Problem Set 1

1. (see also Folland 4.15) If X is a topological space, A ⊂ X is closed, and g ∈ C(A) satisfies
g = 0 on ∂A, then the extension of g to X defined by g(x) = 0 for x ∈ Ac is continuous.
2. (similar to Folland 4.32) correction: Prove that if X is Hausdorff then every se-
quence in X converges to at most one point.
3. (see also Folland 4.57) An open cover U of a topological space X is called locally finite if
each x ∈ X has a neighborhood that intersects only finitely many members of U. If U, V
are open covers of X, V is a refinement of U if for each V ∈ V, there exists U ∈ U with
V ⊂ U . X is called paracompact if every open cover of X has a locally finite refinement.
We say that X is σ-compact if it is the countable union of compact sets.
(a) If X is a σ-compact LCH space, then X is paracompact. In fact, every open cover U
has locally finite refinements {Vα } and {Wα } so that Vα is compact and Wα ⊂ Vα for
all α.
(b) If X is a σ-compact LCH space, for any open cover U of X there is a partition of unity
on X subordinate to U and consisting of compactly supported functions.
4. (see also Folland 1.5) If M is a σ-algebra generated by E, then M is the union of the
σ-algebras generated by F as F ranges over all countable subset of E.
5. (see also Folland 1.12) Let (X, M, µ) be a finite measure space.
(a) If E, F ∈ M and µ(E∆F ) = 0, then µ(E) = µ(F ).
(b) Say that E ∼ F if µ(E∆F ) = 0; then ∼ is an equivalence relation on M.
(c) For E, F ∈ M, define ρ(E, F ) = µ(E∆F ). Then ρ(E, G) ≤ ρ(E, F ) + ρ(F, G), and
hence ρ defines a metric on the space M/ ∼ of equivalence classes.
6. (see also Folland 1.16) Let (X, M, µ) be a measure space. A set E ⊂ X is called locally
measurable if E ∩ A ∈ M for all A ∈ M with µ(A) < ∞. Let M̃ be the collection of all
locally measurable sets. Clearly M ⊂ M̃; if M = M̃, then µ is called saturated.
(a) If µ is σ-finite, then µ is saturated.
(b) M̃ is a σ-algebra.
(c) Define µ̃ on M̃ by µ̃(E) = µ(E) if E ∈ M and µ̃(E) = ∞ otherwise. Then µ̃ is a
saturated measure on M̃. It is called the saturation of µ.
(d) If µ is complete, so is µ̃.
(e) Suppose that µ is semifinite. For E ∈ M̃ , define µ(E) = sup{µ(A) | A ∈ M and A ⊂
E}. Then µ is a saturated measure on M̃ that extends µ.
(f) Let X1 , X2 be disjoint uncountable sets, X = X1 ∪ X2 , and M the σ−algebra of
countable or co-countalbe sets in X. Let µ0 be counting measure on P(X1 ), and
define µ by µ(E) = µ0 (E ∩ X1 ). Then µ is a measure on M, M̃ = P(X), and in the
notation of this problem µ̃ 6= µ.

1. Proof. Let
(
g(x) x∈A
h(x) = .
0 x ∈ Ac
1
2

Then consider h−1 ((a, b)). Suppose first that 0 ∈

/ (a, b). Then
h−1 ((a, b)) = g −1 (a, b) = U ∩ A
where U ⊂ X is open and U ∩ ∂A = ∅. Therefore,
U ∩ A = U ∩ (A \ ∂A) = U ∩ Ao
is open in X.
Next, suppose 0 ∈ (a, b). Then there exists U ⊂ X open with
h−1 ((a, b)) = g −1 (a, b) = U ∩ A ∪ Ac = U \ (U ∩ Ac ) ∪ Ac = U
is open in X. Hence, since the intervals (a, b) generate the open sets on R, h is continuous.

