REAL ANALYSIS AND MEASURE

THEORY

BY

DR. PRATULANANDA DAS

DEPARTMENT OF MATHEMATICS
JADAVPUR UNIVERSITY
KOLKATA- 700032
WEST BENGAL, INDIA
E-mail : pratulananda@yahoo.co.in
 Chapter 5

(Equivalent to three hours of

teaching/modules)

Lebesgue integration of bounded

measurable functions

1
 5.1 Introduction

We start this chapter by recalling the definition of Riemann integral which is

motivated by the notion of area. For a bounded function f (x) defined on [a, b],
Z b
f (x)dx is defined to be the area of the region bounded by the curve y = f (x),
a
x-axis and the two ordinates (vertical lines) x = a and x = b. In order to find
the required area we take a partition P = {a = x0 < x1 < x2 < · · · < xn = b}
of [a, b] and construct the upper and lower sums as
n
X
U (P, f ) = Mi (xi − xi−1 )
i=1

and
n
X
L(P, f ) = mi (xi − xi−1 )
i=1

where Mi = sup f (x) and mi = inf f (x).

[xi ,xi−1 ] [xi ,xi−1 ]

Geometrically the upper and the lower sum represents the sums of the areas
of finite number of circumscribed and inscribed rectangles for the curve respec-
tively. U (P, f ) presents an upper estimate of the area under consideration while
L(P, f ) gives a lower estimate. As the partition becomes finer and finer the
Z b
estimates become better. The Riemann integral f (x)dx is obtained as the
a
common limit of the upper sums and the lower sums as the norm or mesh of
the partition tends to zero and we say that f is Riemann integrable on [a, b].

Technically we proceed in the following way. We define the lower and upper
Riemann integrals (also called Darbaux integrals) as
Z b
f (x)dx = sup L(P, f )
a P ∈π[a,b]

and
Z b
f (x)dx = inf U (P, f )
a P ∈π[a,b]

where π[a, b] stands for the collection of all partitions of [a, b]. f is said to be
Z b Z b Z b
Riemann integrable on [a, b] (i.e. f (x)dx exists) if f (x)dx = f (x)dx.
a a a

However the problem with the definition of Riemann integral is that it can
not be applied to a badly discontinuous but bounded function like χQ . However
if one looks closely at the function χQ , it occurs that there are only countable
number of vertical lines which in any case cannot form a region of positive area
and it seems only natural that the area under χQ should have been zero and it is

2
actually quite simple enough function to be integrable. this is one of the major
weaknesses of Riemann integration, namely its failure to deal with totally dis-
continuous functions. There are more weaknesses of Riemann integration about
which we will discuss in later modules.

This insufficiency of the process of Riemann integration was the motivation

for introduction of new and more efficient integration processes and Lebesgue
integration is one of them.

Moving away from the geometrical interpretation of Riemann integral as

the area of the undergraph (which in any case seems reasonable only for non-
negative and piece wise continuous functions) it is not difficult to realize that
the upper sum is in fact the integral of a step function which is above the curve
y = f (x) (i.e. which is greater or equal to f ) while the lower sum is the integral
of a step function which is below the curve (i.e. is less or equal to f ). At this
point it will be beneficial to recall the definition of a step function.

Definition 5.1. A function φ : [a, b] → R is a step function if there is a finite

collection of pairwise disjoint open intervals {Ik : 1 ≤ k ≤ n} such that φ is
constant on each Ik and
n
[
[a, b] = Ik .
k=1

If the step function takes the value ck on Ik then one can see that φ can be
represented in the form
Xn
φ= ck χIk .
k=1

Though this representation is not exactly correct in the sense that we ignore
the values of φ at finite number of points but it does not affect the value of the
integral which can then be defined as
Z b n
X
φ(x)dx = ck `(Ik )
a k=1

where as before `(Ik ) stands for the length of the interval Ik .

In view of above we can give the alternative definition of Riemann integral

in terms of step functions as follows.

Definition 5.2 Let f be a bounded function defined on [a, b]. The upper and
lower Riemann integrals of f on [a, b] are defined by
Z b Z b
f (x)dx = inf{ φ(x)dx : φ ≥ f, φ is a step function}
a a

3
and Z b Z b
f (x)dx = sup{ ψ(x)dx : ψ ≤ f, ψ is a step function}.
a a
Z b Z b
f is said to be Riemann integrable on [a, b] if f (x)dx = f (x)dx.
a a

5.2. Lebesgue integral of bounded functions

In the last chapter we have seen how the notion of a step function can be ex-
tended to the notion of a simple function and so the natural way of introducing
the new integration process seems replacing the step functions by simple func-
tions in the definition of Riemann integral which precisely gives us the notion
of Lebesgue integration.

