This action might not be possible to undo. Are you sure you want to continue?
1. Dynkin systems
Deﬁnition 1 A dynkin system on a set Ω is a subset D of the
power set P(Ω), with the following properties:
(i) Ω ∈ D
(ii) A, B ∈ D, A ⊆ B ⇒ B \ A ∈ D
(iii) A
n
∈ D, A
n
⊆ A
n+1
, n ≥ 1 ⇒
+∞
n=1
A
n
∈ D
Deﬁnition 2 A σalgebra on a set Ω is a subset F of the power
set P(Ω) with the following properties:
(i) Ω ∈ F
(ii) A ∈ F ⇒ A
c
= Ω \ A ∈ F
(iii) A
n
∈ F, n ≥ 1 ⇒
+∞
n=1
A
n
∈ F
Tutorial 1: Dynkin systems 2
Exercise 1. Let F be a σalgebra on Ω. Show that ∅ ∈ F, that
if A, B ∈ F then A ∪ B ∈ F and also A ∩ B ∈ F. Recall that
B \ A = B ∩ A
c
and conclude that F is also a dynkin system on Ω.
Exercise 2. Let (D
i
)
i∈I
be an arbitrary family of dynkin systems
on Ω, with I = ∅. Show that D
= ∩
i∈I
D
i
is also a dynkin system on
Ω.
Exercise 3. Let (F
i
)
i∈I
be an arbitrary family of σalgebras on Ω,
with I = ∅. Show that F
= ∩
i∈I
F
i
is also a σalgebra on Ω.
Exercise 4. Let A be a subset of the power set P(Ω). Deﬁne:
D(A)
= {D dynkin system on Ω : A ⊆ D}
Show that P(Ω) is a dynkin system on Ω, and that D(A) is not empty.
Deﬁne:
D(A)
=
D∈D(A)
D
Tutorial 1: Dynkin systems 3
Show that D(A) is a dynkin system on Ω such that A ⊆ D(A), and
that it is the smallest dynkin system on Ω with such property, (i.e. if
D is a dynkin system on Ω with A ⊆ D, then D(A) ⊆ D).
Deﬁnition 3 Let A ⊆ P(Ω). We call dynkin system generated
by A, the dynkin system on Ω, denoted D(A), equal to the intersection
of all dynkin systems on Ω, which contain A.
Exercise 5. Do exactly as before, but replacing dynkin systems by
σalgebras.
Deﬁnition 4 Let A ⊆ P(Ω). We call σalgebra generated by
A, the σalgebra on Ω, denoted σ(A), equal to the intersection of all
σalgebras on Ω, which contain A.
Deﬁnition 5 A subset A of the power set P(Ω) is called a πsystem
on Ω, if and only if it is closed under ﬁnite intersection, i.e. if it has
the property:
A, B ∈ A ⇒ A ∩ B ∈ A
Tutorial 1: Dynkin systems 4
Exercise 6. Let A be a πsystem on Ω. For all A ∈ D(A), we deﬁne:
Γ(A)
= {B ∈ D(A) : A ∩ B ∈ D(A)}
1. If A ∈ A, show that A ⊆ Γ(A)
2. Show that for all A ∈ D(A), Γ(A) is a dynkin system on Ω.
3. Show that if A ∈ A, then D(A) ⊆ Γ(A).
4. Show that if B ∈ D(A), then A ⊆ Γ(B).
5. Show that for all B ∈ D(A), D(A) ⊆ Γ(B).
6. Conclude that D(A) is also a πsystem on Ω.
Exercise 7. Let D be a dynkin system on Ω which is also a πsystem.
1. Show that if A, B ∈ D then A∪ B ∈ D.
Tutorial 1: Dynkin systems 5
2. Let A
n
∈ D, n ≥ 1. Consider B
n
= ∪
n
i=1
A
i
. Show that
∪
+∞
n=1
A
n
= ∪
+∞
n=1
B
n
.
3. Show that D is a σalgebra on Ω.
Exercise 8. Let A be a πsystem on Ω. Explain why D(A) is a
σalgebra on Ω, and σ(A) is a dynkin system on Ω. Conclude that
D(A) = σ(A). Prove the theorem:
Theorem 1 (dynkin system) Let C be a collection of subsets of Ω
which is closed under pairwise intersection. If D is a dynkin system
containing C then D also contains the σalgebra σ(C) generated by C.
Tutorial 2: Caratheodory’s Extension 1
2. Caratheodory’s Extension
In the following, Ω is a set. Whenever a union of sets is denoted as
opposed to ∪, it indicates that the sets involved are pairwise disjoint.
Deﬁnition 6 A semiring on Ω is a subset S of the power set P(Ω)
with the following properties:
(i) ∅ ∈ S
(ii) A, B ∈ S ⇒ A∩ B ∈ S
(iii) A, B ∈ S ⇒ ∃n ≥ 0, ∃A
i
∈ S : A \ B =
n
i=1
A
i
The last property (iii) says that whenever A, B ∈ S, there is n ≥ 0
and A
1
, . . . , A
n
in S which are pairwise disjoint, such that A \ B =
A
1
. . . A
n
. If n = 0, it is understood that the corresponding union
is equal to ∅, (in which case A ⊆ B).
Tutorial 2: Caratheodory’s Extension 2
Deﬁnition 7 A ring on Ω is a subset R of the power set P(Ω) with
the following properties:
(i) ∅ ∈ R
(ii) A, B ∈ R ⇒ A ∪ B ∈ R
(iii) A, B ∈ R ⇒ A \ B ∈ R
Exercise 1. Show that A ∩ B = A \ (A \ B) and therefore that a
ring is closed under pairwise intersection.
Exercise 2.Show that a ring on Ω is also a semiring on Ω.
Exercise 3.Suppose that a set Ω can be decomposed as Ω = A
1
A
2
A
3
where A
1
, A
2
and A
3
are distinct from ∅ and Ω. Deﬁne
S
1
= {∅, A
1
, A
2
, A
3
, Ω} and S
2
= {∅, A
1
, A
2
A
3
, Ω}. Show that S
1
and S
2
are semirings on Ω, but that S
1
∩ S
2
fails to be a semiring
on Ω.
Exercise 4. Let (R
i
)
i∈I
be an arbitrary family of rings on Ω, with
I = ∅. Show that R
= ∩
i∈I
R
i
is also a ring on Ω.
Tutorial 2: Caratheodory’s Extension 3
Exercise 5. Let A be a subset of the power set P(Ω). Deﬁne:
R(A)
= {R ring on Ω : A ⊆ R}
Show that P(Ω) is a ring on Ω, and that R(A) is not empty. Deﬁne:
R(A)
=
R∈R(A)
R
Show that R(A) is a ring on Ω such that A ⊆ R(A), and that it is
the smallest ring on Ω with such property, (i.e. if R is a ring on Ω
and A ⊆ R then R(A) ⊆ R).
Deﬁnition 8 Let A ⊆ P(Ω). We call ring generated by A, the
ring on Ω, denoted R(A), equal to the intersection of all rings on Ω,
which contain A.
Exercise 6.Let S be a semiring on Ω. Deﬁne the set R of all ﬁnite
unions of pairwise disjoint elements of S, i.e.
R
= {A : A =
n
i=1
A
i
for some n ≥ 0, A
i
∈ S}
Tutorial 2: Caratheodory’s Extension 4
(where if n = 0, the corresponding union is empty, i.e. ∅ ∈ R). Let
A =
n
i=1
A
i
and B =
p
j=1
B
j
∈ R:
1. Show that A ∩ B =
i,j
(A
i
∩ B
j
) and that R is closed under
pairwise intersection.
2. Show that if p ≥ 1 then A \ B = ∩
p
j=1
(
n
i=1
(A
i
\ B
j
)).
3. Show that R is closed under pairwise diﬀerence.
4. Show that A ∪ B = (A \ B) B and conclude that R is a ring
on Ω.
5. Show that R(S) = R.
Tutorial 2: Caratheodory’s Extension 5
Exercise 7. Everything being as before, deﬁne:
R
= {A : A = ∪
n
i=1
A
i
for some n ≥ 0, A
i
∈ S}
(We do not require the sets involved in the union to be pairwise dis
joint). Using the fact that R is closed under ﬁnite union, show that
R
⊆ R, and conclude that R
= R = R(S).
Deﬁnition 9 Let A ⊆ P(Ω) with ∅ ∈ A. We call measure on A,
any map µ : A → [0, +∞] with the following properties:
(i) µ(∅) = 0
(ii) A ∈ A, A
n
∈ A and A =
+∞
n=1
A
n
⇒ µ(A) =
+∞
n=1
µ(A
n
)
The indicates that we assume the A
n
’s to be pairwise disjoint in
the l.h.s. of (ii). It is customary to say in view of condition (ii) that
a measure is countably additive.
Tutorial 2: Caratheodory’s Extension 6
Exercise 8.If A is a σalgebra on Ω explain why property (ii) can
be replaced by:
(ii)
A
n
∈ A and A =
+∞
n=1
A
n
⇒ µ(A) =
+∞
n=1
µ(A
n
)
Exercise 9. Let A ⊆ P(Ω) with ∅ ∈ A and µ : A → [0, +∞] be a
measure on A.
1. Show that if A
1
, . . . , A
n
∈ A are pairwise disjoint and the union
A =
n
i=1
A
i
lies in A, then µ(A) = µ(A
1
) +. . . +µ(A
n
).
2. Show that if A, B ∈ A, A ⊆ B and B\A ∈ A then µ(A) ≤ µ(B).
Exercise 10. Let S be a semiring on Ω, and µ : S → [0, +∞] be a
measure on S. Suppose that there exists an extension of µ on R(S),
i.e. a measure ¯ µ : R(S) → [0, +∞] such that ¯ µ
S
= µ.
Tutorial 2: Caratheodory’s Extension 7
1. Let A be an element of R(S) with representation A =
n
i=1
A
i
as a ﬁnite union of pairwise disjoint elements of S. Show that
¯ µ(A) =
n
i=1
µ(A
i
)
2. Show that if ¯ µ
: R(S) → [0, +∞] is another measure with
¯ µ
S
= µ, i.e. another extension of µ on R(S), then ¯ µ
= ¯ µ.
Exercise 11. Let S be a semiring on Ω and µ : S → [0, +∞] be a
measure. Let A be an element of R(S) with two representations:
A =
n
i=1
A
i
=
p
j=1
B
j
as a ﬁnite union of pairwise disjoint elements of S.
1. For i = 1, . . . , n, show that µ(A
i
) =
p
j=1
µ(A
i
∩ B
j
)
2. Show that
n
i=1
µ(A
i
) =
p
j=1
µ(B
j
)
Tutorial 2: Caratheodory’s Extension 8
3. Explain why we can deﬁne a map ¯ µ : R(S) → [0, +∞] as:
¯ µ(A)
=
n
i=1
µ(A
i
)
4. Show that ¯ µ(∅) = 0.
Exercise 12. Everything being as before, suppose that (A
n
)
n≥1
is
a sequence of pairwise disjoint elements of R(S), each A
n
having the
representation:
A
n
=
p
n
k=1
A
k
n
, n ≥ 1
as a ﬁnite union of disjoint elements of S. Suppose moreover that
A =
+∞
n=1
A
n
is an element of R(S) with representation A =
p
j=1
B
j
,
as a ﬁnite union of pairwise disjoint elements of S.
1. Show that for j = 1, . . . , p, B
j
= ∪
+∞
n=1
∪
p
n
k=1
(A
k
n
∩ B
j
) and
explain why B
j
is of the form B
j
=
+∞
m=1
C
m
for some sequence
(C
m
)
m≥1
of pairwise disjoint elements of S.
Tutorial 2: Caratheodory’s Extension 9
2. Show that µ(B
j
) =
+∞
n=1
p
n
k=1
µ(A
k
n
∩ B
j
)
3. Show that for n ≥ 1 and k = 1, . . . , p
n
, A
k
n
=
p
j=1
(A
k
n
∩ B
j
)
4. Show that µ(A
k
n
) =
p
j=1
µ(A
k
n
∩ B
j
)
5. Recall the deﬁnition of ¯ µ of exercise (11) and show that it is a
measure on R(S).
Exercise 13.Prove the following theorem:
Theorem 2 Let S be a semiring on Ω. Let µ : S → [0, +∞] be a
measure on S. There exists a unique measure ¯ µ : R(S) → [0, +∞]
such that ¯ µ
S
= µ.
Tutorial 2: Caratheodory’s Extension 10
Deﬁnition 10 We deﬁne an outermeasure on Ω as being any
map µ
∗
: P(Ω) → [0, +∞] with the following properties:
(i) µ
∗
(∅) = 0
(ii) A ⊆ B ⇒ µ
∗
(A) ≤ µ
∗
(B)
(iii) µ
∗
_
+∞
_
n=1
A
n
_
≤
+∞
n=1
µ
∗
(A
n
)
Exercise 14. Show that µ
∗
(A ∪ B) ≤ µ
∗
(A) + µ
∗
(B), where µ
∗
is
an outermeasure on Ω and A, B ⊆ Ω.
Deﬁnition 11 Let µ
∗
be an outermeasure on Ω. We deﬁne:
Σ(µ
∗
)
= {A ⊆ Ω : µ
∗
(T) = µ
∗
(T ∩ A) +µ
∗
(T ∩ A
c
) , ∀T ⊆ Ω}
We call Σ(µ
∗
) the σalgebra associated with the outermeasure µ
∗
.
Note that the fact that Σ(µ
∗
) is indeed a σalgebra on Ω, remains to
be proved. This will be your task in the following exercises.
Tutorial 2: Caratheodory’s Extension 11
Exercise 15. Let µ
∗
be an outermeasure on Ω. Let Σ = Σ(µ
∗
) be
the σalgebra associated with µ
∗
. Let A, B ∈ Σ and T ⊆ Ω
1. Show that Ω ∈ Σ and A
c
∈ Σ.
2. Show that µ
∗
(T ∩ A) = µ
∗
(T ∩ A ∩ B) +µ
∗
(T ∩ A ∩ B
c
)
3. Show that T ∩ A
c
= T ∩ (A ∩ B)
c
∩ A
c
4. Show that T ∩ A ∩ B
c
= T ∩ (A ∩ B)
c
∩ A
5. Show that µ
∗
(T ∩ A
c
) +µ
∗
(T ∩ A ∩ B
c
) = µ
∗
(T ∩ (A ∩ B)
c
)
6. Adding µ
∗
(T∩(A∩B)) on both sides 5., conclude that A∩B ∈ Σ.
7. Show that A∪ B and A \ B belong to Σ.
Exercise 16. Everything being as before, let A
n
∈ Σ, n ≥ 1. Deﬁne
B
1
= A
1
and B
n+1
= A
n+1
\ (A
1
∪. . . ∪A
n
). Show that the B
n
’s are
pairwise disjoint elements of Σ and that ∪
+∞
n=1
A
n
=
+∞
n=1
B
n
.
Tutorial 2: Caratheodory’s Extension 12
Exercise 17. Everything being as before, show that if B, C ∈ Σ and
B ∩ C = ∅, then µ
∗
(T ∩ (B C)) = µ
∗
(T ∩ B) + µ
∗
(T ∩ C) for any
T ⊆ Ω.
Exercise 18.Everything being as before, let (B
n
)
n≥1
be a sequence
of pairwise disjoint elements of Σ, and let B
=
+∞
n=1
B
n
. Let N ≥ 1.
1. Explain why
N
n=1
B
n
∈ Σ
2. Show that µ
∗
(T ∩ (
N
n=1
B
n
)) =
N
n=1
µ
∗
(T ∩ B
n
)
3. Show that µ
∗
(T ∩ B
c
) ≤ µ
∗
(T ∩ (
N
n=1
B
n
)
c
)
4. Show that µ
∗
(T ∩ B
c
) +
+∞
n=1
µ
∗
(T ∩ B
n
) ≤ µ
∗
(T), and:
5. µ
∗
(T) ≤ µ
∗
(T ∩B
c
)+µ
∗
(T ∩B) ≤ µ
∗
(T ∩B
c
)+
+∞
n=1
µ
∗
(T ∩B
n
)
6. Show that B ∈ Σ and µ
∗
(B) =
+∞
n=1
µ
∗
(B
n
).
7. Show that Σ is a σalgebra on Ω, and µ
∗
Σ
is a measure on Σ.
Tutorial 2: Caratheodory’s Extension 13
Theorem 3 Let µ
∗
: P(Ω) → [0, +∞] be an outermeasure on Ω.
Then Σ(µ
∗
), the socalled σalgebra associated with µ
∗
, is indeed a
σalgebra on Ω and µ
∗
Σ(µ
∗
)
, is a measure on Σ(µ
∗
).
Exercise 19. Let R be a ring on Ω and µ : R → [0, +∞] be a
measure on R. For all T ⊆ Ω, deﬁne:
µ
∗
(T)
= inf
_
+∞
n=1
µ(A
n
) , (A
n
) is an Rcover of T
_
where an Rcover of T is deﬁned as any sequence (A
n
)
n≥1
of elements
of R such that T ⊆ ∪
+∞
n=1
A
n
. By convention inf ∅
= +∞.
1. Show that µ
∗
(∅) = 0.
2. Show that if A ⊆ B then µ
∗
(A) ≤ µ
∗
(B).
3. Let (A
n
)
n≥1
be a sequence of subsets of Ω, with µ
∗
(A
n
) < +∞
for all n ≥ 1. Given > 0, show that for all n ≥ 1, there exists
Tutorial 2: Caratheodory’s Extension 14
an Rcover (A
p
n
)
p≥1
of A
n
such that:
+∞
p=1
µ(A
p
n
) < µ
∗
(A
n
) +/2
n
Why is it important to assume µ
∗
(A
n
) < +∞.
4. Show that there exists an Rcover (R
k
) of ∪
+∞
n=1
A
n
such that:
+∞
k=1
µ(R
k
) =
+∞
n=1
+∞
p=1
µ(A
p
n
)
5. Show that µ
∗
(∪
+∞
n=1
A
n
) ≤ +
+∞
n=1
µ
∗
(A
n
)
6. Show that µ
∗
is an outermeasure on Ω.
Tutorial 2: Caratheodory’s Extension 15
Exercise 20. Everything being as before, Let A ∈ R. Let (A
n
)
n≥1
be an Rcover of A and put B
1
= A
1
∩ A, and:
B
n+1
= (A
n+1
∩ A) \ ((A
1
∩ A) ∪ . . . ∪ (A
n
∩ A))
1. Show that µ
∗
(A) ≤ µ(A).
2. Show that (B
n
)
n≥1
is a sequence of pairwise disjoint elements
of R such that A =
+∞
n=1
B
n
.
3. Show that µ(A) ≤ µ
∗
(A) and conclude that µ
∗
R
= µ.
Exercise 21. Everything being as before, Let A ∈ R and T ⊆ Ω.
1. Show that µ
∗
(T) ≤ µ
∗
(T ∩ A) +µ
∗
(T ∩ A
c
).
2. Let (T
n
) be an Rcover of T. Show that (T
n
∩A) and (T
n
∩A
c
)
are Rcovers of T ∩ A and T ∩ A
c
respectively.
3. Show that µ
∗
(T ∩ A) +µ
∗
(T ∩ A
c
) ≤ µ
∗
(T).
Tutorial 2: Caratheodory’s Extension 16
4. Show that R ⊆ Σ(µ
∗
).
5. Conclude that σ(R) ⊆ Σ(µ
∗
).
Exercise 22.Prove the following theorem:
Theorem 4 (caratheodory’s extension) Let R be a ring on Ω
and µ : R → [0, +∞] be a measure on R. There exists a measure
µ
: σ(R) → [0, +∞] such that µ
R
= µ.
Exercise 23. Let S be a semiring on Ω. Show that σ(R(S)) = σ(S).
Exercise 24.Prove the following theorem:
Theorem 5 Let S be a semiring on Ω and µ : S → [0, +∞] be a
measure on S. There exists a measure µ
: σ(S) → [0, +∞] such that
µ
S
= µ.
Tutorial 3: StieltjesLebesgue Measure 1
3. StieltjesLebesgue Measure
Deﬁnition 12 Let A ⊆ P(Ω) and µ : A → [0, +∞] be a map. We
say that µ is ﬁnitely additive if and only if, given n ≥ 1:
A ∈ A, A
i
∈ A, A =
n
i=1
A
i
⇒ µ(A) =
n
i=1
µ(A
i
)
We say that µ is ﬁnitely subadditive if and only if, given n ≥ 1 :
A ∈ A, A
i
∈ A, A ⊆
n
i=1
A
i
⇒ µ(A) ≤
n
i=1
µ(A
i
)
Exercise 1. Let S
= {]a, b] , a, b ∈ R} be the set of all intervals
]a, b], deﬁned as ]a, b] = {x ∈ R, a < x ≤ b}.
1. Show that ]a, b]∩]c, d] =]a ∨ c, b ∧ d]
2. Show that ]a, b]\]c, d] =]a, b ∧ c]∪]a ∨ d, b]
Tutorial 3: StieltjesLebesgue Measure 2
3. Show that c ≤ d ⇒ b ∧ c ≤ a ∨ d.
4. Show that S is a semiring on R.
Exercise 2. Suppose S is a semiring in Ω and µ : S → [0, +∞] is
ﬁnitely additive. Show that µ can be extended to a ﬁnitely additive
map ¯ µ : R(S) → [0, +∞], with ¯ µ
S
= µ.
Exercise 3. Everything being as before, Let A ∈ R(S), A
i
∈ R(S),
A ⊆ ∪
n
i=1
A
i
where n ≥ 1. Deﬁne B
1
= A
1
∩A and for i = 1, . . . , n−1:
B
i+1
= (A
i+1
∩ A) \ ((A
1
∩ A) ∪ . . . ∪ (A
i
∩ A))
1. Show that B
1
, . . . , B
n
are pairwise disjoint elements of R(S)
such that A =
n
i=1
B
i
.
2. Show that for all i = 1, . . . , n, we have ¯ µ(B
i
) ≤ ¯ µ(A
i
).
3. Show that ¯ µ is ﬁnitely subadditive.
4. Show that µ is ﬁnitely subadditive.
Tutorial 3: StieltjesLebesgue Measure 3
Exercise 4. Let F : R → R be a rightcontinuous, nondecreasing
map. Let S be the semiring on R, S = {]a, b] , a, b ∈ R}. Deﬁne the
map µ : S → [0, +∞] by µ(∅) = 0, and:
∀a ≤ b , µ(]a, b])
= F(b) −F(a) (1)
Let a < b and a
i
< b
i
for i = 1, . . . , n and n ≥ 1, with :
]a, b] =
n
i=1
]a
i
, b
i
]
1. Show that there is i
1
∈ {1, . . . , n} such that a
i
1
= a.
2. Show that ]b
i
1
, b] =
i∈{1,...,n}\{i
1
}
]a
i
, b
i
]
3. Show the existence of a permutation (i
1
, . . . , i
n
) of {1, . . . , n}
such that a = a
i
1
< b
i
1
= a
i
2
< . . . < b
i
n
= b.
4. Show that µ is ﬁnitely additive and ﬁnitely subadditive.
Tutorial 3: StieltjesLebesgue Measure 4
Exercise 5. µ being deﬁned as before, suppose a < b and a
n
< b
n
for n ≥ 1 with:
]a, b] =
+∞
n=1
]a
n
, b
n
]
Given N ≥ 1, let (i
1
, . . . , i
N
) be a permutation of {1, . . . , N} with:
a ≤ a
i
1
< b
i
1
≤ a
i
2
< . . . < b
i
N
≤ b
1. Show that
N
k=1
F(b
i
k
) −F(a
i
k
) ≤ F(b) −F(a).
2. Show that
+∞
n=1
µ(]a
n
, b
n
]) ≤ µ(]a, b])
3. Given > 0, show that there is η ∈]0, b −a[ such that:
0 ≤ F(a + η) −F(a) ≤
4. For n ≥ 1, show that there is η
n
> 0 such that:
0 ≤ F(b
n
+ η
n
) −F(b
n
) ≤
2
n
Tutorial 3: StieltjesLebesgue Measure 5
5. Show that [a + η, b] ⊆ ∪
+∞
n=1
]a
n
, b
n
+ η
n
[.
6. Explain why there exist p ≥ 1 and integers n
1
, . . . , n
p
such that:
]a + η, b] ⊆ ∪
p
k=1
]a
n
k
, b
n
k
+ η
n
k
]
7. Show that F(b) −F(a) ≤ 2 +
+∞
n=1
F(b
n
) −F(a
n
)
8. Show that µ : S → [0, +∞] is a measure.
Deﬁnition 13 A topology on Ω is a subset T of the power set
P(Ω), with the following properties:
(i) Ω, ∅ ∈ T
(ii) A, B ∈ T ⇒ A ∩ B ∈ T
(iii) A
i
∈ T , ∀i ∈ I ⇒
i∈I
A
i
∈ T
Tutorial 3: StieltjesLebesgue Measure 6
Property (iii) of deﬁnition (13) can be translated as: for any family
(A
i
)
i∈I
of elements of T , the union ∪
i∈I
A
i
is still an element of T .
Hence, a topology on Ω, is a set of subsets of Ω containing Ω and
the empty set, which is closed under ﬁnite intersection and arbitrary
union.
Deﬁnition 14 A topological space is an ordered pair (Ω, T ), where
Ω is a set and T is a topology on Ω.
Deﬁnition 15 Let (Ω, T ) be a topological space. We say that A ⊆ Ω
is an open set in Ω, if and only if it is an element of the topology T .
We say that A ⊆ Ω is a closed set in Ω, if and only if its complement
A
c
is an open set in Ω.
Deﬁnition 16 Let (Ω, T ) be a topological space. We deﬁne the
borel σalgebra on Ω, denoted B(Ω), as the σalgebra on Ω, gener
ated by the topology T . In other words, B(Ω) = σ(T )
Tutorial 3: StieltjesLebesgue Measure 7
Deﬁnition 17 We deﬁne the usual topology on R, denoted T
R
,
as the set of all U ⊆ R such that:
∀x ∈ U , ∃ > 0 , ]x −, x + [⊆ U
Exercise 6.Show that T
R
is indeed a topology on R.
Exercise 7. Consider the semiring S
= {]a, b] , a, b ∈ R}. Let T
R
be the usual topology on R, and B(R) be the borel σalgebra on R.
1. Let a ≤ b. Show that ]a, b] = ∩
+∞
n=1
]a, b + 1/n[.
2. Show that σ(S) ⊆ B(R).
3. Let U be an open subset of R. Show that for all x ∈ U, there
exist a
x
, b
x
∈ Q such that x ∈]a
x
, b
x
] ⊆ U.
4. Show that U = ∪
x∈U
]a
x
, b
x
].
5. Show that the set I
= {]a
x
, b
x
] , x ∈ U} is countable.
Tutorial 3: StieltjesLebesgue Measure 8
6. Show that U can be written U = ∪
i∈I
A
i
with A
i
∈ S.
7. Show that σ(S) = B(R).
Theorem 6 Let S be the semiring S = {]a, b] , a, b ∈ R}. Then,
the borel σalgebra B(R) on R, is generated by S, i.e. B(R) = σ(S).
Deﬁnition 18 A measurable space is an ordered pair (Ω, F) where
Ω is a set and F is a σalgebra on Ω.
Deﬁnition 19 A measure space is a triple (Ω, F, µ) where (Ω, F)
is a measurable space and µ : F → [0, +∞] is a measure on F.
Tutorial 3: StieltjesLebesgue Measure 9
Exercise 8.Let (Ω, F, µ) be a measure space. Let (A
n
)
n≥1
be a
sequence of elements of F such that A
n
⊆ A
n+1
for all n ≥ 1, and let
A = ∪
+∞
n=1
A
n
(we write A
n
↑ A). Deﬁne B
1
= A
1
and for all n ≥ 1,
B
n+1
= A
n+1
\ A
n
.
1. Show that (B
n
) is a sequence of pairwise disjoint elements of F
such that A =
+∞
n=1
B
n
.
2. Given N ≥ 1 show that A
N
=
N
n=1
B
n
.
3. Show that µ(A
N
) → µ(A) as N → +∞
4. Show that µ(A
n
) ≤ µ(A
n+1
) for all n ≥ 1.
Theorem 7 Let (Ω, F, µ) be a measure space. Then if (A
n
)
n≥1
is a
sequence of elements of F, such that A
n
↑ A, we have µ(A
n
) ↑ µ(A)
1
.
1
i.e. the sequence (µ(A
n
))
n≥1
is nondecreasing and converges to µ(A).
Tutorial 3: StieltjesLebesgue Measure 10
Exercise 9.Let (Ω, F, µ) be a measure space. Let (A
n
)
n≥1
be a
sequence of elements of F such that A
n+1
⊆ A
n
for all n ≥ 1, and let
A = ∩
+∞
n=1
A
n
(we write A
n
↓ A). We assume that µ(A
1
) < +∞.
1. Deﬁne B
n
= A
1
\ A
n
and show that B
n
∈ F, B
n
↑ A
1
\ A.
2. Show that µ(B
n
) ↑ µ(A
1
\ A)
3. Show that µ(A
n
) = µ(A
1
) −µ(A
1
\ A
n
)
4. Show that µ(A) = µ(A
1
) −µ(A
1
\ A)
5. Why is µ(A
1
) < +∞ important in deriving those equalities.
6. Show that µ(A
n
) → µ(A) as n → +∞
7. Show that µ(A
n+1
) ≤ µ(A
n
) for all n ≥ 1.
Theorem 8 Let (Ω, F, µ) be a measure space. Then if (A
n
)
n≥1
is
a sequence of elements of F, such that A
n
↓ A and µ(A
1
) < +∞, we
have µ(A
n
) ↓ µ(A).
Tutorial 3: StieltjesLebesgue Measure 11
Exercise 10.Take Ω = R and F = B(R). Suppose µ is a measure
on B(R) such that µ(]a, b]) = b −a, for a < b. Take A
n
=]n, +∞[.
1. Show that A
n
↓ ∅.
2. Show that µ(A
n
) = +∞, for all n ≥ 1.
3. Conclude that µ(A
n
) ↓ µ(∅) fails to be true.
Exercise 11. Let F : R → R be a rightcontinuous, nondecreasing
map. Show the existence of a measure µ : B(R) → [0, +∞] such that:
∀a, b ∈ R , a ≤ b , µ(]a, b]) = F(b) −F(a) (2)
Exercise 12.Let µ
1
, µ
2
be two measures on B(R) with property (2).
For n ≥ 1, we deﬁne:
D
n
= {B ∈ B(R) , µ
1
(B∩] −n, n]) = µ
2
(B∩] −n, n])}
1. Show that D
n
is a dynkin system on R.
Tutorial 3: StieltjesLebesgue Measure 12
2. Explain why µ
1
(] − n, n]) < +∞ and µ
2
(] − n, n]) < +∞ is
needed when proving 1.
3. Show that S
= {]a, b] , a, b ∈ R} ⊆ D
n
.
4. Show that B(R) ⊆ D
n
.
5. Show that µ
1
= µ
2
.
6. Prove the following theorem.
Theorem 9 Let F : R → R be a rightcontinuous, nondecreasing
map. There exists a unique measure µ : B(R) → [0, +∞] such that:
∀a, b ∈ R , a ≤ b , µ(]a, b]) = F(b) −F(a)
Deﬁnition 20 Let F : R → R be a rightcontinuous, nondecreasing
map. We call stieltjes measure on R associated with F, the unique
measure on B(R), denoted dF, such that:
∀a, b ∈ R , a ≤ b , dF(]a, b]) = F(b) −F(a)
Tutorial 3: StieltjesLebesgue Measure 13
Deﬁnition 21 We call lebesgue measure on R, the unique mea
sure on B(R), denoted dx, such that:
∀a, b ∈ R , a ≤ b , dx(]a, b]) = b −a
Exercise 13. Let F : R → R be a rightcontinuous, nondecreasing
map. Let x
0
∈ R.
1. Show that the limit F(x
0
−) = lim
x<x
0
,x→x
0
F(x) exists and is
an element of R.
2. Show that {x
0
} = ∩
+∞
n=1
]x
0
−1/n, x
0
].
3. Show that {x
0
} ∈ B(R)
4. Show that dF({x
0
}) = F(x
0
) −F(x
0
−)
Tutorial 3: StieltjesLebesgue Measure 14
Exercise 14.Let F : R → R be a rightcontinuous, nondecreasing
map. Let a ≤ b.
1. Show that ]a, b] ∈ B(R) and dF(]a, b]) = F(b) −F(a)
2. Show that [a, b] ∈ B(R) and dF([a, b]) = F(b) −F(a−)
3. Show that ]a, b[∈ B(R) and dF(]a, b[) = F(b−) −F(a)
4. Show that [a, b[∈ B(R) and dF([a, b[) = F(b−) −F(a−)
Exercise 15. Let A be a subset of the power set P(Ω). Let Ω
⊆ Ω.
Deﬁne:
A
Ω
= {A∩ Ω
, A ∈ A}
1. Show that if A is a topology on Ω, A
Ω
is a topology on Ω’.
2. Show that if A is a σalgebra on Ω, A
Ω
is a σalgebra on Ω’.
Tutorial 3: StieltjesLebesgue Measure 15
Deﬁnition 22 Let Ω be a set, and Ω
⊆ Ω. Let A be a subset of
the power set P(Ω). We call trace of A on Ω’, the subset A
Ω
of the
power set P(Ω
) deﬁned by:
A
Ω
= {A∩ Ω
, A ∈ A}
Deﬁnition 23 Let (Ω, T ) be a topological space and Ω
⊆ Ω. We call
induced topology on Ω’, denoted T
Ω
, the topology on Ω’ deﬁned
by:
T
Ω
= {A∩ Ω
, A ∈ T }
In other words, the induced topology T
Ω
is the trace of T on Ω’.
Exercise 16.Let A be a subset of the power set P(Ω). Let Ω
⊆ Ω,
and A
Ω
be the trace of A on Ω’. Deﬁne:
Γ
= {A ∈ σ(A) , A ∩ Ω
∈ σ(A
Ω
)}
where σ(A
Ω
) refers to the σalgebra generated by A
Ω
on Ω’.
Tutorial 3: StieltjesLebesgue Measure 16
1. Explain why the notation σ(A
Ω
) by itself is ambiguous.
2. Show that A ⊆ Γ.
3. Show that Γ is a σalgebra on Ω.
4. Show that σ(A
Ω
) = σ(A)
Ω
Theorem 10 Let Ω
⊆ Ω and A be a subset of the power set P(Ω).
Then, the trace on Ω’ of the σalgebra σ(A) generated by A, is equal
to the σalgebra on Ω’ generated by the trace of A on Ω’. In other
words, σ(A)
Ω
= σ(A
Ω
).
Exercise 17.Let (Ω, T ) be a topological space and Ω
⊆ Ω with its
induced topology T
Ω
.
1. Show that B(Ω)
Ω
= B(Ω
).
2. Show that if Ω
∈ B(Ω) then B(Ω
) ⊆ B(Ω).
Tutorial 3: StieltjesLebesgue Measure 17
3. Show that B(R
+
) = {A∩ R
+
, A ∈ B(R)}.
4. Show that B(R
+
) ⊆ B(R).
Exercise 18.Let (Ω, F, µ) be a measure space and Ω
⊆ Ω
1. Show that (Ω
, F
Ω
) is a measurable space.
2. If Ω
∈ F, show that F
Ω
⊆ F.
3. If Ω
∈ F, show that (Ω
, F
Ω
, µ
Ω
) is a measure space, where
µ
Ω
is deﬁned as µ
Ω
= µ
(F
Ω
)
.
Exercise 19. Let F : R
+
→ R be a rightcontinuous, nondecreasing
map with F(0) ≥ 0. Deﬁne:
¯
F(x)
=
0 if x < 0
F(x) if x ≥ 0
1. Show that
¯
F : R → R is rightcontinuous and nondecreasing.
Tutorial 3: StieltjesLebesgue Measure 18
2. Show that µ : B(R
+
) → [0, +∞] deﬁned by µ = d
¯
F
B(R
+
)
, is a
measure on B(R
+
) with the properties:
(i) µ({0}) = F(0)
(ii) ∀0 ≤ a ≤ b , µ(]a, b]) = F(b) −F(a)
Exercise 20. Deﬁne: C = {{0}} ∪ {]a, b] , 0 ≤ a ≤ b}
1. Show that C ⊆ B(R
+
)
2. Let U be open in R
+
. Show that U is of the form:
U =
i∈I
(R
+
∩]a
i
, b
i
])
where I is a countable set and a
i
, b
i
∈ R with a
i
≤ b
i
.
3. For all i ∈ I, show that R
+
∩]a
i
, b
i
] ∈ σ(C).
4. Show that σ(C) = B(R
+
)
Tutorial 3: StieltjesLebesgue Measure 19
Exercise 21.Let µ
1
and µ
2
be two measures on B(R
+
) with:
(i) µ
1
({0}) = µ
2
({0}) = F(0)
(ii) µ
1
(]a, b]) = µ
2
(]a, b]) = F(b) −F(a)
for all 0 ≤ a ≤ b. For n ≥ 1, we deﬁne:
D
n
= {B ∈ B(R
+
) , µ
1
(B ∩ [0, n]) = µ
2
(B ∩ [0, n])}
1. Show that D
n
is a dynkin system on R
+
with C ⊆ D
n
, where
the set C is deﬁned as in exercise (20).
2. Explain why µ
1
([0, n]) < +∞ and µ
2
([0, n]) < +∞ is important
when proving 1.
3. Show that µ
1
= µ
2
.
4. Prove the following theorem.
Tutorial 3: StieltjesLebesgue Measure 20
Theorem 11 Let F : R
+
→ R be a rightcontinuous, nondecreasing
map with F(0) ≥ 0. There exists a unique µ : B(R
+
) → [0, +∞]
measure on B(R
+
) such that:
(i) µ({0}) = F(0)
(ii) ∀0 ≤ a ≤ b , µ(]a, b]) = F(b) − F(a)
Deﬁnition 24 Let F : R
+
→R be a rightcontinuous, nondecreasing
map with F(0) ≥ 0. We call stieltjes measure on R
+
associated
with F, the unique measure on B(R
+
), denoted dF, such that:
(i) dF({0}) = F(0)
(ii) ∀0 ≤ a ≤ b , dF(]a, b]) = F(b) −F(a)
Tutorial 4: Measurability 1
4. Measurability
Deﬁnition 25 Let A and B be two sets, and f : A → B be a map.
Given A
⊆ A, we call direct image of A
by f the set denoted f(A
),
and deﬁned by f(A
) = {f(x) : x ∈ A
}.
Deﬁnition 26 Let A and B be two sets, and f : A → B be a map.
Given B
⊆ B, we call inverse image of B
by f the set denoted
f
−1
(B
), and deﬁned by f
−1
(B
) = {x : x ∈ A , f(x) ∈ B
}.
Exercise 1. Let A and B be two sets, and f : A → B be a bijection
from A to B. Let A
⊆ A and B
⊆ B.
1. Explain why the notation f
−1
(B
) is potentially ambiguous.
2. Show that the inverse image of B
by f is in fact equal to the
direct image of B
by f
−1
.
3. Show that the direct image of A
by f is in fact equal to the
inverse image of A
by f
−1
.
Tutorial 4: Measurability 2
Deﬁnition 27 Let (Ω, T ) and (S, T
S
) be two topological spaces. A
map f : Ω → S is said to be continuous if and only if:
∀B ∈ T
S
, f
−1
(B) ∈ T
In other words, if and only if the inverse image of any open set in S
is an open set in Ω.
We Write f : (Ω, T ) → (S, T
S
) is continuous, as a way of emphasizing
the two topologies T and T
S
with respect to which f is continuous.
Deﬁnition 28 Let E be a set. A map d : E × E → [0, +∞[ is said
to be a metric on E, if and only if:
(i) ∀x, y ∈ E , d(x, y) = 0 ⇔ x = y
(ii) ∀x, y ∈ E , d(x, y) = d(y, x)
(iii) ∀x, y, z ∈ E , d(x, y) ≤ d(x, z) + d(z, y)
Deﬁnition 29 A metric space is an ordered pair (E, d) where E
is a set, and d is a metric on E.
Tutorial 4: Measurability 3
Deﬁnition 30 Let (E, d) be a metric space. For all x ∈ E and
> 0, we deﬁne the socalled open ball in E:
B(x, )
= {y : y ∈ E , d(x, y) < }
We call metric topology on E, associated with d, the topology T
d
E
deﬁned by:
T
d
E
= {U ⊆ E , ∀x ∈ U, ∃ > 0, B(x, ) ⊆ U}
Exercise 2. Let T
d
E
be the metric topology associated with d, where
(E, d) is a metric space.
1. Show that T
d
E
is indeed a topology on E.
2. Given x ∈ E and > 0, show that B(x, ) is an open set in E.
Exercise 3. Show that the usual topology on R is nothing but the
metric topology associated with d(x, y) = x −y.
Tutorial 4: Measurability 4
Exercise 4. Let (E, d) and (F, δ) be two metric spaces. Show that
a map f : E → F is continuous, if and only if for all x ∈ E and > 0,
there exists η > 0 such that for all y ∈ E:
d(x, y) < η ⇒ δ(f(x), f(y)) <
Deﬁnition 31 Let (Ω, T ) and (S, T
S
) be two topological spaces. A
map f : Ω → S is said to be a homeomorphism, if and only if f is
a continuous bijection, such that f
−1
is also continuous.
Deﬁnition 32 A topological space (Ω, T ) is said to be metrizable,
if and only if there exists a metric d on Ω, such that the associated
metric topology coincides with T , i.e. T
d
Ω
= T .
Tutorial 4: Measurability 5
Deﬁnition 33 Let (E, d) be a metric space and F ⊆ E. We call
induced metric on F, denoted d
F
, the restriction of the metric d
to F ×F, i.e. d
F
= d
F×F
.
Exercise 5.Let (E, d) be a metric space and F ⊆ E. We deﬁne
T
F
= (T
d
E
)
F
as the topology on F induced by the metric topology on
E. Let T
F
= T
d
F
F
be the metric topology on F associated with the
induced metric d
F
on F.
1. Show that T
F
⊆ T
F
.
2. Given A ∈ T
F
, show that A = (∪
x∈A
B(x,
x
)) ∩ F for some
x
> 0, x ∈ A, where B(x,
x
) denotes the open ball in E.
3. Show that T
F
⊆ T
F
.
Theorem 12 Let (E, d) be a metric space and F ⊆ E. Then, the
topology on F induced by the metric topology, is equal to the metric
topology on F associated with the induced metric, i.e. (T
d
E
)
F
= T
d
F
F
.
Tutorial 4: Measurability 6
Exercise 6. Let φ : R →] −1, 1[ be the map deﬁned by:
∀x ∈ R , φ(x)
=
x
x + 1
1. Show that [−1, 0[ is not open in R.
2. Show that [−1, 0[ is open in [−1, 1].
3. Show that φ is a homeomorphism between R and ] −1, 1[.
4. Show that lim
x→+∞
φ(x) = 1 and lim
x→−∞
φ(x) = −1.
Exercise 7. Let
¯
R = [−∞, +∞] = R∪{−∞, +∞}. Let φ be deﬁned
as in exercise (6), and
¯
φ :
¯
R → [−1, 1] be the map deﬁned by:
¯
φ(x) =
φ(x) if x ∈ R
1 if x = +∞
−1 if x = −∞
Deﬁne:
T
¯
R
= {U ⊆
¯
R ,
¯
φ(U) is open in [−1, 1]}
Tutorial 4: Measurability 7
1. Show that
¯
φ is a bijection from
¯
R to [−1, 1], and let
¯
ψ =
¯
φ
−1
.
2. Show that T
¯
R
is a topology on
¯
R.
3. Show that
¯
φ is a homeomorphism between
¯
R and [−1, 1].
4. Show that [−∞, 2[, ]3, +∞], ]3, +∞[ are open in
¯
R.
5. Show that if φ
:
¯
R → [−1, 1] is an arbitrary homeomorphism,
then U ⊆
¯
R is open, if and only if φ
(U) is open in [−1, 1].
Deﬁnition 34 The usual topology on
¯
R is deﬁned as:
T
¯
R
= {U ⊆
¯
R ,
¯
φ(U) is open in [−1, 1]}
where
¯
φ :
¯
R → [−1, 1] is deﬁned by
¯
φ(−∞) = −1,
¯
φ(+∞) = 1 and:
∀x ∈ R ,
¯
φ(x)
=
x
x + 1
Tutorial 4: Measurability 8
Exercise 8. Let φ and
¯
φ be as in exercise (7). Deﬁne:
T
= (T
¯
R
)
R
= {U ∩ R , U ∈ T
¯
R
}
1. Recall why T
is a topology on R.
2. Show that for all U ⊆
¯
R, φ(U ∩ R) =
¯
φ(U)∩] −1, 1[.
3. Explain why if U ∈ T
¯
R
, φ(U ∩ R) is open in ] −1, 1[.
4. Show that T
⊆ T
R
, (the usual topology on R).
5. Let U ∈ T
R
. Show that
¯
φ(U) is open in ] −1, 1[ and [−1, 1].
6. Show that T
R
⊆ T
¯
R
7. Show that T
R
= T
, i.e. that the usual topology on
¯
R induces
the usual topology on R.
8. Show that B(R) = B(
¯
R)
R
= {B ∩ R , B ∈ B(
¯
R)}
Tutorial 4: Measurability 9
Exercise 9.Let d :
¯
R×
¯
R → [0, +∞[ be deﬁned by:
∀(x, y) ∈
¯
R×
¯
R , d(x, y) = φ(x) −φ(y)
where φ is an arbitrary homeomorphism from
¯
R to [−1, 1].
1. Show that d is a metric on
¯
R.
2. Show that if U ∈ T
¯
R
, then φ(U) is open in [−1, 1]
3. Show that for all U ∈ T
¯
R
and y ∈ φ(U), there exists > 0 such
that:
∀z ∈ [−1, 1] , z −y < ⇒ z ∈ φ(U)
4. Show that T
¯
R
⊆ T
d
¯
R
.
5. Show that for all U ∈ T
d
¯
R
and x ∈ U, there is > 0 such that:
∀y ∈
¯
R , φ(x) −φ(y) < ⇒ y ∈ U
6. Show that for all U ∈ T
d
¯
R
, φ(U) is open in [−1, 1].
Tutorial 4: Measurability 10
7. Show that T
d
¯
R
⊆ T
¯
R
8. Prove the following theorem.
Theorem 13 The topological space (
¯
R, T
¯
R
) is metrizable.
Deﬁnition 35 Let (Ω, F) and (S, Σ) be two measurable spaces. A
map f : Ω → S is said to be measurable with respect to F and Σ, if
and only if:
∀B ∈ Σ , f
−1
(B) ∈ F
We Write f : (Ω, F) → (S, Σ) is measurable, as a way of emphasizing
the two σalgebra F and Σ with respect to which f is measurable.
Exercise 10. Let (Ω, F) and (S, Σ) be two measurable spaces. Let
S
be a set and f : Ω → S be a map such that f(Ω) ⊆ S
⊆ S. We
deﬁne Σ
as the trace of Σ on S
, i.e. Σ
= Σ
S
.
1. Show that for all B ∈ Σ, we have f
−1
(B) = f
−1
(B ∩ S
)
Tutorial 4: Measurability 11
2. Show that f : (Ω, F) → (S, Σ) is measurable, if and only if
f : (Ω, F) → (S
, Σ
) is itself measurable.
3. Let f : Ω → R
+
. Show that the following are equivalent:
(i) f : (Ω, F) → (R
+
, B(R
+
)) is measurable
(ii) f : (Ω, F) → (R, B(R)) is measurable
(iii) f : (Ω, F) → (
¯
R, B(
¯
R)) is measurable
Exercise 11. Let (Ω, F), (S, Σ), (S
1
, Σ
1
) be three measurable spaces.
let f : (Ω, F) → (S, Σ) and g : (S, Σ) → (S
1
, Σ
1
) be two measurable
maps.
1. For all B ⊆ S
1
, show that (g ◦ f)
−1
(B) = f
−1
(g
−1
(B))
2. Show that g ◦ f : (Ω, F) → (S
1
, Σ
1
) is measurable.
Tutorial 4: Measurability 12
Exercise 12.Let (Ω, F) and (S, Σ) be two measurable spaces. Let
f : Ω → S be a map. We deﬁne:
Γ
= {B ∈ Σ , f
−1
(B) ∈ F}
1. Show that f
−1
(S) = Ω.
2. Show that for all B ⊆ S, f
−1
(B
c
) = (f
−1
(B))
c
.
3. Show that if B
n
⊆ S, n ≥ 1, then f
−1
(∪
+∞
n=1
B
n
) = ∪
+∞
n=1
f
−1
(B
n
)
4. Show that Γ is a σalgebra on S.
5. Prove the following theorem.
Theorem 14 Let (Ω, F) and (S, Σ) be two measurable spaces, and
A be a set of subsets of S generating Σ, i.e. such that Σ = σ(A).
Then f : (Ω, F) → (S, Σ) is measurable, if and only if:
∀B ∈ A , f
−1
(B) ∈ F
Tutorial 4: Measurability 13
Exercise 13. Let (Ω, T ) and (S, T
S
) be two topological spaces. Let
f : Ω → S be a map. Show that if f : (Ω, T ) → (S, T
S
) is continuous,
then f : (Ω, B(Ω)) → (S, B(S)) is measurable.
Exercise 14.We deﬁne the following subsets of the power set P(
¯
R):
C
1
= {[−∞, c] , c ∈ R}
C
2
= {[−∞, c[ , c ∈ R}
C
3
= {[c, +∞] , c ∈ R}
C
4
= {]c, +∞] , c ∈ R}
1. Show that C
2
and C
4
are subsets of T
¯
R
.
2. Show that the elements of C
1
and C
3
are closed in
¯
R.
3. Show that for all i = 1, 2, 3, 4, σ(C
i
) ⊆ B(
¯
R).
4. Let U be open in
¯
R. Explain why U ∩ R is open in R.
Tutorial 4: Measurability 14
5. Show that any open subset of R is a countable union of open
bounded intervals in R.
6. Let a < b, a, b ∈ R. Show that we have:
]a, b[=
+∞
n=1
]a, b −1/n] =
+∞
n=1
[a + 1/n, b[
7. Show that for all i = 1, 2, 3, 4, ]a, b[∈ σ(C
i
).
8. Show that for all i = 1, 2, 3, 4, {{−∞}, {+∞}} ⊆ σ(C
i
).
9. Show that for all i = 1, 2, 3, 4, σ(C
i
) = B(
¯
R)
10. Prove the following theorem.
Tutorial 4: Measurability 15
Theorem 15 Let (Ω, F) be a measurable space, and f : Ω →
¯
R be
a map. The following are equivalent:
(i) f : (Ω, F) → (
¯
R, B(
¯
R)) is measurable
(ii) ∀B ∈ B(
¯
R) , {f ∈ B} ∈ F
(iii) ∀c ∈ R , {f ≤ c} ∈ F
(iv) ∀c ∈ R , {f < c} ∈ F
(v) ∀c ∈ R , {c ≤ f} ∈ F
(vi) ∀c ∈ R , {c < f} ∈ F
Exercise 15. Let (Ω, F) be a measurable space. Let (f
n
)
n≥1
be a
sequence of measurable maps f
n
: (Ω, F) → (
¯
R, B(
¯
R)). Let g and h be
the maps deﬁned by g(ω) = inf
n≥1
f
n
(ω) and h(ω) = sup
n≥1
f
n
(ω),
for all ω ∈ Ω.
1. Let c ∈ R. Show that {c ≤ g} = ∩
+∞
n=1
{c ≤ f
n
}.
2. Let c ∈ R. Show that {h ≤ c} = ∩
+∞
n=1
{f
n
≤ c}.
Tutorial 4: Measurability 16
3. Show that g, h : (Ω, F) → (
¯
R, B(
¯
R)) are measurable.
Deﬁnition 36 Let (v
n
)
n≥1
be a sequence in
¯
R. We deﬁne:
u
= liminf
n→+∞
v
n
= sup
n≥1
inf
k≥n
v
k
and:
w
= limsup
n→+∞
v
n
= inf
n≥1
sup
k≥n
v
k
Then, u, w ∈
¯
R are respectively called lower limit and upper limit
of the sequence (v
n
)
n≥1
.
Exercise 16. Let (v
n
)
n≥1
be a sequence in
¯
R. for n ≥ 1 we deﬁne
u
n
= inf
k≥n
v
k
and w
n
= sup
k≥n
v
k
. Let u and w be the lower limit
and upper limit of (v
n
)
n≥1
, respectively.
1. Show that u
n
≤ u
n+1
≤ u, for all n ≥ 1.
Tutorial 4: Measurability 17
2. Show that w ≤ w
n+1
≤ w
n
, for all n ≥ 1.
3. Show that u
n
→ u and w
n
→ w as n → +∞.
4. Show that u
n
≤ v
n
≤ w
n
, for all n ≥ 1.
5. Show that u ≤ w.
6. Show that if u = w then (v
n
)
n≥1
converges to a limit v ∈
¯
R,
with u = v = w.
7. Show that if a, b ∈ R are such that u < a < b < w then for all
n ≥ 1, there exist N
1
, N
2
≥ n such that v
N
1
< a < b < v
N
2
.
8. Show that if a, b ∈ R are such that u < a < b < w then there
exist two strictly increasing sequences of integers (n
k
)
k≥1
and
(m
k
)
k≥1
such that for all k ≥ 1, we have v
n
k
< a < b < v
m
k
.
9. Show that if (v
n
)
n≥1
converges to some v ∈
¯
R, then u = w.
Tutorial 4: Measurability 18
Theorem 16 Let (v
n
)
n≥1
be a sequence in
¯
R. Then, the following
are equivalent:
(i) liminf
n→+∞
v
n
= limsup
n→+∞
v
n
(ii) lim
n→+∞
v
n
exists in
¯
R.
in which case:
lim
n→+∞
v
n
= liminf
n→+∞
v
n
= limsup
n→+∞
v
n
Exercise 17. Let f, g : (Ω, F) → (
¯
R, B(
¯
R)) be two measurable
maps, where (Ω, F) is a measurable space.
1. Show that {f < g} = ∪
r∈Q
({f < r} ∩ {r < g}).
2. Show that the sets {f < g}, {f > g}, {f = g}, {f ≤ g}, {f ≥ g}
belong to the σalgebra F.
Tutorial 4: Measurability 19
Exercise 18. Let (Ω, F) be a measurable space. Let (f
n
)
n≥1
be
a sequence of measurable maps f
n
: (Ω, F) → (
¯
R, B(
¯
R)). We deﬁne
g = liminf f
n
and h = limsup f
n
in the obvious way:
∀ω ∈ Ω , g(ω)
= liminf
n→+∞
f
n
(ω)
∀ω ∈ Ω , h(ω)
= limsup
n→+∞
f
n
(ω)
1. Show that g, h : (Ω, F) → (
¯
R, B(
¯
R)) are measurable.
2. Show that g ≤ h, i.e. ∀ω ∈ Ω , g(ω) ≤ h(ω).
3. Show that {g = h} ∈ F.
4. Show that {ω : ω ∈ Ω , lim
n→+∞
f
n
(ω) exists in
¯
R} ∈ F.
5. Suppose Ω = {g = h}, and let f(ω) = lim
n→+∞
f
n
(ω), for all
ω ∈ Ω. Show that f : (Ω, F) → (
¯
R, B(
¯
R)) is measurable.
Tutorial 4: Measurability 20
Exercise 19. Let f, g : (Ω, F) → (
¯
R, B(
¯
R)) be two measurable
maps, where (Ω, F) is a measurable space.
1. Show that −f, f, f
+
= max(f, 0) and f
−
= max(−f, 0) are
measurable with respect to F and B(
¯
R).
2. Let a ∈
¯
R. Explain why the map a+f may not be well deﬁned.
3. Show that (a+f) : (Ω, F) → (
¯
R, B(
¯
R)) is measurable, whenever
a ∈ R.
4. Show that (a.f) : (Ω, F) → (
¯
R, B(
¯
R)) is measurable, for all
a ∈
¯
R. (Recall the convention 0.∞ = 0).
5. Explain why the map f + g may not be well deﬁned.
6. Suppose that f ≥ 0 and g ≥ 0, i.e. f(Ω) ⊆ [0, +∞] and also
g(Ω) ⊆ [0, +∞]. Show that {f + g < c} = {f < c − g}, for all
c ∈ R. Show that f + g : (Ω, F) → (
¯
R, B(
¯
R)) is measurable.
Tutorial 4: Measurability 21
7. Show that f +g : (Ω, F) → (
¯
R, B(
¯
R)) is measurable, in the case
when f and g take values in R.
8. Show that 1/f : (Ω, F) → (
¯
R, B(
¯
R)) is measurable, in the case
when f(Ω) ⊆ R\ {0}.
9. Suppose that f is Rvalued. Show that
¯
f deﬁned by
¯
f(ω) =
f(ω) if f(ω) = 0 and
¯
f(ω) = 1 if f(ω) = 0, is measurable with
respect to F and B(
¯
R).
10. Suppose f and g take values in R. Let
¯
f be deﬁned as in 9.
Show that for all c ∈ R, the set {fg < c} can be expressed as:
({f >0}∩{g < c/
¯
f})({f <0}∩{g > c/
¯
f})({f = 0}∩{f < c})
11. Show that fg : (Ω, F) → (
¯
R, B(
¯
R)) is measurable, in the case
when f and g take values in R.
Tutorial 4: Measurability 22
Exercise 20.Let f, g : (Ω, F) → (
¯
R, B(
¯
R)) be two measurable maps,
where (Ω, F) is a measurable space. Let
¯
f, ¯ g, be deﬁned by:
¯
f(ω)
=
f(ω) if f(ω) ∈ {−∞, +∞}
1 if f(ω) ∈ {−∞, +∞}
¯ g(ω) being deﬁned in a similar way. Consider the partitions of Ω,
Ω = A
1
A
2
A
3
A
4
A
5
and Ω = B
1
B
2
B
3
B
4
B
5
,
where A
1
= {f ∈]0, +∞[}, A
2
= {f ∈] − ∞, 0[}, A
3
= {f = 0},
A
4
= {f = −∞}, A
5
= {f = +∞} and B
1
, B
2
, B
3
, B
4
, B
5
being
deﬁned in a similar way with g. Recall the conventions 0×(+∞) = 0,
(−∞) ×(+∞) = (−∞), etc. . .
1. Show that
¯
f and ¯ g are measurable with respect to F and B(
¯
R).
2. Show that all A
i
’s and B
j
’s are elements of F.
3. Show that for all B ∈ B(
¯
R):
{fg ∈ B} =
5
i,j=1
(A
i
∩ B
j
∩ {fg ∈ B})
Tutorial 4: Measurability 23
4. Show that A
i
∩ B
j
∩ {fg ∈ B} = A
i
∩ B
j
∩ {
¯
f¯ g ∈ B}, in the
case when 1 ≤ i ≤ 3 and 1 ≤ j ≤ 3.
5. Show that A
i
∩ B
j
∩ {fg ∈ B} is either equal to ∅ or A
i
∩ B
j
,
in the case when i ≥ 4 or j ≥ 4.
6. Show that fg : (Ω, F) → (
¯
R, B(
¯
R)) is measurable.
Deﬁnition 37 Let (Ω, T ) be a topological space, and A ⊆ Ω. We
call closure of A in Ω, denoted
¯
A, the set deﬁned by:
¯
A
= {x ∈ Ω : x ∈ U ∈ T ⇒ U ∩ A = ∅}
Exercise 21. Let (E, T ) be a topological space, and A ⊆ E. Let
¯
A
be the closure of A.
1. Show that A ⊆
¯
A and that
¯
A is closed.
2. Show that if B is closed and A ⊆ B, then
¯
A ⊆ B.
Tutorial 4: Measurability 24
3. Show that
¯
A is the smallest closed set in E containing A.
4. Show that A is closed if and only if A =
¯
A.
5. Show that if (E, T ) is metrizable, then:
¯
A = {x ∈ E : ∀ > 0 , B(x, ) ∩ A = ∅}
where B(x, ) is relative to any metric d such that T
d
E
= T .
Exercise 22. Let (E, d) be a metric space. Let A ⊆ E. For all
x ∈ E, we deﬁne:
d(x, A)
= inf{d(x, y) : y ∈ A}
= Φ
A
(x)
where it is understood that inf ∅ = +∞.
1. Show that for all x ∈ E, d(x, A) = d(x,
¯
A).
2. Show that d(x, A) = 0, if and only if x ∈
¯
A.
3. Show that for all x, y ∈ E, d(x, A) ≤ d(x, y) + d(y, A).
Tutorial 4: Measurability 25
4. Show that if A = ∅, d(x, A) −d(y, A) ≤ d(x, y).
5. Show that Φ
A
: (E, T
d
E
) → (
¯
R, T
¯
R
) is continuous.
6. Show that if A is closed, then A = Φ
−1
A
({0})
Exercise 23.Let (Ω, F) be a measurable space. Let (f
n
)
n≥1
be a
sequence of measurable maps f
n
: (Ω, F) → (E, B(E)), where (E, d) is
a metric space. We assume that for all ω ∈ Ω, the sequence (f
n
(ω))
n≥1
converges to some f(ω) ∈ E.
1. Explain why liminf f
n
and limsup f
n
may not be deﬁned in an
arbitrary metric space E.
2. Show that f : (Ω, F) → (E, B(E)) is measurable, if and only if
f
−1
(A) ∈ F for all closed subsets A of E.
3. Show that for all A closed in E, f
−1
(A) = (Φ
A
◦ f)
−1
({0}),
where the map Φ
A
: E →
¯
R is deﬁned as in exercise (22).
Tutorial 4: Measurability 26
4. Show that Φ
A
◦ f
n
: (Ω, F) → (
¯
R, B(
¯
R)) is measurable.
5. Show that f : (Ω, F) → (E, B(E)) is measurable.
Theorem 17 Let (Ω, F) be a measurable space. Let (f
n
)
n≥1
be a
sequence of measurable maps f
n
: (Ω, F) → (E, B(E)), where (E, d)
is a metric space. Then, if the limit f = limf
n
exists on Ω, the map
f : (Ω, F) → (E, B(E)) is itself measurable.
Deﬁnition 38 The usual topology on C, the set of complex num
bers, is deﬁned as the metric topology associated with d(z, z
) = z−z
.
Exercise 24. Let f : (Ω, F) → (C, B(C)) be a measurable map,
where (Ω, F) is a measurable space. Let u = Re(f) and v = Im(f).
Show that u, v, f : (Ω, F) → (
¯
R, B(
¯
R)) are all measurable.
Exercise 25. Deﬁne the subset of the power set P(C):
C
= {]a, b[×]c, d[ , a, b, c, d ∈ R}
Tutorial 4: Measurability 27
where it is understood that:
]a, b[×]c, d[
= {z = x + iy ∈ C , (x, y) ∈]a, b[×]c, d[}
1. Show that any element of C is open in C.
2. Show that σ(C) ⊆ B(C).
3. Let z = x + iy ∈ C. Show that if x < η and y < η then we
have z <
√
2η.
4. Let U be open in C. Show that for all z ∈ U, there are rational
numbers a
z
, b
z
, c
z
, d
z
such that z ∈]a
z
, b
z
[×]c
z
, d
z
[⊆ U.
5. Show that U can be written as U = ∪
+∞
n=1
A
n
where A
n
∈ C.
6. Show that σ(C) = B(C).
7. Let (Ω, F) be a measurable space, and u, v : (Ω, F) → (R, B(R))
be two measurable maps. Show that u+iv : (Ω, F) → (C, B(C))
is measurable.
Tutorial 5: Lebesgue Integration 1
5. Lebesgue Integration
In the following, (Ω, F, µ) is a measure space.
Deﬁnition 39 Let A ⊆ Ω. We call characteristic function of A,
the map 1
A
: Ω → R, deﬁned by:
∀ω ∈ Ω , 1
A
(ω)
=
_
1 if ω ∈ A
0 if ω ∈ A
Exercise 1. Given A ⊆ Ω, show that 1
A
: (Ω, F) → (
¯
R, B(
¯
R)) is
measurable if and only if A ∈ F.
Deﬁnition 40 Let (Ω, F) be a measurable space. We say that a map
s : Ω → R
+
is a simple function on (Ω, F), if and only if s is of
the form :
s =
n
i=1
α
i
1
A
i
where n ≥ 1, α
i
∈ R
+
and A
i
∈ F, for all i = 1, . . . , n.
Tutorial 5: Lebesgue Integration 2
Exercise 2. Show that s : (Ω, F) → (R
+
, B(R
+
)) is measurable,
whenever s is a simple function on (Ω, F).
Exercise 3. Let s be a simple function on (Ω, F) with representation
s =
n
i=1
α
i
1
A
i
. Consider the map φ : Ω → {0, 1}
n
deﬁned by
φ(ω) = (1
A
1
(ω), . . . , 1
A
n
(ω)). For each y ∈ s(Ω), pick one ω
y
∈ Ω
such that y = s(ω
y
). Consider the map ψ : s(Ω) → {0, 1}
n
deﬁned by
ψ(y) = φ(ω
y
).
1. Show that ψ is injective, and that s(Ω) is a ﬁnite subset of R
+
.
2. Show that s =
α∈s(Ω)
α1
{s=α}
3. Show that any simple function s can be represented as:
s =
n
i=1
α
i
1
A
i
where n ≥ 1, α
i
∈ R
+
, A
i
∈ F and Ω = A
1
. . . A
n
.
Tutorial 5: Lebesgue Integration 3
Deﬁnition 41 Let (Ω, F) be a measurable space, and s be a simple
function on (Ω, F). We call partition of the simple function s, any
representation of the form:
s =
n
i=1
α
i
1
A
i
where n ≥ 1, α
i
∈ R
+
, A
i
∈ F and Ω = A
1
. . . A
n
.
Exercise 4. Let s be a simple function on (Ω, F) with two partitions:
s =
n
i=1
α
i
1
A
i
=
m
j=1
β
j
1
B
j
1. Show that s =
i,j
α
i
1
A
i
∩B
j
is a partition of s.
2. Recall the convention 0 × (+∞) = 0 and α × (+∞) = +∞
if α > 0. For all a
1
, . . . , a
p
in [0, +∞], p ≥ 1 and x ∈ [0, +∞],
prove the distributive property: x(a
1
+. . .+a
p
) = xa
1
+. . .+xa
p
.
Tutorial 5: Lebesgue Integration 4
3. Show that
n
i=1
α
i
µ(A
i
) =
m
j=1
β
j
µ(B
j
).
4. Explain why the following deﬁnition is legitimate.
Deﬁnition 42 Let (Ω, F, µ) be a measure space, and s be a simple
function on (Ω, F). We deﬁne the integral of s with respect to µ, as
the sum, denoted I
µ
(s), deﬁned by:
I
µ
(s)
=
n
i=1
α
i
µ(A
i
) ∈ [0, +∞]
where s =
n
i=1
α
i
1
A
i
is any partition of s.
Exercise 5. Let s, t be two simple functions on (Ω, F) with partitions
s =
n
i=1
α
i
1
A
i
and t =
m
j=1
β
j
1
B
j
. Let α ∈ R
+
.
1. Show that s + t is a simple function on (Ω, F) with partition:
s + t =
n
i=1
m
j=1
(α
i
+ β
j
)1
A
i
∩B
j
Tutorial 5: Lebesgue Integration 5
2. Show that I
µ
(s + t) = I
µ
(s) + I
µ
(t).
3. Show that αs is a simple function on (Ω, F).
4. Show that I
µ
(αs) = αI
µ
(s).
5. Why is the notation I
µ
(αs) meaningless if α = +∞ or α < 0.
6. Show that if s ≤ t then I
µ
(s) ≤ I
µ
(t).
Exercise 6. Let f : (Ω, F) → [0, +∞] be a nonnegative and mea
surable map. For all n ≥ 1, we deﬁne:
s
n
=
n2
n
−1
k=0
k
2
n
1
{
k
2
n
≤f<
k+1
2
n
}
+ n1
{n≤f}
(1)
1. Show that s
n
is a simple function on (Ω, F), for all n ≥ 1.
2. Show that equation (1) is a partition s
n
, for all n ≥ 1.
3. Show that s
n
≤ s
n+1
≤ f, for all n ≥ 1.
Tutorial 5: Lebesgue Integration 6
4. Show that s
n
↑ f as n → +∞
1
.
Theorem 18 Let f : (Ω, F) → [0, +∞] be a nonnegative and mea
surable map, where (Ω, F) is a measurable space. There exists a se
quence (s
n
)
n≥1
of simple functions on (Ω, F) such that s
n
↑ f.
Deﬁnition 43 Let f : (Ω, F) → [0, +∞] be a nonnegative and
measurable map, where (Ω, F, µ) is a measure space. We deﬁne the
lebesgue integral of f with respect to µ, denoted
_
fdµ, as:
_
fdµ
= sup{I
µ
(s) : s simple function on (Ω, F) , s ≤ f}
where, given any simple function s on (Ω, F), I
µ
(s) denotes its inte
gral with respect to µ.
1
i.e. for all ω ∈ Ω, the sequence (s
n
(ω))
n≥1
is nondecreasing and converges
to f(ω) ∈ [0, +∞].
Tutorial 5: Lebesgue Integration 7
Exercise 7. Let f : (Ω, F) → [0, +∞] be a nonnegative and mea
surable map.
1. Show that
_
fdµ ∈ [0, +∞].
2. Show that
_
fdµ = I
µ
(f), whenever f is a simple function.
3. Show that
_
gdµ ≤
_
fdµ, whenever g : (Ω, F) → [0, +∞] is
nonnegative and measurable map with g ≤ f.
4. Show that
_
(cf)dµ = c
_
fdµ, if 0 < c < +∞. Explain why
both integrals are well deﬁned. Is the equality still true for
c = 0.
5. For n ≥ 1, put A
n
= {f > 1/n}, and s
n
= (1/n)1
A
n
. Show
that s
n
is a simple function on (Ω, F) with s
n
≤ f. Show that
A
n
↑ {f > 0}.
6. Show that
_
fdµ = 0 ⇒ µ({f > 0}) = 0.
Tutorial 5: Lebesgue Integration 8
7. Show that if s is a simple function on (Ω, F) with s ≤ f, then
µ({f > 0}) = 0 implies I
µ
(s) = 0.
8. Show that
_
fdµ = 0 ⇔ µ({f > 0}) = 0.
9. Show that
_
(+∞)fdµ = (+∞)
_
fdµ. Explain why both inte
grals are well deﬁned.
10. Show that (+∞)1
{f=+∞}
≤ f and:
_
(+∞)1
{f=+∞}
dµ = (+∞)µ({f = +∞})
11. Show that
_
fdµ < +∞ ⇒ µ({f = +∞}) = 0.
12. Suppose that µ(Ω) = +∞ and take f = 1. Show that the
converse of the previous implication is not true.
Tutorial 5: Lebesgue Integration 9
Exercise 8. Let s be a simple function on (Ω, F). let A ∈ F.
1. Show that s1
A
is a simple function on (Ω, F).
2. Show that for any partition s =
n
i=1
α
i
1
A
i
of s, we have:
I
µ
(s1
A
) =
n
i=1
α
i
µ(A
i
∩ A)
3. Let ν : F → [0, +∞] be deﬁned by ν(A) = I
µ
(s1
A
). Show that
ν is a measure on F.
4. Suppose A
n
∈ F, A
n
↑ A. Show that I
µ
(s1
A
n
) ↑ I
µ
(s1
A
).
Exercise 9. Let (f
n
)
n≥1
be a sequence of nonnegative and measur
able maps f
n
: (Ω, F) → [0, +∞], such that f
n
↑ f.
1. Recall what the notation f
n
↑ f means.
2. Explain why f : (Ω, F) → (
¯
R, B(
¯
R)) is measurable.
Tutorial 5: Lebesgue Integration 10
3. Let α = sup
n≥1
_
f
n
dµ. Show that
_
f
n
dµ ↑ α.
4. Show that α ≤
_
fdµ.
5. Let s be any simple function on (Ω, F) such that s ≤ f. Let
c ∈]0, 1[. For n ≥ 1, deﬁne A
n
= {cs ≤ f
n
}. Show that A
n
∈ F
and A
n
↑ Ω.
6. Show that cI
µ
(s1
A
n
) ≤
_
f
n
dµ, for all n ≥ 1.
7. Show that cI
µ
(s) ≤ α.
8. Show that I
µ
(s) ≤ α.
9. Show that
_
fdµ ≤ α.
10. Conclude that
_
f
n
dµ ↑
_
fdµ.
Theorem 19 (Monotone Convergence) Let (Ω, F, µ) be a mea
sure space. Let (f
n
)
n≥1
be a sequence of nonnegative and measurable
maps f
n
: (Ω, F) → [0, +∞] such that f
n
↑ f. Then
_
f
n
dµ ↑
_
fdµ.
Tutorial 5: Lebesgue Integration 11
Exercise 10. Let f, g : (Ω, F) → [0, +∞] be two nonnegative and
measurable maps. Let a, b ∈ [0, +∞].
1. Show that if (f
n
)
n≥1
and (g
n
)
n≥1
are two sequences of non
negative and measurable maps such that f
n
↑ f and g
n
↑ g,
then f
n
+ g
n
↑ f + g.
2. Show that
_
(f + g)dµ =
_
fdµ +
_
gdµ.
3. Show that
_
(af + bg)dµ = a
_
fdµ + b
_
gdµ.
Exercise 11. Let (f
n
)
n≥1
be a sequence of nonnegative and mea
surable maps f
n
: (Ω, F) → [0, +∞]. Deﬁne f =
+∞
n=1
f
n
.
1. Explain why f : (Ω, F) → [0, +∞] is well deﬁned, nonnegative
and measurable.
2. Show that
_
fdµ =
+∞
n=1
_
f
n
dµ.
Tutorial 5: Lebesgue Integration 12
Deﬁnition 44 Let (Ω, F, µ) be a measure space and let P(ω) be a
property depending on ω ∈ Ω. We say that the property P(ω) holds
µalmost surely, and we write P(ω) µa.s., if and only if:
∃N ∈ F , µ(N) = 0 , ∀ω ∈ N
c
, P(ω) holds
Exercise 12. Let P(ω) be a property depending on ω ∈ Ω, such that
{ω ∈ Ω : P(ω) holds} is an element of the σalgebra F.
1. Show that P(ω) , µa.s. ⇔ µ({ω ∈ Ω : P(ω) holds}
c
) = 0.
2. Explain why in general, the righthand side of this equivalence
cannot be used to deﬁned µalmost sure properties.
Exercise 13. Let (Ω, F, µ) be a measure space and (A
n
)
n≥1
be a
sequence of elements of F. Show that µ(∪
+∞
n=1
A
n
) ≤
+∞
n=1
µ(A
n
).
Exercise 14. Let (f
n
)
n≥1
be a sequence of maps f
n
: Ω → [0, +∞].
1. Translate formally the statement f
n
↑ f µa.s.
Tutorial 5: Lebesgue Integration 13
2. Translate formally f
n
→ f µa.s. and ∀n, (f
n
≤ f
n+1
µa.s.)
3. Show that the statements 1. and 2. are equivalent.
Exercise 15. Suppose that f, g : (Ω, F) → [0, +∞] are nonnegative
and measurable with f = g µa.s.. Let N ∈ F, µ(N) = 0 such that
f = g on N
c
. Explain why
_
fdµ =
_
(f1
N
)dµ +
_
(f1
N
c)dµ, all
integrals being well deﬁned. Show that
_
fdµ =
_
gdµ.
Exercise 16. Suppose (f
n
)
n≥1
is a sequence of nonnegative and
measurable maps such that f
n
↑ f µa.s.. Let N ∈ F, µ(N) = 0, such
that f
n
↑ f on N
c
. Deﬁne
¯
f
n
= f
n
1
N
c and
¯
f = f1
N
c.
1. Explain why
¯
f and the
¯
f
n
’s are nonnegative and measurable.
2. Show that
¯
f
n
↑
¯
f.
3. Show that
_
f
n
dµ ↑
_
fdµ.
Tutorial 5: Lebesgue Integration 14
Exercise 17. Let (f
n
)
n≥1
be a sequence of nonnegative and measur
able maps f
n
: (Ω, F) → [0, +∞]. For n ≥ 1, we deﬁne g
n
= inf
k≥n
f
k
.
1. Explain why the g
n
’s are nonnegative and measurable.
2. Show that g
n
↑ liminf f
n
.
3. Show that
_
g
n
dµ ≤
_
f
n
dµ, for all n ≥ 1.
4. Show that if (u
n
)
n≥1
and (v
n
)
n≥1
are two sequences in
¯
R with
u
n
≤ v
n
for all n ≥ 1, then liminf u
n
≤ liminf v
n
.
5. Show that
_
(liminf f
n
)dµ ≤ liminf
_
f
n
dµ, and recall why all
integrals are well deﬁned.
Theorem 20 (Fatou Lemma) Let (Ω, F, µ) be a measure space,
and let (f
n
)
n≥1
be a sequence of nonnegative and measurable maps
f
n
: (Ω, F) → [0, +∞]. Then:
_
(liminf
n→+∞
f
n
)dµ ≤ liminf
n→+∞
_
f
n
dµ
Tutorial 5: Lebesgue Integration 15
Exercise 18. Let f : (Ω, F) → [0, +∞] be a nonnegative and mea
surable map. Let A ∈ F.
1. Recall what is meant by the induced measure space (A, F
A
, µ
A
).
Why is it important to have A ∈ F. Show that the restriction
of f to A, f
A
: (A, F
A
) → [0, +∞] is measurable.
2. We deﬁne the map µ
A
: F → [0, +∞] by µ
A
(E) = µ(A∩E), for
all E ∈ F. Show that (Ω, F, µ
A
) is a measure space.
3. Consider the equalities:
_
(f1
A
)dµ =
_
fdµ
A
=
_
(f
A
)dµ
A
(2)
For each of the above integrals, what is the underlying measure
space on which the integral is considered. What is the map
being integrated. Explain why each integral is well deﬁned.
4. Show that in order to prove (2), it is suﬃcient to consider the
case when f is a simple function on (Ω, F).
Tutorial 5: Lebesgue Integration 16
5. Show that in order to prove (2), it is suﬃcient to consider the
case when f is of the form f = 1
B
, for some B ∈ F.
6. Show that (2) is indeed true.
Deﬁnition 45 Let f : (Ω, F) → [0, +∞] be a nonnegative and mea
surable map, where (Ω, F, µ) is a measure space. let A ∈ F. We call
partial lebesgue integral of f with respect to µ over A, the integral
denoted
_
A
fdµ, deﬁned as:
_
A
fdµ
=
_
(f1
A
)dµ =
_
fdµ
A
=
_
(f
A
)dµ
A
where µ
A
is the measure on (Ω, F), µ
A
= µ(A∩•), f
A
is the restric
tion of f to A and µ
A
is the restriction of µ to F
A
, the trace of F
on A.
Tutorial 5: Lebesgue Integration 17
Exercise 19. Let f, g : (Ω, F) → [0, +∞] be two nonnegative and
measurable maps. Let ν : F → [0, +∞] be deﬁned by ν(A) =
_
A
fdµ,
for all A ∈ F.
1. Show that ν is a measure on F.
2. Show that:
_
gdν =
_
gfdµ
Theorem 21 Let f : (Ω, F) → [0, +∞] be a nonnegative and mea
surable map, where (Ω, F, µ) is a measure space. Let ν : F → [0, +∞]
be deﬁned by ν(A) =
_
A
fdµ, for all A ∈ F. Then, ν is a measure on
F, and for all g : (Ω, F) → [0, +∞] nonnegative and measurable, we
have:
_
gdν =
_
gfdµ
Tutorial 5: Lebesgue Integration 18
Deﬁnition 46 The L
1
spaces on a measure space (Ω, F, µ), are:
L
1
R
(Ω, F, µ)
=
_
f : (Ω, F) → (R, B(R)) measurable,
_
fdµ < +∞
_
L
1
C
(Ω, F, µ)
=
_
f : (Ω, F) → (C, B(C)) measurable,
_
fdµ < +∞
_
Exercise 20. Let f : (Ω, F) → (C, B(C)) be a measurable map.
1. Explain why the integral
_
fdµ makes sense.
2. Show that f : (Ω, F) → (R, B(R)) is measurable, if f(Ω) ⊆ R.
3. Show that L
1
R
(Ω, F, µ) ⊆ L
1
C
(Ω, F, µ).
4. Show that L
1
R
