TOPOLOGY HW 4

CLAY SHONKWILER

24.1
(a) Show that no two of the spaces (0, 1), (0, 1], and [0, 1] are homeomor-
phic.
Proof. Suppose (0, 1) and (0, 1] are homeomorphic, with the homeomor-
phism given by f . If A = (0, 1) − {f −1 (1), then f |A : A → (0, 1) is a
homeomorphism, by Theorem 18.2(d). However, the interval (0, 1) is con-
nected, whereas
A = (0, f −1 (1)) ∪ (f −1 (1), 1)
is not connected, so A and (0, 1) cannot be homeomorphic. From this con-
tradiction, then, we conclude that (0, 1) and (0, 1] are not homeomorphic.
Similarly, suppose g : (0, 1] → [0, 1] is a homeomorphism. Then, if B =
(0, 1] − {g −1 (0), g −1 (1)}, g|B : B → (0, 1) is a homeomorphism. However,
g −1 (0) 6= g −1 (1), so at most one of these can be 1, meaning one must lie in
the interval (0, 1). Suppose, without loss of generality, that g −1 (0) ∈ (0, 1).
Then
B = (0, g −1 (0)) ∪ (g −1 (0), 1] − {g −1 (1)
is not connected, whereas (0, 1) is, so the two cannot be homeomorphic.
From this contradiction, then, we conclude that (0, 1] and [0, 1] are not
homeomorphic.
A similar argument easily demonstrates that (0, 1) and [0, 1] are not home-
omorphic, so we see that no two of the spaces (0, 1), (0, 1], and [0, 1] are
homeomorphic. 
(b) Suppose that there exist imbeddings f : X → Y and g : Y → X.
Show by means of an example that X and Y need not be homeomorphic.
Example: Let f : (0, 1) → [0, 1] be the canonical imbedding and let
g : [0, 1] → (0, 1) such that
x 1
g(x) = + .
3 3
Then g([0, 1]) = [ 31 , 23 ]. Obtain g 0 by restricting the range of g to g([0, 1]) =
[ 13 , 32 ]. We claim that g 0 is a homeomorphism. Since multiplication and addi-
tion are continuous, as are the inclusion map and compositions of continuous
functions, we see that g 0 is continuous, as is g 0−1 , where
 
1
g 0−1 (x) = 3 x − .
3
1
2 CLAY SHONKWILER

Both of these maps are also bijective, so we see that g 0 is indeed a homeo-
morphism, meaning g is an imbedding.
However, as we saw in part (a) above, (0, 1) and [0, 1] are not homeomor-
phic, so two spaces need not be homeomorphic for each to be imbedded in
the other.
♣
n
(c) Show R and R are not homeomorphic if n > 1.
Lemma 0.1. If f : X → Y is a homeomorphism and X is path-connected,
then Y is path-connected.
Proof. Let x, y ∈ Y . Then there exists a continuous path g : [a, b] → X such
that g(a) = f −1 (x) and g(b) = f −1 (b). Define
h := f ◦ g.
Then h : [a, b] → Y is continuous and
h(a) = (f ◦ g)(a) = f (g(a)) = f (f −1 (x)) = x
and
h(b) = (f ◦ g)(b) = f (g(b)) = f (f −1 (y)) = y
since f is bijective. Hence, h is a path from x to y, so, since our choice of x
and y was arbitrary, Y is path connected. 
Proposition 0.2. Rn and R are not homeomorphic if n > 1.
Proof. Suppose f : Rn → R is a homeomorphism. Then, restricting the
domain to Rn −{0} gives a homeomorphism of the punctured euclidean space
to R − {f (0)}. However, the punctured euclidean space is path-connected
(as shown in Example 4), whereas R − {f (0)} is not even connected, let
alone path-connected. To see this, we need only note that
R − {f (0)} = (−∞, f (0)) ∪ (f (0), ∞)
so the open sets (−∞, f (0)) and (f (0), ∞) give a separation of this space.
Hence, by the above lemma, the punctured euclidean space and R − {f (0)}
are not homeomorphic, a contradiction. Therefore, we conclude that Rn and
R are not homeomorphic. 

