Math F651: Homework 9 Solutions April 6, 2007

1. Munkres 26.12
Solution:
Lemma 1. Let p  X Y be a closed map. Suppose y > Y, and suppose U is an open set in X
containing p−1 (y). Then there exists an open set W in Y containing y such that p−1 (W) b U.
Proof. Consider U c . This is a closed set, so p(U c ) is closed. Moreover, since p−1 (y) b U, we
have y ~i np(U c ). Let W be any open set in Y such that y > W b (p(U c ))c . Then
p−1 (W) b p−1 ((p(U c ))c ) = (p−1 (p(U c )))c .
But p−1 (p(U c )) a U c , so (p−1 (p(U c )))c b (U c )c = U. Hence p−1 (W) b U, as required.

Solution:
Suppose p  X Y be a closed surjective perfect map, and suppose Y is compact. We want
to show that X is compact, so let ˜Uα be an open cover of X. For every y > Y we have
p−1 (y) is compact, so there exists a finite subcover C y of p−1 (y). By the lemma there exists
an open set Wy such that y > Wy and p−1 (Wy ) b 8U>C y U. Since Y is compact, we can cover
Y with finitely many sets Wy k (this step uses the fact that p is surjective). But then the union
8C y k is a finite subcover of X.

2. Munkres 29.5
Solution:
Let X and Y be homeomorphic locally compact Hausdorff spaces, with ϕ being a home-
omorphism. Let ϕ‡ be the extension to X ‡ such that ϕ(‡X ) = ‡Y . We claim that ϕ‡ is
continuous. If U b Y ‡ is open, and if ‡Y >~ U, then (ϕ‡ )−1 (U) = ϕ−1 (U), which is open in X
and hence open in X ‡ . On the other hand, if ‡Y > U, then U c = K, where K is compact in
Y. But then
((ϕ‡ )−1 (U))c = (ϕ‡ )−1 (U c ) = (ϕ‡ )−1 (K) = (ϕ)−1 (K)
which is compact in X since ϕ−1 is continuous. Hence (ϕ‡ )−1 (U) is open in X. Hence ϕ‡ is
continuous. Since it is a continuous bijection from a compact space to a Hausdorff space, it
is a homeomorphism.

3. Munkres 29.6 The following lemma will be handy for the following problems.
Lemma 2. If X is a compact Hausdorff space and U = X − ˜p for some p > X, then U ‡ is
homeomorphic to X.
Proof. Define ϕ  U ‡ X by ϕ(x) = x for x x ‡, and ϕ(‡) = p. This is clearly a bijection
from a compact space to a Hausdorff space, so to show it is a homeomorphism, we need only
show it is continuous. Clearly if V is open in X ‡ and does not contain p, then ϕ−1 (V ) = V
which is open in U. On the other hand, if V is open in X ‡ and contains p, then V c is compact
in X and contained in U, and hence compact in U. But then ϕ−1 (V c ) is compact in U. But
then
ϕ−1 (V ) = (ϕ−1 (V c ))c
which is the complement of a compact set in U and therefore an open set in U ‡ . Hence ϕ is
continuous.
 Math F651: Homework 9 Solutions April 6, 2007

Corollary 3. Let Z be a compact Hausdorff space. Suppose Y = Z − ˜p and X is homeomor-

phic to Y. Then X ‡ is homeomorphic to Z.
Proof. By Munkres 29.5 we have X ‡ is homeomorphic to Y ‡ , and by Lemma 2 we have Y ‡
is homeomorphic to Z.

Solution:
Since R is homeomorphic to (0, 2π), it is enough to show that the one point compactification
is homeomorphic to S 1 . Define ϕ  (−π, π) S 1 by ϕ(x) = e 2πix . This map is clearly clearly
injective, and is surjective onto S 1  ˜(−1, 0). Moreover, the branch of ln with branch cut
along then negative real axis is a continuous function and, when restricted to S 1  ˜(−1, 0)
is the inverse function of ϕ. Hence (−π, π) is homeomorphic to S 1  ˜(−1, 0). Since S 1 is
a compact Hausdorff space, we conclude that the one point compactification of (−π, π) is
homeomorphic to S 1 .

4. Munkres 29.8
Solution:
Define f  R+ R+ by f (x) = 1~x. This is a homeomorphism (it is continuous and is its
own inverse). Let A = f (Z+ ). Then A is homeomorphic to Z+ and is ˜1~n  n > Z+ . Let

Y =08A

This is a closed and bounded subset of R and is hence compact. It is also a Hausdorff space.
Therefore the one point compactification of A is homeomorphic to Y, and hence the one
point compactification of Z+ is homeomorphic to Y.

5.

a) Show that every manifold is a locally compact Hausdorff space.

b) Show that the one-point compactification of an n−manifold M is an n−manifold if and

only if there is an precompact open subset V b M such that M  V is homeomorphic
to Rn  B1 (0).