(Ω, F, µ) = {f ∈ L
1
C
(Ω, F, µ) , f(Ω) ⊆ R}
5. Show that L
1
R
(Ω, F, µ) is closed under Rlinear combinations.
6. Show that L
1
C
(Ω, F, µ) is closed under Clinear combinations.
Tutorial 5: Lebesgue Integration 19
Deﬁnition 47 Let u : Ω → R be a realvalued function deﬁned on a
set Ω. We call positive part and negative part of u the maps u
+
and u
−
respectively, deﬁned as u
+
= max(u, 0) and u
−
= max(−u, 0).
Exercise 21. Let f ∈ L
1
C
(Ω, F, µ). Let u = Re(f) and v = Im(f).
1. Show that u = u
+
−u
−
, v = v
+
−v
−
, f = u
+
−u
−
+i(v
+
−v
−
).
2. Show that u = u
+
+ u
−
, v = v
+
+ v
−
3. Show that u
+
, u
−
, v
+
, v
−
, f, u, v, u, v all lie in L
1
R
(Ω, F, µ).
4. Explain why the integrals
_
u
+
dµ,
_
u
−
dµ,
_
v
+
dµ,
_
v
−
dµ are
all well deﬁned.
5. We deﬁne the integral of f with respect to µ, denoted
_
fdµ, as
_
fdµ =
_
u
+
dµ−
_
u
−
dµ+i
__
v
+
dµ −
_
v
−
dµ
_
. Explain why
_
fdµ is a well deﬁned complex number.
Tutorial 5: Lebesgue Integration 20
6. In the case when f(Ω) ⊆ C ∩ [0, +∞] = R
+
, explain why this
new deﬁnition of the integral of f with respect to µ is consistent
with the one already known (43) for nonnegative and measur
able maps.
7. Show that
_
fdµ =
_
udµ+i
_
vdµ and explain why all integrals
involved are well deﬁned.
Deﬁnition 48 Let f = u + iv ∈ L
1
C
(Ω, F, µ) where (Ω, F, µ) is a
measure space. We deﬁne the lebesgue integral of f with respect to
µ, denoted
_
fdµ, as:
_
fdµ
=
_
u
+
dµ −
_
u
−
dµ + i
__
v
+
dµ −
_
v
−
dµ
_
Exercise 22. Let f = u + iv ∈ L
1
C
(Ω, F, µ) and A ∈ F.
1. Show that f1
A
∈ L
1
C
(Ω, F, µ).
Tutorial 5: Lebesgue Integration 21
2. Show that f ∈ L
1
C
(Ω, F, µ
A
).
3. Show that f
A
∈ L
1
C
(A, F
A
, µ
A
)
4. Show that
_
(f1
A
)dµ =
_
fdµ
A
=
_
f
A
dµ
A
.
5. Show that 4. is:
_
A
u
+
dµ −
_
A
u
−
dµ + i
__
A
v
+
dµ −
_
A
v
−
dµ
_
.
Deﬁnition 49 Let f ∈ L
1
C
(Ω, F, µ) , where (Ω, F, µ) is a measure
space. let A ∈ F. We call partial lebesgue integral of f with
respect to µ over A, the integral denoted
_
A
fdµ, deﬁned as:
_
A
fdµ
=
_
(f1
A
)dµ =
_
fdµ
A
=
_
(f
A
)dµ
A
where µ
A
is the measure on (Ω, F), µ
A
= µ(A∩•), f
A
is the restric
tion of f to A and µ
A
is the restriction of µ to F
A
, the trace of F
on A.
Tutorial 5: Lebesgue Integration 22
Exercise 23. Let f, g ∈ L
1
R
(Ω, F, µ) and let h = f + g
1. Show that:
_
h
+
dµ +
_
f
−
dµ +
_
g
−
dµ =
_
h
−
dµ +
_
f
+
dµ +
_
g
+
dµ
2. Show that
_
hdµ =
_
fdµ +
_
gdµ.
3. Show that
_
(−f)dµ = −
_
fdµ
4. Show that if α ∈ R then
_
(αf)dµ = α
_
fdµ.
5. Show that if f ≤ g then
_
fdµ ≤
_
gdµ
6. Show the following theorem.
Theorem 22 For all f, g ∈ L
1
C
(Ω, F, µ) and α ∈ C, we have:
_
(αf + g)dµ = α
_
fdµ +
_
gdµ
Tutorial 5: Lebesgue Integration 23
Exercise 24. Let f, g be two maps, and (f
n
)
n≥1
be a sequence of
measurable maps f
n
: (Ω, F) → (C, B(C)), such that:
(i) ∀ω ∈ Ω , lim
n→+∞
f
n
(ω) = f(ω) in C
(ii) ∀n ≥ 1 , f
n
 ≤ g
(iii) g ∈ L
1
R
(Ω, F, µ)
Let (u
n
)
n≥1
be an arbitrary sequence in
¯
R.
1. Show that f ∈ L
1
C
(Ω, F, µ) and f
n
∈ L
1
C
(Ω, F, µ) for all n ≥ 1.
2. For n ≥ 1, deﬁne h
n
= 2g − f
n
− f. Explain why Fatou
lemma (20) can be applied to the sequence (h
n
)
n≥1
.
3. Show that liminf(−u
n
) = −limsup u
n
.
4. Show that if α ∈ R, then liminf(α + u
n
) = α + liminf u
n
.
5. Show that u
n
→ 0 as n → +∞ if and only if limsupu
n
 = 0.
6. Show that
_
(2g)dµ ≤
_
(2g)dµ −limsup
_
f
n
−fdµ
Tutorial 5: Lebesgue Integration 24
7. Show that limsup
_
f
n
−fdµ = 0.
8. Conclude that
_
f
n
−fdµ → 0 as n → +∞.
Theorem 23 (Dominated Convergence) Let (f
n
)
n≥1
be a se
quence of measurable maps f
n
: (Ω, F) → (C, B(C)) such that f
n
→ f
in C
2
. Suppose that there exists some g ∈ L
1
R
(Ω, F, µ) such that
f
n
 ≤ g for all n ≥ 1. Then f, f
n
∈ L
1
C
(Ω, F, µ) for all n ≥ 1, and:
lim
n→+∞
_
f
n
−fdµ = 0
Exercise 25. Let f ∈ L
1
C
(Ω, F, µ) and put z =
_
fdµ. Let α ∈ C,
be such that α = 1 and αz = z. Put u = Re(αf).
1. Show that u ∈ L
1
R
(Ω, F, µ)
2. Show that u ≤ f
2
i.e. for all ω ∈ Ω, the sequence (f
n
(ω))
n≥1
converges to f(ω) ∈ C
Tutorial 5: Lebesgue Integration 25
3. Show that 
_
fdµ =
_
(αf)dµ.
4. Show that
_
(αf)dµ =
_
udµ.
5. Prove the following theorem.
Theorem 24 Let f ∈ L
1
C
(Ω, F, µ) where (Ω, F, µ) is a measure
space. We have:
¸
¸
¸
¸
_
fdµ
¸
¸
¸
¸
≤
_
fdµ
Tutorial 6: Product Spaces 1
6. Product Spaces
In the following, I is a nonempty set.
Deﬁnition 50 Let (Ω
i
)
i∈I
be a family of sets, indexed by a non
empty set I. We call cartesian product of the family (Ω
i
)
i∈I
the
set, denoted Π
i∈I
Ω
i
, and deﬁned by:
i∈I
Ω
i
= ¦ω : I → ∪
i∈I
Ω
i
, ω(i) ∈ Ω
i
, ∀i ∈ I¦
In other words, Π
i∈I
Ω
i
is the set of all maps ω deﬁned on I, with
values in ∪
i∈I
Ω
i
, such that ω(i) ∈ Ω
i
for all i ∈ I.
Theorem 25 (Axiom of choice) Let (Ω
i
)
i∈I
be a family of sets,
indexed by a nonempty set I. Then, Π
i∈I
Ω
i
is nonempty, if and
only if Ω
i
is nonempty for all i ∈ I
1
.
1
When I is ﬁnite, this theorem is traditionally derived from other axioms.
Tutorial 6: Product Spaces 2
Exercise 1.
1. Let Ω be a set and suppose that Ω
i
= Ω, ∀i ∈ I. We use the
notation Ω
I
instead of Π
i∈I
Ω
i
. Show that Ω
I
is the set of all
maps ω : I → Ω.
2. What are the sets R
R
+
, R
N
, [0, 1]
N
,
¯
R
R
?
3. Suppose I = N
∗
. We sometimes use the notation Π
+∞
n=1
Ω
n
in
stead of Π
n∈N
∗Ω
n
. Let o be the set of all sequences (x
n
)
n≥1
such that x
n
∈ Ω
n
for all n ≥ 1. Is o the same thing as the
product Π
+∞
n=1
Ω
n
?
4. Suppose I = N
n
= ¦1, . . . , n¦, n ≥ 1. We use the notation
Ω
1
. . . Ω
n
instead of Π
i∈{1,...,n}
Ω
i
. For ω ∈ Ω
1
. . . Ω
n
, it
is customary to write (ω
1
, . . . , ω
n
) instead of ω, where we have
ω
i
= ω(i). What is your guess for the deﬁnition of sets such as
R
n
,
¯
R
n
, Q
n
, C
n
.
5. Let E, F, G be three sets. Deﬁne E F G.
Tutorial 6: Product Spaces 3
Deﬁnition 51 Let I be a nonempty set. We say that a family of
sets (I
λ
)
λ∈Λ
, where Λ ,= ∅, is a partition of I, if and only if:
(i) ∀λ ∈ Λ , I
λ
,= ∅
(ii) ∀λ, λ
∈ Λ , λ ,= λ
⇒ I
λ
∩ I
λ
= ∅
(iii) I = ∪
λ∈Λ
I
λ
Exercise 2. Let (Ω
i
)
i∈I
be a family of sets indexed by I, and (I
λ
)
λ∈Λ
be a partition of the set I.
1. For each λ ∈ Λ, recall the deﬁnition of Π
i∈I
λ
Ω
i
.
2. Recall the deﬁnition of Π
λ∈Λ
(Π
i∈I
λ
Ω
i
).
3. Deﬁne a natural bijection Φ : Π
i∈I
Ω
i
→ Π
λ∈Λ
(Π
i∈I
λ
Ω
i
).
4. Deﬁne a natural bijection ψ : R
p
R
n
→ R
p+n
, for all n, p ≥ 1.
Tutorial 6: Product Spaces 4
Deﬁnition 52 Let (Ω
i
)
i∈I
be a family of sets, indexed by a non
empty set I. For all i ∈ I, let c
i
be a set of subsets of Ω
i
. We deﬁne
a rectangle of the family (c
i
)
i∈I
, as any subset A of Π
i∈I
Ω
i
, of the
form A = Π
i∈I
A
i
where A
i
∈ c
i
∪ ¦Ω
i
¦ for all i ∈ I, and such that
A
i
= Ω
i
except for a ﬁnite number of indices i ∈ I. Consequently, the
set of all rectangles, denoted H
i∈I
c
i
, is deﬁned as:
i∈I
c
i
=
_
i∈I
A
i
: A
i
∈ c
i
∪ ¦Ω
i
¦ , A
i
,= Ω
i
for ﬁnitely many i ∈ I
_
Exercise 3. (Ω
i
)
i∈I
and (c
i
)
i∈I
being as above:
1. Show that if I = N
n
and Ω
i
∈ c
i
for all i = 1, . . . , n, then
c
1
H . . . H c
n
= ¦A
1
. . . A
n
: A
i
∈ c
i
, ∀i ∈ I¦.
2. Let A be a rectangle. Show that there exists a ﬁnite subset J
of I such that: A = ¦ω ∈ Π
i∈I
Ω
i
: ω(j) ∈ A
j
, ∀j ∈ J¦ for
some A
j
’s such that A
j
∈ c
j
, for all j ∈ J.
Tutorial 6: Product Spaces 5
Deﬁnition 53 Let (Ω
i
, T
i
)
i∈I
be a family of measurable spaces, in
dexed by a nonempty set I. We call measurable rectangle , any
rectangle of the family (T
i
)
i∈I
. The set of all measurable rectangles
is given by
2
:
i∈I
T
i
=
_
i∈I
A
i
: A
i
∈ T
i
, A
i
,= Ω
i
for ﬁnitely many i ∈ I
_
Deﬁnition 54 Let (Ω
i
, T
i
)
i∈I
be a family of measurable spaces, in
dexed by a nonempty set I. We deﬁne the product σalgebra of
(T
i
)
i∈I
, as the σalgebra on Π
i∈I
Ω
i
, denoted ⊗
i∈I
T
i
, and generated
by all measurable rectangles, i.e.
i∈I
T
i
= σ
_
i∈I
T
i
_
2
Note that Ω
i
∈ F
i
for all i ∈ I.
Tutorial 6: Product Spaces 6
Exercise 4.
1. Suppose I = N
n
. Show that T
1
⊗ . . . ⊗ T
n
is generated by all
sets of the form A
1
. . . A
n
, where A
i
∈ T
i
for all i = 1, . . . , n.
2. Show that B(R) ⊗B(R) ⊗B(R) is generated by sets of the form
AB C where A, B, C ∈ B(R).
3. Show that if (Ω, T) is a measurable space, B(R
+
) ⊗ T is the
σalgebra on R
+
Ω generated by sets of the form BF where
B ∈ B(R
+
) and F ∈ T.
Exercise 5. Let (Ω
i
)
i∈I
be a family of nonempty sets and c
i
be a
subset of the power set T(Ω
i
) for all i ∈ I.
1. Give a generator of the σalgebra ⊗
i∈I
σ(c
i
) on Π
i∈I
Ω
i
.
2. Show that:
σ
_
i∈I
c
i
_
⊆
i∈I
σ(c
i
)
Tutorial 6: Product Spaces 7
3. Let A be a rectangle of the family (σ(c
i
))
i∈I
. Show that if A is
not empty, then the representation A = Π
i∈I
A
i
with A
i
∈ σ(c
i
)
is unique. Deﬁne J
A
= ¦i ∈ I : A
i
,= Ω
i
¦. Explain why J
A
is a
welldeﬁned ﬁnite subset of I.
4. If A ∈ H
i∈I
σ(c
i
), Show that if A = ∅, or A ,= ∅ and J
A
= ∅,
then A ∈ σ(H
i∈I
c
i
).
Exercise 6. Everything being as before, Let n ≥ 0. We assume that
the following induction hypothesis has been proved:
A ∈
i∈I
σ(c
i
), A ,= ∅, cardJ
A
= n ⇒ A ∈ σ
_
i∈I
c
i
_
We assume that A is a non empty measurable rectangle of (σ(c
i
))
i∈I
with cardJ
A
= n + 1. Let J
A
= ¦i
1
, . . . , i
n+1
¦ be an extension of J
A
.
For all B ⊆ Ω
i
1
, we deﬁne:
A
B
=
i∈I
¯
A
i
Tutorial 6: Product Spaces 8
where each
¯
A
i
is equal to A
i
except
¯
A
i
1
= B. We deﬁne the set:
Γ
=
_
B ⊆ Ω
i
1
: A
B
∈ σ
_
i∈I
c
i
__
1. Show that A
Ω
i
1
,= ∅, cardJ
A
Ω
i
1
= n and that A
Ω
i
1
∈ H
i∈I
σ(c
i
).
2. Show that Ω
i
1
∈ Γ.
3. Show that for all B ⊆ Ω
i
1
, we have A
Ω
i
1
\B
= A
Ω
i
1
¸ A
B
.
4. Show that B ∈ Γ ⇒ Ω
i
1
¸ B ∈ Γ.
5. Let B
n
⊆ Ω
i
1
, n ≥ 1. Show that A
∪B
n
= ∪
n≥1
A
B
n
.
6. Show that Γ is a σalgebra on Ω
i
1
.
7. Let B ∈ c
i
1
, and for i ∈ I deﬁne
¯
B
i
= Ω
i
for all i’s except
¯
B
i
1
= B. Show that A
B
= A
Ω
i
1
∩ (Π
i∈I
¯
B
i
).
8. Show that σ(c
i
1
) ⊆ Γ.
Tutorial 6: Product Spaces 9
9. Show that A = A
A
i
1
and A ∈ σ(H
i∈I
c
i
).
10. Show that H
i∈I
σ(c
i
) ⊆ σ(H
i∈I
c
i
).
11. Show that σ(H
i∈I
c
i
) = ⊗
i∈I
σ(c
i
).
Theorem 26 Let (Ω
i
)
i∈I
be a family of nonempty sets indexed by a
nonempty set I. For all i ∈ I, let c
i
be a set of subsets of Ω
i
. Then,
the product σalgebra ⊗
i∈I
σ(c
i
) on the cartesian product Π
i∈I
Ω
i
is
generated by the rectangles of (c
i
)
i∈I
, i.e. :
i∈I
σ(c
i
) = σ
_
i∈I
c
i
_
Exercise 7. Let T
R
denote the usual topology in R. Let n ≥ 1.
1. Show that T
R
H . . . H T
R
= ¦A
1
. . . A
n
: A
i
∈ T
R
¦.
2. Show that B(R) ⊗. . . ⊗B(R) = σ(T
R
H . . . H T
R
).
Tutorial 6: Product Spaces 10
3. Deﬁne (
2
= ¦]a
1
, b
1
] . . . ]a
n
, b
n
] : a
i
, b
i
∈ R¦. Show that
(
2
⊆ o H . . . H o, where o = ¦]a, b] : a, b ∈ R¦, but that the
inclusion is strict.
4. Show that o H . . . H o ⊆ σ((
2
).
5. Show that B(R) ⊗. . . ⊗B(R) = σ((
2
).
Exercise 8. Let Ω and Ω
be two nonempty sets. Let A be a subset
of Ω such that ∅ ,= A ,= Ω. Let c = ¦A¦ ⊆ T(Ω) and c
= ∅ ⊆ T(Ω
).
1. Show that σ(c) = ¦∅, A, A
c
, Ω¦.
2. Show that σ(c
) = ¦∅, Ω
¦.
3. Deﬁne ( = ¦E F , E ∈ c, F ∈ c
¦ and show that ( = ∅.
4. Show that c H c
= ¦AΩ
, Ω Ω
¦.
5. Show that σ(c) ⊗σ(c
) = ¦∅, AΩ
, A
c
Ω
, Ω Ω
¦.
6. Conclude that σ(c) ⊗ σ(c
) ,= σ(() = ¦∅, ΩΩ
¦.
Tutorial 6: Product Spaces 11
Exercise 9. Let n ≥ 1 and p ≥ 1 be two positive integers.
1. Deﬁne T = B(R) ⊗ . . . ⊗B(R)
. ¸¸ .
n
, and ( = B(R) ⊗. . . ⊗B(R)
. ¸¸ .
p
.
Explain why T ⊗( can be viewed as a σalgebra on R
n+p
.
2. Show that T⊗( is generated by sets of the form A
1
. . . A
n+p
where A
i
∈ B(R), i = 1, . . . , n +p.
3. Show that:
B(R)⊗. . .⊗B(R)
. ¸¸ .
n+p
= (B(R)⊗. . .⊗B(R))
. ¸¸ .
n
⊗(B(R)⊗. . .⊗B(R))
. ¸¸ .
p
Exercise 10. Let (Ω
i
, T
i
)
i∈I
be a family of measurable spaces. Let
(I
λ
)
λ∈Λ
, where Λ ,= ∅, be a partition of I. Let Ω = Π
i∈I
Ω
i
and
Ω
= Π
λ∈Λ
(Π
i∈I
λ
Ω
i
).
1. Deﬁne a natural bijection between T(Ω) and T(Ω
).
Tutorial 6: Product Spaces 12
2. Show that through such bijection, A = Π
i∈I
A
i
⊆ Ω, where
A
i
⊆ Ω
i
, is identiﬁed with A
= Π
λ∈Λ
(Π
i∈I
λ
A
i
) ⊆ Ω
.
3. Show that H
i∈I
T
i
= H
λ∈Λ
(H
i∈I
λ
T
i
).
4. Show that ⊗
i∈I
T
i
= ⊗
λ∈Λ
(⊗
i∈I
λ
T
i
).
Deﬁnition 55 Let Ω be set and / be a set of subsets of Ω. We call
topology generated by /, the topology on Ω, denoted T (/), equal
to the intersection of all topologies on Ω, which contain /.
Exercise 11. Let Ω be a set and / ⊆ T(Ω).
1. Explain why T (/) is indeed a topology on Ω.
2. Show that T (/) is the smallest topology T such that / ⊆ T .
3. Show that the metric topology on a metric space (E, d) is gen
erated by the open balls / = ¦B(x, ) : x ∈ E, > 0¦.
Tutorial 6: Product Spaces 13
Deﬁnition 56 Let (Ω
i
, T
i
)
i∈I
be a family of topological spaces, in
dexed by a nonempty set I. We deﬁne the product topology of
(T
i
)
i∈I
, as the topology on Π
i∈I
Ω
i
, denoted ¸
i∈I
T
i
, and generated by
all rectangles of (T
i
)
i∈I
, i.e.
_
i∈I
T
i
= T
_
i∈I
T
i
_
Exercise 12. Let (Ω
i
, T
i
)
i∈I
be a family of topological spaces.
1. Show that U ∈ ¸
i∈I
T
i
, if and only if:
∀x ∈ U , ∃V ∈ H
i∈I
T
i
, x ∈ V ⊆ U
2. Show that H
i∈I
T
i
⊆ ¸
i∈I
T
i
.
3. Show that ⊗
i∈I
B(Ω
i
) = σ(H
i∈I
T
i
).
4. Show that ⊗
i∈I
B(Ω
i
) ⊆ B(Π
i∈I
Ω
i
).
Tutorial 6: Product Spaces 14
Exercise 13. Let n ≥ 1 be a positive integer. For all x, y ∈ R
n
, let:
(x, y)
=
n
i=1
x
i
y
i
and we put x =
_
(x, x).
1. Show that for all t ∈ R, x +ty
2
= x
2
+t
2
y
2
+ 2t(x, y).
2. From x +ty
2
≥ 0 for all t, deduce that [(x, y)[ ≤ x.y.
3. Conclude that x +y ≤ x +y.
Exercise 14. Let (Ω
1
, T
1
), . . . , (Ω
n
, T
n
), n ≥ 1, be metrizable topo
logical spaces. Let d
1
, . . . , d
n
be metrics on Ω
1
, . . . , Ω
n
, inducing the
topologies T
1
, . . . , T
n
respectively. Let Ω = Ω
1
. . . Ω
n
and T be
the product topology on Ω. For all x, y ∈ Ω, we deﬁne:
d(x, y)
=
¸
¸
¸
_
n
i=1
(d
i
(x
i
, y
i
))
2
Tutorial 6: Product Spaces 15
1. Show that d : Ω Ω → R
+
is a metric on Ω.
2. Show that U ⊆ Ω is open in Ω, if and only if, for all x ∈ U there
are open sets U
1
, . . . , U
n
in Ω
1
, . . . , Ω
n
respectively, such that:
x ∈ U
1
. . . U
n
⊆ U
3. Let U ∈ T and x ∈ U. Show the existence of > 0 such that:
(∀i = 1, . . . , n d
i
(x
i
, y
i
) < ) ⇒ y ∈ U
4. Show that T ⊆ T
d
Ω
.
5. let U ∈ T
d
Ω
and x ∈ U. Show that existence of > 0 such that:
x ∈ B(x
1
, ) . . . B(x
n
, ) ⊆ U
6. Show that T
d
Ω
⊆ T .
7. Show that the product topological space (Ω, T ) is metrizable.
Tutorial 6: Product Spaces 16
8. For all x, y ∈ Ω, deﬁne:
d
(x, y)
=
n
i=1
d
i
(x
i
, y
i
)
d
(x, y)
= max
i=1,...,n
d
i
(x
i
, y
i
)
Show that d
, d
are metrics on Ω.
9. Show the existence of α
, β
, α
and β
> 0, such that we have
α
d
≤ d ≤ β
d
and α
d
≤ d ≤ β
d
.
10. Show that d
and d
also induce the product topology on Ω.
Exercise 15. Let (Ω
n
, T
n
)
n≥1
be a sequence of metrizable topological
spaces. For all n ≥ 1, let d
n
be a metric on Ω
n
inducing the topology
T
n
. Let Ω = Π
+∞
n=1
Ω
n
be the cartesian product and T be the product
Tutorial 6: Product Spaces 17
topology on Ω. For all x, y ∈ Ω, we deﬁne:
d(x, y)
=
+∞
n=1
1
2
n
(1 ∧ d
n
(x
n
, y
n
))
1. Show that for all a, b ∈ R
+
, we have 1 ∧ (a +b) ≤ 1 ∧ a +1 ∧ b.
2. Show that d is a metric on Ω.
3. Show that U ⊆ Ω is open in Ω, if and only if, for all x ∈ U, there
is an integer N ≥ 1 and open sets U
1
, . . . , U
N
in Ω
1
, . . . , Ω
N
respectively, such that:
x ∈ U
1
. . . U
N
+∞
n=N+1
Ω
n
⊆ U
4. Show that d(x, y) < 1/2
n
⇒ d
n
(x
n
, y
n
) ≤ 2
n
d(x, y).
5. Show that for all U ∈ T and x ∈ U, there exists > 0 such that
d(x, y) < ⇒ y ∈ U.
Tutorial 6: Product Spaces 18
6. Show that T ⊆ T
d
Ω
.
7. Let U ∈ T
d
Ω
and x ∈ U. Show the existence of > 0 and N ≥ 1,
such that:
N
n=1
1
2
n
(1 ∧ d
n
(x
n
, y
n
)) < ⇒ y ∈ U
8. Show that for all U ∈ T
d
Ω
and x ∈ U, there is > 0 and N ≥ 1
such that:
x ∈ B(x
1
, ) . . . B(x
N
, )
+∞
n=N+1
Ω
n
⊆ U
9. Show that T
d
Ω
⊆ T .
10. Show that the product topological space (Ω, T ) is metrizable.
Tutorial 6: Product Spaces 19
Deﬁnition 57 Let (Ω, T ) be a topological space. A subset H of T
is called a countable base of (Ω, T ), if and only if H is at most
countable, and has the property:
∀U ∈ T , ∃H
⊆ H , U =
_
V ∈H
V
Exercise 16.
1. Show that H = ¦]r, q[ : r, q ∈ Q¦ is a countable base of (R, T
R
).
2. Show that if (Ω, T ) is a topological space with countable base,
and Ω
⊆ Ω, then the induced topological space (Ω
, T
Ω
) also
has a countable base.
3. Show that [−1, 1] has a countable base.
4. Show that if (Ω, T ) and (S, T
S
) are homeomorphic, then (Ω, T )
has a countable base if and only if (S, T
S
) has a countable base.
5. Show that (
¯
R, T
¯
R
) has a countable base.
Tutorial 6: Product Spaces 20
Exercise 17. Let (Ω
n
, T
n
)
n≥1
be a sequence of topological spaces
with countable base. For n ≥ 1, Let ¦V
k
n
: k ∈ I
n
¦ be a countable
base of (Ω
n
, T
n
) where I
n
is a ﬁnite or countable set. Let Ω = Π
∞
n=1
Ω
n
be the cartesian product and T be the product topology on Ω. For
all p ≥ 1, we deﬁne:
H
p
=
_
V
k
1
1
. . . V
k
p
p
+∞
n=p+1
Ω
n
: (k
1
, . . . , k
p
) ∈ I
1
. . . I
p
_
and we put H = ∪
p≥1
H
p
.
1. Show that for all p ≥ 1, H
p
⊆ T .
2. Show that H ⊆ T .
3. For all p ≥ 1, show the existence of an injection j
p
: H
p
→ N
p
.
4. Show the existence of a bijection φ
2
: N
2
→ N.
5. For p ≥ 1, show the existence of an bijection φ
p
: N
p
→ N.
Tutorial 6: Product Spaces 21
6. Show that H
p
is at most countable for all p ≥ 1.
7. Show the existence of an injection j : H → N
2
.
8. Show that H is a ﬁnite or countable set of open sets in Ω.
9. Let U ∈ T and x ∈ U. Show that there is p ≥ 1 and U
1
, . . . , U
p
open sets in Ω
1
, . . . , Ω
p
such that:
x ∈ U
1
. . . U
p
+∞
n=p+1
Ω
n
⊆ U
10. Show the existence of some V
x
∈ H such that x ∈ V
x
⊆ U.
11. Show that H is a countable base of the topological space (Ω, T ).
12. Show that ⊗
+∞
n=1
B(Ω
n
) ⊆ B(Ω).
13. Show that H ⊆ ⊗
+∞
n=1
B(Ω
n
).
14. Show that B(Ω) = ⊗
+∞
n=1
B(Ω
n
)
Tutorial 6: Product Spaces 22
Theorem 27 Let (Ω
n
, T
n
)
n≥1
be a sequence of topological spaces
with countable base. Then, the product space (Π
+∞
n=1
Ω
n
, ¸
+∞
n=1
T
n
) has
a countable base and:
B
_
+∞
n=1
Ω
n
_
=
+∞
n=1
B(Ω
n
)
Exercise 18.
1. Show that if (Ω, T ) has a countable base and n ≥ 1:
B(Ω
n
) = B(Ω) ⊗. . . ⊗B(Ω)
. ¸¸ .
n
2. Show that B(
¯
R
n
) = B(
¯
R) ⊗. . . ⊗B(
¯
R).
3. Show that B(C) = B(R) ⊗ B(R).
Tutorial 6: Product Spaces 23
Deﬁnition 58 We say that a metric space (E, d) is separable, if
and only if there exists a ﬁnite or countable dense subset of E, i.e.
a ﬁnite or countable subset A of E such that E =
¯
A, where
¯
A is the
closure of A in E.
Exercise 19. Let (E, d) be a metric space.
1. Suppose that (E, d) is separable. Let H = ¦B(x
n
,
1
p
) : n, p ≥ 1¦,
where ¦x
n
: n ≥ 1¦ is a countable dense subset in E. Show that
H is a countable base of the metric topological space (E, T
d
E
).
2. Suppose conversely that (E, T
d
E
) has a countable base H. For
all V ∈ H such that V ,= ∅, take x
V
∈ V . Show that the set
¦x
V
: V ∈ H , V ,= ∅¦ is at most countable and dense in E.
3. For all x, y, x
, y
∈ E, show that:
[d(x, y) −d(x
, y
)[ ≤ d(x, x
) +d(y, y
)
Tutorial 6: Product Spaces 24
4. Let T
E×E
be the product topology on E E. Show that the
map d : (E E, T
E×E
) → (R
+
, T
R
+) is continuous.
5. Show that d : (E E, B(E E)) → (
¯
R, B(
¯
R)) is measurable.
6. Show that d : (EE, B(E)⊗B(E)) → (
¯
R, B(
¯
R)) is measurable,
whenever (E, d) is a separable metric space.
7. Let (Ω, T) be a measurable space and f, g : (Ω, T) → (E, B(E))
be measurable maps. Show that Φ : (Ω, T) → EE deﬁned by
Φ(ω) = (f(ω), g(ω)) is measurable with respect to the product
σalgebra B(E) ⊗B(E).
8. Show that if (E, d) is separable, then Ψ : (Ω, T) → (
¯
R, B(
¯
R))
deﬁned by Ψ(ω) = d(f(ω), g(ω)) is measurable.
9. Show that if (E, d) is separable then ¦f = g¦ ∈ T.
10. Let (E
n
, d
n
)
n≥1
be a sequence of separable metric spaces. Show
that the product space Π
+∞
n=1
E
n
is metrizable and separable.
Tutorial 6: Product Spaces 25
Exercise 20. Prove the following theorem.
Theorem 28 Let (Ω
i
, T
i
)
i∈I
be a family of measurable spaces and
(Ω, T) be a measurable space. For all i ∈ I, let f
i
: Ω → Ω
i
be a map,
and deﬁne f : Ω → Π
i∈I
Ω
i
by f(ω) = (f
i
(ω))
i∈I
. Then, the map:
f : (Ω, T) →
_
i∈I
Ω
i
,
i∈I
T
i
_
is measurable, if and only if each f
i
: (Ω, T) → (Ω
i
, T
i
) is measurable.
Exercise 21.
1. Let φ, ψ : R
2
→ R with φ(x, y) = x + y and ψ(x, y) = x.y.
Show that both φ and ψ are continuous.
2. Show that φ, ψ : (R
2
, B(R)⊗B(R))→(
¯
R, B(
¯
R)) are measurable.
3. Let (Ω, T) be a measurable space, and f, g : (Ω, T) → (R, B(R))
be measurable maps. Using the previous results, show that f +g
and f.g are measurable with respect to T and B(R).
Tutorial 7: Fubini Theorem 1
7. Fubini Theorem
Deﬁnition 59 Let (Ω
1
, F
1
) and (Ω
2
, F
2
) be two measurable spaces.
Let E ⊆ Ω
1
× Ω
2
. For all ω
1
∈ Ω
1
, we call ω
1
section of E in Ω
2
,
the set:
E
ω
1
= {ω
2
∈ Ω
2
: (ω
1
, ω
2
) ∈ E}
Exercise 1. Let (Ω
1
, F
1
) and (Ω
2
, F
2
) be two measurable spaces.
Given ω
1
∈ Ω
1
, deﬁne:
Γ
ω
1
= {E ⊆ Ω
1
×Ω
2
, E
ω
1
∈ F
2
}
1. Show that for all ω
1
∈ Ω
1
, Γ
ω
1
is a σalgebra on Ω
1
×Ω
2
.
2. Show that for all ω
1
∈ Ω
1
, F
1
F
2
⊆ Γ
ω
1
.
3. Show that for all ω
1
∈ Ω
1
and E ∈ F
1
⊗F
2
, we have E
ω
1
∈ F
2
.
4. Show that the map ω → 1
E
(ω
1
, ω) is measurable with respect
to F
2
and B(
¯
R), for all ω
1
∈ Ω
1
and E ∈ F
1
⊗F
2
.
Tutorial 7: Fubini Theorem 2
5. Let s be a simple function on (Ω
1
×Ω
2
, F
1
⊗F
2
). Show that for
all ω
1
∈ Ω
1
, the map ω → s(ω
1
, ω) is measurable with respect
to F
2
and B(
¯
R).
6. Let f : (Ω
1
× Ω
2
, F
1
⊗ F
2
) → [0, +∞] be a nonnegative, mea
surable map. Show that for all ω
1
∈ Ω
1
, the map ω → f(ω
1
, ω)
is measurable with respect to F
2
and B(
¯
R).
7. Let f : (Ω
1
× Ω
2
, F
1
⊗ F
2
) → (
¯
R, B(
¯
R)) be a measurable map.
Show that for all ω
1
∈ Ω
1
, the map ω → f(ω
1
, ω) is measurable
with respect to F
2
and B(
¯
R).
8. Show the following theorem:
Theorem 29 Let (E, d) be a metric space, and (Ω
1
, F
1
), (Ω
2
, F
2
)
be two measurable spaces. Let f : (Ω
1
×Ω
2
, F
1
⊗F
2
) → (E, B(E)) be
a measurable map . Then for all ω
1
∈ Ω
1
, the map ω → f(ω
1
, ω) is
measurable with respect to F
2
and B(E).
Tutorial 7: Fubini Theorem 3
Exercise 2. Let (Ω
i
, F
i
)
i∈I
be a family of measurable spaces with
cardI ≥ 2. Let f : (Π
i∈I
Ω
i
, ⊗
i∈I
F
i
) → (E, B(E)) be a measurable
map, where (E, d) be a metric space. Let i
1
∈ I. Put E
1
= Ω
i
1
,
E
1
= F
i
1
, E
2
= Π
i∈I\{i
1
}
Ω
i
, E
2
= ⊗
i∈I\{i
1
}
F
i
.
1. Explain why f can be viewed as a map deﬁned on E
1
×E
2
.
2. Show that f : (E
1
×E
2
, E
1
⊗E
2
) → (E, B(E)) is measurable.
3. For all ω
i
1
∈ Ω
i
1
, show that the map ω → f(ω
i
1
, ω) deﬁned on
Π
i∈I\{i
1
}
Ω
i
is measurable w.r. to ⊗
i∈I\{i
1
}
F
i
and B(E).
Deﬁnition 60 Let (Ω, F, µ) be a measure space. (Ω, F, µ) is said to
be a ﬁnite measure space, or we say that µ is a ﬁnite measure,
if and only if µ(Ω) < +∞.
Tutorial 7: Fubini Theorem 4
Deﬁnition 61 Let (Ω, F, µ) be a measure space. (Ω, F, µ) is said
to be a σﬁnite measure space, or µ a σﬁnite measure, if and
only if there exists a sequence (Ω
n
)
n≥1
in F such that Ω
n
↑ Ω and
µ(Ω
n
) < +∞, for all n ≥ 1.
Exercise 3. Let (Ω, F, µ) be a measure space.
1. Show that (Ω, F, µ) is σﬁnite if and only if there exists a se
quence (Ω
n
)
n≥1
in F such that Ω =
+∞
n=1
Ω
n
, and µ(Ω
n
) < +∞
for all n ≥ 1.
2. Show that if (Ω, F, µ) is ﬁnite, then µ has values in R
+
.
3. Show that if (Ω, F, µ) is ﬁnite, then it is σﬁnite.
4. Let F : R → R be a rightcontinuous, nondecreasing map.
Show that the measure space (R, B(R), dF) is σﬁnite, where
dF is the stieltjes measure associated with F.
Tutorial 7: Fubini Theorem 5
Exercise 4. Let (Ω
1
, F
1
) be a measurable space, and (Ω
2
, F
2
, µ
2
) be
a σﬁnite measure space. For all E ∈ F
1
⊗F
2
and ω
1
∈ Ω
1
, deﬁne:
Φ
E
(ω
1
)
=
Ω
2
1
E
(ω
1
, x)dµ
2
(x)
Let D be the set of subsets of Ω
1
×Ω
2
, deﬁned by:
D
= {E ∈ F
1
⊗F
2
: Φ
E
: (Ω
1
, F
1
) → (
¯
R, B(
¯
R)) is measurable}
1. Explain why for all E ∈ F
1
⊗F
2
, the map Φ
E
is well deﬁned.
2. Show that F
1
F
2
⊆ D.
3. Show that if µ
2
is ﬁnite, A, B ∈ D and A ⊆ B, then B \ A ∈ D.
4. Show that if E
n
∈ F
1
⊗F
2
, n ≥ 1 and E
n
↑ E, then Φ
E
n
↑ Φ
E
.
5. Show that if µ
2
is ﬁnite then D is a dynkin system on Ω
1
×Ω
2
.
6. Show that if µ
2
is ﬁnite, then the map Φ
E
: (Ω
1
, F
1
)→(
¯
R, B(
¯
R))
is measurable, for all E ∈ F
1
⊗F
2
.
Tutorial 7: Fubini Theorem 6
7. Let (Ω
n
2
)
n≥1
in F
2
be such that Ω
n
2
↑ Ω
2
and µ
2
(Ω
n
2
) < +∞.
Deﬁne µ
n
2
= µ
Ω
n
2
2
= µ
2
(• ∩ Ω
n
2
). For E ∈ F
1
⊗F
2
, we put:
Φ
n
E
(ω
1
)
=
Ω
2
1
E
(ω
1
, x)dµ
n
2
(x)
Show that Φ
n
E
: (Ω
1
, F
1
) → (
¯
R, B(
¯
R)) is measurable, and:
Φ
n
E
(ω
1
) =
Ω
2
1
Ω
n
2
(x)1
E
(ω
1
, x)dµ
2
(x)
Deduce that Φ
n
E
↑ Φ
E
.
8. Show that the map Φ
E
: (Ω
1
, F
1
) → (
¯
R, B(
¯
R)) is measurable,
for all E ∈ F
1
⊗F
2
.
9. Let s be a simple function on (Ω
1
× Ω
2
, F
1
⊗ F
2
). Show that
the map ω →
Ω
2
s(ω, x)dµ
2
(x) is well deﬁned and measurable
with respect to F
1
and B(
¯
R).
10. Show the following theorem:
Tutorial 7: Fubini Theorem 7
Theorem 30 Let (Ω
1
, F
1
) be a measurable space, and (Ω
2
, F
2
, µ
2
)
be a σﬁnite measure space. Then for all nonnegative and measurable
map f : (Ω
1
×Ω
2
, F
1
⊗F
2
) → [0, +∞], the map:
ω →
Ω
2
f(ω, x)dµ
2
(x)
is measurable with respect to F
1
and B(
¯
R).
Exercise 5. Let (Ω
i
, F
i
)
i∈I
be a family of measurable spaces, with
cardI ≥ 2. Let i
0
∈ I, and suppose that µ
0
is a σﬁnite measure
on (Ω
i
0
, F
i
0
). Show that if f : (Π
i∈I
Ω
i
, ⊗
i∈I
F
i
) → [0, +∞] is a non
negative and measurable map, then:
ω →
Ω
i
0
f(ω, x)dµ
0
(x)
deﬁned on Π
i∈I\{i
0
}
Ω
i
, is measurable w.r. to ⊗
i∈I\{i
0
}
F
i
and B(
¯
R).
Tutorial 7: Fubini Theorem 8
Exercise 6. Let (Ω
1
, F
1
, µ
1
) and (Ω
2
, F
2
, µ
2
) be two σﬁnite measure
spaces. For all E ∈ F
1
⊗F
2
, we deﬁne:
µ
1
⊗µ
2
(E)
=
Ω
1
Ω
2
1
E
(x, y)dµ
2
(y)
dµ
1
(x)
1. Explain why µ
1
⊗µ
2
: F
1
⊗F
2
→ [0, +∞] is well deﬁned.
2. Show that µ
1
⊗µ
2
is a measure on F
1
⊗F
2
.
3. Show that if A×B ∈ F
1
F
2
, then:
µ
1
⊗µ
2
(A ×B) = µ
1
(A)µ
2
(B)
Exercise 7. Further to ex. (6), suppose that µ : F
1
⊗F
2
→ [0, +∞]
is another measure on F
1
⊗ F
2
with µ(A × B) = µ
1
(A)µ
2
(B), for all
measurable rectangle A×B. Let (Ω
n
1
)
n≥1
and (Ω
n
2
)
n≥1
be sequences
in F
1
and F
2
respectively, such that Ω
n
1
↑ Ω
1
, Ω
n
2
↑ Ω
2
, µ
1
(Ω
n
1
) < +∞
and µ
2
(Ω
n
2
) < +∞. Deﬁne, for all n ≥ 1:
D
n
= {E ∈ F
1
⊗F
2
: µ(E ∩ (Ω
n
1
×Ω
n
2
)) = µ
1
⊗µ
2
(E ∩ (Ω
n
1
×Ω
n
2
))}
Tutorial 7: Fubini Theorem 9
1. Show that for all n ≥ 1, F
1
F
2
⊆ D
n
.
2. Show that for all n ≥ 1, D
n
is a dynkin system on Ω
1
×Ω
2
.
3. Show that µ = µ
1
⊗µ
2
.
4. Show that (Ω
1
×Ω
2
, F
1
⊗F
2
, µ
1
⊗µ
2
) is a σﬁnite measure space.
5. Show that for all E ∈ F
1
⊗F
2
, we have:
µ
1
⊗µ
2
(E) =
Ω
2
Ω
1
1
E
(x, y)dµ
1
(x)
dµ
2
(y)
Exercise 8. Let (Ω
1
, F
1
, µ
1
), . . . , (Ω
n
, F
n
, µ
n
) be n σﬁnite measure
spaces, n ≥ 2. Let i
0
∈ {1, . . . , n} and put E
1
= Ω
i
0
, E
2
= Π
i=i
0
Ω
i
,
E
1
= F
i
0
and E
2
= ⊗
i=i
0
F
i
. Put ν
1
= µ
i
0
, and suppose that ν
2
is
a σﬁnite measure on (E
2
, E
2
) such that for all measurable rectangle
Π
i=i
0
A
i
∈
i=i
0
F
i
, we have ν
2
(Π
i=i
0
A
i
) = Π
i=i
0
µ
i
(A
i
).
Tutorial 7: Fubini Theorem 10
1. Show that ν
1
⊗ ν
2
is a σﬁnite measure on the measure space
(Ω
1
× . . . × Ω
n
, F
1
⊗ . . . ⊗ F
n
) such that for all measurable
rectangles A
1
×. . . ×A
n
, we have:
ν
1
⊗ν
2
(A
1
×. . . ×A
n
) = µ
1
(A
1
) . . . µ
n
(A
n
)
2. Show by induction the existence of a measure µ on F
1
⊗. . .⊗F
n
,
such that for all measurable rectangles A
1
×. . . ×A
n
, we have:
µ(A
1
×. . . ×A
n
) = µ
1
(A
1
) . . . µ
n
(A
n
)
3. Show the uniqueness of such measure, denoted µ
1
⊗. . . ⊗µ
n
.
4. Show that µ
1
⊗. . . ⊗µ
n
is σﬁnite.
5. Let i
0
∈ {1, . . . , n}. Show that µ
i
0
⊗(⊗
i=i
0
µ
i
) = µ
1
⊗. . . ⊗µ
n
.
Tutorial 7: Fubini Theorem 11
Deﬁnition 62 Let (Ω
1
, F
1
, µ
1
), . . . , (Ω
n
, F
n
, µ
n
) be n σﬁnite mea
sure spaces, with n ≥ 2. We call product measure of µ
1
, . . . , µ
n
,
the unique measure on F
1
⊗. . . ⊗F
n
, denoted µ
1
⊗. . . ⊗µ
n
, such that
for all measurable rectangles A
1
×. . . ×A
n
in F
1
. . . F
n
, we have:
µ
1
⊗. . . ⊗µ
n
(A
1
×. . . ×A
n
) = µ
1
(A
1
) . . . µ
n
(A
n
)
This measure is itself σﬁnite.
Exercise 9. Prove that the following deﬁnition is legitimate:
Deﬁnition 63 We call lebesgue measure in R
n
, n ≥ 1, the
unique measure on (R
n
, B(R
n
)), denoted dx, dx
n
or dx
1
. . . dx
n
, such
that for all a
i
≤ b
i
, i = 1, . . . , n, we have:
dx([a
1
, b
1
] ×. . . ×[a
n
, b
n
]) =
n
i=1
(b
i
−a
i
)
Tutorial 7: Fubini Theorem 12
Exercise 10.
1. Show that (R
n
, B(R
n
), dx
n
) is a σﬁnite measure space.
2. For n, p ≥ 1, show that dx
n+p
= dx
n
⊗dx
p
.
Exercise 11. Let (Ω
1
, F
1
, µ
1
) and (Ω
2
, F
2
, µ
2
) be σﬁnite.
1. Let s be a simple function on (Ω
1
×Ω
2
, F
1
⊗F
2
). Show that:
Ω
1
×Ω
2
sdµ
1
⊗µ
2
=
Ω
1
Ω
2
sdµ
2
dµ
1
=
Ω
2
Ω
1
sdµ
1
dµ
2
2. Show the following:
Theorem 31 (Fubini) Let (Ω
1
, F
1
, µ
1
) and (Ω
2
, F
2
, µ
2
) be two σ
ﬁnite measure spaces. Let f : (Ω
1
× Ω
2
, F
1
⊗ F
2
) → [0, +∞] be a
nonnegative and measurable map. Then:
Ω
1
×Ω
2
fdµ
1
⊗µ
2
=
Ω
1
Ω
2
fdµ
2
dµ
1
=
Ω
2
Ω
1
fdµ
1
dµ
2
Tutorial 7: Fubini Theorem 13
Exercise 12. Let (Ω
1
, F
1
, µ
1
), . . . , (Ω
n
, F
n
, µ
n
) be n σﬁnite measure
spaces, n ≥ 2. Let f : (Ω
1
×. . . ×Ω
n
, F
1
⊗ . . . ⊗F
n
) → [0, +∞] be a
nonnegative, measurable map. Let σ be a permutation of N
n
, i.e. a
bijection from N
n
to itself.
1. For all ω ∈ Π
i=σ(1)
Ω
i
, deﬁne:
J
1
(ω)
=
Ω
σ(1)
f(ω, x)dµ
σ(1)
(x)
Explain why J
1
: (Π
i=σ(1)
Ω
i
, ⊗
i=σ(1)
F
i
) → [0, +∞] is a well
deﬁned, nonnegative and measurable map.
2. Suppose J
k
: (Π
i∈{σ(1),...,σ(k)}
Ω
i
, ⊗
i∈{σ(1),...,σ(k)}
F
i
) → [0, +∞]
is a nonnegative, measurable map, for 1 ≤ k < n −2. Deﬁne:
J
k+1
(ω)
=
Ω
σ(k+1)
J
k
(ω, x)dµ
σ(k+1)
(x)
and show that:
J
k+1
: (Π
i∈{σ(1),...,σ(k+1)}
Ω
i
, ⊗
i∈{σ(1),...,σ(k+1)}
F
i
) → [0, +∞]
Tutorial 7: Fubini Theorem 14
is also welldeﬁned, nonnegative and measurable.
3. Propose a rigorous deﬁnition for the following notation:
Ω
σ(n)
. . .
Ω
σ(1)
fdµ
σ(1)
. . . dµ
σ(n)
Exercise 13. Further to ex. (12), Let (f
p
)
p≥1
be a sequence of non
negative and measurable maps:
f
p
: (Ω
1
×. . . ×Ω
n
, F
1
⊗. . . ⊗F
n
) → [0, +∞]
such that f
p
↑ f. Deﬁne similarly:
J
p
1
(ω)
=
Ω
σ(1)
f
p
(ω, x)dµ
σ(1)
(x)
J
p
k+1
(ω)
=
Ω
σ(k+1)
J
p
k
(ω, x)dµ
σ(k+1)
(x) , 1 ≤ k < n −2
1. Show that J
p
1
↑ J
1
.
Tutorial 7: Fubini Theorem 15
2. Show that if J
p
k
↑ J
k
, then J
p
k+1
↑ J
k+1
, 1 ≤ k < n −2.
3. Show that:
Ω
σ(n)
. . .
Ω
σ(1)
f
p
dµ
σ(1)
. . . dµ
σ(n)
↑
Ω
σ(n)
. . .
Ω
σ(1)
fdµ
σ(1)
. . . dµ
σ(n)
4. Show that the map µ : F
1
⊗. . . ⊗F
n
→ [0, +∞], deﬁned by:
µ(E) =
Ω
σ(n)
. . .
Ω
σ(1)
1
E
dµ
σ(1)
. . . dµ
σ(n)
is a measure on F
1
⊗. . . ⊗F
n
.
5. Show that for all E ∈ F
1
⊗. . . ⊗F
n
, we have:
µ
1
⊗. . . ⊗µ
n
(E) =
Ω
σ(n)
. . .
Ω
σ(1)
1
E
dµ
σ(1)
. . . dµ
σ(n)
6. Show the following:
Tutorial 7: Fubini Theorem 16
Theorem 32 Let (Ω
1
, F
1
, µ
1
), . . . , (Ω
n
, F
n
, µ
n
) be n σﬁnite mea
sure spaces, with n ≥ 2. Let f : (Ω
1
×. . .×Ω
n
, F
1
⊗. . .⊗F
n
) → [0, +∞]
be a nonnegative and measurable map. let σ be a permutation of N
n
.
Then:
Ω
1
×...×Ω
n
fdµ
1
⊗. . . ⊗µ
n
=
Ω
σ(n)
. . .
Ω
σ(1)
fdµ
σ(1)
. . . dµ
σ(n)
Exercise 14. Let (Ω, F, µ) be a measure space. Deﬁne:
L
1
= {f : Ω →
¯
R , ∃g ∈ L
1
R
(Ω, F, µ) , f = g µa.s.}
1. Show that if f ∈ L
1
, then f < +∞, µa.s.
2. Suppose there exists A ⊆ Ω, such that A ∈ F and A ⊆ N for
some N ∈ F with µ(N) = 0. Show that 1
A
∈ L
1
and 1
A
is not
measurable with respect to F and B(
¯
R).
3. Explain why if f ∈ L
1
, the integrals
fdµ and
fdµ may not
be well deﬁned.
Tutorial 7: Fubini Theorem 17
4. Suppose that f : (Ω, F) → (
¯
R, B(
¯
R)) is a measurable map with
fdµ < +∞. Show that f ∈ L
1
.
5. Show that if f ∈ L
1
and f = f
1
µa.s. then f
1
∈ L
1
.
6. Suppose that f ∈ L
1
and g
1
, g
2
∈ L
1
R
(Ω, F, µ) are such that
f = g
1
µa.s. and f = g
2
µa.s.. Show that
g
1
dµ =
g
2
dµ.
7. Propose a deﬁnition of the integral
fdµ for f ∈ L
1
which
extends the integral deﬁned on L
1
R
(Ω, F, µ).
Exercise 15. Further to ex. (14), Let (f
n
)
n≥1
be a sequence in L
1
,
and f, h ∈ L
1
, with f
n
→ f µa.s. and for all n ≥ 1, f
n
 ≤ h µa.s..
1. Show the existence of N
1
∈ F, µ(N
1
) = 0, such that for all
ω ∈ N
c
1
, f
n
(ω) → f(ω), and for all n ≥ 1, f
n
(ω) ≤ h(ω).
2. Show the existence of g
n
, g, h
1
∈ L
1
R
(Ω, F, µ) and N
2
∈ F,
µ(N
2
) = 0, such that for all ω ∈ N
c
2
, g(ω) = f(ω), h(ω) = h
1
(ω),
and for all n ≥ 1, g
n
(ω) = f
n
(ω).
Tutorial 7: Fubini Theorem 18
3. Show the existence of N ∈ F, µ(N) = 0, such that for all
ω ∈ N
c
, g
n
(ω) → g(ω), and for all n ≥ 1, g
n
(ω) ≤ h
1
(ω).
4. Show that the Dominated Convergence Theorem can be applied
to g
n
1
N
c, g1
N
c and h
1
1
N
c.
5. Recall the deﬁnition of
f
n
−fdµ when f, f
n
∈ L
1
.
6. Show that
f
n
−fdµ → 0.
Exercise 16. Let (Ω
1
, F
1
, µ
1
) and (Ω
2
, F
2
, µ
2
) be two σﬁnite mea
sure spaces. let f be an element of L
1
R
(Ω
1
×Ω
2
, F
1
⊗ F
2
, µ
1
⊗µ
2
).
1. Let A = {ω
1
∈ Ω
1
:
Ω
2
f(ω
1
, x)dµ
2
(x) < +∞}. Show that
A ∈ F
1
and µ
1
(A
c
) = 0.
2. Show that f(ω
1
, .) ∈ L
1
R
(Ω
2
, F
2
, µ
2
) for all ω
1
∈ A.
3. Show that
¯
I(ω
1
) =
Ω
2
f(ω
1
, x)dµ
2
(x) is well deﬁned for all
ω
1
∈ A. Let I be an arbitrary extension of
¯
I, on Ω
1
.
Tutorial 7: Fubini Theorem 19
4. Deﬁne J = I1
A
. Show that:
J(ω) = 1
A
(ω)
Ω
2
f
+
(ω, x)dµ
2
(x) −1
A
(ω)
Ω
2
f
−
(ω, x)dµ
2
(x)
5. Show that J is F
1
measurable and Rvalued.
6. Show that J ∈ L
1
R
(Ω
1
, F
1
, µ
1
) and that J = I µ
1
a.s.
7. Propose a deﬁnition for the integral:
Ω
1
Ω
2
f(x, y)dµ
2
(y)
dµ
1
(x)
8. Show that
Ω
1
(1
A
Ω
2
f
+
dµ
2
)dµ
1
=
Ω
1
×Ω
2
f
+
dµ
1
⊗µ
2
.
9. Show that:
Ω
1
Ω
2
f(x, y)dµ
2
(y)
dµ
1
(x) =
Ω
1
×Ω
2
fdµ
1
⊗µ
2
10. Prove the following:
Tutorial 7: Fubini Theorem 20
Theorem 33 Let (Ω
1
, F
1
, µ
1
) and (Ω
2
, F
2
, µ
2
) be two σﬁnite mea
sure spaces. Let f ∈ L
1
C
(Ω
1
×Ω
2
, F
1
⊗F
2
, µ
1
⊗µ
2
). Then, the map:
ω
1
→
Ω
2
f(ω
1
, x)dµ
2
(x)
is µ
1
almost surely equal to an element of L
1
C
(Ω
1
, F
1
, µ
1
) and:
Ω
1
Ω
2
f(x, y)dµ
2
(y)
dµ
1
(x) =
Ω
1
×Ω
2
fdµ
1
⊗µ
2
Exercise 17. Let (Ω
1
, F
1
, µ
1
),. . . ,(Ω
n
, F
n
, µ
n
) be n σﬁnite measure
spaces, n ≥ 2. Let f ∈ L
1
C
(Ω
1
×. . . ×Ω
n
, F
1
⊗. . . ⊗F
n
, µ
1
⊗. . . ⊗µ
n
).
Let σ be a permutation of N
n
.
1. For all ω ∈ Π
i=σ(1)
Ω
i
, deﬁne:
J
1
(ω)
=
Ω
σ(1)
f(ω, x)dµ
σ(1)
(x)
Tutorial 7: Fubini Theorem 21
Explain why J
1
is well deﬁned and equal to an element of
L
1
C
(Π
i=σ(1)
Ω
i
, ⊗
i=σ(1)
F
i
, ⊗
i=σ(1)
µ
i
), ⊗
i=σ(1)
µ
i
almost surely.
2. Suppose 1 ≤ k < n −2 and that
¯
J
k
is well deﬁned and equal to
an element of:
L
1
C
(Π
i∈{σ(1),...,σ(k)}
Ω
i
, ⊗
i∈{σ(1),...,σ(k)}
F
i
, ⊗
i∈{σ(1),...,σ(k)}
µ
i
)
⊗
i∈{σ(1),...,σ(k)}
µ
i
almost surely. Deﬁne:
J
k+1
(ω)
=
Ω
σ(k+1)
¯
J
k
(ω, x)dµ
σ(k+1)
(x)
What can you say about J
k+1
.
3. Show that:
Ω
σ(n)
. . .
Ω
σ(1)
fdµ
σ(1)
. . . dµ
σ(n)
is a well deﬁned complex number. (Propose a deﬁnition for it).
Tutorial 7: Fubini Theorem 22
4. Show that:
Ω
σ(n)
. . .
Ω
σ(1)
fdµ
σ(1)
. . . dµ
σ(n)
=
Ω
1
×...×Ω
n
fdµ
1
⊗. . . ⊗µ
n
Tutorial 8: Jensen inequality 1
8. Jensen inequality
Deﬁnition 64 Let a, b ∈
¯
R, with a < b. Let φ : ]a, b[→ R be an
Rvalued function. We say that φ is a convex function, if and only
if, for all x, y ∈]a, b[ and t ∈ [0, 1], we have:
φ(tx + (1 −t)y) ≤ tφ(x) + (1 −t)φ(y)
Exercise 1. Let a, b ∈
¯
R, with a < b. Let φ : ]a, b[→ R be a map.
1. Show that φ : ]a, b[→ R is convex, if and only if for all x
1
, . . . , x
n
in ]a, b[ and α
1
, . . . , α
n
in R
+
with α
1
+. . . +α
n
= 1, n ≥ 1, we
have:
φ(α
1
x
1
+. . . +α
n
x
n
) ≤ α
1
φ(x
1
) +. . . α
n
φ(x
n
)
2. Show that φ : ]a, b[→ R is convex, if and only if for all x, y, z
with a < x < y < z < b we have:
φ(y) ≤
z −y
z −x
φ(x) +
y − x
z −x
φ(z)
Tutorial 8: Jensen inequality 2
3. Show that φ : ]a, b[→ R is convex if and only if for all x, y, z
with a < x < y < z < b, we have:
φ(y) −φ(x)
y −x
≤
φ(z) −φ(y)
z −y
4. Let φ : ]a, b[→R be convex. Let x
0
∈]a, b[, and u, u
, v, v
∈]a, b[
be such that u < u
< x
0
< v < v
. Show that for all x ∈]x
0
, v[:
φ(u
) −φ(u)
u
−u
≤
φ(x) −φ(x
0
)
x −x
0
≤
φ(v
) −φ(v)
v
−v
and deduce that lim
x↓↓x
0
φ(x) = φ(x
0
)
5. Show that if φ : ]a, b[→ R is convex, then φ is continuous.
6. Deﬁne φ : [0, 1] → R by φ(0) = 1 and φ(x) = 0 for all x ∈]0, 1].
Show that φ(tx+(1 −t)y) ≤ tφ(x) +(1−t)φ(y), ∀x, y, t ∈ [0, 1],
but that φ fails to be continuous on [0, 1].
Tutorial 8: Jensen inequality 3
Deﬁnition 65 Let (Ω, T ) be a topological space. We say that (Ω, T )
is a compact topological space if and only if, for all family (V
i
)
i∈I
of open sets in Ω, such that Ω = ∪
i∈I
V
i
, there exists a ﬁnite subset
{i
1
, . . . , i
n
} of I such that Ω = V
i
1
∪ . . . ∪ V
i
n
.
In short, we say that (Ω, T ) is compact if and only if, from any open
covering of Ω, one can extract a ﬁnite subcovering.
Deﬁnition 66 Let (Ω, T ) be a topological space, and K ⊆ Ω. We
say that K is a compact subset of Ω, if and only if the induced
topological space (K, T
K
) is a compact topological space.
Exercise 2. Let (Ω, T ) be a topological space.
1. Show that if (Ω, T ) is compact, it is a compact subset of itself.
2. Show that ∅ is a compact subset of Ω.
3. Show that if Ω
⊆ Ω and K is a compact subset of Ω
, then K
is also a compact subset of Ω.
Tutorial 8: Jensen inequality 4
4. Show that if (V
i
)
i∈I
is a family of open sets in Ω such that
K ⊆ ∪
i∈I
V
i
, then K = ∪
i∈I
(V
i
∩ K) and V
i
∩ K is open in K
for all i ∈ I.
5. Show that K ⊆ Ω is a compact subset of Ω, if and only if for any
family (V
i
)
i∈I
of open sets in Ω such that K ⊆ ∪
i∈I
V
i
, there is
a ﬁnite subset {i
1
, . . . , i
n
} of I such that K ⊆ V
i
1
∪ . . . ∪ V
i
n
.
6. Show that if (Ω, T ) is compact and K is closed in Ω, then K is
a compact subset of Ω.
Exercise 3. Let a, b ∈ R, a < b. Let (V
i
)
i∈I
be a family of open
sets in R such that [a, b] ⊆ ∪
i∈I
V
i
. We deﬁne A as the set of all
x ∈ [a, b] such that [a, x] can be covered by a ﬁnite number of V
i
’s.
Let c = sup A.
1. Show that a ∈ A.
2. Show that there is > 0 such that a + ∈ A.
Tutorial 8: Jensen inequality 5
3. Show that a < c ≤ b.
4. Show the existence of i
0
∈ I and c
, c
with a < c
< c < c
,
such that ]c
, c
] ⊆ V
i
0
.
5. Show that [a, c
] can be covered by a ﬁnite number of V
i
’s.
6. Show that [a, c
] can be covered by a ﬁnite number of V
i
’s.
7. Show that b ∧ c
≤ c and conclude that c = b.
8. Show that [a, b] is a compact subset of R.
Theorem 34 Let a, b ∈ R, a < b. The closed interval [a, b] is a
compact subset of R.
Deﬁnition 67 Let (Ω, T ) be a topological space. We say that (Ω, T )
is a hausdorﬀ topological space, if and only if for all x, y ∈ Ω
with x = y, there exists open sets U and V in Ω, such that:
x ∈ U , y ∈ V , U ∩ V = ∅
Tutorial 8: Jensen inequality 6
Exercise 4. Let (Ω, T ) be a topological space.
1. Show that if (Ω, T ) is hausdorﬀ and Ω
⊆ Ω, then the induced
topological space (Ω
, T
Ω
) is itself hausdorﬀ.
2. Show that if (Ω, T ) is metrizable, then it is hausdorﬀ.
3. Show that any subset of
¯
R is hausdorﬀ.
4. Let (Ω
i
, T
i
)
i∈I
be a family of hausdorﬀ topological spaces. Show
that the product topological space Π
i∈I
Ω
i
is hausdorﬀ.
Exercise 5. Let (Ω, T ) be a hausdorﬀ topological space. Let K be
a compact subset of Ω and suppose there exists y ∈ K
c
.
1. Show that for all x ∈ K, there are open sets V
x
, W
x
in Ω, such
that y ∈ V
x
, x ∈ W
x
and V
x
∩ W
x
= ∅.
2. Show that there exists a ﬁnite subset {x
1
, . . . , x
n
} of K such
that K ⊆ W
y
where W
y
= W
x
1
∪ . . . ∪ W
x
n
.
Tutorial 8: Jensen inequality 7
3. Let V
y
= V
x
1
∩. . .∩V
x
n
. Show that V
y
is open and V
y
∩W
y
= ∅.
4. Show that y ∈ V
y
⊆ K
c
.
5. Show that K
c
= ∪
y∈K
cV
y
6. Show that K is closed in Ω.
Theorem 35 Let (Ω, T ) be a hausdorﬀ topological space. For all
K ⊆ Ω, if K is a compact subset, then it is closed.
Deﬁnition 68 Let (E, d) be a metric space. For all A ⊆ E, we
call diameter of A with respect to d, the element of
¯
R denoted δ(A),
deﬁned as δ(A) = sup{d(x, y) : x, y ∈ A}, with the convention that
δ(∅) = −∞.
Deﬁnition 69 Let (E, d) be a metric space, and A ⊆ E. We say that
A is bounded, if and only if its diameter is ﬁnite, i.e. δ(A) < +∞.
Tutorial 8: Jensen inequality 8
Exercise 6. Let (E, d) be a metric space. Let A ⊆ E.
1. Show that δ(A) = 0 if and only if A = {x} for some x ∈ E.
2. Let φ :
¯
R → [−1, 1] be an increasing homeomorphism. Deﬁne
d
(x, y) = x − y and d
(x, y) = φ(x) − φ(y), for all x, y ∈ R.
Show that d
is a metric on R inducing the usual topology on
R. Show that R is bounded with respect to d
but not with
respect to d
.
3. Show that if K ⊆ E is a compact subset of E, for all > 0,
there is a ﬁnite subset {x
1
, . . . , x
n
} of K such that:
K ⊆ B(x
1
, ) ∪ . . . ∪ B(x
n
, )
4. Show that any compact subset of any metrizable topological
space (Ω, T ), is bounded with respect to any metric inducing
the topology T .
Tutorial 8: Jensen inequality 9
Exercise 7. Suppose K is a closed subset of R which is bounded
with respect to the usual metric on R.
1. Show that there exists M ∈ R
+
such that K ⊆ [−M, M].
2. Show that K is also closed in [−M, M].
3. Show that K is a compact subset of [−M, M].
4. Show that K is a compact subset of R.
5. Show that any compact subset of R is closed and bounded.
6. Show the following:
Theorem 36 A subset of R is compact if and only if it is closed,
and bounded with respect to the usual metric on R.
Tutorial 8: Jensen inequality 10
Exercise 8. Let (Ω, T ) and (S, T
S
) be two topological spaces. Let
f : (Ω, T ) → (S, T
S
) be a continuous map.
1. Show that if (W
i
)
i∈I
is an open covering of f(Ω), then the family
(f
−1
(W
i
))
i∈I
is an open covering of Ω.
2. Show that if (Ω, T ) is a compact topological space, then f(Ω)
is a compact subset of (S, T
S
).
Exercise 9.
1. Show that (
¯
R, T
¯
R
) is a compact topological space.
2. Show that any compact subset of R is a compact subset of
¯
R.
3. Show that a subset of
¯
R is compact if and only if it is closed.
4. Let A be a nonempty subset of
¯
R, and let α = sup A. Show
that if α = −∞, then for all U ∈ T
¯
R
with α ∈ U, there exists
β ∈ R with β < α and ]β, α] ⊆ U. Conclude that α ∈
¯
A.
Tutorial 8: Jensen inequality 11
5. Show that if A is a nonempty closed subset of
¯
R, then we have
supA ∈ A and inf A ∈ A.
6. Consider A = {x ∈ R , sin(x) = 0}. Show that A is closed in
R, but that supA ∈ A and inf A ∈ A.
7. Show that if A is a nonempty, closed and bounded subset of R,
then supA ∈ A and inf A ∈ A.
Exercise 10. Let (Ω, T ) be a compact, nonempty topological space.
Let f : (Ω, T ) → (
¯
R, T
¯
R
) be a continuous map.
1. Show that if f(Ω) ⊆ R, the continuity of f with respect to T
¯
R
is equivalent to the continuity of f with respect to T
R
.
2. Show the following:
Tutorial 8: Jensen inequality 12
Theorem 37 Let f : (Ω, T ) → (
¯
R, T
¯
R
) be a continuous map, where
(Ω, T ) is a nonempty topological space. Then, if (Ω, T ) is compact,
f attains its maximum and minimum, i.e. there exist x
m
, x
M
∈ Ω,
such that:
f(x
m
) = inf
x∈Ω
f(x) , f(x
M
) = sup
x∈Ω
f(x)
Exercise 11. Let a, b ∈ R, a < b. Let f : [a, b] → R be continuous
on [a, b], and diﬀerentiable on ]a, b[, with f(a) = f(b).
1. Show that if c ∈]a, b[ and f(c) = sup
x∈[a,b]
f(x), then f
(c) = 0.
2. Show the following:
Theorem 38 (Rolle) Let a, b ∈ R, a < b. Let f : [a, b] → R be
continuous on [a, b], and diﬀerentiable on ]a, b[, with f(a) = f(b).
Then, there exists c ∈]a, b[ such that f
(c) = 0.
Tutorial 8: Jensen inequality 13
Exercise 12. Let a, b ∈ R, a < b. Let f : [a, b] → R be continuous
on [a, b] and diﬀerentiable on ]a, b[. Deﬁne:
h(x)
= f(x) −(x −a)
f(b) −f(a)
b −a
1. Show that h is continuous on [a, b] and diﬀerentiable on ]a, b[.
2. Show the existence of c ∈]a, b[ such that:
f(b) −f(a) = (b −a)f
(c)
Exercise 13. Let a, b ∈ R, a < b. Let f : [a, b] → R be a map.
Let n ≥ 0. We assume that f is of class C
n
on [a, b], and that f
(n+1)
exists on ]a, b[. Deﬁne:
h(x)
= f(b) −f(x) −
n
k=1
(b −x)
k
k!
f
(k)
(x) −α
(b −x)
n+1
(n + 1)!
where α is chosen such that h(a) = 0.
Tutorial 8: Jensen inequality 14
1. Show that h is continuous on [a, b] and diﬀerentiable on ]a, b[.
2. Show that for all x ∈]a, b[:
h
(x) =
(b −x)
n
n!
(α −f
(n+1)
(x))
3. Prove the following:
Theorem 39 (TaylorLagrange) Let a, b ∈ R, a < b, and n ≥ 0.
Let f : [a, b] → R be a map of class C
n
on [a, b] such that f
(n+1)
exists on ]a, b[. Then, there exists c ∈]a, b[ such that:
f(b) −f(a) =
n
k=1
(b −a)
k
k!
f
(k)
(a) +
(b −a)
n+1
(n + 1)!
f
(n+1)
(c)
Tutorial 8: Jensen inequality 15
Exercise 14. Let a, b ∈
¯
R, a < b and φ : ]a, b[→ R be diﬀerentiable.
1. Show that if φ is convex, then for all x, y ∈]a, b[, x < y, we have:
φ
(x) ≤ φ
(y)
2. Show that if x, y, z ∈]a, b[ with x < y < z, there are c
1
, c
2
∈]a, b[,
with c
1
< c
2
and:
φ(y) −φ(x) = φ
(c
1
)(y −x)
φ(z) −φ(y) = φ
(c
2
)(z −y)
3. Show conversely that if φ
is nondecreasing, then φ is convex.
4. Show that x → e
x
is convex on R.
5. Show that x → −ln(x) is convex on ]0, +∞[.
Tutorial 8: Jensen inequality 16
Deﬁnition 70 we say that a ﬁnite measure space (Ω, F, P) is a
probability space, if and only if P(Ω) = 1.
Deﬁnition 71 Let (Ω, F, P) be a probability space, and (S, Σ) be
a measurable space. We call random variable w.r. to (S, Σ), any
measurable map X : (Ω, F) → (S, Σ).
Deﬁnition 72 Let (Ω, F, P) be a probability space. Let X be a non
negative random variable, or an element of L
1
C
(Ω, F, P). We call
expectation of X, denoted E[X], the integral:
E[X]
=
Ω
XdP
Tutorial 8: Jensen inequality 17
Exercise 15. Let a, b ∈
¯
R, a < b and φ : ]a, b[→R be a convex map.
Let (Ω, F, P) be a probability space and X ∈ L
1
R
(Ω, F, P) be such
that X(Ω) ⊆]a, b[.
1. Show that φ ◦ X : (Ω, F) → (R, B(R)) is measurable.
2. Show that φ◦ X ∈ L
1
R
(Ω, F, P), if and only if E[φ◦ X] < +∞.
3. Show that if E[X] = a, then a ∈ R and X = a Pa.s.
4. Show that if E[X] = b, then b ∈ R and X = b Pa.s.
5. Let m = E[X]. Show that m ∈]a, b[.
6. Deﬁne:
β
= sup
x∈]a,m[
φ(m) −φ(x)
m−x
Show that β ∈ R and that for all z ∈]m, b[, we have:
β ≤
φ(z) −φ(m)
z −m
Tutorial 8: Jensen inequality 18
7. Show that for all x ∈]a, b[, we have φ(m) +β(x −m) ≤ φ(x).
8. Show that for all ω ∈ Ω, φ(m) +β(X(ω) −m) ≤ φ(X(ω)).
9. Show that if φ ◦ X ∈ L
1
R
(Ω, F, P) then φ(m) ≤ E[φ ◦ X].
Theorem 40 (Jensen inequality) Let (Ω, F, P) be a probability
space. Let a, b ∈
¯
R, a < b and φ : ]a, b[→ R be a convex map.
Suppose that X ∈ L
1
R
(Ω, F, P) is such that X(Ω) ⊆]a, b[ and such
that φ ◦ X ∈ L
1
R
(Ω, F, P). Then:
φ(E[X]) ≤ E[φ ◦ X]
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 1
9. L
p
spaces, p ∈ [1, +∞]
In the following, (Ω, F, µ) is a measure space.
Exercise 1. Let f, g : (Ω, F) → [0, +∞] be nonnegative and mea
surable maps. Let p, q ∈ R
+
, such that 1/p + 1/q = 1.
1. Show that 1 < p < +∞ and 1 < q < +∞.
2. For all α ∈]0, +∞[, we deﬁne φ
α
: [0, +∞] → [0, +∞] by:
φ
α
(x)
=
_
x
α
if x ∈ R
+
+∞ if x = +∞
Show that φ
α
is a continuous map.
3. Deﬁne A = (
_
f
p
dµ)
1/p
, B = (
_
g
q
dµ)
1/q
and C =
_
fgdµ.
Explain why A, B and C are well deﬁned elements of [0, +∞].
4. Show that if A = 0 or B = 0 then C ≤ AB.
5. Show that if A = +∞ or B = +∞ then C ≤ AB.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 2
6. We assume from now on that 0 < A < +∞ and 0 < B < +∞.
Deﬁne F = f/A and G = g/B. Show that:
_
Ω
F
p
dµ =
_
Ω
G
p
dµ = 1
7. Let a, b ∈]0, +∞[. Show that:
ln(a) + ln(b) ≤ ln
_
1
p
a
p
+
1
q
b
q
_
and:
ab ≤
1
p
a
p
+
1
q
b
q
Prove this last inequality for all a, b ∈ [0, +∞].
8. Show that for all ω ∈ Ω, we have:
F(ω)G(ω) ≤
1
p
(F(ω))
p
+
1
q
(G(ω))
q
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 3
9. Show that C ≤ AB.
Theorem 41 (H¨older’s inequality) Let (Ω, F, µ) be a measure
space and f, g : (Ω, F) → [0, +∞] be two nonnegative and measurable
maps. Let p, q ∈ R
+
be such that 1/p + 1/q = 1. Then:
_
Ω
fgdµ ≤
__
Ω
f
p
dµ
_
1
p
__
Ω
g
q
dµ
_
1
q
Theorem 42 (CauchySchwarz’s inequality:ﬁrst)
Let (Ω, F, µ) be a measure space and f, g : (Ω, F) → [0, +∞] be two
nonnegative and measurable maps. Then:
_
Ω
fgdµ ≤
__
Ω
f
2
dµ
_
1
2
__
Ω
g
2
dµ
_
1
2
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 4
Exercise 2. Let f, g : (Ω, F) → [0, +∞] be two nonnegative and
measurable maps. Let p ∈]1, +∞[. Deﬁne A = (
_
f
p
dµ)
1/p
and
B = (
_
g
p
dµ)
1/p
and C = (
_
(f +g)
p
dµ)
1/p
.
1. Explain why A, B and C are well deﬁned elements of [0, +∞].
2. Show that for all a, b ∈]0, +∞[, we have:
(a +b)
p
≤ 2
p−1
(a
p
+b
p
)
Prove this inequality for all a, b ∈ [0, +∞].
3. Show that if A = +∞ or B = +∞ or C = 0 then C ≤ A+B.
4. Show that if A < +∞ and B < +∞ then C < +∞.
5. We assume from now that A, B ∈ [0, +∞[ and C ∈]0, +∞[.
Show the existence of some q ∈ R
+
such that 1/p + 1/q = 1.
6. Show that for all a, b ∈ [0, +∞], we have:
(a +b)
p
= (a +b).(a +b)
p−1
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 5
7. Show that:
_
Ω
f.(f +g)
p−1
dµ ≤ AC
p
q
_
Ω
g.(f +g)
p−1
dµ ≤ BC
p
q
8. Show that:
_
Ω
(f +g)
p
dµ ≤ C
p
q
(A +B)
9. Show that C ≤ A +B.
10. Show that the inequality still holds if we assume that p = 1.
Theorem 43 (Minkowski’s inequality) Let (Ω, F, µ) be a mea
sure space and f, g : (Ω, F) → [0, +∞] be two nonnegative and mea
surable maps. Let p ∈ [1, +∞[. Then:
__
Ω
(f +g)
p
dµ
_
1
p
≤
__
Ω
f
p
dµ
_
1
p
+
__
Ω
g
p
dµ
_
1
p
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 6
Deﬁnition 73 The L
p
spaces, p ∈ [1, +∞[, on (Ω, F, µ), are:
L
p
R
(Ω, F, µ)
=
_
f : (Ω, F)→(R, B(R)) measurable,
_
Ω
f
p
dµ <+∞
_
L
p
C
(Ω, F, µ)
=
_
f : (Ω, F)→(C, B(C)) measurable,
_
Ω
f
p
dµ <+∞
_
For all f ∈ L
p
C
(Ω, F, µ), we put:
f
p
=
__
Ω
f
p
dµ
_
1
p
Exercise 3. Let p ∈ [1, +∞[. Let f, g ∈ L
p
C
(Ω, F, µ).
1. Show that L
p
R
(Ω, F, µ) = {f ∈ L
p
C
(Ω, F, µ) , f(Ω) ⊆ R}.
2. Show that L
p
R
(Ω, F, µ) is closed under Rlinear combinations.
3. Show that L
p
C
(Ω, F, µ) is closed under Clinear combinations.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 7
4. Show that f +g
p
≤ f
p
+g
p
.
5. Show that f
p
= 0 ⇔ f = 0 µa.s.
6. Show that for all α ∈ C, αf
p
= α.f
p
.
7. Explain why (f, g) → f −g
p
is not a metric on L
p
C
(Ω, F, µ)
Deﬁnition 74 For all f : (Ω, F) → (C, B(C)) measurable, Let:
f
∞
= inf{M ∈ R
+
, f ≤ M µa.s.}
The L
∞
spaces on a measure space (Ω, F, µ) are:
L
∞
R
(Ω, F, µ)
= {f : (Ω, F) → (R, B(R)) measurable, f
∞
< +∞}
L
∞
C
(Ω, F, µ)
= {f : (Ω, F) → (C, B(C)) measurable, f
∞
< +∞}
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 8
Exercise 4. Let f, g ∈ L
∞
C
(Ω, F, µ).
1. Show that L
∞
R
(Ω, F, µ) = {f ∈ L
∞
C
(Ω, F, µ) , f(Ω) ⊆ R}.
2. Show that f ≤ f
∞
µa.s.
3. Show that f +g
∞
≤ f
∞
+g
∞
.
4. Show that L
∞
R
(Ω, F, µ) is closed under Rlinear combinations.
5. Show that L
∞
C
(Ω, F, µ) is closed under Clinear combinations.
6. Show that f
∞
= 0 ⇔ f = 0 µa.s..
7. Show that for all α ∈ C, αf
∞
= α.f
∞
.
8. Explain why (f, g) → f −g
∞
is not a metric on L
∞
C
(Ω, F, µ)
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 9
Deﬁnition 75 Let p ∈ [1, +∞]. Let K = R or C. For all > 0 and
f ∈ L
p
K
(Ω, F, µ), we deﬁne the socalled open ball in L
p
K
(Ω, F, µ):
B(f, )
= {g : g ∈ L
p
K
(Ω, F, µ), f −g
p
< }
We call usual topology in L
p
K
(Ω, F, µ), the set T deﬁned by:
T
= {U : U ⊆ L
p
K
(Ω, F, µ), ∀f ∈ U, ∃ > 0, B(f, ) ⊆ U}
Note that if (f, g) → f − g
p
was a metric, the usual topology in
L
p
K
(Ω, F, µ), would be nothing but the metric topology.
Exercise 5. Let p ∈ [1, +∞]. Suppose there exists N ∈ F with
µ(N) = 0 and N = ∅. Put f = 1
N
and g = 0
1. Show that f, g ∈ L
p
C
(Ω, F, µ) and f = g.
2. Show that any open set containing f also contains g.
3. Show that L
p
C
(Ω, F, µ) and L
p
R
(Ω, F, µ) are not Hausdorﬀ.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 10
Exercise 6. Show that the usual topology on L
p
R
(Ω, F, µ) is induced
by the usual topology on L
p
C
(Ω, F, µ), where p ∈ [1, +∞].
Deﬁnition 76 Let (E, T ) be a topological space. A sequence (x
n
)
n≥1
in E is said to converge to x ∈ E, and we write x
n
T
→ x, if and only
if, for all U ∈ T such that x ∈ U, there exists n
0
≥ 1 such that:
n ≥ n
0
⇒ x
n
∈ U
When E = L
p
C
(Ω, F, µ) or E = L
p
R
(Ω, F, µ), we write x
n
L
p
→ x.
Exercise 7. Let (E, T ) be a topological space and E
⊆ E. Let
T
= T
E
be the induced topology on E
. Show that if (x
n
)
n≥1
is a
sequence in E
and x ∈ E
, then x
n
T
→ x is equivalent to x
n
T
→ x.
Exercise 8. Let f, g, (f
n
)
n≥1
be in L
p
C
(Ω, F, µ) and p ∈ [1, +∞].
1. Recall what the notation f
n
→ f means.
2. Show that f
n
L
p
→ f is equivalent to f
n
−f
p
→ 0.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 11
3. Show that if f
n
L
p
→ f and f
n
L
p
→ g then f = g µa.s.
Exercise 9. Let p ∈ [1, +∞]. Suppose there exists some N ∈ F such
that µ(N) = 0 and N = ∅. Find a sequence (f
n
)
n≥1
in L
p
C
(Ω, F, µ)
and f, g in L
p
C
(Ω, F, µ), f = g such that f
n
L
p
→ f and f
n
L
p
→ g.
Deﬁnition 77 Let (f
n
)
n≥1
be a sequence in L
p
C
(Ω, F, µ), where
(Ω, F, µ) is a measure space and p ∈ [1, +∞]. We say that (f
n
)
n≥1
is
a cauchy sequence, if and only if, for all > 0, there exists n
0
≥ 1
such that:
n, m ≥ n
0
⇒ f
n
−f
m
p
≤
Exercise 10. Let f, (f
n
)
n≥1
be in L
p
C
(Ω, F, µ) and p ∈ [1, +∞].
Show that if f
n
L
p
→ f, then (f
n
)
n≥1
is a cauchy sequence.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 12
Exercise 11. Let p ∈ [1, +∞], and (f
n
)
n≥1
be cauchy in L
p
C
(Ω, F, µ).
1. Show the existence of n
1
≥ 1 such that:
n ≥ n
1
⇒ f
n
−f
n
1
p
≤
1
2
2. Suppose we have found n
1
< n
2
< . . . < n
k
, k ≥ 1, such that:
∀j ∈ {1, . . . , k} , n ≥ n
j
⇒ f
n
−f
n
j
p
≤
1
2
j
Show the existence of n
k+1
, n
k
< n
k+1
such that:
n ≥ n
k+1
⇒ f
n
−f
n
k+1
p
≤
1
2
k+1
3. Show that there exists a subsequence (f
n
k
)
k≥1
of (f
n
)
n≥1
with:
+∞
k=1
f
n
k+1
−f
n
k
p
< +∞
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 13
Exercise 12. Let p ∈ [1, +∞], and (f
n
)
n≥1
be in L
p
C
(Ω, F, µ), with:
+∞
n=1
f
n+1
−f
n
p
< +∞
We deﬁne:
g
=
+∞
n=1
f
n+1
−f
n

