REAL ANALYSIS AND MEASURE

THEORY

BY

DR. PRATULANANDA DAS

DEPARTMENT OF MATHEMATICS
JADAVPUR UNIVERSITY
KOLKATA- 700032
WEST BENGAL, INDIA
E-mail : [email protected]
 Chapter 4

(Equivalent to four hours of

teaching/modules)

Convergence of sequences of
measurable functions

1
 4.1. Introduction

During the undergraduate level we have come across two notions of conver-
gences of sequences of real functions, one is the notion of pointwise convergence
and the other is the notion of uniform convergence. In this chapter we consider
sequences of real functions and show that with respect to Lebesgue measure, the
pointwise convergence can be generalized into the notion of convergence almost
everywhere which means usual pointwise convergence outside a set of measure
zero while the notion of uniform convergence can be generalized into the notion
of almost uniform convergence which means that uniform convergence happens
outside a set of measure less than any pre-assigned positive number. We also
define a new type of convergence called convergence in measure for sequences of
measurable functions only which looks a little different from the known notions
of undergraduate level.

We prove several basic properties of these convergences and prove a very

important result called ”Egoroff’s Theorem” which shows that for sequences
of measurable functions defined on a set of finite measure almost uniform con-
vergence and convergence almost everywhere are equivalent and are stronger
than convergence in measure. Finally using Egoroff’s Theorem we observe that
measurable functions are nearly continuous which is known as the ”Lusin’s The-
orem”.

4.2. Sequences of measurable functions

In this section we will briefly consider three notions of convergences of se-

quences of measurable functions. We start with the notion of convergence which
generalizes the notion of pointwise convergence.

Definition 4.1. A sequence of real valued functions {fn }n∈N defined on a mea-
surable set E is said to converge to f almost everywhere (in short fn → f a.e.)
if there exists a measurable set A with µ(A) = 0 such that {fn (x)}n∈N converges
to f (x) for all x ∈ Ac .

Theorem 4.2. (i) If fn → f a.e. and f = g a.e. then fn → g a.e.

(ii) If fn → f a.e. and fn → g a.e. then f = g a.e.

(iii) If fn → f a.e. and fn = gn a.e. then gn → f a.e.

Proof: As all the proofs are almost same, we only prove (ii). Since fn → f a.e.
and fn → g a.e. so there exist measurable sets A and B with µ(A) = µ(B) = 0
such that {fn (x)}n∈N converges to f (x) for all x ∈ X − A and {fn (x)}n∈N
converges to g(x) for all x ∈ X − B. Now clearly A ∪ B is a measurable set with

2
µ(A ∪ B) = 0 and ∀ x ∈ (A ∪ B)c we have

f (x) = limfn (x) = g(x)

n

which shows that f = g a.e.

Definition 4.3. A sequence of functions {fn }n∈N defined on a measurable set

E is said to be Cauchy a.e. if there exists a measurable set A with µ(A) = 0
such that {fn (x)}n∈N is a Cauchy sequence for all x ∈ Ac .

Theorem 4.4. If fn → f a.e. then {fn }n∈N is Cauchy a.e. Conversely if

{fn }n∈N is Cauchy a.e. then there exists a function f such that fn → f a.e.
Further if each fn is measurable then we can choose f to be measurable.

Proof: If fn → f a.e. then there exists a measurable set A with µ(A) = 0 such
that fn → f on Ac . So {fn (x)}n∈N is a Cauchy sequence for all x ∈ Ac which
implies that {fn }n∈N is Cauchy a.e.

Conversely if {fn }n∈N is Cauchy a.e. then there exists a measurable set A
with µ(A) = 0 such that {fn (x)}n∈N is a Cauchy sequence for all x ∈ Ac . Let
f (x) = lim fn (x) for all x ∈ Ac and f takes any value on A. Then evidently
n
fn → f a.e.

Finally define f (x) = 0 for all x ∈ Ac and let gn = fn χAc . Then each gn is
measurable and gn → f on X. Then f being the limit of a sequence of measur-
able functions is measurable.

We now consider another type of convergence associated with measure.

Definition 4.5. A sequence of real valued measurable functions {fn }n∈N de-
fined on a measurable set E is said to converge in measure to a measurable
function f if for any ε > 0,

lim µ [{x : |fn (x) − f (x)| ≥ ε}] = 0.

n

In this case we write fn → f in m .

