Professional Documents
Culture Documents
(X, o, ) 86
6.4. L
2
(X, o, ) 87
6.5. L
2
-convergence of Fourier series 93
Appendix A. Extended real numbers 97
Appendix B. Axiom of choice 101
Appendix C. Continuum hypothesis 103
Appendix. References 105
Appendix. Index 107
Chapter 1
Classes of sets
1.1. Semi-algebra and algebra of sets
Concepts and examples:
1.1.1. Denition:
Let X be a nonempty set and let c be a collection of subsets of X. We say
c is a semi-algebra of subsets of X if it has the following properties:
(i) , X c.
(ii) A B c for every A, B c.
(iii) For every A c there exist n N and sets C
1
, C
2
, . . . , C
n
c such
that C
i
C
j
= for i ,= j and A
c
=
n
i=1
C
i
.
1.1.2. Denition:
Let X be a nonempty set and T a collection of subsets of X. The collection
T is called an algebra of subsets of X if T has the following properties:
(i) , X T.
(ii) A B T, whenever A, B T.
(iii) A
c
T, whenever A T.
1.1.3. Examples:
(i) Let X be any nonempty set. The collections , X and T(X) := E [ E
X are trivial examples of algebras of subsets of X. The collection T(X)
is called the power set of X.
1
2 1. Classes of sets
(ii) The collection 1 of all intervals forms a semi-algebra of subsets of R. For
a, b R with a < b, consider the collection
1 of all intervals of the form
(a, b], (, b], (a, ), (, +). We call
1 the collection of all left-open,
right-closed intervals of R. It is easy to check that
1 is also a semi-algebra
of subsets of R.
(iii) The collection
T(1) :=
_
E R E =
n
_
k=1
I
k
, I
k
1, I
k
I
= for k ,= , n N
_
.
is an algebra of subsets of R. So is the class
T(
1) :=
_
E R[ E =
n
_
k=1
I
k
, I
k
1, I
k
I
= for k ,= , n N
_
.
(iv) Let X be any nonempty set. Let
c := E X[ either E or E
c
is nite.
Then c is an algebra of subsets of X.
(v) Let X and Y be two nonempty sets, and T and ( semi-algebras of subsets
of X and Y , respectively. Let
T( = FG [ F T, G (.
Then, T( is a semi-algebra of subsets of XY .
Exercises:
(1.1) Let T be any collection of subsets of a set X. Show that T is an
algebra if and only if the following hold:
(i) , X T.
(ii) A
c
T whenever A T.
(iii) A B T whenever A, B T.
(1.2) Let T be an algebra of subsets of X. Show that
(i) If A, B T then AB := (A B) (B A) T.
(ii) If E
1
, E
2
, . . . , E
n
T then F
1
, F
2
, . . . , F
n
T such that
F
i
E
i
for each i, F
i
F
j
= for i ,= j and
n
i=1
E
i
=
n
j=1
F
j
.
The next set of exercise describes some methods of constructing algebras
and semi-algebras.
1.1. Semi-algebra and algebra of sets 3
(1.3) Let X be a nonempty set. Let , = E X and let c be a semi-
algebra (algebra) of subsets of X. Let
c E := A E [ A c.
Show that c E is a semi-algebra (algebra) of subsets of E. Note
that c E is the collection of those subsets of E which are elements
of c when E c.
(1.4) Let X, Y be two nonempty sets and f : X Y be any map. For
E Y, we write f
1
(E) := x X[ f(x) E. Let c be any
semi-algebra (algebra) of subsets of Y. Show that
f
1
(c) := f
1
(E) [ E c
is a semi-algebra (algebra) of subsets of X.
(1.5) Give examples of two nonempty sets X, Y and algebras T, G of
subsets of X and Y, respectively such that T ( := A B[ A
T, B ( is not an algebra. (It will of course be a semi-algebra,
as shown in example 1.1.3(v).)
(1.6) Let T
I
be a family of algebras of subsets of a set X. Let
T :=
I
T
n1
be a sequence of algebras of subsets of a set X. Under
what conditions on T
n
can you conclude that T :=
n=1
T
n
is also
an algebra?
(1.8) Let c be a semi-algebra of subsets of a set X. A set A X is called
a -set if there exist sets C
i
c, i = 1, 2, . . . , such that C
i
C
j
=
for i ,= j and
i=1
C
i
= A. Prove the following:
(i) For any nite number of sets C, C
1
, C
2
, . . . , C
n
in c, C(
n
i=1
C
i
)
is a nite union of pairwise disjoint sets from c and hence is a
-set.
(ii) For any sequence C
n
n1
of sets in c,
n=1
C
n
is a -set.
(iii) A nite intersection and a countable union of -sets is a -set.
(1.9) Let c be any collection of subsets of a set X. Then there exists a
unique algebra T of subsets of X such that c T and if / is any
other algebra such that c /, then T /. This unique algebra
given is called the algebra generated by c and is denoted by
T(c).
(1.10) Show that he algebra generated by 1, the class of all intervals, is
E R [ E =
n
k=1
I
k
, I
k
I, I
k
I
= if 1 k ,= n.
(1.11) Let c be any semi-algebra of subsets of a set X. Show that T(c), the
algebra generated by c, is given by E X [ E =
n
i=1
C
i
, C
i
c
and C
i
C
j
= for i ,= j, n N.
4 1. Classes of sets
Remark:
Exercise 1.11 gives a description of T(c), the algebra generated by
a semi-algebra c. In general, no description is possible for T(c)
when c is not a semi-algebra.
(1.12) Let X be any nonempty set and c = x [ x X
, X. Is c
a semi-algebra of subsets of X? What is the algebra generated by
c? Does your answer depend upon whether X is nite or not?
(1.13) Let Y be any nonempty set and let X be the set of all sequences
with elements from Y, i.e.,
X = x = x
n
n1
[ x
n
Y, n = 1, 2, . . ..
For any positive integer k let A Y
k
, the k-fold Cartesian product
of Y with itself, and let i
1
< i
2
< < i
k
be positive integers. Let
C(i
1
, i
2
, . . . , i
k
; A) := x = (x
n
)
n1
X [ (x
i
1
, . . . , x
i
k
) A.
We call C(i
1
, i
2
, . . . , i
k
; A) a k-dimensional cylinder set in X
with base A. Prove the following assertions:
(a) Every k-dimensional cylinder can be regarded as a n-dimensional
cylinder also for n k.
(b) Let
/ = E X [ E is an n-dimensional cylinder set for some n.
Then, / , X is an algebra of subsets of X.
(1.14) Let c be any collection of subsets of a set X and let E X. Let
c E := C E [ C c.
Then the following hold:
(a)
c E T(c) E := A E [ A T(c).
Deduce that
T(c E) T(c) E.
(b) Let
/ = A X [ A E T(c E).
Then, / is an algebra of subsets of X, c / and
/ E = T(c E).
(c) Using (a) and (b), deduce that T(c) E = T(c E).
1.2. Sigma algebra and monotone class 5
1.2. Sigma algebra and monotone class
Concepts and examples:
1.2.1. Denition:
Let X be any nonempty set and let o be a class of subsets of X with the
following properties:
(i) and X o.
(ii) A
c
o whenever A o.
(iii)
i=1
A
i
o whenever A
i
o, i = 1, 2, . . . .
Such a class o is called a sigma algebra (written as -algebra) of subsets
of X.
1.2.2. Examples:
(i) Let X be any set. Then , X and T(X) are obvious examples of -
algebras of subsets of X.
(ii) Let X be any uncountable set and let
o = A X [ A orA
c
is countable.
Then o is a -algebra of subsets of X.
(iii) Let X be any set and let c be any class of subsets of X. Let o(c) :=
o,
where the intersection is taken over all -algebras o of subsets of X such
that S c (note that T(X) is one such -algebra). It is easy to see that
o(c) is also a -algebra of subsets of X and o(c) c. In fact, if o is any
-algebra of subsets of X such that o c, then clearly o o(c). Thus
o(c) is the smallest -algebra of subsets of X containing c, and is called
the -algebra generated by c. In general it is not possible to represent
an element of o(c) explicitly in terms of elements of c.
1.2.3. Denition:
Let X be a nonempty set and / be a class of subsets of X. We say / is a
monotone class if
(i)
n=1
A
n
/, whenever A
n
/ and A
n
A
n+1
for n = 1, 2, . . . ,
(ii)
n=1
A
n
/, whenever A
n
/ and A
n
A
n+1
for n = 1, 2, . . . .
1.2.4. Examples:
(i) Clearly, every -algebra is also a monotone class.
6 1. Classes of sets
(ii) Let X be any uncountable set. Let / := A X [ A is countable.
Then / is a monotone class but not a -algebra.
(iii) Let X be any nonempty set and let c be any collection of subsets of X.
Clearly T(X) is a monotone class of subsets of X such that c T(X).
Let /(c) :=
1),
1 o(1) and hence deduce that
o(1) = o(
1) = B
R
.
(1.21) Prove the following statements:
(i) Let 1
r
denote the class of all open intervals of R with rational
endpoints. Show that o(1
r
) = B
R
.
