ≤ d2 (x, y) ≤ M d1 (x, y) ∀x, y ∈ X

So fix U ∈ τd1 and a ∈ U . Then ∃r > 0 such that Bd1 (a, r) ⊂ U . Now if
x ∈ Bd2 (a, rK), then
d2 (x, a)
d1 (x, a) ≤ <r
K
So Bd2 (a, rK) ⊂ Bd1 (a, r) ⊂ U . Hence, U ∈ τd2 as required.
2.16. Example: (The converse of the previous theorem is not true) Let d be the usual
metric on R and
ρ(x, y) := min{|x − y|, 1}
Then
(i) τρ = τd
Proof. Since ρ(x, y) ≤ d(x, y), it follows as above that

Bd (a, r) ⊂ Bρ (a, r)

Hence, τρ ⊂ τd [Check!]. Conversely, if U ∈ τd and a ∈ U , then ∃r > 0 such

that Bd (a, r) ⊂ U . We may assume that r < 1, but in that case,

Bρ (a, r) = Bd (a, r) ⊂ U

so that U ∈ τρ as well. Hence, τd ⊂ τρ as required.

(ii) ρ is not equivalent to d
Proof. Note that ρ(x, y) ≤ 1 for all x, y ∈ R. If ∃M > 0 such that

d(x, y) ≤ M ρ(x, y)

Then this would imply that d(x, y) ≤ M for all x, y ∈ R. This is not true
because d(n, 0) = n for all n ∈ N.

3. Basis for a topology

3.1. Definition: Let (X, τ ) be a topological space. A collection B ⊂ τ of open sets is
called a basis for τ if every member of τ is a union of elements from B. Equivalently,
U ∈ τ if and only if, for each x ∈ U, ∃B ∈ B such that x ∈ B and B ⊂ U .
3.2. Examples:
(i) Let X = R with the usual topology and B be the collection of open intervals.

16
 (ii) Similarly, if (X, d) is any metric space with τ the metric topology. Then B
may denote the set of all balls (of various centers and radii).
3.3. Proposition: Let f : X → Y be a function between two topological spaces, and
suppose B is a basis for τY . Then f is continuous if and only if f −1 (B) ∈ τX for
all B ∈ B
Proof. One direction is clear, so suppose f −1 (B) ∈ τX for all B ∈ B. WTS: f is
continuous, so fix an open set U ∈ τY and we want to show f −1 (U ) ∈ τX . Fix
x ∈ f −1 (U ), then f (x) ∈ U , so ∃Bx ∈ B such that x ∈ Bx , and Bx ⊂ U . Hence,
Vx := f −1 (Bx ) ∈ τX and Vx ⊂ f −1 (U )
This is true for any x ∈ f −1 (U ) so
[
f −1 (U ) = Vx
x∈f −1 (U )

Hence, f −1 (U ) ∈ τX as required.
3.4. Lemma: Let C be a collection of subset of X. Then there is a unique topology τ
on X such that
(i) C ⊂ τ
(ii) If η is any other topology on X such that C ⊂ η, then τ ⊂ η.
ie. τ is the smallest topology containing C. This is called the topology generated
by C.
Proof. Let F be the set set of all topologies η on X such that C ⊂ η. Then F 6= ∅
because P(X) ∈ F. Now set \
τ= η
η∈F

Then check that τ is a topology that satisfies the required conditions.

3.5. Theorem: Let X be a set and B be a collection of subsets of X such that
(a) For each x ∈ X, ∃B ∈ B such that x ∈ B
(b) If B1 , B2 ∈ B and x ∈ B1 ∩ B2 , then ∃B3 ∈ B such that x ∈ B3 and
B3 ⊂ B1 ∩ B2 .
Let τ denote the topology generated by B. Then B is a basis for τ .
Proof. Let η be the collection of all subsets of X that are unions of members of
B. Claim: η is a topology on X. The first three axioms hold trivially, and the last
one follows from property (b) of B.

Now clearly, B ⊂ η, so that η ∈ F of the previous proof. Hence, τ ⊂ η. Further-

more, if µ is any topology that contains B, then η ⊂ µ because µ is closed under
arbitrary unions. Hence, η ⊂ τ as required.
(End of Day 6)

17
4. The Product Topology on X × Y
4.1. Theorem: Let (X, τX ) and (Y, τY ) be two topological spaces. Then there is a
unique topology on X × Y whose basis are sets of the form

U ×V

where U ∈ τX and V ∈ τY . This is called the product topology on X × Y , denoted

by τX×Y
Proof. Let B = {U ×V : U ∈ τX , V ∈ τY }. We check that B satisfies the conditions
of Theorem 3.5.
(i) Clearly, X × Y ∈ B
(ii) If U1 , U2 ∈ τX and V1 , V2 ∈ τY , then

(U1 × V1 ) ∩ (U2 × V2 ) = (U1 ∩ U2 ) × (V1 ∩ V2 ) ∈ B

4.2. Theorem: Suppose (X, dX ) and (Y, dY ) are metric spaces. Define d : (X ×Y )2 → R
by
d((x1 , y1 ), (x2 , y2 )) = max{dX (x1 , x2 ), dY (y1 , y2 )}
Then
(i) d is a metric on X × Y
(ii) The metric topology induced by d coincides with the product topology on
X ×Y
Proof. Part (i) is trivial, so we prove (ii). Let τd denote the metric topology and
τX×Y denote the product topology.
• WTS: τX×Y ⊂ τd : If U = BX (a, δ1 ) and V = BY (b, δ2 ) are open balls in X
and Y respectively, consider

W =U ×V

We claim that W ∈ τd . To see this, fix (x, y) ∈ W , then x ∈ U, y ∈ V , so

dX (x, a) < δ1 and dY (y, b) < δ2

Let r = min{δ1 − dX (x, a), δ2 − dY (y, b)} > 0. We claim that

Bd ((x, y), r) ⊂ W

So choose (u, v) ∈ Bd ((x, y), r), then d((u, v), (x, y)) < r, so that

dX (u, x) < r, and dY (v, y) < r

18
 Hence,

dX (u, a) ≤ dX (u, x) + dX (x, a) < r + dX (x, a) ≤ δ1 − dX (x, a) + dX (x, a) = δ1

Hence, u ∈ U . Similarly, v ∈ V , so that (u, v) ∈ W , proving the claim.

Hence,
U × V ∈ τd
for any open ball U ∈ τdX and V ∈ τdY . But these open balls form a basis for
τdX and τdY respectively. Hence, by Lemma 4.2,

τX×Y ⊂ τd

• WTS: τd ⊂ τX×Y : Let (a, b) ∈ X × Y and r > 0. It suffices to show that

Bd ((a, b), r) ⊂ τX×Y

Note that (x, y) ∈ Bd ((a, b), r) iff

dX (x, a) < r and dY (y, b) < r

Hence,
Bd ((a, b), r) = BX (a, r) × BY (b, r) ∈ τX×Y
This is true for any open d-ball in X × Y , so τd ⊂ τX×Y .

4.3. Remark: Let X1 , X2 , X3 be three topological spaces, then we may define the prod-
uct topology inductively as the product topology on (X1 ×X2 )×X3 where X1 ×X2
has the product topology. Thus, basic open sets in X1 × X2 × X3 are of the form

U1 × U2 × U3

where Ui are open in Xi . The same can be done for finitely many spaces X1 , X2 , . . . , Xn .
4.4. Corollary: The metric topology on Rn induced by the Euclidean metric is the same
as the product topology.
Proof. Theorem 4.4+2.15.
4.5. Definition: Let X, Y be sets. The maps πX : X × Y → X and πY : X × Y → Y
given by
πX (x, y) = x and πY (x, y) = y
are called the projection maps.
4.6. Lemma:
(i) The maps πX and πY are continuous if X × Y is equipped with the product
topology.

19
 (ii) If η is a topology on X × Y such that πX and πY are both continuous, then
τX×Y ⊂ η.
Proof. (i) If U ⊂ X is open, then
−1
πX (U ) = U × Y ∈ τX×Y
and similarly for πY .
(ii) If η is a topology such that πX and πY are continuous, then for any U, V open
in X, Y respectively,
−1
U × V = πX (U ) ∩ πY−1 (V ) ∈ η
Hence, τX×Y ⊂ η.

4.7. Theorem: Let f : Z → X × Y be a function. Then f is continuous if and only if

πX ◦ f and πY ◦ f are continuous.
Proof. If f is continuous then πX ◦ f and πY ◦ f are continuous by the previous
lemma and HW1. Conversely, suppose f1 := πX ◦f and f2 := πY ◦f are continuous,
and WTS: f is continuous. By Prop. 3.3, it suffices to show that f −1 (W ) is open
when W ⊂ X × Y is a basic open set. So write W = U × V where U and V are
open in X and Y respectively. Then
f −1 (W ) = {z ∈ Z : f (z) ∈ U × V } = f1−1 (U ) ∩ f2−1 (V )
which is open by hypothesis.

Q
5. The Product Topology on Xα
Fix topological spaces (Xα , τα ), α ∈ J, where J is a possibly infinite set.
5.1. Remark: The product topology on X × Y has two definitions:
(i) [Theorem 4.1]: The basis sets are of the form U × V where U ∈ τX , V ∈ τY
(ii) [Lemma 4.6]: It is the smallest topology that maps πX and πY continuous.
(End of Day 7)
Q
5.2. Theorem: Let (Xα , τα ) be a family of topological spaces, and let X = Xα . Let
πα : X → Xα be the projection map. Let B be the collection of finite intersections
of the form n
\
πα−1i (Ui )
i=1

for some finite set {α1 , α2 , . . . , αn } ⊂ J and open sets Ui ∈ ταi . Then there is
a unique topology τp on X which has B as a basis. This is called the product
topology on X.

20
 Proof. We once again check the conditions of Theorem 3.5:
(i) If x ∈ X then x ∈ Xα = πα−11 (Xα1 )
Q

(ii) If B1 := ni=1 πα−1i (Ui ) and B2 = m −1

T T
j=1 πβj (Vj ), then B1 ∩ B2 ∈ B

5.3. Lemma: Let {Xα } and X as above, and let τp denote the product topology.
(i) Each πα : (X, τp ) → (Xα , τα ) is continuous.
(ii) If η is a topology on X such that each πα : (X, η) → (Xα , τα ) is continuous,
then τp ⊂ η.
Proof. (i) If Uα ∈ τα , then πα−1 (Uα ) ∈ τp by definition.
(ii) If η is a topology as above, then for any α ∈ J, and Uα ∈ τα , πα−1 (Uα ) ∈ η.
By taking finite intersections, any basic open set in τp is in η. Hence, τp ⊂ η.

5.4. Theorem: Let f : Z → X be a function. Then f is continuous iff πα ◦ f is

continuous for each α ∈ J
Proof. One direction is clear from HW1 and the previous Lemma. For the other,
suppose πα ◦ f is continuous for each α ∈ J and WTS: f is continuous. Then by
3.3, it suffices to show that
f −1 (U ) ∈ τZ
for any basic open set U ⊂ X. Hence, we write U = ni=1 πα−1i (Ui ), whence
T

n
\
−1
f (U ) = (παi ◦ f )−1 (Ui ) ∈ τZ
i=1

Q
5.5. Theorem: Let (Xα ), τα ) be a family of topological spaces, and let X = Xα . Let
B be the collection of sets of the form
Y
Uα

where Uα ∈ τα for each α ∈ J. Then there is a unique topology τB on X which

has B as a basis. This is called the box topology on X.
Proof. Identical to Theorem 5.2.
5.6. Remark/Example:
(i) If J is finite, then the product and box topologies on X coincide.

21
 (ii) The basic open sets of τB are of the form
Y
Uα

where Uα ∈ τα are any open sets. However, the basic open sets in τp are of
the form Y
Uα
where Uα = Xα for all but finitely many α ∈ J
(iii) In general, τp ⊂ τB .
(iv) If J is infinite, they may not coincide. Example: In Rω ,
∞
Y
U := (−1/n, 1/n)
n=1

is open in the box topology, but not in the product topology.

Proof. Consider 0 ∈ U . If U ∈ τp , then there must be a basic open set B such
that 0 ∈ B and B ⊂ U . But if B is a basic open set, then ∃n1 , n2 , . . . , nk ∈ N
and open sets Ui ⊂ R such that
k
\
B= πi−1 (Ui ) = Un1 × Un2 × . . . Unk × R × R × . . .
i=1

Let n = max{ni : 1 ≤ i ≤ k} + 1, and y = (0, 0, 0, . . . , 1, 0, 0, . . .), where 1

occurs in the nth stage, then y ∈ B, but y ∈
/ U . Hence, B is not a subset of
U , so U ∈
/ τp .

6. Closed Sets
6.1. Definition: Let (X, τ ) be a topological space. A subset A ⊂ X is said to be closed
if X \ A is open.
6.2. Examples:
(i) [a, b] is closed in R
(ii) A = {(x, y) ∈ R2 : x ≥ 0, and y ≥ 0} is closed in R2 because R2 \ A =
R × (−∞, 0) ∪ (−∞, 0) × R
(iii) If τ is the discrete topology, then every subset of X is closed.
(iv) If τ is the co-finite (or finite complement) topology on R, then the only closed
sets are finite sets and R.
6.3. Lemma: Let X be a topological space. Then
(i) ∅ and X are closed in X

22
 T
(ii) If {Fα } are closed in X, then so is Fα
(iii) If F1 , F2 are closed in X, then so is F1 ∪ F2
6.4. Theorem: Let Y ⊂ X. A set A ⊂ Y is closed in Y (wrt the subspace topology) if
and only if ∃F ⊂ X closed in X such that A = F ∩ Y
Proof. HW
6.5. Corollary: Let Y ⊂ X. If A ⊂ Y is closed in Y , and Y is closed in X, then A is
closed in X.
(End of Day 8)
6.6. Definition: Let A ⊂ X
(i) The interior of A, int(A) is the union of all open sets contained in A.
(ii) The closure of A, A, is the intersection of all open sets containing A.
6.7. Remark:
(i) int(A) ⊂ A ⊂ A
(ii) A is open iff int(A) = A and A is closed iff A = A
(iii) int(A) is the largest open set contained in A. ie. If U ⊂ A is open in X, then
U ⊂ int(A).
(iv) Similarly, A is the smallest closed set containing A. If F ⊂ X is closed and
A ⊂ F , then A ⊂ F .
(v) If A ⊂ Y ⊂ X, we write clX (A) and clY (A) to denote the closures of A with
respect to X and Y respectively.
6.8. Lemma: Let A ⊂ Y ⊂ X. Then clY (A) = clX (A) ∩ Y
Proof. Note that
\
clY (A) = {F ⊂ Y : F closed, and A ⊂ F }

By Theorem 6.4,
\
clY (A) = {G ∩ Y : G ⊂ X closed in X, and A ⊂ G}

which is clearly clX (A) ∩ Y .

6.9. Theorem: Let A ⊂ X and x ∈ X.
(i) x ∈ A iff, for every open set U containing x, U ∩ A 6= ∅.
(ii) If the topology on X has a basis B, then x ∈ A iff, for every basic open set
B ∈ B, B ∩ A 6= ∅.
Note: An open set U containing a point x is called a neighbourhood of x.

23
 Proof. We only prove (i): If x ∈ A, let U be an open set containing x. If x ∈ A,
then U ∩ A 6= ∅ so there is nothing to prove. If x ∈
/ A, suppose U ∩ A = ∅. Then
F := X \ U is closed, and A ⊂ F . By Remark 6.7, A ⊂ F , so that A ∩ U = ∅,
whence x ∈
/ A. This is a contradiction.