Theorem 4.6. (i) If fn → f in m and f = g a.e. then fn → g in m where g

is measurable.

(ii) If fn → f in m and fn → g in m then f = g a.e.

Proof: (i) Since f = g a.e. so there exists a measurable set A with µ(A) = 0
such that f = g on Ac . Let ε > 0 be given. Now

{x : |fn (x) − g(x)| ≥ ε} ⊂ {x : |fn (x) − f (x)| ≥ ε} ∪ A.

3
Hence

µ [{x : |fn (x) − g(x)| ≥ ε}] ≤ µ [{x : |fn (x) − f (x)| ≥ ε}] + µ(A).

Since µ(A) = 0 and fn → f in m , so it follows that

lim µ [{x : |fn (x) − g(x)| ≥ ε}] = 0.

n

Therefore fn → g in m .

(ii) Let fn → f in m and fn → g in m . Now for any ε > 0 let Aε =

{x : |f (x) − g(x)| ≥ ε}. Then
n εo n εo
Aε ⊂ x : |fn (x) − f (x)| ≥ ∪ x : |fn (x) − g(x)| ≥ .
2 2
So
hn ε oi hn ε oi
µ(Aε ) ≤ µ x : |fn (x) − f (x)| ≥ + µ x : |fn (x) − g(x)| ≥ .
2 2
Therefore µ(Aε ) = 0 for any ε > 0. Observe that we can write
∞
[
{x : f (x) 6= g(x)} = A n1
n=1

and so µ({x : f (x) 6= g(x)}) = 0 which implies that f = g a.e.

Theorem 4.7. If fn → f in m and fn = gn a.e. then gn → f in m .

Proof: Since fn → f in m then for any ε > 0,

hn ε oi
lim µ x : |fn (x) − f (x)| ≥ = 0.
n 2
Again since fn = gn a.e. so there exist measurable sets Bn with µ(Bn ) = 0 such
that fn = gn on Bnc for each n. Now

|gn (x) − f (x)| ≤ |gn (x) − fn (x)| + |fn (x) − f (x)|

which implies that

ε ε
{x : |gn (x) − f (x)| ≥ ε} ⊂ {x : |gn (x) − fn (x)| ≥} ∪ {x : |fn (x) − f (x)| ≥ }
2 2
ε
⊂ Bn ∪ {x : |fn (x) − f (x)| ≥ }.
2
Since µ(Bn ) = 0 so clearly

lim µ [{x : |gn (x) − f (x)| ≥ ε}] = 0.

n

4
Therefore gn → f in m .

Remark 4.8. In general convergence in measure does not imply convergence

a.e. For example, let us consider the sequence of intervals E1 = [0, 1), E2 =
[0, 21 ), E3 = [ 12 , 1), E4 = [0, 13 ), E5 = [ 13 , 23 ), E6 = [ 23 , 1), · · · . Note that
{µ(En )}n∈N is a decreasing sequence converging to zero. Let us consider a se-
quence of functions {fn }n∈N where fn = χEn for all n ∈ N. Then clearly each
fn is measurable and fn → f in m where f (x) = 0 for all x ∈ [0, 1) because
{x ∈ [0, 1) : |fn (x) − f (x)| ≥ ε} = En for each n ∈ N and for 0 < ε < 1. But
fn 9 f anywhere on [0, 1) because for any x ∈ [0, 1), {fn (x)}n∈N consists of
infinite number of 0’s and 1’s.

Theorem 4.9. For a sequence of measurable functions {fn }n∈N and a mea-
surable function f defined on a measurable set A of finite Lebesgue measure,
fn → f a.e. implies fn → f in m .