1. Show that g : (Ω, F) → [0, +∞] is nonnegative and measurable.
2. If p = +∞, show that g ≤
+∞
n=1
f
n+1
−f
n
∞
µa.s.
3. If p ∈ [1, +∞[, show that for all N ≥ 1, we have:
_
_
_
_
_
N
n=1
f
n+1
− f
n

_
_
_
_
_
p
≤
+∞
n=1
f
n+1
−f
n
p
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 14
4. If p ∈ [1, +∞[, show that:
_
_
Ω
g
p
dµ
_
1
p
≤
+∞
n=1
f
n+1
−f
n
p
5. Show that for p ∈ [1, +∞], we have g < +∞ µa.s.
6. Deﬁne A = {g < +∞}. Show that for all ω ∈ A, (f
n
(ω))
n≥1
is
a cauchy sequence in C. We denote z(ω) its limit.
7. Deﬁne f : (Ω, F) → (C, B(C)), by:
f(ω)
=
_
z(ω) if ω ∈ A
0 if ω ∈ A
Show that f is measurable and f
n
→ f µa.s.
8. if p = +∞, show that for all n ≥ 1, f
n
 ≤ f
1
 +g and conclude
that f ∈ L
∞
C
(Ω, F, µ).
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 15
9. If p ∈ [1, +∞[, show the existence of n
0
≥ 1, such that:
n ≥ n
0
⇒
_
Ω
f
n
−f
n
0

