CHAPTER 2

Measurable Functions

In developing the Lebesgue integral we shall be concerned with

classes of real-valued functions defined on a set X. In various applica-
tions the set X may be the unit interval I = [0, 11 consisting of all real
numbers x satisfying 0 c x c 1, it may be the set N = {1, 2, 3, }
of natural numbers; it may be the entire real line R; it may be all of
the plane; or it may be some other set. Since the development of the
integral does not depend on the character of the underlying space X,
we shall make no assumptions about its specific nature.
Given the set X, we single out a family X of subsets of X which are
" well-behaved " in a certain technical sense. To be precise, we shall
assume that this family contains the empty set 0 and the entire set X.
and that X is closed under complementation and countable unions.

2.1 DEFINITION. A family X of subsets of a set X is said to be a

Q-algebra (or a Q-field) in case:

(i) 0, X belong to X.
(ii) If A belongs to X, then the complement W(A) = X \ A belongs
to X.
(iii) If (An} is a sequence of sets in X, then the union Un A,, be-
longs to X.

An ordered pair (X, X) consisting of a set X and a Q-algebra X of

subsets of X is called a measurable space. Any set in X is called an
6
 Measurable Functions 7

X-measurable set, but when the Q-algebra X is fixed (as is generally the
case), the set will usually be said to be measurable.
The reader will recall the rules of De Morgan:

(2.1) u Aa = f l c(Aa}, r n Aa = (J c(A j .

a a a a

It follows from these that the intersection of a sequence of sets in X

also belongs to X.
We shall now give some examples of Q-algebras of subsets.
2.2 EXAMPLES. (a) Let X be any set and let X be the family of all
subsets of X.
(b) Let X be the family consisting of precisely two subsets of X,
namely 0 and X.
(c) Let X= {192939 } be the set N of natural numbers and let
X consist of the subsets
01 { 1, 3, 5, }, {2, 4, 6, }, X.
(d} Let X be an uncountable set and X be the collection of subsets
which are either countable or have countable complements.
(e) If Xl and X2 are Q-algebras of subsets of X, let X3 be the inter-
section of X, and X2; that is, X3 consists of all subsets of X which
belong to both Xl and X2. It is readily checked that X3 is a Q-algebra
(f) Let A be a nonempty collection of subsets of X. We observe
that there is a smallest Q-algebra of subsets of X containing A. To see
this, observe that the family of all subsets of X is a Q-algebra containing
A and the intersection of all the Q-algebras containing A is also a
Q-algebra containing A. This smallest Q-algebra is sometimes called
the Q-algebra generated by A.
(g) Let X be the set R of real numbers. The Borel algebra is the
Q-algebra B generated by all open intervals (a, b) in R. Observe that
the Borel algebra B is also the Q-algebra generated by all closed intervals
[a, b] in R. Any set in B is called a Borel set.
(h} Let X be the set A of extended real numbers. If E is a Borel
subset of R, let
(2 2) El = .E U { - oo} , .E2 = .E U { + oo} , .E3 = .E u {- oo , + oo} ,
S The Elements of Integration

and let B be the collection of all sets E. El, E2, E3 as E varies over B.
It is readily seen that B is a a-algebra and it will be called the extended
Borel algebra.
In the following, we shall consider a fixed measurable space (X, X).
A function f on X to R is said to be X-measurable
2.3 DEFINITION.
(or simply measurable) if for every real number a the set
(2.3) {x E X : f (x) > a}
belongs to X.
The next lemma shows that we could have modified the form of the
sets in defining measurability.
2.4 LEMMA. The following statements are equivalent for a f unction
f on X to R :
(a) For every a c R, the set A,, = {x c X : f (x) > a} belongs to X
(b) For every a c R, the set Ba = {x c X : f(x} C a} belongs to X
(c) For every a E R, the set Ca = {x E X : f (x) >, a} belongs to X
(d) For every a E R, the set Da = {x c X : f(x) < a} belongs to X
PROOF.Since Ba and Aa are complements of each other, statement
(a) is equivalent to statement (b). Similarly, statements (c) and (d)
are equivalent. I f (a) holds, then A,,-,I,, belongs to X for each n and
since
00

n
Ca = n1 Aa--1/n

it follows that Ca E X Hence (a) implies (c). Since

COD

Aa = U Ca+iin,
n-= 1

it follows that (c) implies (a) Q.E.D

2.5 EXAMPLES. (a) Any constant function is measurable. For,

if f(x} = c for all x E X and if a >, c, then
{X E X - f(x) > a}
whereas if a < c, then
{xEX:f(x)>a}=X.
 Measurable Functions 9