Conversely, suppose every open set U containing x has the property that U ∩A 6= ∅.
WTS: x ∈ A. By definition,
\
A = {F : F ⊂ X closed, and A ⊂ F }

So choose F ⊂ X closed such that A ⊂ F . WTS: x ∈ F . Suppose x ∈ / F , then

x ∈ U := X \ F , which is open. Hence, U ∩ A 6= ∅. However, A ⊂ F , so this is
impossible. Hence, x ∈ F as required.
6.10. Corollary: Let (X, d) be a metric space and A ⊂ X. Then x ∈ A if and only if
there is a sequence (xn ) ⊂ A such that xn → x.
Proof. (i) Suppose there is a sequence (xn ) ⊂ A such that xn → x, then, for any
open set U containing x, ∃ > 0 such that B(x, ) ⊂ U . Then ∃N ∈ N such
that xn ∈ B(x, ) for all n ≥ N . Hence, U ∩ A 6= ∅, and so x ∈ A
(ii) Conversely, suppose x ∈ A. Fix n ∈ N and Un := B(x, 1/n). Then Un ∩A 6= ∅
so ∃xn ∈ A such that d(x, xn ) < 1/n. It follows that xn → x.

6.11. Definition: Let (X, τ ) be a topological space and A ⊂ X. A point x ∈ X is said to

be a limit point of A if, for every open set U containing x, U ∩ A contains a point
of A other than x. Equivalently,

x ∈ (A \ {x})

Write A0 for the set of limit points of A.

6.12. Examples:
(i) If A ⊂ R is a finite set, then A has no limit points. Similarly, Z ⊂ R has no
limit points.
(ii) Let τ be the co-finite topology on R, and A = Z, and let x ∈ R be any point.
If U is an open neighbourhood of x, then U ∩(Z\{x}) 6= ∅ because U contains
all but finitely many points of R. Hence, every point of R is a limit point of
Z
(iii) If A = [0, 1], then every point of A is a limit point of A.
(iv) If A = {1/n : n ∈ N}, then 0 is the only limit point of A.
Proof. If x ∈ A0 , then
(a) If x < 0, then U := (x − |x|/2, x + |x|/2) is a neighbourhood of x, and
U ∩ A = ∅. Hence, x ∈/ A0 .

24
 / A0 .
(b) If x > 1, then a similar argument shows that x ∈
(c) If 1 ≥ x > 0, and x ∈
/ A, then ∃N ∈ N such that
1 1
<x<
N +1 N
So if δ = min{1/N − x, x − N1+1 }, then U := (x − δ/2, x + δ/2) is an open
neighbourhood of x such that U ∩ A = ∅
(d) If 1 ≥ x > 0 and x ∈ A, then x = 1/N ofr some N ∈ N. Once again,
1 1
<x<
N +1 N −1
/ A0
so a similar argument shows that x ∈
(e) If x = 0, and U is an open set containing 0, then ∃δ > 0 such that
(−δ, +δ) ⊂ U . Choose N ∈ N such that 1/N < δ, so that 1/N ∈ U , so
that U ∩ (A \ {0}) 6= ∅. Hence, 0 ∈ A0 .

6.13. Theorem: A = A ∪ A0
Proof. (i) A ⊂ A ∪ A0 : Let F := A ∪ A0 and U := X \ F . We claim that U is
open. To see this, fix x ∈ X \ F . Then by definition, ∃ a neighbourhood V
of x such that V ∩ (A \ {x}) = ∅. Furthermore, x ∈ / A, so that V ∩ A = ∅.
Hence, V ⊂ U , so that U is open. Hence, F is closed, and since A ⊂ F , it
follows that A ⊂ F .
(ii) A ∪ A0 ⊂ A: If x ∈ A, then x ∈ A. Also, if x ∈ A0 , then x ∈ A by definition.
Hence, A ∪ A0 ⊂ A.

(End of Day 9)
6.14. Corollary: A set A is closed iff it contains all its limit points.
6.15. Example: Let X = Rω with the box topology, and

A := {(xn ) ∈ X : xn > 0 ∀n ∈ N}

and let 0 = (0, 0, . . .). Then

(i) 0 ∈ A: If U is any basic open set containing 0, then
Y
U= Un

where Un ⊂ R is open and contains 0. Hence, ∃xn ∈ Un such that xn > 0, so

that x := (xn ) ∈ A ∩ U . Hence, A ∩ U 6= ∅.

25
 (ii) Let xm = (xmn ) be a sequence in A. Then consider the diagonal
Q an := xnn > 0,
and the open set Un = (−an , an ) ⊂ R. Define U := Un , so that 0 ∈ U .
However, xm ∈ / U for all m ∈ N. Hence, there is no sequence in A that
converges to 0.
(iii) Hence, the box topology on Rω is not induced by a metric.
6.16. Definition: Let A ⊂ X
(i) A is said to be dense in X if A = X. Equivalently, U ∩ A 6= ∅ for any open
set U ⊂ X
(ii) X is said to be separable if it has a countable dense subset.
6.17. Examples:
(i) Q is dense in R, so R is separable.
Proof. If x ∈ R, δ > 0, then (x−δ, x+δ)∩Q 6= ∅. By Theorem 6.9, Q = R
(ii) If X, Y are topological spaces and A, B are dense in X and Y respectively.
Then A × B is dense in X × Y
Proof. If U ⊂ X and V ⊂ Y are open, then U ∩ A 6= ∅, V ∩ B 6= ∅. Hence,
(U × V ) ∩ (A × B) 6= ∅ as required.
(iii) Hence, Rn is separable because Qn is dense in it.
(iv) Rω is separable with respect to the product topology because

A = {(xn ) ∈ Rω : ∃N ∈ N such that xn = 0∀n ≥ N, xn ∈ Q}

is dense in Rω
Proof. Let
AN = {(xn ) : xn ∈ Q, xn = 0 ∀n ≥ N }
Then AN ∼ N −1
S
=Q , so AN is countable. Hence, A = Q AN is also countable.
Now if U is a basic open set in Rω , then write U = Un , where Un = R for
all n ≥ N . Then Ui ∩ Q 6= ∅ for all 1 ≤ i ≤ N , so choose xi ∈ Ui ∩ Q. Then

x = (x1 , x2 , . . . , xN , 0, 0, . . .)

is in U ∩ A. Hence, U ∩ A 6= ∅, so A = Rω
(v) Rω with the box topology is not separable.
Proof. Suppose A = {y n } is a countable subset of Rω , we show that A is not
dense. For each n ∈ N, write

y n = (y1n , y2n , . . . , ym
n
, . . .)

ynn ∈ R, so choose an open set Un ⊂ R such that ynn ∈

Now, Q / Un . Then
ω n
U := Un is open in R and has the property that y ∈ / U for all n ∈ N.
Hence, A ∩ U = ∅ as required.

26
6.18. Theorem: Let f : X → Y be a function. Then TFAE:
(i) f is continuous.
(ii) For every A ⊂ X, f (A) ⊂ f (A)
(iii) f −1 (B) is closed in X whenever B is closed in Y .
Proof. (i) (i) ⇒ (ii): Suppose f is continuous and y ∈ f (A), then WTS: y ∈
f (A). Write y = f (x) for some x ∈ A, and choose an open set U such that
y ∈ U . Then f −1 (U ) is an open neighbourhood of x. Hence, f −1 (U ) ∩ A 6= ∅,
so choose z ∈ f −1 (U ) ∩ A. Then f (z) ∈ U ∩ f (A). Hence U ∩ f (A) 6= ∅ so
that y ∈ f (A).
(ii) (ii) ⇒ (iii): Suppose B is closed, WTS: A := f −1 (B) is closed. We have
f (A) = f (f −1 (B)) ⊂ B so if x ∈ A, then

f (x) ∈ f (A) ⊂ f (A) ⊂ B = B

Hence, x ∈ f −1 (B) = A. Hence, A ⊂ A whence A = A is closed.

(iii) (iii) ⇒ (i): Take complements and apply the hypothesis.

(End of Day 10)

7. Continuous Functions
7.1. Definition: A function f : X → Y is called a
(i) open map if f (U ) is open whenever U ⊂ X is open.
(ii) homeomorphism if f is bijective, continuous, and f −1 : Y → X is also con-
tinuous. Equivalently, f is bijective, continuous and an open map.
7.2. Examples:
(i) f : R → R given by f (x) = 2x + 3 is a homeomorphism because g(y) :=
1
2
(y − 3) is the inverse.
(ii) Let f : (−1, 1) → R given by f (x) = x/(1 − x2 ). Then f is a homeomorphism
with inverse
2y
g(y) :=
1 + (1 + 4y 2 )1/2

(iii) Let Q = [−1, 1]2 ⊂ R2 and D = {(x, y) ∈ R2 : x2 + y 2 ≤ 1} be the square and

the disc in R2 . Define f : D → Q by f (0, 0) = (0, 0) and if (x, y) 6= (0, 0),
then p
x2 + y 2
f (x, y) = (x, y)
max{|x|, |y|}

27
 and g : Q → D by g(0, 0) = (0, 0) and if (x, y) 6= (0, 0), then

max{|x|, |y|}
g(x, y) = p (x, y)
x2 + y 2

(iv) Let f : [0, 1) → S 1 be f (t) = (cos(t), sin(t)). Then f is bijective and contin-
uous, but not a homeomorphism, because if U = [0, 1/4), then p := f (0) ∈
f (U ) is not an interior point of f (U ).
7.3. Theorem (Rules for constructing Continuous functions) Let X, Y, Z be topological
spaces.
(i) (Constant function): If f : X → Y maps X to a single point y0 ∈ Y , then f
is continuous.
(ii) (Inclusion): If Y ⊂ X has the subspace topology, then the inclusion map
ι : Y → X is continuous.
(iii) (Composites): If f : X → Y and g : Y → Z are continuous, then g ◦ f : X →
Z is continuous.
(iv) (Restricting the domain): If f : X → Y is continuous and A ⊂ X has the
subspace topology, then f |A : A → Y is continuous.
(v) (Restricting the range): Suppose f : X → Y is continuous, and A ⊂ Y has
the subspace topology. If f (X) ⊂ A, then the function g : X → A given by
f is continuous.
(vi) (Expanding the range): Suppose f : X → Y is continuous, and Y ⊂ Z has
the subspace topology, then f : X → Z is continuous.
Proof. (i) If U is an open set, then f −1 (U ) = X if y0 ∈ U and f −1 (U ) = ∅ if
/ Y . In either case, f −1 (U ) is open.
y0 ∈
(ii) If U ⊂ X is open, then ι−1 (U ) = U ∩ Y , which is open in Y by definition.
(iii) HW1.
(iv) f |A = f ◦ ι where ι : A → X is the inclusion map. So apply (iii).
(v) If U ⊂ A is open, then U = V ∩ A for some open set V ⊂ X. Then
g −1 (U ) = f −1 (V ) ∩ f −1 (A) = f −1 (V ) ∩ X = f −1 (V ), which is open in X.
(vi) If U ⊂ Z is open, then f −1 (U ) = f −1 (U ∩ Y ), which is open in X.

7.4. Theorem (Pasting Lemma):

S
(i) Let X = α∈J Uα where Uα is open, and let f : X → Y such that f |Uα : Uα →
Y is continuous for each α ∈ J. Then f : X → Y is continuous.

28
 (ii) Let X = A ∪ B where A and B are closed. Let f : A → Y and g : B → Y
be continuous functions such that f (x) = g(x) for all x ∈ A ∩ B. Then
h : X → Y given by (
f (x) : x ∈ A
h(x) =
g(x) : x ∈ B
is a well-defined continuous function from X to Y .
Proof. (i) If V ⊂ Y is open, then
[ [
f −1 (V ) = f −1 (V ) ∩ Uα = f |−1
Uα (V )
α∈J α∈J

(ii) If C ⊂ Y is a closed set, then [Check!]

h−1 (C) = f −1 (C) ∪ g −1 (C)

which is closed.

7.5. Example:
(i) Define h : R → R by (
0 :x≤0
h(x) =
x :x≥0
defines a continuous function.
(ii) Let f, g : X → R be continuous functions. Then

h1 (x) := min{f (x), g(x)} and h2 (x) := max{f (x), g(x)}

are continuous functions [HW]

(iii) (Part (ii) of the Pasting Lemma fails for infinitely many closed sets). Let
X = {1/n : n ∈ N} ∪ {0}, and A0 = {0}, Ai = {1/i} for i ∈ N. Define
fi : Ai → R by (
0 :i=0
fi =
1 : i 6= 0
Then each fi is continuous, and Ai ∩Aj = ∅ so they agree on the intersections.
However, the function f : X → R obtained by pasting them is not continuous.
(End of Day 11)
7.6. Example: (Stereographic Projection). Consider S 2 = {(x, y, z) ∈ R3 : x2 +y 2 +z 2 =
1}, and fix the north pole N = (0, 0, 1). We claim

S 2 \ {N } ∼
= R2

29
 Consider the plane passing through the equatorial circle. Fix P = (x, y, z) ∈ S 2 .
Draw a line from N through P , and let it meet the plane at the point Q := (u, v, 0).
Now taking ratios, we get
x u
=
y v
y
=v
1−z
x2 + y 2 + z 2 = 1

Solving, we get
2u x
x= 2 2
,u =
1+u +v 1−z
2v y
y= , v =
1 + u2 + v 2 1−z
1 − u2 − v 2
z=
1 + u2 + v 2
This gives a function

F : S 2 \ {N } → R2 and G : R2 → S 2 \ {N }

Note that the map

(u, v) 7→ 1 + u2 + v 2
is continuous from R2 → R \ {0} and

t 7→ 1/t

is continuous from R \ {0} to R. Hence, by composition (See Example 2.7), both

F and G are continuous, and inverses of each other. Hence, they are homeomor-
phisms.
7.7. Remark: The stereographic projection has the property that it preserves angles
(such a map is called a conformal map). This is the same property that the
mercator projection also has.

8. The Quotient Topology

8.1. Remark: Many spaces are constructed from other spaces by gluing, ie. by identi-
fying parts of the space to obtain another space.
(i) A cylinder is obtained from a rectangle by identifying one pair of opposite
edges.
(ii) The torus is obtained from a rectangle in R2 by identifying both pairs of
opposite edges.

30
 (iii) Consider X to be the union of two discs in R2 . If we identify the boundary
of one with the bounday of the other, we obtain the sphere S 2 .
8.2. Definition: Let X be a set.
(i) An equivalence relation on X is a subset R ⊂ X × X such that, for all
x, y, z ∈ X,
(a) (Reflexive): (x, x) ∈ R
(b) (Symmetric): (x, y) ∈ R ⇒ (y, x) ∈ R
(c) (Transitive): {(x, y), (y, z)} ⊂ R ⇒ (x, z) ∈ R
We write x ∼ y iff (x, y) ∈ R.
(ii) For x ∈ X, write
[x] := {y ∈ X : y ∼ x}
for the equivalence class of x. Note that [x] ∩ [y] = ∅ or [x] = [y]. Hence the
equivalence classes partition X.
(iii) Write X/ ∼= {[x] : x ∈ X} to be the set of equivalence classes of (X, ∼),
and let p : X → X ∗ be the map x 7→ [x].
8.3. Examples:
F
(i) If X = α∈J Aα is a partition of X. Write x ∼ y iff ∃α ∈ J such that
{x, y} ⊂ Aα . Then this is an equivalence relation whose equivalence classes
are precisely the Aα .
(ii) Let A ⊂ X. Define x ∼ y iff {x, y} ⊂ A. Then ∼ is an equivalence relation
whose equivalence classes are either A or singleton sets. In this case, we write
X/A := X/ ∼

(iii) If X = [0, 1], then define 0 ∼ 1 and x  y if {x, y} 6= {0, 1}. Then X/ ∼ can
be thought of as gluing the end-points of X.
(iv) If X = R, write x ∼ y iff x − y ∈ Z.
(v) If X = [0, 1]2 , write
(x, 0) ∼ (x, 1), for 0 ≤ x ≤ 1
(0, y) ∼ (1, y), for 0 ≤ y ≤ 1
This gives equivalence classes
[(x, y)] = {(x, y)} : 0 < x, y < 1
[(x, 0)] = {(x, 0), (x, 1)} : 0 < x < 1
[(0, y)] = {(0, y), (1, y)} : 0 < y < 1
[(0, 0)] = {(0, 0), (1, 0), (0, 1), (1, 1)}
ie. Opposite edges of the square are identified, and the vertices collapse to a
single point.