Proof: Suppose on the contrary that fn → f a.e. but fn 9 f in m . Then there

exists a ε > 0 such that lim µ(En ) 6= 0 where En = {x ∈ A : |fn (x)−f (x)| ≥ ε}.
n
This implies that there exists a δ > 0 and a subsequence {Enk }k∈N of the
sequence {En }n∈N such that µ(Enk ) > δ for all k ∈ N. Now we take
∞ [
\ ∞
E = lim sup Enk = Eni .
k
k=1 i=k

Then
∞
[
µ( Eni ) ≥ µ(Enk ) > δ
i=k

for all k ∈ N. So
∞ ∞
!
[ [
µ(E) = µ(lim Eni ) = lim µ Eni ≥ δ.
k k
i=k i=k

Now since fn → f a.e. and µ(E) ≥ δ > 0, so there must exist a point
x ∈ E such that fn (x) → f (x). But from the definition of E it follows that
|fnk (x) − f (x)| ≥ ε for infinitely many indices nk which is a contradiction. This
proves the theorem.

Finally we consider the following notion of convergence which is a general-

ization of the notion of uniform convergence.

Definition 4.10. A sequence of real valued measurable functions {fn }n∈N is

said to converge almost uniformly to a real valued measurable function f defined
on a measurable set E if for any δ > 0 there exists a measurable set F with
µ(F ) < δ such that fn → f uniformly on F c . In this case we write fn → f a.u.

5
Theorem 4.11 (Egoroff ’s Theorem). If {fn }n∈N is a sequence of real valued
measurable functions defined on a measurable set A of finite Lebesgue measure
and fn → f a.e. on A where f is also measurable then fn → f a.u.

Proof: First we assume that fn → f everywhere on A. For m, n ∈ N, we define

 
m 1
Fn = x ∈ A : |fk (x) − f (x)| ≥ for some k ≥ n
m
∞  
[ 1
= x ∈ A : |fk (x) − f (x)| ≥ .
m
k=n

Now fk − f is measurable and so each set in the union on the right hand side is
measurable which implies that each set Fnm is also measurable. Again m
\{Fn }n∈N
m m
is a decreasing sequence of sets. We first observe that lim Fn = Fn = φ.
n
n
For if x ∈ A then as fn (x) → f (x), so we can find r ∈ N such that
1
|fn (x) − f (x)| < for all n ≥ r.
m
\
/ Frm . Since this is true for all x ∈ A, so
This implies that x ∈ Fnm = φ. Again
n
as µ(A) < ∞, so \
lim µ(Fnm ) = µ( Fnm ) = µ(φ) = 0.
n
n
For a given δ > 0 we can then choose n(m) ∈ N such that
m δ
µ(Fn(m) )< .
2m
Now we define [
m
F = Fn(m) .
n
Then F is clearly a measurable set and
∞ ∞
X
m
X δ
µ(F ) = µ(Fn(m) )≤ m
= δ.
m=1 m=1
2
We now show that fn → f uniformly on A − F . Observe that
∞
[
m
A−F = A− Fn(m)
m=1
∞
\
m
= (A − Fn(m) )
m=1
∞  c
\ 1
= x ∈ A : |fk (x) − f (x)| ≥ for some k ≥ n(m)
m=1
m
∞  
\ 1
= x ∈ A : |fk (x) − f (x)| < for all k ≥ n(m) .
m=1
m

6
 1
For ε > 0 given, if we choose m ∈ N such that m < ε then clearly for all
x∈A−F
1
|fk (x) − f (x)| < <ε
m
for all k ≥ n(m). Therefore fn → f a.u.

Finally we assume that fn → f a.e. Then there exists a measurable set E

with µ(E) = 0 such that fn → f on E c . Take gn = fn χEc and g = f χEc . Then
gn , g are all measurable functions and gn → g everywhere on A. So by the first
part there exists a measurable set G with µ(G) < δ such that gn → g uniformly
on A−G. If F = E ∪G then µ(F ) ≤ δ and fn = gn → g = f uniformly on A−F .

One should be careful that the notion of almost uniform convergence is not
same as uniform convergence almost everywhere. To see this let {rk }k∈N be the
sequence of rational numbers in [0, 1] and let fn be the function which takes the
value 1 at the points r1 , r2 , . . . , rn and takes the value 0 at all other points of
[0, 1] for every n ∈ N. Then it is easy to see that fn → f where f = χQ . In
fact the convergence is uniform on the set of irrationals and so outside a set of
Lebesgue measure zero. Thus fn converges to f uniformly almost everywhere.
Now if we take gn = χ(0, n1 ) for each n then it is easy to verify that gn converges
to χQ everywhere and so by Egoroff’s theorem gn converges to χQ almost uni-
formly. But there is no set of Lebesgue measure zero such that gn converges to
χQ uniformly outside that set.