2. Proof. Suppose that xn → x and xn → y with y 6= x. Then for all neighborhoods, U, V
of x, y respectively, there exists N > 0 so that for n > N , xn ∈ U ∩ V . But, since X
is Hausdorff,there exist neighborhoods U, V of x and y respectively so that U ∩ V = ∅, a
contradiction. 
∞
3. (a) Proof. Suppose that {Uα }α∈A is an open cover of X and {Wj }j=1 is an increasing
family of precompact open sets sets so that W j ⊂ Wj+1 and i Wi = X. Then for
S

each j there exists α1j , . . . αnj j so that

nj
[
Wj ⊂ Uαj .
k
i=1
In particular,
nj
[
Wj ⊂ Uαj ∩ Wj+1 .
k
i=1
c
Now, each W j is closed, so Wj is open and in particular,
nj
[
W j \ W j−1 ⊂ (Uαj ∩ Wj+1 ) \ W j−1
i
i=1

and Vij := (Uαj ∩Wj+1 )\W j−1 is open. In addition, Vij is compact since it is contained
i
in the compact set W j+1 . Now, let
V = {Vij | 1 ≤ i ≤ nj , j ∈ N}.
Then
nk
W j−1 ∩ ∪∞ k
k=j ∪i=1 Vi = ∅.

In particular, each W j intersects only finitely many elements of V.

Now, suppose that x ∈ X. Then x has a compact neighborhood F . Then since Wj are
open and ∪j Wj = X, F ⊂ Wj for j large enough. Therefore, F intersects only finitely
many elements in V.
Thus we have constructed a locally finite refinement of U consisting of precompact
sets.
 3

Now, for all x ∈ X, there exists Vx ∈ V so that x ∈ Vx . Then there exists a compact
neighborhood of x, Nx with
Nx ⊂ Vx
In particular, there exists Wx open so that x ∈ Wx ⊂ Nx and
W ⊂ N ⊂ Vx .
Then ∪x∈X Wx = X and hence W 0 := {Wx }x∈X is an open cover of X. Therefore, it
has a locally finite refinement W and by construction for each W ∈ W, there exists
V ∈ V so that W ⊂ V. 
(b) Proof. Now, let {Wα } and {Vα } be locally finite open covers as above. Then let
fα ∈ Cc (X; [0, 1]) with fα ≡ 1 on W α and supp fα ⊂ Vα . Then since {Wα } and {Vα }
are locally finite, the sum
X
g= fα
α
is a finite sum at each x and hence is a well defined continuous function. Then since
Wα is an open cover g > 0 and so
fα
hα = P
α fα

is the desired partition of unity. 

4. Proof. Let
A := {F ⊂ E | F is countable}.
Then we want to show that
[
B := M(F) = M(E).
F ∈A

The inclusion M(E) ⊃ B is clear since M(E) is a σ-algebra containing E and in particular
containing F for any F ⊂ E.
For the opposite inclusion, first observe that E ⊂ B since in particular, M(E) ∈ B for
each E ∈ E. To see that B ⊃ M(E), we need to show that B is a σ-algebra.
It is clear that ∅, X ∈ B and that B is closed under complements since each M(F) is
closed under complements. It remains to check that B is closed under countable unions.
Suppose that {Fi }∞ i=1 ⊂ B. Then for each i there exists Fi so that Fi ∈ M(Fi ) and Fi is a
countable subset of E. Therefore,
!
[
Fi ∈ M Fi
i

and since each Fi is countable and we are taking a countable union,

[
G= Fi
i

is countable and hence M(G) ⊂ B. Therefore B is a σ-algebra and hence M(E) ⊂ B. 

4

5. (a) Proof.
µ(E) = µ(E \ F ) + µ(F ∩ E) = µ(E \ F ) + µ(F ) − µ(F \ E).
Observe that
µ(E∆F ) = µ(E \ F ) + µ(F \ E) = 0.
Therefore, since µ : M → [0, ∞), µ(E \ F ) = µ(F \ E) = 0. Hence,
µ(E) = µ(E \ F ) + µ(F ∩ E) = µ(E \ F ) + µ(F ) − µ(F \ E) = µ(F ).