31
8.4. Lemma: Let X be a topological space, and Y any set. Suppose p : X → Y is a
function. Define
τY := {U ⊂ Y : p−1 (U ) ∈ τX }
Then
(i) τY is a topology on Y ,
(ii) p : X → Y is a continuous function.
(iii) If η is any topology on Y such that p : X → (Y, η) is continuous, then η ⊂ τY .
ie. τY is the largest topology that makes p continuous.
Proof. (i) To see that τY is a topology.
(a) ∅ = p−1 (∅) and X = p−1 (Y ), so ∅, Y ∈ τY
(b) If {Uα : α ∈ J} ⊂ τY , then
[ [
p−1 ( Uα ) = p−1 (Uα ) ∈ τX
S
so Uα ∈ τY .
(c) Similarly, τY is closed under finite intersection.
(ii) Obvious.
(iii) Suppose η is as above, then for any U ∈ η, p−1 (U ) ∈ τX , so U ∈ τY by
definition. Hence, η ⊂ τY .

8.5. Definition: Let X be a set and ∼ an equivalence relation of X. Let p : X → X/ ∼

be the map x 7→ [x]. The quotient topology on X/ ∼ is the topology induced by
p as in the above lemma. ie. A set U ⊂ X/ ∼ is open iff
[
[x]
[x]∈U

is open in X.
8.6. Examples:
(i) If X = [0, 1] with 0 ∼ 1. Then U = {[x] : 0 ≤ x < 1/4} is not an open set
because [
[x] = [0, 1/4) ∪ {1}
[x]∈U

whereas U = {[x] : 0 ≤ x < 1/4, or 3/4 < x ≤ 1} is an open set.

(End of Day 12)
2
(ii) Similarly, if X = [0, 1] with the relation in Example 8.3, then (draw picture
of open set bounded by an edge, and not having a counterpart on the opposite
edge)

32
8.7. (Universal Property of Quotient Spaces): Let X be a set with an equivalence
relation ∼, let X/ ∼ be given the quotient topology, and let p : X → X/ ∼ be the
natural map. Let Y be a topological space, and f : X → Y be a function such
that
x ∼ x0 ⇒ f (x) = f (x0 )
Then ∃ a unique function f : X/ ∼→ Y such that

f =f ◦p

Furthermore, f is continuous iff f is continuous.

Proof. (i) Given f : X → Y as above, define f : X ∗ → Y by

f ([x]) := f (x)

This is well-defined and satisfies f ◦ p = f . Furthermore, if g : X/ ∼→ Y is

any other function such that g ◦ p = f . Then g ◦ p = f ◦ p. But p is surjective,
so g = f , so f is unique.
(ii) Suppose f is continuous, then f = f ◦ p is continuous by 8.4. Conversely,
suppose f is continuous. WTS: f is continuous. So choose an open set U ⊂ Y ,
−1
then WTS: f (U ) ⊂ X/ ∼ is open. By definition, this is equivalent to asking
−1
if p−1 (f (U ) = (f ◦ p)−1 (U ) is open in X, which is true.

8.8. Example:
(i) Let X = [0, 1] with 0 ∼ 1, then X ∗ ∼
= S1
Proof. Define f : X → S 1 by f (x) = e2πix , then f is continuous, and f (0) =
f (1). Hence, we get a continuous function f : X/ ∼→ S 1 as above. We want
to construct an inverse g : S 1 → X/ ∼. Write

A1 = {z ∈ S 1 : Im(z) ≥ 0}, and A2 = {z ∈ S 1 : Im(z) ≤ 0}

Then A1 and A2 are closed sets and A1 ∩ A2 = {±1}. We now use the pasting
lemma. Given z ∈ A1 , ∃ unique t ∈ [0, 1/2] such that z = e2πit . Define
h1 : A1 → [0, 1] by h1 (z) = t. Similarly, if z ∈ A2 , ∃ unique t0 ∈ [1/2, 1] such
0
that z = e2πit , so define h2 (z) = t0 . Note that h1 and h2 are continuous, but
do not agree on A1 ∩ A2 because

h1 (1) = 0, but h2 (1) = 1

Now define gi : Ai → X/ ∼ by gi = p ◦ hi . Then gi are continuous (because

the hi are continuous), and they agree on A1 ∩ A2 . Hence by pasting lemma,
they define a continuous function g : S 1 → X/ ∼. Now note that

g ◦ f ([t]) = g(f (t)) = g(e2πit ) = [t]

33
 and similarly,
(f ◦ g)(z) = z ∀z ∈ S 1
Hence, f is a homeomorphism.
(ii) If X = R and x ∼ y iff x − y ∈ Z, then define f : R → S 1 by f (x) = e2πix .
As above, we get a homeomorphism R/ ∼∼ = S 1.
(iii) Similarly, if X = [0, 1]2 with the equivalence relation in 8.3(v), then X/ ∼∼
=
1 1
S × S . This is the torus.
(End of Day 13)
(iv) Let D2 = {(x, y) ∈ R2 : x2 + y 2 ≤ 1}. Then S 1 ⊂ D2 . We claim

D2 /S 1 ∼
= S2

Proof. (a) Write D2 = int(D2 ) t S 1 . Now define f1 : R2 → int(D2 ) by

1
f1 (x, y) = p (x, y)
x2 + y 2 + 1

Then f1 is a homeomorphism. Let f2 : R2 → S 2 \ {N } be the inverse of

the stereographic projection, so fb = f2 ◦ f1 : int(D2 ) → S 2 \ {N } is a
homeomorphism.
(b) Define f : D2 → S 2 by
(
fb(x) : x ∈ int(D2 )
f (x) =
N : x ∈ S1

We claim that f is continuous. It suffices to check continuity on S 1 ,

so fix x0 ∈ S 1 and an open set U ⊂ S 2 containing N = f (x0 ). Then
∃δ > 0 such that BR3 (N, δ) ∩ S 2 ⊂ U . By definition of the stereographic
projection, ∃R > 0 such that
p
x2 + y 2 > R ⇒ f2 (x, y) ∈ U

Hence, ∃0 < r < 1 such that

p
x2 + y 2 > r ⇒ fb(x, y) ∈ U

Hence, f −1 (U ) contains the set

V = {(x, y) ∈ D2 : x2 + y 2 > r2 }

which is open in D2 and contains x0

34
 (c) Thus, f is continuous. Clearly, x ∼ y if and only if f (x) = f (y), so by
8.7, f induces a map
f : D2 /S 1 → S 2
This map is both continuous and bijective. We will show later this is
enough to conclude that f is a homeomorphism.

8.9. Definition:
(i) Consider X
S n = {(x1 , x2 , . . . , xn+1 ) ∈ Rn+1 : x2i = 1}
Define x ∼ y iff y = −x (antipodal points are identified). Then we define

RP n := S n / ∼

This is called the real projective space.

(ii) Consider X = [0, 1]2 , and define ∼ by (0, y) ∼ (1, 1 − y). The quotient space
X/ ∼ is called the Mobius strip.
(iii) Let X = [0, 1]2 and define ∼ by (0, y) ∼ (1, 1 − y) and (x, 0) ∼ (x, 1). The
quotient space X/ ∼ is called the Klein bottle.

35
III. Properties of Topological Spaces
1. The Hausdorff property
1.1. Definition: A topological space X is said to be Hausdorff (T2 )if, for each x, y ∈ X
and distinct point, then ∃ open sets U, V such that x ∈ U, y ∈ V and U ∩ V = ∅.
1.2. Examples:
(i) Every metric space is Hausdorff.
Proof. If x, y ∈ X such that x 6= y, then δ := d(x, y) > 0, so let U = B(x, δ/2)
and V = B(y, δ/2)
(ii) If X is Hausdorff, and Y ⊂ X, then Y is Hausdorff.
Proof. If x, y ∈ Y are distinct, then ∃U, V ⊂ X open such that x ∈ U, y ∈ V
and U ∩ V = ∅. So let U 0 = U ∩ Y and V 0 = V ∩ Y .
(iii) If X and Y are Hausdorff, then so is X × Y .
Proof. If (x1 , y1 ) 6= (x2 , y2 ), then assume WLOG that x1 6= x2 , so ∃U, V ⊂ X
open such that U ∩V = ∅ and x1 ∈ U, x2 ∈ V . Now consider U 0 = U ×Y, V 0 =
V × Y . Then U 0 ∩ V 0 = ∅ and (x1 , y1 ) ∈ U 0 , (x2 , y2 ) ∈ V 0 .
Q
(iv) Similarly if each Xα is Hausdorff, then so is Xα in either the product or
the box topology.
(v) If X has the indiscrete topology, then it is not Hausdorff.
(vi) If R has the co-finite topology, then it is not Hausdorff.
Proof. Any two open sets must intersect non-trivially.
1.3. Definition: A topological space X is said to be T1 is singleton sets are closed in X.
Equivalently, if x 6= y are distinct points, then ∃ an open set U such that x ∈ U
and y ∈/ U.
1.4. Examples:
(i) If X is T2 , then it is T1
Proof. If x ∈ X, then WTS: X \ {x} is open. But if y ∈ X \ {x}, then by
the Hausdorff property, ∃V open such that y ∈ V and V ⊂ X \ {x}. Hence,
X \ {x} is open as required.
(ii) R with the co-finite topology is T1 but not T2

36
 Proof. If x ∈ R, then by definition, R\{x} is an open set, so {x} is closed.
(iii) If X has the indiscrete topology and |X| ≥ 2, then X is not T1
1.5. Theorem: Let X be Hausdorff, and (xn ) ⊂ X. Then (xn ) can converge to atmost
one point in X.
Proof. If xn → x, and x 6= y, then choose neighbourhoods U, V such that x ∈
U, y ∈ V and U ∩ V = ∅. Then ∃N ∈ N such that xn ∈ U for all n ≥ N . Hence,
at most finitely many xj may lie in V . Hence, (xn ) does not converge to y.
1.6. Example: Recall that if R has the co-finite topology, and xn = n, then for any
open set U ⊂ R, ∃N ∈ N such that xn ∈ U for all n ≥ N. Hence, xn → a for all
a ∈ R.
(End of Day 14)
1.7. Remark/Example:
(i) Let X be a topological space and X ∗ be a quotient space of X. Then a set
A ⊂ X ∗ is closed iff [
[x]
[x]∈A

is closed in X. Hence, X ∗ is T1 if and only each [x] is closed in X.

(ii) For example, all the spaces constructed in the previous section are T1 . How-
ever, if A = Q ⊂ R, then R/Q (the topological space) is not T1 because Q
is not closed in R. Hence, it is not true that if X is Hausdorff, then X ∗ is
Hausdorff.

2. Connectedness
2.1. Definition: Let X be a topological space.
(i) A separation of X is a pair {U, V } of non-empty open sets such that X = U ∪V
and U ∩ V = ∅.
(ii) A space X is said to be connected if it does not have a separation.
(iii) A set A ⊂ X is called cl-open if it is both closed an open.
2.2. Lemma: X is connected iff the only sets in X that are both open and closed are
∅ and X (ie. X has no non-trivial cl-open sets)
Proof. If X has a non-trivial cl-open set U , then V := X \U is cl-open, and {U, V }
is a separation of X. Conversely, if X is not connected, then it has a separation
{U, V } of disjoint non-empty sets. Then U is a non-trivial cl-open set.
2.3. Example:
(i) If X has the indiscrete topology, then X is connected.
(ii) If X has the discrete topology and |X| ≥ 2, then X is disconnected.

37
 (iii) R is connected.
(iv) Q ⊂ R is not connected.
2.4. Lemma: If A ⊂ X is connected, and A ⊂ B ⊂ A, then B is connected. In
particular, A is connected.
Proof. If A ⊂ B ⊂ A has a separation {U, V }, then U 0 := U ∩ A, V 0 := V ∩ A are
disjoint open subsets of A. Furthermore, U 0 6= ∅ because U ⊂ A is open (by II.6.9).
Similarly, V 0 6= ∅, so {U 0 , V 0 } is a separation of A. This is a contradiction.
2.5. Theorem: Any interval in R is connected. In particular, R is connected.
Proof. By the previous lemma, it suffices to consider closed intervals Y = [a, b].
Suppose {U, V } is a separation of Y , then U = U 0 ∩ Y, V = V 0 ∩ Y for some open
sets U 0 , V 0 ⊂ R. Assume WLOG that a ∈ U . Since U is open in Y, ∃δ > 0 such
that [a, a + δ) ⊂ U . Define

c := sup A, where A := {x ∈ [a, b] : [a, x] ⊂ U }

Note that c > a by the above argument. Claim: c ∈ U .

Proof. For each  > 0, c −  is not an upper bound for the set A, so ∃x ∈ A such
that
c−<x<c
Now [a, x] ⊂ U , so Hence, (c − , c + ) ∩ U 6= ∅. Hence, c ∈ clR (U ) by II.6.9. But
Y is closed in R, so c ∈ clY (U ) (II.6.8). But U is closed in Y , so c ∈ U .
Claim: c = b.
Proof. Suppose c < b, then since c ∈ U and U is open in Y, ∃δ > 0 such that
[c, c + δ) ⊂ U ∩ Y . Hence, [a, c + δ/2] ⊂ U , which contradicts the fact that
c = sup A. Hence, c = b.
Thus, [a, b] ⊂ U , so that V is empty.
2.6. Theorem: The only connected subsets of R are intervals.
Proof. Suppose Y ⊂ R is connected is not an interval. Then ∃a < c < b such that
{a, b} ⊂ Y and c ∈/ Y . Hence, U := (−∞, c) ∩ Y and V := (c, ∞) ∩ Y form a
separation of Y .
(End of Day 15)
2.7. Theorem: Let X be a topological space and {Aα : α ∈ J} be a collection of
connected sets such that \
Aα 6= ∅
S
Then A := Aα is connected.

38
 Proof. Let {U, V } be a separation of A, then for any β ∈ J, {U ∩ Aβ , V ∩ Aβ } are
two disjoint cl-open sets in Aβ . By 2.2, either U ∩ Aβ = Aβ or V ∩ Aβ = Aβ . ie.
either Aβ ⊂ U or Aβ ⊂ V . Let

J1 := {α ∈ J : Aα ⊂ U } and J2 = {α ∈ J : Aα ⊂ V }

Since {U, V } is a separation of A, it follows that J1 , J2 are both non-empty. How-

ever, if x ∈ ∩Aα , then x ∈ U ∩ V . This contradicts the fact that U ∩ V = ∅.
2.8. Theorem: Let X, Y be connected, then X × Y is connected.
Proof. Fix a ∈ X, b ∈ Y , then Ya := {a} × Y ∼
= Y is connected, and Xb := X × {b}
is connected. Furthermore, Xa ∩ Yb = {(a, b)} =
6 ∅. Hence, Xb ∪ Ya is connected by
the previous lemma. Now consider Ab := Xb ∪ Ya , b ∈ Y . Then Ab is connected,
and \
X ×Y = Ab = Ya 6= ∅
So by the previous theorem, X × Y is connected.
2.9. Example:
(i) Let X = Rω with the product topology, then X is connected.
Proof. Write

Xn = {(x1 , x2 , . . . , xn , 0, 0, . . .) : xi ∈ R} ⊂ X

Then Xn ∼
= RnT, so Xn is connected by the previous theorems and induction.
Furthermore, Xn = {0} = 6 ∅. Hence,
∞
[
A := Xn
n=1

is connected. We claim: X = A. Fix x = (xn ) ∈ X and an open set U

containing x. Then we may assume that
∞
Y
U := Un
n=1

where Un = R for all n ≥ N . Then for

y := (x1 , x2 , . . . , xN , 0, 0, . . .)

we have y ∈ A and y ∈ U , so U ∩ A 6= ∅. Hence, A = X, so X is connected

by 2.4
(ii) Let X = Rω with the box topology, then X is disconnected.

39
 Proof. Let

A := {(xn ) ∈ Rω : ∃M ∈ N such that |xn | ≤ M ∀n ∈ N}

be the set of all bounded sequences. Then A 6= ∅ and A 6= X. We claim that

A is cl-open, which would prove that Rω is disconnected.
• To see that A is open, fix x = (xn ) ∈ A, and consider
∞
Y
V := (xn − 1, xn + 1)
n=1

Then V is open, and if y = (yn ) ∈ V , then

|yn | < |xn | + 1

so (yn ) ∈ A.
• To see that A is closed, fix x = (xn ) ∈
/ A, and
∞
Y
V := (xn − 1, xn + 1)
n=1

If y = (yn ) ∈ V is bounded, then |xn | ≤ |yn | + 1 would imply that x ∈ A.

