MTH309 (2024-25, E VEN S EMESTER )

P ROBLEM S ET 7

1. Let f 1 , f 2 , . . . be Borel measurable functions on (Ω, F , µ). Suppose

∞ Z
∑ | f n |dµ < ∞.
n =1 Ω

Show that ∑∞ ∞
Ω ( ∑n=1 f n dµ )
R
n=1 f n converges a.e. [ µ ] to a finite valued function, and =
∑∞
R
n=1 Ω f n dµ.

2. Consider a closed bounded interval [ a, b], equipped with the Lebesgue measure λ on
B̄([ a, b]). Let f be a bounded real-valued function on [ a, b]. For a finite partition P of the
interval [ a, b], let U (P ) and L(P ) denote respectively the “upper” and “lower” functions
R
as in the Riemann integration theory. Note that both are simple functions and U (P )dλ
R
and L(P )dλ are exactly the upper and lower Riemann sums of f corresponding to the
partition P . Now if {Pn } is a nested sequence of partitions with ||Pn || ↓ 0 (||P || is the
maximum size of the sets in the partition P ) and if we denote the corresponding sequence of
upper and lower functions as {Un } and { Ln } respectively, then clearly U1 ≥ U2 ≥ · · · f ≥
· · · ≥ L2 ≥ L1 , so that the two sequences converge pointwise to two measurable functions
U and L respectively and U ≥ f ≥ L.

(a) Use DCT to conclude that the sequences of the upper and lower Riemann sums
R R
converge to Udλ and Ldλ respectively (note that f is assumed to be bounded).

(b) Show that, if x ∈ [ a, b] is not an end-point of any of the partitions Pn , then f is

continuous at x if and only if U ( x ) = f ( x ) = L( x ).

1
 Recall that f is said to be “Riemann integrable” on [ a, b] if the sequences of the upper
Riemann sums and the lower Riemann sums converge to a common limit I and this
common limit I does not depend on the particular nested sequence of partitions. In
this case, I is defined to be the value of the Riemann integral.

(c) Show that f is Riemann integrable on [ a, b] if and only if f is continuous λ-a.e. on [ a, b].
R R
(Hint: Note that f is Riemann integrable if Udλ = Ldλ.)

(d) Show that if f is Riemann integrable on [ a, b], then f is also λ-integrable and in that
case, the two integrals have the same value.

3. Let (Ω, F , µ) be a measure space and a < b be any two real numbers.

(a) Suppose f is a real-valued function on ( a, b) × Ω such that (i) for each t ∈ ( a, b), the
function ω → f (t, ω ) is F -measurable and (ii) for each ω ∈ Ω, the function t → f (t, ω )
is continuous on ( a, b). Show that if there is a µ-integrable function g on Ω such that
R
| f (t, ω )| ≤ g(ω ) for all (t, ω ) then φ(t) = Ω f (t, ω )dµ defines real-valued continuous
function.

(b) Suppose f is a real-valued function on ( a, b) × Ω such that (i) for each t ∈ ( a, b), the
function ω → f (t, ω ) is F -measurable and µ-integrable, and (ii) for each ω ∈ Ω, the
∂ f (t,ω )
function t → f (t, ω ) is everywhere differentiable on ( a, b), that is, h(t, ω ) = ∂t .
Show that if there is a µ-integrable function g on Ω such that |h(t, ω )| ≤ g(ω ) for
R
all (t, ω ) then the function defined on ( a, b) by φ(t) = Ω f (t, ω )dµ is everywhere
differentiable and, moreover, φ′ (t) = Ω h(t, ω )dµ.
R

[Result (b) gives conditions for differentiation to be performed under the integral sign,
d
R R ∂ f (t,ω )
that is, conditions for dt Ω f ( t, ω ) dµ = Ω ∂t dµ.]