Definition 5.3. Let s : E → R be a simple function where µ(E) < ∞ and let
n
X
the canonical representation of s be given by s = αk χEk . Then the Lebesgue
k=1
integral of s on E is defined as
Z n
X
sdµ = αk µ(Ek ).
E k=1

For a measurable subset A of E, the Lebesgue integral of s over A is defined

as Z n
X
sdµ = αk µ(A ∩ Ek ).
A k=1

The value of this integral is independent of the canonical representation of

n
X
the simple function s. To see this let s = αk χEk be the canonical represen-
k=1
m
X
tation of s and let s = βj χFj be another representation of s where the sets
j=1
Fj ’s are pairwise disjoint measurable sets whose union is E. Set

Ck = {j : βj = αk } for all k = 1, 2, . . . , n.
[
Then Ek = Fj . Now
j∈Ck

m
X n X
X n
X X n
X
βj µ(Fj ) = βj µ(Fj ) = αk µ(Fj ) = αk µ(Ek ).
j=1 k=1j∈Ck k=1 j∈Ck k=1

4
 We start with some basic properties of Lebesgue integrals of simple functions
in the next theorem.

Theorem 5.4. Let s and t be two simple functions defined on a measurable

set E of finite Lebesgue measure. Then
Z Z
(i) For any c ∈ R, csdµ = c sdµ.
Z ZE Z E
(ii) (s + t)dµ = sdµ + tdµ.
E E E
Z Z
(iii) If s ≤ t a.e. then sdµ ≤ tdµ.
ZE ZE
(iv) If s = t a.e. then sdµ = tdµ.
Z E Z E

(v) If A ⊂ B then sdµ ⊂ sdµ.

A B
Z Z Z
(vi) If A, B are disjoint then sdµ = sdµ + sdµ.
A∪B A B
Z Z
(vii) We have | sdµ| ≤ |s|dµ.
E E

Proof: We only prove (ii).

Let α1 , α2 , · · · , αn be the n distinct values of s and β1 , β2 , · · · , βm be the m

distinct values of t. Then we can write
n
X
s= αi χEi
i=1

where as before Ei = {x ∈ E : s(x) = αi } and

n
X
t= βj χFj
i=1

where Fj = {x ∈ E : t(x) = βj }. If Aij = Ei ∩ Fj then the collection of

measurable sets {Aij : 1 ≤ i ≤ n, 1 ≤ j ≤ m} is pairwise disjoint whose union

5
is E and on Aij , (s + t)(x) = αi + βj . Then
Z XZ X
(s + t)dµ = (s + t)dµ = (αi + βj )µ(Aij )
E i,j Aij i,j
  !
n
X Xm m
X n
X
= αi  µ(Aij ) + βj µ(Aij )
i=1 j=1 j=1 i=1
n
X m
X
= αi µ(Ei ) + βj µ(Fj )
i=1 j=1
Z Z
= sdµ + tdµ.
E E

This completes the proof.

All other proofs are easy and are left to the reader.

We are now in a position to define the Lebesgue integral of a bounded mea-

surable function.

Definition 5.5. Let f be a bounded function defined on a measurable set E

with finite Lebesgue measure. The upper and lower Lebesgue integrals of f on
E are defined by
Z Z
f dµ = inf{ tdµ : t ≥ f, t is a simple function}
E E

and Z Z
f dµ = sup{ sdµ : s ≤ f, s is a simple function}.
E E
Z Z
f is said to be Lebesgue integrable on E if f dµ = f dµ.
E E

Theorem 5.6. Let f and g be two bounded measurable functions defined on a

measurable set E of finite Lebesgue measure. Then
Z Z
(i) For any c ∈ R, cf dµ = c f dµ.
Z ZE Z E
(ii) (f + g)dµ = f dµ + gdµ.
E E E
Z Z
(iii) If f ≤ g a.e. then f dµ ≤ gdµ.
ZE ZE
(iv) If f = g a.e. then f dµ = gdµ.
E E

6
 Z Z
(v) If A ⊂ B then f dµ ⊂ f dµ.
A
ZB Z Z
(vi) If A, B are disjoint then f dµ = f dµ + f dµ.
A∪B A B
Z Z
(vii) We have | f dµ| ≤ |f |dµ.
E E

The proofs follow from Definition 5.5 and Theorem 5.4 and are left to the
reader.