This is a contradiction, so V ⊂ X \ A. Hence, X \ A is open, so A is
closed.

2.10. Theorem: Let f : X → Y be a continuous function. If X is connected, then so is

f (X) (ie. the continuous image of a connected set is connected).
Proof. If f (X) has a separation {U, V }, then {f −1 (U ), f −1 (V )} would be open
sets, and
X = f −1 (f (X)) = f −1 (U ∪ V ) = f −1 (U ) ∪ f −1 (V )
and
f −1 (U ) ∩ f −1 (V ) = f −1 (U ∩ V ) = f −1 (∅) = ∅
so {f −1 (U ), f −1 (V )} would be a separation of X. Since X is connected, this cannot
happen.
2.11. Corollary: If X is connected, and ∼ and equivalence relation on X, then X/ ∼ is
connected.
2.12. (Intermediate Value Theorem): Let f : [a, b] → R be a continuous function and
d ∈ R such that f (a) < d < f (b). Then ∃c ∈ [a, b] such that f (c) = d.
Proof. By the previous theorems, f ([a, b]) is a connected subset of R, and is hence
an interval. In particular, f (a), f (b) ∈ f ([a, b]), so d ∈ f ([a, b]). This implies the
result.

40
2.13. Theorem: Rn ∼ = R iff n = 1
(It is true that Rn ∼
= Rm implies that n = m, but that is much harder to prove.)
Proof. Assume n > 1 and f : Rn → R is a homeomorphism. We will show that
Rn \ {0} is connected, so f (Rn \ {0}) = f (Rn ) \ {f (0)} must be connected. But
G
f (Rn \ {0}) = f (Rn ) \ {f (0)} = R \ {c} = (−∞, c) (c, ∞)

which is disconnected. This is a contradiction

(End of Day 16)

3. Path Connectedness
3.1. Definition: Let X be a topological space.
(i) A path between two points x, y ∈ X is a continuous function f : [0, 1] → X
such that f (0) = x, f (1) = y.
(ii) A space X is said to be path connected if any two points in X are connected
by a path.
3.2. Remark: Every interval [a, b] is homeomorphic to [0, 1] (via the map t 7→ at + (1 −
t)b), so we may as well write f : [a, b] → X is the above definition.
3.3. Theorem: A path connected space is connected.
Proof. If {U, V } is a separation for X, then choose x ∈ U, y ∈ V . By hypothesis,
there is path f : [0, 1] → X such that f (0) = x, f (1) = y. Consider U 0 := f −1 (U )
and V 0 := f −1 (V ). Then these are non-empty open sets and [0, 1] = f −1 (X) =
f −1 (U ) ∪ f −1 (V ), so [0, 1] must be disconnected. This contradicts 2.5.
3.4. Theorem: If f : X → Y is continuous, and X is path connected, then f (X) is
path connected.
Proof. Given u, v ∈ f (X), write u = f (x), v = f (y) for some x, y ∈ X. Let
g : [0, 1] → X be a path from x to y, then f ◦ g is path from u to v.
3.5. Corollary: If X is path connected, then any quotient space on X is path connected.
3.6. Definition: A set X ⊂ Rn is said to be convex if, for any x, y ∈ X and 0 ≤ t ≤ 1,
the point z := tx + (1 − t)y ∈ X.
3.7. Lemma: Any convex subset of Rn is path connected. In particular, Rn , and every
(closed or open) ball in Rn is path connected.
Proof. Consider the straight line path f : [0, 1] → X by f (t) := tx + (1 − t)y and
check that this is continuous.

41
 3.8. Let X be a topological space and {Aα : α ∈ J} be a collection of path connected
sets such that, for any two α, β ∈ J, ∃γ ∈ J such that

Aα ∩ Aγ 6= ∅ and Aβ ∩ Aγ 6= ∅
S
Then A := Aα is path connected.
Proof. Fix x, y ∈ A, then ∃α, β ∈ J such that x ∈ Aα , y ∈ Aβ . Let γ ∈ J as in
the hypothesis, and z1 ∈ Aα ∩ Aγ , z2 ∈ Aβ ∩ Aγ . Since Aα is path connected, ∃f1 :
[0, 1] → Aα continuous such that f1 (0) = x, f1 (1) = z1 . Similarly, ∃f2 : [1, 2] → Aγ
such that f2 (1) = z1 , f2 (2) = z2 , and ∃f3 : [2, 3] → Aβ such that f3 (2) = z2 and
f3 (3) = y. Define h : [0, 3] → A by

f1 (x) : 0 ≤ x ≤ 1

h(x) = f2 (x) : 1 ≤ x ≤ 2

f3 (x) : 2 ≤ x ≤ 3


Then h is continuous by pasting lemma and II.7.3, and h(0) = x, h(3) = y. So by

3.2, A is path connected.
3.9. Examples:
(i) If n > 1, then Rn \ {0} is path connected.
Proof. For each 1 ≤ i ≤ n, let

Ai := {x ∈ Rn : xi > 0}, and Bi := {x ∈ Rn : xi < 0}

Then Ai and Bi are convex (check!) and satisfy the hypotheses of 3.8. Hence,
[
Rn \ {0} = Ai ∪ Bi

is path connected.
(ii) S n ⊂ Rn+1 is path connected.
Proof. The map g : Rn+1 \ {0} → S n given by x 7→ x/d(x, 0) is a continuous
surjective map. So apply 3.4.
(iii) The following quotient spaces are all path connected: The Torus, The Mobius
strip, the Klein bottle, the real projective space.
Q
3.10. Theorem: If each Xα is path connected, then Xα is path connected with the
product topology.
Q
Proof. Given x = (xα ), y = (yα ) ∈ X := Xα , for each β ∈ J, there is a path
fβ : [0, 1] → Xβ such that f (0) = xβ and f (1) = yβ . Define f : [0, 1] → X by
f (t) = (fα (t)), then f is continuous because each component of f is continuous.
And clearly f (0) = x, f (1) = y, so X is path connected.

42
3.11. Remark: Note that the above result is not true with the box topology: Rω is not
connected with the box topology, so cannot be path connected. (See Example 2.9)
(End of Day 17)
3.12. Example (The Topologists’ Sine Curve): Define

S := {(x, sin(1/x)) : 0 < x ≤ 1} ⊂ R2

and let X = S. Then note that

X = S ∪ {0} × [−1, 1]

Then X is connected, but not path connected.

Proof. The map f : (0, 1] → S given by x 7→ (x, sin(1/x)) is continuous, and (0, 1]
is connected. Hence, S is connected (Note: In fact, S is path connected). By
Lemma 2.4, X is connected. We claim there is no path from (0, 0) to any point of
S. Suppose f : [0, 1] → X is such a path, consider

A = {t ∈ [0, 1] : f (t) ∈ {0} × [−1, 1]}

and let a := sup(A). By hypothesis, a < 1. Consider f |[a,1] : [a, 1] → X and

write f (t) = (x(t), y(t)). Then x(0) = 0 and x(t) > 0 for all t > a, so that
y(t) = sin(1/x(t)) for all t > a. We claim: ∃(tn ) ⊂ [a, 1] such that tn → a and
y(tn ) = (−1)n .

For n ∈ N fixed, choose 0 < u < x(a + 1/n) such that sin(1/u) = (−1)n . By the
intermediate value theorem, ∃a < tn < a + 1/n such that f (tn ) = (tn , (−1)n ). This
proves the claim.
Hence, tn → 0 and f (tn ) = (tn , (−1)n ) does not converge. Hence, f is not contin-
uous.
3.13. Remark:
(i) The above example also shows that even if A is path connected, then A may
not be path connected (compare with 2.4)
(ii) There are two other examples similar to the topologists’ sine curve:
(a) The deleted infinite broom: For n ∈ N, let Ln denote the line segment in
R2 connecting (0, 0) to (1, 1/n). Let
∞
[
S := Ln , and X := S \ {(0, 1)}
n=1

Then S is called the infinite broom, and X the deleted infinite broom.
Once again, X is connected, but not path connected.

43
 (b) The deleted comb space: Define
∞
[
D := ([0, 1] × {0}) ∪ ({1/n} × [0, 1]) ∪ [0, 1]
n=1

and X := D\{(0, 1)}. Then D is called the comb space, and X the deleted
comb space. Once again, X is connected, but not path connected.

(End of Day 18)

4. Local Connectedness
4.1. Definition: Let X be a topological space. Write x ∼ y if there is a connected
subspace A ⊂ X such that {x, y} ⊂ A.
4.2. Lemma: The above relation is an equivalence relation, and the equivalence classes
are the maximal connected subsets of X (ie. if C is an equivalence class, and B
is a connected set such that C ⊂ B, then C = B). These equivalence classes are
called the connected components of X.
Proof. That this is an equivalence class is easy to see. For any x ∈ X,

[x] = {y ∈ X : x ∼ y}
= {y ∈ X : ∃Ay connected, such that {x, y} ⊂ Ay }
[
= Ay
y∈[x]

T
Each Ay is connected, and Ay ⊃ {x} = 6 ∅, so by 2.7, [x] is connected. Further-
more, if B is a connected set such that [x] ⊂ B, and y ∈ B, then {x, y} ⊂ B, so
by definition, y ∈ [x]. Hence, [x] is maximal as well.
4.3. Definition: Let X be a topological space. Write x ∼h y if there is a path f :
[0, 1] → X such that f (0) = x, f (1) = y.
4.4. Lemma: The above relation is an equivalence relation, and the equivalence classes
are the maximal path connected subsets of X. These are called the path compo-
nents of X.
Proof. To show that ∼h is an equivalence relation:
(i) x ∼ x: Consider the constant path
(ii) x ∼ y ⇒ y ∼ x: If f : [0, 1] → X is such that f (0) = x, f (1) = y, take
g(s) := f (1 − s), then g is continuous, g(0) = y, g(1) = y.
(iii) If x ∼ y, y ∼ z: To show that x ∼ z, simply use the pasting lemma as in 3.8
to join the two paths.

44
 That the equivalence classes are path connected, and maximal is exactly as in
4.2.
4.5. Examples:
(i) If X is connected, it has only one component.
(ii) If X = Q, then the connected components are singletons.
Proof. If A ⊂ X has at least two points, then ∃a, b ∈ A and x ∈ R \ Q such
that a < x < b. Hence, U := (−∞, x) ∩ A and V := (x, ∞) ∩ A forms a
separation of A, so A is disconnected. Hence, the only connected sets are
singletons.
4.6. Definition.
(i) A topological space X is said to be locally connected if, for each x ∈ X and
each open set U 3 x, ∃ an open neighbourhood V ⊂ U of x that is connected.
(ii) We define locally path connected similarly.
(End of Day 19)
4.7. Examples:
(i) Locally path connected implies locally connected.
(ii) A = (0, 1) t (2, 3) is locally (path) connected, but not connected.
(iii) If A = {0} ∪ {1/n : n ∈ N} ⊂ R, then A is not locally connected because, for
any 1 > δ > 0, B(0, δ) ∩ A is a finite set, and hence disconnected.
(iv) However, connected does not imply local connectedness: Consider the topol-
ogists’ sine curve X from 3.12, and x = (0, 1) ∈ X. Fix δ < 1 and consider
U = B(x, δ) ∩ X. Then U is a disjoint union of infinitely many line segments
U = tLn . Each such Ln is a cl-open set in U , so U is disconnected.
(v) Similarly, path connectedness does not imply local path connectedness: De-
fine
∞   
[ 1
X= , y) : y ∈ R ∪ {(0, y) : y ∈ R} ∪ {(x, 0) : x ∈ R}
n=1
n

Then X is clearly path connected, but if x = (0, 1) ∈ X, and δ < 1, then

U = B(x, δ) ∩ X is once again a disjoint union of line segments. Hence, U is
not path connected either.
4.8. Lemma:
(i) If X is locally connected, then components are open sets. Hence each com-
ponent is cl-open.
(ii) If X is locally path connected, then each path component is open in X.
Hence, each path component is cl-open.

45
 Proof. We prove (i), because (ii) is identical: If C is a component of x and x ∈ C,
then ∃ a connected neighbourhood U of x. It follows that U ⊂ C, so C is open.
Now if each component is open, and X is a disjoint union of components, then
each component must also be closed.
4.9. Theorem: Let X be a topological space.
(i) Every path component is contained in a connected component of X.
(ii) If X is locally path connected, then the components and path components
coincide.
Proof. (i) is obvious, so we prove (ii): Let P be a path component, and x ∈ P ,
then P ⊂ Cx , the connected component of x. Also, P is a cl-open set in X, so P
is cl-open in Cx . Since Cx is connected, it follows that P = Cx .
4.10. Corollary: If X is connected and locally path connected, then it is path connected.
4.11. Examples:
(i) If X ⊂ Rn is open, then it is locally path connected.
Proof. Let x ∈ X, then ∃ a n-cell V := ni=1 (ai , bi ) ⊂ X such that x ∈ V .
Q
But each (ai , bi ) ⊂ R is path connected by 3.7, so V is path connected by
3.10.
(ii) More generally, if X is locally connected, and Y ⊂ X is open, then Y is
locally connected.

(End of Day 20)

5. Compactness
5.1. Remark: Consider some nice properties of the interval [0, 1]:
(i) If f : [0, 1] → R is continuous, then f is bounded.
(ii) If f : [0, 1] → R is continuous, then it is uniformly continuous. ie. For all
 > 0, ∃δ > 0 such that |x − y| < δ implies |f (x) − f (y)| < .
(iii) Every sequence in [0, 1] has a convergent subsequence.
Note that these properties are also shared by other sets, for instance, finite
sets. Compactness is a generalization of finiteness in the context of topological
spaces.
(iv) Example: If f : (0, 1) → R is given by f (x) = 1/x, then f is not uniformly
continuous, and is not bounded. ie. [0, 1] should be compact, but (0, 1) should
not.
5.2. Definition: Let X be a topological space.

46
 (i) A collection U of subsets of X is called an open cover for X if every member
of U is open, and, for each x ∈ X, ∃U ∈ U such that x ∈ U .
(ii) Let U and V be open covers of X. We say V is a subcover of U if V ⊂ U.
5.3. Examples:
(i) {X} is an open cover for X. Similarly, the topology τ (or any basis of τ ) is
an open cover for X.
(ii) If U is an open cover for X, and W ⊂ τ is any collection of open sets, then
U ∪ W is an open cover, and U is a subcover of U ∪ W.
(iii) If X is a metric space. For each x ∈ X, choose δx > 0. Then U := {B(x, δx ) :
x ∈ X} is an open cover for X.
(iv) If U is an open cover for X, and V is an open cover for Y , then W := {U ×V :
U ∈ U, V ∈ V} is an open cover for X × Y .
(v) If U is an open cover for X, and X ∗ is any quotient space of X, then V :=
{π(U ) : U ∈ U} is an open cover for X ∗ (where π : X → X ∗ denotes the
quotient map).
5.4. Definition: A topological space X is said to be compact if, whenever U is an open
cover for X, ∃ finitely many elements V := {U1 , U2 , . . . , Un } ⊂ U such that V is an
open cover for X. ie. Every open cover of X has a finite subcover.
5.5. Examples:
(i) Any finite set is compact.
Proof. If U is an open cover for X, then U ⊂ P(X), which is itself finite.
Hence, U is finite.
(ii) (0, 1) is not compact.
Proof. Let Un := (1/n, 1), then {Un } is an open cover without a finite sub-
cover.
5.6. Theorem: [0, 1] ⊂ R is compact.
Proof. Let U be an open cover for [0, 1]. Since 0 ∈ [0, 1], ∃U ∈ U such that 0 ∈ U .
Hence, ∃δ > 0 such that [0, δ) ⊂ U . Now define

A := {x ∈ [0, 1] : [0, x] is contained in finitely many elements of U}

Then, by the above argument, δ/2 ∈ A. So define

c := sup(A)

We claim that c = 1. If c < 1, then c ∈ [0, 1], so ∃V ∈ U such that c ∈ V . Hence,

∃δ > 0 such that (c − δ, c + δ) ⊂ V . Since c = sup(A), c − δ is not an upper bound
for A. Hence, ∃x ∈ A such that

c−δ <x≤c

47
 Now, [a, x] is covered by finitely many members of U, say {U1 , U2 , . . . , Uk }. Also,
[x, c+δ/2] ⊂ (c−δ, c+δ) ⊂ V . Hence, [a, c+δ/2] is covered by {U1 , U2 , . . . , Uk , V }.
In particular,
c + δ/2 ∈ A
contradicting the fact that c = sup(A). Thus, c = 1, and the proof is complete.
5.7. Theorem: A closed subspace of a compact space is compact.
Proof. Let Y ⊂ X be a closed and X compact. Let U be an open cover for Y .
Then for each V ∈ U, ∃V 0 ⊂ X open such that V = V 0 ∩ Y . Consider
[
U 0 := {V 0 : V ∈ U} {X \ Y }

This is an open cover for X, so has a finite subcover V ⊂ U 0 . Consider

{W ∩ Y : W ∈ V}

then this is a cover of Y that is finite, and a subcover of U [Check!]