(ii) Let 1
d
denote the class of all subintervals of [0, 1] with dyadic
endpoints (i.e., points of the form m/2
n
for some integers m and
n). Show that o(1
d
) = B
R
[0, 1].
(1.22) Let c be any class of subsets of X. Prove the following:
(i) If c is an algebra which is also a monotone class, show that c is
a -algebra.
(ii) c /(c) o(c).
(1.23) (-algebra monotone class theorem ) Let / be an algebra of
subsets of a set X. Then , o(/) = /(/).
Prove the above statement by proving the following:
(i) /(/) o(/).
(ii) Show that /(/) is closed under complements by proving that
for
B := E X [ E
c
/(/),
/ B, and B is a monotone class. Hence deuce that /(/)
B.
(iii) For F /(/), let
/(F) := A X [ A F /(/).
Show that E /(F) i F /(E), /(F) is a monotone class,
and / /(F) whenever F /. Hence, /(/) /(F), for
F /.
8 1. Classes of sets
(iv) Using (iii), deduce that /(/) /(E) for every E /(/),i.e.,
/(/) is closed under unions also. Now use exercise (1.22) to
deduce that o(/) /(/).
Chapter 2
Measure
2.1. Set functions
Concepts and examples:
2.1.1. Denition:
Let c be a class of subsets of a set X. A function : c [0, +] is called
a set function. Further,
(i) is said to be monotone if (A) (B) whenever A, B c and
A B.
(ii) is said to be nitely additive if
_
n
_
i=1
A
i
_
=
n
i=1
(A
i
).
whenever A
1
, A
2
, . . . , A
n
c are such that A
i
A
j
= for i ,= j
and
n
i=1
A
i
c.
(iii) is said to be countably additive if
_
_
n=1
A
n
_
=
n=1
(A
n
)
whenever A
1
, A
2
, . . . in c with A
i
A
j
= for i ,= j and
n=1
A
n
c.
9
10 2. Measure
(iv) is said to be countably subadditive if
(A)
n=1
(A
n
).
whenever A c, A =
n=1
A
n
with A
n
c for every n.
(v) is called a measure on c if c with () = 0 and is countably
additive on c.
Here are some more examples of nitely/countably additive set func-
tions:
2.1.2.Example:
Let X be any innite set and let x
n
X, n = 1, 2, . . . . Let p
n
n1
be a
sequence of nonnegative real numbers. For any A X, dene
(A) :=
{i|x
i
A}
p
i
.
It is easy to show that is a countably additive set function on the algebra
T(X). We say is a discrete measure with mass p
i
at x
i
. The measure
is nite (i.e., (X) < +) i
i=1
p
i
< +. If
i=1
p
i
= 1, the measure
is called a discrete probability measure/distribution. Note that
(x
i
) = p
i
i and (x) = 0 if x ,= x
i
. So, one can regard as a set
function dened on the subsets of the set Y := x
n
: n 1. Some of the
special cases when X = 0, 1, 2, . . . are:
(a) Binomial distribution: Y := 0, 1, 2, . . . , n and, for 0 < p < 1,
p
k
=
_
n
k
_
p
k
(1 p)
nk
, 0 k n.
(b) Poisson distribution: Y := 0, 1, 2, . . . and
p
k
:=
k
e
/k!
for k = 0, 1, 2, . . . , where > 0.
(c) Uniform distribution: Y := 1, 2, . . . , n,
p
k
:= 1/k k.
An important example of a set function on the collection of intervals in
R.
2.1.3 Example:
We denote the set of real numbers by R. Let R
[x 0 = [0, +) +.
Dene the function : 1 [0, ] by
(I(a, b)) :=
_
[b a[ if a, b R,
+ if either a = or b = + or both.
The function , as dened above, is called the length function and has
the following properties:
Property (1): () = 0.
Property (2): (I) (J) if I J.
This is called the monotonicity property of (or one says that is
monotone).
Property (3): Let I 1 be such that I =
n
i=1
J
i
, where J
i
J
j
= for
i ,= j. Then
(I) =
n
i=1
(J
i
).
This property of is called the nite additivity of , or one says that
is nitely additive.
Property (4): Let I 1 be a nite interval such that I
i=1
I
i
, where
I
i
1. Then
(I)
i=1
(I
i
).
Property (5): Let I 1 be a nite interval such that I =
n=1
I
n
, where
I
n
I and I
n
I
m
= for n ,= m. Then
(I) =
n=1
(I
n
).
Property (6): Let I 1 be any interval. Then
(I) =
n=
(I [n, n + 1)).
Property (7): Let I 1 be any interval such that I =
n=1
I
n
, I
n
1
and I
n
I
m
= for n ,= m. Then
(I) =
n=1
(I
n
).
12 2. Measure
This property of is called the countable additivity of , or one says
that is countably additive.
Property (8): Let I 1 and I
n=1
I
n
, I
n
1. Then
(I)
n=1
(I
n
).
This property of is called the countable subadditivity of , or one
says that is countably subadditive.
Property (9): (I) = (I + x), for every I 1 and x R, where
I +x := y +x [ y I.
This property of the length function is called translation invariance, or
one says that is translation invariant.
Exercises:
(2.1) Let X be any countably innite set and let
c = x [ x X.
Show that the algebra generated by c is
T(c) := A X [ A or A
c
is nite.
Let : T(c) [0, ) be dened by
(A) :=
_
0 if A is nite,
1 if A
c
is nite.
Show that is nitely additive but not countably additive. If X is
an uncountable set, show that is also countably additive.
(2.2) Let X = N, the set of natural numbers. For every nite set A X,
let #A denote the number of elements in A. Dene for A X,
n
(A) :=
#m : 1 m n, m A
n
.
Show that
n
is countably additive for every n on T(X). In a sense,
n
is the proportion of integers between 1 to n which are in A. Let
c = A X [ lim
n
n
(A) exists.
Show that c is closed under taking complements, nite disjoint
unions and proper dierences. Is it an algebra?
2.1. Set functions 13
(2.3) Let :
1 (0, 1] [0, ] be dened by
(a, b] :=
_
b a if a ,= 0, 0 < a < b 1,
+ otherwise.
(Recall that
1 (0, 1] is the class of all left-open right-closed inter-
vals in (0, 1].)
Show that is nitely additive. Is countably additive also?
(2.4) Let X be a nonempty set.
(a) Let : T(X) [0, ) be a nitely additive set function
such that (A) = 0 or 1 for every A T(X). Let
| = A T(X) [ (A) = 1.
Show that | has the following properties:
(i) , |.
(ii) If A X and B A, then B |.
(iii) If A, B |, then A B |.
(iv) For every A T(X), either A | or A
c
|.
(Any | T(X) satisfying (i) to (iv) is called an ultralter
in X.)
(b) Let | be any ultralter in X. Dene : T(X) [0, ) by
(A) :=
_
1 if A |,
0 if A , |.
Show that is nitely additive.
(2.5) Let / be an algebra of subsets of a set X.
(i) Let
1
,
2
be measures on /, and let and be nonnegative
real numbers. Show that
1
+
2
is also a measure on /.
(ii) For any two measures
1
,
2
on /, we say
1
2
if
1
(E)
2
(E), E /.
Let
n
n1
be a sequence of measures on / such that
n
n+1
, n 1.
Dene E /,
(E) := lim
n
n
(E).
Show that is also a measure on / and E B,
(E) = sup
n
(E) [ n 1.
(2.6) Let X be a compact topological space and / be the collection of
all those subsets of X which are both open and closed. Show that
/ is an algebra of subsets of X. Further, every nitely additive set
function on / is also countably additive.
14 2. Measure
2.2. Countably additive set functions on
intervals
Concepts and examples:
We saw in the previous section that the length function is a countably ad-
ditive set function on the class of all intervals. One can ask the question:
do there exist countably additive set functions on intervals, other than the
length function? The answer is given by the following:
2.2.1. Proposition:
Let F : R R be a monotonically increasing function. Let
F
:
1
[0, ] be dened by
F
(a, b ] := F(b) F(a),
F
(, b ] := lim
x
[F(b) F(x)],
F
(a, ) := lim
x
[F(x) F(a)],
F
(, ) := lim
x
[F(x) F(x)].
Then,
F
is a well-dened nitely additive set function on
1. Further,
F
is countably additive if F is right continuous.
One calls
F
the set function induced by F.
The converse of proposition 2.2.1 is also true.
2.2.2. Proposition:
Let :
1 [0, ] be a nitely additive set function such that (a, b] < +
for every a, b R. Then there exists a monotonically increasing function
F : R R such that (a, b] = F(b) F(a) a, b R. If is also
countably additive, then F is right-continuous.
(Hint: Dene F as follows:
F(x) :=
_
_
_
(0, x] if x > 0,
0 if x = 0,
(x, 0] if x < 0.)
2.2.3. Remarks:
(i) In case (R) < +, a more canonical choice for the required function F
in proposition 2.2.2 is given by F(x) := (, x], x R.