Theorem 5.7.
Z For any bounded measurable function f : E → R+ where
µ(E) < ∞, f dµ = 0 if and only if f = 0 almost everywhere on E.
E

The proof is left as a self-check exercise.

We now show that for a bounded function defined on a measurable set of

finite Lebesgue measure, Lebesgue integrability is equivalent to measurability.

Theorem 5.8. A bounded function f : E → R defined on a measurable set E

of finite Lebesgue measure is Lebesgue integrable if and only if it is measurable.

Proof: The proof of the theorem is trivially true when µ(E) = 0. So we assume
that µ(E) is positive. First suppose that f is Lebesgue integrable on E. Then
Z Z
from the definition f dµ = f dµ where
E E

Z Z
f dµ = inf{ tdµ : t ≥ f, t is a simple function}
E E

and Z Z
f dµ = sup{ sdµ : s ≤ f, s is a simple function}.
E E

Then for each k ∈ N we can find two simple functions sk and tk , sk ≤ f ≤ tk

such that Z
1
(tk − sk )dµ < .
E k
Let t = inf tk and s = sup sk . Then clearly s and t are bounded measurable
k k
functions defined on E and s ≤ f ≤ t. Now observe that
Z Z
1
(t − s)dµ ≤ (tk − sk )dµ < ∀k ∈ N
E E k
Z
which implies that (t − s)dµ = 0. Then applying Theorem 5.7 we can con-
E
clude that s = t almost everywhere. But then s = f = t almost everywhere,

7
Since the functions s and t are measurable so by Lemma 3.6 f is also measurable.

Conversely suppose that f is a bounded measurable function defined on a

measurable set E of finite Lebesgue measure. Since f is bounded, we can find
a closed interval [A, B] such that f (E) ⊂ [A, B]. Let ε > 0 be given. Choose a
partition P = {A − 1 = l0 < l1 < l2 < · · · < ln−1 < ln = B} of [A − 1, B] with
ε
norm less that 2µ(E) .

Let Er = {x ∈ E : lr−1 < f (x) ≤ lr } for r = 1, 2, 3, . . . , n. Since f is

measurable so each set Er is measurable and Ei ∩ Ej = φ if i 6= j. Further
[n
Er = E. We consider two simple functions s and t as follows:
r=1

n
X
s= lr−1 χEr
r=1

and
n
X
t= lr χEr
r=1

Clearly s ≤ f ≤ t and
Z n n
X ε X ε ε
(t − s)dµ ≤ (lr − lr−1 )µ(Er ) < µ(Er ) = µ(E) = .
E r=1
2µ(E) r=1 2µ(E) 2

But this implies that we must have

Z Z
inf{ tdµ : t ≥ f, t is a simple function} = sup{ sdµ : s ≤ f, s is a simple function}
E E

and so f must be Lebesgue integrable on E.

Recall that a sequence of real valued functions {fn }n∈N defined on a set A
is said to be uniformly bounded if there is a positive real number M such that
|fn (x)| ≤ M for all x ∈ A.

Theorem 5.9 (Bounded Convergence Theorem). Let {fn }n∈N be a uni-

formly bounded sequence of Lebesgue integrable functions defined on a measur-
able set E of finite Lebesgue measure. Let fn → f a.e. on E. Then f is also
Lebesgue integrable on E and
Z Z
f dµ = lim fn dµ.
E n→∞ E

Proof: Since fn → f a.e. on E so there is a measurable subset B of E

of Lebesgue measure zero such that fn → f on E − B. As before setting

8
fn = f = 0 on the exceptional set B we see that fn → f on E. Note that
this does not change the value of any integral under consideration. Now let
|fn (x)| ≤ M ∀n ∈ N and ∀ x ∈ E where M > 0 is a real number. Then
|f (x)| ≤ M ∀x ∈ E. Note that f is bounded and measurable on E and so is
Lebesgue integrable on E.