(b) If E E X, then the characteristic function XE' defined by

XE(x) - 1, x cz E,
= 0, x 0 E,
is measurable. In fact, {x E X : XE(x} > a} is either X, E, or
(c) If X is the set R of real numbers, and X is the Borel algebra B,
then any continuous function f on R to R is Borel measurable (that is,
B-measurable). In fact, if f is continuous, then {x E R : f(x) > a} is
an open set in B and hence is the union of a sequence of open intervals.
Therefore, it belongs to B.
(d) If X = R and X = B, then any monotone function is Borel
measurable. For, suppose that f is monotone increasing in the sense
that x c x' implies f (x) c f(x'). Then {x E R : J '(x) > a} consists of
a half-line which is either of the form {x c R : x > a} or the form
a}, orisRor0.
Certain simple algebraic combinations of measurable functions are
measurable, as we shall now show.
2.6 LEMMA. Let f and g be measurable real-valued functions and let
c be a real number. Then the functions
cf, f2,
, f+ g, fg, Ill,
are also measurable.
PROOF. (a) If c = 0, the statement is trivial. If c > 0, then
{x E X : cf(x) > a} = {x E X : f (x) > a/c} E X.
The case c < 0 is handled similarly.
(b) If a < 0, then {x E X : (1(x))2 > a} = X; if a >, 0, then
{x E X : f (X))2 > a}
= {x E X : f (x) > '\/a-} U {x E X : f (x) < - V a} .
(c) By hypothesis, if r is a rational number, then
Sr = {x E X : f (x) > r} f1 {x E X : g(x) > a - r}
belongs to X. Since it is readily seen that
{xEX: (f+g)(x) > a) _U{Sr: rrational},
it follows that f + g is measurable.
10 The Elements of Integration

(d) Since fg = J[(f + g)2 - (f - g)2J, it follows from parts (a), (b),
and (c) thatfg is measurable.
(e) If a < 0, then {x c- X : > a} = X, whereas if a >, 0, then
I

{x E X : I > a} = {x E X : f(X) > a} U {x E X : f(X) < - a}.

Thus the function If I is measurable. Q.E.D.

If f is any function on X to R, let f + and f - be the nonnegative

functions defined on X by
(2.4) f + (x} = sup {.f (x) , O} , f - (x} = sup { -.f (x), 0}.
The function f + is called the positive part off and f - is called the
negative part off. It is clear that
(2.5) f =f+ -f- and 111 =f+
+f-
and it follows from these identities that
(2.6) f + = 2(I fI +J)-5 f = RI A _J)-
In view of the preceding lemma we infer that f is measurable if and
only if f + and f - are measurable.
The preceding discussion pertained to real-valued functions defined
on a measurable space. However, in dealing with sequences of
measurable functions we often wish to form suprema, limits, etc., and
it is technically convenient to allow the extended real numbers -oo, +oo
to be taken as values. Hence we wish to define measurability for
extended real-valued functions and we do this exactly as in Definition 2.3.
2.7 DEFINITION. An extended real-valued function on X is X-
measurable in case the set {x c X : f(x) > a} belongs to X for each real
number a. The collection of all extended real-valued X-measurable
functions on X is denoted by M(X, X) .
Observe that if f c M(X, X), then
Co

{x c x : f(x) = +oo} _ fl {x c X : f(x) > n},

n=1

x E X : f(X) _ 00} _x n=1

c X :f(x)
U -n ,

so that both of these sets belong to X.