2.3. Set functions on algebras 15
(ii) Propositions 2.2.1 and 2.2.2 completely characterize the non-trivial count-
ably additive set functions on intervals in terms of functions F : R R
which are monotonically increasing and right continuous. Such functions
are called distribution functions on R. The set function
F
induced by
the distribution function F is non-trivial in the sense that it assigns nite
non-zero values to bounded intervals.
Exercises:
(2.7) Let F(x) = [x], the integral part of x, x R. Describe the set
function
F
.
(2.8) R. Show that F
1
:= F + is also a distribution function and
F
=
F
1
. Is the converse true?
(2.9) (i) Let c be a collection of subsets of a set X and : c [0, ]
be a set function. If is a measure on c, show that is nitely
additive. Is monotone? Countably subadditive?
(ii) If c be a semi-algebra, then is countably subadditive i A
c with A
i=1
A
i
, A
i
c implies
(A)
i=1
(A
i
).
2.3. Set functions on algebras
Concepts and examples:
In this section, we give some general properties of a set function dened
on an algebra / of subsets of an arbitrary set X.
2.3.1. Theorem:
Let / be an algebra of subsets of a set X and let : / [0, ] be a set
function. Then the following hold:
(i) If is nitely additive and (B) < +, then (B A) = (B) (A)
for every A, B / with A B. In particular, () = 0 if is nitely
additive and (B) < + for some B /.
(ii) If is nitely additive, then is also monotone.
(iii) Let () = 0. Then is countably additive i is both nitely additive
and countably subadditive.
16 2. Measure
Another characterization of countable additivity of set functions dened
on algebras is given in the next theorem.
2.3.2. Theorem:
Let / be an algebra of subsets of a set X and let : / [0, ] be such
that () = 0.
(a) If is countably additive then the following hold:
(i) For any A /, if A =
n=1
A
n
, where A
n
/ and A
n
A
n+1
n, then
(A) = lim
n
(A
n
).
This is called the continuity from below of at A.
(ii) For any A /, if A =
n=1
A
n
, where A
n
/ with A
n
A
n+1
n and (A
n
) < + for some n, then
lim
n
(A
n
) = (A).
This is called the continuity from above of at A.
Conversely,
(b) If is nitely additive and (i) holds, then is countably additive.
(c) If (X) < +, is nitely additive and (ii) holds, then is count-
ably additive.
Exercises:
(2.10 ) (i) In the proofs of part (ii) and part (c) of theorem 2.3.2, where do
you think we used the hypothesis that (X) < +? Do you think
this condition is necessary?
(ii) Let / be an algebra of subsets of a set X and : / [0, ] be
a nitely additive set function such that (X) < +. Show that
the following statements are equivalent:
(a) lim
k
(A
k
) = 0, whenever A
k
k1
is a sequence in / with
A
k
A
k+1
k, and
k=1
A
k
= .
(b) is countably additive.
(2.11 ) Extend the claim of theorem 2.3.2 when / is only a semi-algebra
of subsets of X.
(Hint: Use exercise 1.8)
(2.12 ) Let / be a -algebra and : / [0, ] be a measure. For any
sequence E
n
n1
in /, show that
2.4. Uniqueness problem for measures 17
(i) (liminf
n
E
n
) liminf
n
(E
n
).
(ii) (limsup
n
E
n
) limsup
n
(E
n
).
(Hint: For a sequence E
n
n1
of subsets of a set X,
liminf
n
E
n
:=
_
n=1
k=n
E
k
limsup
n
E
n
:=
n=1
_
k=n
E
k
. )
2.4. Uniqueness problem for measures
Concepts and examples:
The problem that we want to analyze is the following: Let be a -nite
measure on an algebra / of subsets of X. Let
1
and
2
be two measures
on o(/), the -algebra generated by the algebra /, such that
1
(A) =
2
(A) A /. Is
1
=
2
?
2.4.1. Denition:
Let c be a collection of subsets of X and let : c [0, ] be a set function.
We say is totally nite (or just nite) if (A) < + A c. The
set function is said to be sigma nite (written as -nite) if there exist
pairwise disjoint sets X
n
c, n = 1, 2, . . . , such that (X
n
) < + for every
n and X =
n=1
X
n
.
2.4.2. Examples:
(i) The length function on the class of intervals is -nite.
(ii) Let / = /(
(A) :=
_
+ if A , = ,
0 if either A = or A = .
It is easy to check that
is also a measure on /
c
is a measure on X o(/) = o(X /) and is not -nite.
18 2. Measure
2.4.3. Proposition:
Let
1
and
2
be totally-nite measures on a -algebra o. Then the class
/ = E o [
1
(E) =
2
(E) has the following properties:
(i) / is a monotone class.
(ii) If o = o(/), and
1
(A) =
2
(A) A /, then
1
(A) =
2
(A) A
o(/).
Exercises:
(2.13 ) Let / be an algebra of subsets of a set X. Let
1
and
2
be -nite
measures on a -algebra o(/) such that
1
(A) =
2
(A) A /.
Then,
1
(A) =
2
(A) A o(/).
(2.14) Show that a measure dened on an algebra / of subsets of a set
X is nite if and only if (X) < +.
Chapter 3
Construction of
measures
Concepts and examples:
3.1. Extension from semi-algebra to the
generated algebra
3.1.1. Denition:
Let c
i
, i = 1, 2 be classes of subsets of a set X, with c
1
c
2
. Let function
1
: c
1
[0, +] and
2
: c
2
[0, +] be set functions. The set
function
2
is called an extension of
1
if
1
(E) =
2
(E) for every E
1
.
3.1.2. Examples:
In example 2.4.2(ii), each
n
i=1
E
i
for pairwise disjoint sets E
1
, . . . , E
n
c, dene
(E) :=
i=1
(E
i
).]
Exercises:
(3.1) Let c be a collection of subsets of a set X and : c [0, ] be a
set function. If is a measure on c, show that is nitely additive.
Is monotone? Countably subadditive?
(3.2) If c be a semi-algebra, then is countably subadditive i A c
with A
i=1
A
i
, A
i
c implies
(A)
i=1
(A
i
).
(3.3) Using theorem 3.1.3, show that length function, which is initially
dened on the semi-algebra 1 of all intervals, can be uniquely ex-
tended to a set function on T(1), the algebra generated. It is
worth mentioning a result due to S.M. Ulam (1930) which, under
the assumption of the continuum hypothesis, implies that it is
not possible to extend the length function to all subsets of R.
Theorem (Ulam): Let be a measure dened on all subsets of
R such that ((n, n + 1]) < n Z and (x) = 0 for every
x R. Then (E) = 0 for every E R.
3.1.3. Remark:
Ulams theorem shows the impossibility of extending the length function
from intervals to all subsets of R, assuming the continuum hypothesis. We
shall see later that similar results can be proved if one assumes the axiom
of choice. Ulams result uses the property of that (x) = 0 x R,
and the fact that ([n, n + 1]) < + for every n Z. In the later results
we shall use the translation invariance property of the length function .
3.2. Extension from algebra to the generated
-algebra
Concepts and Examples:
Given an arbitrary measure on an algebra / of subsets of a set X, our
aim is to try to extend to a class of subsets of X which is larger than
3.2. Extension from algebra to the generated -algebra 21
/. Intuitively, sets A in / are those whose size (A) can be measured
accurately. The approximate size of any set E X is given by the outer
measure as dened next. Recall that, for any nonempty set A [0, +],
we write inf(A) := inf A [0, +) if A [0, +) ,= , and inf(A) := +
otherwise.
3.2.1. Denition:
Let / be an algebra of subsets of a set X and : / [0, ] be a measure
on /. For E X, dene
(E) := inf
_
i=1
(A
i
) A
i
/,
_
i=1
A
i
E
_
.
The set function
() = 0 and
(A) 0 A X.
(ii)
is monotone, i.e.,
(A)
(B) whenever A B X.
(iii)
(A)
i=1
(A
i
) whenever A =
_
i=1
A
i
.
(iv)
is an extension of , i.e.,
(A) = (A) if A /.
3.2.3. Remarks:
(i) A set function dened on all subsets of a set X is called an outer
measure if has properties (i), (ii) and (iii) in proposition 3.2.2. The outer
measure
(A) (A).
In other words,
n
i=1
(I
i
) 1. This will imply
(A) = (A) = 0.
Further,
(E) = inf
_
i=1
(A
i
) A
i
/, A
i
A
j
= for i ,= j and
_
i=1
A
i
E
_
.
(3.7) Let X be any nonempty set and let / be any algebra of subsets of
X. Let x
0
X be xed. For A /, dene
(A) :=
_
0 if x
0
, A,
1 if x
0
A.
Show that is countably additive. Let
(A) = 1 if x
0
A. Can you conclude that
(A) = 1 implies
x
0
A? Show that this is possible if x
0
/.
3.3. Choosing nice sets: Measurable sets
Concepts and examples:
In the previous section we dened the notion of
(Y ) =
(Y E) +
(Y E
c
). (3.1)
We denote by o
i E
c
o
(Y E) and
(Y E
c
) add up to give the size
.