Choose ε > 0. By Egoroff’s Theorem there is a measurable subset A of E

ε
such that µ(A) < 4M and fn converges to f uniformly on E − A. Subsequently
we can find a positive integer N ∈ N such that
ε
|fn (x) − f (x)| <
2µ(E) + 1

for all n ≥ N and for all x ∈ E − A. Then for n ≥ N we get

Z Z Z
| f dµ − fn dµ| ≤ |fn − f |dµ
E E E
Z Z
= |fn − f |dµ + |fn − f |dµ
A E−A
εµ(E − A)
< 2M µ(A) +
2µ(E) + 1
< ε.

This shows that Z Z

f dµ = lim fn dµ.
E n→∞ E

We can now pause for a moment and try to understand what are the advan-
tages of this new integration process over Riemenn integration.

One of the main deficiencies of Riemann integration is that it is not preserved

under pointwise convergence. To see this, consider the sequence of functions
{fn }n∈N defined on [0, 1] where

fn (x) = 1 if x = r1 , r2 , . . . , rn
= 0 otherwise

where {rn }n∈N is the sequence of rational numbers in [0, 1]. Note that the
sequence of functions {fn }n∈N is uniformly bounded and each function fn is
Riemann integrable. But this sequence converges pointwise to the function χQ
on [0, 1] which is not Riemann integrable. Theorem 5.9 is one instance which
shows why Lebesgue integration is more efficient that Riemann integration.

We already have an example of a function, namely, χQ on [0, 1] which is

not Riemann integrable but being a simple function is Lebesgue integrable
with the value of the integral 0. In fact we have shown that any bounded

9
measurable function defined on a measurable set of finite Lebesgue measure
is Lebesgue integrable. There are more examples of functions which may not
be Riemann integrable but are Lebesgue integrable. Now the natural question
arises what about those functions which are already Riemann integrable, will
they be Lebesgue integrable ? the answer is ”YES”. Recall the Lebesgue crite-
ria for Riemann integration which states that a bounded function f defined on
[a, b] is Riemann integrable if and only if the set of discontinuities of f is a set
of Lebesgue measure zero. But then f is continuous almost everywhere on [a, b]
and so by Theorem 3.7 f is measurable and so is Lebesgue integrable.

However we can give a direct proof also which is given below.

Theorem 5.10. Every Riemann integrable function on closed interval [a, b] is

Lebesgue integrable on [a, b] and
Z b Z b
(L) f (x)dx = (R) f (x)dx.
a a

Proof: Let π[a, b] denote the set of all partitions of [a, b] and for any P ∈ π[a, b],
as usual, let L(P, f ) and U (P, f ) be the Riemann lower and upper sums of f
respectively. Then we know that
Z b
sup {L(P, f ) : P ∈ π[a, b]} = (R) f (x)dx
a

and
Z b
inf {U (P, f ) : P ∈ π[a, b]} = (R) f (x)dx.
a

Evidently we can choose sequences of partitions {Pk }k∈N and {Pk0 }k∈N of [a, b]
for which Z b
lim L(Pk , f ) = (R) f (x)dx
k a

and
Z b
lim U (Pk0 , f ) = (R) f (x)dx.
k a

Let θ1 = P1 ∪ P10 ,
θ2 = P1 ∪ P10 ∪ P2 ∪ P20 ,
θ3 = P1 ∪ P10 ∪ P2 ∪ P20 ∪ P3 ∪ P30 and
so on. Then {θ1 ⊂ θ2 ⊂ θ3 · · · } is a sequence of partitions of [a, b] and
Z b
lim L(θk , f ) = (R) f (x)dx
k a

and
Z b
lim U (θk , f ) = (R) f (x)dx.
k a

10
Consider a typical partition θk = {a = x0 < x1 < · · · < xn−1 < xn = b}. Let
mi = inf f (x), Mi = sup f (x) for i = 1, 2, · · · , n. Let
x∈[xi−1 ,xi ] x∈[xi−1 ,xi ]

lk (x) = mi if xi−1 ≤ x < xi i = 1, 2, · · · , n

= mn at x = b.

and

uk (x) = Mi if xi−1 ≤ x < xi for i = 1, 2, · · · , n

= Mn at x = b.