 Measurable Functions 11

The following lemma is often useful in treating extended real-valued

functions.
2.8 LEMMA. An extended real-valued function f is measurable if and
only if the sets
A = {xEX : f(x) = +oo}, B = {xEX : f(x) _ -oo}
belong to X and the real-valued function fl defined by
fi(x} = f(x), if x 0 A U B,
= 0, ifxEA UB,
is measurable.
PROOF. If f is in M(X, X), it has already been noted that A and B
belong to X. Let a E R and a >, 0, then
{x c X : fl (x) > a} = {x c X : f (x) > a} 1 A.
If a < 0, then
{x E X (x) > a} = {x c X : f (x) > a} U B.
Hence fl is measurable.
Conversely, if A, B E X and fl is measurable, then
{x c X : f (x) > a} = {x c X : f, (x) > a} U A
when a >, 0, and
{xEX:f(x) > a} = {xEX:fl(x) > a}\B
when a < 0. Therefore f is measurable. Q.E.D.

It is a consequence of Lemmas 2.6 and 2.8 that if f is in M(X, X),

then the functions
cf, f2, IfII f+j f
also belong to M(X, X).
The only comment that need be made is that we adopt the convention
that 0(±oo) = 0 so that cf vanishes identically when c = 0. If f and g
belong to M(X, X), then the sum f + g is not well-defined by the
formula (f + g)(x) = f (x) + g (x) on the sets
El = {x E X : f (x) = - oo and g(x) = + oo} ,
E2 = {x E X : f (x) = +oo and g(x) _ -00}1
12 The Elements of Integration

both of which belong to X. However, if we define f + g to be zero

on .El U E2, the resulting function on X is measurable. We shall
return to the measurability of the productfg after the next result.
2.9 LEMMA. Let (fn) be a sequence in M (X, X) and define the
functions
f(x) = inffn(x), F(x) = supfn(x),
f*(x) = lim inf fn(x), F*(x) = lim sup f7t(x).
Then f, F, f *, and F* belong to X1 (X, X).
PROOF. Observe that
Go

{xeX: f(x)>a}= n=1

n {xeX:ff(x}>
{x c- X : F(x) > a} = U {x E X : fn(x) > U},
n=1

so that f and F are measurable when all the fn are. Since

f * (x) = sup inf fm(x) ,

n>,1 m,n

F*(x) = inf sup fm(x} ,

n>,1 m>n

the measurability off * and F* is also established. Q.E.D.

2.10 COROLLARY. If (fn) is a sequence in M (X, X) which converges

to f on X, then f is in M(X, X).
PROOF. In this case f(x) = lim fn(x) = lim inf f,(x). Q.E.D.

We now return to the measurability of the product f g when f, g

belong to M(X, X). If n e N, let, fn be the "truncation off" defined by
fn(x) =1(x) , if I .f (x) I c n,
=n, if f(x)>n,
= -n, if f(x) < -n.
Let gm be defined similarly. It is readily seen that fn and gm are
measurable (see Exercise 2. K). It follows from Lemma 2.6 that the
product fn gm is measurable. Since
f(X) gm(x) = lirafn(x) gm(x), x E X,
n
 Measurable Functions 13

it follows from Corollary 2.10 that f g.. belongs to M(X, X). Since

(fg)(x) = R X) g(x) = lint f (x) gm(x), x E X,

m

another application of Corollary 2.10 shows that f g belongs to M(X, X) .

It has been seen that the limit of a sequence of functions in M(X, X)
belongs to M(X, X) . We shall now prove that a nonnegative function
f in M(X, X) is the limit of a monotone increasing sequence (p,) in
M(X, X). Moreover, each g , can be chosen to be nonnegative and to
assume only a finite number of real values.
2.11 LEMMA. If f is a nonnegative function in M(X, X), then there
exists a sequence (g) in M(X, X) such that
(a) 0 c 9)n(x) c 9)n + 2 (x) for x c- X, n c- N.
(b) f (x) = lim 9).(x) for each x c- X.
(c) Each 9)n has only a finite number of real values.
PROOF. Let n be a fixed natural number. If k = 0, 1, ... , n2n - 1,
let Eke be the set
.Eke = {x E X : k2 - n C f(X) < (k + 1) 2 - n},

and if k = n2n, let .Eke be the set {x E X : f (x) > n} . We observe that
the sets {.Eke : k = 0, 1, ... , n2n} are disjoint, belong to X, and have
union equal to X. If we define 9pn to be equal to k2 - n on Eke, then 9)n
belongs to M(X, X). It is readily established that the properties (a),
(b), (c) hold. Q.E.D.