p
dµ ≤ 1
Deduce from Fatou’s lemma that f −f
n
0
∈ L
p
C
(Ω, F, µ).
10. Show that for p ∈ [1, +∞], f ∈ L
p
C
(Ω, F, µ).
11. Suppose that f
n
∈ L
p
R
(Ω, F, µ), for all n ≥ 1. Show the exis
tence of f ∈ L
p
R
(Ω, F, µ), such that f
n
→ f µa.s.
Exercise 13. Let p ∈ [1, +∞], and (f
n
)
n≥1
be in L
p
C
(Ω, F, µ), with:
+∞
n=1
f
n+1
−f
n
p
< +∞
1. Does there exist f ∈ L
p
C
(Ω, F, µ) such that f
n
→ f µa.s.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 16
2. Suppose p = +∞. Show that for all n < m, we have:
f
m+1
−f
n
 ≤
m
k=n
f
k+1
− f
k
∞
µa.s.
3. Suppose p = +∞. Show that for all n ≥ 1, we have:
f −f
n
∞
≤
+∞
k=n
f
k+1
−f
k
∞
4. Suppose p ∈ [1, +∞[. Show that for all n < m, we have:
__
Ω
f
m+1
−f
n

p
dµ
_
1
p
≤
m
k=n
f
k+1
−f
k
p
5. Suppose p ∈ [1, +∞[. Show that for all n ≥ 1, we have:
f −f
n
p
≤
+∞
k=n
f
k+1
−f
k
p
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 17
6. Show that for p ∈ [1, +∞], we also have f
n
L
p
→ f.
7. Suppose conversely that g ∈ L
p
C
(Ω, F, µ) is such that f
n
L
p
→ g.
Show that f = g µa.s.. Conclude that f
n
→ g µa.s..
Theorem 44 Let (Ω, F, µ) be a measure space. Let p ∈ [1, +∞],
and (f
n
)
n≥1
be a sequence in L
p
C
(Ω, F, µ) such that:
+∞
n=1
f
n+1
−f
n
p
< +∞
Then, there exists f ∈ L
p
C
(Ω, F, µ) such that f
n
→ f µa.s. Moreover,
for all g ∈ L
p
C
(Ω, F, µ), the convergence f
n
→ g µa.s. and f
n
L
p
→ g
are equivalent.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 18
Exercise 14. Let f, (f
n
)
n≥1
be in L
p
C
(Ω, F, µ) such that f
n
L
p
→ f,
where p ∈ [1, +∞].
1. Show that there exists a subsequence (f
n
k
)
k≥1
of (f
n
)
n≥1
, with:
+∞
k=1
f
n
k+1
−f
n
k
p
< +∞
2. Show that there exists g ∈ L
p
C
(Ω, F, µ) such that f
n
k
→ g µa.s.
3. Show that f
n
k
L
p
→ g and g = f µa.s.
4. Conclude with the following:
Theorem 45 Let (f
n
)
n≥1
be in L
p
C
(Ω, F, µ) and f ∈ L
p
C
(Ω, F, µ)
such that f
n
L
p
→ f, where p ∈ [1, +∞]. Then, we can extract a sub
sequence (f
n
k
)
k≥1
of (f
n
)
n≥1
such that f
n
k
→ f µa.s.
Tutorial 9: L
p
spaces, p ∈ [1, +∞] 19
Exercise 15. Prove the last theorem for L
p
R
(Ω, F, µ).
Exercise 16. Let p ∈ [1, +∞], and (f
n
)
n≥1
be cauchy in L
p
C
(Ω, F, µ).
1. Show that there exists a subsequence (f
n
k
)
k≥1
of (f
n
)
n≥1
and
f belonging to L
p
C
(Ω, F, µ), such that f
n
k
L
p
→ f.
2. Using the fact that (f
n
)
n≥1
is cauchy, show that f
n
L
p
→ f.
Theorem 46 Let p ∈ [1, +∞]. Let (f
n
)
n≥1
be a cauchy sequence in
L
p
C
(Ω, F, µ). Then, there exists f ∈ L
p
C
(Ω, F, µ) such that f
n
L
p
→ f.
Exercise 17. Prove the last theorem for L
p
R
(Ω, F, µ).
Tutorial 10: Bounded Linear Functionals in L
2
1
10. Bounded Linear Functionals in L
2
In the following, (Ω, T, µ) is a measure space.
Deﬁnition 78 Let (x
n
)
n≥1
be a sequence in an arbitrary set. We
call subsequence of (x
n
)
n≥1
, any sequence of the form (x
φ(n)
)
n≥1
,
where φ : N
∗
→ N
∗
is a strictly increasing map.
Exercise 1. Let (E, d) be a metric space, with metric topology T .
Let (x
n
)
n≥1
be a sequence in E. For all n ≥ 1, let F
n
be the closure
of the set ¦x
k
: k ≥ n¦.
1. Show that for all x ∈ E, x
n
J
→ x is equivalent to:
∀ > 0 , ∃n
0
≥ 1 , n ≥ n
0
⇒ d(x
n
, x) ≤
2. Show that (F
n
)
n≥1
is a decreasing sequence of closed sets in E.
3. Show that if F
n
↓ ∅, then (F
c
n
)
n≥1
is an open covering of E.
Tutorial 10: Bounded Linear Functionals in L
2
2
4. Show that if (E, T ) is compact then ∩
+∞
n=1
F
n
,= ∅.
5. Show that if (E, T ) is compact, there exists x ∈ E such that for
all n ≥ 1 and > 0, we have B(x, ) ∩ ¦x
k
, k ≥ n¦ ,= ∅.
6. By induction, construct a subsequence (x
n
p
)
p≥1
of (x
n
)
n≥1
such
that x
n
p
∈ B(x, 1/p) for all p ≥ 1.
7. Conclude that if (E, T ) is compact, any sequence (x
n
)
n≥1
in E
has a convergent subsequence.
Exercise 2. Let (E, d) be a metric space, with metric topology T .
We assume that any sequence (x
n
)
n≥1
in E has a convergent subse
quence. Let (V
i
)
i∈I
be an open covering of E. For x ∈ E, let:
r(x)
Z
= sup¦r > 0 : B(x, r) ⊆ V
i
, for some i ∈ I¦
1. Show that ∀x ∈ E, ∃i ∈ I, ∃r > 0, such that B(x, r) ⊆ V
i
.
2. Show that ∀x ∈ E, r(x) > 0.
Tutorial 10: Bounded Linear Functionals in L
2
3
Exercise 3. Further to ex. (2), suppose inf
x∈E
r(x) = 0.
1. Show that for all n ≥ 1, there is x
n
∈ E such that r(x
n
) < 1/n.
2. Extract a subsequence (x
n
k
)
k≥1
of (x
n
)
n≥1
converging to some
x
∗
∈ E. Let r
∗
> 0 and i ∈ I be such that B(x
∗
, r
∗
) ⊆ V
i
. Show
that we can ﬁnd some k
0
≥ 1, such that d(x
∗
, x
n
k
0
) < r
∗
/2 and
r(x
n
k
0
) ≤ r
∗
/4.
3. Show that d(x
∗
, x
n
k
0
) < r
∗
/2 implies that B(x
n
k
0
, r
∗
/2) ⊆ V
i
.
Show that this contradicts r(x
n
k
0
) ≤ r
∗
/4, and conclude that
inf
x∈E
r(x) > 0.
Exercise 4. Further to ex. (3), Let r
0
with 0 < r
0
< inf
x∈E
r(x).
Suppose that E cannot be covered by a ﬁnite number of open balls
with radius r
0
.
1. Show the existence of a sequence (x
n
)
n≥1
in E, such that for all
n ≥ 1, x
n+1
,∈ B(x
1
, r
0
) ∪ . . . ∪ B(x
n
, r
0
).
Tutorial 10: Bounded Linear Functionals in L
2
4
2. Show that for all n > m we have d(x
n
, x
m
) ≥ r
0
.
3. Show that (x
n
)
n≥1
cannot have a convergent subsequence.
4. Conclude that there exists a ﬁnite subset ¦x
1
, . . . , x
n
¦ of E such
that E = B(x
1
, r
0
) ∪ . . . ∪ B(x
n
, r
0
).
5. Show that for all x ∈ E, we have B(x, r
0
) ⊆ V
i
for some i ∈ I.
6. Conclude that (E, T ) is compact.
7. Prove the following:
Theorem 47 Let (E, T ) be a metrizable topological space. Then
(E, T ) is compact, if and only if for every sequence (x
n
)
n≥1
in E,
there exists a subsequence (x
n
k
)
k≥1
of (x
n
)
n≥1
, and some x ∈ E,
such that x
n
k
J
→ x.
Tutorial 10: Bounded Linear Functionals in L
2
5
Exercise 5. Let a, b ∈ R , a < b and (x
n
)
n≥1
be a sequence in ]a, b[.
1. Show that (x
n
)
n≥1
has a convergent subsequence.
2. Can we conclude that ]a, b[ is a compact subset of R?
Exercise 6. Let E = [−M, M] . . . [−M, M] ⊆ R
n
, where n ≥ 1
and M ∈ R
+
. Let T
R
n be the usual product topology on R
n
, and
T
E
= (T
R
n)
]E
be the induced topology on E.
1. Let (x
p
)
p≥1
be a sequence in E. Let x ∈ E. Show that x
p
J
E
→ x
is equivalent to x
p
J
R
n
→ x.
2. Propose a metric on R
n
, inducing the topology T
R
n.
3. Let (x
p
)
p≥1
be a sequence in R
n
. Let x ∈ R
n
. Show that
x
p
J
R
n
→ x if and only if, x
i
p
J
R
→ x
i
for all i ∈ N
n
.
Tutorial 10: Bounded Linear Functionals in L
2
6
Exercise 7. Further to ex. (6), suppose (x
p
)
p≥1
is a sequence in E.
1. Show the existence of a subsequence (x
φ(p)
)
p≥1
of (x
p
)
p≥1
, such
that x
1
φ(p)
J
[−M,M]
→ x
1
for some x
1
∈ [−M, M].
2. Explain why the above convergence is equivalent to x
1
φ(p)
J
R
→ x
1
.
3. Suppose that 1 ≤ k ≤ n − 1 and (y
p
)
p≥1
= (x
φ(p)
)
p≥1
is a
subsequence of (x
p
)
p≥1
such that:
∀j = 1, . . . , k , x
j
φ(p)
J
R
→ x
j
for some x
j
∈ [−M, M]
Show the existence of a subsequence (y
ψ(p)
)
p≥1
of (y
p
)
p≥1
such
that y
k+1
ψ(p)
J
R
→ x
k+1
for some x
k+1
∈ [−M, M].
4. Show that φ ◦ ψ : N
∗
→ N
∗
is strictly increasing.
5. Show that (x
φ◦ψ(p)
)
p≥1
is a subsequence of (x
p
)
p≥1
such that:
∀j = 1, . . . , k + 1 , x
j
φ◦ψ(p)
J
R
→ x
j
∈ [−M, M]
Tutorial 10: Bounded Linear Functionals in L
2
7
6. Show the existence of a subsequence (x
φ(p)
)
p≥1
of (x
p
)
p≥1
, and
x ∈ E, such that x
φ(p)
J
E
→ x
7. Show that (E, T
E
) is a compact topological space.
Exercise 8. Let A be a closed subset of R
n
, n ≥ 1, which is bounded
with respect to the usual metric of R
n
.
1. Show that A ⊆ E = [−M, M]. . .[−M, M], for some M ∈ R
+
.
2. Show from E ¸ A = E ∩ A
c
that A is closed in E.
3. Show (A, (T
R
n)
]A
) is a compact topological space.
4. Conversely, let A is a compact subset of R
n
. Show that A is
closed and bounded.
Theorem 48 A subset of R
n
, n ≥ 1, is compact if and only if it is
closed and bounded (with respect to the usual metric).
Tutorial 10: Bounded Linear Functionals in L
2
8
Exercise 9. Let n ≥ 1. Consider the map:
φ :
_
C
n
→ R
2n
(a
1
+ ib
1
, . . . , a
n
+ ib
n
) → (a
1
, b
1
, . . . , a
n
, b
n
)
1. Recall the expressions of the usual metrics d
C
n
and d
R
2n of C
n
and R
2n
respectively.
2. Show that for all z, z
t
∈ C
n
, d
C
n(z, z
t
) = d
R
2n(φ(z), φ(z
t
)).
3. Show that φ is a homeomorphism from C
n
to R
2n
.
4. Show that a subset K of C
n
is compact, if and only if φ(K) is
a compact subset of R
2n
.
5. Show that K is closed, if and only if φ(K) is closed.
6. Show that K is bounded, if and only if φ(K) is bounded.
7. Show that a subset K of C
n
is compact, if and only if it is closed
and bounded (with respect to the usual metric).
Tutorial 10: Bounded Linear Functionals in L
2
9
Deﬁnition 79 Let (E, d) be a metric space. A sequence (x
n
)
n≥1
in
E, is said to be a cauchy sequence (relative to the metric d), if and
only if, for all > 0, there exists n
0
≥ 1 such that:
n, m ≥ n
0
⇒ d(x
n
, x
m
) ≤
Deﬁnition 80 We say that a metric space (E, d) is complete, if
and only if, for all (x
n
)
n≥1
cauchy sequence in E, there exists x ∈ E
such that (x
n
)
n≥1
converges to x.
Exercise 10.
1. Explain why strictly speaking, given p ∈ [1, +∞], deﬁnition (77)
of Cauchy sequences in L
p
C
(Ω, T, µ) is not a covered by deﬁni
tion (79).
2. Explain why L
p
C
(Ω, T, µ) is not a complete metric space, despite
theorem (46) and deﬁnition (80).
Tutorial 10: Bounded Linear Functionals in L
2
10
Exercise 11. Let (z
k
)
k≥1
be a Cauchy sequence in C
n
, n ≥ 1, with
respect to the usual metric d(z, z
t
) = z −z
t
, where:
z
Z
=
¸
¸
¸
_
n
i=1
[z
i
[
2
1. Show that the sequence (z
k
)
k≥1
is bounded, i.e. that there exists
M ∈ R
+
such that z
k
 ≤ M, for all k ≥ 1.
2. Deﬁne B = ¦z ∈ C
n
, z ≤ M¦. Show that δ(B) < +∞, and
that B is closed in C
n
.
3. Show the existence of a subsequence (z
k
p
)
p≥1
of (z
k
)
k≥1
such
that z
k
p
J
C
n
→ z for some z ∈ B.
4. Show that for all > 0, there exists p
0
≥ 1 and n
0
≥ 1 such
that d(z, z
k
p
0
) ≤ /2 and:
k ≥ n
0
⇒ d(z
k
, z
k
p
0
) ≤ /2
Tutorial 10: Bounded Linear Functionals in L
2
11
5. Show that z
k
J
C
n
→ z.
6. Conclude that C
n
is complete with respect to its usual metric.
7. For which theorem of Tutorial 9 was the completeness of Cused?
Exercise 12. Let (x
k
)
k≥1
be a sequence in R
n
such that x
k
J
C
n
→ z,
for some z ∈ C
n
.
1. Show that z ∈ R
n
.
2. Show that R
n
is complete with respect to its usual metric.
Theorem 49 For all n ≥ 1, C
n
and R
n
are complete with respect
to their usual metrics.
Tutorial 10: Bounded Linear Functionals in L
2
12
Exercise 13. Let (E, d) be a metric space, with metric topology T .
Let F ⊆ E, and
¯
F denote the closure of F.
1. Explain why, for all x ∈
¯
F and n ≥ 1, we have F∩B(x, 1/n) ,= ∅.
2. Show that for all x ∈
¯
F, there exists a sequence (x
n
)
n≥1
in F,
such that x
n
J
→ x.
3. Show conversely that if there is a sequence (x
n
)
n≥1
in F with
x
n
J
→ x, then x ∈
¯
F.
4. Show that F is closed if and only if for all sequence (x
n
)
n≥1
in
F such that x
n
J
→ x for some x ∈ E, we have x ∈ F.
5. Explain why (F, T
]F
) is metrizable.
6. Show that if F is complete with respect to the metric d
]F·F
,
then F is closed in E.
Tutorial 10: Bounded Linear Functionals in L
2
13
7. Let d
¯
R
be a metric on
¯
R, inducing the usual topology T
¯
R
. Show
that d
t
= (d
¯
R
)
]R·R
is a metric on R, inducing the topology T
R
.
8. Find a metric on [−1, 1] which induces its usual topology.
9. Show that ¦−1, 1¦ is not open in [−1, 1].
10. Show that ¦−∞, +∞¦ is not open in
¯
R.
11. Show that R is not closed in
¯
R.
12. Let d
R
be the usual metric of R. Show that d
t
= (d
¯
R
)
]R·R
and d
R
induce the same topology on R, but that however, R
is complete with respect to d
R
, whereas it cannot be complete
with respect to d
t
.
Tutorial 10: Bounded Linear Functionals in L
2
14
Deﬁnition 81 Let H be a Kvector space, where K = R or C. We
call innerproduct on H, any map ¸, ) : H H → K with the
following properties:
(i) ∀x, y ∈ H , ¸x, y) = ¸y, x)
(ii) ∀x, y, z ∈ H , ¸x + z, y) = ¸x, y) +¸z, y)
(iii) ∀x, y ∈ H, ∀α ∈ K , ¸αx, y) = α¸x, y)
(iv) ∀x ∈ H , ¸x, x) ≥ 0
(v) ∀x ∈ H , (¸x, x) = 0 ⇔ x = 0)
where for all z ∈ C, ¯ z denotes the complex conjugate of z. For all
x ∈ H, we call norm of x, denoted x, the number deﬁned by:
x
Z
=
_
¸x, x)
Exercise 14. Let ¸, ) be an innerproduct on a Kvector space H.
1. Show that for all y ∈ H, the map x → ¸x, y) is linear.
Tutorial 10: Bounded Linear Functionals in L
2
15
2. Show that for all x ∈ H, the map y → ¸x, y) is linear if K = R,
and conjugatelinear if K = C.
Exercise 15. Let ¸, ) be an innerproduct on a Kvector space H.
Let x, y ∈ H. Let A = x
2
, B = [¸x, y)[ and C = y
2
. let α ∈ K
be such that [α[ = 1 and:
B = α¸x, y)
1. Show that A, B, C ∈ R
+
.
2. For all t ∈ R, show that ¸x −tαy, x −tαy) = A −2tB + t
2
C.
3. Show that if C = 0 then B
2
≤ AC.
4. Suppose that C ,= 0. Show that P(t) = A − 2tB + t
2
C has a
minimal value which is in R
+
, and conclude that B
2
≤ AC.
5. Conclude with the following:
Tutorial 10: Bounded Linear Functionals in L
2
16
Theorem 50 (CauchySchwarz’s inequality:second) Let H be
a Kvector space, where K = R or C, and ¸, ) be an innerproduct
on H. Then, for all x, y ∈ H, we have:
[¸x, y)[ ≤ x.y
Exercise 16. For all f, g ∈ L
2
C
(Ω, T, µ), we deﬁne:
¸f, g)
Z
=
_
Ω
f ¯ gdµ
1. Use the ﬁrst cauchyschwarz inequality (42) to prove that for all
f, g ∈ L
2
C
(Ω, T, µ), we have f ¯ g ∈ L
1
C
(Ω, T, µ). Conclude that
¸f, g) is a welldeﬁned complex number.
2. Show that for all f ∈ L
2
C
(Ω, T, µ), we have f
2
=
_
¸f, f).
3. Make another use of the ﬁrst cauchyschwarz inequality to show
that for all f, g ∈ L
2
C
(Ω, T, µ), we have:
[¸f, g)[ ≤ f
2
.g
2
(1)
Tutorial 10: Bounded Linear Functionals in L
2
17
4. Go through deﬁnition (81), and indicate which of the proper
ties (i) − (v) fails to be satisﬁed by ¸, ). Conclude that ¸, )
is not an innerproduct on L
2
C
(Ω, T, µ), and therefore that in
equality (*) is not a particular case of the second cauchyschwarz
inequality (50).
5. Let f, g ∈ L
2
C
(Ω, T, µ). By considering
_
([f[+t[g[)
2
dµ for t ∈ R,
imitate the proof of the second cauchyschwarz inequality to
show that:
_
Ω
[fg[dµ ≤
__
Ω
[f[
2
dµ
_
1
2
__
Ω
[g[
2
dµ
_
1
2
6. Let f, g : (Ω, T) → [0, +∞] nonnegative and measurable. Show
that if
_
f
2
dµ and
_
g
2
dµ are ﬁnite, then f and g are µalmost
surely equal to elements of L
2
C
(Ω, T, µ). Deduce from 5. a new
proof of the ﬁrst CauchySchwarz inequality:
_
Ω
fgdµ ≤
__
Ω
f
2
dµ
_
1
2
__
Ω
g
2
dµ
_
1
2
Tutorial 10: Bounded Linear Functionals in L
2
18
Exercise 17. Let ¸, ) be an inner product on a Kvector space H.
1. Show that for all x, y ∈ H, we have:
x + y
2
= x
2
+y
2
+¸x, y) +¸x, y)
2. Using the second cauchyschwarz inequality (50), show that:
x + y ≤ x +y
3. Show that d
¸,)
(x, y) = x − y deﬁnes a metric on H.
Deﬁnition 82 Let H be a Kvector space, where K = R or C,
and ¸, ) be an innerproduct on H. We call norm topology on H,
denoted T
¸,)
, the metric topology associated with d
¸,)
(x, y) = x−y.
Deﬁnition 83 We call hilbert space (over K), where K = R
or C, any ordered pair (H, ¸, )), where H is a Kvector space, and
¸, ) is an inner product on H for which the metric space (H, d
¸,)
) is
complete, where d
¸,)
(x, y) = x −y.
Tutorial 10: Bounded Linear Functionals in L
2
19
Exercise 18. Let (H, ¸, )) be a hilbert space over K and let /
be a closed linear subspace of H, (closed with respect to the norm
topology T
¸,)
). Deﬁne [, ] = ¸, )
],·,
.
1. Show that [, ] is an innerproduct on the Kvector space /.
2. With obvious notations, show that d
[,]
= (d
¸,)
)
],·,
.
3. Deduce that T
[,]
= (T
¸,)
)
],
.
Exercise 19. Further to ex. (18), Let (x
n
)
n≥1
be a cauchy sequence
in /, with respect to the metric d
[,]
.
1. Show that (x
n
)
n≥1
is a cauchy sequence in H.
2. Explain why there exists x ∈ H such that x
n
J
·,·
→ x.
3. Explain why x ∈ /.
4. Explain why we also have x
n
J
[·,·]
→ x.
Tutorial 10: Bounded Linear Functionals in L
2
20
5. Explain why (/, ¸, )
],·,
) is a hilbert space over K.
Exercise 20. For all z, z
t
∈ C
n
, n ≥ 1, we deﬁne:
¸z, z
t
)
Z
=
n
i=1
z
i
¯ z
i
t
1. Show that ¸, ) is an innerproduct on C
n
.
2. Show that the metric d
¸,)
is equal to the usual metric of C
n
.
3. Conclude that (C
n
, ¸, )) is a hilbert space over C.
4. Show that R
n
is a closed subset of C
n
.
5. Show however that R
n
is not a linear subspace of C
n
.
6. Show that (R
n
, ¸, )
]R
n
·R
n) is a hilbert space over R.
Tutorial 10: Bounded Linear Functionals in L
2
21
Deﬁnition 84 Let K = R or C. The usual innerproduct in
K
n
, denoted ¸, ), is deﬁned as:
∀x, y ∈ K
n
, ¸x, y) =
n
i=1
x
i
¯ y
i
Theorem 51 The spaces C
n
and R
n
, n ≥ 1, together with their
usual innerproducts, are hilbert spaces over C and R respectively.
Deﬁnition 85 Let H be a Kvector space, where K = R or C. Let
( ⊆ H. We say that ( is a convex subset or H, if and only if, for
all x, y ∈ ( and t ∈ [0, 1], we have tx + (1 −t)y ∈ (.
Tutorial 10: Bounded Linear Functionals in L
2
22
Exercise 21. Let (H, ¸, )) be a hilbert space over K. Let ( ⊆ H be
a nonempty closed convex subset of H. Let x
0
∈ H. Deﬁne:
δ
min
Z
= inf¦x −x
0
 : x ∈ (¦
1. Show the existence of a sequence (x
n
)
n≥1
in ( such that
x
n
−x
0
 → δ
min
.
2. Show that for all x, y ∈ H, we have:
x −y
2
= 2x
2
+ 2y
2
−4
_
_
_
_
x + y
2
_
_
_
_
2
3. Explain why for all n, m ≥ 1, we have:
δ
min
≤
_
_
_
_
x
n
+ x
m
2
−x
0
_
_
_
_
4. Show that for all n, m ≥ 1, we have:
x
n
−x
m

2
≤ 2x
n
−x
0

2
+ 2x
m
−x
0

2
−4δ
2
min
Tutorial 10: Bounded Linear Functionals in L
2
23
5. Show the existence of some x
∗
∈ H, such that x
n
J
·,·
→ x
∗
.
6. Explain why x
∗
∈ (
7. Show that for all x, y ∈ H, we have [ x −y [ ≤ x −y.
8. Show that x
n
−x
0
 → x
∗
−x
0
.
9. Conclude that we have found x
∗
∈ ( such that:
x
∗
−x
0
 = inf¦x −x
0
 : x ∈ (¦
10. Let y
∗
be another element of ( with such property. Show that:
x
∗
−y
∗

2
≤ 2x
∗
−x
0

2
+ 2y
∗
−x
0

2
−4δ
2
min
11. Conclude that x
∗
= y
∗
.
Tutorial 10: Bounded Linear Functionals in L
2
24
Theorem 52 Let (H, ¸, )) be a hilbert space over K, where K = R
or C. Let ( be a nonempty, closed and convex subset of H. For all
x
0
∈ H, there exists a unique x
∗
∈ ( such that:
x
∗
−x
0
 = inf¦x −x
0
 : x ∈ (¦
Deﬁnition 86 Let (H, ¸, )) be a hilbert space over K, where K = R
or C. Let ( ⊆ H. We call orthogonal of (, the subset of H denoted
(
⊥
and deﬁned by:
(
⊥
Z
= ¦ x ∈ H : ¸x, y) = 0 , ∀y ∈ ( ¦
Exercise 22. Let (H, ¸, )) be a hilbert space over K and ( ⊆ H.
1. Show that (
⊥
is a linear subspace of H, even if ( isn’t.
2. Show that φ
y
: H → K deﬁned by φ
y
(x) = ¸x, y) is continuous.
3. Show that (
⊥
= ∩
y∈Ç
φ
−1
y
(¦0¦).
Tutorial 10: Bounded Linear Functionals in L
2
25
4. Show that (
⊥
is a closed subset of H, even if ( isn’t.
5. Show that ∅
⊥
= ¦0¦
⊥
= H.
6. Show that H
⊥
= ¦0¦.
Exercise 23. Let (H, ¸, )) be a hilbert space over K. Let / be a
closed linear subspace of H, and x
0
∈ H.
1. Explain why there exists x
∗
∈ / such that:
x
∗
−x
0
 = inf¦ x −x
0
 : x ∈ / ¦
2. Deﬁne y
∗
= x
0
−x
∗
∈ H. Show that for all y ∈ / and α ∈ K:
y
∗

2
≤ y
∗
−αy
2
3. Show that for all y ∈ / and α ∈ K, we have:
0 ≤ −α¸y, y
∗
) −α¸y, y
∗
) +[α[
2
.y
2
Tutorial 10: Bounded Linear Functionals in L
2
26
4. For all y ∈ /¸ ¦0¦, taking α = ¸y, y
∗
)/y
2
, show that:
0 ≤ −
[¸y, y
∗
)[
2
y
2
5. Conclude that x
∗
∈ /, y
∗
∈ /
⊥
and x
0
= x
∗
+ y
∗
.
6. Show that /∩ /
⊥
= ¦0¦
7. Show that x
∗
∈ / and y
∗
∈ /
⊥
with x
0
= x
∗
+y
∗
, are unique.
Theorem 53 Let (H, ¸, )) be a hilbert space over K, where K = R
or C. Let / be a closed linear subspace of H. Then, for all x
0
∈ H,
there is a unique decomposition:
x
0
= x
∗
+ y
∗
where x
∗
∈ / and y
∗
∈ /
⊥
.
Tutorial 10: Bounded Linear Functionals in L
2
27
Deﬁnition 87 Let H be a Kvector space, where K = R or C.
We call linear functional, any map λ : H → K, such that for all
x, y ∈ H and α ∈ K:
λ(x + αy) = λ(x) + αλ(y)
Exercise 24. Let λ be a linear functional on a Khilbert
1
space H.
1. Suppose that λ is continuous at some point x
0
∈ H. Show the
existence of η > 0 such that:
∀x ∈ H , x −x
0
 ≤ η ⇒ [λ(x) −λ(x
0
)[ ≤ 1
Show that for all x ∈ H with x ,= 0, we have [λ(ηx/x)[ ≤ 1.
2. Show that if λ is continuous at x
0
, there exits M ∈ R
+
, with:
∀x ∈ H , [λ(x)[ ≤ Mx (2)
1
Norm vector spaces are introduced later in these tutorials.
Tutorial 10: Bounded Linear Functionals in L
2
28
3. Show conversely that if (2) holds, λ is continuous everywhere.
Deﬁnition 88 Let (H, ¸, )) be a hilbert
2
space over K = R or C.
Let λ be a linear functional on H. Then, the following are equivalent:
(i) λ : (H, T
¸,)
) → (K, T
K
) is continuous
(ii) ∃M ∈ R
+
, ∀x ∈ H , [λ(x)[ ≤ M.x
In which case, we say that λ is a bounded linear functional.
Exercise 25. Let (H, ¸, )) be a hilbert space over K. Let λ be a
bounded linear functional on H, such that λ(x) ,= 0 for some x ∈ H,
and deﬁne / = λ
−1
(¦0¦).
1. Show the existence of x
0
∈ H, such that x
0
,∈ /.
2. Show the existence of x
∗
∈ / and y
∗
∈ /
⊥
with x
0
= x
∗
+y
∗
.
2
Norm vector spaces are introduced later in these tutorials.
Tutorial 10: Bounded Linear Functionals in L
2
29
3. Deduce the existence of some z ∈ /
⊥
such that z = 1.
4. Show that for all α ∈ K¸ ¦0¦ and x ∈ H, we have:
λ(x)
¯ α
¸z, αz) = λ(x)
5. Show that in order to have:
∀x ∈ H , λ(x) = ¸x, αz)
it is suﬃcient to choose α ∈ K¸ ¦0¦ such that:
∀x ∈ H ,
λ(x)z
¯ α
−x ∈ /
6. Show the existence of y ∈ H such that:
∀x ∈ H , λ(x) = ¸x, y)
7. Show the uniqueness of such y ∈ H.
Tutorial 10: Bounded Linear Functionals in L
2
30
Theorem 54 Let (H, ¸, )) be a hilbert space over K, where K = R
or C. Let λ be a bounded linear functional on H. Then, there exists
a unique y ∈ H such that: ∀x ∈ H , λ(x) = ¸x, y).
Deﬁnition 89 Let K = R or C. We call Kvector space, any set
H, together with operators ⊕ and ⊗ for which there exits an element
0
1
∈ H such that for all x, y, z ∈ H and α, β ∈ K, we have:
(i) 0
1
⊕x = x
(ii) ∃(−x) ∈ H , (−x) ⊕x = 0
1
(iii) x ⊕(y ⊕z) = (x ⊕y) ⊕z
(iv) x ⊕y = y ⊕x
(v) 1 ⊗x = x
(vi) α ⊗(β ⊗x) = (αβ) ⊗x
(vii) (α + β) ⊗x = (α ⊗x) ⊕(β ⊗x)
(viii) α ⊗(x ⊕y) = (α ⊗x) ⊕(α ⊗y)
Tutorial 10: Bounded Linear Functionals in L
2
31
Exercise 26. For all f ∈ L
2
K
(Ω, T, µ), deﬁne:
H
Z
= ¦ [f] : f ∈ L
2
K
(Ω, T, µ) ¦
where [f] = ¦g ∈ L
2
K
(Ω, T, µ) : g = f, µa.s.¦. Let 0
1
= [0], and for
all [f], [g] ∈ H, and α ∈ K, we deﬁne:
[f] ⊕[g]
Z
= [f + g]
α ⊗[f]
Z
= [αf]
We assume f, f
t
, g and g
t
are elements of L
2
K
(Ω, T, µ).
1. Show that for f = g µa.s. is equivalent to [f] = [g].
2. Show that if [f] = [f
t
] and [g] = [g
t
], then [f + g] = [f
t
+ g
t
].
3. Conclude that ⊕ is welldeﬁned.
4. Show that ⊗ is also welldeﬁned.
5. Show that (H, ⊕, ⊗) is a Kvector space.
Tutorial 10: Bounded Linear Functionals in L
2
32
Exercise 27. Further to ex. (26), we deﬁne for all [f], [g] ∈ H:
¸[f], [g])
1
Z
=
_
Ω
f ¯ gdµ
1. Show that ¸, )
1
is welldeﬁned.
2. Show that ¸, )
1
is an innerproduct on H.
3. Show that (H, ¸, )
1
) is a hilbert space over K.
4. Why is ¸f, g)
Z
=
_
Ω
f ¯ gdµ not an innerproduct on L
2
K
(Ω, T, µ)?
Exercise 28. Further to ex. (27), Let λ : L
2
K
(Ω, T, µ) → K be a
continuous linear functional
3
. Deﬁne Λ : H → K by Λ([f]) = λ(f).
3
As deﬁned in these tutorials, L
2
K
(Ω, F, µ) is not a hilbert space (not even a
norm vector space). However, both L
2
K
(Ω, F, µ) and K have natural topologies
and it is therefore meaningful to speak of continuous linear functional. Note
however that we are slightly outside the framework of deﬁnition (88).
Tutorial 10: Bounded Linear Functionals in L
2
33
1. Show the existence of M ∈ R
+
such that:
∀f ∈ L
2
K
(Ω, T, µ) , [λ(f)[ ≤ M.f
2
2. Show that if [f] = [g] then λ(f) = λ(g).
3. Show that Λ is a well deﬁned bounded linear functional on H.
4. Conclude with the following:
Theorem 55 Let λ : L
2
K
(Ω, T, µ) → K be a continuous linear func
tional, where K = R or C. Then, there exists g ∈ L
2
K
(Ω, T, µ) such
that:
∀f ∈ L
2
K
(Ω, T, µ) , λ(f) =
_
Ω
f ¯ gdµ
Tutorial 11: Complex Measures 1
11. Complex Measures
In the following, (Ω, T) denotes an arbitrary measurable space.
Deﬁnition 90 Let (a
n
)
n≥1
be a sequence of complex numbers. We
say that (a
n
)
n≥1
has the permutation property if and only if, for
all bijections σ : N
∗
→ N
∗
, the series
+∞
k=1
a
σ(k)
converges in C
1
Exercise 1. Let (a
n
)
n≥1
be a sequence of complex numbers.
1. Show that if (a
n
)
n≥1
has the permutation property, then the
same is true of (Re(a
n
))
n≥1
and (Im(a
n
))
n≥1
.
2. Suppose a
n
∈ R for all n ≥ 1. Show that if
+∞
k=1
a
k
converges:
+∞
k=1
[a
k
[ = +∞ ⇒
+∞
k=1
a
+
k
=
+∞
k=1
a
−
k
= +∞
1
which excludes ±∞ as limit.
Tutorial 11: Complex Measures 2
Exercise 2. Let (a
n
)
n≥1
be a sequence in R, such that the series
+∞
k=1
a
k
converges, and
+∞
k=1
[a
k
[ = +∞. Let A > 0. We deﬁne:
N
+
= ¦k ≥ 1 : a
k
≥ 0¦ , N
−
= ¦k ≥ 1 : a
k
< 0¦
1. Show that N
+
and N
−
are inﬁnite.
2. Let φ
+
: N
∗
→ N
+
and φ
−
: N
∗
→ N
−
be two bijections. Show
the existence of k
1
≥ 1 such that:
k
1
k=1
a
φ
+
(k)
≥ A
3. Show the existence of an increasing sequence (k
p
)
p≥1
such that:
k
p
k=k
p−1
+1
a
φ
+
(k)
≥ A
for all p ≥ 1, where k
0
= 0.
Tutorial 11: Complex Measures 3
4. Consider the permutation σ : N
∗
→ N
∗
deﬁned informally by:
(φ
−
(1), φ
+
(1), . . . , φ
+
(k
1
)
. ¸¸ .
, φ
−
(2), φ
+
(k
1
+ 1), . . . , φ
+
(k
2
)
. ¸¸ .
, . . .)
representing (σ(1), σ(2), . . .). More speciﬁcally, deﬁne k
∗
0
= 0
and k
∗
p
= k
p
+p for all p ≥ 1. For all n ∈ N
∗
and p ≥ 1 with:
2
k
∗
p−1
< n ≤ k
∗
p
(1)
we deﬁne:
σ(n) =
_
φ
−
(p) if n = k
∗
p−1
+ 1
φ
+
(n −p) if n > k
∗
p−1
+ 1
(2)
Show that σ : N
∗
→ N
∗
is indeed a bijection.
5. Show that if
+∞
k=1
a
σ(k)
converges, there is N ≥ 1, such that:
n ≥ N , p ≥ 1 ⇒
¸
¸
¸
¸
¸
n+p
k=n+1
a
σ(k)
¸
¸
¸
¸
¸
< A
2
Given an integer n ≥ 1, there exists a unique p ≥ 1 such that (1) holds.
Tutorial 11: Complex Measures 4
6. Explain why (a
n
)
n≥1
cannot have the permutation property.
7. Prove the following theorem:
Theorem 56 Let (a
n
)
n≥1
be a sequence of complex numbers such
that for all bijections σ : N
∗
→ N
∗
, the series
+∞
k=1
a
σ(k)
converges.
Then, the series
+∞
k=1
a
k
converges absolutely, i.e.
+∞
k=1
[a
k
[ < +∞
Deﬁnition 91 Let (Ω, T) be a measurable space and E ∈ T. We
call measurable partition of E, any sequence (E
n
)
n≥1
of pairwise
disjoint elements of T, such that E = ¬
n≥1
E
n
.
Tutorial 11: Complex Measures 5
Deﬁnition 92 We call complex measure on a measurable space
(Ω, T) any map µ : T → C, such that for all E ∈ T and (E
n
)
n≥1
measurable partition of E, the series
+∞
n=1
µ(E
n
) converges to µ(E).
The set of all complex measures on (Ω, T) is denoted M
1
(Ω, T).
Deﬁnition 93 We call signed measure on a measurable space
(Ω, T), any complex measure on (Ω, T) with values in R.
3
Exercise 3.
1. Show that a measure on (Ω, T) may not be a complex measure.
2. Show that for all µ ∈ M
1
(Ω, T) , µ(∅) = 0.
3. Show that a ﬁnite measure on (Ω, T) is a complex measure with
values in R
+
, and conversely.
3
In these tutorials, signed measure may not have values in {−∞, +∞}.
Tutorial 11: Complex Measures 6
4. Let µ ∈ M
1
(Ω, T). Let E ∈ T and (E
n
)
n≥1
be a measurable
partition of E. Show that:
+∞
n=1
[µ(E
n
)[ < +∞
5. Let µ be a measure on (Ω, T) and f ∈ L
1
C
(Ω, T, µ). Deﬁne:
∀E ∈ T , ν(E)
=
_
E
fdµ
Show that ν is a complex measure on (Ω, T).
Deﬁnition 94 Let µ be a complex measure on a measurable space
(Ω, T). We call total variation of µ, the map [µ[ : T → [0, +∞],
deﬁned by:
∀E ∈ T , [µ[(E)
= sup
+∞
n=1
[µ(E
n
)[
where the ’sup’ is taken over all measurable partitions (E
n
)
n≥1
of E.
Tutorial 11: Complex Measures 7
Exercise 4. Let µ be a complex measure on (Ω, T).
1. Show that for all E ∈ T, [µ(E)[ ≤ [µ[(E).
2. Show that [µ[(∅) = 0.
Exercise 5. Let µ be a complex measure on (Ω, T). Let E ∈ T and
(E
n
)
n≥1
be a measurable partition of E.
1. Show that there exists (t
n
)
n≥1
in R, with t
n
< [µ[(E
n
) for all n.
2. Show that for all n ≥ 1, there exists a measurable partition
(E
p
n
)
p≥1
of E
n
such that:
t
n
<
+∞
p=1
[µ(E
p
n
)[
3. Show that (E
p
n
)
n,p≥1
is a measurable partition of E.
4. Show that for all N ≥ 1, we have
N
n=1
t
n
≤ [µ[(E).
Tutorial 11: Complex Measures 8
5. Show that for all N ≥ 1, we have:
N
n=1
[µ[(E
n
) ≤ [µ[(E)
6. Suppose that (A
p
)
p≥1
is another arbitrary measurable partition
of E. Show that for all p ≥ 1:
[µ(A
p
)[ ≤
+∞
n=1
[µ(A
p
∩ E
n
)[
7. Show that for all n ≥ 1:
+∞
p=1
[µ(A
p
∩ E
n
)[ ≤ [µ[(E
n
)
8. Show that:
+∞
p=1
[µ(A
p
)[ ≤
+∞
n=1
[µ[(E
n
)
Tutorial 11: Complex Measures 9
9. Show that [µ[ : T → [0, +∞] is a measure on (Ω, T).
Exercise 6. Let a, b ∈ R, a < b. Let F ∈ C
1
([a, b]; R), and deﬁne:
∀x ∈ [a, b] , H(x)
=
_
x
a
F
(t)dt
1. Show that H ∈ C
1
([a, b]; R) and H
= F
.
2. Show that:
F(b) −F(a) =
_
b
a
F
(t)dt
3. Show that:
1
2π
_
+π/2
−π/2
cos θdθ =
1
π
4. Let u ∈ R
n
and τ
u
: R
n
→ R
n
be the translation τ
u
(x) = x+u.
Show that the Lebesgue measure dx on (R
n
, B(R
n
)) is invariant
by translation τ
u
, i.e. dx(¦τ
u
∈ B¦) = dx(B) for all B ∈ B(R
n
).
Tutorial 11: Complex Measures 10
5. Show that for all f ∈ L
1
C
(R
n
, B(R
n
), dx), and u ∈ R
n
:
_
R
n
f(x +u)dx =
_
R
n
f(x)dx
6. Show that for all α ∈ R, we have:
_
+π
−π
cos
+
(α −θ)dθ =
_
+π−α
−π−α
cos
+
θdθ
7. Let α ∈ R and k ∈ Z such that k ≤ α/2π < k + 1. Show:
−π −α ≤ −2kπ −π < π −α ≤ −2kπ +π
8. Show that:
_
−2kπ−π
−π−α
cos
+
θdθ =
_
−2kπ+π
π−α
cos
+
θdθ
Tutorial 11: Complex Measures 11
9. Show that:
_
+π−α
−π−α
cos
+
θdθ =
_
−2kπ+π
−2kπ−π
cos
+
θdθ =
_
+π
−π
cos
+
θdθ
10. Show that for all α ∈ R:
1
2π
_
+π
−π
cos
+
(α −θ)dθ =
1
π
Exercise 7. Let z
1
, . . . , z
N
be N complex numbers. Let α
k
∈ R be
such that z
k
= [z
k
[e
iα
k
, for all k = 1, . . . , N. For all θ ∈ [−π, +π], we
deﬁne S(θ) = ¦k = 1, . . . , N : cos(α
k
−θ) > 0¦.
1. Show that for all θ ∈ [−π, +π], we have:
¸
¸
¸
¸
¸
¸
k∈S(θ)
z
k
¸
¸
¸
¸
¸
¸
=
¸
¸
¸
¸
¸
¸
k∈S(θ)
z
k
e
−iθ
¸
¸
¸
¸
¸
¸
≥
k∈S(θ)
[z
k
[ cos(α
k
−θ)
Tutorial 11: Complex Measures 12
2. Deﬁne φ : [−π, +π] → R by φ(θ) =
N
k=1
[z
k
[ cos
+
(α
k
− θ).
Show the existence of θ
0
∈ [−π, +π] such that:
φ(θ
0
) = sup
θ∈[−π,+π]
φ(θ)
3. Show that:
1
2π
_
+π
−π
φ(θ)dθ =
1
π
N
k=1
[z
k
[
4. Conclude that:
1
π
N
k=1
[z
k
[ ≤
¸
¸
¸
¸
¸
¸
k∈S(θ
0
)
z
k
¸
¸
¸
¸
¸
¸
Tutorial 11: Complex Measures 13
Exercise 8. Let µ ∈ M
1
(Ω, T). Suppose that [µ[(E) = +∞ for
some E ∈ T. Deﬁne t = π(1 +[µ(E)[) ∈ R
+
.
1. Show that there is a measurable partition (E
n
)
n≥1
of E, with:
t <
+∞
n=1
[µ(E
n
)[
2. Show the existence of N ≥ 1 such that:
t <
N
n=1
[µ(E
n
)[
3. Show the existence of S ⊆ ¦1, . . . , N¦ such that:
N
n=1
[µ(E
n
)[ ≤ π
¸
¸
¸
¸
¸
n∈S
µ(E
n
)
¸
¸
¸
¸
¸
4. Show that [µ(A)[ > t/π, where A = ¬
n∈S
E
n
.
Tutorial 11: Complex Measures 14
5. Let B = E ¸ A. Show that [µ(B)[ ≥ [µ(A)[ −[µ(E)[.
6. Show that E = A ¬ B with [µ(A)[ > 1 and [µ(B)[ > 1.
7. Show that [µ[(A) = +∞ or [µ[(B) = +∞.
Exercise 9. Let µ ∈ M
1
(Ω, T). Suppose that [µ[(Ω) = +∞.
1. Show the existence of A
1
, B
1
∈ T, such that Ω = A
1
¬ B
1
,
[µ(A
1
)[ > 1 and [µ[(B
1
) = +∞.
2. Show the existence of a sequence (A
n
)
n≥1
of pairwise disjoint
elements of T, such that [µ(A
n
)[ > 1 for all n ≥ 1.
3. Show that the series
+∞
n=1
µ(A
n
) does not converge to µ(A)
where A = ¬
+∞
n=1
A
n
.
4. Conclude that [µ[(Ω) < +∞.
Tutorial 11: Complex Measures 15
Theorem 57 Let µ be a complex measure on a measurable space
(Ω, T). Then, its total variation [µ[ is a ﬁnite measure on (Ω, T).
Exercise 10. Show that M
1
(Ω, T) is a Cvector space, with:
(λ +µ)(E)
= λ(E) +µ(E)
(αλ)(E)
= α.λ(E)
where λ, µ ∈ M
1
(Ω, T), α ∈ C, and E ∈ T.
Deﬁnition 95 Let H be a Kvector space, where K = R or C. We
call norm on H, any map N : H → R
+
, with the following properties:
(i) ∀x ∈ H , (N(x) = 0 ⇔ x = 0)
(ii) ∀x ∈ H, ∀α ∈ K , N(αx) = [α[N(x)
(iii) ∀x, y ∈ H , N(x +y) ≤ N(x) +N(y)
Tutorial 11: Complex Measures 16
Exercise 11.
1. Explain why .
p
may not be a norm on L
p
K
(Ω, T, µ).
2. Show that  =
_
¸, ) is a norm, when ¸, ) is an innerproduct.
3. Show that µ
= [µ[(Ω) deﬁnes a norm on M
1
(Ω, T).
Exercise 12. Let µ ∈ M
1
(Ω, T) be a signed measure. Show that:
µ
+
=
1
2
([µ[ +µ)
µ
−
=
1
2
([µ[ −µ)
are ﬁnite measures such that:
µ = µ
+
−µ
−
, [µ[ = µ
+
+µ
−
Tutorial 11: Complex Measures 17
Exercise 13. Let µ ∈ M
1
(Ω, T) and l : R
2
→ R be a linear map.
1. Show that l is continuous.
2. Show that l ◦ µ is a signed measure on (Ω, T).
4
3. Show that all µ ∈ M
1
(Ω, T) can be decomposed as:
µ = µ
1
−µ
2
+i(µ
3
−µ
4
)
where µ
1
, µ
2
, µ
3
, µ
4
are ﬁnite measures.
4
l ◦ µ refers strictly speaking to l(Re(µ), Im(µ)).
Tutorial 12: RadonNikodym Theorem 1
12. RadonNikodym Theorem
In the following, (Ω, T) is an arbitrary measurable space.
Deﬁnition 96 Let µ and ν be two (possibly complex) measures on
(Ω, T). We say that ν is absolutely continuous with respect to µ,
and we write ν << µ, if and only if, for all E ∈ T:
µ(E) = 0 ⇒ ν(E) = 0
Exercise 1. Let µ be a measure on (Ω, T) and ν ∈ M
1
(Ω, T). Show
that ν << µ is equivalent to [ν[ << µ.
Exercise 2. Let µ be a measure on (Ω, T) and ν ∈ M
1
(Ω, T). Let
> 0. Suppose there exists a sequence (E
n
)
n≥1
in T such that:
∀n ≥ 1 , µ(E
n
) ≤
1
2
n
, [ν(E
n
)[ ≥
Deﬁne:
E
= limsup
n≥1
E
n
=
n≥1
_
k≥n
E
k
Tutorial 12: RadonNikodym Theorem 2
1. Show that:
µ(E) = lim
n→+∞
µ
_
_
_
k≥n
E
k
_
_
= 0
2. Show that:
[ν[(E) = lim
n→+∞
[ν[
_
_
_
k≥n
E
k
_
_
≥
3. Let λ be a measure on (Ω, T). Can we conclude in general that:
λ(E) = lim
n→+∞
λ
_
_
_
k≥n
E
k
_
_
4. Prove the following:
Tutorial 12: RadonNikodym Theorem 3
Theorem 58 Let µ be a measure on (Ω, T) and ν be a complex
measure on (Ω, T). The following are equivalent:
(i) ν << µ
(ii) [ν[ << µ
(iii) ∀ > 0, ∃δ > 0, ∀E ∈ T, µ(E) ≤ δ ⇒ [ν(E)[ <
Exercise 3. Let µ be a measure on (Ω, T) and ν ∈ M
1
(Ω, T) such
that ν << µ. Let ν
1
= Re(ν) and ν
2
= Im(ν).
1. Show that ν
1
<< µ and ν
2
<< µ.
2. Show that ν
+
1
, ν
−
1
, ν
+
2
, ν
−
2
are absolutely continuous w.r. to µ.
Exercise 4. Let µ be a ﬁnite measure on (Ω, T) and f ∈ L
1
C
(Ω, T, µ).
Let S be a closed subset of C. We assume that for all E ∈ T such
that µ(E) > 0, we have:
1
µ(E)
_
E
fdµ ∈ S
Tutorial 12: RadonNikodym Theorem 4
1. Show the existence of a sequence (D
n
) of closed discs in C with:
S
c
=
+∞
_
n=1
D
n
Let α
n
∈ C, r
n
> 0 be such that D
n
= ¦z ∈ C : [z −α
n
[ ≤ r
n
¦.
2. Suppose µ(E
n
) > 0 for some n ≥ 1, where E
n
= ¦f ∈ D
n
¦.
Show that:
¸
¸
¸
¸
1
µ(E
n
)
_
E
n
fdµ −α
n
¸
¸
¸
¸
≤
1
µ(E
n
)
_
E
n
[f −α
n
[dµ ≤ r
n
3. Show that for all n ≥ 1, µ(¦f ∈ D
n
¦) = 0.
4. Prove the following:
Tutorial 12: RadonNikodym Theorem 5
Theorem 59 Let µ be a ﬁnite measure on (Ω, T), f ∈ L
1
C
(Ω, T, µ).
Let S be a closed subset of C such that for all E ∈ T with µ(E) > 0,
we have:
1
µ(E)
_
E
fdµ ∈ S
Then, f ∈ S µa.s.
Exercise 5. Let µ be a σﬁnite measure on (Ω, T). Let (E
n
)
n≥1
be
a sequence in T such that E
n
↑ Ω and µ(E
n
) < +∞ for all n ≥ 1.
Deﬁne w : (Ω, T) → (R, B(R)) as:
w
=
+∞
n=1
1
2
n
1
1 +µ(E
n
)
1
E
n
1. Show that for all ω ∈ Ω, 0 < w(ω) ≤ 1.
2. Show that w ∈ L
1
R
(Ω, T, µ).
Tutorial 12: RadonNikodym Theorem 6
Exercise 6. Let µ be a σﬁnite measure on (Ω, T) and ν be a ﬁnite
measure on (Ω, T), such that ν << µ. Let w ∈ L
1
R
(Ω, T, µ) be such
that 0 < w ≤ 1. We deﬁne ¯ µ =
_
wdµ, i.e.
∀E ∈ T , ¯ µ(E)
=
_
E
wdµ
1. Show that ¯ µ is a ﬁnite measure on (Ω, T).
2. Show that φ = ν + ¯ µ is also a ﬁnite measure on (Ω, T).
3. Show that for all f ∈ L
1
C
(Ω, T, φ), we have f ∈ L
1
C
(Ω, T, ν),
fw ∈ L
1
C
(Ω, T, µ), and:
_
Ω
fdφ =
_
Ω
fdν +
_
Ω
fwdµ
4. Show that for all f ∈ L
2
C
(Ω, T, φ), we have:
_
Ω
[f[dν ≤
_
Ω
[f[dφ ≤
__
Ω
[f[
2
dφ
_
1
2
(φ(Ω))
1
2
Tutorial 12: RadonNikodym Theorem 7
5. Show that L
2
C
(Ω, T, φ) ⊆ L
1
C
(Ω, T, ν), and for f ∈ L
2
C
(Ω, T, φ):
¸
¸
¸
¸
_
Ω
fdν
¸
¸
¸
¸
≤
_
φ(Ω).f
2
6. Show the existence of g ∈ L
2
C
(Ω, T, φ) such that:
∀f ∈ L
2
C
(Ω, T, φ) ,
_
Ω
fdν =
_
Ω
fgdφ (1)
7. Show that for all E ∈ T such that φ(E) > 0, we have:
1
φ(E)
_
E
gdφ ∈ [0, 1]
8. Show the existence of g ∈ L
2
C
(Ω, T, φ) such that g(ω) ∈ [0, 1]
for all ω ∈ Ω, and (1) still holds.
9. Show that for all f ∈ L
2
C
(Ω, T, φ), we have:
_
Ω
f(1 −g)dν =
_
Ω
fgwdµ
Tutorial 12: RadonNikodym Theorem 8
10. Show that for all n ≥ 1 and E ∈ T,
f
= (1 +g +. . . +g
n
)1
E
∈ L
2
C
(Ω, T, φ)
11. Show that for all n ≥ 1 and E ∈ T,
_
E
(1 −g
n+1
)dν =
_
E
g(1 +g +. . . +g
n
)wdµ
12. Deﬁne:
h
= gw
_
+∞
n=0
g
n
_
Show that if A = ¦0 ≤ g < 1¦, then for all E ∈ T:
ν(E ∩ A) =
_
E
hdµ
13. Show that ¦h = +∞¦ = A
c
and conclude that µ(A
c
) = 0.
14. Show that for all E ∈ T, we have ν(E) =
_
E
hdµ.
Tutorial 12: RadonNikodym Theorem 9
15. Show that if µ is σﬁnite on (Ω, T), and ν is a ﬁnite measure
on (Ω, T) such that ν << µ, there exists h ∈ L
1
R
(Ω, T, µ), such
that h ≥ 0 and:
∀E ∈ T , ν(E) =
_
E
hdµ
16. Prove the following:
Theorem 60 (RadonNikodym:1) Let µ be a σﬁnite measure on
(Ω, T). let ν be a complex measure on (Ω, T) such that ν << µ. Then,
there exists some h ∈ L
1
C
(Ω, T, µ) such that:
∀E ∈ T , ν(E) =
_
E
hdµ
If ν is a signed measure on (Ω, T), we can assume h ∈ L
1
R
(Ω, T, µ).
If ν is a ﬁnite measure on (Ω, T), we can assume h ≥ 0.
Tutorial 12: RadonNikodym Theorem 10
Exercise 7. Let f = u +iv ∈ L
1
C
(Ω, T, µ), such that:
∀E ∈ T ,
_
E
fdµ = 0
where µ is a measure on (Ω, T).
1. Show that:
_
Ω
u
+
dµ =
_
{u≥0}
udµ
2. Show that f = 0 µa.s.
3. State and prove some uniqueness property in theorem (60).
Exercise 8. Let µ and ν be two σﬁnite measures on (Ω, T) such
that ν << µ. Let (E
n
)
n≥1
be a sequence in T such that E
n
↑ Ω and
ν(E
n
) < +∞ for all n ≥ 1. We deﬁne:
∀n ≥ 1 , ν
n
= ν
E
n
= ν(E
n
∩ )
Tutorial 12: RadonNikodym Theorem 11
1. Show that there exists h
n
∈ L
1
R
(Ω, T, µ) with h
n
≥ 0 and:
∀E ∈ T , ν
n
(E) =
_
E
h
n
dµ (2)
for all n ≥ 1.
2. Show that for all E ∈ T,
_
E
h
n
dµ ≤
_
E
h
n+1
dµ
3. Show that for all n, p ≥ 1,
µ(¦h
n
−h
n+1
>
1
p
¦) = 0
4. Show that h
n
≤ h
n+1
µa.s.
5. Show the existence of a sequence (h
n
)
n≥1
in L
1
R
(Ω, T, µ) such
that 0 ≤ h
n
≤ h
n+1
for all n ≥ 1 and with (2) still holding.
Tutorial 12: RadonNikodym Theorem 12
6. Let h = sup
n≥1
h
n
. Show that:
∀E ∈ T , ν(E) =
_
E
hdµ (3)
7. Show that for all n ≥ 1,
_
E
n
hdµ < +∞.
8. Show that h < +∞ µa.s.
9. Show there exists h : (Ω, T) → R
+
measurable, while (3) holds.
10. Show that for all n ≥ 1, h ∈ L
1
R
(Ω, T, µ
E
n
).
Theorem 61 (RadonNikodym:2) Let µ and ν be two σﬁnite
measures on (Ω, T) such that ν << µ. There exists a measurable
map h : (Ω, T) → (R
+
, B(R
+
)) such that:
∀E ∈ T , ν(E) =
_
E
hdµ
Tutorial 12: RadonNikodym Theorem 13
Exercise 9. Let h, h
: (Ω, T) → [0, +∞] be two nonnegative and
measurable maps. Let µ be a σﬁnite measure on (Ω, T). We assume:
∀E ∈ T ,
_
E
hdµ =
_
E
h
dµ
Let (E
n
)
n≥1
be a sequence in T with E
n
↑ Ω and µ(E
n
) < +∞ for
all n ≥ 1. We deﬁne F
n
= E
n
∩ ¦h ≤ n¦ for all n ≥ 1.
1. Show that for all n and E ∈ T,
_
E
hdµ
F
n
=
_
E
h
dµ
F
n
< +∞.
2. Show that for all n, p ≥ 1, µ(F
n
∩ ¦h > h
+ 1/p¦) = 0.
3. Show that for all n ≥ 1, µ(¦F
n
∩ ¦h ,= h
¦) = 0.
4. Show that µ(¦h ,= h
¦ ∩ ¦h < +∞¦) = 0.
5. Show that h = h
µa.s.
6. State and prove some uniqueness property in theorem (61).
Tutorial 12: RadonNikodym Theorem 14
Exercise 10. Take Ω = ¦∗¦ and T = T(Ω) = ¦∅, ¦∗¦¦. Let µ be
the measure on (Ω, T) deﬁned by µ(∅) = 0 and µ(¦∗¦) = +∞. Let
h, h
: (Ω, T) → [0, +∞] be deﬁned by h(∗) = 1 ,= 2 = h
(∗). Show
that we have:
∀E ∈ T ,
_
E
hdµ =
_
E
h
dµ
Explain why this does not contradict the previous exercise.
Exercise 11. Let µ be a complex measure on (Ω, T).
1. Show that µ << [µ[.
2. Show the existence of some h ∈ L
1
C
(Ω, T, [µ[) such that:
∀E ∈ T , µ(E) =
_
E
hd[µ[
3. If µ is a signed measure, can we assume h ∈ L
1
R
(Ω, T, [µ[)?
Tutorial 12: RadonNikodym Theorem 15
Exercise 12. Further to ex. (11), deﬁne A
r
= ¦[h[ < r¦ for all r > 0.
1. Show that for all measurable partition (E
n
)
n≥1
of A
r
:
+∞
n=1
[µ(E
n
)[ ≤ r[µ[(A
r
)
2. Show that [µ[(A
r
) = 0 for all 0 < r < 1.
3. Show that [h[ ≥ 1 [µ[a.s.
4. Suppose that E ∈ T is such that [µ[(E) > 0. Show that:
¸
¸
¸
¸
1
[µ[(E)
_
E
hd[µ[
¸
¸
¸
¸
≤ 1
5. Show that [h[ ≤ 1 [µ[a.s.
6. Prove the following:
Tutorial 12: RadonNikodym Theorem 16
Theorem 62 For all complex measure µ on (Ω, T), there exists h
belonging to L
1
C
(Ω, T, [µ[) such that [h[ = 1 and:
∀E ∈ T , µ(E) =
_
E
hd[µ[
If µ is a signed measure on (Ω, T), we can assume h ∈ L
1
R
(Ω, T, [µ[).
Exercise 13. Let A ∈ T, and (A
n
)
n≥1
be a sequence in T.
1. Show that if A
n
↑ A then 1
A
n
↑ 1
A
.
2. Show that if A
n
↓ A then 1
A
n
↓ 1
A
.
3. Show that if 1
A
n
→ 1
A
, then for all µ ∈ M
1
(Ω, T):
µ(A) = lim
n→+∞
µ(A
n
)
Tutorial 12: RadonNikodym Theorem 17
Exercise 14. Let µ be a measure on (Ω, T) and f ∈ L
1
C
(Ω, T, µ).
1. Show that ν =
_
fdµ ∈ M
1
(Ω, T).
2. Let h ∈ L
1
C
(Ω, T, [ν[) be such that [h[ = 1 and ν =
_
hd[ν[.
Show that for all E, F ∈ T:
_
E
f1
F
dµ =
_
E
h1
F
d[ν[
3. Show that if g : (Ω, T) → (C, B(C)) is bounded and measurable:
∀E ∈ T ,
_
E
fgdµ =
_
E
hgd[ν[
4. Show that:
∀E ∈ T , [ν[(E) =
_
E
f
¯
hdµ
5. Show that for all E ∈ T,
_
E
Re(f
¯
h)dµ ≥ 0 ,
_
E
Im(f
¯
h)dµ = 0
Tutorial 12: RadonNikodym Theorem 18
6. Show that f
¯
h ∈ R
+
µa.s.
7. Show that f
¯
h = [f[ µa.s.
8. Prove the following:
Theorem 63 Let µ be a measure on (Ω, T) and f ∈ L
1
C
(Ω, T, µ).
Then, ν =
_
fdµ deﬁned by:
∀E ∈ T , ν(E)
=
_
E
fdµ
is a complex measure on (Ω, T) with total variation:
∀E ∈ T , [ν[(E) =
_
E
[f[dµ
Tutorial 12: RadonNikodym Theorem 19
Exercise 15. Let µ ∈ M
1
(Ω, T) be a signed measure. Suppose that
h ∈ L
1
R
(Ω, T, [µ[) is such that [h[ = 1 and µ =
_
hd[µ[. Deﬁne
A = ¦h = 1¦ and B = ¦h = −1¦.
1. Show that for all E ∈ T, µ
+
(E) =
_
E
1
2
(1 +h)d[µ[.
2. Show that for all E ∈ T, µ
−
(E) =
_
E
1
2
(1 −h)d[µ[.
3. Show that µ
+
= µ
A
= µ(A ∩ ).
4. Show that µ
−
= −µ
B
= −µ(B ∩ ).
Theorem 64 (Hahn Decomposition) Let µ be a signed measure
on (Ω, T). There exist A, B ∈ T, such that A ∩ B = ∅, Ω = A ¬ B
and for all E ∈ T, µ
+
(E) = µ(A ∩ E) and µ
−
(E) = −µ(B ∩ E).
Tutorial 12: RadonNikodym Theorem 20
Deﬁnition 97 Let µ be a complex measure on (Ω, T). We deﬁne:
L
1
C
(Ω, T, µ)
= L
1
C
(Ω, T, [µ[)
and for all f ∈ L
1
C
(Ω, T, µ), the lebesgue integral of f with respect
to µ, is deﬁned as:
_
fdµ
=
_
fhd[µ[
where h ∈ L
1
C
(Ω, T, [µ[) is such that [h[ = 1 and µ =
_
hd[µ[.
Exercise 16. Let µ be a complex measure on (Ω, T).
1. Show that for all f : (Ω, T) → (C, B(C)) measurable:
f ∈ L
1
C
(Ω, T, µ) ⇔
_
[f[d[µ[ < +∞
2. Show that for f ∈ L
1
C
(Ω, T, µ),
_
fdµ is unambiguously deﬁned.
3. Show that for all E ∈ T, 1
E
∈ L
1
C
(Ω, T, µ) and
_
1
E
dµ = µ(E).
Tutorial 12: RadonNikodym Theorem 21
4. Show that if µ is a ﬁnite measure, then [µ[ = µ.
5. Show that if µ is a ﬁnite measure, deﬁnition (97) of integral
and space L
1
C
(Ω, T, µ) is consistent with that already known
for measures.
6. Show that L
1
C
(Ω, T, µ) is a Cvector space and that:
_
(f +αg)dµ =
_
fdµ +α
_
gdµ
for all f, g ∈ L
1
C
(Ω, T, µ) and α ∈ C.
7. Show that for all f ∈ L
1
C
(Ω, T, µ), we have:
¸
¸
¸
¸
_
fdµ
¸
¸
¸
¸
≤
_
[f[d[µ[
Tutorial 12: RadonNikodym Theorem 22
Exercise 17. Let µ, ν ∈ M
1
(Ω, T), let α ∈ C.
1. Show that [αν[ = [α[.[ν[
2. Show that [µ +ν[ ≤ [µ[ +[ν[
3. Show that L
1
C
(Ω, T, µ) ∩ L
1
C
(Ω, T, ν) ⊆ L
1
C
(Ω, T, µ +αν)
4. Show that for all E ∈ T:
_
1
E
d(µ +αν) =
_
1
E
dµ +α
_
1
E
dν
5. Show that for all f ∈ L
1
C
(Ω, T, µ) ∩ L
1
C
(Ω, T, ν):
_
fd(µ +αν) =
_
fdµ +α
_
fdν
Tutorial 12: RadonNikodym Theorem 23
Exercise 18. Let µ = µ
1
+iµ
2
∈ M
1
(Ω, T).
1. Show that [µ
1
[ ≤ [µ[ and [µ
2
[ ≤ [µ[.
2. Show that [µ[ ≤ [µ
1
[ +[µ
2
[.
3. Show that L
1
C
(Ω, T, µ) = L
1
C
(Ω, T, µ
1
) ∩ L
1
C
(Ω, T, µ
2
).
4. Show that:
L
1
C
(Ω, T, µ
1
) = L
1
C
(Ω, T, µ
+
1
) ∩ L
1
C
(Ω, T, µ
−
1
)
L
1
C
(Ω, T, µ
2
) = L
1
C
(Ω, T, µ
+
2
) ∩ L
1
C
(Ω, T, µ
−
2
)
5. Show that for all f ∈ L
1
C
(Ω, T, µ):
_
fdµ =
_
fdµ
+
1
−
_
fdµ
−
1
+i
__
fdµ
+
2
−
_
fdµ
−
2
_
Tutorial 12: RadonNikodym Theorem 24
Exercise 19. Let µ ∈ M
1
(Ω, T). Let A ∈ T. Let h ∈ L
1
C
(Ω, T, [µ[)
be such that [h[ = 1 and µ =
_
hd[µ[. Recall that µ
A
= µ(A∩ ) and
µ
A
= µ
(F
A
)
where T
A
= ¦A∩ E , E ∈ T¦ ⊆ T.
1. Show that we also have T
A
= ¦E : E ∈ T , E ⊆ A¦.
2. Show that µ
A
∈ M
1
(Ω, T) and µ
A
∈ M
1
(A, T
A
).
3. Let E ∈ T and (E
n
)
n≥1
be a measurable partition of E. Show:
+∞
n=1
[µ
A
(E
n
)[ ≤ [µ[
A
(E)
4. Show that we have [µ
A
[ ≤ [µ[
A
.
5. Let E ∈ T and (E
n
)
n≥1
be a measurable partition of A ∩ E.
Show that:
+∞
n=1
[µ(E
n
)[ ≤ [µ
A
[(A ∩ E)
Tutorial 12: RadonNikodym Theorem 25
6. Show that [µ
A
[(A
c
) = 0.
7. Show that [µ
A
[ = [µ[
A
.
8. Let E ∈ T
A
and (E
n
)
n≥1
be an T
A
measurable partition of E.
Show that:
+∞
n=1
[µ
A
(E
n
)[ ≤ [µ[
A
(E)
9. Show that [µ
A
[ ≤ [µ[
A
.
10. Let E ∈ T
A
⊆ T and (E
n
)
n≥1
be a measurable partition of E.
Show that (E
n
)
n≥1
is also an T
A
measurable partition of E,
and conclude:
+∞
n=1
[µ(E
n
)[ ≤ [µ
A
[(E)
11. Show that [µ
A
[ = [µ[
A
.
12. Show that µ
A
=
_
hd[µ
A
[.
Tutorial 12: RadonNikodym Theorem 26
13. Show that h
A
∈ L
1
C
(A, T
A
, [µ
A
[) and µ
A
=
_
h
A
d[µ
A
[.
14. Show that for all f ∈ L
1
C
(Ω, T, µ), we have:
f1
A
∈ L
1
C
(Ω, T, µ) , f ∈ L
1
C
(Ω, T, µ
A
) , f
A
∈ L
1
C
(A, T
A
, µ
A
)
and:
_
f1
A
dµ =
_
fdµ
A
=
_
f
A
dµ
A
Deﬁnition 98 Let f ∈ L
1
C
(Ω, T, µ) , where µ is a complex measure
on (Ω, T). let A ∈ T. We call partial lebesgue integral of f with
respect to µ over A, the integral denoted
_
A
fdµ, deﬁned as:
_
A
fdµ
=
_
(f1
A
)dµ =
_
fdµ
A
=
_
(f
A
)dµ
A
where µ
A
is the complex measure on (Ω, T), µ
A
= µ(A ∩ ), f
A
is
the restriction of f to A and µ
A
is the restriction of µ to T
A
, the
trace of T on A.
Tutorial 12: RadonNikodym Theorem 27
Exercise 20. Prove the following:
Theorem 65 Let f ∈ L
1
C
(Ω, T, µ), where µ is a complex measure
on (Ω, T). Then, ν =
_
fdµ deﬁned as:
∀E ∈ T , ν(E)
=
_
E
fdµ
is a complex measure on (Ω, T), with total variation:
∀E ∈ T , [ν[(E) =
_
E
[f[d[µ[
Moreover, for all measurable map g : (Ω, T) → (C, B(C)), we have:
g ∈ L
1
C
(Ω, T, ν) ⇔ gf ∈ L
1
C
(Ω, T, µ)
and when such condition is satisﬁed:
_
gdν =
_
gfdµ
Tutorial 12: RadonNikodym Theorem 28
Exercise 21. Let (Ω
1
, T
1
), . . . , (Ω
n
, T
n
) be n measurable spaces,
where n ≥ 2. Let µ
1
∈ M
1
(Ω
1
, T
1
), . . ., µ
n
∈ M
1
(Ω
n
, T
n
). For
all i ∈ N
n
, let h
i
belonging to L
1
C
(Ω
i
, T
i
, [µ
i
[) be such that [h
i
[ = 1
and µ
i
=
_
h
i
d[µ
i
[. For all E ∈ T
1
⊗. . . ⊗T
n
, we deﬁne:
µ(E)
=
_
E
h
1
. . . h
n
d[µ
1
[ ⊗. . . ⊗[µ
n
[
1. Show that µ ∈ M
1
(Ω
1
. . . Ω
n
, T
1
⊗. . . ⊗T
n
)
2. Show that for all measurable rectangle A
1
. . . A
n
:
µ(A
1
. . . A
n
) = µ
1
(A
1
) . . . µ
n
(A
n
)
3. Prove the following:
Tutorial 12: RadonNikodym Theorem 29
Theorem 66 Let µ
1
, . . . , µ
n
be n complex measures on measurable
spaces (Ω
1
, T
1
), . . . , (Ω
n
, T
n
) respectively, where n ≥ 2. There exists
a unique complex measure µ
1
⊗. . .⊗µ
n
on (Ω
1
. . .Ω
n
, T
1
⊗. . .⊗T
n
)
such that for all measurable rectangle A
1
. . . A
n
, we have:
µ
1
⊗. . . ⊗µ
n
(A
1
. . . A
n
) = µ
1
(A
1
) . . . µ
n
(A
n
)
Exercise 22. Further to theorem (66),
1. Show that [µ
1
⊗. . . ⊗µ
n
[ = [µ
1
[ ⊗. . . ⊗[µ
n
[.
2. Show that µ
1
⊗. . . ⊗µ
n
 = µ
1
 . . . µ
n
.
3. Show that for all E ∈ T
1
⊗. . . ⊗T
n
:
µ
1
⊗. . . ⊗µ
n
(E) =
_
E
h
1
. . . h
n
d[µ
1
⊗. . . ⊗µ
n
[
4. Let f ∈ L
1
C
(Ω
1
. . . Ω
n
, T
1
⊗. . . ⊗T
n
, µ
1
⊗. . . ⊗µ
n
). Show:
_
fdµ
1
⊗. . . ⊗µ
n
=
_
fh
1
. . . h
n
d[µ
1
[ ⊗. . . ⊗[µ
n
[
Tutorial 12: RadonNikodym Theorem 30
5. let σ be a permutation of ¦1, . . . , n¦. Show that:
_
fdµ
1
⊗. . . ⊗µ
n
=
_
Ω
σ(n)
. . .
_
Ω
σ(1)
fdµ
σ(1)
. . . dµ
σ(n)
Tutorial 13: Regular Measure 1
13. Regular Measure
In the following, K denotes R or C.
Deﬁnition 99 Let (Ω, T) be a measurable space. We say that a map
s : Ω → C is a complex simple function on (Ω, T), if and only if
it is of the form:
s =
n
i=1
α
i
1
A
i
where n ≥ 1, α
i
∈ C and A
i
∈ T for all i ∈ N
n
. The set of all
complex simple functions on (Ω, T) is denoted S
C
(Ω, T). The set of
all Rvalued complex simple functions in (Ω, T) is denoted S
R
(Ω, T).
Recall that a simple function on (Ω, T), as deﬁned in (40), is just a
nonnegative element of S
R
(Ω, T).
Tutorial 13: Regular Measure 2
Exercise 1. Let (Ω, T, µ) be a measure space and p ∈ [1, +∞[.
1. Suppose s : Ω → C is of the form
s =
n
i=1
α
i
1
A
i
where n ≥ 1, α
i
∈ C, A
i
∈ T and µ(A
i
) < +∞ for all i ∈ N
n
.
Show that s ∈ L
p
C
(Ω, T, µ) ∩ S
C
(Ω, T).
2. Show that any s ∈ S
C
(Ω, T) can be written as:
s =
n
i=1
α
i
1
A
i
where n ≥ 1, α
i
∈ C¸ ¦0¦, A
i
∈ T and A
i
∩ A
j
= ∅ for i ,= j.
3. Show that any s ∈ L
p
C
(Ω, T, µ) ∩ S
C
(Ω, T) is of the form:
s =
n
i=1
α
i
1
A
i
Tutorial 13: Regular Measure 3
where n ≥ 1, α
i
∈ C, A
i
∈ T and µ(A
i
) < +∞, for all i ∈ N
n
.
4. Show that L
∞
C
(Ω, T, µ) ∩ S
C
(Ω, T) = S
C
(Ω, T).
Exercise 2. Let (Ω, T, µ) be a measure space and p ∈ [1, +∞[. Let
f be a nonnegative element of L
p
R
(Ω, T, µ).
1. Show the existence of a sequence (s
n
)
n≥1
of nonnegative func
tions in L
p
R
(Ω, T, µ) ∩ S
R
(Ω, T) such that s
n
↑ f.
2. Show that:
lim
n→+∞
[s
n
−f[
p
dµ = 0
3. Show that there exists s ∈ L
p
R
(Ω, T, µ) ∩ S
R
(Ω, T) such that
f −s
p
≤ , for all > 0.
4. Show that L
p
K
(Ω, T, µ) ∩ S
K
(Ω, T) is dense in L
p
K
(Ω, T, µ).
Tutorial 13: Regular Measure 4
Exercise 3. Let (Ω, T, µ) be a measure space. Let f be a non
negative element of L
∞
R
(Ω, T, µ). For all n ≥ 1, we deﬁne:
s
n
=
n2
n
−1
k=0
k
2
n
1
{k/2
n
≤f<(k+1)/2
n
}
+n1
{n≤f}
1. Show that for all n ≥ 1, s
n
is a simple function.
2. Show there exists n
0
≥ 1 and N ∈ T with µ(N) = 0, such that:
∀ω ∈ N
c
, 0 ≤ f(ω) < n
0
3. Show that for all n ≥ n
0
and ω ∈ N
c
, we have:
0 ≤ f(ω) −s
n
(ω) <
1
2
n
4. Conclude that:
lim
n→+∞
f −s
n

