PC2 (2023/2024)

M EASURE AND I NTEGRATION

E XERCISE S ET 03 (Solutions)

1. Let (X, 6, µ) be a measure space. Let A, B, C ∈ 6. Prove the following relations:

µ(A ∪ B) + µ(A ∩ B) = µ(A) + µ(B),

µ(A ∪ B) = µ(A ∩ B) + µ(A1B),

µ(A1B) = 0 H⇒ µ(A) = µ(B),

µ(A1B) ≤ µ(A1C) + µ(C1B).

Solution. (i) We have

A ∪ B = (A ∩ B) ∪ (A \ (A ∩ B)) ∪ (B \ (A ∩ B)) ,

where the right side is a union of 3 disjoint measurable sets. Then, by the additivity of µ, we have

µ (A ∪ B) = µ (A ∩ B) + µ (A \ (A ∩ B)) + µ (B \ (A ∩ B)) .

Then adding µ (A ∩ B) on both sides of the last equality, we have

µ (A ∪ B) + µ (A ∩ B) = µ (A \ (A ∩ B)) + µ (A ∩ B) + µ (B \ (A ∩ B)) + µ (A ∩ B)
= µ (A) + µ (B) .

(ii) By Problem 1(b) in the Exercise set 1, we have A ∪ B = (A ∩ B) ∪ (A△B). Now A ∩ B and A△B
are measurable disjoint sets. Then, by the additivity of µ, we have

µ(A ∪ B) = µ(A ∩ B) + µ(A△B).

(iii) As A1B = (A \ B) ∪ (B \ A) is the union of 2 measurable disjoint sets, we obtain

µ(A \ B) + µ(B \ A) = µ(A△B) = 0.

By the non-negativity of the measure µ, we deduce that µ(A \ B) = 0 and µ(B \ A) = 0. On the
other hand, we have

µ(A) = µ((A \ B) ∪ (A ∩ B)) = µ(A \ B) + µ(A ∩ B) = µ(A ∩ B).

Similarly, we have

µ(B) = µ((B \ A) ∪ (A ∩ B)) = µ(B \ A) + µ(A ∩ B) = µ(A ∩ B).

Finally, we deduce that µ(A) = µ(A ∩ B) = µ(B).

1
 (iv) By By Problem 1(c) in the Exercise set 1 we have A1B ⊂ (A1C) ∪ (C1B). Then, by the
subadditivity of the measure µ, we have

µ(A1B) ≤ µ((A1C) ∪ (C1B)) ≤ µ(A1C) + µ(C1B).

2. Let (X, 6, µ) be a finite measure space, i.e., µ(X ) < +∞. Let (An )n∈N , (Bn )n∈N ⊂ 6 be two sequences
of measurable sets.

(a) If Bn ⊂ An for every n ∈ N, then show that

! ! +∞
[ [ X
µ An − µ Bn ≤ (µ (An ) − µ (Bn )) .
n∈N n∈N n=0
 
(b) If µ(An ) = µ(X ) for every n ∈ N, then show that µ = µ(X ).
T
An
n∈N

Solution. (a) We proved in the lectures that

! !
[ [ [
An \ Bn ⊂ (An \ Bn ) .
n∈N n∈N n∈N

By using monotonicity and σ -subadditivity of µ, and considering that µ(X ) < +∞, we obtain
! ! ! !!
[ [ [ [
µ An − µ Bn = µ An \ Bn
n∈N n∈N n∈N n∈N
! +∞ +∞
[ X X
≤µ (An \Bn ) ≤ µ ( An \Bn ) = (µ ( An ) − µ ( Bn )) .
n∈N n=0 n=0

(b) Since µ(X ) < +∞, it follows that µ(Acn )= µ(X\ An ) = µ(X ) − µ(An ) = 0 for all n ∈ N.
Consequently, σ -subadditivity results in µ µ(Acn ) = 0. Therefore,
S c P
An ≤
n∈N n∈N
! !
\ [
µ(X ) − µ An =µ Acn = 0.
n∈N n∈N

n o
3. Let (X, 6, µ) be a probability space, i.e., µ(X ) = 1. Let M = A ∈ 6 : µ(A) ∈ {0, 1} . Show that
M is a σ -algebra on X .

Solution. (i) Since X ∈ 6 and µ(X ) = 1, it follows that X ∈ M.

(ii) Let A ∈ M. Then A ∈ 6 and µ(A) ∈ {0, 1}. Hence, Ac ∈ 6 and µ(Ac ) = µ(X \ A) =
µ(X ) − µ(A) = 1 − µ(A) ∈ {0, 1}. So, Ac ∈ M.

 {An}n∈N ⊂ 6 and µ(An ) ∈ {0, 1} for every n ∈ N. If µ(An ) = 0 for all

(iii) Let {An }n∈N ⊂ M. Then
n ∈ N. Then 0 ≤ µ µ (An ) = 0. Thus
S P S
An ≤ An ∈ M. If there exists n 0 ∈ N such
n∈N n∈N   n∈N  
that µ An 0 = 1, then 1 = µ An 0 ≤ µ An ≤ µ(X ) = 1, and so µ


S S
An = 1. Thus
S n∈N n∈N
An ∈ M.
n∈N S
In all cases we have An ∈ M.
n∈N
Conclusion. M is a σ -algebra on X .