(ii) For every Y X,
(Y )
(Y E) +
(Y E
c
).
(iii) For every Y X, with
(Y ) < +,
(Y )
(Y E) +
(Y E
c
).
(iv) For every A /,
(A)
(A E) +
(A E
c
).
We give an equivalent denition of measurable sets when (X) < +.
3.3.3. Theorem:
Let (X) < +. Then E X is
-measurable i
(X) = (E) +(E
c
)
Next, we check that o
.
(ii) o
, and
restricted to
o
is nitely additive.
24 3. Construction of measures
(iii) If A
n
o
, n = 1, 2, . . . , then
n=1
A
n
o
and
restricted to
o
is countably additive.
(iv) Let ^ := E X [
(E) = 0. Then ^ o
.
This together with proposition 2.4.3 gives us the following:
3.3.5. Theorem:
Let be a measure on an algebra / of subsets of a set A. If is -nite, then
there exists a unique extension of to a measure on o(/), the -algebra
generated by /.
3.3.6. Remark:
Theorems 3.2.2.(iv) and 3.3.4 together give us a method of constructing an
extension of a measure dened on an algebra / to a class o
o(A) /.
Exercises:
(3.8) Identify the collection of
1) as given in proposition
2.2.1. Let
F
itself denote the unique extension of
F
to /
F
, the
-algebra of
F
-measurable sets, as given by theorem 3.3.4. Show
that
(i) B
R
/
F
.
(ii)
F
(x) = F(x) lim
yx
F(y). Deduce that the function F is
continuous at x i
F
(x) = 0.
(iii) Let F be dierentiable with bounded derivative. If A R is
a null set, then
F
(A) = 0.
The measure
F
is called the Lebesgue-Stieltjes measure in-
duced by the distribution function F.
3.4. Completion of a measure space
Concepts and examples:
Theorem 3.3.4 showed that, given a -nite measure on an algebra / of
subsets of a set X, can be extended to a unique measure
on the -
algebra o
of
i=1
A
i
, A
i
/.
3.4.1. Proposition:
For every set E X,
(E) = inf
(A) [ A A
, E A
= inf
(A) [ A o(/), E A
= inf
(A) [ A o
, E A.
3.4.2. Proposition:
For every E X, there exists a set F o(/) such that E F,
(E) =
(F) and
(F E) = 0.
The set F is called a measurable cover of E.
3.4.3. Corollary:
Let E X. Then there exists a set K E, K o(/), such that
(A) = 0
for every set A E K.
The set K is called a measurable kernel of E.
3.4.4. Denition:
Let X be a nonempty set, o a -algebra of subsets of X and a measure on
o. The pair (X, o) is called a measurable space and the triple (X, o, ) is
called a measure space. Elements of o are normally called measurable
sets.
Till now what we have done is that, given a measure on an algebra / of
subsets of a set X, we have constructed the measure spaces (X, o(/),
),
(X, o
(E) = 0, then E o
. This
property is called the completeness of the measure space (X, o
).
The measure space (X, o(/),
(E) = 0 by
o
= o(/) ^ := E N [ E o(/), N ^.
One calls (X, o
)
and (X, o
on T(X) and on /.
3.4.7. Proposition:
Let be a measure on an algebra / of subsets of a set X and let
be the
induced outer measure. Let E o
be such that
/ such that
(E F
) < .
3.4.8. Note:
Whenever (X, o, ) is a nite measure space with (X) = 1, it is called a
probability space and the measure is called a probability. The reason
for this terminology is that the triple (X, o, ) plays a fundamental role
in the axiomatic theory of probability. It gives a mathematical model for
analyzing statistical experiments. The set X represents the set of all possible
outcomes of the experiment, the -algebra o represents the collection of
events of interest in that experiment, and for every E o, the nonnegative
number (E) is the probability that the event E occurs. For more details
see Kolmogorov [9] and Parthasarathy [10].
Exercises:
(3.11 ) Let E X, and let G
1
, G
2
be two measurable covers of E. Show
that
(G
1
G
2
) = 0.
(3.12) Let E
1
E
2
E
3
. . . be subsets of X. Then
_
_
n=1
E
n
_
= lim
n
(E
n
).
(3.13 ) Let E X, and let K
1
, K
2
be two measurable kernels of E. Show
that
(K
1
K
2
) = 0.
3.5. The Lebesgue measure 27
(3.14 ) Let ^ := E X [
= o(/) ^ := E N [ E o(/), N ^,
where o
is the -algebra of
(A) =
(E) := inf
_
i=1
(I
i
) I
i
1 i, I
i
I
j
= for i ,= j and E
_
i=1
I
i
_
.
The -algebra of
to / or
B
R
by itself. The measure space (R, /, ) is called the Lebesgue mea-
sure space and is called the Lebesgue measure. We note that since
on 1 is -nite (e.g., R =
+
n=
(n, n + 1]), the extension of to B
R
is
unique. It is natural to ask the question:
What is the relation between the classes B
R
, / and T(R)?
As a special case of theorem 3.4.6, we have / = B
R
^, where
^ := N R [ N E B
R
, (E) = 0.
Thus,
B
R
/ T(R).
The question arises:
Is B
R
a proper subset of /?
That is, are there sets in / which are not Borel sets?. First of all, we note
that Cantors ternary set C ^ /. Further, E C, then
(E) = 0
and hence E /. In other words, T(C) /. Thus, the cardinality of /
28 3. Construction of measures
is at least 2
c
(here c denotes the cardinality of the real line, also called the
cardinality of the continuum). Since / T(R), we get the cardinality
of / to be 2
c
. On the other hand, B
R
is the -algebra generated by all
open intervals of R with rational endpoints. One can show that the -
algebra B
R
of Borel subsets of R has cardinality c, that of the continuum.
Thus, there exist sets which are Lebesgue measurable but are not Borel sets.
The actual construction of such sets is not easy. One such class of sets is
called analytic sets. An analytic set is a set which can be represented as a
continuous image of a Borel set. For a detailed discussion on analytic sets,
see Srivastava [38], Parthasarathy [29].
Since /, the class of all Lebesgue measurable subsets of R, has 2
c
ele-
ments, i.e., same as that of T(R), the natural question arises:
Is / = T(R)?
We stated earlier that, if we assume the continuum hypothesis, it is not
possible to dene a countably additive set function on T(R) such that
(x) = 0 x R. In particular, if we assume the continuum hypothesis,
we cannot extend to all subsets of R. Hence / , = T(R). What can be said
if one does not assume the continuum hypothesis? To answer this question,
one can either try to construct a set E R such that E , / ,or, assuming
that such a set exists, try to see whether one can reach a contradiction.
G. Vitali (1905), F. Bernstein (1908), H. Rademacher (1916) and others
constructed such sets assuming the axiom of choice (see appendix B). The
example of Vitali used the translation invariance property of the Lebesgue
measure, and that of Bernstein used the regularity properties of the Lebesgue
measure. Rademacher proved that every set of positive outer Lebesgue
measure includes a Lebesgue nonmeasurable set. Even today, more and
more nonmeasurable sets with additional properties are being constructed.
For example, one can construct nonmeasurable subsets A of R such that
(AI) =
which
is a nite disjoint union of open intervals and is such that
(E F
) < .
We give next some more characterizations of Lebesgue measurable sets.
3.5.2. Theorem:
For any set E R the following statements are equivalent:
(i) E /, i.e., E is Lebesgue measurable.
(ii) For every > 0, there exists an open set G
such that
E G
and
(G
E) < .
(iii) For every > 0, there exists a closed set F
such that
F
E and
(E F
) < .
(iv) There exists a G
(G E) = 0.
(v) There exists an F
(E F) = 0.
[Hint: Prove the following implications:
(i) = (ii) = (iv) = (i)
and
(i) = (iii) = (v) = (i).]
3.5.3. Note:
Theorem 3.5.2 tells us the relation between /, the class of Lebesgue mea-
surable sets, and the topologically nice sets, e.g., open sets and closed sets.
The property that for E / and > 0, there exists an open set G E
with (G E) < can be stated equivalently as:
(E) = inf(U) [ U open, U E.
30 3. Construction of measures
This is called the outer regularity of . Other examples of outer regular
measures on R (in fact any metric space) are given in the exercise 3.25.
Another topologically nice class of subsets of R is that of compact subsets
of R. It is natural to ask the question: does there exist a relation between
/ and the class of compact subsets of R? Let K be any compact subset of
R. Since K is closed (and bounded), clearly K B
R
/ and (K) < +.
It is natural to ask the question: can one obtain (E) for a set E B
R
, if
(E) < +, from the knowledge of (K), K compact in R? The answer is
given by the next proposition.
3.5.4. Proposition:
Let E / with 0 < (E) < + and let > 0 be given. Then there exists a
compact set K E such that (E K) < .
On the set R, we have the group structure given by the binary operation of
the addition of two real numbers. We analyze the behavior of on / under
the map y y + x, y R and x R xed. We saw that A + x is a
Lebesgue measurable set whenever A is Lebesgue measurable and x R. It
is natural to ask the question: for A / and x R, is (A + x) = (A)?