∞
= 0
5. Show the following:
Tutorial 13: Regular Measure 5
Theorem 67 Let (Ω, T, µ) be a measure space and p ∈ [1, +∞].
Then, L
p
K
(Ω, T, µ) ∩ S
K
(Ω, T) is dense in L
p
K
(Ω, T, µ).
Exercise 4. Let (Ω, T ) be a metrizable topological space, and µ be
a ﬁnite measure on (Ω, B(Ω)). We deﬁne Σ as the set of all B ∈ B(Ω)
such that for all > 0, there exist F closed and G open in Ω, with:
F ⊆ B ⊆ G , µ(G ¸ F) ≤
Given a metric d on (Ω, T ) inducing the topology T , we deﬁne:
d(x, A)
= inf¦d(x, y) : y ∈ A¦
for all A ⊆ Ω and x ∈ Ω.
1. Show that x → d(x, A) from Ω to
¯
R is continuous for all A ⊆ Ω.
2. Show that if F is closed in Ω, x ∈ F is equivalent to d(x, F) = 0.
Tutorial 13: Regular Measure 6
Exercise 5. Further to exercise (4), we assume that F is a closed
subset of Ω. For all n ≥ 1, we deﬁne:
G
n
= ¦x ∈ Ω : d(x, F) <
1
n
¦
1. Show that G
n
is open for all n ≥ 1.
2. Show that G
n
↓ F.
3. Show that F ∈ Σ.
4. Was it important to assume that µ is ﬁnite?
5. Show that Ω ∈ Σ.
6. Show that if B ∈ Σ, then B
c
∈ Σ.
Tutorial 13: Regular Measure 7
Exercise 6. Further to exercise (5), let (B
n
)
n≥1
be a sequence in Σ.
Deﬁne B = ∪
+∞
n=1
B
n
and let > 0.
1. Show that for all n, there is F
n
closed and G
n
open in Ω, with:
F
n
⊆ B
n
⊆ G
n
, µ(G
n
¸ F
n
) ≤
2
n
2. Show the existence of some N ≥ 1 such that:
µ
+∞
n=1
F
n
¸
N
n=1
F
n
≤
3. Deﬁne G = ∪
+∞
n=1
G
n
and F = ∪
N
n=1
F
n
. Show that F is closed,
G is open and F ⊆ B ⊆ G.
4. Show that:
G¸ F ⊆ G¸
+∞
n=1
F
n
¬
+∞
n=1
F
n
¸ F
Tutorial 13: Regular Measure 8
5. Show that:
G¸
+∞
n=1
F
n
⊆
+∞
n=1
G
n
¸ F
n
6. Show that µ(G¸ F) ≤ 2.
7. Show that Σ is a σalgebra on Ω, and conclude that Σ = B(Ω).
Theorem 68 Let (Ω, T ) be a metrizable topological space, and µ be
a ﬁnite measure on (Ω, B(Ω)). Then, for all B ∈ B(Ω) and > 0,
there exist F closed and G open in Ω such that:
F ⊆ B ⊆ G , µ(G¸ F) ≤
Deﬁnition 100 Let (Ω, T ) be a topological space. We denote C
b
K
(Ω)
the Kvector space of all continuous, bounded maps φ : Ω → K,
where K = R or K = C.
Tutorial 13: Regular Measure 9
Exercise 7. Let (Ω, T ) be a metrizable topological space with some
metric d. Let µ be a ﬁnite measure on (Ω, B(Ω)) and F be a closed
subset of Ω. For all n ≥ 1, we deﬁne φ
n
: Ω → R by:
∀x ∈ Ω , φ
n
(x)
= 1 −1 ∧ (nd(x, F))
1. Show that for all p ∈ [1, +∞], we have C
b
K
(Ω) ⊆ L
p
K
(Ω, B(Ω), µ).
2. Show that for all n ≥ 1, φ
n
∈ C
b
R
(Ω).
3. Show that φ
n
→ 1
F
.
4. Show that for all p ∈ [1, +∞[, we have:
lim
n→+∞
[φ
n
−1
F
[
p
dµ = 0
5. Show that for all p ∈ [1, +∞[ and > 0, there exists φ ∈ C
b
R
(Ω)
such that φ −1
F

p
≤ .
Tutorial 13: Regular Measure 10
6. Let ν ∈ M
1
(Ω, B(Ω)). Show that C
b
C
(Ω) ⊆ L
1
C
(Ω, B(Ω), ν) and:
ν(F) = lim
n→+∞
φ
n
dν
7. Prove the following:
Theorem 69 Let (Ω, T ) be a metrizable topological space and µ, ν
be two complex measures on (Ω, B(Ω)) such that:
∀φ ∈ C
b
R
(Ω) ,
φdµ =
φdν
Then µ = ν.
Exercise 8. Let (Ω, T ) be a metrizable topological space and µ be
a ﬁnite measure on (Ω, B(Ω)). Let s ∈ S
C
(Ω, B(Ω)) be a complex
simple function:
s =
n
i=1
α
i
1
A
i
Tutorial 13: Regular Measure 11
where n ≥ 1, α
i
∈ C, A
i
∈ B(Ω) for all i ∈ N
n
. Let p ∈ [1, +∞[.
1. Show that given > 0, for all i ∈ N
n
there is a closed subset F
i
of Ω such that F
i
⊆ A
i
and µ(A
i
¸ F
i
) ≤ . Let:
s
=
n
i=1
α
i
1
F
i
2. Show that:
s −s

p
≤
n
i=1
[α
i
[
1
p
3. Conclude that given > 0, there exists φ ∈ C
b
C
(Ω) such that:
φ −s
p
≤
4. Prove the following:
Tutorial 13: Regular Measure 12
Theorem 70 Let (Ω, T ) be a metrizable topological space and µ be
a ﬁnite measure on (Ω, B(Ω)). Then, for all p ∈ [1, +∞[, C
b
K
(Ω) is
dense in L
p
K
(Ω, B(Ω), µ).
Deﬁnition 101 A topological space (Ω, T ) is said to be σcompact
if and only if, there exists a sequence (K
n
)
n≥1
of compact subsets of
Ω such that K
n
↑ Ω.
Exercise 9. Let (Ω, T ) be a metrizable and σcompact topological
space, with metric d. Let Ω
be open in Ω. For all n ≥ 1, we deﬁne:
F
n
= ¦x ∈ Ω : d(x, (Ω
)
c
) ≥ 1/n¦
Let (K
n
)
n≥1
be a sequence of compact subsets of Ω such that K
n
↑ Ω.
1. Show that for all n ≥ 1, F
n
is closed in Ω.
2. Show that F
n
↑ Ω
.
Tutorial 13: Regular Measure 13
3. Show that F
n
∩ K
n
↑ Ω
.
4. Show that F
n
∩ K
n
is closed in K
n
for all n ≥ 1.
5. Show that F
n
∩ K
n
is a compact subset of Ω
for all n ≥ 1
6. Prove the following:
Theorem 71 Let (Ω, T ) be a metrizable and σcompact topological
space. Then, for all Ω
open subset of Ω, the induced topological space
(Ω
, T
Ω
) is itself metrizable and σcompact.
Deﬁnition 102 Let (Ω, T ) be a topological space and µ be a measure
on (Ω, B(Ω)). We say that µ is locally ﬁnite, if and only if, every
x ∈ Ω has an open neighborhood of ﬁnite µmeasure, i.e.
∀x ∈ Ω , ∃U ∈ T , x ∈ U , µ(U) < +∞
Tutorial 13: Regular Measure 14
Deﬁnition 103 Let µ be a measure on a topological space (Ω, T ).
We say that µ is innerregular, if and only if, for all B ∈ B(Ω):
µ(B) = sup¦µ(K) : K ⊆ B , K compact¦
We say that µ is outerregular, if and only if, for all B ∈ B(Ω):
µ(B) = inf¦µ(G) : B ⊆ G , G open¦
We say that µ is regular if it is both inner and outerregular.
Exercise 10. Let (Ω, T ) be a topological space and µ be a locally
ﬁnite measure on (Ω, B(Ω)). Let K be a compact subset of Ω.
1. Show the existence of open sets V
1
, . . . , V
n
with µ(V
i
) < +∞ for
all i ∈ N
n
and K ⊆ V
1
∪ . . . ∪ V
n
, where n ≥ 1.
2. Conclude that µ(K) < +∞.
Tutorial 13: Regular Measure 15
Exercise 11. Let (Ω, T ) be a metrizable and σcompact topological
space. Let µ be a ﬁnite measure on (Ω, B(Ω)). Let (K
n
)
n≥1
be a
sequence of compact subsets of Ω such that K
n
↑ Ω. Let B ∈ B(Ω).
We deﬁne α = sup¦µ(K) : K ⊆ B , K compact¦.
1. Show that given > 0, there exists F closed in Ω such that
F ⊆ B and µ(B ¸ F) ≤ .
2. Show that F ¸ (K
n
∩ F) ↓ ∅.
3. Show that K
n
∩ F is closed in K
n
.
4. Show that K
n
∩ F is compact.
5. Conclude that given > 0, there exists K compact subset of Ω
such that K ⊆ F and µ(F ¸ K) ≤ .
6. Show that µ(B) ≤ µ(K) + 2.
7. Show that µ(B) ≤ α and conclude that µ is innerregular.
Tutorial 13: Regular Measure 16
Exercise 12. Let (Ω, T ) be a metrizable and σcompact topological
space. Let µ be a locally ﬁnite measure on (Ω, B(Ω)). Let (K
n
)
n≥1
be
a sequence of compact subsets of Ω such that K
n
↑ Ω. Let B ∈ B(Ω),
and α ∈ R be such that α < µ(B).
1. Show that µ(K
n
∩ B) ↑ µ(B).
2. Show the existence of K ⊆ Ω compact, with α < µ(K ∩ B).
3. Let µ
K
= µ(K ∩ ). Show that µ
K
is a ﬁnite measure, and
conclude that µ
K
(B) = sup¦µ
K
(K
∗
) : K
∗
⊆ B , K
∗
compact¦.
4. Show the existence of a compact subset K
∗
of Ω, such that
K
∗
⊆ B and α < µ(K ∩ K
∗
).
5. Show that K
∗
is closed in Ω.
6. Show that K ∩ K
∗
is closed in K.
7. Show that K ∩ K
∗
is compact.
Tutorial 13: Regular Measure 17
8. Show that α < sup¦µ(K
) : K
⊆ B , K
compact¦.
9. Show that µ(B) ≤ sup¦µ(K
) : K
⊆ B , K
compact¦.
10. Conclude that µ is innerregular.
Exercise 13. Let (Ω, T ) be a metrizable topological space.
1. Show that (Ω, T ) is separable if and only if it has a countable
base.
2. Show that if (Ω, T ) is compact, for all n ≥ 1, Ω can be covered
by a ﬁnite number of open balls with radius 1/n.
3. Show that if (Ω, T ) is compact, then it is separable.
Tutorial 13: Regular Measure 18
Exercise 14. Let (Ω, T ) be a metrizable and σcompact topological
space with metric d. Let (K
n
)
n≥1
be a sequence of compact subsets
of Ω such that K
n
↑ Ω.
1. For all n ≥ 1, give a metric on K
n
inducing the topology T
K
n
.
2. Show that (K
n
, T
K
n
) is separable. Let (x
p
n
)
p≥1
be a countable
dense family of (K
n
, T
K
n
).
3. Show that (x
p
n
)
n,p≥1
is a countable dense family of (Ω, T ), and
conclude with the following:
Theorem 72 Let (Ω, T ) be a metrizable and σcompact topological
space. Then, (Ω, T ) is separable and has a countable base.
Exercise 15. Let (Ω, T ) be a metrizable and σcompact topological
space. Let µ be a locally ﬁnite measure on (Ω, B(Ω)). Let H be a
countable base of (Ω, T ). We deﬁne H
= ¦V ∈ H : µ(V ) < +∞¦.
Tutorial 13: Regular Measure 19
1. Show that for all U open in Ω and x ∈ U, there is U
x
open in
Ω such that x ∈ U
x
⊆ U and µ(U
x
) < +∞.
2. Show the existence of V
x
∈ H such that x ∈ V
x
⊆ U
x
.
3. Conclude that H
is a countable base of (Ω, T ).
4. Show the existence of a sequence (V
n
)
n≥1
of open sets in Ω with:
Ω =
+∞
n=1
V
n
, µ(V
n
) < +∞ , ∀n ≥ 1
Exercise 16. Let (Ω, T ) be a metrizable and σcompact topological
space. Let µ be a locally ﬁnite measure on (Ω, B(Ω)). Let (V
n
)
n≥1
a
sequence of open subsets of Ω such that:
Ω =
+∞
n=1
V
n
, µ(V
n
) < +∞ , ∀n ≥ 1
Let B ∈ B(Ω) and α = inf¦µ(G) : B ⊆ G , G open¦.
Tutorial 13: Regular Measure 20
1. Given > 0, show that there exists G
n
open in Ω such that
B ⊆ G
n
and µ
V
n
(G
n
¸ B) ≤ /2
n
, where µ
V
n
= µ(V
n
∩ ).
2. Let G = ∪
+∞
n=1
(V
n
∩G
n
). Show that G is open in Ω, and B ⊆ G.
3. Show that G¸ B = ∪
+∞
n=1
V
n
∩ (G
n
¸ B).
4. Show that µ(G) ≤ µ(B) +.
5. Show that α ≤ µ(B).
6. Conclude that is µ outerregular.
7. Show the following:
Theorem 73 Let µ be a locally ﬁnite measure on a metrizable and
σcompact topological space (Ω, T ). Then, µ is regular, i.e.:
µ(B) = sup¦µ(K) : K ⊆ B , K compact¦
= inf¦µ(G) : B ⊆ G , G open¦
for all B ∈ B(Ω).
Tutorial 13: Regular Measure 21
Exercise 17. Show the following:
Theorem 74 Let Ω be an open subset of R
n
, where n ≥ 1. Any
locally ﬁnite measure on (Ω, B(Ω)) is regular.
Deﬁnition 104 We call strongly σcompact topological space, a
topological space (Ω, T ), for which there exists a sequence (V
n
)
n≥1
of
open sets with compact closure, such that V
n
↑ Ω.
Deﬁnition 105 We call locally compact topological space, a topo
logical space (Ω, T ), for which every x ∈ Ω has an open neighborhood
with compact closure, i.e. such that:
∀x ∈ Ω , ∃U ∈ T : x ∈ U ,
¯
U is compact
Tutorial 13: Regular Measure 22
Exercise 18. Let (Ω, T ) be a σcompact and locally compact topo
logical space. Let (K
n
)
n≥1
be a sequence of compact subsets of Ω
such that K
n
↑ Ω.
1. Show that for all n ≥ 1, there are open sets V
n
1
, . . . , V
n
p
n
, p
n
≥ 1,
such that K
n
⊆ V
n
1
∪ . . . ∪ V
n
p
n
and
¯
V
n
1
, . . . ,
¯
V
n
p
n
are compact
subsets of Ω.
2. Deﬁne W
n
= V
n
1
∪. . . ∪V
n
p
n
and V
n
= ∪
n
k=1
W
k
, for n ≥ 1. Show
that (V
n
)
n≥1
is a sequence of open sets in Ω with V
n
↑ Ω.
3. Show that
¯
W
n
=
¯
V
n
1
∪ . . . ∪
¯
V
n
p
n
for all n ≥ 1.
4. Show that
¯
W
n
is compact for all n ≥ 1.
5. Show that
¯
V
n
is compact for all n ≥ 1
6. Conclude with the following:
Tutorial 13: Regular Measure 23
Theorem 75 A topological space (Ω, T ) is strongly σcompact, if
and only if it is σcompact and locally compact.
Exercise 19. Let (Ω, T ) be a topological space and Ω
be an open
subset of Ω. Let A ⊆ Ω
. We denote
¯
A
Ω
the closure of A in the
induced topological space (Ω
, T
Ω
), and
¯
A its closure in Ω.
1. Show that A ⊆ Ω
∩
¯
A.
2. Show that Ω
∩
¯
A is closed in Ω
.
3. Show that
¯
A
Ω
⊆ Ω
∩
¯
A.
4. Let x ∈ Ω
∩
¯
A. Show that if x ∈ U
∈ T
Ω
, then A ∩ U
,= ∅.
5. Show that
¯
A
Ω
= Ω
∩
¯
A.
Tutorial 13: Regular Measure 24
Exercise 20. Let (Ω, d) be a metric space.
1. Show that for all x ∈ Ω and > 0, we have:
B(x, ) ⊆ ¦y ∈ Ω : d(x, y) ≤ ¦
2. Take Ω = [0, 1/2[∪¦1¦. Show that B(0, 1) = [0, 1/2[.
3. Show that [0, 1/2[ is closed in Ω.
4. Show that B(0, 1) = [0, 1/2[.
5. Conclude that B(0, 1) ,= ¦y ∈ Ω : [y[ ≤ 1¦ = Ω.
Exercise 21. Let (Ω, d) be a locally compact metric space. Let Ω
be an open subset of Ω. Let x ∈ Ω
.
1. Show there exists U open with compact closure, such that x ∈ U.
2. Show the existence of > 0 such that B(x, ) ⊆ U ∩ Ω
.
Tutorial 13: Regular Measure 25
3. Show that B(x, /2) ⊆
¯
U.
4. Show that B(x, /2) is closed in
¯
U.
5. Show that B(x, /2) is a compact subset of Ω.
6. Show that B(x, /2) ⊆ Ω
.
7. Let U
= B(x, /2) ∩ Ω
= B(x, /2). Show x ∈ U
∈ T
Ω
, and:
¯
U
Ω
= B(x, /2)
8. Show that the induced topological space Ω
is locally compact.
9. Prove the following:
Theorem 76 Let (Ω, T ) be a metrizable and strongly σcompact
topological space. Then, for all Ω
open subset of Ω, the induced topo
logical space (Ω
, T
Ω
) is itself metrizable and strongly σcompact.
Tutorial 13: Regular Measure 26
Deﬁnition 106 Let (Ω, T ) be a topological space, and φ : Ω → C be
a map. We call support of φ, the closure of the set ¦φ ,= 0¦, i.e.:
supp(φ)
= ¦ω ∈ Ω : φ(ω) ,= 0¦
Deﬁnition 107 Let (Ω, T ) be a topological space. We denote C
c
K
(Ω)
the Kvector space of all continuous map with compact support
φ : Ω → K, where K = R or K = C.
Exercise 22. Let (Ω, T ) be a topological space.
1. Show that 0 ∈ C
c
K
(Ω).
2. Show that C
c
K
(Ω) is indeed a Kvector space.
3. Show that C
c
K
(Ω) ⊆ C
b
K
(Ω).
Exercise 23. let (Ω, d) be a locally compact metric space. let K be
a compact subset of Ω, and G be open in Ω, with K ⊆ G.
Tutorial 13: Regular Measure 27
1. Show the existence of open sets V
1
, . . . , V
n
such that:
K ⊆ V
1
∪ . . . ∪ V
n
and
¯
V
1
, . . . ,
¯
V
n
are compact subsets of Ω.
2. Show that V = (V
1
∪. . . ∪V
n
)∩G is open in Ω, and K ⊆ V ⊆ G.
3. Show that
¯
V ⊆
¯
V
1
∪ . . . ∪
¯
V
n
.
4. Show that
¯
V is compact.
5. We assume K ,= ∅ and V ,= Ω, and we deﬁne φ : Ω → R by:
∀x ∈ Ω , φ(x)
=
d(x, V
c
)
d(x, V
c
) +d(x, K)
6. Show that φ is welldeﬁned and continuous.
7. Show that ¦φ ,= 0¦ = V .
8. Show that φ ∈ C
c
R
(Ω).
Tutorial 13: Regular Measure 28
9. Show that 1
K
≤ φ ≤ 1
G
.
10. Show that if K = ∅, there is φ ∈ C
c
R
(Ω) with 1
K
≤ φ ≤ 1
G
.
11. Show that if V = Ω then Ω is compact.
12. Show that if V = Ω, there φ ∈ C
c
R
(Ω) with 1
K
≤ φ ≤ 1
G
.
Theorem 77 Let (Ω, T ) be a metrizable and locally compact topolog
ical space. Let K be a compact subset of Ω, and G be an open subset
of Ω such that K ⊆ G. Then, there exists φ ∈ C
c
R
(Ω), realvalued
continuous map with compact support, such that:
1
K
≤ φ ≤ 1
G
Exercise 24. Let (Ω, T ) be a metrizable and strongly σcompact
topological space. Let µ be a locally ﬁnite measure on (Ω, B(Ω)). Let
B ∈ B(Ω) be such that µ(B) < +∞. Let p ∈ [1, +∞[.
1. Show that C
c
K
(Ω) ⊆ L
p
K
(Ω, B(Ω), µ).
Tutorial 13: Regular Measure 29
2. Let > 0. Show the existence of K compact and G open, with:
K ⊆ B ⊆ G , µ(G¸ K) ≤
3. Where did you use the fact that µ(B) < +∞?
4. Show the existence of φ ∈ C
c
R
(Ω) with 1
K
≤ φ ≤ 1
G
.
5. Show that:
[φ −1
B
[
p
dµ ≤ µ(G¸ K)
6. Conclude that for all > 0, there exists φ ∈ C
c
R
(Ω) such that:
φ −1
B

p
≤
7. Let s ∈ S
C
(Ω, B(Ω)) ∩ L
p
C
(Ω, B(Ω), µ). Show that for all > 0,
there exists φ ∈ C
c
C
(Ω) such that φ −s
p
≤ .
8. Prove the following:
Tutorial 13: Regular Measure 30
Theorem 78 Let (Ω, T ) be a metrizable and strongly σcompact
topological space
1
. Let µ be a locally ﬁnite measure on (Ω, B(Ω)).
Then, for all p ∈ [1, +∞[, the space C
c
K
(Ω) of Kvalued, continuous
maps with compact support, is dense in L
p
K
(Ω, B(Ω), µ).
Exercise 25. Prove the following:
Theorem 79 Let Ω be an open subset of R
n
, where n ≥ 1. Then,
for any locally ﬁnite measure µ on (Ω, B(Ω)) and p ∈ [1, +∞[, C
c
K
(Ω)
is dense in L
p
K
(Ω, B(Ω), µ).
1
i.e. a metrizable topological space for which there exists a sequence (V
n
)
n≥1
of open sets with compact closure, such that V
n
↑ Ω.
Tutorial 14: Maps of Finite Variation 1
14. Maps of Finite Variation
Deﬁnition 108 We call total variation of a map b : R
+
→ C the
map b : R
+
→ [0, +∞] deﬁned as:
∀t ∈ R
+
, b(t)
= b(0) + sup
n
i=1
b(t
i
) −b(t
i−1
)
where the ’sup’ is taken over all ﬁnite t
0
≤ . . . ≤ t
n
in [0, t], n ≥ 1.
We say that b is of ﬁnite variation, if and only if:
∀t ∈ R
+
, b(t) < +∞
We say that b is of bounded variation, if and only if:
sup
t∈R
+
b(t) < +∞
Warning: The notation b can be misleading: it can refer to the map
t → b(t)(modulus), or to the map t → b(t) (total variation).
Tutorial 14: Maps of Finite Variation 2
Exercise 1. Let a : R
+
→ R
+
be nondecreasing with a(0) ≥ 0.
1. Show that a = a and a is of ﬁnite variation.
2. Show that the limit a(∞) = lim
t→+∞
a(t) exists in
¯
R.
3. Show that a is of bounded variation if and only if a(∞) < +∞.
Exercise 2. Let b = b
1
+ ib
2
: R
+
→ C be a map.
1. Show that b
1
 ≤ b and b
2
 ≤ b.
2. Show that b ≤ b
1
 +b
2
.
3. Show that b is of ﬁnite variation if and only if b
1
, b
2
are.
4. Show that b is of bounded variation if and only if b
1
, b
2
are.
5. Show that b(0) = b(0).
Tutorial 14: Maps of Finite Variation 3
Exercise 3. Let b : R
+
→ R be continuous and diﬀerentiable, such
that b
is bounded on each compact interval. Show that b is of ﬁnite
variation.
Exercise 4. Show that if b : R
+
→ C is of class C
1
, i.e. continu
ous and diﬀerentiable with continuous derivative, then b is of ﬁnite
variation.
Exercise 5. Let f : (R
+
, B(R
+
)) → (C, B(C)) be a measurable map,
with
t
0
f(s)ds < +∞ for all t ∈ R
+
. Let b : R
+
→ C deﬁned by:
∀t ∈ R
+
, b(t)
=
R
+
f1
[0,t]
ds
1. Show that b is of ﬁnite variation and:
∀t ∈ R
+
, b(t) ≤
t
0
f(s)ds
2. Show that f ∈ L
1
C
(R
+
, B(R
+
), ds) ⇒b is of bounded variation.
Tutorial 14: Maps of Finite Variation 4
Exercise 6. Show that if b, b
: R
+
→ C are maps of ﬁnite variation,
and α ∈ C, then b + αb
is also a map of ﬁnite variation. Prove the
same result when the word ’ﬁnite’ is replaced by ’bounded’.
Exercise 7. Let b : R
+
→ C be a map. For all t ∈ R
+
, let S(t)
be the set of all ﬁnite subsets A of [0, t], with cardA ≥ 2. For all
A ∈ S(t), we deﬁne:
S(A)
=
n
i=1
b(t
i
) −b(t
i−1
)
where it is understood that t
0
, . . . , t
n
are such that:
t
0
< t
1
< . . . < t
n
and A = {t
0
, . . . , t
n
} ⊆ [0, t]
1. Show that for all t ∈ R
+
, if s
0
≤ . . . ≤ s
p
(p ≥ 1) is a ﬁnite
sequence in [0, t], then if:
S
=
p
j=1
b(s
j
) −b(s
j−1
)
Tutorial 14: Maps of Finite Variation 5
either S = 0 or S = S(A) for some A ∈ S(t).
2. Conclude that:
∀t ∈ R
+
, b(t) = b(0) + sup{S(A) : A ∈ S(t)}
3. Let A ∈ S(t) and s ∈ [0, t]. Show that S(A) ≤ S(A ∪ {s}).
4. Let A, B ∈ S(t). Show that:
A ⊆ B ⇒ S(A) ≤ S(B)
5. Show that if t
0
≤ . . . ≤ t
n
, n ≥ 1, and s
0
≤ . . . ≤ s
p
, p ≥ 1, are
ﬁnite sequence in [0, t] such that:
{t
0
, . . . , t
n
} ⊆ {s
0
, . . . , s
p
}
then:
n
i=1
b(t
i
) −b(t
i−1
) ≤
p
j=1
b(s
j
) −b(s
j−1
)
Tutorial 14: Maps of Finite Variation 6
Exercise 8. Let b : R
+
→ C be of ﬁnite variation. Let s, t ∈ R
+
,
with s ≤ t. We deﬁne:
δ
= sup
n
i=1
b(t
i
) −b(t
i−1
)
where the ’sup’ is taken over all ﬁnite t
0
≤ . . . ≤ t
n
, n ≥ 1, in [s, t].
1. let s
0
≤ . . . ≤ s
p
and t
0
≤ . . . ≤ t
n
be ﬁnite sequences in [0, s]
and [s, t] respectively, where n, p ≥ 1. Show that:
p
j=1
b(s
j
) −b(s
j−1
) +b(t
0
) −b(s
p
) +
n
i=1
b(t
i
) −b(t
i−1
)
is less or equal than b(t) −b(0).
2. Show that δ ≤ b(t) −b(s).
3. Let t
0
≤ . . . ≤ t
n
be a ﬁnite sequence in [0, t], where n ≥ 1, and
Tutorial 14: Maps of Finite Variation 7
suppose that s = t
j
for some 0 < j < n. Show that:
n
i=1
b(t
i
) −b(t
i−1
) ≤ b(s) − b(0) + δ (1)
4. Show that inequality (1) holds, for all t
0
≤ . . . ≤ t
n
in [0, t].
5. Prove the following:
Theorem 80 Let b : R
+
→ C be a map of ﬁnite variation. Then,
for all s, t ∈ R
+
, s ≤ t, we have:
b(t) −b(s) = sup
n
i=1
b(t
i
) −b(t
i−1
)
where the ’sup’ is taken over all ﬁnite t
0
≤ . . . ≤ t
n
, n ≥ 1, in [s, t].
Exercise 9. Let b : R
+
→ C be a map of ﬁnite variation. Show that
b is nondecreasing with b(0) ≥ 0, and b = b.
Tutorial 14: Maps of Finite Variation 8
Deﬁnition 109 Let b : R
+
→ R be a map of ﬁnite variation. Let:
b
+
=
1
2
(b + b)
b
−
=
1
2
(b −b)
b
+
, b
−
are respectively the positive, negative variation of b.
Exercise 10. Let b : R
+
→ R be a map of ﬁnite variation.
1. Show that b = b
+
+b
−
and b = b
+
−b
−
.
2. Show that b
+
(0) = b
+
(0) ≥ 0 and b
−
(0) = b
−
(0) ≥ 0.
3. Show that for all s, t ∈ R
+
, s ≤ t, we have:
b(t) −b(s) ≤ b(t) −b(s)
4. Show that b
+
and b
−
are nondecreasing.
Tutorial 14: Maps of Finite Variation 9
Exercise 11. Let b : R
+
→ C be of ﬁnite variation. Show the
existence of b
1
, b
2
, b
3
, b
4
: R
+
→ R
+
, nondecreasing with b
i
(0) ≥ 0,
such that b = b
1
−b
2
+i(b
3
−b
4
). Show conversely that if b : R
+
→ C
is a map with such decomposition, then it is of ﬁnite variation.
Exercise 12. Let b : R
+
→ C be a rightcontinuous map of ﬁnite
variation, and x
0
∈ R
+
.
1. Show that b(x
0
+) = lim
t↓↓x
0
b(t) = inf
x
0
<t
b(t) ∈ R.
2. Show that b(x
0
) ≤ b(x
0
+).
3. Given > 0, show the existence of y
0
∈ R
+
, x
0
< y
0
, such that:
u ∈]x
0
, y
0
] ⇒ b(u) −b(x
0
) ≤ /2
u ∈]x
0
, y
0
] ⇒ b(y
0
) −b(u) ≤ /2
Tutorial 14: Maps of Finite Variation 10
Exercise 13. Further to exercise (12), let t
0
≤ . . . ≤ t
n
, n ≥ 1, be
a ﬁnite sequence in [0, y
0
], such that x
0
= t
j
for some 0 < j < n −1.
We choose j to be the maximum index satisfying this condition, so
that x
0
< t
j+1
≤ y
0
.
1. Show that
j
i=1
b(t
i
) −b(t
i−1
) ≤ b(x
0
) −b(0).
2. Show that b(t
j+1
) −b(t
j
) ≤ /2.
3. Show that
n
i=j+2
b(t
i
) −b(t
i−1
) ≤ b(y
0
) −b(t
j+1
) ≤ /2.
4. Show that for all ﬁnite sequence t
0
≤ . . . ≤ t
n
, n ≥ 1, in [0, y
0
]:
n
i=1
b(t
i
) −b(t
i−1
) ≤ b(x
0
) −b(0) +
5. Show that b(y
0
) ≤ b(x
0
) + .
6. Show that b(x
0
+) ≤ b(x
0
) and that b is rightcontinuous.
Tutorial 14: Maps of Finite Variation 11
Exercise 14. Let b : R
+
→ C be a leftcontinuous map of ﬁnite
variation, and let x
0
∈ R
+
\ {0}.
1. Show that b(x
0
−) = lim
t↑↑x
0
b(t) = sup
t<x
0
b(t) ∈ R.
2. Show that b(x
0
−) ≤ b(x
0
).
3. Given > 0, show the existence of y
0
∈]0, x
0
[, such that:
u ∈ [y
0
, x
0
[ ⇒ b(x
0
) −b(u) ≤ /2
u ∈ [y
0
, x
0
[ ⇒ b(u) −b(y
0
) ≤ /2
Exercise 15. Further to exercise (14), let t
0
≤ . . . ≤ t
n
, n ≥ 1, be
a ﬁnite sequence in [0, x
0
], such that y
0
= t
j
for some 0 < j < n −1,
and x
0
= t
n
. We denote k = max{i : j ≤ i , t
i
< x
0
}.
1. Show that j ≤ k ≤ n −1 and t
k
∈ [y
0
, x
0
[.
2. Show that
j
i=1
b(t
i
) −b(t
i−1
) ≤ b(y
0
) −b(0).
Tutorial 14: Maps of Finite Variation 12
3. Show that
k
i=j+1
b(t
i
) − b(t
i−1
) ≤ b(t
k
) − b(y
0
) ≤ /2,
where if j = k, the corresponding sum is zero.
4. Show that
n
i=k+1
b(t
i
) −b(t
i−1
) = b(x
0
) −b(t
k
) ≤ /2.
5. Show that for all ﬁnite sequence t
0
≤ . . . ≤ t
n
, n ≥ 1, in [0, x
0
]:
n
i=1
b(t
i
) −b(t
i−1
) ≤ b(y
0
) −b(0) +
6. Show that b(x
0
) ≤ b(y
0
) + .
7. Show that b(x
0
) ≤ b(x
0
−) and that b is leftcontinuous.
8. Prove the following:
Theorem 81 Let b : R
+
→ C be a map of ﬁnite variation. Then:
b is rightcontinuous ⇒ b is rightcontinuous
b is leftcontinuous ⇒ b is leftcontinuous
b is continuous ⇒ b is continuous
Tutorial 14: Maps of Finite Variation 13
Exercise 16. Let b : R
+
→ R be an Rvalued map of ﬁnite variation.
1. Show that if b is rightcontinuous, then so are b
+
and b
−
.
2. State and prove similar results for leftcontinuity and continuity.
Exercise 17. Let b : R
+
→ C be a rightcontinuous map of ﬁnite
variation. Show the existence of b
1
, b
2
, b
3
, b
4
: R
+
→ R
+
, right
continuous and nondecreasing maps with b
i
(0) ≥ 0, such that:
b = b
1
−b
2
+ i(b
3
−b
4
)
Exercise 18. Let b : R
+
→ C be a rightcontinuous map. Let
t ∈ R
+
. For all p ≥ 1, we deﬁne:
S
p
= b(0) +
2
p
k=1
b(kt/2
p
) −b((k −1)t/2
p
)
1. Show that for all p ≥ 1, S
p
≤ S
p+1
and deﬁne S = sup
p≥1
S
p
.
Tutorial 14: Maps of Finite Variation 14
2. Show that S ≤ b(t).
Exercise 19. Further to exercise (18), let t
0
< . . . < t
n
be a ﬁnite
sequence of distinct elements of [0, t]. Let > 0.
1. Show that for all i = 0, . . . , n, there exists p
i
≥ 1 and
q
i
∈ {0, 1, . . . , 2
p
i
} such that:
0 ≤ t
0
≤
q
0
t
2
p
0
< t
1
≤
q
1
t
2
p
1
< . . . < t
n
≤
q
n
t
2
p
n
≤ t
and:
b(t
i
) −b(q
i
t/2
p
i
) ≤ , ∀i = 0, . . . , n
2. Show the existence of p ≥ 1, and k
0
, . . . , k
n
∈ {0, . . . , 2
p
} with:
0 ≤ t
0
≤
k
0
t
2
p
< t
1
≤
k
1
t
2
p
< . . . < t
n
≤
k
n
t
2
p
≤ t
and:
b(t
i
) −b(k
i
t/2
p
) ≤ , ∀i = 0, . . . , n
Tutorial 14: Maps of Finite Variation 15
3. Show that:
n
i=1
b(k
i
t/2
p
) −b(k
i−1
t/2
p
) ≤ S
p
−b(0)
4. Show that:
n
i=1
b(t
i
) −b(t
i−1
) ≤ S −b(0) + 2n
5. Show that:
n
i=1
b(t
i
) −b(t
i−1
) ≤ S −b(0)
6. Conclude that b(t) ≤ S.
7. Prove the following:
Tutorial 14: Maps of Finite Variation 16
Theorem 82 Let b : R
+
→ C be rightcontinuous or leftcontinuous.
Then, for all t ∈ R
+
:
b(t) = b(0) + lim
n→+∞
2
n
k=1
b(kt/2
n
) −b((k −1)t/2
n
)
Exercise 20. Let b : R
+
→ R
+
be deﬁned by b = 1
Q
+. Show that:
+∞ = b(1) = lim
n→+∞
2
n
k=1
b(k/2
n
) −b((k −1)/2
n
) = 0
Exercise 21. b : R
+
→ C is rightcontinuous of bounded variation.
1. Let b = b
1
+ ib
2
. Explain why db
1