5.8. (The tube lemma): Let X, Y be topological spaces with Y compact. Let x0 ∈ X,
and suppose N ⊂ X × Y is open such that

x0 × Y ⊂ N

Then ∃W ⊂ X open such that x0 ∈ W and

W ×Y ⊂N

Note: A set of the form W × Y is called a tube about x0 × Y

Proof. For each (x0 , y) ∈ x0 × Y , choose a basic open set Uy × Vy such that
(x0 , y) ∈ Uy × Vy and
Uy × Vy ⊂ N
The collection {Uy × Vy : y ∈ Y } forms an open cover for x0 × Y ∼
= Y . Hence, it
has a finite subcover

{U1 × V1 , U2 × V2 , . . . , Un × Vn }

Consider W := U1 ∩ U2 ∩ . . . ∩ Un , then if x ∈ W and y ∈ Y , then ∃1 ≤ i ≤ n such

that (x0 , y) ∈ Ui × Vi ⊂ N . Hence, (x, y) ∈ Ui × Vi , so

(x, y) ∈ N

So W × Y ⊂ N
(End of Day 21)
5.9. Theorem: The finite product of compact spaces is compact.

48
 Proof. By induction, we prove it for two spaces, so let X, Y be compact, and let
U = {Uα } be an open cover for X × Y . Fix x0 ∈ X, then U is an open cover for
x0 × Y . Since x0 × Y ∼
= Y is compact, it has a finite subcover {U1 , U2 , . . . , Un }.
Let
N := U1 ∪ U2 ∪ . . . ∪ Un
then N is an open set containing x0 × Y . Let W ⊂ X be an open set such that

W ×Y ⊂N

as in the previous lemma. Then W × Y is covered by finitely many sets of U,

namely {U1 , U2 , . . . , Un }.

Hence, for each x ∈ X, there is an open neighbourhood Wx of x such that Wx × Y

is covered by finitely many elements of U. Now the collection {Wx : x ∈ X} forms
an open cover for X, so has a finite subcover {W1 , W2 , . . . , Wn }. Now each Wi × Y
is covered by finitely many elements of U, so
n
[
Wi × Y
i=1

is covered by finitely many elements of U. But

n
[
X ×Y ⊂ Wi × Y
i=1

so this completes the proof.

5.10. Definition: A collection C of subsets of X is said to have the finite intersection
property if, for each finite subcollection {C1 , C2 , . . . , Cn } ⊂ C, the intersection

C1 ∩ C2 ∩ . . . ∩ Cn

is non-empty.
5.11. Theorem: Let X be a topological space, then X is compact iff, for every collection
C of closed sets with the finite intersection property,
\
C 6= ∅
C∈C

Proof. Define U by
U := {X \ C : C ∈ C}
Then
(i) U is a collection of open sets.

49
 (ii) U is an open cover for X if and only if
\
C=∅
C∈C

(iii) A finite subcollection {U1 , U2 , . . . , Un } of U covers X if and only if, the cor-
responding subcollection Ci := X \ Ui has the property that

C1 ∩ C2 ∩ . . . ∩ Cn = ∅

Now suppose X is compact: If C has the finite intersection property and

\
C=∅
C∈C

then U is a cover for X. By compactness, it must have a finite subcover. By (iii),

this would violate the finite intersection property.

The converse is similar.

5.12. Corollary: Let X be a compact topological space. Let {Ci } be a sequence of
non-empty closed subsets of X such that

C1 ⊃ C2 ⊃ . . . ⊃ Ci ⊃ Ci+1 ⊃ . . .

(Such a sequence is called a nested sequence of closed sets.) Then

\
Cn 6= ∅
n∈N

6. Compact Subsets of Rn
6.1. Example: Fix real numbers ai < bi for 1 ≤ i ≤ n, then
n
Y
X := [ai , bi ]
i=1

is compact in Rn . Such a set is called a n-cell.

Proof. Any set of the form [a, b] ⊂ R is homeomorphic to [0, 1] via the map

t 7→ tb + (1 − t)a

Hence, [a, b] is compact. Hence, X is compact by Theorem 5.9.

50
6.2. Definition: Let X be a metric space and Y ⊂ X. Y is said to be bounded if
∃M > 0 such that
d(x, y) ≤ M ∀x, y ∈ Y
By the triangle inequality, this is equivalent to: ∃x0 ∈ X and M 0 > 0 such that

d(x0 , y) ≤ M 0 ∀y ∈ Y

6.3. Lemma: Let X be a metric space and Y ⊂ X be a compact set, then Y is bounded.
Proof. Fix x0 ∈ Y . Then consider

U := {B(x0 , r) ∩ Y : r > 0}

If y ∈ Y , then ∃r > 0 such that d(x0 , y) < r, so U is an open cover for Y . Hence
it has a finite subcover {B(x0 , r1 ) ∩ Y, . . . , B(x0 , rn ) ∩ Y }. Let

M := max{ri : 1 ≤ i ≤ n} > 0

Then for any y ∈ Y, ∃1 ≤ i ≤ n such that y ∈ B(x0 , ri ) ∩ Y , so d(x0 , y) < ri ≤ M .

Hence, Y is bounded.
Recall: Let X be a set. Two metrics d1 and d2 on X are said to be equivalent if
∃K, M > 0 such that

Kd1 (x, y) ≤ d2 (x, y) ≤ M d1 (x, y) ∀x, y ∈ X

Note: If a set Y ⊂ X is bounded with respect to d1 , then it is bounded with

respect to d2 and vice versa.
6.4. Lemma: Let X be a Hausdorff space and Y ⊂ X compact, then Y is closed.
Proof. If x ∈
/ Y , then for each y ∈ Y, ∃ open sets Uy and Vy such that x ∈ Uy , y ∈ Vy
and Uy ∩ Vy = ∅. Now {Vy : y ∈ Y } is an open cover of Y , which must have a
finite subcover {Vy1 , Vy2 , . . . , Vyn }. Set
n
\
U := Uyi
i=1

Then U is open, x ∈ U , and U ∩ Vyi = ∅ for all i. Hence, U ∩ Y = ∅, so U ⊂ Y c ,

whence Y c is open.
6.5. (Heine-Borel Theorem): Let X ⊂ Rn , then X is compact if and only if X is both
closed and bounded (wrt the Euclidean metric).
Proof. If X is compact, X is closed and bounded by the previous two lemmas. If
X is closed and bounded, and is non-empty, fix x0 ∈ X, then

X − x0 := {a − x0 : a ∈ X}

51
 is homeomorphic to X and contains 0. To show that X is compact, it suffices to
show that X − x0 is compact, so we may assume WLOG that 0 ∈ X. Since X is
bounded with respect to the Euclidean metric, it is bounded with respect to the
sup-metric because they are equivalent (Example II.2.14). Hence, ∃M > 0 such
that
max{|yi | : 1 ≤ i ≤ n} = d∞ (0, y) ≤ M ∀y ∈ X
Hence, if y ∈ X, then |yi | ≤ M for all 1 ≤ i ≤ n. ie. X is contained in the set
n
Y
Z := [−M, M ]
i=1

Now Z is compact because it is an n-cell. Since X ⊂ Z and X is closed in Rn , X

is closed in Z (Why?). Hence X is compact by 5.8.
6.6. Example: Let X = Z with the discrete metric
(
1 : x 6= y
d(x, y) =
0 :x=y

Then X is closed and bounded, but not compact. Hence, the above theorem does
not hold for all metric spaces.
(End of Day 22)

7. Continuous Functions on Compact Sets

7.1. Theorem: Let f : X → Y be a continuous function, and X compact. Then f (X)
is compact.
Proof. If U is an open cover for f (X), then

V := {f −1 (U ) : U ∈ U}

is an open cover for X [Check!]. Let {f −1 (U1 ), f −1 (U2 ), . . . , f −1 (Un )} be a finite

subcover of V, then {U1 , U2 , . . . , Un } is a finite subcover of f (X) [Check!].
7.2. Theorem: If X is compact and X ∗ is a quotient space of X, then X ∗ is compact.
Proof. The quotient map π : X → X ∗ is surjective and continuous, so the previous
theorem applies.
7.3. Definition: Let f : X → R be a function.
(i) We say that f is bounded below if ∃m ∈ R such that f (x) ≥ m for all x ∈ X.
(ii) Similarly, we define f to be bounded above.
(iii) If f is bounded below, we say that f attains its infimum at a point x0 ∈ X if

f (x0 ) ≤ f (x) ∀x ∈ X

52
 (iv) We say that f attains its supremum at x1 if

f (x) ≤ f (x1 ) ∀x ∈ X

The points x0 and x1 (if they exist, and they need not be unique) are called
extreme points of f .
7.4. Example:
(i) Let f : (0, 1) → R be given by f (x) = 1/x, then f is not bounded above.
(ii) Let f : R → R by f (x) = e−x , then f is bounded below, but it does not
attain its infimum 0.
7.5. (Extreme Value Theorem): Let X be compact and f : X → R continuous, then
∃x0 , x1 ∈ X such that

f (x0 ) ≤ f (x) ≤ f (x1 ) ∀x ∈ X

Proof. Since f (X) is compact, by the Heine-Borel theorem, it is closed and bounded.
In particular,
m := inf{f (x) : x ∈ X}
exists and is finite. m is a limit point of f (X) and f (X) is closed, so m ∈
f (X). Hence, ∃x0 ∈ X such that f (x0 ) = m. The proof for the upper bound is
analogous.
7.6. Theorem: Let f : X → Y be a continuous, bijective function. If X is compact,
and Y is Hausdorff, then f is a homeomorphism.
Proof. We want to show that f is an open map. It suffices to show that f is a
closed map. If F ⊂ X is closed, then F is compact. Hence, f (F ) is compact in Y ,
so f (F ) is closed in Y .
7.7. Example:
(i) This completes the proof from Example II.8.8,

D2 /S 1 ∼
= S2

(ii) In the Mid-Sem Exam (Q. 5), we had

p
A := {(x, y) : 1 ≤ x2 + y 2 ≤ 2}

and we had constructed a continuous bijective function f : S 1 × [1, 2] → A.

Note that S 1 ×[1, 2] is compact and A is Hausdorff, so f is a homeomorphism.
7.8. Definition: Let (X, d) be a metric space and A ⊂ X. Given x ∈ X, define the
distance of x from A as

d(x, A) := inf{d(x, y) : y ∈ A}

53
 7.9. Lemma: The function p : X → R given by p(x) := d(x, A) is a continuous function.
Furthermore, p(x) = 0 if and only if x ∈ A
Proof. (i) If x1 , x2 ∈ X, y ∈ A

d(x1 , A) ≤ d(x1 , y) ≤ d(x1 , x2 ) + d(x2 , y)

This is true for all y ∈ A, so

d(x1 , A) ≤ d(x1 , x2 ) + d(x2 , A)

so
d(x1 , A) − d(x2 , A) ≤ d(x1 , x2 )
By symmetry, d(x2 , A) − d(x1 , A) ≤ d(x1 , x2 ) so

|d(x1 , A) − d(x2 , A)| ≤ d(x1 , x2 )

From this continuity follows [Why?]

(ii) Suppose x ∈ A, then ∃yn ∈ A such that d(x, yn ) → 0. Hence, d(x, A) = 0.
Conversely, if d(x, A) = 0, then for each n ∈ N, 1/n is not a lower bound for
the set
{d(x, y) : y ∈ A}
So ∃yn ∈ A such that d(x, yn ) < 1/n. Clearly, yn → x, so x ∈ A

7.10. Definition: Let (X, d) be a metric space and A ⊂ X. The diameter of A is defined
as
diam(A) := sup{d(x, y) : x, y ∈ A}

7.11. (Lebesgue Number Lemma): Let U be an open cover of a metric space (X, d). If
X is compact, ∃δ > 0 such that if A ⊂ X such that diam(A) < δ, then ∃U ∈ U
such that A ⊂ U .
Any number δ as above is called a Lebesgue number for the cover U. Note if δ is
a Lebesgue number for U and δ 0 < δ, then δ 0 is also a Lebesgue number for U
Proof. Let {U1 , U2 , . . . , Un } be a finite subcover of U and define Ai := X \ Ui .
Define f : X → R by
n
1X
f (x) = d(x, Ai )
n i=1
Then f is continuous by the previous lemma, so it must attain its minimum at
some point x ∈ X. Now, ∃Ui such that x ∈ Ui , so x ∈ / Ai so by the previous
lemma, d(x, Ai ) > 0, whence f (x) > 0, so if δ := f (x), then

f (y) ≥ δ ∀y ∈ X

54
 Now if A is a set of diameter less than δ, then fix x0 ∈ A, then

A ⊂ B(x0 , δ)

Now, assume that d(x0 , Am ) is the maximum of {d(x0 , A1 ), d(x0 , A2 ), . . . , d(x0 , An )}.
Then
δ ≤ f (x0 ) ≤ d(x0 , Cm )
Hence, for each y ∈ Cm , d(x0 , y) ≥ δ, whence

B(x0 , δ) ⊂ X \ Cm = Um ⇒ A ⊂ Um

(End of Day 23)

7.12. Definition: Let f : X → Y be a continuous function between two metric spaces.
We say that f is uniformly continuous if, for each  > 0, ∃δ > 0 such that

dX (x1 , x2 ) < δ ⇒ dY (f (x1 ), f (x2 )) < 

7.13. Example: Let f : (0, 1) → R given by f (x) = 1/x, then f is not uniformly
continuous.
7.14. Theorem: Let f : X → Y be a continuous function between metric spaces. If X
is compact, then f is uniformly continuous.
Proof. Consider  > 0 and set

V := {B(y, /2) : y ∈ Y }

Then V is an open cover for Y , so

U := {f −1 (B(y, /2)) : y ∈ Y }

is an open cover for X. Let δ > 0 be a Lebesgue number for U. Then if x1 , x2 ∈ X

such that dX (x1 , x2 ) < δ, then A := {x1 , x2 } has diameter < δ, so ∃y ∈ Y such
that
A ⊂ f −1 (B(y, /2))
Hence, {f (x1 ), f (x2 )} ⊂ B(y, /2) so by the triangle inequality,

dY (f (x1 ), f (x2 )) < 

55
8. Compactness in Metric Spaces
8.1. Definition: Let X be a topological space.
(i) X is said to be sequentially compact if, for any sequence (xn ) ⊂ X, there is
a subsequence (xnk ) of (xn ) that converges to a point in X.
(ii) Recall: If A ⊂ X. A point x ∈ X is called a limit point of A if, for each open
set U containing x, U ∩ (A \ {x}) 6= ∅
(iii) X is said to be limit point compact if every infinite subset of X has a limit
point in X.
8.2. Lemma: If X is compact, then it is limit point compact.
Proof. Let A ⊂ X be an infinite set, and suppose A has no limit point. Then, for
each x ∈ X, there is an open set Ux containing x such that Ux ∩ (A \ {x}) = ∅.
Then, U := {Ux : x ∈ X} is an open cover for X which has a finite subcover
{Ux1 , Ux2 , . . . , Uxn }. Then each Ux1 contains atmost one point of A (possibly xi ).
Hence A is finite.
8.3. Example: Let Y = {1, 2} with the indiscrete topology τY = {∅, Y }, and let

X := N × Y

with the product topology, where N is given the usual discrete topology. Then X
is limit point compact but not compact.
Proof. If A ⊂ X is any non-empty set, and assume that (n, 1) ∈ A. If U is an open
set containing (n, 2), then U contains a basic open neighbourhood W = {n} × Y ,
so
(n, 1) ∈ W ∩ (A \ {(n, 2)})
whence U ∩ (A \ {(n, 1)}) 6= ∅.