The following theorem shows that in general almost uniform convergence is

stronger than the notions of convergence almost everywhere and convergence in
measure.

Theorem 4.12. fn → f a.u. ⇒ fn → f a.e. and also fn → f in m if in addition

fn , f are measurable.

Proof: Since fn → f a.u., so for any m ∈ N we can find a measurable set Fm

∞
\
1 c
such that µ(Fm ) < m and on Fm , fn → f uniformly. Let F = Fm . Then
m=1
clearly µ(F ) = 0 and fn (x) → f (x) for all x ∈ F c and so fn → f a.e.

Now let fn , f be all measurable and fn → f a.u. Let ε > 0 be given and

En = {x : |fn (x) − f (x)| ≥ ε} .

1
Let δ > 0 be given. Choose m ∈ N such that m < δ. Let Fm be the set as
c
defined before. Now as fn → f uniformly on Fm so there exists a n0 ∈ N such
that
|fn (x) − f (x)| < ε

7
 c c
for all n ≥ n0 and x ∈ Fm . Hence Fm ∩ En = φ for all n ≥ n0 . Then En ⊂ Fm
for all n ≥ n0 and so µ(En ) < µ(Fm ) < δ for all n ≥ n0 . this shows that

lim µ(En ) = 0
n

and so fn → f in m .

4.3. Lusin’s theorem

In the last section of this chapter we prove an important result on measur-

able functions which states that a measurable function is nearly continuous.

Lemma 4.13. Let s be a simple function defined on (a, b). Then for any ε > 0
there is a closed set F ⊂ (a, b) such that the restriction of s on F , s|F is con-
tinuous on F and µ[(a, b) − F ] < ε.

Proof: For simplicity let us write I = (a, b) and let the canonical representation
of s be given by
Xk
f= αi χEi .
i=1

Since each set Ei is measurable, by Theorem 2.5 (iv) we can find a closed set
k
[
ε
Fi ⊂ Ei such that µ(Ei − Fi ) < 2k . Now set F = Fi . Then F is a closed set
i=1
contained in I and
k
X
µ(I − F ) ≤ µ(Ei − Fi ) < ε.
i=1

Now if x ∈ F then x ∈ Fi for some i. Further, since Fi ’s are closed and pairwise
disjoint so we can find an open interval J containing x such that J ∩ F = J ∩ Fi .
From definition it follows that s is constant on J ∩ F and so s|F is continuous
at x. Since this is true for any x ∈ F it readily follows that s|F is continuous
on F .

Lemma 4.14. Let f be a measurable function defined on (a, b). Then for any
ε > 0 there is a closed set F ⊂ (a, b) such that the restriction of f on F , f |F is
continuous on F and µ[(a, b) − F ] < ε.

Proof: Since f is measurable, so by Theorem 3.14 there exists a sequence of

simple functions {sn }n∈N defined on (a, b) which converges to f pointwise. Let
ε > 0 be given. For each n ∈ N, by the above lemma, we can find a closed set
ε
Hn ⊂ (a, b) such that sn |Hn is continuous on Hn where µ[(a, b) − Hn ] < 2n+1 .

8
 ∞
\
Take H = Hn . Then H is a closed subset of (a, b) and we have
n=1

∞ ∞
X X ε ε
µ([(a, b) − H]) ≤ µ([(a, b) − Hn ]) ≤ n+1
= .
n=1 n=1
2 2

Again since sn → f on the measurable set H where µ(H) < ∞, so by Egoroff’s

Theorem and Theorem 2.5 (iv) we can find another closed set F ⊂ H such that
µ(H − F ) < 2ε and sn |F converges uniformly to f on F . Now evidently

µ([(a, b) − F ]) ≤ µ([(a, b) − H]) + µ(H − F ) < ε.

Observe that as each sn |F is continuous on F and sn |F converges uniformly to

f on F so f |F is continuous on F . This completes the proof of the result.

For the final result in this chapter we will need the following version of an
well known theorem from Topology which we state here without proof. The
general statement (a characterization of normality) and the detailed proof can
be seen in any standard book on general topology.