+
, db
1

−
, db
2

+
and db
2

−
are all welldeﬁned measures on (R
+
, B(R
+
)).
2. Is this still true, if b is rightcontinuous of ﬁnite variation?
3. Show that db
1

+
, db
1

−
, db
2

+
and db
2

−
are ﬁnite measures.
Tutorial 14: Maps of Finite Variation 17
4. Let db = db
1

+
−db
1

−
+ i(db
2

+
−db
2

−
). Show that db is a
welldeﬁned complex measure on (R
+
, B(R
+
)).
5. Show that db({0}) = b(0).
6. Show that for all s, t ∈ R
+
, s ≤ t, db(]s, t]) = b(t) −b(s).
7. Show that lim
t→+∞
b(t) exists in C. We denote b(∞) this limit.
8. Show that db(R
+
) = b(∞).
9. Proving the uniqueness of db, justify the following:
Deﬁnition 110 Let b : R
+
→ C be a rightcontinuous map of
bounded variation. There exists a unique complex measure db on
(R
+
, B(R
+
)), such that:
(i) db({0}) = b(0)
(ii) ∀s, t ∈ R
+
s ≤ t , db(]s, t]) = b(t) −b(s)
db is called the complex stieltjes measure associated with b.
Tutorial 14: Maps of Finite Variation 18
Exercise 22. Show that if a : R
+
→ R
+
is rightcontinuous, non
decreasing with a(0) ≥ 0 and a(∞) < +∞, then deﬁnition (110) of
da coincide with the already known deﬁnition (24).
Exercise 23. b : R
+
→ C is rightcontinuous of ﬁnite variation.
1. Let b = b
1
+ ib
2
. Explain why db
1

+
, db
1

−
, db
2

+
and db
2

−
are all welldeﬁned measures on (R
+
, B(R
+
)).
2. Why is it in general impossible to deﬁne:
db
= db
1

+
−db
1

−
+ i(db
2

+
−db
2

−
)
Warning: it does not make sense to write something like ’db’, unless
b is either rightcontinuous, nondecreasing and b(0) ≥ 0, or b is a
rightcontinuous map of bounded variation.
Exercise 24. Let b : R
+
→ C be a map. For all T ∈ R
+
, we deﬁne
b
T
: R
+
→ C as b
T
(t) = b(T ∧ t) for all t ∈ R
+
.
1. Show that for all T ∈ R
+
, b
T
 = b
T
.
Tutorial 14: Maps of Finite Variation 19
2. Show that if b is of ﬁnite variation, then for all T ∈ R
+
, b
T
is
of bounded variation, and we have b
T
(∞) = b(T) < +∞.
3. Show that if b is rightcontinuous and of ﬁnite variation, for all
T ∈ R
+
, db
T
is the unique complex measure on R
+
, with:
(i) db
T
({0}) = b(0)
(ii) ∀s, t ∈ R
+
, s ≤ t , db
T
(]s, t]) = b(T ∧ t) −b(T ∧ s)
4. Show that if b is Rvalued and of ﬁnite variation, for all T ∈ R
+
,
we have b
T

+
= (b
+
)
T
and b
T

−
= (b
−
)
T
.
5. Show that if b is rightcontinuous and of bounded variation, for
all T ∈ R
+
, we have db
T
= db
[0,T]
= db([0, T] ∩ · )
6. Show that if b is rightcontinuous, nondecreasing with b(0) ≥ 0,
for all T ∈ R
+
, we have db
T
= db
[0,T]
= db([0, T] ∩ · )
Tutorial 14: Maps of Finite Variation 20
Exercise 25. Let µ, ν be two ﬁnite measures on R
+
, such that:
(i) µ({0}) ≤ ν({0})
(ii) ∀s, t ∈ R
+
, s ≤ t , µ(]s, t]) ≤ ν(]s, t])
We deﬁne a, c : R
+
→ R
+
by a(t) = µ([0, t]) and c(t) = ν([0, t]).
1. Show that a and c are rightcontinuous, nondecreasing with
a(0) ≥ 0 and c(0) ≥ 0.
2. Show that da = µ and dc = ν.
3. Show that a ≤ c.
4. Deﬁne b : R
+
→ R
+
by b = c − a. Show that b is right
continuous, nondecreasing with b(0) ≥ 0.
5. Show that da + db = dc.
6. Conclude with the following:
Tutorial 14: Maps of Finite Variation 21
Theorem 83 Let µ, ν be two ﬁnite measures on (R
+
, B(R
+
)) with:
(i) µ({0}) ≤ ν({0})
(ii) ∀s, t ∈ R
+
, s ≤ t , µ(]s, t]) ≤ ν(]s, t])
Then µ ≤ ν, i.e. for all B ∈ B(R
+
), µ(B) ≤ ν(B).
Exercise 26. b : R
+
→ C is rightcontinuous of bounded variation.
1. Show that db({0}) = b(0) = db({0}). Let s, t ∈ R
+
, s ≤ t.
2. Let t
0
≤ . . . ≤ t
n
be a ﬁnite sequence in [s, t]. Show:
n
i=1
b(t
i
) −b(t
i−1
) ≤ db(]s, t])
3. Show that b(t) −b(s) ≤ db(]s, t]).
4. Show that db ≤ db.
Tutorial 14: Maps of Finite Variation 22
5. Show that L
1
C
(R
+
, B(R
+
), db) ⊆ L
1
C
(R
+
, B(R
+
), db).
6. Show that R
+
is metrizable and strongly σcompact.
7. Show that C
c
C
(R
+
), C
b
C
(R
+
) are dense in L
1
C
(R
+
, B(R
+
), db).
8. Let h ∈ L
1
C
(R
+
, B(R
+
), db). Given > 0, show the existence
of φ ∈ C
b
C
(R
+
) such that
φ −hdb ≤ .
9. Show that 
hdb ≤ 
φdb + .
10. Show that:
φdb −
hdb
≤
φ −hdb ≤
φ −hdb
11. Show that
φdb ≤
hdb + .
12. For all n ≥ 1, we deﬁne:
φ
n
= φ(0)1
{0}
+
n2
n
−1
k=0
φ(k/2
n
)1
]k/2
n
,(k+1)/2
n
]
Tutorial 14: Maps of Finite Variation 23
Show there is M ∈ R
+
, such that φ
n
(x) ≤ M for all x and n.
13. Using the continuity of φ, show that φ
n
→ φ.
14. Show that lim
φ
n
db =
φdb.
15. Show that lim
φ
n
db =
φdb.
16. Show that for all n ≥ 1:
φ
n
db = φ(0)b(0) +
n2
n
−1
k=0
φ(k/2
n
)(b((k + 1)/2
n
) −b(k/2
n
))
17. Show that 
φ
n
db ≤
φ
n
db for all n ≥ 1.
18. Show that 
φdb ≤
φdb.
19. Show that 
hdb ≤
hdb + 2.
20. Show that 
hdb ≤
hdb for all h ∈ L
1
C
(R
+
, B(R
+
), db).
Tutorial 14: Maps of Finite Variation 24
21. Let B ∈ B(R
+
) and h ∈ L
1
C
(R
+
, B(R
+
), db) be such that
h = 1 and db =
hdb. Show that db(B) =
B
¯
hdb.
22. Conclude that db ≤ db.
Exercise 27. b : R
+
→ C is rightcontinuous of ﬁnite variation.
1. Show that for all T ∈ R
+
, db
T
 = db
T
 and db
T
 = db
T
.
2. Show that db
T
= db
[0,T]
= db([0, T] ∩ · ), and conclude:
Theorem 84 If b : R
+
→ C is rightcontinuous of bounded varia
tion, the total variation of its associated complex stieltjes measure, is
equal to the stieltjes measure associated with its total variation, i.e.
db = db
If b : R
+
→ C is rightcontinuous of ﬁnite variation, then for all
T ∈ R
+
, b
T
deﬁned by b
T
(t) = b(T ∧t), is rightcontinuous of bounded
variation, and we have db
T
 = db([0, T] ∩ · ).
Tutorial 14: Maps of Finite Variation 25
Deﬁnition 111 Let b : R
+
→ E be a map, where E is a topological
space. We say that b is cadlag with respect to E, if and only if b is
rightcontinuous, and the limit:
b(t−) = lim
s↑↑t
b(s)
exists in E, for all t ∈ R
+
\ {0}. In the case when E = C, given b
cadlag, we deﬁne b(0−) = 0, and for all t ∈ R
+
:
∆b(t)
= b(t) −b(t−)
Exercise 28. Let b : R
+
→ E be cadlag, where E is a topological
space. Suppose b has values in E
⊆ E.
1. Explain why b may not be cadlag with respect to E
.
2. Show that b is cadlag with respect to
¯
E
.
3. Show that b : R
+
→ R is cadlag ⇔ it is cadlag w.r. to C.
Tutorial 14: Maps of Finite Variation 26
Exercise 29.
1. Show that if b : R
+
→ C is cadlag, then b is continuous with
b(0) = 0 if and only if ∆b(t) = 0 for all t ∈ R
+
.
2. Show that if a : R
+
→ R
+
is rightcontinuous, nondecreasing
with a(0) ≥ 0, then a is cadlag (w.r. to R) with ∆a ≥ 0.
3. Show that any linear combination of cadlag maps is itself cadlag.
4. Show that if b : R
+
→ C is a rightcontinuous map of ﬁnite
variation, then b is cadlag.
5. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing with
a(0) ≥ 0. Show that da({t}) = ∆a(t) for all t ∈ R
+
.
6. Let b : R
+
→ C be a rightcontinuous map of bounded varia
tion. Show that db({t}) = ∆b(t) for all t ∈ R
+
.
Tutorial 14: Maps of Finite Variation 27
7. Let b : R
+
→ C be a rightcontinuous map of ﬁnite variation.
Let T ∈ R
+
. Show that:
∀t ∈ R
+
, b
T
(t−) =
b(t−) if t ≤ T
b(T) if T < t
Show that ∆b
T
= (∆b)1
[0,T]
, and db
T
({t}) = ∆b(t)1
[0,T]
(t).
Exercise 30. Let b : R
+
→ C be a cadlag map and T ∈ R
+
.
1. Show that if t → b(t−) is not bounded on [0, T], there exists a
sequence (t
n
)
n≥1
in [0, T] such that b(t
n
) → +∞.
2. Suppose from now on that b is not bounded on [0, T]. Show the
existence of a sequence (t
n
)
n≥1
in [0, T], such that t
n
→ t for
some t ∈ [0, T], and b(t
n
) → +∞.
3. Deﬁne R = {n ≥ 1 : t ≤ t
n
} and L = {n ≥ 1 : t
n
< t}. Show
that R and L cannot be both ﬁnite.
Tutorial 14: Maps of Finite Variation 28
4. Suppose that R is inﬁnite. Show the existence of n
1
≥ 1, with:
t
n
1
∈ [t, t + 1[∩[0, T]
5. If R is inﬁnite, show there is n
1
< n
2
< . . . such that:
t
n
k
∈ [t, t +
1
k
[∩[0, T] , ∀k ≥ 1
6. Show that b(t
n
k
) → +∞.
7. Show that if L is inﬁnite, then t > 0 and there is an increasing
sequence n
1
< n
2
< . . ., such that:
t
n
k
∈]t −
1
k
, t[∩[0, T] , ∀k ≥ 1
8. Show that: b(t
n
k
) → +∞.
9. Prove the following:
Tutorial 14: Maps of Finite Variation 29
Theorem 85 Let b : R
+
→ C be a cadlag map. Let T ∈ R
+
. Then
b and t → b(t−) are bounded on [0, T], i.e. there exists M ∈ R
+
such
that:
b(t) ∨ b(t−) ≤ M , ∀t ∈ [0, T]
Tutorial 15: Stieltjes Integration 1
15. Stieltjes Integration
Deﬁnition 112 b : R
+
→ C is rightcontinuous of ﬁnite variation.
The stieltjes L
1
spaces associated with b are deﬁned as:
L
1
C
(b)
=
_
f : R
+
→ C measurable,
_
fdb < +∞
_
L
1,loc
C
(b)
=
_
f : R
+
→ C measurable,
_
t
0
fdb < +∞, ∀t ∈ R
+
_
Warning : In these tutorials,
_
t
0
. . . refers to
_
[0,t]
. . ., i.e. the domain
of integration is always [0, t], not ]0, t], [0, t[, or ]0, t[.
Exercise 1. b : R
+
→ C is rightcontinuous of ﬁnite variation.
1. Propose a deﬁnition for L
1
R
(b) and L
1,loc
R
(b).
2. Is L
1
C
(b) the same thing as L
1
C
(R
+
, B(R
+
), db)?
3. Is it meaningful to speak of L
1
C
(R
+
, B(R
+
), db)?
Tutorial 15: Stieltjes Integration 2
4. Show that L
1
C
(b) = L
1
C
(b) and L
1,loc
C
(b) = L
1,loc
C
(b).
5. Show that L
1
C
(b) ⊆ L
1,loc
C
(b).
Exercise 2. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. For all f ∈ L
1,loc
C
(a), we deﬁne f.a : R
+
→ C as:
f.a(t)
=
_
t
0
fda , ∀t ∈ R
+
1. Explain why f.a : R
+
→ C is a welldeﬁned map.
2. Let t ∈ R
+
, (t
n
)
n≥1
be a sequence in R
+
with t
n
↓↓ t. Show:
lim
n→+∞
_
f1
[0,t
n
]
da =
_
f1
[0,t]
da
3. Show that f.a is rightcontinuous.
Tutorial 15: Stieltjes Integration 3
4. Let t ∈ R
+
and t
0
≤ . . . ≤ t
n
be a ﬁnite sequence in [0, t]. Show:
n
i=1
f.a(t
i
) −f.a(t
i−1
) ≤
_
]0,t]
fda
5. Show that f.a is a map of ﬁnite variation with:
f.a(t) ≤
_
t
0
fda , ∀t ∈ R
+
Exercise 3. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f ∈ L
1
C
(a).
1. Show that f.a is a rightcontinuous map of bounded variation.
2. Show d(f.a)([0, t]) = ν([0, t]), for all t ∈ R
+
, where ν =
_
fda.
3. Prove the following:
Tutorial 15: Stieltjes Integration 4
Theorem 86 Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f ∈ L
1
C
(a). The map f.a : R
+
→ C deﬁned by:
f.a(t)
=
_
t
0
fda , ∀t ∈ R
+
is a rightcontinuous map of bounded variation, and its associated
complex stieltjes measure is given by d(f.a) =
_
fda, i.e.
d(f.a)(B) =
_
B
fda , ∀B ∈ B(R
+
)
Exercise 4. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f ∈ L
1,loc
R
(a), f ≥ 0.
1. Show f.a is rightcontinuous, nondecreasing with f.a(0) ≥ 0.
2. Show d(f.a)([0, t]) = µ([0, t]), for all t ∈ R
+
, where µ =
_
fda.
3. Prove that d(f.a)([0, T] ∩ · ) = µ([0, T] ∩ · ), for all T ∈ R
+
.
Tutorial 15: Stieltjes Integration 5
4. Prove with the following:
Theorem 87 Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f ∈ L
1,loc
R
(a), f ≥ 0. The map f.a : R
+
→ R
+
deﬁned by:
f.a(t)
=
_
t
0
fda , ∀t ∈ R
+
is rightcontinuous, nondecreasing with (f.a)(0) ≥ 0, and its associ
ated stieltjes measure is given by d(f.a) =
_
fda, i.e.
d(f.a)(B) =
_
B
fda , ∀B ∈ B(R
+
)
Exercise 5. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f ∈ L
1,loc
C
(a) and T ∈ R
+
.
1. Show that
_
f1
[0,T]
da =
_
fda
[0,T]
=
_
fda
T
.
Tutorial 15: Stieltjes Integration 6
2. Show that f1
[0,T]
∈ L
1
C
(a) and f ∈ L
1
C
(a
T
).
3. Show that (f.a)
T
= f.(a
T
) = (f1
[0,T]
).a.
4. Show that for all B ∈ B(R
+
):
d(f.a)
T
(B) =
_
B
fda
T
=
_
B
f1
[0,T]
da
5. Explain why it does not in general make sense to write:
d(f.a)
T
= d(f.a)([0, T] ∩ · )
6. Show that for all B ∈ B(R
+
):
d(f.a)
T
(B) =
_
B
f1
[0,T]
da , ∀B ∈ B(R
+
)
7. Show that d(f.a)
T
 = df.a([0, T] ∩ · )
Tutorial 15: Stieltjes Integration 7
8. Show that for all t ∈ R
+
f.a(t) = (f.a)(t) =
_
t
0
fda
9. Show that f.a is of bounded variation if and only if f ∈ L
1
C
(a).
10. Show that ∆(f.a)(0) = f(0)∆a(0).
11. Let t > 0, (t
n
)
n≥1
be a sequence in R
+
with t
n
↑↑ t. Show:
lim
n→+∞
_
f1
[0,t
n
]
da =
_
f1
[0,t[
da
12. Show that ∆(f.a)(t) = f(t)∆a(t) for all t ∈ R
+
.
13. Show that if a is continuous with a(0) = 0, then f.a is itself
continuous with (f.a)(0) = 0.
14. Prove with the following:
Tutorial 15: Stieltjes Integration 8
Theorem 88 Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f ∈ L
1,loc
C
(a). The map f.a : R
+
→ C deﬁned by:
f.a(t)
=
_
t
0
fda , ∀t ∈ R
+
is rightcontinuous of ﬁnite variation, and we have f.a = f.a, i.e.
f.a(t) =
_
t
0
fda , ∀t ∈ R
+
In particular, f.a is of bounded variation if and only if f ∈ L
1
C
(a).
Furthermore, we have ∆(f.a) = f∆a.
Exercise 6. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
1. Prove the equivalence between the following:
(i) db << da
Tutorial 15: Stieltjes Integration 9
(ii) db
T
 << da , ∀T ∈ R
+
(iii) db
T
<< da , ∀T ∈ R
+
2. Does it make sense in general to write db << da?
Deﬁnition 113 Let a: R
+
→R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
We say that b is absolutely continuous with respect to a, and we
write b << a, if and only if, one of the following holds:
(i) db << da
(ii) db
T
 << da , ∀T ∈ R
+
(iii) db
T
<< da , ∀T ∈ R
+
In other words, b is absolutely continuous w.r. to a, if and only if the
stieltjes measure associated with the total variation of b is absolutely
continuous w.r. to the stieltjes measure associated with a.
Tutorial 15: Stieltjes Integration 10
Exercise 7. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let b : R
+
→ C be rightcontinuous of ﬁnite variation,
absolutely continuous w.r. to a, i.e. with b << a.
1. Show that for all T ∈ R
+
, there exits f
T
∈ L
1
C
(a) such that:
db
T
(B) =
_
B
f
T
da , ∀B ∈ B(R
+
)
2. Suppose that T, T
∈ R
+
and T ≤ T
. Show that:
_
B
f
T
da =
_
B∩[0,T]
f
T
da , ∀B ∈ B(R
+
)
3. Show that f
T
= f
T
1
[0,T]
daa.s.
4. Show the existence of a sequence (f
n
)
n≥1
in L
1
C
(a), such that
for all 1 ≤ n ≤ p, f
n
= f
p
1
[0,n]
and:
∀n ≥ 1 , db
n
(B) =
_
B
f
n
da , ∀B ∈ B(R
+
)
Tutorial 15: Stieltjes Integration 11
5. We deﬁne f : (R
+
, B(R
+
)) → (C, B(C)) by:
∀t ∈ R
+
, f(t)
= f
n
(t) for any n ≥ 1 : t ∈ [0, n]
Explain why f is unambiguously deﬁned.
6. Show that for all B ∈ B(C), {f ∈ B} = ∪
+∞
n=1
[0, n] ∩ {f
n
∈ B}.
7. Show that f : (R
+
, B(R
+
)) → (C, B(C)) is measurable.
8. Show that f ∈ L
1,loc
C
(a) and that we have:
b(t) =
_
t
0
fda , ∀t ∈ R
+
9. Prove the following:
Tutorial 15: Stieltjes Integration 12
Theorem 89 Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let b : R
+
→ C be a rightcontinuous map of ﬁnite
variation. Then, b is absolutely continuous w.r. to a, i.e. db << da,
if and only if there exists f ∈ L
1,loc
C
(a) such that b = f.a, i.e.
b(t) =
_
t
0
fda , ∀t ∈ R
+
If b is Rvalued, we can assume that f ∈ L
1,loc
R
(a).
If b is nondecreasing with b(0) ≥ 0, we can assume that f ≥ 0.
Exercise 8. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. Let f, g ∈ L
1,loc
C
(a) be such that f.a = g.a, i.e.:
_
t
0
fda =
_
t
0
gda , ∀t ∈ R
+
Tutorial 15: Stieltjes Integration 13
1. Show that for all T ∈ R
+
and B ∈ B(R
+
):
d(f.a)
T
(B) =
_
B
f1
[0,T]
da =
_
B
g1
[0,T]
da
2. Show that for all T ∈ R
+
, f1
[0,T]
= g1
[0,T]
daa.s.
3. Show that f = g daa.s.
Exercise 9. b : R
+
→ C is rightcontinuous of ﬁnite variation.
1. Show the existence of h ∈ L
1,loc
C
(b) such that b = h.b.
2. Show that for all B ∈ B(R
+
) and T ∈ R
+
:
db
T
(B) =
_
B
hdb
T
=
_
B
hdb
T

3. Show that h = 1 db
T
a.s. for all T ∈ R
+
.
4. Show that for all T ∈ R
+
, db([0, T] ∩ {h = 1}) = 0.
Tutorial 15: Stieltjes Integration 14
5. Show that h = 1 dba.s.
6. Prove the following:
Theorem 90 Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
There exists h ∈ L
1,loc
C
(b) such that h = 1 and b = h.b, i.e.
b(t) =
_
t
0
hdb , ∀t ∈ R
+
Deﬁnition 114 b : R
+
→ C is rightcontinuous of ﬁnite variation.
For all f ∈ L
1
C
(b), the stieltjes integral of f with respect to b, is
deﬁned as:
_
fdb
=
_
fhdb
where h ∈ L
1,loc
C
(b) is such that h = 1 and b = h.b.
Tutorial 15: Stieltjes Integration 15
Warning : the notation
_
fdb of deﬁnition (114) is controversial and
potentially confusing: ’db’ is not in general a complex measure on R
+
,
unless b is of bounded variation.
Exercise 10. b : R
+
→ C is rightcontinuous of ﬁnite variation.
1. Show that if f ∈ L
1
C
(b), then
_
fhdb is welldeﬁned.
2. Explain why, given f ∈ L
1
C
(b),
_
fdb is unambiguously deﬁned.
3. Show that if b is rightcontinuous, nondecreasing with b(0) ≥ 0,
deﬁnition (114) of
_
fdb coincides with that of an integral w.r.
to the stieltjes measure db.
4. Show that if b is a rightcontinuous map of bounded variation,
deﬁnition (114) of
_
fdb coincides with that of an integral w.r.
to the complex stieltjes measure db.
Tutorial 15: Stieltjes Integration 16
Exercise 11. Let b : R
+
→ C be a rightcontinuous map of ﬁnite
variation. For all f ∈ L
1,loc
C
(b), we deﬁne f.b : R
+
→ C as:
f.b(t)
=
_
t
0
fdb , ∀t ∈ R
+
1. Explain why f.b : R
+
→ C is a welldeﬁned map.
2. If b is rightcontinuous, nondecreasing with b(0) ≥ 0, show this
deﬁnition of f.b coincides with that of theorem (88).
3. Show f.b = (fh).b, where h ∈ L
1,loc
C
(b), h = 1, b = h.b.
4. Show that f.b : R
+
→ C is rightcontinuous of ﬁnite variation,
with f.b = f.b, i.e.
f.b(t) =
_
t
0
fdb , ∀t ∈ R
+
5. Show that f.b is of bounded variation if and only if f ∈ L
1
C
(b).
Tutorial 15: Stieltjes Integration 17
6. Let t > 0, (t
n
)
n≥1
be a sequence in R
+
such that t
n
↑↑ t. Show:
lim
n→+∞
_
fh1
[0,t
n
]
db =
_
fh1
[0,t[
db
7. Show that ∆(f.b) = f∆b.
8. Show that if b is continuous with b(0) = 0, then f.b is itself
continuous with (f.b)(0) = 0.
9. Prove the following:
Tutorial 15: Stieltjes Integration 18
Theorem 91 Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
For all f ∈ L
1,loc
C
(b), the map f.b : R
+
→ C deﬁned by:
f.b(t)
=
_
t
0
fdb , ∀t ∈ R
+
is rightcontinuous of ﬁnite variation, and we have f.b = f.b, i.e.
f.b(t) =
_
t
0
fdb , ∀t ∈ R
+
In particular, f.b is of bounded variation if and only if f ∈ L
1
C
(b).
Furthermore, we have ∆(f.b) = f∆b.
Exercise 12. Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
Let f ∈ L
1,loc
C
(b) and T ∈ R
+
.
1. Show that
_
f1
[0,T]
db =
_
fdb
[0,T]
=
_
fdb
T
.
2. Show that f1
[0,T]
∈ L
1
C
(b) and f ∈ L
1
C
(b
T
).
Tutorial 15: Stieltjes Integration 19
3. Show b
T
= h.b
T
, where h ∈ L
1,loc
C
(b), h = 1, b = h.b.
4. Show that (f.b)
T
= f.(b
T
) = (f1
[0,T]
).b
5. Show that df.b(B) =
_
B
fdb for all B ∈ B(R
+
).
6. Let g : R
+
→ C be a measurable map. Show the equivalence:
g ∈ L
1,loc
C
(f.b) ⇔ gf ∈ L
1,loc
C
(b)
7. Show that d(f.b)
T
(B) =
_
B
fhdb
T
 for all B ∈ B(R
+
).
8. Show that db
T
=
_
hdb
T
 and conclude that:
d(f.b)
T
(B) =
_
B
fdb
T
, ∀B ∈ B(R
+
)
9. Let g ∈ L
1,loc
C
(f.b). Show that g ∈ L
1
C
((f.b)
T
) and:
_
g1
[0,t]
d(f.b)
T
=
_
gf1
[0,t]
db
T
, ∀t ∈ R
+
Tutorial 15: Stieltjes Integration 20
10. Show that g.
_
(f.b)
T
_
= (gf).(b
T
).
11. Show that (g.(f.b))
T
= ((gf).b)
T
.
12. Show that g.(f.b) = (gf).b
13. Prove the following:
Theorem 92 Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
For all f ∈ L
1,loc
C
(b) and g : (R
+
, B(R
+
)) → (C, B(C)) measurable
map, we have the equivalence:
g ∈ L
1,loc
C
(f.b) ⇔ gf ∈ L
1,loc
C
(b)
and when such condition is satisﬁed, g.(f.b) = (fg).b, i.e.
_
t
0
gd(f.b) =
_
t
0
gfdb , ∀t ∈ R
+
Tutorial 15: Stieltjes Integration 21
Exercise 13. Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
let f, g ∈ L
1,loc
C
(b) and α ∈ C. Show that f + αg ∈ L
1,loc
C
(b), and:
(f + αg).b = f.b + αg.b
Exercise 14. Let b, c : R
+
→ C be two rightcontinuous maps of
ﬁnite variations. Let f ∈ L
1,loc
C
(b) ∩ L
1,loc
C
(c) and α ∈ C.
1. Show that for all T ∈ R
+
, d(b + αc)
T
= db
T
+ αdc
T
.
2. Show that for all T ∈ R
+
, db + αc
T
≤ db
T
+αdc
T
.
3. Show that db + αc ≤ db +αdc.
4. Show that f ∈ L
1,loc
C
(b + αc).
5. Show d(f.(b + αc))
T
(B) =
_
B
fd(b + αc)
T
for all B ∈ B(R
+
).
6. Show that d(f.(b + αc))
T
= d(f.b)
T
+ αd(f.c)
T
.
Tutorial 15: Stieltjes Integration 22
7. Show that (f.(b + αc))
T
= (f.b)
T
+ α(f.c)
T
8. Show that f.(b + αc) = f.b + α(f.c).
Exercise 15. Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
1. Show that db ≤ db
1
 + db
2
, where b = b
1
+ ib
2
.
2. Show that db
1
 ≤ db and db
2
 ≤ db.
3. Show that f ∈ L
1,loc
C
(b), if and only if:
f ∈ L
1,loc
C
(b
1