However, the open cover {{n} × Y : n ∈ N} does not have a finite subcover, so X
is not compact.
8.4. Lemma: Let X be Hausdorff, A ⊂ X and x ∈ X a limit point of A. Then for any
open neighbourhood U of x, U ∩ (A \ {x}) is infinite.
Proof. Suppose U ∩(A\{x}) is finite, then write U ∩A = {a1 , a2 , . . . , an }. For each
i, there are open sets Vi , Wi such that x ∈ Vi and ai ∈ Wi such that Vi ∩ Wi = ∅.
If n
\
V := Vi
i=1

Then V is an open set containing x and V ∩ (A \ {x}) = ∅, so x cannot be a limit

point of A.

56
8.5. Definition: A metric space X is said to be totally bounded if, for each  > 0, there
are finitely many points {x1 , x2 , . . . , xn } ⊂ X such that

{B(xi , ) : 1 ≤ i ≤ n}

covers X. Such a collection of open set is called an -net of X.

8.6. Lemma: If X is sequentially compact, then it is totally bounded.
Proof. Suppose X is not totally bounded, then ∃ > 0 for which there is no finite
epsilon net. In particular, if x1 ∈ X, then X 6= B(x1 , ), so ∃x2 ∈ X such that

d(x1 , x2 ) ≥ 

Now, {B(x1 , ), B(x2 , )} is not an open cover for X, so ∃x3 ∈ X such that

d(x3 , x1 ) and d(x3 , x2 ) ≥ 

Thus proceeding, we obtain a sequence (xn ) ⊂ X such that if m > n, then

d(xm , xn ) ≥ 

Such a sequence cannot have a convergent subsequence [Why?] contradicting the

fact that X is sequentially compact.
8.7. (Lebesgue Number Lemma - II): If X is a sequentially compact metric space and
U is an open cover for X, then ∃δ > 0 such that, for any y ∈ X, ∃U ∈ U such that
B(y, ) ⊂ U .
Proof. Suppose U does not have a Lebesgue number, then δ = 1/n does not work.
So ∃xn ∈ X such that B(xn , 1/n) is not contained in any single member of U.
Then (xn ) has a convergent subsequence xnk → x. Now x ∈ X, so ∃U ∈ U such
that x ∈ U . Choose δ > 0 such that B(x, δ) ⊂ U , then ∃nk ∈ N such that

d(xnk , x) < δ/2 and 1/nk < δ/2

Then by the triangle inequality

B(xnk , 1/nk ) ⊂ B(x, δ) ⊂ U

This contradicts the assumption on the xn .

(End of Day 24)
8.8. Theorem: If X is a metric space, then TFAE:
(i) X is compact
(ii) X is limit point compact.
(iii) X is sequentially compact.

57
 Proof. (i) ⇒ (ii): Lemma 8.2.
(ii) ⇒ (iii): If (xn ) ⊂ X is a sequence, then let A := {xn }. If A is finite, then there
is a subsequence (nk ) ⊂ N such that xnk is constant, and hence convergent.
Suppose A is infinite, then it has a limit point x. In particular,
B(x, 1) ∩ (A \ {x}) 6= ∅
so choose n1 ∈ N such that xn1 ∈ B(x, 1). Now,
B(x, 1/2) ∩ (A \ {x}) 6= ∅
By the previous lemma, B(x, 1/2) ∩ (A \ {x}) is infinite. In particular,
B(x, 1/2) ∩ (A \ {x, x1 , x2 , . . . , xn1 }) 6= ∅
So ∃n2 > n1 such that
xn2 ∈ B(x, 1/2) ∩ (A \ {x})
Thus proceeding, for each k ∈ N, we choose nk > nk−1 such that
xnk ∈ B(x, 1/k) ∩ (A \ {x})
Now d(x, xnk ) < 1/k, so xnk → x.
(iii) ⇒ (i): If X is sequentially compact, choose an open cover U of X. By the
Lebesgue Number Lemma II, ∃δ > 0 such that any ball of radius δ is contained
in a single member of U. However, X is totally bounded by Lemma 8.6, so
finitely many balls {B(x1 , δ), B(x2 , δ), . . . , B(xn , δ)} cover X. Hence, finitely
many members of U cover X.

8.9. (Bolzano-Weierstrass): Every bounded sequence in Rn has a convergent subse-

quence.
Proof. If (xm ) ⊂ Rn is bounded, then ∃M ≥ 0 such that
n
Y
(xm ) ⊂ [−M, M ] =: Z
i=1

Z is compact, so it is sequentially compact.

8.10. Example: Let
X := {(xn ) ∈ Rω : (xn ) is bounded}
Define a metric on X by
d(x, y) := sup{|xn − yn | : n ∈ N}
This is a well-defined metric on X. Now consider en to be the standard basis
vector in X. Then d(en , 0) = 1, so {en } is a bounded sequence in X. However, en
does not have a convergent subsequence because d(en , em ) = 1 if n 6= m.

58
9. Local Compactness
9.1. Definition: A topological space X is said to be locally compact if, for each x ∈ X,
there is an open neighbourhood V of x such that V is compact.
9.2. Examples:
(i) Every compact space is locally compact.
(ii) R is locally compact because every closed interval [a, b] = (a, b) is compact.
(iii) Q is not locally compact because if V ⊂ Q is open, then ∃a < b in R such
that (a, b) ∩ Q ⊂ V . If s ∈ R \ Q is an irrational such that a < s < b, then
there is a sequence (xn ) ⊂ V that converges to s in R, so (xn ) cannot have a
convergent subsequence. Hence, V cannot be compact.
(iv) Rω with the product topology is not locally compact, because if V is a non-
empty open set, then V contains an open set of the form

(a1 , b1 ) × (a2 , b2 ) × . . . × (an , bn ) × R × R × . . .

If V were compact, then

[a1 , b1 ] × [a2 , b2 ] × . . . × [an , bn ] × R × R × . . .

would be compact, but it is not [Check! Use the fact that R is not compact].
9.3. Theorem: Let X be a topological space, then ∃ a compact space Y such that
(i) X ⊂ Y
(ii) Y \ X is a singleton.
Proof. Define Y := X t {∞} as a new set, and define τY as the collection of sets
U satisfying one of the two following properties:
(i) U ⊂ X is open in X
(ii) ∞ ∈ U and Y \ U is compact in X
We show that τY is a topology on Y , and that Y is compact.
(i) ∅ ∈ τY because ∅ ∈ τX
(ii) Y ∈ τY because Y \ Y = ∅ is compact in X
S
(iii) If {Uα } is a collection of members of τY , we set U := Uα consider two cases:
(a) If ∞ ∈
/ U , then U ∈ τX so U ∈ τY
(b) If ∞ ∈ U , then choose I ⊂ J such that ∞ ∈ Uβ iff β ∈ I, so Uβ = Y \ Cβ
for all β ∈ J, where Cβ ⊂ X is compact, then
! !
[ [ [
Uα = (Y \ Cβ ) ∪ Uγ
α∈J β∈I γ∈I c

59
 T
Now β∈I Cβ is compact, so
[
Y \ Cβ
β∈I

is in τY , so U ∈ τY .
(iv) If U1 , U2 ∈ τY , we WTS: U1 ∩ U2 ∈ τY . Consider cases again:
(a) If ∞ ∈
/ U1 ∪ U2 , then U1 ∩ U2 ∈ τX ⊂ τY
(b) If ∞ ∈ U1 , ∞ ∈
/ U2 , then U1 = Y \ C for C ⊂ X compact, so

U1 ∩ U2 = (Y \ C) ∩ U2 = (X \ C) ∩ U2 ∈ τX ⊂ τY

(c) Similarly if ∞ ∈ U2 \ U1
(d) If ∞ ∈ U1 ∩ U2 , then Ui = (Y \ Ci ) as above, so

U1 ∩ U2 = Y \ (C1 ∪ C2 )

but C1 ∪ C2 is compact in X.
We now show that Y is compact: Suppose U is an open cover for Y , then
∃U ∈ U such that ∞ ∈ U , so U = Y \ C for some compact C ⊂ X. There
are finitely many elements {U1 , U2 , . . . , Un } of U that cover C, so

{U1 , U2 , . . . , Un } ∪ {U }

covers Y .
(End of Day 25)

9.4. Lemma: If X is a locally compact and Hausdorff, then the space constructed above
is Hausdorff.
Proof. If x, y ∈ Y with x 6= y. If x, y ∈ X, then we use the fact that X is Hausdorff
to produce open sets as required. So assume y = ∞, then choose a neighbourhood
V of x such that V is compact. Then U := X \ V is an open neighbourhood of y
and U ∩ V = ∅. So Y is Hausdorff.
9.5. Theorem: If X is locally compact and Hausdorff, and suppose Y1 and Y2 are two
spaces such that
(i) Both Y1 and Y2 are compact.
(ii) X ⊂ Y1 and X ⊂ Y2
(iii) Y1 \ X is a singleton and Y2 \ X is a singleton.
Then there is a homeomorphism p : Y1 → Y2 such that p|X = idX

60
 Proof. Suppose Y1 \ X = {y1 } and Y2 \ X = {y2 }, then define p : Y1 → Y2 by
(
z :z∈X
p(z) =
y2 : z = y1

Then p is clearly a well-defined bijection. Also, if U ⊂ Y2 is an open set such that

U ⊂ X, then p−1 (U ) = U ⊂ Y1 is open. If U ⊂ Y2 is open and ∞ ∈ Y2 , then
F := Y2 \ U = X \ U is closed in Y2 . But Y2 is compact, so F is compact in Y2 .
Since F ⊂ X, F is compact in X. But X ⊂ Y1 , so F is compact in Y1 . But Y1 is
Hausdorff, so F is closed in Y1 . Hence, Y1 \ F = p−1 (U ) is open in Y1 . Hence, p is
continuous. But p : Y1 → Y2 is a continuous bijection from a compact space to a
Hausdorff space, so it is a homeomorphism.
9.6. Definition: Given a locally compact Hausdorff space, we have shown that ∃ a
compact space Y such that X ⊂ Y and Y \ X is a singleton. Furthermore, Y is
unique in the sense of 9.5. This space Y is called the one-point compactification
of X, and is denoted by X + .
9.7. Example: If X = Rn , then X + ∼= Sn
Proof. The stereographic projection gives a continuous injective map p : X → S n ,
and is a homeomorphism onto its range p(X) = S n \ {N }. Identifying X with
p(X), we see that S n satisfies the conditions of Theorem 9.3. By Theorem 9.5,
Sn ∼
= X +.
Note: For n = 2, S 2 , thought of as (R2 )+ is referred to as the Riemann sphere.

61
IV. Separation Axioms
1. Regular Spaces
Assume that all spaces are T1 : Singleton sets are closed.
1.1. Definition: A topological space X is said to be regular if, for any closed set A ⊂ X
and any x ∈ / A, there are open sets U, V ⊂ X such that A ⊂ U, x ∈ V and
U ∩ V = ∅.
1.2. Example:
(i) Every regular space is Hausdorff.
(ii) Let K = {1/n : n ∈ N} ⊂ R and define a topology on R as follows: Define
B1 := { open intervals in R}
B2 := {(a, b) \ K : a < b in R}
Then B := B1 ∪ B2 forms a basis for a topology on R (HW 4), which we
denote by τK . Then RK := (R, τK ) is Hausdorff but not regular.
Proof. RK is Hausdorff because distinct points can be separated by open
intervals. To see that RK is not regular, note that K is closed in RK and
0∈ / K. However, if U is an open set containing 0, then U must contain a
basic open set around 0. It cannot contain sets of the form (−r, r) because
they intersect K. So suppose (−r, r) \ K ⊂ U . Let n ∈ N such that 1/n < r.
Let V be an open set containing K and choose a basic open set (a, b) around
1/n contained in V . Then
1/n ∈ (a, b) and 1/n < r ⇒ ((a, b) \ K) ∩ (−r, r) 6= ∅
Hence, U ∩ V 6= ∅, so K and 0 cannot be separated.
(End of Day 26)
1.3. Theorem: Every compact Hausdorff space is regular.
Proof. If X is compact and A ⊂ X closed, x ∈ / A, then A is compact. For each
y ∈ A, there are open sets Uy , Vy such that x ∈ Uy , y ∈ Vy and Uy ∩ Vy = ∅.
Now {Vy ∩ A : y ∈ A} forms an open cover for A. Choose a finite subcover
{Vyi ∩ A : 1 ≤ i ≤ n} and consider
n
\ n
[
U := Uyi and V := Vyi
i=1 i=1

62
 Then U and V are open, A ⊂ V, x ∈ U and U ∩ V = ∅.
1.4. Theorem: X is regular iff, for each x ∈ X and an open neighbourhood U of x,
there is an open neighbourhood V of x such that V ⊂ U .
Proof. Suppose X is regular, and x ∈ X, U an open neighbourhood of X. Then,
X \ U is closed and does not contain x, so there are open sets V, W such that
x ∈ V, X \ U ⊂ W and V ∩ W = ∅. We claim that V ⊂ U . If y ∈ / U , then y ∈ W
and W ∩ V = ∅, so y ∈
/ V . Hence, V ⊂ U .

Conversely, suppose the given condition holds and x ∈ X, A ⊂ X closed and x ∈

/ A.
Then U := X \ A is an open set containing x, so there is an open set V such that
V ⊂ U . Then W := X \ V is open, contains A and V ∩ W = ∅.
1.5. Corollary: Every subspace of a regular space is regular.
Proof. If Y ⊂ X, where X is regular, suppose U is an open neighbourhood of x
in Y , then U = U 0 ∩ Y for some open set U 0 ⊂ X. Choose V 0 ⊂ X open such that
V 0 ⊂ U 0 . Now take V := V 0 ∩ Y , which is open in Y , contains x and by II.6.8,

clY (V ) = clX (V ) ∩ Y ⊂ clX (V 0 ) ∩ Y ⊂ U 0 ∩ Y = U

1.6. Corollary: Every locally compact Hausdorff space is regular.

Proof. Let X be locally compact and Hausdorff, and X ⊂ X + its one point com-
pactification. X + is regular, so X must also be regular.
1.7. Corollary: Any product of regular spaces is regular.
Q
Proof. Suppose Xα is regular for all α ∈ J, and X := α∈J Xα . Let x := (xα ∈ X
and U ⊂ X an open neighbourhood of x. Then we may assume that U is a basic
open set of the form Y
Uα1 × Uα2 × . . . × Uαn × Xβ
β

Now xαi ∈ Uαi , so there are open sets Vαi such that Vαi ⊂ Uαi . Then
Y
V := Vα1 × Vα2 × . . . × Vαn × Xβ
β

is an open neighbourhood of x such that V ⊂ U [Why?]

63
2. Normal Spaces
2.1. Definition: A topological space X is said to be normal if, whenever A and B are
disjoint closed sets, there are open sets U, V such that A ⊂ U, B ⊂ V and U ∩V = ∅
2.2. Lemma: X is normal iff, given a closed set A ⊂ X and an open set U containing
A, there is an open set V containing A such that V ⊂ U
Proof. HW.
2.3. Theorem: Every metric space is normal.
Proof. If A, B ⊂ X are disjoint closed sets. For each a ∈ A, a ∈
/ B, so ∃a > 0
such that B(a, a ) ⊂ X \ B. Define
[
U := B(a, a /2)
a∈A

Then U is open and it contains A. Similarly, define

[
V := B(b, b /2)
b∈B

where b is chosen as above. Then, if z ∈ U ∩ V , then ∃a ∈ A, b ∈ B such that

z ∈ B(a, a /2) ∩ B(b, b /2)

Assume WLOG that a ≤ b , then by triangle inequality,

a b
d(a, b) ≤ d(a, z) + d(z, b) < + ≤ a
2 2
Hence, B(a, a ) ∩ B 6= ∅ contradicting the choice of a .
2.4. Theorem: Every compact Hausdorff space is normal.
Proof. Let X be a compact Hausdorff space and A, B ⊂ X disjoint closed sets.
By 1.3, X is regular, so for each a ∈ A, there are open sets Ua and Va such that

a ∈ Ua , B ⊂ Va and Ua ∩ Va = ∅

So {Ua : a ∈ A} is an open cover for A. But A is compact, so there is a finite

subcover {Ua1 , Ua2 , . . . , Uak }. Define
k
[ n
\
U := Uai and V := Vai
i=1 i=1

Then U, V are open, A ⊂ U, B ⊂ V and U ∩ V = ∅ [Check!].