COMPLEX-VALUED FUNCTIONS

It is frequently important to consider complex-valued functions

defined on X and to have a notion of measurability for such functions.
We observe that if f is a complex-valued function defined on X, then
there exist two uniquely determined real-valued functions f, , f2 such
that
f = A +if2.
(Indeed, f, (x) = Ref (x) , f2(x) = Im f (x) , for x c- X.) We define the
14 The Elements of Integration

complex-valued function f to be measurable if and only if its real and

imaginary parts Jj and f2, respectively, are measurable. It is easy to
see that sums, products, and limits of complex-valued measurable
functions are also measurable.

FUNCTIONS BETWEEN MEASURABLE SPACES

In the sequel we shall require the notion of measurability only for
real- and complex-valued functions. In some work, however, one
wishes to define measurability for a function f from one measurable
space (X, X) into another measurable space (Y, Y). In this case one
says that f is measurable in case the set
f -' (E) = {x c- X : f (x) e E}
belongs to X for every set E belonging to Y. Although this definition
of measurability appears to differ from Definition 2.3, it is not difficult
to show (see Exercise 2.P) that Definition 2.3 is equivalent to this
definition in the case that Y = R and Y = B.
This definition of measurability shows very clearly the close analogy
between the measurable functions on a measurable space and continuous
functions on a topological space.

EXERCISES
2.A. Show that [a, b] =nn,= 1 (a -- 1 In, b + 1 In). Hence any Q-
algebra of subsets of R which contains all open intervals also contains
all closed intervals. Similarly, (a, b) = U n =1 [a + 1In, b - 1 In], so
that any a-algebra containing all closed intervals also contains all
open intervals.
2.B. Show that the Borel algebra B is also generated by the collection
of all half-open intervals (a, b] = {x e R : a < x c b}. Also show
that B is generated by the collection of all half-rays {x c- R : x > a),
ac- R.
2.C. Let (An) be a sequence of subsets of a set X. Let Efl = 0 and
for n c- N, let
n
En =k=1
U A k, F n = A n 1 .En - 1
 Measurable runctlons 1)

Show that (En) is a monotone increasing sequence of sets and that (Fn)
is a disjoint sequence of sets (that is, Fn n Fm if n m) such that
CO

U En= n=1
CO CO

Fn= Y An-
n=1 n n=1
2.D. Let (An) be a sequence of subsets of a set X. If A consists of
all x e X which belong to infinitely many of the sets An, show that
[Go I GO

A= m=1 n=m
The set A is often called the limit superior of the sets (An) and denoted
by lim sup An An A.

2.E. Let (An) be a sequence of subsets of a set X. If B consists of

all x e X which belong to all but a finite number of the sets An, show
that
B=m=u1 [nOm']'
The set B is often called the limit inferior of the sets (An) and denoted
by lim inf An-
2-F. If (En) is a sequence of subsets of a set X which is monotone
increasing (that is, .E1 9 E2 g E3 c ), show that
CO

lim sup En = 1J En = lim inf En.

n=1

2.G. If (Fn) is a sequence of subsets of a set X which is monotone

decreasing (that is, F1 F2 F3 2 .), show that
CO

lim sup Fn= nFn = lim inf Fn.