1. 24.8
(a) Is a product of path-connected spaces necessarily path-connected?
Answer: Yes. Suppose X and Y are path-connected. Let x1 × y1 , x2 ×
y2 ∈ X ×Y . Now, we know that X ×y1 is homeomorphic to X and, therefore,
is path-connected. Hence, there exists a continuous map f : [0, 1] → X × y1
such that
f (0) = x1 × y1 f (1) = x2 × y1 .
Also, x2 × Y is homeomorphic to Y and is, therefore, path connected, so
there exists a continuous map g : [0, 1] → x2 × Y such that
g(0) = x2 × y1 g(1) = x2 × y2 .
 TOPOLOGY HW 4 3

Now, define
f x2 0 ≤ x ≤ 12


h(x) =
g x2 + 12 1


2 ≤ x ≤ 1.
By the pasting lemma, then, h is continuous. Furthermore,
 
0
h(0) = f = f (0) = x1 × y1
2
and  
1 1
h(1) = g + = g(1) = x2 × y2 .
2 2
Hence, h is a path from x1 × y1 to x2 × y2 . Since our choice of x1 × y1 and
x2 × y2 was arbitrary, we see that X × Y is path-connected.
♣
(b) If A ⊂ X and A is path-connected, is A necessarily path connected?
Answer: No. In Example 7, we saw that, if A = {x×sin(1/x)|0 < x ≤ 1},
then A, the topologist’s sine curve, is not path connected. To see that A is
path-connected, let s, y ∈ A. Then x = a × sin(1/a) for some a ∈ (0, 1] and
y = b × sin(1/b) for some b ∈ (0, 1]. Define the map f : [a, b] → A by
f (z) = z × sin(1/z).
Then f is continuous since its coordinate functions are continuous and
f (a) = a × sin(1/a) = x and f (b) = b × sin(1/b) = y, so f is a path
from x to y. Since our choice of x and y was arbitrary, we see that A is
path-connected.
♣
(c) If f : X → Y is continuous and X is path-connected, is f (X) neces-
sarily path connected?
Answer: Yes. Let f be continuous and let y1 , y2 ∈ f (X). Let x1 ∈
f −1 (y1 ) and x2 ∈ f −1 (y2 ). Then, since X is path-connected, there exists a
continuous map g : [a, b] → X such that g(a) = x1 and g(b) = x2 . Define
h := f ◦ g. Then
h(a) = (f ◦ g)(a) = f (g(a)) = f (x1 ) = y1
and
h(b) = (f ◦ g)(b) = f (g(b)) = f (x2 ) = y2 .
Furthermore, h is continuous, since f and g are, so h is a path from y1 o
y2 . Since our choice of y1 and y2 was arbitrary, we conclude that f (X) is
path-connected.
♣
T
(d)SIf {Aα } is a collection of path-connected subspaces of X and if Aα 6=
∅, is Aα necessarily path-connected?S T
Answer: Yes. Let x, y ∈ Xα and let z ∈ Aα . Then x ∈ Aβ and
y ∈ Aγ for some β and γ. Furthermore, z ∈ Aβ , z ∈ Aγ . Since Aβ is path
connected, there exists a path f from x to z. Since Aγ is path connected,
4 CLAY SHONKWILER

there exists a path g from z to y. Using the pasting lemma, we can glue
these two paths together to make a path h from x to y (much as we did in
part (a) above).
♣