Theorem 4.15 (Tietze Extension Theorem). Let F be a closed subset

of R. If f : E → R is continuous on F then there is a continuous function
g : R → R such that f = g on F .

Theorem 4.16 (Lusin’s Theorem). If f : R → R is a measurable function

then for any ε > 0 there is a closed set F and a continuous function g : R → R
such that f = g on F and µ(F c ) < ε.
∞
[
Proof: Note that R can be written in the form R = In ∪ N where {In }n∈N
n=1
is the sequence of all open intervals of the form (m, m + 1) for integers m. Note
that µ(N) = 0. Let ε > 0 be given. Now applying Theorem 4.14 on f |In for
each n we can find a closed set Fn ⊂ In such that µ(In − Fn ) < 2εn and f |Fn is
∞
[
continuous on Fn . Set F = Fn . Then observe that F is a closed set and
n=1

∞ ∞
X X ε
µ(F c ) ≤ µ(In − Fn ) + µ(N) ≤ = ε.
n=1 n=1
2n

Now f |F is continuous on the closed set F and so by Tietze Extension Theorem,

there is a continuous function g : R → R such that f = g on F . This completes
the proof of the theorem.

4.4. Self-check Exercises

9
Exercise 4.1. If fn → f almost everywhere and fn ≥ 0 almost everywhere
then show that f ≥ 0 almost everywhere.

Solution: We can find a set E with µ(E) = 0 such that fn (x) → f (x) for all
x ∈ E c . Again since fn ≥ 0 almost everywhere so for each n there is a set[ En
with µ(En ) = 0 such that fn (x) ≥ 0 for all x ∈ Enc . Consequently G = E ∪ En
n
is a set with µ(G) = 0 such that for all x ∈ Gc we have

f (x) = lim fn (x) ≥ 0.

n

Exercise 4.2. A sequence of measurable functions {fn }n∈N defined over a

measurable set D is said to be Cauchy in measure if for any ε > 0,

µ({x ∈ D : |fm (x) − fn (x)| ≥ ε}) → 0

as m, n → ∞. Prove that if fn → f in measure then {fn } is Cauchy in measure.

Solution: let us write Emn = {x ∈ D : |fm (x) − fn (x)| ≥ ε} and En = {x ∈

D : |f (x) − fn (x)| ≥ 2ε }. Since we have

|fn (x) − fm (x)| ≤ |fn (x) − f (x)| + |fm (x) − f (x)|

so we have Emn ⊂ Em ∪ En which implies that

µ(Emn ) ≤ µ(Em ) + µ(En ) → 0 if m, n → ∞.

Exercise 4.3. For measurable functions fn , f defined over a measurable set

D, if fn → f in measure and fn ≥ 0 almost everywhere then show that f ≥ 0
almost everywhere.

Solution: Let En be a set with µ(En ) = 0 such that fn ≥ 0 on Enc . Redefining

gn = fn .χEn we see that fn = gn almost everywhere where each gn is measur-
able and gn ≥ 0 everywhere. Since fn → f in measure it is easy to check that
gn → f in measure.

Let ε > 0 be given. Write A = {x ∈ D : f (x) ≤ −ε} and as always

En = {x ∈ D : |gn (x) − f (x)| ≥ ε}. For x ∈ A, note that

−ε ≥ f (x) = [f (x) − gn (x)] + gn (x) ≥ f (x) − gn (x)

which shows that |f (x) − gn (x)| ≥ ε and so x ∈ En . This shows that A ⊂ En

for all n. Since gn → f in measure so µ(En ) → 0 as n → ∞ which implies that
µ(A) = 0. Now the result follows from the fact that
∞
[ 1
{x ∈ D : f (x) < 0} = {x ∈ D : f (x) ≤ }
k
k=1

10
being countable union of µ-null sets, has also measure zero.

Exercise 4.4. Let f : R → R be a function. If for each η > 0 there is a closed

set F and a continuous function h : R → R such that µ(F c ) < η and f = h on
F then prove that f is measurable.

