MTH 304 Final Solutions
1. Consider the multiplicative matrix groups
a b 4
GL(2, R) = | (a, b, c, d) ∈ R and ad − bc 6= 0
c d
and
a b 4
SL(2, R) = | (a, b, c, d) ∈ R and ad − bc = 1 .
c d
(a) Show that GL(2, R) and SL(2, R) are topological groups.
(b) Show that GL(2, R) is homeomorphic to an open subspace of R4 ,
while SL(2, R) is homeomorphic to a closed subspace of R4 . [Hint:
Consider the determinant map Det : R4 → R.]
(c) Show that GL(2, R) is not connected, and SL(2, R) is noncom-
pact.
Solution. (a) As both G and H are subsets of R4 , they inherit the
subspace topology from the standard topology in the ambient space.
We know from MTH 301 that both G = GL(2, R) and H = SL(2, R)
are multiplicative groups, and H E G. So it suffices to show that G
is a topological group. In other words, we need to establish that the
matrix product operation
ϕ
ϕ : G × G → G : (A, B) 7−
→ AB,
and the matrix inversion operation
I
7− A−1
I:G→G : A→
are continuous maps. We know that for two matrices
a1 b1 a2 b2
A= ,B= ∈ G,
c1 d1 c2 d2
we have
a1 a2 + b1 c2 a1 b2 + b1 d2
ϕ(A, B) = AB = ,
c1 a2 + d1 c2 c1 b2 + d1 d2
1
a b
and for C = ∈ G, we have
c d
−1 1 d −b
I(C) = C = .
ad − bc −c a
Viewing ϕ as a map
ϕ
R4 × R4 → R4 : (A, B) 7−
→ ((AB)11 , (AB)12 , (AB)21 , (AB)22 ),
(where A = (a1 , b1 , c1 , d1 ), B = (a2 , b2 , c2 , d2 ) ∈ R4 ) we see that its
four component functions of the form (A, B) 7→ (AB)ij are multivari-
able polynomials, and hence ϕ is continuous. Similarly, viewing I as a
function R4 → R4 , we see that its four components functions are ratio-
nal functions (i.e. polynomial/polynomial), which have the a common
denominator ad − bc 6= 0. Hence, I is a continuous function, and this
shows that G is a topological group.
(b) Now consider the determinant map
a b Det
Det : G → R : 7−−→ ad − bc.
c d
Once again, viewing Det as a map R4 → R, we see that as ad − bc is
a polynomial, and so Det is continuous. It is apparent that as R is a
T1 space,
A = Det−1 (R \ {0}) and Det−1 ({1}),
from which (b) follows.
(c) The disconnectedness of G follows from the fact that the open
subsets
Det−1 ((−∞, 0)) and Det−1 ((0, ∞))
form a separation for G. To see the noncompactness of H, it suffices
to show that H is unbounded under the standard metric in R4 (by the
Heine-Borel property). For any n ∈ N, consider the matrix An ∈ H
defined by
n 0
An = .
0 1/n
p
Then kAn k = n2 + 1/n2 , and
lim kAn k = ∞,
n→∞
which shows that {An | n ∈ N} is an unbounded subset of H. As H
has an unbounded subset, it is unbounded.
2
2. Show that if X is separable, then every collection of disjoint open sets
in X is countable.
Solution. Let A be a countable dense subset of X such that Ā = X.
Let {Uα }α∈J be an arbitrary collection of disjoint open sets in X.
Since Ā = X, for each α, there exists an xα ∈ A ∩ Uα . Moreover,
the fact that the open sets in {Uα }α∈J are mutually disjoint implies
that xα 6= xβ , whenever α 6= β. As A is countable, Y = {xα | α ∈ J}
is a countable subset of X. Since Y is bijective with J, J has to be
countable.
3. (a) Define the one-point compactification of a locally compact Haus-
dorff space.
(b) Show that the open point compactification of N is homeomorphic
to {1/n | n ∈ N} ∪ {0}.
Solution. (a) See 1.2 (xxx) in the Lesson Plan.
(b) Let K = {1/n | n ∈ N}. The inversion map
ι
ι : R \ {0} → R \ {0} : x →
7− 1/x
is a homeomorphism, as it is a rational function. This shows that
ι|N : N → K
is a homeomorphism. As K is a closed subset of the locally compact
Hausdorff space R, K is locally compact. Hence, it follows from 1.2
(xxxi) that K has a one-point compacitification. But that fact that
K = K ∪ {0} is compact space (being a closed and bounded subspace
of R) and Hausdorff implies that K is the unique one point compact-
ification of K, up to homeomorphism.
Let X ∗ = N ∪ {∞} be the one-point compactification of X = N. Then
by defining ι(∞) = 0, the map ι extends to a bijective map
ι̂ : X ∗ → Y,
where Y = K ∪ {0}. Since X ∗ compact and Y is Hausdorff, it suffices
to show that ι̂ is continuous, and in particular, ι̂ is continuous at ∞.
Let U be a neighborhood of 0 in Y . Then by definition Y \ U is
compact, which implies that ι̂−1 (Y \ U ) = ι−1 (K \ U ) is compact, and
so
ι̂−1 (U ) = X \ ι̂−1 (Y \ U )
is a open neighborhood of ∞, which is mapped into U . This shows
that ι̂ is continuous, and hence a homeomorphism.
3
4. Let X be a nonempty compact Hausdorff space without isolated points.
(a) Show that for each nonempty open U ⊂ X, and x ∈ X, there
exists an open set V ⊂ U such that x ∈
/ V.
(b) Show that there exists no surjective map f : N → X. [Hint:
Consider x1 = f (1) and U = X and apply (a) to get a V . Now
take x2 = f (2) and U = V , and so on. Finally, use the finite
intersection property.]