2
 n o
4. Let X =]0, +∞[ and let C = ]k − 1, k] : k ∈ N , where N = {1, 2, . . .}. Let 6 be the family of all
arbitrary unions of members of C. For every A ∈ 6, define µ(A) to be the number of elements of C that
constitute A.

(a) Explain why 6 is a σ -algebra on X .

(b) Show that µ is a measure on 6.
6 be a decreasing
(c) Let {An }n∈N ⊂   sequence defined by An =]n, +∞[ for n ∈ N. Check that
lim µ(An ) ̸= µ lim An .
n→∞ n→+∞

Solution. (a) Notice that C is a partition of X . Then by Problem 2 from Exercise set 2, we have that 6
is a σ -algebra on X , and in addition, 6 = σ (C).
(b) It is clear that µ is non-negative and µ(∅) = 0. Let (An )n≥1 be a sequence of pairwise disjoint
elements of 6. Then for each n ∈ N there exists K n ⊂ N s.th. An = ]k − 1, k] and µ(An ) =
S
k∈K n
Card(K n ). Notice that {K n }n∈N must be disjoint as {An }n∈N is disjoint. Therefore,
[ [ [ [
An = ]k − 1, k] = ]k − 1, k],
S
n∈N n∈N k∈K n k∈ Kn
n∈N

and hence ! !
[ [ X X
µ An = Card Kn = Card(K n ) = µ(An ).
n∈N n∈N n∈N n∈N

Thus µ is a measure on 6.
 
]n, +∞[= ∅, and hence µ
T
(c) Since {An }n∈N is decreasing, it follows that lim An = lim An =
n→+∞ n∈N n→+∞
µ(∅) = 0. On the other hand, we have
[
An =]n, +∞[= ]k − 1, k], ∀n ∈ N,
k∈[n+1,+∞[∩N

which gives µ(An ) = Card([n + 1, +∞[∩N) = +∞ for every n ∈ N. So, lim µ(An ) = +∞.
  n→+∞

Thus lim µ(An ) = +∞ ̸= 0 = µ lim An .

n→∞ n→+∞

5. Suppose that X is an uncountable set. We say that a subset A ⊂ X is co-countable if its complement Ac
in X is countable. Let 6 be the family of all the countable and the co-countable subsets of X . For every
A ∈ 6, define µ(A) by 
0 if A is countable,
µ(A) =
1 if A is co-countable.
Show that µ is a measure on 6.

Solution. By problems 6 and 7, we know that 6 is a σ -algebra on X , and in addition, 6 = σ ({{x} : x ∈ X }).
Clearly, µ(∅) = 0 as ∅ is countable. Let now {An }n∈N ⊂ 6 be a pairwise disjoint sequence. Then for

3
 each n ∈ N, we have either An is countable or An is co-countable. If An is countable for every n ∈ N,
S
then An is also countable as a countable union of countable sets. Therefore,
n∈N
!
[ X
µ An =0= µ(An ).
n∈N n∈N

If there exists n 0 ∈ N s.th. An 0 is co-countable, i.e., Acn 0 is countable. Since {An }n∈N is pairwise disjoint,
it follows, for every n ̸= n 0 , that An ⊂ Acn 0 , and therefore, An is countable for every n ̸= n 0 . So,
!
[ X X
1 = µ(An 0 ) ≤ µ An ≤ µ(An ) = µ(An 0 ) + µ(An ) = 1 + 0 = 1,
n∈N n∈N n̸=n 0
 
which yields µ µ(An ). In all cases we have
S P
An =1=
n∈N n∈N
!
[ X
µ An = µ(An ).
n∈N n∈N

Thus µ is a measure on 6.

6. Let (X, 6) be a measurable space. Let (µk )k≥1 be a sequence of measures on 6 and let (αk )k≥1 be a
+∞
sequence of positive numbers. Show that µ = αk µk is a measure on 6.
P
k=1

Solution. In order to solve this problem and next problem, we need the following lemma, which is the
Fubini’s theorem for infinite series with non-negative terms.

Lemma. Let (an,k )n∈N be a double sequence of real numbers s.th. an,k ≥ 0 for all n, k ∈ N. Then
k∈N
XX XX
an,k = an,k .
n∈N k∈N k∈N n∈N

Proof of Lemma. First, notice that for every L ∈ N, we have

L
XX X
an,k = (an,1 + an,2 + · · · + an,L )
n∈N k=1 n∈N
(1)
X X X L X
X
= an,1 + an,2 + · · · + an,L = an,k .
n∈N n∈N n∈N k=1 n∈N

As the terms are all non-negative, we have

L
X X
an,k ≤ an,k , ∀L ∈ N.
k=1 k∈N

Hence, from this and (1), we obtain

L X
X L
XX XX
an,k = an,k ≤ an,k , ∀L ∈ N.
k=1 n∈N n∈N k=1 n∈N k∈N

4
 Letting L → +∞ yields
XX XX
an,k ≤ an,k . (2)
k∈N n∈N n∈N k∈N

By interchanging the role of n and k in (2), we obtain

XX XX
an,k ≤ an,k .
n∈N k∈N k∈N n∈N

Combining this with (2) proves the lemma.