Solution: By the given condition, for each k ∈ N we can find a closed set Fk
and a continuous function hk : R → R such that µ(Fkc ) < k1 and f = hk on Fk .
[∞
Defining F = Fk we see that µ(F c ) = 0. Since each hk is continuous and so
k=1
is measurable. Then for each k, f |Fk is measurable. We now verify that f |F is
also measurable. For any real number α
∞
[ ∞
[
{x ∈ F : f (x) > α} = {x ∈ Fk : f (x) > α} = {x ∈ Fk : f |Fk (x) > α}.
k=1 k=1

Since each set on the right hand side is measurable so it follows that f |F is
measurable. Finally since µ(F ) = 0, so it can be easily checked that f is also
measurable.

Exercise 4.5. For every measurable function f : [a, b] → R prove that there
exists a sequence of continuous functions {fn }n∈N on [a, b] which converges to
f almost everywhere.

Solution: We use Lusin’s Theorem to get for each positive integer k, a closed
set Fk ⊂ [a, b] and a continuous function hk : [a, b] → R such that f = hk on Fk
and µ([a, b] − Fk ) < k1 . Observe that

1
lim µ({x ∈ [a, b] : |hn (x) − f (x)| ≥ }) ≤ lim µ([a, b] − Fn ) = 0.
n→∞ k n→∞

Therefore in view of above we can choose an increasing sequence of positive

integers {nk }k∈N such that µ(Bk ) < 21k where we write

1
Bk = {x ∈ [a, b] : |hnk (x) − f (x)| ≥ }.
k
∞ [
\ ∞
Let B = Bk . Using continuity of Lebesgue measure it is easy to check
n=1 k=n
that µ(B) = 0. We now show that hnk → f as k → ∞ on [a, b] − B. Let
x ∈ [a, b] − B and let ε > 0 be given. Now x ∈
/ B implies that there is a positive
integer M such that x ∈ / Bk for all k ≥ M . We can choose K > M such that
1
K < ε. Since x ∈ / Bk for all k ≥ K so

1
|hnk (x) − f (x)| < < ε ∀ k ≥ K.
K

11
Summary: In this chapter we have defined three types of convergences with
respect to Lebesgue measure, namely, pointwise convergence almost everywhere,
convergence in measure and almost uniform convergence. Besides proving sev-
eral basic properties, we have primarily proved two main results. One is ”Ego-
roff’s Theorem” which shows that on sets of finite measure, almost uniform
convergence and convergence almost everywhere are equivalent for sequences of
measurable functions. The second result is ”Lusin’s Theorem” which gives a
very important property of measurable functions showing that they are nearly
continuous.

Following are the main definitions and points to be remembered.

• A sequence of real valued functions {fn }n∈N is said to converge to f al-

most everywhere (in short fn → f a.e.) if there exists a measurable set A with
µ(A) = 0 such that {fn (x)}n∈N converges to f (x) for all x ∈ Ac .

• A sequence of real valued measurable functions {fn }n∈N is said to converge

in measure to a measurable function f if for any ε > 0,

lim µ [{x : |fn (x) − f (x)| ≥ ε}] = 0.

n

In this case we write fn → f in m .

• A sequence of real valued functions {fn }n∈N is said to converge to a real

valued function f almost uniformly if for any δ > 0 there exists a measurable
set F with µ(F ) < δ such that fn → f uniformly on F c . In this case we write
fn → f a.u.

• (Egoroff ’s Theorem) If {fn }n∈N is a sequence of real valued measurable

functions defined on a measurable set A of finite Lebesgue measure and fn → f
a.e. on A where f is also measurable then fn → f a.u.

Consequently for a sequence of real valued measurable functions defined on a

measurable set A of finite Lebesgue measure, convergence a.e. and convergence
a.u. are equivalent and are stronger than convergence in measure.

• (Lusin’s Theorem) If f : R → R is a measurable function then for any

ε > 0 there is a closed set F and a continuous function g : R → R such that
f = g on F and µ(F c ) < ε.

Acknowledgement: While writing this chapter the author has followed the
excellent book by R.A. Gordon and also in certain places the book by S.K.
Berberian. The full references of these books is given in ”Learn More” section.

12
 SELF ASSESSMENT

Example 4.1. Which of the following statements is always true?

(a) Convergence almost everywhere implies convergence in measure.

(b) Convergence in measure implies convergence almost everywhere.
(c) Convergence almost everywhere implies convergence almost uniformly.
(d) Convergence almost uniformly implies both convergence in measure and
convergence almost everywhere.