25.1
What are the components and path components of R` ? What are the
continuous maps f : R → R` ?
Answer: Each component and path component of R` consists of a single
point. To see this, suppose not. Then there exists a component containing
two points, x and y. This means there exists a connected subspace A ⊂ R`
containing x and y. However, x ∈ A ∩ (−∞, y) and y ∈ A ∩ [y, ∞), both
of these subsets are open in A, and A is equal to their union, so A can
be separated by A ∩ (−∞, y) and A ∩ [y, ∞), a contradiction. Hence, we
conclude that each component (and, therefore, path component) of R` is a
point.
Furthermore, since the continuous image of a connected space is con-
nected, we can conclude that the continuous maps f : R → R` are just the
constant maps. This is because R is connected, so it’s continuous image
in R` must be connected. The only connected subspaces of R` are single
points, so such a continuous map must map all of R to a single point.
♣

26.1
(a) Let τ and τ 0 be two topologies on the set X; suppose that τ 0 ⊃ τ .
What does compactness of X under one of these topologies imply about
compactness under the other?
Answer: If X is compact under τ 0 , then it must be compact under τ . To
see this, suppose A is an open cover of X in τ . Then A is also an open cover
of X in τ 0 , since every open set in τ is open in τ 0 . Since X is compact under
τ 0 , A contains a finite subcover of X. Since our choice of A was arbitrary,
we conclude that X is compact under τ .
On the other hand, if X = R, τ is the trivial topology and τ 0 is the
discrete topology, then τ 0 ⊃ τ and X is compact under τ , but not under
τ 0 . To see this last, we merely construct the open cover {{x}|x ∈ R}, which
certainly contains no finite subcover.
♣
0
(b) Show that if X is compact Hausdorff under both τ and τ , then either
τ and τ 0 are equal or they are not comparable.
Proof. Suppose τ 6= τ 0 and that τ and τ 0 are comparable. Suppose, without
loss of generality, that τ ( τ 0 . Then there exists U ∈ τ 0 such that U ∈
/ τ.
Therefore, X − U is not closed in X under τ . Since X is compact under
τ , the contrapositive of Theorem 26.3 implies that X − U is not compact.
 TOPOLOGY HW 4 5

Let {Uα } be an open cover of X − U in τ . Since τ ⊂ τ 0 , Uα ∈ τ 0 for all

α. Furthermore, since U ∈ τ 0 , X − U is closed in X under τ 0 . Therefore,
by Theorem 26.2, it is compact, so the open cover {Uα } contains a finite
subcover {Uαi }ni=1 . Each Uα ∈ τ , so it is certainly true that Uαi ∈ τ for all
i = 1, . . . , n. That is to say,
{Uαi }ni=1 ⊆ {Uα }α∈J
is an open cover of X − U in τ . In other words, the open cover {Uα }
contains a finite subcover. Since our choice of open cover was arbitrary, we
conclude that, in fact, X −U is compact under τ , a contradiction. From this
contradiction, we conclude that either τ = τ 0 or τ and τ 0 are not comparable.


26.8
Let f : X → Y ; let Y be compact Hausdorff. Then f is continuous if and
only if the graph of f ,
Gf = {x × f (x)|x ∈ X}
is closed in X × Y .
Proof. (⇒) Suppose f is continuous. To show Gf is closed, it suffices to show
that X × Y − Gf is open in X × Y . Therefore, let x0 × y ∈ (X × Y ) − Gf .
Clearly, y 6= f (x0 ) so, since Y is Hausdorff, there exist open neighborhoods
Vy and Vf (x0 ) of y and f (x0 ), respectively, such that
Vy ∩ Vf (x0 ) = ∅.
Since f is continuous, there exists an open neighborhood U of x0 such that
f (U ) ⊆ Vf (x0 ) .
Note that this implies that f (U ) ∩ Vy = ∅. Also, note that x0 × y ∈ U × Vy
and that U × Vy is open in X × Y .
Now, we want to show that U × Vy ⊆ (X × Y ) − Gf . Let z × w ∈ U × Vy .
Then, since f (U ) and Vy are disjoint, w ∈ / f (U ). However, since z ∈ U ,
f (z) ∈ f (U ). Therefore, w 6= f (z), so
z×w ∈
/ Gf .
In other words,
z × w ∈ (X × Y ) − Gf .
Therefore, we can conclude that, indeed, (X × Y ) − Gf is open, meaning
that Gf is closed in X × Y .
(⇐) On the other hand, suppose that Gf is closed in X × Y . Let x0 ∈ X
and let V be an open neighborhood of f (x0 ) in Y . Since V is open in Y ,
Y − V is closed and so, since X closed in X, x × (Y − V ) is closed in X × Y .
Since Gf closed in X × Y ,
Gf ∩ (X × (Y − V ))
6 CLAY SHONKWILER