Then {lk }k∈N and {uk }k∈N are two sequences of simple functions on [a, b] with
Z b Z b
(L) lk (x)dx = L(θk , f ) and (L) uk (x)dx = U (θk , f ).
a a

Clearly
l1 ≤ l 2 ≤ · · · ≤ f ≤ · · · ≤ u 3 ≤ u 2 ≤ u 1 .
From monotonicity and boundedness of the two sequences it follows that lim lk
k
and lim uk exist. Let us define
k

l(x) = lim lk (x) ∀x ∈ [a, b]

k

and
u(x) = lim uk (x) ∀x ∈ [a, b].
k

Since each lk and uk are measurable, so by Theorem 3.4 (v) both the functions
l and u are measurable and further l ≤ f ≤ u. Since f is bounded on [a, b]
so there exists a λ > 0 such that |f (x)| ≤ λ ∀x ∈ [a, b]. Subsequently for all
k ∈ N, |lk (x)| ≤ λ, |uk (x)| ≤ λ for all x ∈ [a, b]. Hence by Bounded Convergence
Theorem Z b Z b
lim(L) lk (x)dx = (L) l(x)dx
k a a
and Z b Z b
lim(L) uk (x)dx = (L) u(x)dx.
k a a
But evidently
Z b Z b
lim(L) lk (x)dx = (R) f (x)dx
k a a

and
Z b Z b
lim(L) uk (x)dx = (R) f (x)dx.
k a a

11
Since f is Riemann integrable, so
Z b Z b Z b Z b
(L) l(x)dx = (R) f (x)dx = (R) f (x)dx = (L) u(x)dx.
a a a a

Then Z b
(L) (u(x) − l(x))dx = 0.
a

But as u ≥ l soit follows from Theorem 3.7 u(x) = l(x) a.e. on [a, b] and so
u(x) = f (x) = l(x) a.e. on [a, b]. Since u and l are measurable so by Lemma
3.6 f is also measurable on [a, b] and hence is Lebesgue integrable on [a, b].
Obviously we have
Z b Z b
(L) f (x)dx = (R) f (x)dx.
a a

There exist improper Riemann integrable functions for which above is not
true. Let f : [0, ∞) → R be defined by

sin x
f (x) = , x 6= 0
x
= 0, x = 0.
Z ∞ Z ∞
sin x sin x
Then (R) dx exists and its value is π2 . But (L) dx does not
0 x Z ∞ 0 x
sin x
exist finitely. For this we show that (L) dx does not exist finitely.
0 x
Indeed
Z ∞ ∞ Z (k+1)π
sin x X sin x
(L) dx = (L) dx
0 x kπ x
k=0
∞ Z (k+1)π
X sin x
≥ (L) dx
kπ x
k=1
∞ Z (k+1)π
X sin x
= (R) dx
kπ x
k=1
∞ Z (k+1)π
X 1
≥ (R) |sin x| dx
(k + 1)π kπ
k=1
∞
2 1
X
=
π (k + 1)
k=1
= ∞.

12
 5.3. Self-check Exercises

Exercise 5.1. If h : [a, b] → R is a step function then show that for every ε > 0,
Z b
there is a continuous function f : [a, b] → R satisfying |h − f | < ε.
a
n
X
Solution: We can write h in the form h = ci χJi where ci ∈ R and Ji are
i=1
disjoint subintervals of [a, b] whose union is [a, b] (we actually ignore a finite
number of points at most which do no harm). For each i we can construct a
continuous piecewise linear function fi : [a, b] → R for which
Z b
ε
|χJi − fi | < .
a n(1 + |ci |)
n
X
Then the function f = ci fi is continuous on [a, b] and
i=1

b n Z b n
|ci |ε
Z X X
|h − f | ≤ |χJi − fi | < < ε.
a i=1 a i=1
n(1 + |ci |)

Exercise 5.2. If s : [a, b] → R is a simple function then show that for every
Z b
ε > 0, there is a continuous function f : [a, b] → R satisfying |h − f | < ε.
a

Solution: Since any simple function is a linear combination of characteristic

functions of measurable sets, for simplicity let us take s = χE where E is a
measurable subset of [a, b]. We first show that we can find a step function h
having the desired property.

Now we can find a finite collection of pairwise disjoint open intervals {Ir :
n
[
1 ≤ r ≤ n} in (a, b) such that the open set U = Ir has the property that
r=1
µ(E∆U ) < ε. Subsequently h = χU is a step function having the property that
Z b Z b
|χE − h| = χE∆U = µ(E∆U ) < ε.
a a

It is now clear that for any simple function s on [a, b] we can first find a step
Z b
ε
function h on [a, b] such that |s − h| < . Again by the above exercise we
a 2
Z b
ε
can find a continuous function f on [a, b] satisfying |h − f | < . Combining
a 2
we have the desired result.