The answer is given by the following:
3.5.5. Theorem (Translation invariance property):
Let E /. Then E +x / for every x R, and (E +x) = (E).
We saw that Lebesgue measure is the unique extension of the length function
from the class 1 of intervals to B
R
, the -algebra of Borel subsets of R. This
gave us a measure on B
R
with the following properties:
(i) For every nonempty open set U, (U) > 0.
(ii) For every compact set K, (K) < +.
(iii) For every E B
R
,
(E) = inf(U) [ U open, U E,
= sup(C) [ C E, C closed.
If (E) < +, then we also have
(E) = sup(K) [ K E, K compact.
(iv) For every E B
R
and x R, E +x B
R
and (E +x) = (E).
Thus the Lebesgue measure is a translation invariant -nite regular
measure on B
R
. The question arises: are there other -nite measures on
B
R
with these properties? Obviously, if c > 0 then c dened by (c)(E) :=
c(E), E B
R
, is also a -nite measure and is translation invariant. In
fact the following hold:
3.5. The Lebesgue measure 31
3.5.6. Theorem:
Let be a measure on B
R
such that
(i) (U) > 0 for every nonempty open set U R.
(ii) (K) < + for every compact set K R.
(iii) (E +x) = (E), E B
R
and x R.
Then there exists a positive real number c such that (E) = c(E)
E B
R
.
3.5.7. Note:
In fact the above theorem has a far-reaching generalization to abstract topo-
logical groups. Let us recall that the set of real numbers R is a group under
the binary operation +, the addition of real numbers. Also, there is a topol-
ogy on R which respects the group structure, i.e., the maps (t, s) t +s
and t t from R R R and R R, respectively, are continuous
when R R is given the product topology. In an abstract setting, if G is a
set with a binary operation and a topology T such that (G, ) is a group
and the maps G G G, (g, h) g.h and G G, g g
1
are
continuous with respect to the product topology on G G, one calls G a
topological group. Given a topological group, let B
G
denote the -algebra
generated by open subsets of G, called the -algebra of Borel subsets of G.
The question arises: does there exist a -nite measure on G such that it
has the properties as given in theorem 3.5.5? A celebrated theorem due to A.
Haar states that such a measure exists and is unique up to a multiplicative
(positive) constant if G is locally-compact. Such a measure is called a (right)
Haar measure on G. Theorem 3.5.5 then states that for the topological
group R, the Lebesgue measure is a Haar measure. Consider the group
(R 0, ), where R 0 = t R[t ,= 0 and is the usual multiplication
of real numbers. Let R 0 be given the subspace topology from R. It is
easy to show that R 0 is a topological group and, E B
R\{0}
,
(E) :=
_
E
1
[x[
d(x) (3.2)
is a Haar measure on R 0.
3.5.8. Note:
In the previous sections we have seen how the general extension theory,
as developed earlier, can be applied to the particular situation when the
semi-algebra is that of intervals and the set function is the length function.
More generally, if we consider the semi-algebra
1 of left-open right-closed
intervals in R and consider F : R R as a monotonically increasing
right continuous function, then we can construct a countably additive set
function
F
on the semi-algebra
1, as in example 1.1.2. Using theorem
32 3. Construction of measures
3.4.6, we can construct a complete measure
F
on a -algebra of subsets
of R which includes B
R
. This measure
F
is called the Lebesgue-Stieltjes
measure induced by the function F. Note that
F
has the property that
F
(a, b] < + a, b R, a < b. Conversely, given a measure on B
R
such that (a, b] < + a, b R, a < b, we can restrict it to
1 and,
using proposition 2.2.2, dene a monotonically increasing right continuous
function F : R R such that the unique Lebesgue-Stieltjes measure
F
induced by F is nothing but (by the uniqueness of the extension). Thus
measures on B
R
which have the property that (a, b] < + a < b
can be looked upon as a Lebesgue-Stieltjes measure
F
for some F. We
point out that it is possible to nd dierent F
1
, F
2
: R R such that
both are monotonically increasing and right continuous and
F
1
=
F
2
. If
is nite measure, i.e., (R) < +, then it is easy to see that F(x) :=
(, x], x R, is a monotonically increasing right continuous function
such that =
F
. This F is called the distribution function of . When
(R) = 1, is called a probability and its distribution function F, which is
monotonically increasing and is right continuous with lim
x
[F(x) F(x)]
=
F
(R) = 1, is called a probability distribution function.
Exercises:
(3.15) Let 1
0
denote the collection of all open intervals of R. For E X,
show that
(E) = inf
_
i=1
(I
i
) I
i
1
0
i, for i ,= j and E
_
i=1
I
i
_
.
(3.16) Let E R and let > 0 be arbitrary. Show that there exists an
open set U
E such that (U
E) ?
(3.17) For E R, let
diameter(E) := sup[x y[ [ x, y E.
Show that
(E) diameter(E).
(3.18) Show that for E R,
n1
of intervals such that E
n=1
and
n=1
E) < . Such sets are called Lebesgue null sets.Prove
the following:
(i) Every singleton set x, x R, is a null set. Also every nite
set is a null set.
(ii) Any countably innite set S = x
1
, x
2
, x
3
, . . . is a null set.
(iii) Q, the set of rational numbers, is a null subset of R.
3.5. The Lebesgue measure 33
(iv) Every subset of a null set is also a null set.
(iv) Let A
1
, A
2
, . . . , A
n
, . . . be null sets. Then
n=1
A
n
is a null
set.
(v) Let E [a, b] be any set which has only a nite number of
limit points. Can E be uncountable? Can you say E is a null
set?
(vi) Let E be a null subset of R and x R. What can you say about
the sets E +x := y +x [ y E and xE := xy [ y E?
(vii) Let I be an interval having at least two distinct points. Show
that I is not a null set.
(viii) If E contains an interval of positive length, show that it is not
a null set. Is the converse true, i.e., if E R is not a null set,
then does E contain an interval of positive length?
(ix) Show that Cantors ternary set is an uncountable null set.
(3.19) Let E [0, 1] be such that
n1
be any increasing sequence of subsets (not necessarily
measurable) of R. Then,
_
_
n=1
E
n
_
= lim
n
(E
n
).
(3.22) Let E R. Show that the following statements are equivalent:
(i) E /.
(ii)
(I) =
(E I) +
(E
c
I) for every interval I.
(iii) E [n, n + 1) / for every n Z.
(iv)
(E [n, n + 1)) +
(E
c
[n, n + 1)) = 1 for every n Z.
(3.23) Let A / and x R. Using theorem 4.2.2, show that
(i) A+x /, where A+x := y +x [ y A.
(ii) A /, where A := y [ y A.
(3.24) Let (X, d) be any metric space and let be a measure on B
X
, the
-algebra generated by open subsets of X, called the -algebra of
Borel subsets of X. The measure is called outer regular if
E B
X
,
(E) = inf(U) [ U open, U E,
= sup(C) [ C closed, C E. (3.3)
(i) If (X) < +, show that is outer regular i for every E
B
X
and > 0 given, there exist an open set U
and a closed
34 3. Construction of measures
set C
such that
U
E C
and (U
) < .
(ii) For A X, let
d(x, A) := infd(x, y) [ y A.
Show that for every A X, x d(x, A) is a uniformly con-
tinuous function.
(Hint: [d(x, A) d(y, A)[ d(x, y) x, y.)
(iii) Let (X) < + and
o := E B
X
[ (3.3) holds for E.
Show that o is a -algebra of subsets of X.
(iv) Let C be any closed set in X. Show that C o.
(Hint: C =
n=1
x X : d(x, C) < 1/n.)
(v) Show that is outer regular on B
X
.
(3.25) Let E B
R
. Show that E +x B
R
for every x R.
(3.26) Let E / and x R. Let
xE := xy [ y E and E := x [ x E.
Show that E, xE / for every x E. Compute (xE) and
(E) in terms of (E).
(3.27) Example (Vitali)(Existence of nonmeasurable sets:
Dene a relation on [0,1] as follows: for x, y [0, 1], we say x is
related to y, written as x y, if x y is a rational. Prove the
following:
(i) Show that is an equivalence relation on [0, 1].
(ii) Let E
I
denote the set of equivalence classes of elements
of [0, 1]. Using the axiom of choice, choose exactly one element
x
[ I.
Let r
1
, r
2
, . . . , r
n
, . . . denote an enumeration of the rationals in
[1, 1]. Let
E
n
:= r
n
+E, n = 1, 2, . . . .
Show that E
n
E
m
= for n ,= m and E
n
[1, 2] for every n.
Deuce that
[0, 1]
_
n=1
E
n
[1, 2].
(iii) Show that E is not Lebesgue measurable.
Chapter 4
Integration
Unless stated otherwise, we shall work on a xed -nite measure space
(X, o, ).