Example 4.2. Which of the following statements is always true?

(a) The limit function of an almost everywhere convergent sequence of func-

tions is unique.
(b) The limit function of an almost everywhere convergent sequence of func-
tions is measurable if every function of the sequence is so.
(c) The limit function of an almost everywhere convergent sequence of func-
tions is not unique but are equal almost everywhere.
(d) The limit function of an almost uniform convergent sequence of functions
is unique.

Example 4.3. Which of the following statements is always true?

(a) Convergence almost everywhere implies convergence in measure if the mea-

sure space is endowed with counting measure.
(b) Convergence almost everywhere implies convergence in measure if the mea-
sure space is endowed with a σ-finite measure.
(c) Convergence almost everywhere implies convergence in measure if the mea-
sure space is endowed with a finite measure.
(d) Convergence almost everywhere implies convergence almost uniformly if
the measure space is endowed with counting measure.

Example 4.4. Let fn and gn be two sequences of measurable functions defined

on a Lebesgue measurable set E such that fn → f and gn → g in measure on
E. Which of the following statements is true ?

13
 (a) fn · ϕ → f · ϕ in measure on E, for each measurable function ϕ defined on
E.
(b) fn 2 → f 2 in measure on E.
(c) fn · gn → f · g in measure on E.
(d) There exists a subsequence fnk of fn such that fnk → f pointwise a.e. on
E.

Example 4.5. Let fn be a sequence of measurable functions defined on a

Lebesgue measurable set E such that fn → f in measure on E and g be a mea-
surable function on E that is finite a.e. on E. Which of the following statements
is false ?

(a) If µ(E) = ∞, then there exists a subsequence fnk such that fnk 6→ f in
measure on E.
(b) Every subsequence fnk converges in measure to f .
(c) If fn → g in measure on E then f = g a.e. on E
(d) If f = g a.e. on E then fn → g in measure on E.

Example 4.6. Which of the following statements is true ?

(a) Let fn : R → R be a sequence of Lebesgue measurable functions such

that fn converges pointwise to a function f : R → R on R. Then for any
ε > 0 and δ > 0 there exists a Lebesgue measurable set A ⊂ R such that
µ(A) < δ and there exists a N ∈ N such that for all x ∈/ A and n ≥ N ,
|fn (x) − f (x)| < ε.
(b) Let fn : R → R be a sequence of Lebesgue measurable functions such that
fn converges pointwise a.e. to a function f : R → R on R. Then for any
ε > 0 and δ > 0 there exists a Lebesgue measurable set A ⊂ R such that
µ(A) < δ and there exists a N ∈ N such that for all x ∈/ A and n ≥ N ,
|fn (x) − f (x)| < ε.
(c) Let fn : R → R be a sequence of Lebesgue measurable functions such that
fn converges pointwise a.e. to a function f : R → R on R. Then for any
ε > 0 there exists a Lebesgue measurable set A ⊂ R such that µ(A) < ε
and fn converges to f uniformly on E \ A.
(d) Let fn : [0, 1] → R be a sequence of Lebesgue measurable functions such
that fn converges pointwise a.e. to a function f : [0, 1] → R on [0, 1].
Then for any ε > 0 there exists a Lebesgue measurable set A ⊂ [0, 1] such
that µ(A) < ε and fn converges to f uniformly on E \ A.

14
 1
Example 4.7. Let fn be a sequence of functions defined by fn (x) = (x4 + nx ) 2
on E = (0, ∞) for each n ∈ N. Let f (x) = x2 on E. Which of the following
statements is true ?

(a) fn → f in measure and also fn 2 → f 2 in measure on E.

(b) fn → f in measure but fn 2 6→ f 2 in measure on E.

(c) fn 6→ f in measure but fn 2 → f 2 in measure on E.

(d) fn 6→ f in measure and also fn 2 6→ f 2 in measure on E.

Example 4.8. Let fn be a sequence of measurable functions defined on a

Lebesgue measurable set E. Which of the following statements is true ?

(a) If fn → 0 in measure on E then fn → 0 pointwise a.e. on E.

(b) If fn → 0 uniformly on E then fn → 0 in measure on E.