+
) ∩ L
1,loc
C
(b
1

−
) ∩ L
1,loc
C
(b
2

+
) ∩ L
1,loc
C
(b
2

−
)
4. Show that if f ∈ L
1,loc
C
(b), for all t ∈ R
+
:
_
t
0
fdb =
_
t
0
fdb
1

+
−
_
t
0
fdb
1

−
+i
__
t
0
fdb
2

+
−
_
t
0
fdb
2

−
_
Tutorial 15: Stieltjes Integration 23
Exercise 16. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. We deﬁne c : R
+
→ [0, +∞] as:
c(t)
= inf{s ∈ R
+
: t < a(s)} , ∀t ∈ R
+
where it is understood that inf ∅ = +∞. Let s, t ∈ R
+
.
1. Show that t < a(s) ⇒ c(t) ≤ s.
2. Show that c(t) < s ⇒ t < a(s).
3. Show that c(t) ≤ s ⇒ t < a(s + ) , ∀ > 0.
4. Show that c(t) ≤ s ⇒ t ≤ a(s).
5. Show that c(t) < +∞ ⇔ t < a(∞).
6. Show that c is nondecreasing.
7. Show that if t
0
∈ [a(∞), +∞[, c is rightcontinuous at t
0
.
Tutorial 15: Stieltjes Integration 24
8. Suppose t
0
∈ [0, a(∞)[. Given > 0, show the existence of
s ∈ R
+
, such that c(t
0
) ≤ s < c(t
0
) + and t
0
< a(s).
9. Show that t ∈ [t
0
, a(s)[ ⇒ c(t
0
) ≤ c(t) ≤ c(t
0
) + .
10. Show that c is rightcontinuous.
11. Show that if a(∞) = +∞, then c is a map c : R
+
→ R
+
which
is rightcontinuous, nondecreasing with c(0) ≥ 0.
12. We deﬁne ¯ a(s) = inf{t ∈ R
+
: s < c(t)} for all s ∈ R
+
. Show
that for all s, t ∈ R
+
, s < c(t) ⇒ a(s) ≤ t.
13. Show that a ≤ ¯ a.
14. Show that for all s, t ∈ R
+
and > 0:
a(s + ) ≤ t ⇒ s < s + ≤ c(t)
15. Show that for all s, t ∈ R
+
and > 0, a(s +) ≤ t ⇒ ¯ a(s) ≤ t.
Tutorial 15: Stieltjes Integration 25
16. Show that ¯ a ≤ a and conclude that:
a(s) = inf{t ∈ R
+
: s < c(t)}
Exercise 17. Let f : R
+
→
¯
R be a nondecreasing map. Let α ∈ R.
We deﬁne:
x
0
= sup{x ∈ R
+
: f(x) ≤ α}
1. Explain why x
0
= −∞ if and only if {f ≤ α} = ∅.
2. Show that x
0
= +∞ if and only if {f ≤ α} = R
+
.
3. We assume from now on that x
0
= ±∞. Show that x
0
∈ R
+
.
4. Show that if f(x
0
) ≤ α then {f ≤ α} = [0, x
0
].
5. Show that if α < f(x
0
) then {f ≤ α} = [0, x
0
[.
6. Conclude that f : (R
+
, B(R
+
)) → (
¯
R, B(
¯
R)) is measurable.
Tutorial 15: Stieltjes Integration 26
Exercise 18. Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. We deﬁne c : R
+
→ [0, +∞] as:
c(t)
= inf{s ∈ R
+
: t < a(s)} , ∀t ∈ R
+
1. Let f : R
+
→ [0, +∞] be nonnegative and measurable. Show
(f ◦ c)1
{c<+∞}
is welldeﬁned, nonnegative and measurable.
2. Let t, u ∈ R
+
, and ds be the lebesgue measure on R
+
. Show:
_
a(t)
0
(1
[0,u]
◦ c)1
{c<+∞}
ds ≤
_
1
[0,a(t∧u)]
1
{c<+∞}
ds
3. Show that:
_
a(t)
0
(1
[0,u]
◦ c)1
{c<+∞}
ds ≤ a(t ∧ u)
4. Show that:
a(t ∧ u) =
_
a(t)
0
1
[0,a(u)[
ds =
_
a(t)
0
1
[0,a(u)[
1
{c<+∞}
ds
Tutorial 15: Stieltjes Integration 27
5. Show that:
a(t ∧ u) ≤
_
a(t)
0
(1
[0,u]
◦ c)1
{c<+∞}
ds
6. Show that:
_
t
0
1
[0,u]
da =
_
a(t)
0
(1
[0,u]
◦ c)1
{c<+∞}
ds
7. Deﬁne:
D
t
=
_
B ∈ B(R
+
) :
_
t
0
1
B
da =
_
a(t)
0
(1
B
◦ c)1
{c<+∞}
ds
_
Show that D
t
is a dynkin system on R
+
, and D
t
= B(R
+
).
8. Show that if f : R
+
→ [0, +∞] is nonnegative measurable:
_
t
0
fda =
_
a(t)
0
(f ◦ c)1
{c<+∞}
ds , ∀t ∈ R
+
Tutorial 15: Stieltjes Integration 28
9. Let f : R
+
→ C be measurable. Show that (f ◦ c)1
{c<+∞}
is
itself welldeﬁned and measurable.
10. Show that if f ∈ L
1,loc
C
(a), then for all t ∈ R
+
, we have:
(f ◦ c)1
{c<+∞}
1
[0,a(t)]
∈ L
1
C
(R
+
, B(R
+
), ds)
and furthermore:
_
t
0
fda =
_
a(t)
0
(f ◦ c)1
{c<+∞}
ds
11. Show that we also have:
_
t
0
fda =
_
(f ◦ c)1
[0,a(t)[
ds
12. Conclude with the following:
Tutorial 15: Stieltjes Integration 29
Theorem 93 Let a : R
+
→ R
+
be rightcontinuous, nondecreasing
with a(0) ≥ 0. We deﬁne c : R
+
→ [0, +∞] as:
c(t)
= inf{s ∈ R
+
: t < a(s)} , ∀t ∈ R
+
Then, for all f ∈ L
1,loc
C
(a), we have:
_
t
0
fda =
_
a(t)
0
(f ◦ c)1
{c<+∞}
ds , ∀t ∈ R
+
where ds is the lebesgue measure on R
+
.
Tutorial 16: Diﬀerentiation 1
16. Diﬀerentiation
Deﬁnition 115 Let (Ω, T ) be a topological space. A map f : Ω →
¯
R
is said to be lowersemicontinuous (l.s.c), if and only if:
∀λ ∈ R , {λ < f} is open
We say that f is uppersemicontinuous (u.s.c), if and only if:
∀λ ∈ R , {f < λ} is open
Exercise 1. Let f : Ω →
¯
R be a map, where Ω is a topological space.
1. Show that f is l.s.c if and only if {λ < f} is open for all λ ∈
¯
R.
2. Show that f is u.s.c if and only if {f < λ} is open for all λ ∈
¯
R.
3. Show that every open set U in
¯
R can be written:
U = V
+
∪ V
−
∪
i∈I
]α
i
, β
i
[
Tutorial 16: Diﬀerentiation 2
for some index set I, α
i
, β
i
∈ R, V
+
= ∅ or V
+
=]α, +∞],
(α ∈ R) and V
−
= ∅ or V
−
= [−∞, β[, (β ∈ R).
4. Show that f is continuous if and only if it is both l.s.c and u.s.c.
5. Let u : Ω → R and v : Ω →
¯
R. Let λ ∈ R. Show that:
{λ < u +v} =
(λ
1
, λ
2
) ∈ R
2
λ
1
+λ
2
= λ
{λ
1
< u} ∩ {λ
2
< v}
6. Show that if both u and v are l.s.c, then u +v is also l.s.c.
7. Show that if both u and v are u.s.c, then u + v is also u.s.c.
8. Show that if f is l.s.c, then αf is l.s.c, for all α ∈ R
+
.
9. Show that if f is u.s.c, then αf is u.s.c, for all α ∈ R
+
.
10. Show that if f is l.s.c, then −f is u.s.c.
Tutorial 16: Diﬀerentiation 3
11. Show that if f is u.s.c, then −f is l.s.c.
12. Show that if V is open in Ω, then f = 1
V
is l.s.c.
13. Show that if F is closed in Ω, then f = 1
F
is u.s.c.
Exercise 2. Let (f
i
)
i∈I
be an arbitrary family of maps f
i
: Ω →
¯
R,
deﬁned on a topological space Ω.
1. Show that if all f
i
’s are l.s.c, then f = sup
i∈I
f
i
is l.s.c.
2. Show that if all f
i
’s are u.s.c, then f = inf
i∈I
f
i
is u.s.c.
Exercise 3. Let (Ω, T ) be a metrizable and σcompact topological
space. Let µ be a locally ﬁnite measure on (Ω, B(Ω)). Let f be an
element of ∈ L
1
R
(Ω, B(Ω), µ), such that f ≥ 0.
1. Let (s
n
)
n≥1
be a sequence of simple functions on (Ω, B(Ω)) such
that s
n
↑ f. Deﬁne t
1
= s
1
and t
n
= s
n
− s
n−1
for all n ≥ 2.
Show that t
n
is a simple function on (Ω, B(Ω)), for all n ≥ 1.
Tutorial 16: Diﬀerentiation 4
2. Show that f can be written as:
f =
+∞
n=1
α
n
1
A
n
where α
n
∈ R
+
\ {0} and A
n
∈ B(Ω), for all n ≥ 1.
3. Show that µ(A
n
) < +∞, for all n ≥ 1.
4. Show that there exist K
n
compact and V
n
open in Ω such that:
K
n
⊆ A
n
⊆ V
n
, µ(V
n
\ K
n
) ≤
α
n
2
n+1
for all > 0 and n ≥ 1.
5. Show the existence of N ≥ 1 such that:
+∞
n=N+1
α
n
µ(A
n
) ≤
2
Tutorial 16: Diﬀerentiation 5
6. Deﬁne u =
N
n=1
α
n
1
K
n
. Show that u is u.s.c.
7. Deﬁne v =
+∞
n=1
α
n
1
V
n
. Show that v is l.s.c.
8. Show that we have 0 ≤ u ≤ f ≤ v.
9. Show that we have:
v = u +
+∞
n=N+1
α
n
1
K
n
+
+∞
n=1
α
n
1
V
n
\K
n
10. Show that
vdµ ≤
udµ + < +∞.
11. Show that u ∈ L
1
R
(Ω, B(Ω), µ).
12. Explain why v may fail to be in L
1
R
(Ω, B(Ω), µ).
13. Show that v is µa.s. equal to an element of L
1
R
(Ω, B(Ω), µ).
14. Show that
(v −u)dµ ≤ .
Tutorial 16: Diﬀerentiation 6
15. Prove the following:
Theorem 94 (VitaliCaratheodory) Let (Ω, T ) be a metrizable
and σcompact topological space. Let µ be a locally ﬁnite measure
on (Ω, B(Ω)) and f be an element of L
1
R
(Ω, B(Ω), µ). Then, for all
> 0, there exist maps u, v : Ω →
¯
R, which are µa.s. equal to
elements of L
1
R
(Ω, B(Ω), µ), such that u ≤ f ≤ v, u is u.s.c, v is l.s.c,
and furthermore:
(v −u)dµ ≤
Deﬁnition 116 We call connected topological space, a topolog
ical space (Ω, T ), for which the only subsets of Ω which are both open
and closed, are Ω and ∅.
Tutorial 16: Diﬀerentiation 7
Exercise 4. Let (Ω, T ) be a topological space.
1. Show that (Ω, T ) is connected if and only if whenever Ω = AB
where A, B are disjoint open sets, we have A = ∅ or B = ∅.
2. Show that (Ω, T ) is connected if and only if whenever Ω = AB
where A, B are disjoint closed sets, we have A = ∅ or B = ∅.
Deﬁnition 117 Let (Ω, T ) be a topological space, and A ⊆ Ω. We
say that A is a connected subset of Ω, if and only if the induced
topological space (A, T
A
) is connected.
Exercise 5. Let A be open and closed in R, with A = ∅ and A
c
= ∅.
1. Let x ∈ A
c
. Show that A∩[x, +∞[ or A∩] −∞, x] is nonempty.
2. Suppose B = A ∩ [x, +∞[= ∅. Show that B is closed and that
we have B = A∩]x, +∞[. Conclude that B is also open.
Tutorial 16: Diﬀerentiation 8
3. Let b = inf B. Show that b ∈ B (and in particular b ∈ R).
4. Show the existence of > 0 such that ]b −, b +[⊆ B.
5. Conclude with the following:
Theorem 95 The topological space (R, T
R
) is connected.
Exercise 6. Let (Ω, T ) be a topological space and A ⊆ Ω be a
connected subset of Ω. Let B be a subset of Ω such that A ⊆ B ⊆
¯
A.
We assume that B = V
1
V
2
where V
1
, V
2
are disjoint open sets in B.
1. Show there is U
1
, U
2
open in Ω, with V
1
= B∩U
1
, V
2
= B∩U
2
.
2. Show that A∩ U
1
= ∅ or A∩ U
2
= ∅.
3. Suppose that A ∩ U
1
= ∅. Show that
¯
A ⊆ U
c
1
.
4. Show then that V
1
= B ∩ U
1
= ∅.
Tutorial 16: Diﬀerentiation 9
5. Conclude that B and
¯
A are both connected subsets of Ω.
Exercise 7. Prove the following:
Theorem 96 Let (Ω, T ), (Ω
, T
) be two topological spaces, and f
be a continuous map, f : Ω → Ω
. If (Ω, T ) is connected, then f(Ω)
is a connected subset of Ω
.
Deﬁnition 118 Let A ⊆
¯
R. We say that A is an interval, if and
only if for all x, y ∈ A with x ≤ y, we have [x, y] ⊆ A, where:
[x, y]
= {z ∈
¯
R : x ≤ z ≤ y}
Exercise 8. Let A ⊆
¯
R.
1. If A is an interval, and α = inf A, β = sup A, show that:
]α, β[⊆ A ⊆ [α, β]
Tutorial 16: Diﬀerentiation 10
2. Show that A is an interval if and only if, it is of the form [α, β],
[α, β[, ]α, β] or ]α, β[, for some α, β ∈
¯
R.
3. Show that an interval of the form ] − ∞, α[, where α ∈ R, is
homeomorphic to ] −1, α
[, for some α
∈ R.
4. Show that an interval of the form ]α, +∞[, where α ∈ R, is
homeomorphic to ]α
, 1[, for some α
∈ R.
5. Show that an interval of the form ]α, β[, where α, β ∈ R and
α < β, is homeomorphic to ] −1, 1[.
6. Show that ] −1, 1[ is homeomorphic to R.
7. Show an nonempty open interval in R, is homeomorphic to R.
8. Show that an open interval in R, is a connected subset of R.
9. Show that an interval in R, is a connected subset of R.
Tutorial 16: Diﬀerentiation 11
Exercise 9. Let A ⊆ R be a nonempty connected subset of R, and
α = inf A, β = sup A. We assume there exists x
0
∈ A
c
∩]α, β[.
1. Show that A∩]x
0
, +∞[ or A∩] −∞, x
0
[ is empty.
2. Show that if A∩]x
0
, +∞[= ∅, then β cannot be sup A.
3. Show that ]α, β[⊆ A ⊆ [α, β].
4. Show the following:
Theorem 97 For all A ⊆ R, A is a connected subset of R , if and
only if A is an interval.
Exercise 10. Prove the following:
Theorem 98 Let f : Ω → R be a continuous map, where (Ω, T )
is a connected topological space. Let a, b ∈ Ω such that f(a) ≤ f(b).
Then, for all z ∈ [f(a), f(b)], there exists x ∈ Ω such that z = f(x).
Tutorial 16: Diﬀerentiation 12
Exercise 11. Let a, b ∈ R, a < b, and f : [a, b] → R be a map such
that f
(x) exists for all x ∈ [a, b].
1. Show that f
: ([a, b], B([a, b])) → (R, B(R)) is measurable.
2. Show that f
∈ L
1
R
([a, b], B([a, b]), dx) is equivalent to:
b
a
f
(t)dt < +∞
3. We assume from now on that f
∈ L
1
R
([a, b], B([a, b]), dx). Given
> 0, show the existence of g : [a, b] →
¯
R, almost surely equal
to an element of L
1
R
([a, b], B([a, b]), dx), such that f
≤ g and g
is l.s.c, with:
b
a
g(t)dt ≤
b
a
f
(t)dt +
4. By considering g +α for some α > 0, show that without loss of
generality, we can assume that f
< g with the above inequality
still holding.
Tutorial 16: Diﬀerentiation 13
5. We deﬁne the complex measure ν =
gdx ∈ M
1
([a, b], B([a, b])).
Show that:
∀
> 0 , ∃δ > 0 , ∀E ∈ B([a, b]) , dx(E) ≤ δ ⇒ ν(E) <
6. For all η > 0 and x ∈ [a, b], we deﬁne:
F
η
(x)
=
x
a
g(t)dt −f(x) +f(a) +η(x − a)
Show that F
η
: [a, b] → R is a continuous map.
7. η being ﬁxed, let x = sup F
−1
η
({0}). Show that x ∈ [a, b] and
F
η
(x) = 0.
8. We assume that x ∈ [a, b[. Show the existence of δ > 0 such
that for all t ∈]x, x +δ[∩[a, b], we have:
f
(x) < g(t) and
f(t) −f(x)
t −x
< f
(x) +η
Tutorial 16: Diﬀerentiation 14
9. Show that for all t ∈]x, x+δ[∩[a, b], we have F
η
(t) > F
η
(x) = 0.
10. Show that there exists t
0
such that x < t
0
< b and F
η
(t
0
) > 0.
11. Show that if F
η
(b) < 0 then x cannot be sup F
−1
η
({0}).
12. Conclude that F
η
(b) ≥ 0, even if x = b.
13. Show that f(b) −f(a) ≤
b
a
f
(t)dt, and conclude:
Theorem 99 (Fundamental Calculus) Let a, b ∈ R, a < b, and
f : [a, b] → R be a map which is diﬀerentiable at every point of [a, b],
and such that:
b
a
f
(t)dt < +∞
Then, we have:
f(b) −f(a) =
b
a
f
(t)dt
Tutorial 16: Diﬀerentiation 15
Exercise 12. Let α > 0, and k
α
: R
n
→ R
n
deﬁned by k
α
(x) = αx.
1. Show that k
α
: (R
n
, B(R
n
)) → (R
n
, B(R
n
)) is measurable.
2. Show that for all B ∈ B(R
n
), we have:
dx({k
α
∈ B}) =
1
α
n
dx(B)
3. Show that for all > 0 and x ∈ R
n
:
dx(B(x, )) =
n
dx(B(0, 1))
Deﬁnition 119 Let µ be a complex measure on (R
n
, B(R
n
)), n ≥ 1,
with total variation µ. We call maximal function of µ, the map
Mµ : R
n
→ [0, +∞], deﬁned by:
∀x ∈ R
n
, (Mµ)(x)
= sup
>0
µ(B(x, ))
dx(B(x, ))
where B(x, ) is the open ball in R
n
, of center x and radius , with
respect to the usual metric of R
n
.
Tutorial 16: Diﬀerentiation 16
Exercise 13. Let µ be a complex measure on (R
n
, B(R
n
)).
1. Let λ ∈ R. Show that if λ < 0, then {λ < Mµ} = R
n
.
2. Show that if λ = 0, then {λ < Mµ} = R
n
if µ = 0, and
{λ < Mµ} is the empty set if µ = 0.
3. Suppose λ > 0. Let x ∈ {λ < Mµ}. Show the existence of > 0
such that µ(B(x, )) = tdx(B(x, )), for some t > λ.
4. Show the existence of δ > 0 such that ( +δ)
n
<
n
t/λ.
5. Show that if y ∈ B(x, δ), then B(x, ) ⊆ B(y, +δ).
6. Show that if y ∈ B(x, δ), then:
µ(B(y, +δ)) ≥
n
t
( +δ)
n
dx(B(y, +δ)) > λdx(B(y, +δ))
7. Conclude that B(x, δ) ⊆ {λ < Mµ}, and that the maximal
function Mµ : R
n
→ [0, +∞] is l.s.c, and therefore measurable.
Tutorial 16: Diﬀerentiation 17
Exercise 14. Let B
i
= B(x
i
,
i
), i = 1, . . . , N, N ≥ 1, be a ﬁnite
collection of open balls in R
n
. Assume without loss of generality that
N
≤ . . . ≤
1
. We deﬁne a sequence (J
k
) of sets by J
0
= {1, . . . , N}
and for all k ≥ 1:
J
k
=
J
k−1
∩ {j : j > i
k
, B
j
∩ B
i
k
= ∅} if J
k−1
= ∅
∅ if J
k−1
= ∅
where we have put i
k
= min J
k−1
, whenever J
k−1
= ∅.
1. Show that if J
k−1
= ∅ then J
k
⊂ J
k−1
(strict inclusion), k ≥ 1.
2. Let p = min{k ≥ 1 : J
k
= ∅}. Show that p is welldeﬁned.
3. Let S = {i
1
, . . . , i
p
}. Explain why S is well deﬁned.
4. Suppose that 1 ≤ k < k
≤ p. Show that i
k
∈ J
k
.
5. Show that (B
i
)
i∈S
is a family of pairwise disjoint open balls.
6. Let i ∈ {1, . . . , N} \ S, and deﬁne k
0
to be the minimum of the
set {k ∈ N
p
: i ∈ J
k
}. Explain why k
0
is welldeﬁned.
Tutorial 16: Diﬀerentiation 18
7. Show that i ∈ J
k
0
−1
and i
k
0
≤ i.
8. Show that B
i
∩ B
i
k
0
= ∅.
9. Show that B
i
⊆ B(x
i
k
0
, 3
i
k
0
).
10. Conclude that there exists a subset S of {1, . . . , N} such that
(B
i
)
i∈S
is a family of pairwise disjoint balls, and:
N
i=1
B(x
i
,
i
) ⊆
i∈S
B(x
i
, 3
i
)
11. Show that:
dx
N
i=1
B(x
i
,
i
)
≤ 3
n
i∈S
dx(B(x
i
,
i
))
Tutorial 16: Diﬀerentiation 19
Exercise 15. Let µ be a complex measure on R
n
. Let λ > 0 and K
be a nonempty compact subset of {λ < Mµ}.
1. Show that K can be covered by a ﬁnite collection B
i
= B(x
i
,
i
),
i = 1, . . . , N of open balls, such that:
∀i = 1, . . . , N , λdx(B
i
) < µ(B
i
)
2. Show the existence of S ⊆ {1, . . . , N} such that:
dx(K) ≤ 3
n
λ
−1
µ
i∈S
B(x
i
,
i
)
3. Show that dx(K) ≤ 3
n
λ
−1
µ
4. Conclude with the following:
Theorem 100 Let µ be a complex measure on (R
n
, B(R
n
)), n ≥ 1,
with maximal function Mµ. Then, for all λ ∈ R
+
\ {0}, we have:
dx({λ < Mµ}) ≤ 3
n
λ
−1
µ
Tutorial 16: Diﬀerentiation 20
Deﬁnition 120 Let f ∈ L
1
C
(R
n
, B(R
n
), dx), and µ be the complex
measure µ =
fdx on R
n
, n ≥ 1. We call maximal function of f,
denoted Mf, the maximal function Mµ of µ.
Exercise 16. Let f ∈ L
1
C
(R
n
, B(R
n
), dx), n ≥ 1.
1. Show that for all x ∈ R
n
:
(Mf)(x) = sup
>0
1
dx(B(x, ))
B(x,)
fdx
2. Show that for all λ > 0, dx({λ < Mf}) ≤ 3
n
λ
−1
f
1
.
Deﬁnition 121 Let f ∈ L
1
C
(R
n
, B(R
n
), dx), n ≥ 1. We say that
x ∈ R
n
is a lebesgue point of f, if and only if we have:
lim
↓↓0
1
dx(B(x, ))
B(x,)
f(y) −f(x)dy = 0
Tutorial 16: Diﬀerentiation 21
Exercise 17. Let f ∈ L
1
C
(R
n
, B(R
n
), dx), n ≥ 1.
1. Show that if f is continuous at x ∈ R
n
, then x is a Lebesgue
point of f.
2. Show that if x ∈ R
n
is a Lebesgue point of f, then:
f(x) = lim
↓↓0
1
dx(B(x, ))
B(x,)
f(y)dy
Exercise 18. Let n ≥ 1 and f ∈ L
1
C
(R
n
, B(R
n
), dx). For all > 0
and x ∈ R
n
, we deﬁne:
(T
f)(x)
=
1
dx(B(x, ))
B(x,)
f(y) −f(x)dy
and we put, for all x ∈ R
n
:
(Tf)(x)
= limsup
↓↓0
(T
f)(x)
= inf
>0
sup
u∈]0,[
(T
u
f)(x)
Tutorial 16: Diﬀerentiation 22
1. Given η > 0, show the existence of g ∈ C
c
C
(R
n
) such that:
f −g
1
≤ η
2. Let h = f −g. Show that for all > 0 and x ∈ R
n
:
(T
h)(x) ≤
1
dx(B(x, ))
B(x,)
hdx +h(x)
3. Show that Th ≤ Mh +h.
4. Show that for all > 0, we have T
f ≤ T
g +T
h.
5. Show that Tf ≤ Tg +Th.
6. Using the continuity of g, show that Tg = 0.
7. Show that Tf ≤ Mh +h.
8. Show that for all α > 0, {2α < Tf} ⊆ {α < Mh} ∪ {α < h}.
9. Show that dx({α < h}) ≤ α
−1
h
1
.
Tutorial 16: Diﬀerentiation 23
10. Conclude that for all α > 0 and η > 0, there is N
α,η
∈ B(R
n
)
such that {2α < Tf} ⊆ N
α,η
and dx(N
α,η
) ≤ η.
11. Show that for all α > 0, there exists N
α
∈ B(R
n
) such that
{2α < Tf} ⊆ N
α
and dx(N
α
) = 0.
12. Show there is N ∈ B(R
n
), dx(N) = 0, such that {Tf > 0} ⊆ N.
13. Conclude that Tf = 0 , dx−a.s.
14. Conclude with the following:
Theorem 101 Let f ∈ L
1
C
(R
n
, B(R
n
), dx), n ≥ 1. Then, dxalmost
surely, any x ∈ R
n
is a lebesgue points of f, i.e.
dxa.s. , lim
↓↓0
1
dx(B(x, ))
B(x,)
f(y) −f(x)dy = 0
Tutorial 16: Diﬀerentiation 24
Exercise 19. Let (Ω, F, µ) be a measure space and Ω
∈ F. We
deﬁne F
= F
Ω
and µ
= µ
F
. For all map f : Ω
→ [0, +∞] (or C),
we deﬁne
˜
f : Ω → [0, +∞] (or C), by:
˜
f(ω)
=
f(ω) if ω ∈ Ω
0 if ω ∈ Ω
1. Show that F
⊆ F and conclude that µ
is therefore a well
deﬁned measure on (Ω
, F
).
2. Let A ∈ F
and 1
A
be the characteristic function of A deﬁned
on Ω
. Let 1
A
be the characteristic function of A deﬁned on Ω.
Show that
˜
1
A
= 1
A
.
3. Let f : (Ω
, F
) → [0, +∞] be a nonnegative and measurable
map. Show that
˜
f : (Ω, F) → [0, +∞] is also nonnegative and
measurable, and that we have:
Ω
fdµ
=
Ω
˜
fdµ
Tutorial 16: Diﬀerentiation 25
4. Let f ∈ L
1
C
(Ω
, F
, µ
). Show that
˜
f ∈ L
1
C
(Ω, F, µ), and:
Ω
fdµ
=
Ω
˜
fdµ
Deﬁnition 122 Let b : R
+
→ C be a rightcontinuous map of ﬁnite
variation. We say that b is absolutely continuous, if and only if it
is absolutely continuous with respect to a(t) = t.
Exercise 20. Let b : R
+
→ C be rightcontinuous of ﬁnite variation.
1. Show that b is absolutely continuous, if and only if there is
f ∈ L
1,loc
C
(t) such that b(t) =
t
0
f(s)ds, for all t ∈ R
+
.
2. Show that b absolutely continuous ⇒b continuous with b(0) = 0.
Tutorial 16: Diﬀerentiation 26
Exercise 21. Let b : R
+
→ C be an absolutely continuous map.
Let f ∈ L
1,loc
C
(t) be such that b = f.t. For all n ≥ 1, we deﬁne
f
n
: R → C by:
f
n
(t)
=
f(t)1
[0,n]
(t) if t ∈ R
+
0 if t < 0
1. Let n ≥ 1. Show f
n
∈ L
1
C
(R, B(R), dx) and for all t ∈ [0, n]:
b(t) =
t
0
f
n
dx
2. Show the existence of N
n
∈ B(R) such that dx(N
n
) = 0, and
for all t ∈ N
c
n
, t is a Lebesgue point of f
n
.
3. Show that for all t ∈ R, and > 0:
1
t+
t
f
n
(s) −f
n
(t)ds ≤
2
dx(B(t, ))
B(t,)
f
n
(s) −f
n
(t)ds
Tutorial 16: Diﬀerentiation 27
4. Show that for all t ∈ N
c
n
, we have:
lim
↓↓0
1
t+
t
f
n
(s)ds = f
n
(t)
5. Show similarly that for all t ∈ N
c
n
, we have:
lim
↓↓0
1
t
t−
f
n
(s)ds = f
n
(t)
6. Show that for all t ∈ N
c
n
∩ [0, n[, b
(t) exists and b
(t) = f(t).
1
7. Show the existence of N ∈ B(R
+
), such that dx(N) = 0, and:
∀t ∈ N
c
, b
(t) exists with b
(t) = f(t)
8. Conclude with the following:
1
b
(0) being a r.h.s derivative only.
Tutorial 16: Diﬀerentiation 28
Theorem 102 A map b : R
+
→ C is absolutely continuous, if and
only if there exists f ∈ L
1,loc
C
(t) such that:
∀t ∈ R
+
, b(t) =
t
0
f(s)ds
in which case, b is almost surely diﬀerentiable with b
= f dxa.s.
Tutorial 17: Image Measure 1
17. Image Measure
In the following, K denotes R or C. We denote /
n
(K), n ≥ 1,
the set of all n nmatrices with Kvalued entries. We recall that
for all M = (m
ij
) ∈ /
n
(K), M is identiﬁed with the linear map
M : K
n
→ K
n
uniquely determined by:
∀j = 1, . . . , n , Me
j
=
n
i=1
m
ij
e
i
where (e
1
, . . . , e
n
) is the canonical basis of K
n
, i.e. e
i
= (0, .,
i
1 , ., 0).
Exercise 1. For all α ∈ K, let H
α
∈ /
n
(K) be deﬁned by:
H
α
=
α
1 0
0
.
.
.
1
Tutorial 17: Image Measure 2
i.e. by H
α
e
1
= αe
1
, H
α
e
j
= e
j
, for all j ≥ 2. For k, l ∈ ¦1, . . . , n¦,
we deﬁne the matrix Σ
kl
∈ /
n
(K) by Σ
kl
e
k
= e
l
, Σ
kl
e
l
= e
k
and
Σ
kl
e
j
= e
j
, for all j ∈ ¦1, . . . , n¦ ¸ ¦k, l¦. If n ≥ 2, we deﬁne the
matrix U ∈ /
n
(K) by:
U
=
1 0
1 1 0
0
.
.
.
1
i.e. by Ue
1
= e
1
+e
2
, Ue
j
= e
j
for all j ≥ 2. If n = 1, we put U = 1.
We deﬁne ^
n
(K) = ¦H
α
: α ∈ K¦ ∪ ¦Σ
kl
: k, l = 1, . . . , n¦ ∪ ¦U¦,
and /
n
(K) to be the set of all ﬁnite products of elements of ^
n
(K):
/
n
(K)
= ¦M∈/
n
(K): M = Q
1
. . . . .Q
p
, p ≥ 1 , Q
j
∈ ^
n
(K) , ∀j¦
We shall prove that /
n
(K) = /
n
(K).
Tutorial 17: Image Measure 3
1. Show that if α ∈ K¸ ¦0¦, H
α
is nonsingular with H
−1
α
= H
1/α
2. Show that if k, l = 1, . . . , n, Σ
kl
is nonsingular with Σ
−1
kl
= Σ
kl
.
3. Show that U is nonsingular, and that for n ≥ 2:
U
−1
=
1 0
−1 1 0
0
.
.
.
1
4. Let M = (m
ij
) ∈ /
n
(K). Let R
1
, . . . , R
n
be the rows of M:
M
=
R
1
R
2
.
.
.
R
n
Tutorial 17: Image Measure 4
Show that for all α ∈ K:
H
α
.M =
αR
1
R
2
.
.
.
R
n
Conclude that multiplying M by H
α
from the left, amounts to
multiplying the ﬁrst row of M by α.
5. Show that multiplying M by H
α
from the right, amounts to
multiplying the ﬁrst column of M by α.
6. Show that multiplying M by Σ
kl
from the left, amounts to swap
ping the rows R
l
and R
k
.
7. Show that multiplying M by Σ
kl
from the right, amounts to
swapping the columns C
l
and C
k
.
Tutorial 17: Image Measure 5
8. Show that multiplying M by U
−1
from the left ( n ≥ 2), amounts
to subtracting R
1
to R
2
, i.e.:
U
−1
.
R
1
R
2
.
.
.
R
n
=
R
1
R
2
−R
1
.
.
.
R
n
9. Show that multiplying M by U
−1
from the right (for n ≥ 2),
amounts to subtracting C
2
to C
1
.
10. Deﬁne U
= Σ
12
.U
−1
.Σ
12
, (n ≥ 2). Show that multiplying M
by U
from the right, amounts to subtracting C
1
to C
2
.
11. Show that if n = 1, then indeed we have /
1
(K) = /
1
(K).
Tutorial 17: Image Measure 6
Exercise 2. Further to exercise (1), we now assume that n ≥ 2, and
make the induction hypothesis that /
n−1
(K) = /
n−1
(K).
1. Let O
n
∈ /
n
(K) be the matrix with all entries equal to zero.
Show the existence of Q
1
, . . . , Q
p
∈ ^
n−1
(K), p ≥ 1, such that:
O
n−1
= Q
1
. . . . .Q
p
2. For k = 1, . . . , p, we deﬁne Q
k
∈ /
n
(K), by:
Q
k
=
0
Q
k
.
.
.
0
0 . . . 0 1
Tutorial 17: Image Measure 7
Show that Q
k
∈ ^
n
(K), and that we have:
Σ
1n
.Q
1
. . . . .Q
p
.Σ
1n
=
1 0 . . . 0
0
.
.
. O
n−1
0
3. Conclude that O
n
∈ /
n
(K).
4. We now consider M = (m
ij
) ∈ /
n
(K), M ,= O
n
. We want to
show that M ∈ /
n
(K). Show that for some k, l ∈ ¦1, . . . , n¦:
H
−1
m
kl
.Σ
1k
.M.Σ
1l
=
1 ∗ . . . ∗
∗
.
.
. ∗
∗
5. Show that if H
−1
m
kl
.Σ
1k
.M.Σ
1l
∈ /
n
(K), then M ∈ /
n
(K).
Conclude that without loss of generality, in order to prove that
Tutorial 17: Image Measure 8
M lies in /
n
(K) we can assume that m
11
= 1.
6. Let i = 2, . . . , n. Show that if m
i1
,= 0, we have:
H
−1
m
i1
.Σ
2i
.U
−1
.Σ
2i
.H
−1
1/m
i1
.M =
1 ∗ . . . ∗
∗
0 ← i ∗
∗
7. Conclude that without loss of generality, we can assume that
m
i1
= 0 for all i ≥ 2, i.e. that M is of the form:
M =
1 ∗ . . . ∗
0
.
.
. ∗
0
8. Show that in order to prove that M ∈ /
n
(K), without loss of
Tutorial 17: Image Measure 9
generality, we can assume that M is of the form:
M =
1 0 . . . 0
0
.
.
. M
0
9. Prove that M ∈ /
n
(K) and conclude with the following:
Theorem 103 Given n ≥ 2, any n nmatrix with values in K is
a ﬁnite product of matrices Q of the following types:
(i) Qe
1
= αe
1
, Qe
j
= e
j
, ∀j = 2, . . . , n , (α ∈ K)
(ii) Qe
l
= e
k
, Qe
k
= e
l
, Qe
j
= e
j
, ∀j ,= k, l , (k, l ∈ N
n
)
(iii) Qe
1
= e
1
+e
2
, Qe
j
= e
j
, ∀j = 2, . . . , n
where (e
1
, . . . , e
n
) is the canonical basis of K
n
.
Tutorial 17: Image Measure 10
Deﬁnition 123 Let X : (Ω, T) → (Ω
, T
) be a measurable map,
where (Ω, T) and (Ω
, T
) are two measurable spaces. Given a measure
µ (possibly complex) on (Ω, T), we call distribution of X under µ,
or law of X under µ, or image measure of µ by X, the measure
(possibly complex) denoted µ
X
or X(µ) on (Ω
, T
), deﬁned by:
∀B ∈ T
, µ
X
(B)
= µ(¦X ∈ B¦) = µ(X
−1
(B))
Exercise 3. Let X : (Ω, T) → (Ω
, T
) be a measurable map, where
(Ω, T) and (Ω
, T
) are two measurable spaces.
1. Show that if µ is a measure on (Ω, T), µ
X
is a welldeﬁned
measure on (Ω
, T
).
2. Show that if µ is a complex measure on (Ω, T), µ
X
is a well
deﬁned complex measure on (Ω
, T
).
3. Let B ∈ T
. Show that if (E
n
)
n≥1
is a measurable partition of
B, then (X
−1
(E
n
))
n≥1
is a measurable partition of X
−1
(B).
Tutorial 17: Image Measure 11
4. Show that if µ is a complex measure on (Ω, T), then [µ
X
[ ≤ [µ[
X
.
5. Let Y : (Ω
, T
) → (Ω
, T
) be a measurable map, where
(Ω
, T
) is another measurable space. Show that for all (possi
bly complex) measure µ on (Ω, T), we have:
Y (X(µ)) = (Y ◦ X)(µ) = (µ
X
)
Y
= µ
(Y ◦X)
Deﬁnition 124 Let µ be a measure (possibly complex) on R
n
, n ≥ 1.
We say that µ is invariant by translation, if and only if for all
a ∈ R
n
, and associated translation mapping τ
a
: R
n
→ R
n
deﬁned by
τ
a
(x) = a +x, we have τ
a
(µ) = µ.
Exercise 4. Let µ be a measure (possibly complex) on (R
n
, B(R
n
)).
1. Show that τ
a
: (R
n
, B(R
n
)) → (R
n
, B(R
n
)) is measurable.
2. Show τ
a
(µ) is therefore a welldeﬁned (possibly complex) mea
sure on (R
n
, B(R
n
)), for all a ∈ R
n
.
Tutorial 17: Image Measure 12
3. Show that τ
a
(dx) = dx for all a ∈ R
n
.
4. Show the lebesgue measure on R
n
is invariant by translation.
Exercise 5. Let k
α
: R
n
→ R
n
be deﬁned by k
α
(x) = αx, α > 0.
1. Show that k
α
: (R
n
, B(R
n
)) → (R
n
, B(R
n
)) is measurable.
2. Show that k
α
(dx) = α
−n
dx.
Exercise 6. Show the following:
Theorem 104 (Integral Projection 1) Let X: (Ω, T) → (Ω
, T
)
be a measurable map, where (Ω, T), (Ω
, T
) are measurable spaces.
Let µ be a measure on (Ω, T). Then, for all f : (Ω
, T
) → [0, +∞]
nonnegative and measurable, we have:
Ω
f ◦ Xdµ =
Ω
fdX(µ)
Tutorial 17: Image Measure 13
Exercise 7. Show the following:
Theorem 105 (Integral Projection 2) Let X: (Ω, T) → (Ω
, T
)
be a measurable map, where (Ω, T), (Ω
, T
) are measurable spaces.
Let µ be a measure on (Ω, T). Then, for all f : (Ω
, T
) → (C, B(C))
measurable, we have the equivalence:
f ◦ X ∈ L
1
C
(Ω, T, µ) ⇔ f ∈ L
1
C
(Ω
, T
, X(µ))
in which case, we have:
Ω
f ◦ Xdµ =
Ω
fdX(µ)
Exercise 8. Further to theorem (105), suppose µ is in fact a complex
measure on (Ω, T). Show that:
Ω
[f[d[X(µ)[ ≤
Ω
[f ◦ X[d[µ[ (1)
Conclude with the following:
Tutorial 17: Image Measure 14
Theorem 106 (Integral Projection 3) Let X: (Ω, T) → (Ω
, T
)
be a measurable map, where (Ω, T), (Ω
, T
) are measurable spaces.
Let µ be a complex measure on (Ω, T). Then, for all measurable map
f : (Ω
, T
) → (C, B(C)), we have:
f ◦ X ∈ L
1
C
(Ω, T, µ) ⇒ f ∈ L
1
C
(Ω
, T
, X(µ))
and when the lefthand side of this implication is satisﬁed:
Ω
f ◦ Xdµ =
Ω
fdX(µ)
Exercise 9. Let X : (Ω, T) → (R
n
, B(R
n
)) be a measurable map
with distribution µ = X(P), where (Ω, T, P) is a probability space.
1. Show that X is integrable, if and only if:
+∞
−∞
[x[dµ(x) < +∞
Tutorial 17: Image Measure 15
2. Show that if X is integrable, then:
E[X] =
+∞
−∞
xdµ(x)
3. Show that:
E[X
2
] =
+∞
−∞
x
2
dµ(x)
Exercise 10. Let µ be a locally ﬁnite measure on (R
n
, B(R
n
)), which
is invariant by translation. For all a = (a
1
, . . . , a
n
) ∈ (R
+
)
n
, we deﬁne
Q
a
= [0, a
1
[. . . [0, a
n
[, and in particular Q = Q
(1,...,1)
= [0, 1[
n
.
1. Show that µ(Q
a
) < +∞ for all a ∈ (R
+
)
n
, and µ(Q) < +∞.
2. Let p = (p
1
, . . . , p
n
) where p
i
≥ 1 is an integer for all i’s. Show:
Q
p
=
k = (k
1
, . . . , k
n
)
0 ≤ k
i
< p
i
[k
1
, k
1
+ 1[. . . [k
n
, k
n
+ 1[
Tutorial 17: Image Measure 16
3. Show that µ(Q
p
) = p
1
. . . p
n
µ(Q).
4. Let q
1
, . . . , q
n
≥ 1 be n positive integers. Show that:
Q
p
=
k = (k
1
, . . . , k
n
)
0 ≤ k
i
< q
i
[
k
1
p
1
q
1
,
(k
1
+ 1)p
1
q
1
[. . . [
k
n
p
n
q
n
,
(k
n
+ 1)p
n
q
n
[
5. Show that µ(Q
p
) = q
1
. . . q
n
µ(Q
(p
1
/q
1
,...,p
n
/q
n
)
)
6. Show that µ(Q
r
) = r
1
. . . r
n
µ(Q), for all r ∈ (Q
+
)
n
.
7. Show that µ(Q
a
) = a
1
. . . a
n
µ(Q), for all a ∈ (R
+
)
n
.
8. Show that µ(B) = µ(Q)dx(B), for all B ∈ (, where:
(
= ¦[a
1
, b
1
[. . . [a
n
, b
n
[ , a
i
, b
i
∈ R , a
i
≤ b
i
, ∀i ∈ N
n
¦
9. Show that B(R
n
) = σ(().
Tutorial 17: Image Measure 17
10. Show that µ = µ(Q)dx, and conclude with the following:
Theorem 107 Let µ be a locally ﬁnite measure on (R
n
, B(R
n
)). If
µ is invariant by translation, then there exists α ∈ R
+
such that:
µ = αdx
Exercise 11. Let T : R
n
→ R
n
be a linear bijection.
1. Show that T and T
−1
are continuous.
2. Show that for all B ⊆ R
n
, the inverse image T
−1
(B) = ¦T ∈ B¦
coincides with the direct image:
T
−1
(B)
= ¦y : y = T
−1
(x) for some x ∈ B¦
3. Show that for all B ⊆ R
n
, the direct image T(B) coincides with
the inverse image (T
−1
)
−1
(B) = ¦T
−1
∈ B¦.
4. Let K ⊆ R
n
be compact. Show that ¦T ∈ K¦ is compact.
Tutorial 17: Image Measure 18
5. Show that T(dx) is a locally ﬁnite measure on (R
n
, B(R
n
)).
6. Let τ
a
be the translation of vector a ∈ R
n
. Show that:
T ◦ τ
T
−1
(a)
= τ
a
◦ T
7. Show that T(dx) is invariant by translation.
8. Show the existence of α ∈ R
+
, such that T(dx) = αdx. Show
that such constant is unique, and denote it by ∆(T).
9. Show that Q = T([0, 1]
n
) ∈ B(R
n
) and that we have:
∆(T)dx(Q) = T(dx)(Q) = 1
10. Show that ∆(T) ,= 0.
11. Let T
1
, T
2
: R
n
→ R
n
be two linear bijections. Show that:
(T
1
◦ T
2
)(dx) = ∆(T
1
)∆(T
2
)dx
and conclude that ∆(T
1
◦ T
2
) = ∆(T
1
)∆(T
2
).
Tutorial 17: Image Measure 19
Exercise 12. Let α ∈ R ¸ ¦0¦. Let H
α
: R
n
→ R
n
be the linear
bijection uniquely deﬁned by H
α
(e
1
) = αe
1
, H
α
(e
j
) = e
j
for j ≥ 2.
1. Show that H
α
(dx)([0, 1]
n
) = [α[
−1
.
2. Conclude that ∆(H
α
) = [ det H
α
[
−1
.
Exercise 13. Let k, l ∈ N
n
and Σ : R
n
→ R
n
be the linear bijection
uniquely deﬁned by Σ(e
k
) = e
l
, Σ(e
l
) = e
k
, Σ(e
j
) = e
j
, for j ,= k, l.
1. Show that Σ(dx)([0, 1]
n
) = 1.
2. Show that Σ.Σ = I
n
. (Identity mapping on R
n
).
3. Show that [ det Σ[ = 1.
4. Conclude that ∆(Σ) = [ det Σ[
−1
.
Exercise 14. Let n ≥ 2 and U : R
n
→ R
n
be the linear bijection
uniquely deﬁned by U(e
1
) = e
1
+ e
2
and U(e
j
) = e
j
for j ≥ 2. Let
Q = [0, 1[
n
.
Tutorial 17: Image Measure 20
1. Show that:
U
−1
(Q) = ¦x ∈ R
n
: 0 ≤ x
1
+x
2
< 1 , 0 ≤ x
i
< 1 , ∀i ,= 2¦
2. Deﬁne:
Ω
1
= U
−1
(Q) ∩ ¦x ∈ R
n
: x
2
≥ 0¦
Ω
2
= U
−1
(Q) ∩ ¦x ∈ R
n
: x
2
< 0¦
Show that Ω
1
, Ω
2
∈ B(R
n
).
3. Let τ
e
2
be the translation of vector e
2
. Draw a picture of Q, Ω
1
,
Ω
2
and τ
e
2
(Ω
2
) in the case when n = 2.
4. Show that if x ∈ Ω
1
, then 0 ≤ x
2
< 1.
5. Show that Ω
1
⊆ Q.
6. Show that if x ∈ τ
e
2
(Ω
2
), then 0 ≤ x
2
< 1.
7. Show that τ
e
2
(Ω
2
) ⊆ Q.
Tutorial 17: Image Measure 21
8. Show that if x ∈ Q and x
1
+x
2
< 1 then x ∈ Ω
1
.
9. Show that if x ∈ Q and x
1
+x
2
≥ 1 then x ∈ τ
e
2
(Ω
2
).
10. Show that if x ∈ τ
e
2
(Ω
2
) then x
1
+x
2
≥ 1.
11. Show that τ
e
2
(Ω
2
) ∩ Ω
1
= ∅.
12. Show that Q = Ω
1
¬ τ
e
2
(Ω
2
).
13. Show that dx(Q) = dx(U
−1
(Q)).
14. Show that ∆(U) = 1.
15. Show that ∆(U) = [ det U[
−1
.
Exercise 15. Let T : R
n
→ R
n
be a linear bijection, (n ≥ 1).
1. Show the existence of linear bijections Q
1
, . . . , Q
p
: R
n
→ R
n
,
p ≥ 1, with T = Q
1
◦. . . ◦Q
p
, ∆(Q
i
) = [ det Q
i
[
−1
for all i ∈ N
p
.
Tutorial 17: Image Measure 22
2. Show that ∆(T) = [ det T[
−1
.
3. Conclude with the following:
Theorem 108 Let n ≥ 1 and T : R
n
→ R
n
be a linear bijection.
Then, the image measure T(dx) of the lebesgue measure on R
n
is:
T(dx) = [ det T[
−1
dx
Exercise 16. Let f : (R
2
, B(R
2
)) → [0, +∞] be a nonnegative and
measurable map. Let a, b, c, d ∈ R such that ad −bc ,= 0. Show that:
R
2
f(ax +by, cx +dy)dxdy = [ad −bc[
−1
R
2
f(x, y)dxdy
Exercise 17. Let T : R
n
→ R
n
be a linear bijection. Show that for
all B ∈ B(R
n
), we have T(B) ∈ B(R
n
) and:
dx(T(B)) = [ det T[dx(B)
Tutorial 17: Image Measure 23
Exercise 18. Let V be a linear subspace of R
n
and p = dimV . We
assume that 1 ≤ p ≤ n−1. Let u
1
, . . . , u
p
be an orthonormal basis of
V , and u
p+1
, . . . , u
n
be such that u
1
, . . . , u
n
is an orthonormal basis
of R
n
. For i ∈ N
n
, Let φ
i
: R
n
→ R be deﬁned by φ
i
(x) = ¸u
i
, x).
1. Show that all φ
i
’s are continuous.
2. Show that V =
n
j=p+1
φ
−1
j
(¦0¦).
3. Show that V is a closed subset of R
n
.
4. Let Q = (q
ij
) ∈ /
n
(R) be the matrix uniquely deﬁned by
Qe
j
= u
j
for all j ∈ N
n
, where (e
1
, . . . , e
n
) is the canonical
basis of R
n
. Show that for all i, j ∈ N
n
:
¸u
i
, u
j
) =
n
k=1
q
ki
q
kj
5. Show that Q
T
.Q = Q.Q
T
= I
n
and conclude that [ det Q[ = 1.
Tutorial 17: Image Measure 24
6. Show that dx(¦Q ∈ V ¦) = dx(V ).
7. Show that ¦Q ∈ V ¦ = span(e
1
, . . . , e
p
).
1
8. For all m ≥ 1, we deﬁne:
E
m
=
n−1
[−m, m] . . . [−m, m] ¦0¦
Show that dx(E
m
) = 0 for all m ≥ 1.
9. Show that dx(span(e
1
, . . . , e
n−1
)) = 0.
10. Conclude with the following:
Theorem 109 Let n ≥ 1. Any linear subspace V of R
n
is a closed
subset of R
n
. Moreover, if dimV ≤ n −1, then dx(V ) = 0.
1
i.e. the linear subspace of R
n
generated by e
1
, . . . , e
p
.
Tutorial 18: The Jacobian Formula 1
18. The Jacobian Formula
In the following, K denotes R or C.
Deﬁnition 125 We call Knormed space, an ordered pair (E, N),
where E is a Kvector space, and N : E → R
+
is a norm on E.
See deﬁnition (89) for vector space, and deﬁnition (95) for norm.
Exercise 1. Let (H, ¸, )) be a Khilbert space, and   =
_
¸, ).
1. Show that   is a norm on H.
2. Show that (H,  ) is a Knormed space.
Exercise 2. Let (E,  ) be a Knormed space:
1. Show that d(x, y) = x −y deﬁnes a metric on E.
2. Show that for all x, y ∈ E, we have [ x −y [ ≤ x −y.
Tutorial 18: The Jacobian Formula 2
Deﬁnition 126 Let (E,  ) be a Knormed space, and d be the
metric deﬁned by d(x, y) = x −y. We call norm topology on E,
denoted T
·
, the topology on E associated with d.
Exercise 3. Let E, F be two Knormed spaces, and l : E → F be a
linear map. Show that the following are equivalent:
(i) l is continuous (w.r. to the norm topologies)
(ii) l is continuous at x = 0.
(iii) ∃K ∈ R
+
, ∀x ∈ E , l(x) ≤ Kx
(iv) sup¦l(x) : x ∈ E , x = 1¦ < +∞
Deﬁnition 127 Let E, F be Knormed spaces. The Kvector space
of all continuous linear maps l : E → F is denoted L
K
(E, F).
Exercise 4. Show that L
K
(E, F) is indeed a Kvector space.
Tutorial 18: The Jacobian Formula 3
Exercise 5. Let E, F be Knormed spaces. Given l ∈ L
K
(E, F), let:
l
= sup¦l(x) : x ∈ E , x = 1¦ < +∞
1. Show that:
l = sup¦l(x) : x ∈ E , x ≤ 1¦
2. Show that:
l = sup
_
l(x)
x
: x ∈ E , x ,= 0
_
3. Show that l(x) ≤ l.x, for all x ∈ E.
4. Show that l is the smallest K ∈ R
+
, such that:
∀x ∈ E , l(x) ≤ Kx
5. Show that l → l is a norm on L
K
(E, F).
6. Show that (L
K
(E, F),  ) is a Knormed space.
Tutorial 18: The Jacobian Formula 4
Deﬁnition 128 Let E, F be Rnormed spaces and U be an open
subset of E. We say that a map φ : U → F is diﬀerentiable at
some a ∈ U, if and only if there exists l ∈ L
R
(E, F) such that, for all
> 0, there exists δ > 0, such that for all h ∈ E:
h ≤ δ ⇒ a + h ∈ U and φ(a + h) −φ(a) −l(h) ≤ h
Exercise 6. Let E, F be two Rnormed spaces, and U be open in E.
Let φ : U → F be a map and a ∈ U.
1. Suppose that φ : U → F is diﬀerentiable at a ∈ U, and that
l
1
, l
2
∈ L
R
(E, F) satisfy the requirement of deﬁnition (128).
Show that for all > 0, there exists δ > 0 such that:
∀h ∈ E , h ≤ δ ⇒ l
1
(h) −l
2
(h) ≤ h
2. Conclude that l
1
−l
2
 = 0 and ﬁnally that l
1
= l
2
.
Tutorial 18: The Jacobian Formula 5
Deﬁnition 129 Let E, F be Rnormed spaces and U be an open
subset of E. Let φ : U → F be a map and a ∈ U. If φ is diﬀerentiable
at a, we call diﬀerential of φ at a, the unique element of L
R
(E, F),
denoted dφ(a), satisfying the requirement of deﬁnition (128). If φ is
diﬀerentiable at all points of U, the map dφ : U → L
R
(E, F) is also
called the diﬀerential of φ.
Deﬁnition 130 Let E, F be Rnormed spaces and U be an open
subset of E. A map φ : U → F is said to be of class C
1
, if and only
if dφ(a) exists for all a ∈ U, and the diﬀerential dφ : U → L
R
(E, F)
is a continuous map.
Exercise 7. Let E, F be two Rnormed spaces and U be open in E.
Let φ : U → F be a map, and a ∈ U.
1. Show that φ diﬀerentiable at a ⇒ φ continuous at a.
2. If φ is of class C
1
, explain with respect to which topologies the
diﬀerential dφ : U → L
R
(E, F) is said to be continuous.
Tutorial 18: The Jacobian Formula 6
3. Show that if φ is of class C
1
, then φ is continuous.
4. Suppose that E = R. Show that for all a ∈ U, φ is diﬀerentiable
at a ∈ U, if and only if the derivative:
φ
(a)
= lim
t=0,t→0
φ(a + t) −φ(a)
t
exists in F, in which case dφ(a) ∈ L
R
(R, F) is given by:
∀t ∈ R , dφ(a)(t) = t.φ
(a)
In particular, φ
(a) = dφ(a)(1).
Exercise 8. Let E, F, G be three Rnormed spaces. Let U be open
in E and V be open in F. Let φ : U → F and ψ : V → G be two maps
such that φ(U) ⊆ V . We assume that φ is diﬀerentiable at a ∈ U,
and we put:
l
1
= dφ(a) ∈ L
R
(E, F)
Tutorial 18: The Jacobian Formula 7
We assume that ψ is diﬀerentiable at φ(a) ∈ V , and we put:
l
2
= dψ(φ(a)) ∈ L
R
(F, G)
1. Explain why ψ ◦ φ : U → G is a welldeﬁned map.
2. Given > 0, show the existence of η > 0 such that:
η(η +l
1
 +l
2
) ≤
3. Show the existence of δ
2
> 0 such that for all h
2
∈ F with
h
2
 ≤ δ
2
, we have φ(a) + h
2
∈ V and:
ψ(φ(a) + h
2
) −ψ ◦ φ(a) −l
2
(h
2
) ≤ ηh
2

4. Show that if h
2
∈ F and h
2
 ≤ δ
2
, then for all h ∈ E, we have:
ψ(φ(a) +h
2
) −ψ◦φ(a) −l
2
◦l
1
(h) ≤ ηh
2
+l
2
.h
2
−l
1
(h)
5. Show the existence of δ > 0 such that for all h ∈ E with h ≤ δ,
we have a+h ∈ U and φ(a+h)−φ(a)−l
1
(h) ≤ ηh, together
with φ(a + h) −φ(a) ≤ δ
2
.
Tutorial 18: The Jacobian Formula 8
6. Show that if h ∈ E is such that h ≤ δ, then a + h ∈ U and:
ψ ◦ φ(a+h)−ψ ◦ φ(a)−l
2
◦ l
1
(h)≤ηφ(a+h)−φ(a)+ηl
2
.h
≤η(η +l
1
 +l
2
)h
≤h
7. Show that l
2
◦ l
1
∈ L
R
(E, G)
8. Conclude with the following:
Theorem 110 Let E, F, G be three Rnormed spaces, U be open in
E and V be open in F. Let φ : U → F and ψ : V → G be two maps
such that φ(U) ⊆ V . Let a ∈ U. Then, if φ is diﬀerentiable at a ∈ U,
and ψ is diﬀerentiable at φ(a) ∈ V , then ψ ◦ φ is diﬀerentiable at
a ∈ U, and furthermore:
d(ψ ◦ φ)(a) = dψ(φ(a)) ◦ dφ(a)
Tutorial 18: The Jacobian Formula 9
Exercise 9. Let E, F, G be three Rnormed spaces. Let U be open
in E and V be open in F. Let φ : U → F and ψ : V → G be two
maps of class C
1
such that φ(U) ⊆ V .
1. For all (l
1
, l
2
) ∈ L
R
(F, G) L
R
(E, F), we deﬁne:
N
1
(l
1
, l
2
)
= l
1
 +l
2

N
2
(l
1
, l
2
)
=
_
l
1

2
+l
2

2
N
∞
(l
1
, l
2
)
= max(l
1
, l
2
)
Show that N
1
, N
2
, N
∞
are all norms on L
R
(F, G) L
R
(E, F).
2. Show they induce the product topology on L
R
(F, G)L
R
(E, F).
3. We deﬁne the map H : L
R
(F, G) L
R
(E, F) → L
R
(E, G) by:
∀(l
1
, l
2
) ∈ L
R
(F, G) L
R
(E, F) , H(l
1
, l
2
)
= l
1
◦ l
2
Show that H(l
1
, l
2
) ≤ l
1
.l
2
, for all l
1
, l
2
.
Tutorial 18: The Jacobian Formula 10
4. Show that H is continuous.
5. We deﬁne K : U → L
R
(F, G) L
R
(E, F) by:
∀a ∈ U , K(a)
= (dψ(φ(a)), dφ(a))
Show that K is continuous.
6. Show that ψ ◦ φ is diﬀerentiable on U.
7. Show that d(ψ ◦ φ) = H ◦ K.
8. Conclude with the following:
Theorem 111 Let E, F, G be three Rnormed spaces, U be open in
E and V be open in F. Let φ : U → F and ψ : V → G be two maps
of class C
1
such that φ(U) ⊆ V . Then, ψ ◦ φ : U → G is of class C
1
.
Tutorial 18: The Jacobian Formula 11
Exercise 10. Let E be an Rnormed space. Let a, b ∈ R, a < b.
Let f : [a, b] → E and g : [a, b] → R be two continuous maps which
are diﬀerentiable at every points of ]a, b[. We assume that:
∀t ∈]a, b[ , f
(t) ≤ g
(t)
1. Given > 0, we deﬁne φ
: [a, b] → R by:
φ
(t)
= f(t) −f(a) −g(t) + g(a) −(t −a)
for all t ∈ [a, b]. Show that φ
is continuous.
2. Deﬁne E
= ¦t ∈ [a, b] : φ
(t) ≤ ¦, and c = sup E
. Show that:
c ∈ [a, b] and φ
(c) ≤
3. Show the existence of h > 0, such that:
∀t ∈ [a, a + h[∩[a, b] , φ
(t) ≤
4. Show that c ∈]a, b].
Tutorial 18: The Jacobian Formula 12
5. Suppose that c ∈]a, b[. Show the existence of t
0
∈]c, b] such that:
_
_
_
_
f(t
0
) −f(c)
t
0
−c
_
_
_
_
≤ f
(c) + /2 and g
(c) ≤
g(t
0
) −g(c)
t
0
−c
+ /2
6. Show that f(t
0
) −f(c) ≤ g(t
0
) −g(c) + (t
0
−c).
7. Show that f(c) −f(a) ≤ g(c) −g(a) + (c −a) + .
8. Show that f(t
0
) −f(a) ≤ g(t
0
) −g(a) + (t
0
−a) + .
9. Show that c cannot be the supremum of E
unless c = b.
10. Show that f(b) −f(a) ≤ g(b) −g(a) + (b −a) + .
11. Conclude with the following:
Tutorial 18: The Jacobian Formula 13
Theorem 112 Let E be an Rnormed space. Let a, b ∈ R, a < b.
Let f : [a, b] → E and g : [a, b] → R be two continuous maps which
are diﬀerentiable at every point of ]a, b[, and such that:
∀t ∈]a, b[ , f
(t) ≤ g
(t)
Then:
f(b) −f(a) ≤ g(b) −g(a)
Deﬁnition 131 Let n ≥ 1 and U be open in R
n
. Let φ : U → E
be a map, where E is an Rnormed space. For all i = 1, . . . , n, we
say that φ has an ith partial derivative at a ∈ U, if and only if the
limit:
∂φ
∂x
i
(a)
= lim
h=0,h→0
φ(a + he
i
) −φ(a)
h
exists in E, where (e
1
, . . . , e
n
) is the canonical basis of R
n
.
Tutorial 18: The Jacobian Formula 14
Exercise 11. Let n ≥ 1 and U be open in R
n
. Let φ : U → E be a
map, where E is an Rnormed space.
1. Suppose φ is diﬀerentiable at a ∈ U. Show that for all i ∈ N
n
:
lim
h=0,h→0
1
he
i

φ(a + he
i
) −φ(a) −dφ(a)(he
i
) = 0
2. Show that for all i ∈ N
n
,
∂φ
∂x
i
(a) exists, and:
∂φ
∂x
i
(a) = dφ(a)(e
i
)
3. Conclude with the following:
Theorem 113 Let n ≥ 1 and U be open in R
n
. Let φ : U → E be
a map, where E is an Rnormed space. Then, if φ is diﬀerentiable at
a ∈ U, for all i = 1, . . . , n,
∂φ
∂x
i
(a) exists and we have:
∀h
= (h
1
, . . . , h
n
) ∈ R
n
, dφ(a)(h) =
n
i=1
∂φ
∂x
i
(a)h
i
Tutorial 18: The Jacobian Formula 15
Exercise 12. Let n ≥ 1 and U be open in R
n
. Let φ : U → E be a
map, where E is an Rnormed space.
1. Show that if φ is diﬀerentiable at a, b ∈ U, then for all i ∈ N
n
:
_
_
_
_
∂φ
∂x
i
(b) −
∂φ
∂x
i
(a)
_
_
_
_
≤ dφ(b) −dφ(a)
2. Conclude that if φ is of class C
1
on U, then
∂φ
∂x
i
exists and is
continuous on U, for all i ∈ N
n
.
Exercise 13. Let n ≥ 1 and U be open in R
n
. Let φ : U → E be a
map, where E is an Rnormed space. We assume that
∂φ
∂x
i
exists on
U, and is continuous at a ∈ U, for all i ∈ N
n
. We deﬁne l : R
n
→ E
by:
∀h
= (h
1
, . . . , h
n
) ∈ R
n
, l(h)
=
n
i=1
∂φ
∂x
i
(a)h
i
1. Show that l ∈ L
R
(R
n
, E).
Tutorial 18: The Jacobian Formula 16
2. Given > 0, show the existence of η > 0 such that for all h ∈ R
n
with h < η, we have a + h ∈ U, and:
∀i = 1, . . . , n ,
_
_
_
_
∂φ
∂x
i
(a + h) −
∂φ
∂x
i
(a)
_
_
_
_
≤
3. Let h = (h
1
, . . . , h
n
) ∈ R
n
be such that h < η. (e
1
, . . . , e
n
)
being the canonical basis of R
n
, we deﬁne k
0
= a and for i ∈ N
n
:
k
i
= a +
i
j=1
h
i
e
i
Show that k
0
, . . . , k
n
∈ U, and that we have:
φ(a+h)−φ(a)−l(h)=
n
i=1
_
φ(k
i−1
+ h
i
e
i
)−φ(k
i−1
)−h
i
∂φ
∂x
i
(a)
_
4. Let i ∈ N
n
. Assume that h
i
> 0. We deﬁne f
i
: [0, h
i
] → E by:
∀t ∈ [0, h
i
] , f
i
(t)
= φ(k
i−1
+ te
i
) −φ(k
i−1
) −t
∂φ
∂x
i
(a)
Tutorial 18: The Jacobian Formula 17
Show f
i
is welldeﬁned, f
i
(t) exists for all t ∈ [0, h
i
], and:
∀t ∈ [0, h
i
] , f
i
(t) =
∂φ
∂x
i
(k
i−1
+ te
i
) −
∂φ
∂x
i
(a)
5. Show f
i
is continuous on [0, h
i
], diﬀerentiable on ]0, h
i
[, with:
∀t ∈]0, h
i
[ , f
i
(t) ≤
6. Show that:
_
_
_
_
φ(k
i−1
+ h
i
e
i
) −φ(k
i−1
) −h
i
∂φ
∂x
i
(a)
_
_
_
_
≤ [h
i
[
7. Show that the previous inequality still holds if h
i
≤ 0.
8. Conclude that for all h ∈ R
n
with h < η, we have:
φ(a + h) −φ(a) −l(h) ≤
√
nh
9. Prove the following:
Tutorial 18: The Jacobian Formula 18
Theorem 114 Let n ≥ 1 and U be open in R
n
. Let φ : U → E be
a map, where E is an Rnormed space. If, for all i ∈ N
n
∂φ
∂x
i
exists
on U and is continuous at a ∈ U, then φ is diﬀerentiable at a ∈ U.
Exercise 14. Let n ≥ 1 and U be open in R
n
. Let φ : U → E be a
map, where E is an Rnormed space. We assume that for all i ∈ N
n
,
∂φ
∂x
i
exists and is continuous on U.
1. Show that φ is diﬀerentiable on U.
2. Show that for all a, b ∈ U and h ∈ R
n
:
(dφ(b) −dφ(a))(h) ≤
_
n
i=1
_
_
_
_
∂φ
∂x
i
(b) −
∂φ
∂x
i
(a)
_
_
_
_
2
_
1/2
h
3. Show that for all a, b ∈ U:
dφ(b) −dφ(a) ≤
_
n
i=1
_
_
_
_
∂φ
∂x
i
(b) −
∂φ
∂x
i
(a)
_
_
_
_
2
_
1/2
Tutorial 18: The Jacobian Formula 19
4. Show that dφ : U → L
R
(R
n
, E) is continuous.
5. Prove the following:
Theorem 115 Let n ≥ 1 and U be open in R
n
. Let φ : U → E be
a map, where E is an Rnormed space. Then, φ is of class C
1
on U,
if and only if for all i = 1, . . . , n,
∂φ
∂x
i
exists and is continuous on U.
Exercise 15. Let E, F be two Rnormed spaces and l ∈ L
R
(E, F).
Let U be open in E and l
U
be the restriction of l to U. Show that
l
U
is of class C
1
on U, and that we have:
∀x ∈ U , d(l
U
)(x) = l
Tutorial 18: The Jacobian Formula 20
Exercise 16. Let E
1
, . . . , E
p
, (p ≥ 1), be p Rnormed spaces. Let
E = E
1
. . . E
p
. For all x = (x
1
, . . . , x
p
) ∈ E, we deﬁne:
x
1
=
p
i=1
x
i

x
2
=
¸
¸
¸
_
p
i=1
x
i

2
x
∞
= max
i=1,...,p
x
i

1. Show that .
1
, .
2
and .
∞
are all norms on E.
2. Show .
1
, .
2
and .
∞
induce the product topology on E.
3. Conclude that E is also an Rnormed space, and that the norm
topology on E is exactly the product topology on E.
Tutorial 18: The Jacobian Formula 21
Exercise 17. Let E, F
1
, . . . , F
p
, (p ≥ 1) be p + 1 Rnormed spaces,
U be open in E, F = F
1
. . . F
p
and φ : U → F be a map.
1. For i = 1, . . . , p, let p
i
: F → F
i
be the canonical projection.
Show that p
i
∈ L
R
(F, F
i
). We put φ
i
= p
i
◦ φ.
2. For i = 1, . . . , p, let u
i
: F
i
→ F be deﬁned by:
∀x
i
∈ F
i
, u
i
(x
i
)
= (0, . . . ,
i
¸..¸
x
i
, . . . , 0)
Show that u
i
∈ L
R
(F
i
, F) and φ =
p
i=1
u
i
◦ φ
i
.
3. Show that if φ is diﬀerentiable at a ∈ U, then for all i = 1, . . . , p,
φ
i
: U → F
i
is diﬀerentiable at a ∈ U and dφ
i
(a) = p
i
◦ dφ(a).
4. Show that if for all i = 1, . . . , p, φ
i
is diﬀerentiable at a ∈ U,
then φ is diﬀerentiable at a ∈ U and:
dφ(a) =
p
i=1
u
i
◦ dφ
i
(a)
Tutorial 18: The Jacobian Formula 22
5. Suppose that φ is diﬀerentiable at a, b ∈ U. We assume that F
is given the norm (x
1
, . . . , x
p
)
2
=
_
p
i=1
x
i