4.1. Integral of nonnegative simple measurable
functions
Concepts and examples:
4.1.1. Denition:
Let s : X [0, ] be dened by
s(x) =
n
i=1
a
i
A
i
(x), x X,
where n is some positive integer; a
1
, a
2
, . . . , a
n
are nonnegative extended real
numbers; A
i
o for every i; A
i
A
j
= for i ,= j; and
n
i=1
A
i
= X. Such a
function s is called a nonnegative simple measurable function on (X, o)
and
n
i=1
a
i
A
i
(x) is called a representation of s. We say
n
i=1
a
i
A
i
is
the standard representation of s if a
1
, a
2
, . . . , a
n
are all distinct.
We denote by L
+
0
the class of all nonnegative simple measurable func-
tions on (X, o).
Note that s L
+
0
i s takes only a nite number of distinct values, say
a
1
, a
2
, . . . , a
n
, the value a
i
being taken on the set A
i
o, i = 1, 2, . . . , n.
And in that case its standard representation is
n
i=1
a
i
A
i
. Also note that
the class L
+
0
depends only upon the set X and the -algebra o; the measure
plays no part in the denition of functions in L
+
0
.
35
36 4. Integration
4.1.2. Examples:
(i) Clearly, if s(x) c for some c [0, +], then s L
+
0
.
(ii) For A X, consider
A
: X [0, +], the indicator function of the
set A, i.e.,
A
(x) = 1 if x A and
A
(x) = 0 if x , A. Then
A
L
+
0
i
A o, for
A
= a
A
+b
A
c with a = 1 and b = 0.
(iii) Let A, B o. Then s =
A
B
L
+
0
, since s =
AB
.
(iv) Let A, B o. If A B = , then clearly
A
+
B
=
AB
L
+
0
.
4.1.3. Denition:
For s L
+
0
with a representation s =
n
i=1
a
i
A
i
, we dene
_
s(x)d(x),
the integral of s with respect to , by
_
s(x)d(x) :=
n
i=1
a
i
(A
i
).
The integral
_
s(x)d(x) is also denoted by
_
sd. It is well-dened (see
exercise (4.1)).
4.1.4. Proposition:
For s, s
1
, s
2
L
+
0
and R with 0, the following hold:
(i) 0
_
sd +.
(ii) s L
+
0
and
_
(s)d =
_
sd.
(iii) s
1
+s
2
L
+
0
and
_
(s
1
+s
2
)d =
_
s
1
d +
_
s
2
d.
(iv) For E o we have s
E
L
+
0
, and the set function
E (E) :=
_
s
E
d
is a measure on o. Further, (E) = 0 whenever (E) = 0, E o.
Before we proceed further, we observe that
_
s(x)d(x) is well-dened,(see
Exercise 4.1). The properties of
_
sd, for s L
+
0
, are given by the next
proposition.
4.1. Integral of nonnegative simple measurable functions 37
4.1.5. Proposition:
Let s L
+
0
. Then the following hold:
(i) If s
n
n1
is any increasing sequence in L
+
0
such that lim
n
s
n
(x)
= s(x), x X, then
_
sd = lim
n
_
s
n
d.
(ii)
_
sd = sup
_
_
s
d [ 0 s
s, s
L
+
0
_
.
Exercises:
(4.1) Show that for s L
+
0
,
_
s(x)d(x) is well-dened by proving the
following: let
s =
n
i=1
a
i
A
i
=
m
j=1
b
j
B
j
,
where A
1
, . . . , A
n
and B
1
, . . . , B
m
are partitions of X by ele-
ments of o, then
(ii)
s =
n
i=1
a
i
m
j=1
A
i
B
j
=
m
j=1
b
j
n
i=1
A
i
B
j
.
(ii)
n
i=1
a
i
(A
i
) =
m
j=1
b
j
(B
j
).
(iii)
_
s(x)d(x) is independent of the representation of the func-
tion s(x) =
n
i=1
a
i
A
i
.
(4.2) Let A, B o. Express the functions [
A
B
[ and
A
+
B
AB
as indicator functions of sets in o and hence deduce that they
belong to L
+
0
.
(4.3) Let s
1
, s
2
L
+
0
. Prove the following:
(i) If s
1
s
2
, then
_
s
1
d
_
s
2
d.
(ii) Let x X,
(s
1
s
2
)(x) := maxs
1
(x), s
2
(x) and (s
1
s
2
)(x) := mins
1
(x), s
2
(x).
Then s
1
s
2
and s
1
s
2
L
+
0
with
_
(s
1
s
2
)d
_
s
i
d
_
(s
1
s
2
)d, i = 1, 2.
38 4. Integration
(4.4) Express the functions
A
B
and
A
B
, for A, B o, in terms
of the functions
A
and
B
.
(4.5) Let X = (0, 1], o = B
(0,1]
, the -algebra of Borel subsets of (0, 1]
and = , the Lebesgue measure restricted to o. For x (0, 1], if
x has non-terminating dyadic expansion x =
n=1
x
n
/2
n
. Let
f
i
(x) :=
_
+1 if x
i
= 1,
1 if x
i
= 0, i = 1, 2, . . . .
Show that for every i, there exists simple function s
i
L
+
0
such
that f
i
= s
i
1. Compute
_
s
i
d.
(4.6) Let s : X R
.
(4.7) For s
1
, s
2
L
+
0
show that x[ s
1
(x) s
2
(x) o. Can you say
that the sets x X[ s
1
(x) > s
2
(x), x X[ s
1
(x) s
2
(x) and
x X[ s
1
(x) = s
2
(x) are also elements of o?
(4.8) Let s
1
, s
2
L
+
0
be real valued and s
1
s
2
. Let = s
1
s
2
. Show
that L
+
0
. Can you say that
_
d =
_
s
1
d
_
s
2
d?
(4.9) Let s
n
n1
and s
n1
be sequences in L
+
0
such that for each
x X, both s
n
(x)
n1
and s
n
(x)
n1
are increasing and
lim
n
s
n
(x) = lim
n
s
n
(x).
Show that
lim
n
_
s
n
d = lim
n
_
s
n
d.
(Hint: Apply exercise (4.3) and proposition 4.1.5 to s
n
s
n
for
all xed m to deduce that
_
s
m
d lim
n
_
s
n
d.)
(4.10) Show that in general L
+
0
need not be closed under limiting oper-
ations. For example, consider the Lebesgue measure space (R, /, )
and construct a sequence s
n
n1
in L
+
0
such that lim
n
s
n
(x) = f(x)
exists but f , L
+
0
.
4.2. Integral of nonnegative measurable
functions
Concepts and examples:
Having dened the integral for functions s L
+
0
, i.e., nonnegative simple
measurable functions, we would like to extend it to a larger class.
4.2. Integral of nonnegative measurable functions 39
4.2.1. Denition:
(i) A nonnegative function f : X R
is said to be o-measurable if
there exists an increasing sequence of functions s
n
n1
in L
+
0
such that
f(x) = lim
n
s
n
(x) x X.
If the underlying -algebra is clear from the context, a o-measurable
function is also called measurable. We denote the set of all nonnegative
measurable functions by L
+
.
(ii) For a function f L
+
, we dene the integral of f with respect to by
_
f(x)d(x) := lim
n
_
s
n
(x)d(x).
It follows from exercise (4.9) that for f L
+
,
_
fd is well-dened. Clearly,
L
+
0
L
+
and
_
sd for an element s L
+
0
is the same as
_
sd, for s as an
element of L
+
. The next proposition gives a characterization of functions in
L
+
and the integrals of its elements. Another (intrinsic) characterization of
L
+
will be given in the next section.
4.2.2. Proposition:
Let f : X R
n1
be an increasing sequence of functions in L
+
, and f(x) :=
lim
n
f
n
(x), x X. Then f L
+
and
_
fd = lim
n
_
f
n
d.
4.2.6. Remark:
If f
n
n1
is a sequence in L
+
decreasing to a function f L
+
, then the
equality
_
fd = lim
n
_
f
n
d need not hold. For example, let X = R,
o = / and = , the Lebesgue measure. Let f
n
=
[n,)
. Then f
n
L
+
0
L
+
, and f
n
n1
decreases to f 0. Clearly,
_
f
n
d = + for every n and
_
fd = 0. In fact, at this stage it is not clear whether f L
+
whenever
f
n
n1
decreases to f, with each f
n
L
+
. That this is true will be shown
as a consequence of the characterization of L
+
proved in the next section.
4.2. Integral of nonnegative measurable functions 41
Exercises:
(4.11) Let f L
+
and let s
n
n1
be in L
+
0
and such that s
n
(x)
n1
is decreasing and x X, lim
n
s
n
(x) = f(x). Can you conclude
that
_
fd = lim
n
_
s
n
d?
(4.12) Let f L
+
. Show that
_
fd = sup
__
sd 0 s(x) f(x) for a.e. x(), s L
+
0
_
.
(4.13) Let f
n
n1
be an increasing sequence of functions in L
+
such that
f(x) := lim
n
f
n
(x) exists for a.e. x(). Show that f L
+
and
_
fd = lim
n
_
f
n
d,
where f(x) is dened as an arbitrary constant for all those x for
which lim
n
f
n
(x) does not converge.