(b) Proof. We need to check that if E ∼ F and F ∼ G, then E ∼ G. For this, observe
that
E∆G = E \ G ∪ G \ E
= (E \ F ∪ F ∩ E) \ G ∪ (G \ F ∪ G ∩ F ) \ E
⊂E\F ∪F \G∪F \E∪G\F
(1) = E∆F ∪ G∆F
Therefore,
0 ≤ µ(E∆G) ≤ µ(E∆F ) + µ(G∆F ) = 0
and E ∼ G. 
(c) Proof. It follows form (1) that
µ(E∆G) ≤ µ(E∆F ) + µ(G∆F ).
Therefore,
ρ(E, G) ≤ ρ(E, F ) + ρ(F, G).

6. (a) Proof. Suppose that µ is σ-finite and X = ∪n An with µ(An ) < ∞. Next, let E ∈ M̃.
Then E ∩ An ∈ M and hence
E = ∪n E ∩ An ∈ M.
In particular, M̃ ⊂ M. 
(b) Proof. Since M ⊂ M̃, ∅, X ∈ M̃ and thus we need only check that M̃ is closed
under countable unions, finite intersections, and taking complements. Suppose that
{Ei }∞
i=1 ∈ M̃ and let A ∈ M with µ(A) < ∞. Then,

(∪i Ei ) ∩ A = ∪i (Ei ∩ A) ∈ M
Therefore, ∪i Ei ∈ M̃. Similarly,
(∩ni=1 ) ∩ A = ∩ni=1 (Ei ∩ A) ∈ M
and hence ∩ni=1 ∈ M̃. Now,
E1c ∩ A = (E1 ∪ Ac )c = ((E1 ∩ A) ∪ (E1 ∩ Ac ) ∩ Ac )c = ((E1 ∩ A) ∪ Ac )c .
But, E1 ∩ A, Ac ∈ M. Therefore, their complement is in M and hence E1c ∈ M̃. 
 5

(c) Proof. We need to check that µ̃ is a measure on M̃. First, observe that µ̃(∅) = µ(∅) =
0. So, suppose that {Ai }∞ ∞
i=1 ⊂ M̃ are disjoint. Let A = i Ai . If {Ai }i=1 ⊂ M, then
S
X X
µ̃(A) = µ(A) = µ(Ai ) = µ̃(Ai ).
i i

Therefore, we assume that there exists j with Aj ∈

/ M. Then µ̃(Aj ) = ∞ and hence
X
∞ = µ̃(Aj ) ≤ µ̃(Ai ) ≤ ∞
i

Now, there are two cases, A ∈ M or A ∈

/ M. If A ∈
/ M, then µ̃(A) = ∞ and hence
X
µ̃(A) = µ̃(Ai ) = ∞.
i

So, assume that A ∈ M. Then µ(A) = ∞ since if A ∈ M̃ and A ⊂ B ∈ M with

µ(B) < ∞, then A ∈ M. Indeed, by part (a), this remains true even if A ⊂ ∪n Bn
with µ(Bn ) < ∞. In particular, we still have
X
µ̃(A) = µ(A) = ∞ = µ̃(Ai ).
i


(d) Proof. Suppose that A ⊂ B and B ∈ M̃ with µ̃(B) = 0. Then by definition B ∈ M
and µ(B) = 0. In particular, A ⊂ B ∈ M with µ(B) = 0. Hence, if µ is complete,
A ∈ M ⊂ M̃ and hence µ̃ is complete. 
(e) Proof. First, observe that if E ∈ M, then for all A ∈ M with A ⊂ E, µ(A) ≤ µ(E)
and hence µ(E) = µ(E). In particular, µ extends µ. To see that µ is saturated, suppose
that E ⊂ X has E ∩ A ∈ M̃ for all A ∈ M̃ with µ(A) < ∞. Then in particular, for
A ∈ M with µ(A) < ∞, E ∩ A ∈ M̃ and hence since µ(A) < ∞,
E ∩ A = (E ∩ A) ∩ A ∈ M.
That is, E ∈ M̃.
Now, it is clear that if A, B ∈ M̃, with A ⊂ B, then µ(A) ≤ µ(B).
Suppose that {Ei }∞ i=1 ⊂ M̃ are disjoint. Then let E = ∪i Ei . If µ(Ei ) = ∞ for some
i, then by monotonicity,
X
(2) µ(E) ≥ µ(Ei ) = ∞ = µ(Ei ).
i