2
. Show that
for all i ∈ N
p
:
dφ
i
(b) −dφ
i
(a) ≤ dφ(b) −dφ(a)
6. Show that:
dφ(b) −dφ(a) ≤
¸
¸
¸
_
p
i=1
dφ
i
(b) −dφ
i
(a)
2
7. Show that φ is of class C
1
⇔ φ
i
is of class C
1
for all i ∈ N
p
.
8. Explain why this conclusion would still hold, if F were given the
norm .
1
or .
∞
instead of .
2
.
9. Conclude with theorem (116)
Tutorial 18: The Jacobian Formula 23
Theorem 116 Let E, F
1
, . . . , F
p
, (p ≥ 1), be p+1 Rnormed spaces
and U be open in E. Let F be the Rnormed space F = F
1
. . . F
p
and φ = (φ
1
, . . . , φ
p
) : U → F be a map. Then, φ is diﬀerentiable at
a ∈ U, if and only if dφ
i
(a) exists for all i ∈ N
p
, in which case:
∀h ∈ E , dφ(a)(h) = (dφ
1
(a)(h), . . . , dφ
p
(a)(h))
Also, φ is of class C
1
on U ⇔ φ
i
is of class C
1
on U, for all i ∈ N
p
.
Theorem 117 Let φ = (φ
1
, . . . , φ
n
) : U → R
n
be a map, where
n ≥ 1 and U is open in R
n
. We assume that φ is diﬀerentiable at
a ∈ U. Then, for all i, j = 1, . . . , n,
∂φ
i
∂x
j
(a) exists, and we have:
dφ(a) =
_
_
_
∂φ
1
∂x
1
(a) . . .
∂φ
1
∂x
n
(a)
.
.
.
.
.
.
∂φ
n
∂x
1
(a) . . .
∂φ
n
∂x
n
(a)
_
_
_
Moreover, φ is of class C
1
on U, if and only if for all i, j = 1, . . . , n,
∂φ
i
∂x
j
exists and is continuous on U.
Tutorial 18: The Jacobian Formula 24
Exercise 18. Prove theorem (117)
Deﬁnition 132 Let φ = (φ
1
, . . . , φ
n
) : U → R
n
be a map, where
n ≥ 1 and U is open in R
n
. We assume that φ is diﬀerentiable at
a ∈ U. We call jacobian of φ at a, denoted J(φ)(a), the determinant
of the diﬀerential dφ(a) at a, i.e.
J(φ)(a) = det
_
_
_
∂φ
1
∂x
1
(a) . . .
∂φ
1
∂x
n
(a)
.
.
.
.
.
.
∂φ
n
∂x
1
(a) . . .
∂φ
n
∂x
n
(a)
_
_
_
Deﬁnition 133 Let n ≥ 1 and Ω, Ω
be open in R
n
. A bijection
φ : Ω → Ω
is called a C
1
diﬀeomorphism between Ω and Ω
, if and
only if φ : Ω → R
n
and φ
−1
: Ω
→ R
n
are both of class C
1
.
Tutorial 18: The Jacobian Formula 25
Exercise 19. Let Ω and Ω
be open in R
n
. Let φ : Ω → Ω
be a
C
1
diﬀeomorphism, ψ = φ
−1
, and I
n
be the identity mapping of R
n
.
1. Explain why J(ψ) : Ω
→ R and J(φ) : Ω → R are continuous.
2. Show that dφ(ψ(x)) ◦ dψ(x) = I
n
, for all x ∈ Ω
.
3. Show that dψ(φ(x)) ◦ dφ(x) = I
n
, for all x ∈ Ω.
4. Show that J(ψ)(x) ,= 0 for all x ∈ Ω
.
5. Show that J(φ)(x) ,= 0 for all x ∈ Ω.
6. Show that J(ψ) = 1/(J(φ) ◦ ψ) and J(φ) = 1/(J(ψ) ◦ φ).
Deﬁnition 134 Let n ≥ 1 and Ω ∈ B(R
n
), be a borel set in R
n
. We
deﬁne the lebesgue measure on Ω, denoted dx
Ω
, as the restriction
to B(Ω) of the lebesgue measure on R
n
, i.e the measure on (Ω, B(Ω))
deﬁned by:
∀B ∈ B(Ω) , dx
Ω
(B)
= dx(B)
Tutorial 18: The Jacobian Formula 26
Exercise 20. Show that dx
Ω
is a welldeﬁned measure on (Ω, B(Ω)).
Exercise 21. Let n ≥ 1 and Ω, Ω
be open in R
n
. Let φ : Ω → Ω
be a C
1
diﬀeomorphism and ψ = φ
−1
. Let a ∈ Ω
. We assume that
dψ(a) = I
n
, (identity mapping on R
n
), and given > 0, we denote:
B(a, )
= ¦x ∈ R
n
: a −x < ¦
where . is the usual norm in R
n
.
1. Why are dx
Ω
, φ(dx
Ω
) welldeﬁned measures on (Ω
, B(Ω
)).
2. Show that for > 0 suﬃciently small, B(a, ) ∈ B(Ω
).
3. Show that it makes sense to investigate whether the limit:
lim
↓↓0
φ(dx
Ω
)(B(a, ))
dx
Ω
(B(a, ))
does exists in R.
Tutorial 18: The Jacobian Formula 27
4. Given r > 0, show the existence of
1
> 0 such that for all
h ∈ R
n
with h ≤
1
, we have a + h ∈ Ω
, and:
ψ(a + h) −ψ(a) −h ≤ rh
5. Show for all h ∈ R
n
with h ≤
1
, we have a + h ∈ Ω
, and:
ψ(a + h) −ψ(a) ≤ (1 + r)h
6. Show that for all ∈]0,
1
[, we have B(a, ) ⊆ Ω
, and:
ψ(B(a, )) ⊆ B(ψ(a), (1 + r))
7. Show that dφ(ψ(a)) = I
n
.
8. Show the existence of
2
> 0 such that for all k ∈ R
n
with
k ≤
2
, we have ψ(a) + k ∈ Ω, and:
φ(ψ(a) + k) −a −k ≤ rk
Tutorial 18: The Jacobian Formula 28
9. Show for all k ∈ R
n
with k ≤
2
, we have ψ(a) +k ∈ Ω, and:
φ(ψ(a) + k) −a ≤ (1 + r)k
10. Show for all ∈]0,
2
(1 + r)[, we have B(ψ(a),
1+r
) ⊆ Ω, and:
B(ψ(a),
1 + r
) ⊆ ¦φ ∈ B(a, )¦
11. Show that if B(a, ) ⊆ Ω
, then ψ(B(a, )) = ¦φ ∈ B(a, )¦.
12. Show if 0 < <
0
=
1
∧
2
(1 + r), then B(a, ) ⊆ Ω
, and:
B(ψ(a),
1 + r
) ⊆ ¦φ ∈ B(a, )¦ ⊆ B(ψ(a), (1 + r))
13. Show that for all ∈]0,
0
[:
(i) dx(B(ψ(a),
1 + r
)) = (1 + r)
−n
dx
Ω
(B(a, ))
(ii) dx(B(ψ(a), (1 + r))) = (1 + r)
n
dx
Ω
(B(a, ))
(iii) dx(¦φ ∈ B(a, )¦) = φ(dx
Ω
)(B(a, ))
Tutorial 18: The Jacobian Formula 29
14. Show that for all ∈]0,
0
[, B(a, ) ⊆ Ω
, and:
(1 + r)
−n
≤
φ(dx
Ω
)(B(a, ))
dx
Ω
(B(a, ))
≤ (1 + r)
n
15. Conclude that:
lim
↓↓0
φ(dx
Ω
)(B(a, ))
dx
Ω
(B(a, ))
= 1
Exercise 22. Let n ≥ 1 and Ω, Ω
be open in R
n
. Let φ : Ω → Ω
be
a C
1
diﬀeomorphism and ψ = φ
−1
. Let a ∈ Ω
. We put A = dψ(a).
1. Show that A : R
n
→ R
n
is a linear bijection.
2. Deﬁne Ω
= A
−1
(Ω). Show that this deﬁnition does not depend
on whether A
−1
(Ω) is viewed as inverse , or direct image.
3. Show that Ω
is an open subset of R
n
.
Tutorial 18: The Jacobian Formula 30
4. We deﬁne
˜
φ : Ω
→ Ω
by
˜
φ(x) = φ ◦ A(x). Show that
˜
φ is a
C
1
diﬀeomorphism with
˜
ψ =
˜
φ
−1
= A
−1
◦ ψ.
5. Show that d
˜
ψ(a) = I
n
.
6. Show that:
lim
↓↓0
˜
φ(dx
Ω
)(B(a, ))
dx
Ω
(B(a, ))
= 1
7. Let > 0 with B(a, ) ⊆ Ω
. Justify each of the following steps:
˜
φ(dx
Ω
)(B(a, )) = dx
Ω
(¦
˜
φ ∈ B(a, )¦) (1)
= dx(¦
˜
φ ∈ B(a, )¦) (2)
= dx(¦x ∈ Ω
: φ ◦ A(x) ∈ B(a, )¦) (3)
= dx(¦x ∈ Ω
: A(x) ∈ φ
−1
(B(a, ))¦) (4)
= dx(¦x ∈ R
n
: A(x) ∈ φ
−1
(B(a, ))¦)(5)
= A(dx)(¦φ ∈ B(a, )¦) (6)
= [ det A[
−1
dx(¦φ ∈ B(a, )¦) (7)
Tutorial 18: The Jacobian Formula 31
= [ det A[
−1
dx
Ω
(¦φ ∈ B(a, )¦) (8)
= [ det A[
−1
φ(dx
Ω
)(B(a, )) (9)
8. Show that:
lim
↓↓0
φ(dx
Ω
)(B(a, ))
dx
Ω
(B(a, ))
= [ det A[
9. Conclude with the following:
Theorem 118 Let n ≥ 1 and Ω, Ω
be open in R
n
. Let φ : Ω → Ω
be a C
1
diﬀeomorphism and ψ = φ
−1
. Then, for all a ∈ Ω
, we have:
lim
↓↓0
φ(dx
Ω
)(B(a, ))
dx
Ω
(B(a, ))
= [J(ψ)(a)[
where J(ψ)(a) is the jacobian of ψ at a, B(a, ) is the open ball in R
n
,
and dx
Ω
, dx
Ω
are the lebesgue measures on Ω and Ω
respectively.
Tutorial 18: The Jacobian Formula 32
Exercise 23. Let n ≥ 1 and Ω, Ω
be open in R
n
. Let φ : Ω → Ω
be a C
1
diﬀeomorphism and ψ = φ
−1
.
1. Let K ⊆ Ω
be a compact subset of Ω
such that dx
Ω
(K) = 0.
Given > 0, show the existence of V open in Ω
, such that
K ⊆ V ⊆ Ω
, and dx
Ω
(V ) ≤ .
2. Explain why V is also open in R
n
.
3. Show that M = sup
x∈K
dψ(x) < +∞.
4. For all x ∈ K, show there is
x
> 0 such that B(x,
x
) ⊆ V , and
for all h ∈ R
n
with h ≤ 3
x
, we have x + h ∈ Ω
, and:
ψ(x + h) −ψ(x) ≤ (M + 1)h
5. Show that for all x ∈ K, B(x, 3
x
) ⊆ Ω
, and:
ψ(B(x, 3
x
)) ⊆ B(ψ(x), 3(M + 1)
x
)
6. Show that ψ(B(x, 3
x
)) = ¦φ ∈ B(x, 3
x
)¦, for all x ∈ K.
Tutorial 18: The Jacobian Formula 33
7. Show the existence of ¦x
1
, . . . , x
p
¦ ⊆ K, (p ≥ 0), such that:
K ⊆ B(x
1
,
x
1
) ∪ . . . ∪ B(x
p
,
x
p
)
8. Show the existence of S ⊆ ¦1, . . . , p¦ such that the B(x
i
,
x
i
)’s
are pairwise disjoint for i ∈ S, and:
K ⊆
_
i∈S
B(x
i
, 3
x
i
)
9. Show that ¦φ ∈ K¦ ⊆ ∪
i∈S
B(ψ(x
i
), 3(M + 1)
x
i
).
10. Show that φ(dx
Ω
)(K) ≤
i∈S
3
n
(M + 1)
n
dx(B(x
i
,
x
i
)).
11. Show that φ(dx
Ω
)(K) ≤ 3
n
(M + 1)
n
dx(V ).
12. Show that φ(dx
Ω
)(K) ≤ 3
n
(M + 1)
n
.
13. Conclude that φ(dx
Ω
)(K) = 0.
14. Show that φ(dx
Ω
) is a locally ﬁnite measure on (Ω
, B(Ω
)).
Tutorial 18: The Jacobian Formula 34
15. Let B ∈ B(Ω
) be such that dx
Ω
(B) = 0. Show that:
φ(dx
Ω
)(B) = sup¦φ(dx
Ω
)(K) : K ⊆ B , K compact ¦
16. Show that φ(dx
Ω
)(B) = 0.
17. Conclude with the following:
Theorem 119 Let n ≥ 1, Ω, Ω
be open in R
n
, and φ : Ω → Ω
be
a C
1
diﬀeomorphism. Then, the image measure φ(dx
Ω
), by φ of the
lebesgue measure on Ω, is absolutely continuous with respect to dx
Ω
,
the lebesgue measure on Ω
, i.e.:
φ(dx
Ω
) << dx
Ω
Tutorial 18: The Jacobian Formula 35
Exercise 24. Let n ≥ 1 and Ω, Ω
be open in R
n
. Let φ : Ω → Ω
be a C
1
diﬀeomorphism and ψ = φ
−1
.
1. Explain why there exists a sequence (V
p
)
p≥1
of open sets in Ω
,
such that V
p
↑ Ω
and for all p ≥ 1, the closure of V
p
in Ω
, i.e.
¯
V
Ω
p
, is compact.
2. Show that each V
p
is also open in R
n
, and that
¯
V
Ω
p
=
¯
V
p
.
3. Show that φ(dx
Ω
)(V
p
) < +∞, for all p ≥ 1.
4. Show that dx
Ω
and φ(dx
Ω
) are two σﬁnite measures on Ω
.
5. Show there is h : (Ω
, B(Ω
)) → (R
+
, B(R
+
)) measurable, with:
∀B ∈ B(Ω
) , φ(dx
Ω
)(B) =
_
B
hdx
Ω
6. For all p ≥ 1, we deﬁne h
p
= h1
V
p
, and we put:
∀x ∈ R
n
,
˜
h
p
(x)
=
_
h
p
(x) if x ∈ Ω
0 if x ,∈ Ω
Tutorial 18: The Jacobian Formula 36
Show that:
_
R
n
˜
h
p
dx =
_
Ω
h
p
dx
Ω
= φ(dx
Ω
)(V
p
) < +∞
and conclude that
˜
h
p
∈ L
1
R
(R
n
, B(R
n
), dx).
7. Show the existence of some N ∈ B(R
n
), such that dx(N) = 0
and for all x ∈ N
c
and p ≥ 1, we have:
˜
h
p
(x) = lim
↓↓0
1
dx(B(x, ))
_
B(x,)
˜
h
p
dx
8. Put N
= N ∩ Ω
. Show that N
∈ B(Ω
) and dx
Ω
(N
) = 0.
9. Let x ∈ Ω
and p ≥ 1 be such that x ∈ V
p
. Show that if > 0 is
such that B(x, ) ⊆ V
p
, then dx(B(x, )) = dx
Ω
(B(x, )), and:
_
B(x,)
˜
h
p
dx =
_
R
n
1
B(x,)
˜
h
p
dx =
_
Ω
1
B(x,)
h
p
dx
Ω
Tutorial 18: The Jacobian Formula 37
10. Show that:
_
Ω
1
B(x,)
h
p
dx
Ω
=
_
Ω
1
B(x,)
hdx
Ω
= φ(dx
Ω
)(B(x, ))
11. Show that for all x ∈ Ω
¸ N
, we have:
h(x) = lim
↓↓0
φ(dx
Ω
)(B(x, ))
dx
Ω
(B(x, ))
12. Show that h = [J(ψ)[ dx
Ω
a.s. and conclude with the following:
Theorem 120 Let n ≥ 1 and Ω, Ω
be open in R
n
. Let φ : Ω → Ω
be a C
1
diﬀeomorphism and ψ = φ
−1
. Then, the image measure by φ
of the lebesgue measure on Ω, is equal to the measure on (Ω
, B(Ω
))
with density [J(ψ)[ with respect to the lebesgue measure on Ω
, i.e.:
φ(dx
Ω
) =
_
[J(ψ)[dx
Ω
Tutorial 18: The Jacobian Formula 38
Exercise 25. Prove the following:
Theorem 121 (Jacobian Formula 1) Let n ≥ 1 and φ : Ω → Ω
be a C
1
diﬀeomorphism where Ω, Ω
are open in R
n
. Let ψ = φ
−1
.
Then, for all f : (Ω
, B(Ω
)) → [0, +∞] nonnegative and measurable:
_
Ω
f ◦ φdx
Ω
=
_
Ω
f[J(ψ)[dx
Ω
and:
_
Ω
(f ◦ φ)[J(φ)[dx
Ω
=
_
Ω
fdx
Ω
Exercise 26. Prove the following:
Tutorial 18: The Jacobian Formula 39
Theorem 122 (Jacobian Formula 2) Let n ≥ 1 and φ : Ω → Ω
be a C
1
diﬀeomorphism where Ω, Ω
are open in R
n
. Let ψ = φ
−1
.
Then, for all measurable map f : (Ω
, B(Ω
)) → (C, B(C)), we have
the equivalence:
f ◦ φ ∈ L
1
C
(Ω, B(Ω), dx
Ω
) ⇔ f[J(ψ)[ ∈ L
1
C
(Ω
, B(Ω
), dx
Ω
)
in which case:
_
Ω
f ◦ φdx
Ω
=
_
Ω
f[J(ψ)[dx
Ω
and, furthermore:
(f ◦ φ)[J(φ)[ ∈ L
1
C
(Ω, B(Ω), dx
Ω
) ⇔ f ∈ L
1
C
(Ω
, B(Ω
), dx
Ω
)
in which case:
_
Ω
(f ◦ φ)[J(φ)[dx
Ω
=
_
Ω
fdx
Ω
Tutorial 18: The Jacobian Formula 40
Exercise 27. Let f : R
2
→[0, +∞], with f(x, y) = exp(−(x
2
+y
2
)/2).
1. Show that:
_
R
2
f(x, y)dxdy =
__
+∞
−∞
e
−u
2
/2
du
_
2
2. Deﬁne:
∆
1
= ¦(x, y) ∈ R
2
: x > 0 , y > 0¦
∆
2
= ¦(x, y) ∈ R
2
: x < 0 , y > 0¦
and let ∆
3
and ∆
4
be the other two open quarters of R
2
. Show:
_
R
2
f(x, y)dxdy =
_
∆
1
∪...∪∆
4
f(x, y)dxdy
3. Let Q : R
2
→ R
2
be deﬁned by Q(x, y) = (−x, y). Show that:
_
∆
1
f(x, y)dxdy =
_
∆
2
f ◦ Q
−1
(x, y)dxdy
Tutorial 18: The Jacobian Formula 41
4. Show that:
_
R
2
f(x, y)dxdy = 4
_
∆
1
f(x, y)dxdy
5. Let D
1
=]0, +∞[]0, π/2[⊆ R
2
, and deﬁne φ : D
1
→ ∆
1
by:
∀(r, θ) ∈ D
1
, φ(r, θ)
= (r cos θ, r sin θ)
Show that φ is a bijection and that ψ = φ
−1
is given by:
∀(x, y) ∈ ∆
1
, ψ(x, y) = (
_
x
2
+ y
2
, arctan(y/x))
6. Show that φ is a C
1
diﬀeomorphism, with:
∀(r, θ) ∈ D
1
, dφ(r, θ) =
_
cos θ −r sin θ
sin θ r cos θ
_
and:
∀(x, y) ∈ ∆
1
, dψ(x, y) =
_
x
√
x
2
+y
2
y
√
x
2
+y
2
−y
x
2
+y
2
x
x
2
+y
2
_
Tutorial 18: The Jacobian Formula 42
7. Show that J(φ)(r, θ) = r, for all (r, θ) ∈ D
1
.
8. Show that J(ψ)(x, y) = 1/(
_
x
2
+ y
2
), for all (x, y) ∈ ∆
1
.
9. Show that:
_
∆
1
f(x, y)dxdy =
π
2
10. Prove the following:
Theorem 123 We have:
1
√
2π
_
+∞
−∞
e
−u
2
/2
du = 1
Tutorial 19: Fourier Transform 1
19. Fourier Transform
Exercise 1. We deﬁne the maps ψ : R
2
→ C and φ : R → C:
∀(u, x) ∈ R
2
, ψ(u, x)
= e
iux−x
2
/2
∀u ∈ R , φ(u)
=
+∞
−∞
ψ(u, x)dx
1. Show that for all u ∈ R, the map x → ψ(u, x) is measurable.
2. Show that for all u ∈ R, we have:
+∞
−∞
[ψ(u, x)[dx =
√
2π < +∞
and conclude that φ is well deﬁned.
3. Let u ∈ R and (u
n
)
n≥1
be a sequence in R converging to u.
Show that φ(u
n
) → φ(u) and conclude that φ is continuous.
Tutorial 19: Fourier Transform 2
4. Show that:
+∞
0
xe
−x
2
/2
dx = 1
5. Show that for all u ∈ R, we have:
+∞
−∞
∂ψ
∂u
(u, x)
dx = 2 < +∞
6. Let a, b ∈ R, a < b. Show that:
e
ib
−e
ia
=
b
a
ie
ix
dx
7. Let a, b ∈ R, a < b. Show that:
[e
ib
−e
ia
[ ≤ [b −a[
8. Let a, b ∈ R, a ,= b. Show that for all x ∈ R:
ψ(b, x) −ψ(a, x)
b −a
≤ [x[e
−x
2
/2
Tutorial 19: Fourier Transform 3
9. Let u ∈ R and (u
n
)
n≥1
be a sequence in R converging to u,
with u
n
,= u for all n. Show that:
lim
n→+∞
φ(u
n
) −φ(u)
u
n
−u
=
+∞
−∞
∂ψ
∂u
(u, x)dx
10. Show that φ is diﬀerentiable with:
∀u ∈ R , φ
(u) =
+∞
−∞
∂ψ
∂u
(u, x)dx
11. Show that φ is of class C
1
.
12. Show that for all (u, x) ∈ R
2
, we have:
∂ψ
∂u
(u, x) = −uψ(u, x) −i
∂ψ
∂x
(u, x)
13. Show that for all u ∈ R:
+∞
−∞
∂ψ
∂x
(u, x)
dx < +∞
Tutorial 19: Fourier Transform 4
14. Let a, b ∈ R, a < b. Show that for all u ∈ R:
ψ(u, b) −ψ(u, a) =
b
a
∂ψ
∂x
(u, x)dx
15. Show that for all u ∈ R:
+∞
−∞
∂ψ
∂x
(u, x)dx = 0
16. Show that for all u ∈ R:
φ
(u) = −uφ(u)
Exercise 2. Let o be the set of functions deﬁned by:
o
= ¦h : h ∈ C
1
(R, R) , ∀u ∈ R , h
(u) = −uh(u)¦
1. Let φ be as in ex. (1). Show that Re(φ) and Im(φ) lie in o.
Tutorial 19: Fourier Transform 5
2. Given h ∈ o, we deﬁne g : R → R, by:
∀u ∈ R , g(u)
= h(u)e
u
2
/2
Show that g is of class C
1
with g
= 0.
3. Let a, b ∈ R, a < b. Show the existence of c ∈]a, b[, such that:
g(b) −g(a) = g
(c)(b −a)
4. Conclude that for all h ∈ o, we have:
∀u ∈ R , h(u) = h(0)e
−u
2
/2
5. Prove the following:
Theorem 124 For all u ∈ R, we have:
1
√
2π
+∞
−∞
e
iux−x
2
/2
dx = e
−u
2
/2
Tutorial 19: Fourier Transform 6
Deﬁnition 135 Let µ
1
, . . . , µ
p
be complex measures on R
n
,
1
where
n, p ≥ 1. We call convolution of µ
1
, . . . , µ
p
, denoted µ
1
. . . µ
p
, the
image measure of the product measure µ
1
⊗. . . ⊗µ
p
by the measurable
map S : (R
n
)
p
→ R
n
deﬁned by:
S(x
1
, . . . , x
p
)
= x
1
+. . . +x
p
In other words, µ
1
. . . µ
p
is the complex measure on R
n
, deﬁned by:
µ
1
. . . µ
p
= S(µ
1
⊗. . . ⊗µ
p
)
Exercise 3. Let µ, ν be complex measures on R
n
.
1. Show that for all B ∈ B(R
n
):
µ ν(B) =
R
n
×R
n
1
B
(x +y)dµ ⊗ν(x, y)
1
An obvious shortcut to saying ’complex measures on (R
n
, B(R
n
))’.
Tutorial 19: Fourier Transform 7
2. Show that for all B ∈ B(R
n
):
µ ν(B) =
R
n
R
n
1
B
(x +y)dµ(x)
dν(y)
3. Show that for all B ∈ B(R
n
):
µ ν(B) =
R
n
R
n
1
B
(x +y)dν(x)
dµ(y)
4. Show that µ ν = ν µ.
5. Let f : R
n
→ C be bounded and measurable. Show that:
R
n
fdµ ν =
R
n
×R
n
f(x + y)dµ ⊗ν(x, y)
Tutorial 19: Fourier Transform 8
Exercise 4. Let µ, ν be complex measures on R
n
. Given B ⊆ R
n
and x ∈ R
n
, we deﬁne B −x = ¦y ∈ R
n
, y +x ∈ B¦.
1. Show that for all B ∈ B(R
n
) and x ∈ R
n
, B −x ∈ B(R
n
).
2. Show x → µ(B−x) is measurable and bounded, for B ∈ B(R
n
).
3. Show that for all B ∈ B(R
n
):
µ ν(B) =
R
n
µ(B −x)dν(x)
4. Show that for all B ∈ B(R
n
):
µ ν(B) =
R
n
ν(B −x)dµ(x)
Tutorial 19: Fourier Transform 9
Exercise 5. Let µ
1
, µ
2
, µ
3
be complex measures on R
n
.
1. Show that for all B ∈ B(R
n
):
µ
1
(µ
2
µ
3
)(B) =
R
n
×R
n
1
B
(x +y)dµ
1
⊗(µ
2
µ
3
)(x, y)
2. Show that for all B ∈ B(R
n
) and x ∈ R
n
:
R
n
1
B
(x +y)dµ
2
µ
3
(y) =
R
n
×R
n
1
B
(x +y +z)dµ
2
⊗µ
3
(y, z)
3. Show that for all B ∈ B(R
n
):
µ
1
(µ
2
µ
3
)(B) =
R
n
×R
n
×R
n
1
B
(x +y +z)dµ
1
⊗µ
2
⊗µ
3
(x, y, z)
4. Show that µ
1
(µ
2
µ
3
) = µ
1
µ
2
µ
3
= (µ
1
µ
2
) µ
3
Tutorial 19: Fourier Transform 10
Deﬁnition 136 Let n ≥ 1 and a ∈ R
n
. We deﬁne δ
a
: B(R
n
)→R
+
:
∀B ∈ B(R
n
) , δ
a
(B)
= 1
B
(a)
δ
a
is called the dirac probability measure on R
n
, centered in a.
Exercise 6. Let n ≥ 1 and a ∈ R
n
.
1. Show that δ
a
is indeed a probability measure on R
n
.
2. Show for all f : R
n
→ [0, +∞] nonnegative and measurable:
R
n
fdδ
a
= f(a)
3. Show if f : R
n
→ C is measurable, f ∈L
1
C
(R
n
, B(R
n
), δ
a
) and:
R
n
fdδ
a
= f(a)
Tutorial 19: Fourier Transform 11
4. Show that for any complex measure µ on R
n
:
µ δ
0
= δ
0
µ = µ
5. Let τ
a
(x) = a+x deﬁne the translation of vector a in R
n
. Show
that for any complex measure µ on R
n
:
µ δ
a
= δ
a
µ = τ
a
(µ)
Exercise 7. Let n ≥ 1 and µ, ν be complex measures on R
n
. We
assume that ν << dx, i.e. that ν is absolutely continuous with respect
to the lebesgue measure on R
n
.
1. Show there is f ∈ L
1
C
(R
n
, B(R
n
), dx), such that ν =
fdx.
2. Show that for all B ∈ B(R
n
), we have:
µ ν(B) =
R
n
ν(B −x)dµ(x)
Tutorial 19: Fourier Transform 12
3. Show that for all B ∈ B(R
n
) and x ∈ R
n
:
ν(B −x) =
R
n
1
B
(y)f(y −x)dy
4. Show that for all B ∈ B(R
n
) the map:
(x, y) → 1
B
(y)f(y −x)
lies in L
1
C
(R
n
R
n
, B(R
n
) ⊗B(R
n
), µ ⊗dy).
5. Show that for all B ∈ B(R
n
), we have:
µ ν(B) =
B
R
n
f(y −x)dµ(x)
dy
6. Given y ∈ R
n
, we deﬁne:
g(y)
=
R
n
f(y −x)dµ(x)
Show that g(y) is welldeﬁned for dyalmost all y ∈ R
n
.
Tutorial 19: Fourier Transform 13
7. Deﬁne an element ¯ g of L
1
C
(R
n
, B(R
n
), dx), with g = ¯ g dx−a.s.
8. Show that µ ν is absolutely continuous w.r. to the lebesgue
measure on R
n
, with density g.
Theorem 125 Let µ, ν be two complex measures on R
n
, n ≥ 1. If
ν << dx, i.e. ν is absolutely continuous with respect to the lebesgue
measure on R
n
, with density f∈ L
1
C
(R
n
, B(R
n
), dx), then the convo
lution µ ν = ν µ is itself absolutely continuous with respect to the
lebesgue measure on R
n
, with density:
g(y) =
R
n
f(y −x)dµ(x) , dy −a.s.
In other words, µ ν = ν µ =
gdx.
Exercise 8. Further to theorem (125), show that if µ =
hdx for
some h ∈ L
1
C
(R
n
, B(R
n
), dx), then:
g(y) =
R
n
f(y −x)h(x)dx , dy −a.s.
Tutorial 19: Fourier Transform 14
Deﬁnition 137 Let µ be a complex measure on (R
n
, B(R
n
)), n ≥ 1.
We call fourier transform of µ, the map Tµ : R
n
→ C deﬁned by:
∀u ∈ R
n
, Tµ(u)
=
R
n
e
iu,x
dµ(x)
where ¸, ) is the usual innerproduct in R
n
.
Exercise 9. Further to deﬁnition (137):
1. Show that Tµ is welldeﬁned.
2. Show that Tµ ∈ C
b
C
(R
n
), i.e Tµ is continuous and bounded.
3. Show that for all a, u ∈ R
n
, we have ∀u ∈ R
n
, Tδ
a
(u) = e
iu,a
.
4. Let µ be the probability measure on (R, B(R)) deﬁned by:
∀B ∈ B(R) , µ(B)
=
1
√
2π
B
e
−x
2
/2
dx
Show that Tµ(u) = e
−u
2
/2
, for all u ∈ R.
Tutorial 19: Fourier Transform 15
Exercise 10. Let µ
1
, . . . , µ
p
be complex measures on R
n
, n, p ≥ 1.
1. Show that for all u ∈ R
n
, we have:
T(µ
1
. . .µ
p
)(u) =
(R
n
)
p
e
iu,x
1
+...+x
p
dµ
1
⊗. . .⊗µ
p
(x
1
, . . . , x
p
)
2. Show that T(µ
1
. . . µ
p
) = Π
p
j=1
Tµ
j
.
Exercise 11. Let n ≥ 1, σ > 0 and g
σ
: R
n
→ R
+
deﬁned by:
∀x ∈ R
n
, g
σ
(x)
=
1
(2π)
n
2
σ
n
e
−x
2
/2σ
2
1. Show that:
R
n
g
σ
(x)dx = 1
2. Show that for all u ∈ R
n
, we have:
R
n
g
σ
(x)e
iu,x
dx = e
−σ
2
u
2
/2
Tutorial 19: Fourier Transform 16
3. Show that P
σ
=
g
σ
dx is a probability on R
n
with fourier
transform:
∀u ∈ R
n
, TP
σ
(u) = e
−σ
2
u
2
/2
4. Show that for all x ∈ R
n
, we have:
g
σ
(x) =
1
(2π)
n
R
n
e
ix,u−σ
2
u
2
/2
du
Exercise 12. Further to ex. (11), let µ be a complex measure on R
n
.
1. Show that µ P
σ
=
φ
σ
dx where:
φ
σ
(x) =
R
n
g
σ
(x −y)dµ(y) , dx −a.s.
2. Show that we also have:
φ
σ
(x) =
R
n
g
σ
(y −x)dµ(y) , dx −a.s.
Tutorial 19: Fourier Transform 17
3. Show that:
φ
σ
(x) =
1
(2π)
n
R
n
R
n
e
iy−x,u−σ
2
u
2
/2
du
dµ(y) , dx −a.s.
4. Show that:
φ
σ
(x) =
1
(2π)
n
R
n
e
−ix,u−σ
2
u
2
/2
(Tµ)(u)du
5. Show that if µ, ν are two complex measures on R
n
such that
Tµ = Tν, then for all σ > 0, we have µ P
σ
= ν P
σ
.
Deﬁnition 138 Let (Ω, T ) be a topological space. Let (µ
k
)
k≥1
be a
sequence of complex measures on (Ω, B(Ω)). We say that the sequence
(µ
k
)
k≥1
narrowly converges to a complex measure µ on (Ω, B(Ω)),
and we write µ
k
→ µ narrowly, if and only if:
∀f ∈ C
b
R
(Ω) , lim
k→+∞
fdµ
k
=
fdµ
Tutorial 19: Fourier Transform 18
Exercise 13. Further to deﬁnition (138):
1. Show that µ
k
→ µ narrowly, is equivalent to:
∀f ∈ C
b
C
(Ω) , lim
k→+∞
fdµ
k
=
fdµ
2. Show that if (Ω, T ) is metrizable and ν is a complex measure on
(Ω, B(Ω)) such that µ
k
→ µ and µ
k
→ ν narrowly, then µ = ν.
Theorem 126 On a metrizable topological space, the narrow limit
when it exists, of any sequence of complex measures, is unique.
Exercise 14.
1. Show that on (R, B(R)), we have δ
1/n
→ δ
0
narrowly.
2. Show there is B ∈ B(R), such that δ
1/n
(B) ,→ δ
0
(B).
Tutorial 19: Fourier Transform 19
Exercise 15. Let n ≥ 1. Given σ > 0, let P
σ
be the probability
measure on (R
n
, B(R
n
)) deﬁned as in ex. (11). Let (σ
k
)
k≥1
be a
sequence in R
+
such that σ
k
> 0 and σ
k
→ 0.
1. Show that for all f ∈ C
b
R
(R
n
), we have:
R
n
f(x)g
σ
k
(x)dx =
1
(2π)
n
2
R
n
f(σ
k
x)e
−x
2
/2
dx
2. Show that for all f ∈ C
b
R
(R
n
), we have:
lim
k→+∞
R
n
f(x)g
σ
k
(x)dx = f(0)
3. Show that P
σ
k
→ δ
0
narrowly.
Tutorial 19: Fourier Transform 20
Exercise 16. Let µ, ν be two complex measures on R
n
. Let (ν
k
)
k≥1
be a sequence of complex measures on R
n
, which narrowly converges
to ν. Let f ∈ C
b
R
(R
n
), and φ : R
n
→ R be deﬁned by:
∀y ∈ R
n
, φ(y)
=
R
n
f(x +y)dµ(x)
1. Show that:
R
n
fdµ ν
k
=
R
n
×R
n
f(x + y)dµ ⊗ν
k
(x, y)
2. Show that:
R
n
fdµ ν
k
=
R
n
φdν
k
3. Show that φ ∈ C
b
C
(R
n
).
4. Show that:
lim
k→+∞
R
n
φdν
k
=
R
n
φdν
Tutorial 19: Fourier Transform 21
5. Show that:
lim
k→+∞
R
n
fdµ ν
k
=
R
n
fdµ ν
6. Show that µ ν
k
→ µ ν narrowly.
Theorem 127 Let µ, ν be two complex measures on R
n
, n ≥ 1. Let
(ν
k
)
k≥1
be a sequence of complex measures on R
n
. Then:
ν
k
→ ν narrowly ⇒ µ ν
k
→ µ ν narrowly
Exercise 17. Let µ, ν be two complex measures on R
n
, such that
Tµ = Tν. For all σ > 0, let P
σ
be the probability measure on
(R
n
, B(R
n
)) as deﬁned in ex. (11). Let (σ
k
)
k≥1
be a sequence in R
+
such that σ
k
> 0 and σ
k
→ 0.
1. Show that µ P
σ
k
= ν P
σ
k
, for all k ≥ 1.
2. Show that µ P
σ
k
→ µ δ
0
narrowly.
Tutorial 19: Fourier Transform 22
3. Show that (µ P
σ
k
)
k≥1
narrowly converges to both µ and ν.
4. Prove the following:
Theorem 128 Let µ, ν be two complex measures on R
n
. Then:
Tµ = Tν ⇒ µ = ν
i.e. the fourier transform is an injective mapping on M
1
(R
n
, B(R
n
)).
Deﬁnition 139 Let (Ω, T, P) be a probability space. Given n ≥ 1,
and a measurable map X : (Ω, T) → (R
n
, B(R
n
)), the mapping φ
X
deﬁned as:
∀u ∈ R
n
, φ
X
(u)
= E[e
iu,X
]
is called the characteristic function
2
of the random variable X.
2
Do not confuse with the characteristic function 1
A
of a set A, deﬁnition (39).
Tutorial 19: Fourier Transform 23
Exercise 18. Further to deﬁnition (139):
1. Show that φ
X
is welldeﬁned, bounded and continuous.
2. Show that we have:
∀u ∈ R
n
, φ
X
(u) =
R
n
e
iu,x
dX(P)(x)
3. Show φ
X
is the fourier transform of the image measure X(P).
4. Show the following:
Theorem 129 Let X, Y : (Ω, T) → (R
n
, B(R
n
)), n ≥ 1, be two
random variables on a probability space (Ω, T, P). If X and Y have
the same characteristic functions, i.e.
∀u ∈ R
n
, E[e
iu,X
] = E[e
iu,Y
]
then X and Y have the same distributions, i.e.
∀B ∈ B(R
n
) , P(¦X ∈ B¦) = P(¦Y ∈ B¦)
Tutorial 19: Fourier Transform 24
Deﬁnition 140 Let n ≥ 1. Given α = (α
1
, . . . , α
n
) ∈ N
n
, we deﬁne
the modulus of α, denoted [α[, by [α[ = α
1
+. . . +α
n
. Given x ∈ R
n
and α ∈ N
n
, we put:
x
α
= x
α
1
1
. . . x
α
n
n
where it is understood that x
α
j
j
= 1 whenever α
j
= 0. Given a map
f : U → C, where U is an open subset of R
n
, we denote ∂
α
f the
[α[th partial derivative, when it exists:
∂
α
f
=
∂
α
f
∂x
α
1
1
. . . ∂x
α
n
n
Note that ∂
α
f = f, whenever [α[ = 0. Given k ≥ 0, we say that f is
of class C
k
, if and only if for all α ∈ N
n
with [α[ ≤ k, ∂
α
f exists
and is continuous on U.
Exercise 19. Explain why def. (140) is consistent with def. (130).
Tutorial 19: Fourier Transform 25
Exercise 20. Let µ be a complex measure on R
n
, and α ∈ N
n
, with:
R
n
[x
α
[d[µ[(x) < +∞ (1)
Let x
α
µ the complex measure on R
n
deﬁned by x
α
µ =
x
α
dµ.
1. Explain why the above integral (1) is welldeﬁned.
2. Show that x
α
µ is a welldeﬁned complex measure on R
n
.
3. Show that the total variation of x
α
µ is given by:
∀B ∈ B(R
n
) , [x
α
µ[(B) =
B
[x
α
[d[µ[(x)
4. Show that the fourier transform of x
α
µ is given by:
∀u ∈ R
n
, T(x
α
µ)(u) =
R
n
x
α
e
iu,x
dµ(x)
Tutorial 19: Fourier Transform 26
Exercise 21. Let µ be a complex measure on R
n
. Let β ∈ N
n
with
[β[ = 1, and:
R
n
[x
β
[d[µ[(x) < +∞
Let x
β
µ be the complex measure on R
n
deﬁned as in ex. (20).
1. Show that there is j ∈ N
n
with x
β
= x
j
for all x ∈ R
n
.
2. Show that for all u ∈ R
n
,
∂Fµ
∂u
j
(u) exists and that we have:
∂Tµ
∂u
j
(u) = i
R
n
x
j
e
iu,x
dµ(x)
3. Conclude that ∂
β
Tµ exists and that we have:
∂
β
Tµ = iT(x
β
µ)
4. Explain why ∂
β
Tµ is continuous.
Tutorial 19: Fourier Transform 27
Exercise 22. Let µ be a complex measure on R
n
. Let k ≥ 0 be an
integer. We assume that for all α ∈ N
n
, we have:
[α[ ≤ k ⇒
R
n
[x
α
[d[µ[(x) < +∞ (2)
In particular, if [α[ ≤ k, the measure x
α
µ of ex. (20) is welldeﬁned.
We claim that for all α ∈ N
n
with [α[ ≤ k, ∂
α
Tµ exists, and:
∂
α
Tµ = i
α
T(x
α
µ)
1. Show that if k = 0, then the property is obviously true. We
assume the property is true for some k ≥ 0, and that the above
integrability condition (2) holds for k + 1.
2. Let α
∈ N
n
be such that [α
[ ≤ k + 1. Explain why if [α
[ ≤ k,
then ∂
α
Tµ exists, with:
∂
α
Tµ = i
α

T(x
α
µ)
3. We assume that [α
[ = k + 1. Show the existence of α, β ∈ N
n
such that α +β = α
, [α[ = k and [β[ = 1.
Tutorial 19: Fourier Transform 28
4. Explain why ∂
α
Tµ exists, and:
∂
α
Tµ = i
α
T(x
α
µ)
5. Show that:
R
n
[x
β
[d[x
α
µ[(x) < +∞
6. Show that ∂
β
T(x
α
µ) exists, with:
∂
β
T(x
α
µ) = iT(x
β
(x
α
µ))
7. Show that ∂
β
(∂
α
Tµ) exists, with:
∂
β
(∂
α
Tµ) = i
α+1
T(x
β
(x
α
µ))
8. Show that x
β
(x
α
µ) = x
α
µ.
9. Conclude that the property is true for k + 1.
10. Show the following:
Tutorial 19: Fourier Transform 29
Theorem 130 Let µ be a complex measure on R
n
, n ≥ 1. Let k ≥ 0
be an integer such that for all α ∈ N
n
with [α[ ≤ k, we have:
R
n
[x
α
[d[µ[(x) < +∞
Then, the fourier transform Tµ is of class C
k
on R
n
, and for all
α ∈ N
n
with [α[ ≤ k, we have:
∀u ∈ R
n
, ∂
α
Tµ(u) = i
α
R
n
x
α
e
iu,x
dµ(x)
In particular:
R
n
x
α
dµ(x) = i
−α
∂
α
Tµ(0)
Tutorial 20: Gaussian Measures 1
20. Gaussian Measures
/
n
(R) is the set of all n nmatrices with real entries, n ≥ 1.
Deﬁnition 141 A matrix M ∈ /
n
(R) is said to be symmetric,
if and only if M = M
t
. M is orthogonal, if and only if M is
nonsingular and M
−1
= M
t
. If M is symmetric, we say that M is
nonnegative, if and only if:
∀u ∈ R
n
, ¸u, Mu) ≥ 0
Theorem 131 Let Σ ∈ /
n
(R), n ≥ 1, be a symmetric and non
negative real matrix. There exist λ
1
, . . . , λ
n
∈ R
+
and P ∈ /
n
(R)
orthogonal matrix, such that:
Σ = P.
λ
1
0
.
.
.
0 λ
n
. P
t
In particular, there exists A ∈ /
n
(R) such that Σ = A.A
t
.
Tutorial 20: Gaussian Measures 2
As a rare exception, theorem (131) is given without proof.
Exercise 1. Given n ≥ 1 and M ∈ /
n
(R), show that we have:
∀u, v ∈ R
n
, ¸u, Mv) = ¸M
t
u, v)
Exercise 2. Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be a symmetric
and nonnegative matrix. Let µ
1
be the probability measure on R:
∀B ∈ B(R) , µ
1
(B) =
1
√
2π
B
e
−x
2
/2
dx
Let µ = µ
1
⊗. . . ⊗µ
1
be the product measure on R
n
. Let A ∈ /
n
(R)
be such that Σ = A.A
t
. We deﬁne the map φ : R
n
→ R
n
by:
∀x ∈ R
n
, φ(x)
= Ax +m
1. Show that µ is a probability measure on (R
n
, B(R
n
)).
2. Explain why the image measure P = φ(µ) is welldeﬁned.
3. Show that P is a probability measure on (R
n
, B(R
n
)).
Tutorial 20: Gaussian Measures 3
4. Show that for all u ∈ R
n
:
TP(u) =
R
n
e
iu,φ(x)
dµ(x)
5. Let v = A
t
u. Show that for all u ∈ R
n
:
TP(u) = e
iu,m−v
2
/2
6. Show the following:
Theorem 132 Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be a sym
metric and nonnegative real matrix. There exists a unique complex
measure on R
n
, denoted N
n
(m, Σ), with fourier transform:
TN
n
(m, Σ)(u)
=
R
n
e
iu,x
dN
n
(m, Σ)(x) = e
iu,m−
1
2
u,Σu
for all u ∈ R
n
. Furthermore, N
n
(m, Σ) is a probability measure.
Tutorial 20: Gaussian Measures 4
Deﬁnition 142 Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be
a symmetric and nonnegative real matrix. The probability measure
N
n
(m, Σ) on R
n
deﬁned in theorem (132) is called the ndimensional
gaussian measure or normal distribution, with mean m ∈ R
n
and covariance matrix Σ.
Exercise 3. Let n ≥ 1 and m ∈ R
n
. Show that N
n
(m, 0) = δ
m
.
Exercise 4. Let m ∈ R
n
. Let Σ ∈ /
n
(R) be a symmetric and
nonnegative real matrix. Let A ∈ /
n
(R) be such that Σ = A.A
t
.
A map p : R
n
→ C is said to be a polynomial, if and only if, it is a
ﬁnite linear complex combination of maps x → x
α
,
1
for α ∈ N
n
.
1. Show that for all B ∈ B(R), we have:
N
1
(0, 1)(B) =
1
√
2π
B
e
−x
2
/2
dx
1
See deﬁnition (140).
Tutorial 20: Gaussian Measures 5
2. Show that:
+∞
−∞
[x[dN
1
(0, 1)(x) < +∞
3. Show that for all integer k ≥ 1:
1
√
2π
+∞
0
x
k+1
e
−x
2
/2
dx =
k
√
2π
+∞
0
x
k−1
e
−x
2
/2
dx
4. Show that for all integer k ≥ 0:
+∞
−∞
[x[
k
dN
1
(0, 1)(x) < +∞
5. Show that for all α ∈ N
n
:
R
n
[x
α
[dN
1
(0, 1) ⊗. . . ⊗N
1
(0, 1)(x) < +∞
Tutorial 20: Gaussian Measures 6
6. Let p : R
n
→ C be a polynomial. Show that:
R
n
[p(x)[dN
1
(0, 1) ⊗. . . ⊗N
1
(0, 1)(x) < +∞
7. Let φ : R
n
→ R
n
be deﬁned by φ(x) = Ax + m. Explain why
the image measure φ(N
1
(0, 1) ⊗. . . ⊗N
1
(0, 1)) is welldeﬁned.
8. Show that φ(N
1
(0, 1) ⊗. . . ⊗ N
1
(0, 1)) = N
n
(m, Σ).
9. Show if β ∈ N
n
and [β[ = 1, then x → φ(x)
β
is a polynomial.
10. Show that if α
∈ N
n
and [α
[ = k+1, then φ(x)
α
= φ(x)
α
φ(x)
β
for some α, β ∈ N
n
such that [α[ = k and [β[ = 1.
11. Show that the product of two polynomials is a polynomial.
12. Show that for all α ∈ N
n
, x → φ(x)
α
is a polynomial.
13. Show that for all α ∈ N
n
:
R
n
[φ(x)
α
[dN
1
(0, 1) ⊗. . . ⊗N
1
(0, 1)(x) < +∞
Tutorial 20: Gaussian Measures 7
14. Show the following:
Theorem 133 Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be a sym
metric and nonnegative real matrix. Then, for all α ∈ N
n
, the map
x → x
α
is integrable with respect to the gaussian measure N
n
(m, Σ):
R
n
[x
α
[dN
n
(m, Σ)(x) < +∞
Exercise 5. Let m ∈ R
n
. Let Σ = (σ
ij
) ∈ /
n
(R) be a symmetric
and nonnegative real matrix. Let j, k ∈ N
n
. Let φ be the fourier
transform of the gaussian measure N
n
(m, Σ), i.e.:
∀u ∈ R
n
, φ(u)
= e
iu,m−
1
2
u,Σu
1. Show that:
R
n
x
j
dN
n
(m, Σ)(x) = i
−1
∂φ
∂u
j
(0)
Tutorial 20: Gaussian Measures 8
2. Show that:
R
n
x
j
dN
n
(m, Σ)(x) = m
j
3. Show that:
R
n
x
j
x
k
dN
n
(m, Σ)(x) = i
−2
∂
2
φ
∂u
j
∂u
k
(0)
4. Show that:
R
n
x
j
x
k
dN
n
(m, Σ)(x) = σ
jk
−m
j
m
k
5. Show that:
R
n
(x
j
−m
j
)(x
k
− m
k
)dN
n
(m, Σ)(x) = σ
jk
Tutorial 20: Gaussian Measures 9
Theorem 134 Let n ≥ 1 and m ∈ R
n
. Let Σ = (σ
ij
) ∈ /
n
(R)
be a symmetric and nonnegative real matrix. Let N
n
(m, Σ) be the
gaussian measure with mean m and covariance matrix Σ. Then, for
all j, k ∈ N
n
, we have:
R
n
x
j
dN
n
(m, Σ)(x) = m
j
and:
R
n
(x
j
−m
j
)(x
k
−m
k
)dN
n
(m, Σ)(x) = σ
jk
Deﬁnition 143 Let n ≥ 1. Let (Ω, T, P) be a probability space. Let
X : (Ω, T) → (R
n
, B(R
n
)) be a measurable map. We say that X
is an ndimensional gaussian or normal vector, if and only if its
distribution is a gaussian measure, i.e. X(P) = N
n
(m, Σ) for some
m ∈ R
n
and Σ ∈ /
n
(R) symmetric and nonnegative real matrix.
Exercise 6. Show the following:
Tutorial 20: Gaussian Measures 10
Theorem 135 Let n ≥ 1. Let (Ω, T, P) be a probability space. Let
X : (Ω, T) → R
n
be a measurable map. Then X is a gaussian vector,
if and only if there exist m ∈ R
n
and Σ ∈ /
n
(R) symmetric and
nonnegative real matrix, such that:
∀u ∈ R
n
, E[e
iu,X
] = e
iu,m−
1
2
u,Σu
where ¸, ) is the usual innerproduct on R
n
.
Deﬁnition 144 Let X : (Ω, T) →
¯
R (or C) be a random variable
on a probability space (Ω, T, P). We say that X is integrable, if and
only if we have E[[X[] < +∞. We say that X is squareintegrable,
if and only if we have E[[X[
2
] < +∞.
Exercise 7. Further to deﬁnition (144), suppose X is Cvalued.
1. Show X is integrable if and only if X ∈ L
1
C
(Ω, T, P).
2. Show X is squareintegrable, if and only if X ∈ L
2
C
(Ω, T, P).
Tutorial 20: Gaussian Measures 11
Exercise 8. Further to deﬁnition (144), suppose X is
¯
Rvalued.
1. Show that X is integrable, if and only if X is Palmost surely
equal to an element of L
1
R
(Ω, T, P).
2. Show that X is squareintegrable, if and only if X is Palmost
surely equal to an element of L
2
R
(Ω, T, P).
Exercise 9. Let X, Y : (Ω, T) → (R, B(R)) be two squareintegrable
random variables on a probability space (Ω, T, P).
1. Show that both X and Y are integrable.
2. Show that XY is integrable
3. Show that (X−E[X])(Y −E[Y ]) is a welldeﬁned and integrable.
Tutorial 20: Gaussian Measures 12
Deﬁnition 145 Let X, Y : (Ω, T) → (R, B(R)) be two square
integrable random variables on a probability space (Ω, T, P). We de
ﬁne the covariance between X and Y , denoted cov(X, Y ), as:
cov(X, Y )
= E[(X −E[X])(Y −E[Y ])]
We say that X and Y are uncorrelated if and only if cov(X, Y ) = 0.
If X = Y , cov(X, Y ) is called the variance of X, denoted var(X).
Exercise 10. Let X, Y be two square integrable, real random variable
on a probability space (Ω, T, P).
1. Show that cov(X, Y ) = E[XY ] −E[X]E[Y ].
2. Show that var(X) = E[X
2
] −E[X]
2
.
3. Show that var(X +Y ) = var(X) + 2cov(X, Y ) +var(Y )
4. Show that X and Y are uncorrelated, if and only if:
var(X +Y ) = var(X) +var(Y )
Tutorial 20: Gaussian Measures 13
Exercise 11. Let X be an ndimensional normal vector on some
probability space (Ω, T, P), with law N
n
(m, Σ), where m ∈ R
n
and
Σ = (σ
ij
) ∈ /
n
(R) is a symmetric and nonnegative real matrix.
1. Show that each coordinate X
j
: (Ω, T) → R is measurable.
2. Show that E[[X
α
[] < +∞ for all α ∈ N
n
.
3. Show that for all j = 1, . . . , n, we have E[X
j
] = m
j
.
4. Show that for all j, k = 1, . . . , n, we have cov(X
j
, X
k
) = σ
jk
.
Theorem 136 Let X be an ndimensional normal vector on a prob
ability space (Ω, T, P), with law N
n
(m, Σ). Then, for all α ∈ N
n
, X
α
is integrable. Moreover, for all j, k ∈ N
n
, we have:
E[X
j
] = m
j
and:
cov(X
j
, X
k
) = σ
jk
where (σ
ij
) = Σ.
Tutorial 20: Gaussian Measures 14
Exercise 12. Show the following:
Theorem 137 Let X : (Ω, T) → (R, B(R)) be a real random vari
able on a probability space (Ω, T, P). Then, X is a normal random
variable, if and only if it is square integrable, and:
∀u ∈ R , E[e
iuX
] = e
iuE[X]−
1
2
u
2
var(X)
Exercise 13. Let X be an ndimensional normal vector on a prob
ability space (Ω, T, P), with law N
n
(m, Σ). Let A ∈ /
d,n
(R) be an
d n real matrix, (n, d ≥ 1). Let b ∈ R
n
and Y = AX +b.
1. Show that Y : (Ω, T) → (R
d
, B(R
d
)) is measurable.
2. Show that the law of Y is N
d
(Am +b, A.Σ.A
t
)
3. Conclude that Y is an R
d
valued normal random vector.
Tutorial 20: Gaussian Measures 15
Theorem 138 Let X be an ndimensional normal vector with law
N
n
(m, Σ) on a probability space (Ω, T, P), (n ≥ 1). Let d ≥ 1 and
A ∈ /
d,n
(R) be an dn real matrix. Let b ∈ R
d
. Then, Y = AX+b
is an ddimensional normal vector, with law:
Y (P) = N
d
(Am +b, A.Σ.A
t
)
Exercise 14. Let X : (Ω, T) → (R
n
, B(R
n
)) be a measurable map,
where (Ω, T, P) is a probability space. Show that if X is a gaussian
vector, then for all u ∈ R
n
, ¸u, X) is a normal random variable.
Exercise 15. Let X : (Ω, T) → (R
n
, B(R
n
)) be a measurable map,
where (Ω, T, P) is a probability space. We assume that for all u ∈ R
n
,
¸u, X) is a normal random variable.
1. Show that for all j = 1, . . . , n, X
j
is integrable.
2. Show that for all j = 1, . . . , n, X
j
is square integrable.
3. Explain why given j, k = 1, . . . , n, cov(X
j
, X
k
) is welldeﬁned.
Tutorial 20: Gaussian Measures 16
4. Let m ∈ R
n
be deﬁned by m
j
= E[X
j
], and u ∈ R
n
. Show:
E[¸u, X)] = ¸u, m)
5. Let Σ = (cov(X
i
, X
j
)). Show that for all u ∈ R
n
, we have:
var(¸u, X)) = ¸u, Σu)
6. Show that Σ is a symmetric and nonnegative nn real matrix.
7. Show that for all u ∈ R
n
:
E[e
iu,X
] = e
iE[u,X]−
1
2
var(u,X)
8. Show that for all u ∈ R
n
:
E[e
iu,X
] = e
iu,m−
1
2
u,Σu
9. Show that X is a normal vector.
10. Show the following:
Tutorial 20: Gaussian Measures 17
Theorem 139 Let X : (Ω, T) → (R
n
, B(R
n
)) be a measurable map
on a probability space (Ω, T, P). Then, X is an ndimensional normal
vector, if and only if, any linear combination of its coordinates is itself
normal, or in other words ¸u, X) is normal for all u ∈ R
n
.
Exercise 16. Let (Ω, T) = (R
2
, B(R
2
)) and µ be the probability
on (R, B(R)) deﬁned by µ =
1
2
(δ
0
+ δ
1
). Let P = N
1
(0, 1) ⊗ µ, and
X, Y : (Ω, T) → (R, B(R)) be the canonical projections deﬁned by
X(x, y) = x and Y (x, y) = y.
1. Show that P is a probability measure on (Ω, T).
2. Explain why X and Y are measurable.
3. Show that X has the distribution N
1
(0, 1).
4. Show that P(¦Y = 0¦) = P(¦Y = 1¦) =
1
2
.
5. Show that P
(X,Y )
= P.
Tutorial 20: Gaussian Measures 18
6. Show for all φ : (R
2
, B(R
2
)) → C measurable and bounded:
E[φ(X, Y )] =
1
2
(E[φ(X, 0)] +E[φ(X, 1)])
7. Let X
1
= X and X
2
be deﬁned as:
X
2
= X1
{Y =0}
−X1
{Y =1}
Show that E[e
iuX
2
] = e
−u
2
/2
for all u ∈ R.
8. Show that X
1
(P) = X
2
(P) = N
1
(0, 1).
9. Explain why cov(X
1
, X
2
) is welldeﬁned.
10. Show that X
1
and X
2
are uncorrelated.
11. Let Z =
1
2
(X
1
+X
2
). Show that:
∀u ∈ R , E[e
iuZ
] =
1
2
(1 +e
−u
2
/2
)
Tutorial 20: Gaussian Measures 19
12. Show that Z cannot be gaussian.
13. Conclude that although X
1
, X
2
are normally distributed, (and
even uncorrelated), (X
1
, X
2
) is not a gaussian vector.
Exercise 17. Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be a
symmetric and nonnegative real matrix. Let A ∈ /
n
(R) be such
that Σ = A.A
t
. We assume that Σ is nonsingular. We deﬁne
p
m,Σ
: R
n
→ R
+
by:
∀x ∈ R
n
, p
m,Σ
(x)
=
1
(2π)
n
2
det(Σ)
e
−
1
2
x−m,Σ
−1
(x−m)
1. Explain why det(Σ) > 0.
2. Explain why
det(Σ) = [ det(A)[.
3. Explain why A is nonsingular.
Tutorial 20: Gaussian Measures 20
4. Let φ : R
n
→ R
n
be deﬁned by:
∀x ∈ R
n
, φ(x)
= A
−1
(x −m)
Show that for all x ∈ R
n
, ¸x −m, Σ
−1
(x −m)) = φ(x)
2
.
5. Show that φ is a C
1
diﬀeomorphism.
6. Show that φ(dx) = [ det(A)[dx.
7. Show that:
R
n
p
m,Σ
(x)dx = 1
8. Let µ =
p
m,Σ
dx. Show that:
∀u ∈ R
n
, Tµ(u) =
1
(2π)
n
2
R
n
e
iu,Ax+m−x
2
/2
dx
9. Show that the fourier transform of µ is therefore given by:
∀u ∈ R
n
, Tµ(u) = e
iu,m−
1
2
u,Σu
Tutorial 20: Gaussian Measures 21
10. Show that µ = N
n
(m, Σ).
11. Show that N
n
(m, Σ) << dx, i.e. that N
n
(m, Σ) is absolutely
continuous w.r. to the lebesgue measure on R
n
.
Exercise 18. Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be a sym
metric and nonnegative real matrix. We assume that Σ is singular.
Let u ∈ R
n
be such that Σu = 0 and u ,= 0. We deﬁne:
B
= ¦x ∈ R
n
, ¸u, x) = ¸u, m)¦
Given a ∈ R
n
, let τ
a
: R
n
→ R
n
be the translation of vector a.
1. Show B = τ
−1
−m
(u
⊥
), where u
⊥
is the orthogonal of u in R
n
.
2. Show that B ∈ B(R
n
).
3. Explain why dx(u
⊥
) = 0. Is it important to have u ,= 0?
4. Show that dx(B) = 0.
Tutorial 20: Gaussian Measures 22
5. Show that φ : R
n
→ R deﬁned by φ(x) = ¸u, x), is measurable.
6. Explain why φ(N
n
(m, Σ)) is a welldeﬁned probability on R.
7. Show that for all α ∈ R, we have:
Tφ(N
n
(m, Σ))(α) =
R
n
e
iαu,x
dN
n
(m, Σ)(x)
8. Show that φ(N
n
(m, Σ)) is the dirac distribution on (R, B(R))
centered on ¸u, m), i.e. φ(N
n
(m, Σ)) = δ
u,m
.
9. Show that N
n
(m, Σ)(B) = 1.
10. Conclude that N
n
(m, Σ) cannot be absolutely continuous with
respect to the lebesgue measure on (R
n
, B(R
n
)).
11. Show the following:
Tutorial 20: Gaussian Measures 23
Theorem 140 Let n ≥ 1 and m ∈ R
n
. Let Σ ∈ /
n
(R) be a
symmetric and nonnegative real matrix. Then, the gaussian measure
N
n
(m, Σ) is absolutely continuous with respect to the lebesgue measure
on (R
n
, B(R
n
)), if and only if Σ is nonsingular, in which case for
all B ∈ B(R
n
), we have:
N
n
(m, Σ)(B) =
1
(2π)
n
2
det(Σ)
B
e
−
1
2
x−m,Σ
−1
(x−m)
dx
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.