13
Exercise 5.3. If f : [a, b] → R is Lebesgue integrable and f = g − h where g
and h are non-negative Lebesgue-integrable functions then prove that
Z b Z b Z b
f= g− h.
a a a

Solution: We can write

f + − f − = f = g − h i.e. f + + h = g + f −

and so
Z b Z b Z b Z b Z b Z b
−
+
f + h= +
(f + h) = (g + f ) = g+ f−
a a a a a a

which implies
Z b Z b Z b Z b Z b
f= f+ − f− = g− h.
a a a a a

Exercise 5.4. Prove Theorem 5.7.

Z
Solution: First assume that f dµ > 0. Then there exists a simple function
E Z
s : E → R+ with 0 ≤ s ≤ f such that sdµ > 0. But this implies that
E
µ ({x ∈ E : s(x) > 0}) > 0. Since {x ∈ E : f (x) > 0} ⊇ {x ∈ E : s(x) > 0},
so µ ({x ∈ E : f (x) > 0}) > 0. Hence f 6= 0 a.e. on E.

To prove the converse, let f 6= 0 a.e. on E. Then µ ({x ∈ E : f (x) > 0}) >
∞
[ 1
0. Now {x ∈ E : f (x) > 0} = {x ∈ E : f (x) ≥ }. Thus 0 <
n=1
n
∞  
X 1
µ ({x ∈ E : f (x) > 0}) ≤ µ {x ∈ E : f (x) ≥ } . Hence there is at least
n=1
n
one n ∈ N for which µ (En ) > 0 where En = {x ∈ E : f (x) ≥ n1 }. Clearly then
Z Z
1 µ(En )
s = n1 χEn is a simple function ≤ f . Thus f dµ ≥ χEn dµ = > 0.
E E n n
Exercise 5.5. Let E ⊂ R have Lebesgue measure zero. ShowR that if f is a
bounded function on E, then f is Lebesgue measurable and f = 0.
E

Solution: Clearly f is measurable, since for each c ∈ R, {x ∈ E : f (x) ≤ c} ⊂

E is a set of Lebesgue measure 0. It suffices to consider f non-negative since we
can consider the positive and negative parts of f separately. Since f is bounded

14
on E, there exists a M > 0 such that f ≤ M on E. Therefore, by monotonicity
we conclude that Z Z
0 ≤ f ≤ M = M · µ(E) = 0.
E E

Summary: Starting from the notion of Riemann integration of bounded func-

tions we have shown that a much improved process of integration, namely
Lebesgue integration, can be defined for bounded functions which have obvi-
ous advantages and we have established its basic properties.

Following are the main definitions and points to be remembered.

• Let s : E → R be a simple function where µ(E) < ∞ and let the canonical
n
X
representation of s be given by s = αk χEk . Then the Lebesgue integral of s
k=1
on E is defined as Z n
X
sdµ = αk µ(Ek ).
E k=1

For a measurable subset A of E the Lebesgue integral of s on A is defined

as Z n
X
sdµ = αk µ(A ∩ Ek ).
A k=1

• Let f be a bounded function defined on a measurable set E with finite

Lebesgue measure. The upper and lower Lebesgue integrals of f on E are
defined by
Z Z
f dµ = inf{ tdµ : t ≥ f, t is a simple function}
E E

and Z Z
f dµ = sup{ sdµ : s ≤ f, s is a simple function}.
E E
Z Z
f is said to be Lebesgue integrable on E if f dµ = f dµ.
E E

• A bounded function f : E → R defined on a measurable set E of finite

Lebesgue measure is Lebesgue integrable if and only if it is measurable.

• All the usual properties of Riemann integration hold for this new integra-
tion process.

15
 • (Bounded Convergence Theorem). Let {fn }n∈N be a uniformly bounded
sequence of Lebesgue integrable functions defined on a measurable set E of finite
Lebesgue measure. Let fn → f a.e. on E. Then f is also Lebesgue integrable
on E and Z Z
f dµ = lim fn dµ.
E n→∞ E

• Every bounded Riemann integrable function on [a, b] is Lebesgue integrable

on [a, b] and
Z b Z b
(L) f (x)dx = (R) f (x)dx.
a a

Acknowledgement: While writing this chapter the author has followed the
excellent books by H.L. Royden et al and R.A. Gordon. The full references of
these books is given in ”Learn More” section.