is closed in X × Y . Since the projection π1 : X × Y → X is closed,

U = π1 (Gf ∩ (X × (Y − V ))) = {x ∈ X|f (x) ∈
/ V}

is closed in X. Note that U = X − f −1 (V ), so we see that f −1 (V ) is open

in X. Also, x ∈ f −1 (V ) and f (f −1 (V )) ⊆ V . Hence, f is continuous at
x. Since our choice of x was arbitrary, we see that f is continuous at every
point in X, which is to say that f is continuous.
Having demonstrated both directions, we conclude that f is continuous
if and only if it’s graph is closed in X × Y . 

27.6
(a) Show that the Cantor set C is totally disconnected.

Proof. Suppose not. Then there is an interval [a, b] ⊆ C. Let N ∈ N such

that  
1
N > log3 .
b−a
Then, for n > N , 31n < b − a. However, since C = ∩Aj , C, if it contains
intervals at all, must contain intervals of length less than 31n . Hence, C
contains no intervals, so C is totally disconnected. 

(b) Show that C is compact.

Proof. We show, by induction, that each An is closed in [0, 1]. Clearly,

A0 = [0, 1] is closed in [0, 1]. Now, suppose Ak is closed. Then Ak = [0, 1]−U ,
where U is open. Define
∞  
[ 1 + 3i 2 + 3i
Vk = , .
3k+1 3k+1
i=0

Then
Ak+1 = Ak − Vk = ([0, 1] − U ) − Vk = [0, 1] − (U ∪ Vk ) .
Since U and Vk are open, so is U ∪Vk , so Ak+1 is closed. Hence, by induction,
An is closed for all n ∈ N. Therefore, since C is an intersection of closed
sets, C is closed. C is also clearly bounded, so, by Theorem 27.3, C is
compact. 

(c) Show that each set An is a union of finitely many disjoint closed
intervals of length 1/3n ; and show that the end points of these intervals lie
in C.

Proof. By induction. Clearly, A0 = [0, 1] is the union of a single closed

interval of length 1 = 1/30 . 0 ∈ C, since 0 < 1+3k
3n for all n, k ∈ N ∪ {0}.
 TOPOLOGY HW 4 7