Solution 1. The argument is analogous to the one used in the proof
of Theorem 27.7 (Page 174) in Munkres.
5. If every real-valued continuous function on a metric space X is bounded,
then show that X is compact. [Hint: If X is not compact, consider
a sequence (xn ) with no covergent subsequence. Consider the map
xn 7→ n, and use the Tietze extension theorem.]
Solution. We know that a metric space is compact iff its sequentially
compact. Suppose that X is not compact. Then X is not sequentially
compact, which implies that there exists a sequence (xn ) in X that has
no convergent subsequence. This implies that the set A = {xn | n ∈ N}
has no limit points, and so A is closed in X. Define a function
f
f : A → N(⊂ R) : xn 7−
→ n.
Then clearly, f is a continuous, as its a bijective map between two
discrete subspaces. Moreover, as X is metrizable, it is normal, and by
the Tietze’s extension theorem, f extends to a continuous map
fˆ : X → R,
which is clearly unbounded.
6. Consider the standard quotient map q : R2 → S 1 × S 1 (≈ R2 /Z2 )
induced by the equivalence relation ∼ on R2 defined by
(x1 , y1 ) ∼ (x2 , y2 ) ⇐⇒ (x2 − x1 , y2 − y1 ) ∈ Z2 .
Consider the multiplicative matrix group
a b 4
SL(2, Z) = | (a, b, c, d) ∈ Z and ad − bc = 1 .
c d
4
a b
For a fixed matrix A = ∈ SL(2, Z), define a map MA : R2 →
c d
R2 given by
a b x ax + by
MA ((x, y)) = = ,
c d y cx + dy
for any (x, y) ∈ R2 .
(a) Show that MA induces a map M fA : S 1 × S 1 → S 1 × S 1 , which is
a homeomorphism.
(b) Show that for a fixed A ∈ SL(2, Z), we have q ◦ MA = M fA ◦ q.
Solution. (a) & (b) We know that by definition, q((x, y)) =
[(x, y)], where [(x, y)] denotes the equivalence class of (x, y) under
∼, given by
[(x, y)] = {(x + p, y + q) | p, q ∈ Z}.
We define
fA : S 1 × S 1 → S 1 × S 1
M
by
fA ([(x, y)]) = [MA ((x, y))], for all [(x, y)] ∈ S 1 × S 1 .
M
Then M fA is well-defined, for if [(x, y)] = [(x0 , y 0 )], then there
exists p, q ∈ Z such that (x + p, y + q) = (x0 , y 0 ), so that
fA ([(x0 , y 0 )]) =
M [MA (x0 , y 0 )]
= [MA (x + p, y + q)]
= [a(x + p) + b(y + q), c(x + p) + d(y + q)]
= [MA (x, y) + MA (p, q)]
= [MA (x, y)]
= M
fA ([(x, y)].
Moreover, for all (x, y) ∈ R2 , we have that
(q ◦ MA )((x, y)) = q(MA ((x, y))
= [MA ((x, y))]
= M
fA ([x, y])
= M
fA (q((x, y))
= fA ◦ q)((x, y)).
(M
5
This shows that the diagram
M
R2 A
−−−−→ R2
p
y
p
y (*)
M
S 1 × S 1 −−−−
A
→ S1 × S1
f
is commutative.
Since A ∈ SL(2, Z), MA is an invertible linear map, hence a
homeomorphism. Now for any [(x0 , y 0 )] ∈ S 1 × S 1 , there exits
(x, y) ∈ R2 such that MA ((x, y)) = (x0 , y 0 ), which implies that
fA ([(x, y)]) = [MA ((x, y))] = [(x0 , y 0 )],
M
and so it follows that M
fA is surjective. Moreover, we have that
M fA ([(x0 , y 0 )])
fA ([(x, y)]) = M =⇒ [MA ((x, y))] = [MA ((x0 , y 0 ))]
=⇒ (MfA ◦ q)((x, y)) = (M fA ◦ q)((x0 , y 0 ))
=⇒ (q ◦ MA )((x, y)) = (q ◦ MA )((x0 , y 0 ))
=⇒ [MA ((x, y))] = [MA ((x0 , y 0 ))]
=⇒ [(x, y)] = [(x0 , y 0 )],
from which the injectivity of M fA follows.
Since q is a open and continuous map (why?), for an open set U ⊂
S 1 × S 1 , we have that q −1 (U ) is open in R2 , so that MA (q −1 (U )
is open in R2 , and so
(q ◦ MA ◦ q −1 )(U ) = M
fA (U )
is open in S 1 × S 1 . Finally, for an open set V in S 1 × S 1 , we have
that
f−1 (V ) = (q ◦ MA ◦ q −1 )−1 (V ) = (q ◦ M −1 q −1 )(V ),
M A A
which is open in S 1 × S 1 . This shows that M
fA is a homeomor-
phism.
Solution 2. Alternatively, one could consider the diagram
R2 (**)
q◦MA
q
&
M
/ S1 × S1
fA
S1 × S1
6
that is equivalent to the diagram (*) above. As the map q ◦ MA
is constant on each fiber of q (why?), by 1.10 (x) of the Lesson
Plan, there exists a map M fA as indicated in (**), that makes
the diagram commute. Consequently, (*) is also commutative.
Moreover, since q and MA are continuous maps, we have that
q ◦ MA is continuous, and once again, 1.10 (x) would imply that
M
fA is continuous.
The fact that q is an open map (from class) and MA is a homeo-
morphism implies that q◦MA is a surjective, continuous and open
map, and hence a quotient map. Finally, by 1.10 (xi), q ◦ MA will
induce a homeomorphism S 1 × S 1 → S 1 × S 1 , which in this case
is precisely the map MfA .