January 11, 2025

NOTES ON
LEBESGUE MEASURE & INTEGRATION

M.THAMBAN NAIR

1. Intrduction

1.1. Shortcomings of Riemann integral. Recall that a bounded function f : [a, b] → R is

Riemann integrable iff there exists a unique γf ∈ R such that
L(P, f ) ≤ γf ≤ U (P, f )
for all partitions of [a, b], and in that case
Z b
f (x)dx := γf
a

is called the Riemann integral of f . In the above, by a partition P of [a, b], we mean a finite
set {xi : i = 0, 1, . . . , k} of points in [a, b] such that
a = x0 < x1 < · · · < xk = b,
and
k
X k
X
L(P, f ) = mi ℓ(Ii ), U (P, f ) = Mi ℓ(Ii ),
i=1 i=1
where
Ii := [xi−1 , xi ], mi = inf f (x), Mi := sup f (x).
x∈Ii x∈Ii

0, x ∈ [a, b] ∩ Q,
Example 1.1. Taking f : [a, b] → R defined by f (x) = we see that
1, x ∈ [a, b] \ Q,
L(P, f ) = 0 and U (P, f ) = b − a for all partition P of [a, b]. Therefore, this f is not Riemann
integrable. ♢

0, x ∈ [a, b] ∩ Q,
Example 1.2. Taking f : [a, b] → R defined by f (x) = we see that
x, x ∈ [a, b] \ Q,
L(P, f ) = 0 and U (P, f ) ≥ (b2 − a2 )/2 for all partition P of [a, b] (Exercise). Therefore, this f
is not Riemann integrable. ♢

However, a sequence (fn ) of Riemann inegrable functions on [a, b] can be constructed such that
fn (x) → f (x) as na → ∞ for each x ∈ [a, b] (Exercise).
1
2 M.THAMBAN NAIR

1.2. Looking for a notion of a measure. In view of the above observations, we look for a
new integral which generalizes the notion of Riemann integral so that above referred unpleasant
situations does not arise. The new integral which we are going to define is called the Lebesgue
inegral over certain distinguished sets called Lebesgue measurable sets, which include the set
of all intervals, and many more sets of practical importance. The main idea of developing such a
frame work is to introduce the notion of a measure of a set, say

µ(E) for E ⊆ R,

which generalizes the notion of length of an interval, that is, we must have

µ(I) = ℓ(I) for every inteerval I.

The properties that we are looking for such a µ are the following:

(1) µ(I) = ℓ(I) for every interval,

(2) A ⊆ B =⇒ µ(A) ≤ µ(B),
∞
[ X ∞
(3) µ( Ai ) ≤ µ(Ai ),
i=1 i=1
(4) µ(E + x) = µ(E) for every x ∈ R,
(5) A ∩ B = ∅ =⇒ µ(A ∪ B) = µ(A) + µ(B).

Remark 1.3. Note that (2) implies that µ(∅) = 0 (Exercise) and (5) implies that for any finite
number of disjoint sets A1 , . . . , An , µ( ni=1 An ) = ni=1 µ(Ai ) (Exercise).
S P
♢

But, there is no function µ : 2R → [0, ∞] satisfying all the above properties1.: In order
to show this, first let assume that there is a function µ : 2R → [0, ∞] satisfying all the properties
(1)-(5). Now, let {An } be a pairwise disjoint family of subsets of R. Then, by (2), (3), (5), we
obtain
X∞ ∞
[ n
[ Xn
µ(Ai ) ≥ µ( Ai ) ≥ µ( Ai ) = µ(Ai ) ∀ n ∈ N
i=1 i=1 i=1 i=1

so that we obtain
n
X ∞
[ ∞
X
µ(Ai ) ≤ µ( Ai ) ≤ µ(Ai ) ∀ n ∈ N.
i=1 i=1 i=1

1Wemay recall from set theory that, if A and B are any two nonempty sets, then B A denotes the set all functions
from A to B. Now, given any nonempty set S, if P(S) is the power set of S, that is, family of all subsets of a set
S, then the function φ : P(S) → {0, 1}S defined by

φ(A) = χA ,

is one-one and onto (Exercise). Here χA denotes the characteristic function of A, i.e., χA (x) = 1 if x ∈ A and
χA (x) = 0 if x ∈ S \ A. Thus, P(S) can be identified with {0, 1}S , which is again denoted by 2S
 MEASURE & INTEGRATION 3

Thus, we obtain
∞
[ ∞
X
µ( Ai ) = µ(Ai ) (∗)
i=1 i=1

for every pairwise disjoint family {An } of subsets R.

Theorem 1.4. There does not exist a set function µ : 2R → [0, ∞] satisfying all the above
properties (1)-(5).

Proof. Suppose there exists a set function µ : 2R → [0, ∞] satisfying all the above properties
(1)-(6). We shall arrive at a contradiction.
For x, y ∈ R, define x ∼ y iff x − y ∈ Q. Then ∼ is an equivalent relation2. Hence, in particular,
[0, 1] is a disjoint union of equivalent classes. Let E0 be the subset of [0, 1] containing one element
each from each equivalent class. Such a set exists, by axiom of choice3. Now, let {r1 , r2 , . . .} be
an enumeration of rational numbers in [−1, 1], and let En = E0 + rn for n ∈ N. Then it can be
seen that
∞
[
[0, 1] ⊆ En ⊆ [−1, 2]
n=1

and {En } is a pairwise disjoint family of sets. Hence, by the properties (1), (2), (3), we have
∞
[ 
1⊆µ En ≤ 3.
n=1

Now, by property (∗) which was derived from the earlier properties, we obtain
∞
[ ∞
X
µ( Ei ) = µ(Ei )
i=1 i=1

and by property (4), we have µ(En ) = µ(E0 ) for all n ∈ N. Thus, we arrive at a contradiction! □

We are going to define a function called Lebesgue outer measure on R which has properties
(1)-(4), and then consider a particular family of subsets of R on which the outer measure is going
to satisfy (5), and hence (*) also.

2Equivalence relation on a set is a relation which is Reflexive, Symmetric and Transitive.

3Axiom of choice: Give any family {Sα : α ∈ Λ} of non-empty sets Sα , there exists a set S consisting one
S
element each from each set. In other words, there is a map ψ : Λ → α∈Λ Sα such that ψ(α) ∈ Sα , so that we
have S = {ψ(α) : α ∈ Λ}
4 M.THAMBAN NAIR

2. Lebesgue Outer Measure and Lebesgue Measurable Sets

Department of Mathematics, BITS Pilani, K.K. Birla Goa Campus, Zuarinagar, Goa 403726,
INDIA

Email address: mtnair@goa.bits-pilani.ac.in