(4.14) Let (X) < , and let f L
+
be a bounded function. Let P :=
E
1
, E
2
, . . . , E
n
be such that
n
i=1
E
i
= X, E
i
E
j
= for i ,= j
and E
i
o i. Such a P is called a measurable partition of X.
Given a measurable partition P = E
1
, . . . , E
n
, dene
M
i
:= supf(x) [ x E
i
and m
i
:= inff(x) [ x E
i
.
Let
P
:=
n
i=1
m
i
E
i
and
P
:=
n
i=1
M
i
E
i
.
Prove the following:
(i) For every partition P, show that
P
,
P
L
+
0
and
P
f
P
.
(ii)
_
fd = sup
__
P
d[ P is a measurable partition of X
_
,
= inf
__
P
d[ P is a measurable partition of X
_
.
(This gives an equivalent way of dening
_
fd, in a way sim-
ilar to that for the Riemann integral.)
(iii) Let
= sup
__
0
sd[ s L
+
0
, s f
_
and
= inf
__
sd[ s L
+
0
, f s
_
.
42 4. Integration
Show that
= sup
__
p
d [ P is a measurable partition of X
_
and
= inf
__
p
d [ P is a measurable partition of X
_
.
(iv) Deduce that f L
+
implies =
_
fd = .
Note:
Exercise 4.14 tells us that for f L
+
, in dening
_
fd it is enough
to consider approximations of f from below, as the approximations
from above will also give the same value for
_
fd. This is because
f L
+
, i.e., f is nonnegative measurable. The converse is also
true, i.e., if = , for f : X [0, ], then f L
+
. To see this,
rst note that
= < M((X)) < ,
where M is such that [f(x)[ M x X. Thus for every n we
can choose functions
n
,
n
L
+
0
such that
n
f
n
and
_
(
n
n
)d <
1
n
.
Let := sup
n
and := inf
n
. Then , L
+
(see corollary
4.3.11) and n,
_
( )d
_
(
n
n
)d <
1
n
.
Thus
_
( )d = 0
and by proposition 4.3.3, = = f a.e. . Hence f L
+
.
4.3. Intrinsic characterization of nonnegative
measurable functions
Concepts and examples:
Recall, f L
+
means f : X R
n1
of functions in L
+
0
such that
s
n
(x)
n1
increases to f(x) for every x X. Note that the measure
plays no part in the denition of the functions in L
+
. It is only to dene
the integral of functions f L
+
that we need the measure . We want to
characterize the functions in L
+
intrinsically.
4.3. Intrinsic characterization of nonnegative measurable functions 43
4.3.1. Proposition:
Let s : X R
.
(iii) s
1
[t, ] o t R
.
(iv) s
1
(I) o for every interval I in R
.
In view of the above proposition it is natural to ask the following ques-
tion: does proposition 4.3.1 remain true if s L
+
0
is replaced by any f L
+
?
The answer is yes, as proved in the next proposition.
4.3.2. Proposition:
Let f : X R
and f
1
(E) o for every E B
R
.
The statements (ii) to (vi) of proposition 4.3.2 describe the elements of
L
+
intrinsically. This proposition is used very often to check the measura-
bility of nonnegative functions. Once again we emphasize the fact that for
a nonnegative function f : X R
are o-measurable.
We denote by L the class of all o-measurable functions on X. If the
underlying -algebra is clear from the context, we call a o-measurable
function to be measurable.
(ii) If f L is such that both f
+
, f
L
+
0
, we call f a simple measurable
function.
We denote the class of all simple measurable functions by L
0
. Note
that s L
0
i both s
+
, s
L
+
0
.
4.3.5. Proposition:
Let (X, o) be a measurable space and f : X R
n1
of real valued functions on X such that
n, s
+
n
and s
n
are both nonnegative simple measurable functions on (X, o)
and lim
n
s
n
(x) = f(x) x X.
(iii) f satises any one (and hence all) of the statements (ii) to (vi) of propo-
sition 4.3.2.
4.3.6. Examples:
(i) Let f : X R
_
x X [ f(x) < c/ if > 0,
X if = 0, c 0,
if = 0, c > 0,
x X [ f(x) < c/ if < 0.
(iii) Let X be a topological space and o be the -algebra of Borel subsets
of X, i.e., the -algebra generated by the open sets. Let f : X R be
any continuous function. Then f is o-measurable.
We next describe some more properties of measurable functions.
4.3.7. Proposition:
Let f, g be measurable functions. Then each of the sets x X[ f(x) >
4.3. Intrinsic characterization of nonnegative measurable functions 45
g(x), x X[ f(x) g(x), x X[ f(x) < g(x), x X[ f(x) g(x)
and x X[ f(x) = g(x) o.
4.3.8. Proposition:
Let f, g : X R
be arbitrary. Let
A := x X[ f(x) = +, g(X) = x X[f(x) = , g(x) = +.
Dene x X
(f +g)(x) :=
_
f(x) +g(x) if x , A,
if x A.
Then f +g : X R
be such that Rx
R
[ (x) B
R
, R. Then f is also measurable.
4.3.10. Proposition:
Let f
n
: X R
n1
converges to f, then f is a measurable function.
4.3.11. Corollary:
Let f
n
n1
be a sequence in L
+
. Then each of the functions sup
n
f
n
, inf
n
f
n
,
limsup
n
f
n
and liminf
n
f
n
is in L
+
. In particular, if lim
n
f
n
=: f exists, then
f L
+
.
Recall that in theorem 4.2.5 we analyzed the limit of
_
f
n
d for an in-
creasing sequence of nonnegative measurable functions. We analyze next the
behavior of
_
f
n
d when f
n
n1
is not necessarily an increasing sequence
of nonnegative measurable functions.
4.3.12. Theorem (Fatous lemma):
Let f
n
n1
be a sequence of nonnegative measurable functions. Then
_
_
liminf
n
f
n
_
d liminf
n
_
f
n
d.
4.3.13. Proposition:
Let (X, o) be a measurable space and let f : X R
be o-measurable. Let
be a measure on (X, o). Let g : X R
n1
of simple measurable functions such that
s
n
n1
converges uniformly to f.
(Hint: If 0 f(x) M x, then [s
n
(x) f(x)[ < 1/2
n
n n
0
,
where n
0
M and the s
n
s are as in proposition 4.3.2.)
(4.17) Let f : X R
n1
and s
n
n1
such that
0 s
n
(n) s
n+1
(x) f(x) s
n+1
(x) s
n
(x)
and lim
n
s
n
(x) = f(x) = lim
n
s
n
(x) x X.
(4.18) Let f and g : X R
_
1/f(x) if f(x) , 0, +, ,
if f(x) = 0,
if f(x) = ,
if f(x) = +.
Then 1/f is a measurable function.
(v) Let R
n1
be a sequence of functions in L
+
and let
n=1
f
n
(x) =:
f(x), x X. Show that f L
+
and
_
fd =
n=1
_
f
n
d.
(4.24) Show that each of the functions f : R R dened below is /-
measurable, and compute
_
fd :
(i) f(x) :=
_
0 if x 0,
1/x if x > 0.
(ii) f(x) :=
Q
(x), the indicator function of Q, the set of rationals.
48 4. Integration
(iii) f(x) :=
_
_
0 if x > 1 or x < 0 or x [0, 1] and x is rational,
n if x is an irrational, 0 < x < 1 and in the
decimal expansion of x, the rst nonzero
entry is at the (n + 1)
th
place.
(4.25) Let f, g L
+
with f g. Show that (fg) L
+
and
_
fd
_
gd.
Can you conclude that
_
(f g)d =
_
fd
_
gd?
(4.26) Let f, f
n
L
+
, n = 1, 2, . . ., be such that 0 f
n
f. If
lim
n
f
n
(x) = f(x), can you deduce that
_
fd = lim
n
_
f
n
d?
(4.27) Let f L. For x X and n 1, dene
f
n
(x) :=
_
_
_
f(x) if [f(x)[ n,
n if f(x) > n,
n if f(x) < n.
Prove the following:
(i) f
n
L and [f
n
(x)[ n n and x X.
(ii) lim
n
f
n
(x) = f(x) x X.
(iii) [f
n
(x)[ := min[f
n
(x)[, n := ([f[ n)(x) is an element of L
+
and
lim
n
_
[f
n
[d =
_
[f[d.
Note:
For f L, the sequence f
n
n1
as dened in exercise 4.27 is called
the truncation sequence of f. The truncation sequence is useful
in proving results about functions in the class L.
(4.28) Let f L and
(E) := x X[ f(x) E, E B
R
.
Show that is a measure on (R, B
R
). Further, if g : R R is any
nonnegative B
R
-measurable function, i.e., g
1
(A) B
R
A B
R
,
then g f L and
_
g d =
_
(g f) d.
The measure is usually denoted by f
1
and is called the distri-
bution of the measurable function f.