n=1

2.H. If (An) is a sequence of subsets of X, show that

0 C lim inf An C lim sup An C X.
Give an example of a sequence (An) such that
lim inf An = Q, lim sup An = X.
Give an example of a sequence (An) which is neither monotone
increasing or decreasing, but is such that
liminfAn=limsupAn.
16 The Elements of Integration

when this equality holds, the common value is called the limit of (An)
and is denoted by lim A.
2.1. Give an example of a function f on X to R which is not X-
measurable, but is such that the functions I f I and f2 are X-measurable.
2.J. If a, b, c are real numbers, let mid (a, b, c) denote the "value in
the middle." Show that
mid (a, b, c} = inf sup b), sup c), sup {b, c}}.
If f, , f2, f3 are X-measurable functions on X to R and if g is defined
for xeXby
g(x) = mid (fi(x), f2(x), f3 (x)) ,
then g is X-measurable.
2.K. Show directly (without using the preceding exercise) that if f
is measurable and A > 0, then the truncation fA defined by
fA(x) =1 (x), if I f(x) I c A,
-A, if f(x)>A,
- A, if f(x) < -A
is measurable.
2.L. Let f be a nonnegative X-measurable function on X which is
bounded (that is, there exists a constant K such that 0 c f (x) c K for
all x in X). Show that the sequence (pn) constructed in Lemma 2.11
converges uniformly on X to f.
2. M. Let f be a function defined on a set X with values in a set Y.
If E is any subset of Y, let
f -'(E) = {x e X : f (x) e E} .
Show that f '(0) = 0, f -' (Y) = X. If E and F are subsets of Y,
then
f -'(.E 1 F) = f -'(.E) 1 f -'(F).
If {E } is any nonempty collection of subsets of Y, then

f - ' ua E = U fa -' (EE), .f -' f

a
1 E = nf-,(E,.).
In particular it follows that if Y is a Q-algebra of subsets of Y, then
If ' (E) :.E e Y} is a Q-algebra of subsets of X.
 Measurable Functions 17

2.N. Let f be a function defined on a set X with values in a set Y.

Let X be a a-algebra of subsets of X and let Y = {E 9 Y : f -'(.E) E X}.
Show that Y is a a-algebra.
2.0. Let (X, X) be a measurable space and f be defined on X to Y.
Let A be a collection of subsets of Y such that f -'(E) e X for every
E e A. Show that f '(F) e X for any set F which belongs to the
a-algebra generated by A. (Hint: Use the preceding exercise.)
2.P. Let (X, X) be a measurable space and f be a real-valued function
defined on X. Show that f is X-measurable if and only if f -'(E) e X
for every Borel set E.
2.Q. Let (X, X) be a measurable space, f be an X-measurable function
on X to R and let 9) be a continuous function on R to R. Show that
the composition 9) of, defined by (g) of)(x) = gp[ f(x)] , is X-measurable.
(Hint: If 9) is continuous, then 9) -' (E) e B for each E e B.)
2.R. Let f be as in the preceding exercise and let be a Borel
measurable function. Show that of is X-measurable.
2.S. Let f be a complex-valued function defined on a measurable
space (X, X). Show that f is X-measurable if and only if

Ix eX:a< Ref(x) <b, c<Imf(x)<d}

belongs to X for all real numbers a, b, c, d. More generally, f is
X-measurable if and only if f -'(G) E X for every open set G in the
complex plane C.
2.T. Show that sums, products, and limits of complex-valued
measurable functions are measurable.
2.U. Show that a function f on X to R (or to R) is X-measurable if
and only if the set Aa in Lemma 2.4(a) belongs to X for each rational
number a; or, if and only if the set Ba in Lemma 2.4(b) belongs to X
for each rational number a; etc.
2.V. A nonempty collection M of subsets of a set X is called a
monotone class if, for each monotone increasing sequence (E,,) in M
and each monotone decreasing sequence (F7) in M, the sets

u,
n=1
En, flFn
n=1
18 The Elements of Integration

belong to M. Show that a Q-algebra is a monotone class. Also, if

A is a nonempty collection of subsets of X, then there is a smallest
monotone class containing A. (This smallest monotone class is called
the monotone class generated by A.)
2.W. If A is a nonempty collection of subsets of X, then the Q-algebra
S generated by A contains the monotone class M generated by A .
Show that the inclusion A 9 M c S may be proper.