Now, for our problem, we have

+∞
X +∞
X
µ(∅) = αk µk (∅) = 0 = 0.
k=1 k=1

Let {An }n∈N ⊂ 6 be pairwise disjoint. Then, by using the Lemma, we obtain
! +∞ ! +∞ +∞
[ X [ X X
µ An = αk µk An = αk µk (An )
n∈N k=1 n∈N k=1 n=1
+∞
XX +∞ +∞
XX +∞
= αk µk (An ) = αk µk (An )
k=1 n=1 n=1 k=1
+∞
X
= µ(An ).
n=1

Thus, µ is also a measure on 6.

7. Let (X, P(X )) be a measurable space, where X is countable. Show that a set-function µ : P(X ) → R+
is a measure on P(X ) if and only if there exists a sequence (αx )x∈X ⊂ [0, +∞] such that
X
µ(A) = αx , ∀A ∈ P(X ).
x∈A

Solution. Assume that µ is a measure. Since X = {x} and X is countable, it follows from the σ -
S
x∈X
additivity of µ that !
[ X
µ(A) = µ {x} = µ({x}) ∀A ∈ P(X ),
x∈A x∈A

which implies that αx = µ({x}) for every x ∈ X . Now, assume that exists a sequence (αx )x∈X such that
X X
µ(A) = αx = αx χ A (x), ∀A ∈ P(X ).
x∈A x∈X

clearly, µ(∅) = αx χ∅ (x) = 0 = 0. Let {An }n∈N ⊂ P(X ) be pairwise disjoint. We know, from
P P
x∈X x∈X
problem 2 of Exercise set 1, that
+∞
X
χ+∞
S = χ Ak .
Ak
k=0 k=0

5
 Therefore, !
∞
[ X
µ Ak = αx χ+∞
S (x)
Ak
k=0 x∈X k=0
∞
!
X X
= αx χ Ak (x)
x∈X k=0
∞
!
X X
= αx χ Ak (x) By previous Lemma
k=0 x∈X
X∞
= µ (Ak ) .
k=0
Thus µ is a measure on P(X ).

8. (Borel-Cantelli Lemma) Let (X, 6, µ) be a measure space,

 and let(An )n∈N ⊂ 6 be a sequence of
µ (An ) < +∞, then µ lim sup An = 0.
P
measurable sets. Show that if
n∈N n→+∞

+∞
µ (An ) < +∞ we deduce µ (Ak ) → 0 as n → +∞.
P P
Solution. From
n∈N k=n
+∞
Ak for every n ∈ N. Clearly {Bn }n∈N is non-increasing, and µ(B1 ) ≤ µ(An ) < +∞. So,
S P
Let Bn =
k=n n∈N
by using the properties of the measure, we obtain
  ! ! +∞
\[ \ X
µ lim sup An = µ Ak = µ Bn = lim µ(Bn ) ≤ lim µ(Ak ) = 0.
n→+∞ n→+∞ n→+∞
n∈N k≥n n∈N k=n

Supplementary Problems

1. Let (X, 6, µ) be a finite measure space. Let F = {E λ : λ ∈ 3} ⊂ 6 be a disjoint collection, where 3 is

an arbitrary index set, and that µ (E λ ) > 0 for every λ ∈ 3. Show that F is a countable collection.

2. Let (X, 6, µ) be a measure space. Let 3 is an arbitrary index set, we say that a collection {Aλ : λ ∈ 3} ⊂
6 is almost disjoint if λ1 , λ2 ∈ 3 and λ1 ̸= λ2 , then µ Aλ1 ∩ Aλ2 = 0.

 
(a) Show that if {An }n∈N ⊂ 6 is almost disjoint, then µ µ (An ).
S P
An =
n∈N n∈N
 
(b) Show that if {An }n∈N ⊂ 6 satisfies µ µ (An ) and µ (An ) < ∞ for every n ∈ N
S P
An =
n∈N n∈N
then {An }n∈N is almost disjoint.
(c) Show if we remove, in (b), the condition that µ (An ) < ∞ for every n ∈ N then the condition
 that 
µ µ (An ) alone does not imply that {An }n∈N is almost disjoint.
S P
An =
n∈N n∈N

3. Let (X, 6, µ) be a finite measure space. Let A j ⊂ 6 be a finite collection of measurable sets.

1≤ j≤n
Prove that     
n
[ n
X \

 X 

µ Aj = (−1)k+1
µ Aj .
 
j=1 k=1  J ⊂{1,...,n} j∈J 
card J =k