(End of Day 27)

64
2.5. Theorem: A closed subspace of a normal space is normal.
Proof. If Y ⊂ X is closed and X is normal. We use Lemma 2.2. Suppose A ⊂ Y
is closed and U ⊂ Y an open set such that A ⊂ U . Then write U = U 0 ∩ Y for
some open set U 0 ⊂ X. Since A is closed in Y and Y is closed in X, A is closed in
X. Hence, there is an open set V 0 ⊂ X such that A ⊂ V 0 and V 0 ⊂ U 0 . Now set

V := V 0 ∩ Y

Then A ⊂ V and by II.6.8,

clY (V ) = clX (V ) ∩ Y ⊂ clX (V 0 ) ∩ Y ⊂ U 0 ∩ Y = U

2.6. Examples:
(i) Every normal space is regular. Hence, every normal space is Hausdorff.
(ii) Let X = R with the topology whose basis are sets of the form

[a, b)

where −∞ < a < b ≤ ∞ (See Quiz 1). This topology is denoted by τ` and it
contains the usual topology. It follows that R` := (R, τ` ) is normal.
(iii) X := R` × R` is thus a product of regular spaces, so it is regular. However,
it is not normal [without proof]. Hence,
(a) The product of normal spaces is not necessarily normal.
(b) This is an example of a space that is regular but not normal.
2.7. (Urysohn’s Lemma for metric spaces): Let (X, d) be a metric space and A, B ⊂ X
disjoint closed sets. Then ∃f : X → [0, 1] continuous such that

f (x) = 0 ∀x ∈ A and f (y) = 1 ∀y ∈ B

Proof. Recall that x 7→ d(x, A) is continuous and d(x, A) = 0 iff x ∈ A. Define

f : X → [0, 1] by
d(x, A)
f (x) =
d(x, A) + d(x, B)
Note that the denominator is non-zero because A ∩ B = ∅. Now check that f
satisfies the required properties.
2.8. Lemma: Let X be a normal space and A, B ⊂ X disjoint closed sets. Let P :=
Q ∩ [0, 1], then there is a sequence of open sets {Up : p ∈ P } such that
(i) A ⊂ U0 and U1 = X \ B
(ii) For all p, q ∈ P, p < q ⇒ Up ⊂ Uq

65
 Proof. Define U1 := X \ B. Since A ⊂ U1 , define U0 by Lemma 2.2 such that

A ⊂ U0 and U0 ⊂ U1

Now arrange P in a sequence {0, 1, p1 , p2 , . . .}. We wish to define Up1 : Note that
0 < p1 < 1 and U0 ⊂ U1 , so by Lemma 2.2, there is an open set Up1 such that

U0 ⊂ Up1 and Up1 ⊂ U1

Now we proceed by induction. Having define {U0 , U1 , Up1 , . . . , Upn }, we wish to

define Upn+1 . Since 0 < pn+1 < 1, choose an immediate predecession pi and an
immediate successor pj among {0, 1, p1 , p2 , . . . , pn }. Note that Upi ⊂ Upj . So by
Lemma 2.2, there is an open set Upn+1 such that

Upi ⊂ Upn+1 and Upn+1 ⊂ Upj

By induction, we define Up for all p ∈ P satisfying (i) and (ii).

2.9. Lemma: Let X be a normal space and A, B ⊂ X disjoint closed sets. Let {Up :
p ∈ Q ∩ [0, 1]} be a sequence of open sets as in the previous lemma. Define Up = ∅
if p < 0 and Uq = X if q > 1. Now define f : X → R by

f (x) := inf Q(x)

where Q(x) := {p ∈ Q ∩ [0, 1] : x ∈ Up }.

(i) f (x) ∈ [0, 1] for all x ∈ X.
(ii) For any r ∈ Q, x ∈ Ur ⇒ f (x) ≤ r, and
(iii) x ∈
/ Ur ⇒ f (x) ≥ r
Proof. Note that f is well-defined because, for any x ∈ X, x ∈ Up for all p > 1,
so (1, ∞) ∩ Q ⊂ Q(x). Hence, f (x) ≤ 1. Similarly, x ∈/ Up for all p < 0. Hence,
f (x) ≥ 0.

If x ∈ Ur , then for any p > r, x ∈ Up . Hence,

(r, ∞) ∩ Q ⊂ Q(x)

Since the infimum of a subset is greater than the infimum of a super set, f (x) ≤ r.
Similarly, if x ∈
/ Ur , then x ∈
/ Us for all s < r. Hence,

Q(x) ⊂ (r, ∞) ∩ Q

As before, this implies f (x) ≥ r

2.10. (Urysohn’s Lemma): Let X be a normal space and A, B ⊂ X disjoint closed sets.
Then ∃f : X → [0, 1] continuous such that

f (x) = 0 ∀x ∈ A and f (y) = 1 ∀y ∈ B

66
 Proof. Let {Up : p ∈ Q} and f : X → R defined as above. For any x ∈ X, and
r < 0, x ∈ / Ur , so f (x) ≥ 0. Similarly, f (x) ≤ 1. Furthermore, if x ∈ A, then
x ∈ U0 , so f (x) = 0. Similarly, f (y) = 1 for all y ∈ B. It suffices to show that f
is continuous.

Fix x0 ∈ X and U an open set containing f (x0 ). WTS: ∃ an open set V ⊂ X

containing x0 such that f (V ) ⊂ U . Choose c, d ∈ R such that (c, d) ⊂ U . Now
there exists p, r ∈ Q such that [p, r] ⊂ (c, d) ⊂ U , and let

V := Ur \ Up

Note that V is open, and if z ∈ V , then z ∈ Ur and z ∈

/ Up . So by the previous
lemma,
p ≤ f (x) ≤ r
Hence, f (V ) ⊂ U as required.
(End of Day 28)
2.11. Corollary: Let X be a normal space and A, B ⊂ X disjoint closed sets. Given
a, b ∈ R with a < b, ∃f : X → [a, b] continuous such that

f |A = a and f |B = b

Proof. Simply compose the function g : X → [0, 1] produced by Urysohn’s lemma

with the map [0, 1] → [a, b] given by

t 7→ (1 − t)a + tb

3. Tietze’s extension Theorem

3.1. Definition: Let (X, d) be a metric space.
(i) A sequence (xn ) ⊂ X is said to be Cauchy if, for each  > 0, ∃N ∈ N such
that d(xn , xm ) <  for all n, m ≥ N .
(ii) X is said to be complete if every Cauchy sequence in X converges to a point
in X.
3.2. Examples:
(i) Every convergent sequence is Cauchy.
√
(ii) Let X = Qc , and xn := 2/n, then (xn ) is Cauchy, but does not converge to
a point in X. Hence X is not complete.
(iii) X = (0, 1) is not complete because (1/n) is Cauchy but not convergent.

67
3.3. Lemma: Let (X, d) be a metric space and (xn ) ⊂ X Cauchy. Then (xn ) is bounded.
ie. ∃x0 ∈ X and M ≥ 0 such that d(xn , x0 ) ≤ M for all n ∈ N
Proof. Fix  = 1, then ∃N ∈ N such that
d(xn , xm ) < 1 ∀n, m ≥ 1
For x0 ∈ X fixed, let
M := max{d(x0 , xi ) : 1 ≤ i ≤ N } + 1
Then for any n ∈ N, if n ≤ N , then d(xn , x0 ) ≤ M . And if n ≥ N , then
d(xn , x0 ) ≤ d(xn , xN ) + d(xN , x0 ) ≤ M

3.4. Lemma: Let (X, d) be a metric space and (xn ) a Cauchy sequence. If (xn ) has a
convergent subsequence, then (xn ) converges.
Proof. Suppose xnk → x is a convergent subsequence. For any  > 0, choose
N ∈ N such that
d(xn , xm ) < /2 ∀n, m ≥ N
Now choose K ∈ N such that
d(xni , x) < /2 ∀i ≥ K
Hence, N0 := max{N, nK } has the property that
d(xn , x) <  ∀n ≥ N0

3.5. Lemma: Every sequence in R has a monotone subsequence.

Proof. Let (xn ) ⊂ R and suppose (xn ) has no monotone increasing subsequence.
We show that (xn ) has a monotone decreasing subsequence. We claim: ∃n1 ∈ N
such that xn < xn1 for all n > n1 .
Proof. Suppose not, then set n1 = 1. Then ∃n2 > n1 and xn2 such that xn2 > xn1 .
Similarly, ∃n3 > n2 such that xn3 > xn2 and so on. Thus, we produce a subsequence
(xnk ) that is strictly increasing. This contradicts the assumption that (xn ) has no
increasing subsequence.
Now choose n1 ∈ N such that xn < xn1 for all n > n1 . Now consider the sub-
sequence {xn1 , xn1 +1 , xn1 +2 , . . .}. By the same argument as above, ∃n2 > n1 such
that xn < xn2 for all n > n2 . In particular,
xn2 < xn1
and
xn < xn2 ∀n > n2
Thus proceeding (by induction) there is a subsequence (xnk ) that is strictly de-
creasing.

68
3.6. Theorem: R is complete.
Proof. Let (xn ) ⊂ R be Cauchy, then by the previous lemmas, (xn ) is bounded and
has a monotone subsequence. But every monotone bounded subsequence in R is
convergent (to its supremum or infimum). Some the previous lemma applies.
3.7. Definition: Let X be a topological space and (Y, d) a metric space.
(i) A function f : X → Y is said to be bounded if f (X) is a bounded subset of
Y (ie. ∃y0 ∈ X and M ≥ 0 such that d(f (x), y0 ) ≤ M for all x ∈ X.
(ii) Let Cb (X, Y ) denote the set of all continuous, bounded functions f : X → Y
3.8. Theorem: Define d∞ : Cb (X, Y ) × Cb (X, Y ) → R by

d∞ (f, g) := sup{d(f (x), g(x)) : x ∈ X}

Then this defines a metric on Cb (X, Y ).

Proof. HW
(End of Day 29)
3.9. Theorem: If (Y, d) is a complete metric space, and (Cb (X, Y ), d∞ ) is complete.
Proof. Let (fn ) ⊂ Cb (X, Y ) be a Cauchy sequence. For any x ∈ X,

d(fn (x), fm (x)) ≤ d∞ (fn , fm )

Hence, (fn (x)) is Cauchy in Y . Hence, ∃zx ∈ Y such that fn (x) → zx . Define
f : X → Y by f (x) = zx . We claim that f is continuous and bounded.
(i) Since (fn ) is Cauchy, it is bounded. Hence, ∃M ≥ 0 such that

sup d(fn (x), 0) ≤ M ∀n ∈ N

x∈X

For any x ∈ X fixed, fn (x) → f (x). Hence, d(f (x), 0) ≤ M [Why?]. Hence,
f is bounded.
(ii) To see that fn → f wrt d∞ : Fix  > 0, then ∃N ∈ N such that

d∞ (fn , fm ) < /2 ∀n, m ≥ M

Hence for x ∈ X fixed,

d(fn (x), fm (x)) < /2 ∀n, m ≥ N

Let m → ∞, then
d(fn (x), f (x)) ≤ /2 ∀n ≥ N
Hence, d∞ (fn , f ) <  ∀n ≥ N . Hence, fn → f in d∞

69
 (iii) To see that f is continuous: Let x0 ∈ X and  > 0, then ∃N ∈ N such that

d∞ (fn , f ) < /3 ∀n ≥ N

Since fN is continuous, ∃U ⊂ X open such that x0 ∈ U and

d(fN (y), fN (x0 )) < /3 ∀y ∈ U

Hence, for all y ∈ U ,

d(f (y), f (x0 )) < 

3.10. Corollary: Let X be any topological space. The set Cb (X) := Cb (X, R) is a
complete metric space with respect to the metric

d∞ (f, g) := sup |f (x) − g(x)|

x∈X

3.11. (Tietze’s Extension Theorem): Let X be a normal topological space and Y ⊂ X

closed. Let f : Y → R be a continuous function, then ∃h : X → R continuous
such that
h(y) = f (y) ∀y ∈ Y
(h is called a continuous extension of f )
Proof. Assume first that f is bounded and

c := sup{|f (y)| : y ∈ Y }

Define
E0 := {x ∈ X : f (x) ≤ −c/3} = f −1 (−∞, −c/3]
F0 := {x ∈ X : f (x) ≥ c/3} = f −1 [c/3, ∞)
Then E0 and F0 are disjoint closed sets. By Corollary 2.11, ∃g0 : X → R such that

−c/3 ≤ g0 (x) ≤ c/3 ∀x ∈ X

and
g0 |E0 = −c/3 and g0 |F0 = c/3
Hence,
|g0 (x)| ≤ c/3 ∀x ∈ X
|f (y) − g0 (y)| ≤ 2c/3 ∀y ∈ Y
Let f1 := f − g0 . Then by the above argument, ∃g1 : X → R continuous such that
|g1 (x)| ≤ 2c/9 ∀x ∈ X
|f (y) − g0 (y) − g1 (y)| ≤ 4c/9 ∀y ∈ Y

70
 Thus proceeding, we obtain a sequence (gn ) of continuous functions such that

|gn (x)| ≤ 2n c/3n+1 ∀x ∈ X

|f (y) − hn (y)| ≤ 2n+1 c/3n+1 ∀y ∈ Y

where hn := g0 + g1 + . . . + gn . Now note that if m > n,

 
 Xn 
|hn (x) − hm (x)| =  gi (x)
 
 
i=m+1
Xn
≤ |gi (x)|
i=m+1
n
X 2i c 2m+1 c
≤ ≤
i=m+1
3i+1 3m+1

Hence,
2m+1 c
d∞ (hn , hm ) ≤ m+1
3
Since the RHS goes to zero, (hn ) form a Cauchy sequence in Cb (X, R). By the
previous lemma, ∃h ∈ Cb (X, R) such that hn → h. Now if y ∈ Y , then

2n+1 c
|f (y) − hn (y)| ≤
3n+1
Letting n → ∞, we see that h = f on Y .

Now suppose f is not bounded. Let g : R → (−1, 1) be a homeomorphism (is

there one?). Now define fe := g ◦ f . Now fe is bounded, so ∃e
h : X → R continuous
−1
h|Y = fe. Now define h := g ◦ e
such that e h, and check that h satisfies the required
conditions.

(End of Day 30)

4. Urysohn Metrization Theorem

4.1. Definition: A topological space (X, τ ) is said to be metrizable if there exists a
metric d on X such that τ = τd .
4.2. Theorem: Rω with the product topology is metrizable.
Proof. Let d : R × R → R be the metric given by

d(a, b) = min{|a − b|, 1}

71
Define D : Rω × Rω → R by
 
d(xi , yi )
D(x, y) := sup
i

Then [Check!] that D is a metric on Rω . We claim that the product topology τp

on Rω coincides with τD
(i) τp ⊂ τD : Let U be a basic open set in τp of the form

U := U1 × U2 × . . . × Un × R × R × . . .