(c) If µ(E) < ∞ and fn → 0 in measure on E then fn → 0 pointwise a.e. on

E.
(d) If fn → 0 pointwise a.e. on E then fn → 0 in measure on E.

Example 4.9. Do convergence almost uniformly implies uniform convergence

almost everywhere?

Example 4.10. Do uniform convergence almost everywhere implies conver-

gence almost uniformly?

SOLUTIONS

•Solution of Example 4.1: Ans: (d).

•Solution of Example 4.2: Ans: (c).

•Solution of Example 4.3: Ans: (c).

15
•Solution of Example 4.4: Ans: (d).
( For the assertions (a), (b) and (c), µ(E) < ∞ is required.)

•Solution of Example 4.5: Ans: (a).

( If fn → f in measure on E then every subsequence fnk converges in measure
to f (E does not necessarily need to have finite measure). Now if fn → g in
measure then there is a subsequence fnk → g pointwise a.e. on E. Also since
fnk → f in measure, so there exists a subsequence fnkm tof pointwise a.e. on E.
It follows that fnkm converges pointwise a.e. to both f and g on E and hence
f = g a.e. on E.
Now if f = g a.e. on E, then consider E0 = {x ∈ E : f (x) = g(x)} and so
µ(E \ E0 ) = 0. Now for each ε > 0 we have

µ ({x ∈ E : |fn (x) − f (x)| ≥ ε}) = µ ({x ∈ E0 : |fn (x) − f (x)| ≥ ε})
+µ ({x ∈ E \ E0 : |fn (x) − f (x)| ≥ ε})
= µ ({x ∈ E0 : |fn (x) − f (x)| ≥ ε})
= µ ({x ∈ E0 : |fn (x) − g(x)| ≥ ε})
= µ ({x ∈ E0 : |fn (x) − g(x)| ≥ ε})
+µ ({x ∈ E \ E0 : |fn (x) − g(x)| ≥ ε})
= µ ({x ∈ E : |fn (x) − g(x)| ≥ ε})

since µ(E \ E0 ) = 0. Hence fn → g in measure on E.)

• Solution of Example 4.6: Ans: (d).

( (a),(b) and (c) are true for finite measure spaces only.)

• Solution of Example 4.7: Ans: (b).

•Solution of Example 4.8: Ans: (b).

• Solution of Example 4.9: False.

• Solution of Example 4.10: True.

16
 LEARN MORE

In his 1944 text, Lectures on the Theory of Functions, J.E. Littlewood out-
lined three theorems that give great insight into the essentials of measure theory.
These three principles give one an intuitive way of thinking about measurable
sets and functions and are known as Littlewood’s Three Principles.
Principle One: Every measurable set of finite outer measure is almost the
finite, disjoint union of open intervals:

Let µ∗ (A) < ∞. A is measurable if and only If for each ε > 0 there
is a finite collection of disjoint open intervals {Ik : 1 ≤ k ≤ n} such that
[n
∗
µ (A∆( Ik )) < ε. (see Exercice 2.8 and 2.9)
k=1
Principle Two: Every pointwise-convergent sequence of measurable func-
tions is nearly uniformly convergent:

Theorem 4.11 (Egoroff ’s Theorem). If {fn }n∈N is a sequence of real valued

measurable functions defined on a measurable set A of finite Lebesgue measure
and fn → f a.e. on A where f is also measurable then fn → f a.u.

Principle Three: Every measurable function is nearly continuous:

Theorem 4.16 (Lusin’s Theorem). If f : R → R is a measurable function

then for any ε > 0 there is a closed set F and a continuous function g : R → R
such that f = g on F and µ(F c ) < ε.

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:

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

Publishing house, New York, 1963.

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

American Mathematical Society, 1994.

17
 5. P.R. Halmos, Measure Theory, Reprint, Springer Verlag, 1974.

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

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

Cambridge,Mass, 1953.

8. I.P. Natanson, Theory of functions of a real variable, Volume 1 and 2,

Frederick Ungar Publishing Co., New York, 1964.

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.

Web Resources with links:

http://www.kurims.kyoto u.ac.jp/EMIS/journals/AUA/pdf
/51 501 5 lesnic characterizations of the functions.pdf

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

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

18