16
 SELF ASSESSMENT

Example 5.1. Let f be a Lebesgue measurable function defined on a measur-

able set E. Which of the following statements is true ?
Z
(a) If f exists then f is Lebesgue integrable on E.
E
Z
(b) f may not exist but f is Lebesgue integrable on E.
E

(c) f is Lebesgue integrable only if E is of finite measure.

Z
(d) If f is Lebesgue integrable on E then f exists finitely.
E

Example 5.2. Let fn be a sequence of Lebesgue measurable functions converg-

ing to f pointwise. Which of the following statements is true ?

(a) f is always Lebesgue integrable.

(b) f is always Lebesgue integrable if each fn is a simple function.
(c) f is never Lebesgue integrable.
(d) f is Lebesgue integrable if the sequence {fn }n is uniformly bounded.

Example 5.3. Let f be a bounded measurable function defined on [0, 1]. Which
of the following statements is true ?

(a) If |f | is Riemann integrable then f is Riemann integrable.

(b) f is Riemann integrable but |f | is not Riemann integrable.
(c) |f | is Lebesgue integrable but f is not Lebesgue integrable.
(d) |f | is Lebesgue integrable if and only if f is Lebesgue integrable.
Example 5.4. Is it true that the Bounded Convergence Theorem hold for the
Riemann integral?

Example 5.5. Let f be R a nonnegative bounded measurable function on a set

E of finite measure. If f = 0 then is it true that f = 0 a.e. on E?
E

Example 5.6. Consider the function f = χR−Q on [3, 6]. Which of the follow-
ing is true ?

17
 (a) f is Riemann integrable.
(b) f is neither Riemann integrable nor lebesgue integrable.
(c) f is Lebesgue integrable with with the value of the integral 3.

(d) f is lebesgue integrable with the value of the integral 0.

Example 5.7. Which of the following statements is true ?

(a) A Riemann integrable function can be discontinuous on a set of positive

Lebesgue outer measure.
(b) A Lebesgue integrable function can be discontinuous on a set of positive
Lebesgue outer measure.
(c) A Riemann integrable function can have at most countable number of
discontinuities.
(d) A Lebesgue integrable function must be continuous almost everywhere.

Example 5.8. Is it true that the characteristic function of a non-measurable

subset of [0, 1] can be Lebesgue integrable ?

Example 5.9. Which of the following statements is false ?

(a) Every bounded Lebesgue integrable function defined on [a, b] is Riemann

integrable.
(b) Every bounded Riemann integrable function defined on [a, b] is Lebesgue
integrable.
(c) Every monotonic function defined on a bounded closed interval is Lebesgue
integrable.
(d) A function which is not monotonic at any point of [a, b] can be Lebesgue
integrable.

Example 5.10. Is Bounded convergence Theorem true for a sequence of func-

tions {fn }n∈N which is not uniformly bounded ?

SOLUTIONS

18
• Solution of Example 5.1: Ans: (d).

• Solution of Example 5.2: Ans: (d).

• Solution of Example 5.3: Ans: (d).

• Solution of Example 5.4: False.

( Consider the sequence of functions
n
X
fn = χ{qk }
k=1

where {qk } is an enumeration of the rationals

R in [0, 1]. It is easy to see for any
n ∈ N, fn is Riemann integrable and fn = 0. But fn converges pointwise to
[0,1]
f = χQ on [0, 1] which is not Riemann integrable on [0, 1].)

• Solution of Example 5.5: True.

( By Chebychev’s inequality, we have
Z
1
µ({x ∈ E : f (x) > }) ≤ n f = 0, for all n ∈ N.
n
E

1
S
Since {x ∈ E : f (x) > 0} = {x ∈ E : f (x) > n} is a countable union sets of
n
measure zero, it follows that µ({x ∈ E : f (x) > 0}) = 0 and hence f = 0 a.e.
on E.)

• Solution of Example 5.6: Ans: (c).

• Solution of Example 5.7: Ans: (b).

• Solution of Example 5.8: False.

• Solution of Example 5.9: Ans: (a).