Solution, part a:
For each x > M, there is an open set U x that is homeomorphic to Rn via a homeomorphism
ϕ x taking x to 0. Let K = ϕ−1
x (B1 (0)). Since B1 (0) is compact in R we have K is compact in
n

U. Hence K is compact in M and therefore closed in M (since M is Hausdorff). Moreover,

K contains ϕ−1 x (B1 (0)) which is open in U x (and hence open in M since U x is open) and
contains x. Hence we have shown that every point x in M has a compact set that contains
an open set containing x. Hence M is locally compact. It is Hausdorff by definition.
Solution, part b:
Let M be an n−manifold. Then M ‡ is a compact Hausdorff space and is hence compact.
Moreover, every compact locally Euclidean space is second countable. So it is enough to

2
 Math F651: Homework 9 Solutions April 6, 2007

show that M ‡ is locally Euclidean if and only if there is an precompact open subset V b
M such that M  V is homeomorphic to Rn  B1 (0). Finally, every point in M 9 M ‡ is
contained in an open (in M and therefore in M ‡ ) set that is homeomorphic to Rn . Hence
we have reduced the problem to showing that ‡ has a neighbourhood homeomorphic to Rn
if and only if if and only if there is an precompact open subset V b M such that M  V is
homeomorphic to Rn  B1 (0).
Suppose such a set V exists. Let Y = 0M V . Then Y is homeomorphic to Rn  B1 (0), which
2
is homeomorphic to the closed punctured ball B1 (0)  ˜0 via the map x ( x~ SxS . Since
B1 (0) is compact, we conclude that the one-point compactification of Y is homeomorphic
to B1 (0) by a homeomorphism taking ‡Y to 0. s Let K = M ‡  V . This is a closed set in M ‡
and is hence compact. Since Y is homeomorphic to (M ‡  V )  ˜‡M  we conclude that Y ‡
is homeomorphic to M ‡  V by a homeomorphism taking ‡Y to ‡M .
Composing homeomorphisms, we conclude that there is a homeomorphism ϕ from M ‡ V
to B1 (0) taking ‡M to 0.
Let W = M ‡  V̄ . This set is open in M ‡ and contained in M ‡  V . It is therefore open
in M ‡  V . Hence ϕ(W) is an open set in B1 (0) that contains 0, and hence contains some
Bє (0). Let U = ϕ−1 (Bє (0)). This set is in M ‡  V̄ and is contained in W. Since W is open in
M ‡ , we conclude that U is open in M ‡ . But U is homeomorphic to Bє (0) (by ϕ restricted to
U) and hence ‡M has a neighbourhood that is homeomorphic to and open subset of Rn .
Conversely, suppose that M ‡ is locally Euclidean. Let U be an open set in M ‡ that contains
‡ and is homeomorphic to Rn by a homeomorphism ϕ. Without loss of generality we can
assume that ϕ(‡) = 0. Let W = ϕ−1 (Rn  B1 (0). This is an open set in U and hence open
in M ‡ . It does not contain ‡, so it is also open in M. Let V = (M ‡  U) 8 W. This is an
open subset of M. All the limit points of V in M ‡ must be in U, since V contains M ‡  U.
But it is easy to see that these are exactly ϕ−1 (S n−1 . Hence the closure of V in M ‡ must be
(M ‡  U) 8 ϕ−1 (Rn  B1 (0). This is compact, since its complement is ϕ−1 (B1 (0)), which is
an open subset of ‡. Hence V is a precompact open neighbourhood in M ‡ , and its closure
does not contain ‡. Hence it is a precompact open neighbourhood in M, and M  V is
homeomorphic to B1 (0)  0 which is homeomorphic to Rn  B1 (0). So V is the set we seek.

6. Munkres p.145 Number 1.

Solution:
Let G be a topological group and let µ and ι be the multiplication and inversion maps. Since
the identity map id and the inversion map ι are continuous, we have id ι  G  G G  G
is continuous. Hence µ X (id ι)  G  G G is continuous. But µ X (id ι)(x, y) = xy −1 .
Hence the map taking x and y to xy −1 is continuous.
Conversely, suppose the map taking x and y to xy −1 is continuous. Let g denote this map.
Define h  G G  G by h(x) = (1, x) where 1 is the group identity. This is a continuous
map (and indeed a topological embedding). But then ι = g X h and hence ι is continuous.
But also, µ = g X (id ι) and hence µ is also continuous.

3
 Math F651: Homework 9 Solutions April 6, 2007

7. Munkres p.146 Number 3.

Solution:
Let H be a subgroup of G. Since inversion and mutiplication both take H and H  H respec-
tively to H, and since these are continuous maps from G and G  G to G, their restrictions
to H are also continuous. Hence H is a topological group.
We now show that H is a subgroup. By the above we will have shown that it is a topological
group. Suppose h > H and consider L h  G G given by L h (g) = h ċ g. This map is a
homeomorphism, so
H̄ = L h (H) = L hœ (H̄).
We conclude that if h > H and h œ > H̄, then h ċ h œ > H̄. Similarly, h œ ċ h > H̄.
Now suppose h œ > H̄. Then by the above we see that L hœ (H) is a subset of H̄. Hence

H̄ = H̄ c L hœ (H) = L hœ (H̄).

Since h œ > H̄ is arbitrary, we conclude that the product of two elements of H̄ is again in H̄.
Now inversion ι  G G, which is a homeomorphism. Then

ι(H̄) = ι(H) = H̄

since ι(H) = H. Hence the inverse of an element of H̄ is again in H̄.