4. Let µ be a measure on the Borel σ-field on [0, ∞). Show that, if e−λ0 x dµ( x ) < ∞ for some
R

λ0 > 0 then e−λ0 x dµ( x ) < ∞ for all λ > λ0 . Further the function ϕ(λ) = e−λx dµ( x ) is
R R

infinitely differentiable on (λ0 , ∞) with ϕ(n) (λ) = (−1)n x n e−λx dµ( x ) for all n ≥ 1 and
R

λ > λ0 .

5. Let µ be a measure on the Borel σ-field on R such that the function f ( x ) = x is µ-integrable.
Show that for every t ∈ R, the function x → sin tx is Borel measurable and µ-integrable
R
and, moreover, the real valued function ϕ(t) = sin txdµ( x ) is everywhere differentiable
with ϕ′ (t) = x cos txdµ( x ).
R

2
6. If µ is a measure on (Ω, F ) and A1 , A2 , . . . is a sequence of sets in F . Use Fatou’s lemma to
show that
µ(lim inf An ) ≤ lim inf µ( An ).
n→∞ n→∞

If µ is finite, show that

µ(lim sup An ) ≥ lim sup µ( An ).
n→∞ n→∞

7. (Here are slightly stronger versions of MCT and Fatou’s Lemma.) Let { f n } be a sequence of
extended real valued measurable functions on a measure space (Ω, A, µ).

(a) Show that if f n ↑ f and if for some measurable g with g− dµ < ∞, f n ≥

R
g for all n,
R R R R
then (the integrals f n dµ, f dµ all exist and) f n dµ ↑ f dµ.

(b) Show that if f n ↓ f and if for some measurable g with g+ dµ < ∞, f n ≤

R
g for all n,
R R R R
then (the integrals f n dµ, f dµ all exist and) f n dµ ↓ f dµ.

(c) Show that if for some measurable g with g− dµ < ∞, f n ≥ g for all n, then
R R
lim inf f n dµ ≤
R
lim inf f dµ.

(d) Show that if for some measurable g with g+ dµ < ∞, f n ≤ g for all n, then
R R
lim sup f n dµ ≥
R
lim sup f dµ.

8. Give an example of a sequence of Lebesgue integrable functions f n converging everywhere

to a Lebesgue integrable function f , such that

Z ∞ Z ∞
lim f n ( x )dx < f ( x )dx.
n→∞ −∞ −∞

Thus the conditions of Fatou’s lemma and dominated convergence theorem cannot be
dropped.

9. Let (Ω, F , µ) be a measure space and ( E, E ) a measurable space. Suppose ν is a function on

Ω × E , taking values in [0, 1], such that (i) for every fixed ω ∈ Ω, the set function ν(ω, B),
B ∈ E is a measure on E and (ii) for every B ∈ E , the function ω 7→ ν( B, ω ) is A-measurable.
R
For B ∈ E , define η ( B) = Ω ν(·, B)dµ. (This integral exists —- why?)

(a) Show that η defines a measure on E and that for any non-negative measurable function
f on ( E, E ), the function g : ω 7→ E f dν(ω, ·) on Ω is non-negative A-measurable.
R

(First assume that f is a non-negative simple function)

3
 (b) With f as in (a) and denoting g to be the function on Ω as defined there, show that
R R
E f dη = Ω gdµ. (Again, start by assuming that f is a non-negative simple function)

10. Let (Ω, F , P) be a measure space and g be a non-negative Borel measurable function on Ω.
Define a measure λ on F as

Z
λ( A) = gdµ for all A ∈ F .
A

Show that if f is a Borel measurable function on Ω, then

Z Z
f dλ = f gdµ
Ω Ω

in the sense that if one integrals exists so does the other, and the two integrals are equal.

11. The function defined on [0, 1] by


 1 if x is irrational
f (x) =
 0 ifx is rational.

is the standard example of a function that is Lebesgue integrable (it is 1 a.e.) but not Riemann
integrable. But what is wrong with the following reasoning?

If we consider the behavior of f on the irrationals, f assumes the constant value 1 and is
therefore continuous. Since the rationals have Lebesgue measure 0, f is therefore continuous
almost everywhere and hence is Riemann integrable.