• Solution of Example 5.10: False.

let us define fn (x) = n.χ[0, n1 ] for every n ∈ N and f is the constant function
which takes the value 0 on [0, 1]. Then fn → f pointwise on [0, 1] but {fn } is
not uniformly bounded and
Z 1 Z 1
f dµ 6= lim fn dµ.
0 n→∞ 0

19
 LEARN MORE

The history of integration was probably started by Greek Philosopher Archimedes

around the 2nd century B.C., but did not start to gain momentum until the 17th
century. Italian Mathematician Bonaventura Cavalieri began working with indi-
visibles in the computation of areas by the early 1700s. British mathematician
Isaac Newton is generally credited with the discovery of calculus. But there
is also a widespread view that it was the German mathematician Gottfriend
Liebniz who discovered the relationship between the integral and the derivative,
and really invented integral calculus. Both mathematicians independently made
their contributions at the end of 1700s. Nevertheless, the calculus they devel-
oped was very different from what we read today in books. The whole treatment
was based on the existence of infinitesimals, units of length that were ”infinitely
small.” The problem with this infinitesimal calculus was that nobody could
develop a rigorous treatment of the notion of ”infinitesimal”. French mathe-
matician Augustin-Louis Cauchy and German mathematician Karl Weierstrass
in 19th century eliminated the infinitesimal by constructing the formal  − δ
notion of limit after which the integral was actually defined rigorously. In the
middle of 19th century Bernhard Riemann developed the integral (known as
Riemann integral after his name) which we frequently use today. This inte-
gral is constructed by taking the limit of approximations, called Riemann sums,
which are based on partitioning the domain of the function. However we are also
familiar with another equivalent definition of Riemann integral which was pro-
posed and studied by French mathematician Jean Gaston Darbaux in the end
of nineteenth century with the help of lower and upper Darbaux sums. Actually
this approach is more fundamental in the introduction of Lebesgue integral later.

However, it was realized quickly that it is not too difficult to find functions
which are not Riemann (or Darbaux) integrable. In his thesis in 1902, Henri
Lebesgue first developed the idea of Lebesgue measure extending the notion of
length of intervals and applied it to develop a new integral in which the focus was
on the range of the function, instead of on the domain. The thesis of Lebesgue
is considered as one of the greatest achievements of the twentieth century. The
distinction between the two approaches can be seen by envisioning the graph
of a real function f whose range is the reals. Whereas Riemann or Darbaux
focused on partitioning the x-axis, Lebesgues approach was to partition the y-
axis instead.

The students are encouraged to consult the following excellent books and
web resources to have more in depth study about the course materials as well
as to know further.

Reference Books:

20
 1. G. De Barra, Measure theory and integration, Reprint, New age interna-
tional (P) Ltd., New Delhi, 1987.

2. S.K. Berberian, Measure and integration, Chelsea Publishing Co., New

York, 1965.

3. J.C. Burkill, The Lebesgue integral, Cambridge University press, England,

1951.

4. C. Caratheodory, Algebraic theory of measure and integration, Chelsea

Publishing house, New York, 1963.

5. R.A. Gordon, The integrals of Lebesgue, Denjoy, Perron and Henstock,

American Mathematical Society, 1994.

6. P.R. Halmos, Measure Theory, Reprint, Springer Verlag, 1974.

7. P.K. Jain, V.P. Gupta, Lebesgue measure and integration, Wiley Eastern
Ltd., New Delhi (New Age international Ltd.), 2011.

8. M.E. Munroe, Introduction to measure and integration, Addison-Wesley,

Cambridge,Mass, 1953.

9. I.K. Rana, An Introduction to Measure and Integration (2nd ed.), Narosa

Publishing House, New Delhi, 2004.

10. H. L. Royden and P.M. Fitzpatrick, Real Analysis Fourth edition, Pren-
tice Hall, New York, 2010.

11. W. Rudin, Principles of Mathematical Analysis, Third edition, Mcgrow-

Hill, New York, 1976.

12. W. Rudin, Real and complex analysis, Third edition, Mcgraw Hill, New
York, 1986.

Web Resources with links:

http://en.wikibooks.org/wiki/UMD Analysis Qualifying Exam/Aug08 Real

http://en.wikibooks.org/wiki/Subject:University level mathematics books

http://www.math.iitb.ac.in/ ars/week7 8.pdf

www.math.tifr.res.in/ publ/ln/tifr12.pdf

21
https://www.math.ucdavis.edu/.../measure theory/measure theory.html

www.math.chalmers.se/ borell/MeasureTheory.pdf

https://math.berkeley.edu/ rieffel/measinteg.html

https://www.math.ucdavis.edu/ hunter/measure.../measure theory.html

https://rutherglen.science.mq.edu.au/wchen/lnilifolder/lnili.html

22