Also, 3n−1 ∈ N for all n, so 3n−1 − 32 ∈

/ N. Hence, there exists k ∈ N such
that
k < 3n−1 − 23 < k + 23
3n −2
k < 3 < k + 23
3k < 3n − 2 < 3k + 2
n
2 + 3k < 3 < 4 + 3k
2+3k 1+3(k+1)
3n < 1 < 3n
for all n ∈ N. Hence, the endpoint 1 ∈ C.
Now, suppose Ak−1 is a union of finitely many disjoint closed intervals
of length 1/3k−1 and that the endpoints of these intervals lie in C. Denote
Ak−1 by
n
[
Ak−1 = [aj , bj ].
1
Note that, if (c, d) ⊂ [a, b],
[a, b] − (c, d) = [a, c] ∪ [d, b].
Now,  
Ak = Ak−1 − ∞ 1+3j 2+3j
S
j=0 3k
, 3k
Sn S∞  1+3j 2+3j 
= 1 [ai , bi ] − j=0 3k , 3k
S  
= ∞ n 1+3j 2+3j
T
j=0 [a
1 i i, b ] − 3k
, 3k
which is simply an intersection of unions of closed intervals, which is itself
a union of closed intervals. Furthermore, these closed intervals have length
1/3k since we are deleting the middle third from intervals of length 1/3k−1 .
Finally, an argument similar to the one used to show that 1 ∈ C shows that
each of these endpoints is in C. Therefore, by induction, each set An is a
union of finitely many disjoint closed intervals of length 1/3n and the end
points of these intervals lie in C. 
(d) Show that C has no isolated points.
Proof. Since every basic open set in C is of the form (a, b) ∩ C where (a, b)
is open in [0, 1], it suffices to show that no set of this form is a singleton
set (a simple modification takes care of basis elements of the form [0, b) and
(a, 1]). Let (a, b) ∈ [0, 1] such that (a, b) ∩ C 6= ∅. Let x ∈ (a, b) ∩ C. Since
we showed in part (c) above that each An is a finite union of closed intervals
of length 1/3n and that the endpoints of these intervals are in C, we can
find such an endpoint not equal to x that is in (a, b) ∩ C.  
Let p = min{x − a, b − x}. Choose N ∈ N such that N > log3 p1 . Then,
for m > N , x ∈ Am , so x lies in an interval [c, d] of length 1/3m . Note that
max{x − c, d − x} < p,
so [c, d] ⊂ (a, b). From (c), we know that c, d ∈ C, so
{x, c, d} ∈ (a, b) ∩ C.
8 CLAY SHONKWILER

Even if x is equal to one of these endpoints, we still see that (a, b) ∩ C is not
a singleton set.
Since our choice of basis element (a, b) was arbitrary, we conclude that no
singleton set is open in C, meaning C has no isolated points. 

(e) Show that C is uncountable.

Proof. C is clearly non-empty, since we showed in part (c) that it contains

endpoints of closed intervals; specifically, 0 ∈ C. We showed in (b) that C
is compact and in (d) that C has no isolated points. Furthermore, we know
that C is Hausdorff since [0, 1] is. Therefore, we can use Theorem 27.7 to
conclude that C is uncountable. 

A
Let {Xα }α∈J be an arbitrary collection
Q of connected topological spaces.
Show that in the product topology X = α∈J Xα is connected.
Q
Proof. Let (yα ) = y ∈ Xα . Let Aβ = {(aα )α∈J |aα = yα for all α 6= β}.
Then let
[
A= Aβ .
β∈J

Since y ∈ Aβ for all β ∈ J and each Aβ is connected, A is connected. Now,

Q Q
we want to show that α∈J Xα = A. Let x ∈ Xα and let
Y
Ux = Uα
α∈J

be a basis element of the product topology containing x. Then Uα = Xα for

all but finitely many α. Let α1 , . . . αn be the indices for which this equality
does not hold. Let
a = (aα )α∈J
where aαi = xαi and aγ = yγ for all γ ∈
/ {α1 , . . . , αn }. Then, clearly,
n
[
a∈ Aα j ⊂ A
1

and, since yγ ∈ Uγ = Xγ for all γ ∈

/ {α1 , . . . , αn },

a ∈ Ux .
S
Hence, we can conclude that α∈J Xα is connected, since it is the closure
of a connected set. 
 TOPOLOGY HW 4 9

B
Suppose f : (0, 1) → (0, 1) is a continuous map. Does f have a fixed
point?
Answer: No. As a counter-example, consider f (x) = x2 . For c ∈ (0, 1),
c − f (c) = c − c2 = c(1 − c) < c
since 0 < 1 − c < 1. Hence, f (c) 6= c for all c ∈ (0, 1), so f has no fixed
points.
♣
DRL 3E3A, University of Pennsylvania
E-mail address: [email protected]