Let x = (xi ) ∈ U , so for 1 ≤ i ≤ n, xi ∈ Ui , so ∃i > 0 such that

(xi − i , xi + i ) ⊂ Ui

Assume i < 1 for all i, and let  := min{i /i : 1 ≤ i ≤ n}, then we claim that

BD (x, ) ⊂ U

To see this, suppose y = (yi ) ∈ BD (x, ), then for 1 ≤ i ≤ n,

d(xi , yi )
≤ D(x, y) < 
i
Hence, d(xi , yi ) ≤ i < 1, so |xi − yi | < i . Hence, yi ∈ Ui for all 1 ≤ i ≤ n.
Hence, y ∈ U , so
BD (x, ) ⊂ U
Thus, U is a union of sets of the form BD (x, ), and so U ∈ τD . Since U is a
generic basic open set, it follows that τp ⊂ τD .
(ii) τD ⊂ τp : Let U ∈ τD be open, and x ∈ U . Then ∃ > 0 such that BD (x, ) ⊂
U . Choose N ∈ N such that 1/N < , and consider

V := (x1 − , x1 + ) × . . . × (xN − , xN + ) × R × . . .

We claim that V ⊂ BD (x, ). To see this, suppose y = (yi ) ∈ V , then for
i ≥ N,
d(xi , yi ) 1
≤
i N
because d(xi , yi ) ≤ 1. Furthermore, if 1 ≤ i ≤ N , then

d(xi , yi ) d(xi , yi ) 1
≤ ≤ <
i i Ni
Hence, D(x, y) < . This is true for any y ∈ V , so V ⊂ BD (x, ) ⊂ U . Hence,
U is a union of open sets in τp , and so U ∈ τp . Thus, τD ⊂ τp as well.

72
4.3. Definition: A topological space is called second countable if it has a countable
basis.
4.4. Example:
(i) Rn is second countable.
(ii) If R is given the discrete metric, then it is not second countable.
(iii) Every second countable space is separable.
Proof. Let {Bn : n ∈ N} be a countable basis for X. For each n ∈ N, choose
xn ∈ Bn and let D := {xn : n ∈ N}. Then D is dense in X, because if U is
any non-empty open set, then ∃n ∈ N such that Bn ⊂ U , so xn ∈ U which
implies D ∩ U 6= ∅.
(iv) Any separable metric space is second countable.
Proof. Let (X, d) be a separable metric space and A := {xn } be a countable
dense subset of X. Let Bm,n := B(xm , 1/n), then we claim that B := {Bm,n }
forms a basis for τd .
(a) If x ∈ X, then ∃xm ∈ A such that d(xm , x) < 1. Hence, x ∈ Bm,1 . So B
covers X.
(b) Furthermore, if x ∈ Bm1 ,n1 ∩ Bm2 ,n2 then let α := min{1/2n1 , 1/2n2 }.
Choose m3 ∈ N such that d(x, xm3 ) < α and let n3 ∈ N such that
1/n3 < α, then [Check!]

Bm3 ,n3 ⊂ Bm1 ,n1 ∩ Bm2 ,n2

and x ∈ Bm3 ,n3

(c) Thus, B forms a basis for some topology τ on X. Since B ⊂ τd , it follows
that τ ⊂ τd .
(d) However, if U ∈ τd and x ∈ U , then ∃ > 0 such that Bd (x, ) ⊂ U .
Now choose m ∈ N such that d(x, xm ) < /2, and let n ∈ N such that
1/n < /2, then x ∈ Bm,n and Bm,n ⊂ Bd (x, ) ⊂ U . Hence, every U ∈ τd
is obtained as a union of elements of B.
Hence, B is a basis for τd .
4.5. Lemma: Every regular space with a countable basis is normal.
Proof. Let X be a regular space with a countable basis B, and let A, B ⊂ X be
two closed disjoint sets. WTS: ∃ open sets U and V such that A ⊂ U, B ⊂ V and
U ∩ V = ∅.
(i) For each x ∈ A, x ∈
/ B, so there is an open sets U, V such that x ∈ U, B ⊂ V
and U ∩ V = ∅. Since X is regular, there is an open set W such that x ∈ W
and W ⊂ U . Choose a basic open set Bx ∈ B such that x ∈ Bx and Bx ⊂ W .
Thus,
Bx ∩ B = ∅

73
 Thus, we obtain an open cover {Bx : x ∈ A} for A which is countable, so we
denote it by {Un : n ∈ N}. Note that

Un ∩ B = ∅ ∀n ∈ N

Similarly, we obtain an open cover {Vn : n ∈ N} of B which is countable such

that
Vn ∩ A = ∅ ∀n ∈ N
S S
(ii) If U := Un and V := Vn , then A ⊂ U, B ⊂ V , but U and V need not be
disjoint. So define
"n # "n #
[ [
Un0 := Un \ Vn and Vn0 := Vn \ Un
i=1 i=1

Then each Un0 and Vn0 is open.

(iii) If x ∈ A, then ∃n ∈ N such that x ∈ Un . But Vi ∩ A = ∅ for all i. Hence,
x ∈ Un0 . Thus, {Un0 : n ∈ N} forms an open cover for A. Define
∞
[
0
U := Un0
n=1

Then A ⊂ U 0 . Similarly, if
∞
[
V 0 := Vn0
n=1
0
Then B ⊂ V .
(iv) We claim that U 0 ∩ V 0 = ∅. Suppose x ∈ U 0 ∩ V 0 , then ∃n, m ∈ N such that
x ∈ Un0 and x ∈ Vm0 . Assume n > m, then x ∈ / Vm by definition of Un0 . This
is a contradiction, so U 0 ∩ V 0 = ∅.

(End of Day 31)

4.6. Lemma: Let X be a regular space with a countable basis. Then there is a sequence
of functions fn : X → [0, 1] such that, for any x0 ∈ X and open set U containing
x0 , ∃n ∈ N such that fn (x0 ) = 1 and fn = 0 on X \ U .
Proof. Note that X is normal so Urysohn’s lemma applies. Let {Bn : n ∈ N} be a
countable basis for X. Define

D := {(n, m) ∈ N × N : Bn ⊂ Bm }

For each (n, m) ∈ D, Urysohn’s lemma implies that there is a function gn,m : X →
[0, 1] such that
gn,m |Bn = 1 and gn,m |X\Bm = 0

74
 This collection {gn,m } = {fn } is countable, and it satisfies the required condition:
If x0 ∈ X and U is an open set such that x0 ∈ U , then ∃ a basic open set Bm
such that x0 ∈ Bm and Bm ⊂ U . Furthermore, by regularity, ∃ a basic open set
Bn such that x0 ∈ Bn and Bn ⊂ Bm . Now

gn,m (x0 ) = 1 and gn,m |X\U = 0

4.7. (Urysohn’s Metrization Theorem): Every regular space with a countable basis is
metrizable.
Proof. (i) We construct a continuous function F : X → Rω as follows: Let {fn }
be a sequence as in the previous lemma, and define

F (x) := (fn (x))

Then F is continuous because each coordinate function fn is continuous.

(ii) F is injective: If x 6= y, then there is an open set U such that x ∈ U and
y ∈/ U . Choose n ∈ N such that fn (x) = 1 and fn |X\U = 0. In particular,
fn (y) = 0. Hence, F (x) 6= F (y).
(iii) Let Z := F (X). We claim that F : X → Z is a homeomorphism. F is clearly
surjective, so it suffices to show that F is an open map. Let U ⊂ X be an
open set. WTS: F (U ) ⊂ Z is open. Fix z ∈ F (U ), then ∃x ∈ U such that

F (x) = z

Choose n ∈ N such that fn (x) = 1 and fn |X\U = 0. Define

V := πn−1 ((0, ∞)) ⊂ Rω

and set
W := V ∩ Z
Then W is open in Z since V is open in Rω . Furthermore, fn (x) > 0, so
z ∈ W . We claim: W ⊂ F (U ). To see this, fix y ∈ W , then ∃x0 ∈ X such
that F (x0 ) = y. Now, πn (y) > 0, but

πn (y) = πn (F (x0 )) = fn (x0 )

Since fn = 0 on X \ U , it follows that x0 ∈ U . Hence, x0 ∈ F (U ). Thus,

W ⊂ F (U ). Hence, every z ∈ F (U ) is an interior point of F (U ), so F (U ) is
open.
(iv) Thus, F : X → Z is a homeomorphism. Since Z ⊂ Rω and Rω is metrizable,
it follows that Z is metrizable, and so X is too.

75
4.8. Corollary: Every compact Hausdorff space with a countable basis is metrizable.
4.9. Example:
(i) Every metric space is certainly regular, but need not have a countable basis
(See 4.4(ii)).
(ii) Let K = {1/n : n ∈ N}. Define

B1 := { open intervals in R with rational end-points}

B2 := {(a, b) \ K : a < b in Q}

Then B := B1 ∪ B2 forms a basis for a topology on R, which we denote by τK .

Then RK := (R, τK ) is Hausdorff, has a countable basis, but is not metriz-
able because it is not regular. Thus, regularity is necessary for Urysohn’s
metrization theorem to hold.

5. Imbedding of Manifolds
5.1. An m-manifold is a topological space X with a countable basis such that for each
x ∈ X, there is a neighbourhood Ux of x such that Ux is homeomorphic with an
open subset of Rm .
5.2. Examples:
(i) Rm is an m-manifold. So is any open subset of Rm .
(ii) [0, 1] is not a 1-manifold, because any neighbourhood of 0 is of the form [0, δ),
which is not homeomorphic to an open subset of R.
(iii) S 1 is a 1-manifold. In general, S m is an m-manifold (without proof)
(iv) A 1-manifold is called a curve, and a 2-manifold is called a surface.
(v) The torus S 1 × S 1 is a surface. In general, if X and Y are manifolds, then so
is X × Y .
(End of Day 32)
5.3. Theorem: Let X be an m-manifold. Then X is
(i) Locally path connected.
(ii) Locally compact.
(iii) Regular
(iv) Metrizable.
Proof. (i) Let x ∈ X and U an open neighbourhood of x. WTS: ∃V ⊂ U open
such that x ∈ V and V is path connected. To see this, choose a neighbourhood
Ux of x and a homeomorphism

g : Ux → Ux0 ⊂ Rm

76
 where Ux0 is open in Rm . Then Ux ∩ U is open and

g|Ux ∩U : Ux ∩ U → g(Ux0 ∩ U ) ⊂ Rm

is a homeomorphism. Since g(Ux0 ∩ U ) is an open subset of Rm containing

g(x), and Rm is locally path connected, there is an open set V 0 ⊂ g(Ux0 ∩ U )
that is path connected and containing g(x). Then V := g −1 (V 0 ) is open, path
connected, contains x and V ⊂ U .
(ii) Local compactness is identical to part (i).
(iii) Let x ∈ X and an open set U containing x. WTS: ∃V open such that x ∈ V
and V ⊂ U . Choose Ux open and a homeomorphism

g : Ux → Ux0 ⊂ Rm

as before. Since U ∩ Ux is open in Ux ,

g(U ∩ Ux ) ⊂ Ux0

is open and contains g(x). Since Ux0 ⊂ Rm and Rm is regular, Ux0 is regular
by 1.5. Hence, there is an open set V 0 such that g(x) ∈ V 0 and

V 0 ⊂ g(U ∩ Ux )

Then V := g −1 (V 0 ) is open, contains x and since g is a local homeomorphism

V = g −1 (V 0 ) = g −1 (V 0 ) ⊂ g −1 (g(U ∩ Ux )) ⊂ U ∩ Ux ⊂ U

Hence, X is regular.
(iv) X has a countable basis, so Urysohn’s metrization theorem applies.

5.4. Definition: Let X be a topological space.

(i) Let f : X → R be a function. The support of f is the set

supp(f ) := {x ∈ X : f (x) 6= 0}

(ii) Let U := {U1 , U2 , . . . , Un } be an open cover for X. A partition of unity

dominated by U is a family of continuous functions fi : X → R such that
(a) supp(fi ) ⊂ Ui for all 1 ≤ i ≤ n
(b) For each x ∈ X, f1 (x) + f2 (x) + . . . + fn (x) = 1
5.5. Lemma: Let X be a normal space and U := {U1 , U2 , . . . , Un } be an open cover for
X. Then there is an open cover V := {V1 , V2 , . . . , Vn } such that

Vi ⊂ Ui

for all 1 ≤ i ≤ n

77
 Proof. We induct on n: If n = 1, then U1 = X so take V1 = U1 . If n ≥ 2, note
that "n #
[
A := X \ Ui
i=2

is closed and A ⊂ U1 . Since X is normal, there is an open set V1 such that

A ⊂ V1 and V1 ⊂ U1

The collection {V1 , U2 , . . . , Un } now covers X. Proceeding by induction, suppose

that we have produced a cover

{V1 , V2 , . . . , Vk−1 , Uk , Uk+1 , . . . , Un }

such that Vi ⊂ Ui for all 1 ≤ i ≤ k − 1. Let

" k−1 ! n
!#
[ [
A := X \ Vi ∪ Uj
i=1 j=k+1

Then A is closed and contained in Uk . Choose Vk open such that A ⊂ Vk and

Vk ⊂ Uk . Now {V1 , V2 , . . . , Vk , Uk+1 , . . . , Un } forms an open cover. Proceeding
thus, we exhaust all Ui ’s.
5.6. Theorem: Let X be a normal space and U be a finite open cover for X. Then
there is a partition of unity dominated by U.
Proof. Let U := {U1 , U2 , . . . , Un } be an open cover for X. Choose a cover V :=
{V1 , V2 , . . . , Vn } such that Vi ⊂ Ui and an open cover W := {W1 , W2 , . . . , Wn }
such that Wi ⊂ Vi for all 1 ≤ i ≤ n. By Urysohn’s lemma, there exist function
ψi : X → [0, 1] such that

ψi |Wi = 1 and ψi |X\Vi = 0

Then
supp(ψi ) ⊂ Vi ⊂ Ui
For any x ∈ X, ∃1 ≤ i ≤ n such that x ∈ Wi , so ψi (x) = 1. Hence, define
fi : X → R by
ψi (x)
fi (x) :=
ψ1 (x) + ψ2 (x) + . . . + ψn (x)
The denominator is never zero, so fi is continuous, and is a partition of unity
dominated by U.
5.7. (Imbedding Theorem): Let X be a compact m-manifold, then ∃N ∈ N and an
injective map
F : X → RN
such that F : X → F (X) is a homeomorphism. (ie. F is an imbedding of X into
Rn )

78
Proof. For each x ∈ X, ∃ an open set Ux that is homeomorphic to an open subset
of Rm . Choose a finite subcover {U1 , U2 , . . . , Un } and homeomorphisms

gi : Ui → Vi

where Vi ⊂ Rm is open. Let {f1 , f2 , . . . , fn } be a partition of unity dominated by

U. Let Ai := supp(fi ) ⊂ Ui and define hi : X → Rm by
(
fi (x)gi (x) : x ∈ Ui
hi (x) :=
0 : x ∈ X \ Ai

If x ∈ (X \ Ai ) ∩ Ui , then fi (x) = 0, so both definitions agree. So by pasting

lemma, hi is continuous. Define

× . . . R} × |Rm × Rm {z
F : X → |R × R{z × . . . × Rm}
n times n times

by
x 7→ (f1 (x), f2 (x), . . . , fn (x), h1 (x), h2 (x), . . . , hn (x))
Then F is continuous. Suppose we show that F is injective, then since X is
compact,
F : X → F (X)
will be a homeomorphism. So suppose x, y ∈ X such that F (x) = F (y), then
choose 1 ≤ i ≤ n such that fi (x) > 0. Then x ∈ Ui and fi (x) = fi (y) > 0 and
hi (x) = hi (y) implies that
gi (x) = gi (y)
But gi : Ui → Vi is a homeomorphism, so x = y as required.
(End of Day 33)
Review
(End of Day 34)

79
V. Instructor Notes
0.1. This time, the semester was shortened by one week to accommodate an earlier
convocation.
0.2. Therefore, I was unable to cover Tychonoff’s theorem, which is unfortunate. I did
however discuss the theorem, and how the tube lemma proof does not work. The
only other topic on the syllabus I did not cover was Lindeloff spaces, which is not
a major loss. One extra week and I would have done all of it.
0.3. The student response seemed alright, but no questions were forthcoming which
made it harder to judge.

80
Bibliography
[Crossley] M.D. Crossley, Essential Topology, Springer-Verlag (2005)

[Munkres] J. Munkres, Topology (2nd Ed.)

81