Therefore, we may assume µ(Ei ) < ∞. First we prove superadditivity. For this, let
Bi ∈ M with µ(Bi ) + 2−i  ≥ µ(Ei ). We have
µ(E) = sup{µ(A) | A ⊂ E, A ∈ M}
X
≥ sup{ µ(Ai ) | Ai ⊂ Ei , Ai ∈ M}
i
X X
≥ µ(Bi ) ≥ µ(Ei ) − 
i i
6

Since  > 0 was arbitrary,

X
µ(E) ≥ µ(Ei ).
i

Lemma 0.1. For all E ∈ M̃, there exists An ∈ M with An ⊂ E so that µ(An ) < ∞,
and µ(An ) → µ(E).
Proof. First, suppose that µ(E) < ∞. Then by definition, there exists An ∈ M with
An ⊂ E and
µ(An ) ≤ µ(E) ≤ µ(An ) + 2−n .
Next, suppose that µ(E) = ∞ and the conclusion is false. Then there exists A∞ ⊂ E
with A∞ ∈ M and µ(A∞ ) = ∞. Moreover,
M := sup{µ(A) | A ⊂ A∞ , 0 < µ(A) < ∞} < ∞.
First, note that the set over which we take the supremum is nonempty by semifinite-
ness. Therefore, there exists An ∈ M increasing with µ(An ) ↑ M and An ⊂ A∞ .
Let B1 = ∪n An . Then B1 ∈ M with B1 ⊂ A∞ and 0 < µ(B1 ) = M . On the other
hand µ(A∞ \ B1 ) = ∞ and hence by semifiniteness there exists B2 ⊂ A∞ \ B1 so that
0 < µ(B2 ) < ∞. In particular,
∞ > µ(B1 ∪ B2 ) = µ(B1 ) + µ(B2 ) > M
but B1 ∪ B2 ⊂ A∞ , a contradiction. 
Now, let An ⊂ E with µ(An ) < ∞, µ(An ) → µ(E).
!
[
µ(An ) = µ An ∩ Ei .
i
Now, µ(An ) < ∞, so An ∩ Ei ∈ M and hence
!
[ X X
µ Am ∩ Ei = µ(An ∩ Ei ) ≤ µ(Ei )
i i i
where in the last line, we have used the definition of µ. Letting n → ∞, we have that
X
µ(E) ≤ µ(Ei )
i
as desired.

(f) Proof. To see that µ is a measure, first observe that µ(∅) = µ0 (∅) = 0. Next, suppose
that {Ei }∞
i=1 ⊂ M are disjoint and let E = ∪i Ei . Then
X X
µ(E) = µ0 (E ∩ X1 ) = µ0 (∪i (Ei ∩ X1 )) = µ0 (Ei ∩ X1 ) = µ(Ei ).
i i
since µ0 is a measure on the countable and cocountable sets on X1 and Ei ∩ X1 lies in
this σ-algebra. Indeed, if Ei is countable, then clearly so is Ei ∩X1 . If Ei is cocountable
then so is Ei ∩ X1 since X1 is uncountable.
Now, consider X1 . Then X2 ∈ / M since X2 is not countable and X1 = X \ X2 is not
countable. Suppose that A ∈ M and µ(A) < ∞. Then A ∩ X1 is a finite set and hence
 7

A is countable. In particular, A ∩ X2 is countable and hence lies in M. Therefore,

X2 ∈ M̃. However, if A ⊂ X2 and A ∈ M, then A ∩ X1 = ∅ and hence µ(A) = 0.
Therefore,
µ(X2 ) = sup{µ(A) | A ⊂ X2 , A ∈ M} = 0.
On the other hand µ̃(X2 ) = ∞.