4.3. Intrinsic characterization of nonnegative measurable functions 49
(4.29) Let f L
+
be a bounded function, say f(x) N x X and for
some N N. Show that
_
f d = lim
n
N2
n
k=1
k 1
2
n
_
x [
k 1
2
n
f(x) <
k
2
n
_
. (4.1)
a b
m
M
Figure: Area below the curve y = f(x) from inside
Note that here, for every n N, we are taking the partition
0, 1/2
n
, 2/2
n
, . . . , N of the range of f and using it to induce the
measurable partition P
n
of the domain [X[ of f, given by
X
n
k
[ 1 k N2
n
, where X
n
k
:= f
1
[(k 1)/2
n
, k/2
n
).
Using this partition P
n
, we form the sums on the right hand side
of (4.1), approximating the area below the curve y = f(x) from
inside (see Figure). Thus in our integral we use partitions of the
range, whereas in Riemann integration we use partitions of the
domain. In the words of H. Lebesgue, the situation is similar to
the problem of nding the total amount in a cash box containing
say n coins, the ith coin being of denomination
i
, 1 i n.
One method of nding the total amount in the box is to add the
value of each coin, i.e., the sum
n
i=1
i
. The other method is to
separate out coins of similar denominations. If there are k
j
coins
of denominations
j
, then the total amount is
j
k
j
j
.
(4.30) Let (X, o, ) be a measure space and let (X,
o, ) be its completion.
Let f : X R be an
o-measurable function. Show that there
exists an o-measurable function f : X R such that f(x) = f(x)
for a.e. x().
(4.31) Let (X, o, ) be a complete measure space and f
n
n1
be a se-
quence of o-measurable functions on X. Let f be a function on
50 4. Integration
X such that f(x) = lim
n
f
n
(x) for a.e. x(). Show that f is o-
measurable.
4.4. Integrable functions
Concepts and examples:
Given a measure space (X, o, ) in section 4.2, we dened
_
fd, the integral
for functions f L
+
, i.e., f : X R
, f nonnegative measurable. If f
is measurable, but not necessarily nonnegative, we can write f = f
+
f
.
Since both f
+
and f
d are
dened. Thus it is reasonable (as the integral is expected to be linear) to
dene the integral of f to be
_
fd :=
_
f
+
d
_
f
d =
_
f
+
d = +. To overcome this diculty, we
introduce the following denition:
4.4.1. Denition:
A measurable function f : X R
d.
We denote by L
1
(X, o, ) (or simply by L
1
(X) or L
1
()) the space of all
-integrable functions on X.
The properties of the space L
1
(X, o, ) and of the map f
_
fd,
f L
1
(X, o, ), are given in the next proposition.
4.4.2. Proposition:
For f, g L and a, b R, the following hold:
(i) If [f(x)[ g(x) for a.e. x() and g L
1
(), then f L
1
().
(ii) If f(x) = g(x) for a.e. x() and f L
1
(), then g L
1
() and
_
fd =
_
gd.
(iii) If f L
1
(), then af L
1
() and
_
(af)d = a
_
fd.
(iv) If f and g L
1
(), then f +g L
1
() and
_
(f +g)d =
_
fd +
_
gd.
4.4. Integrable functions 51
(v) If f, g L
1
(), then (af +bg) L
1
() and
_
(af +bg)d = a
_
fd +b
_
gd.
4.4.3. Proposition:
Let f L
1
(). For every E o, let
(E) :=
_
E
[f[d and (E) :=
_
E
fd.
Then the following hold:
(i) If (E) = 0, then (E) = 0.
(ii) If (E) = 0, then (E) = 0.
(iii) lim
(E)0
(E) = 0, i.e., given any > 0, there exists > 0 such that
(E) < whenever, for E o, (E) < .
(iv) If (E) = 0 E, then f(x) = 0 for a.e. x() on E.
4.4.4. Remark:
It is easy to see that (iii) in the above proposition implies (ii). In fact (ii)
also implies (iii). To see this, suppose (iii) does not hold. Then there exist
an > 0 and sets E
n
o, n 1, such that (E
n
) < 2
n
but (E
n
) . Let
A
n
=
k=n
E
k
. Then A
n
n1
is a decreasing sequence in o and (A
n
)
(E
n
) 2
n
. Thus by theorem 2.3.2,
n=1
A
n
_
= lim
n
(A
n
) = 0.
On the other hand, (A
n
) (E
n
) , , contradicting (ii).
We prove next the most frequently used theorem which allows us to
interchange the operations of integration and limits.
4.4.5. Theorem (Lebesgues dominated convergence theorem):
Let f
n
n1
be a sequence of measurable functions and let g L
1
() be
such that n, [f
n
(x)[ g(x) for a.e. x(). Let f
n
(x)
n1
converge to
f(x) for a.e. x(). Then the following hold:
(i) f L
1
().
(ii)
_
f d = lim
n
_
f
n
d.
(iii) lim
n
_
[f
n
f[d = 0.
We state another version of this theorem, which is applicable to series
of functions.
52 4. Integration
4.4.6. Corollary:
Let f
n
n1
be a sequence of functions in L
1
() such that
n=1
_
[f
n
[d
< +. Then f(x) :=
n=1
f
n
(x) exists for a.e. x(), f L
1
() and
_
fd =
n=1
_
f
n
d.
The following is a variation of the dominated convergence theorem for
nite measure spaces. See exercise (4.38) also.
4.4.7. Theorem (Bounded convergence):
Let (X, o, ) be a nite measure space and f, f
1
, f
2
, . . . be measurable func-
tions. Suppose there exists M > 0 such that [f
n
(x)[ M a.e. x() and
f
n
(x) f(x) a.e. x(). Then f, f
n
L
1
(X, o, ) and
_
fd = lim
n
_
f
n
d.
4.4.8. Notes:
(i) The monotone convergence theorem and the dominated convergence theo-
rem (along with its variations and versions) are the most important theorems
used for the interchange of integrals and limits.
(ii) Simple function technique: This is an important technique (similar
to the -algebra technique) used very often to prove results about integrable
and nonnegative measurable functions. Suppose we want to show that a
certain claim () holds for all integrable functions. Then technique is the
following:
(1) Show that () holds for nonnegative simple measurable functions.
(2) Show that () holds for nonnegative measurable / integrable func-
tions by approximating them by nonnegative simple measurable
functions and using (1).
(3) Show that () holds for integrable functions f, by using (2) and the
fact that for f L
1
, f = f
+
f
and both f
+
, f
L
1
.
The proof of the next proposition is an illustration of this technique.
4.4.9. Proposition:
Let (X, o, ) be a -nite measure space and f L
1
(X, o, ) be nonnegative.
For every E o, let
(E) :=
_
E
fd.
4.4. Integrable functions 53
Then is a nite measure on o. Further, fg L
1
(X, o, ) for every g
L
1
(X, o, ), and
_
fd =
_
fgd.
We shall see some more applications of the dominated convergence the-
orem in the remaining sections. In the next section we look at some special
properties of L
1
(X, o, ) in the particular case when X = R, o = /, the
-algebra of Lebesgue measurable sets, and = , the Lebesgue measure.
As an application of the dominated convergence theorem, we exhibit the
possibility of interchanging the order of integration and dierentiation in
the next theorem.
4.4.10. Theorem:
Let f
t
L
1
() for every t (a, b) R. Let t
0
(a, b) be such that for
a.e. x(), t f
t
(x) is dierentiable in a neighborhood U of t
0
and there
exists a function g L
1
() such that
df
t
dt
(x) g(x) for a.e. x() and for
every t U. Then (t) :=
_
f
t
(x)d(x) is dierentiable at t
0
and
(t
0
) =
_
_
_
df
t
dt
(x)
_
t
0
_
d(x).
Exercises:
(4.32) For f L, prove the following:
(i) f L
1
() i [f[ L
1
(). Further, in either case
_
fd
_
[f[d.
(ii) If f L
1
(), then [f(x)[ < + for a.e. x().
(4.33) Let (X) < +and let f L be such that [f(x)[ M for a.e. x()
and for some M. Show that f L
1
().
(4.34) Let f L
1
() and E o. Show that
E
f L
1
(), where
_
E
fd :=
_
E
fd.
Further, if E, F o are disjoint sets, show that
_
EF
fd =
_
E
fd +
_
F
fd.
54 4. Integration
(4.35) Let f L
1
() and E
i
o, i 1, be such that E
i
E
j
= for
i ,= j. Show that the series
i=1
_
E
i
fd is absolutely convergent,
and if E :=
i=1
E
i
, then
i=1
_
E
i
fd =
_
E
fd.
(4.36) (i) For every > 0 and f L
1
(), show that
x X [ [f(x)[
1
_
[f[d < .
This is called Chebyshevs inequality.
(ii) Let f L
1
(), and let there exist M > 0 such that
1
(E)
_
E
fd
M
for every E o with 0 < (E) < . Show that [f(x)[ M for
a.e. x().
(4.37) Let be the Lebesgue measure on R, and let f L
1
(R, /, ) be
such that
_
(,x)
f(t)d(t) = 0, x R.
Show that f(x) = 0 for a.e. ()x R.
(4.38) Let I R be an interval and t I, let f
t
L. Let g L
1
()
be such that t, [f
t
(x)[ g(x) for a.e. x(). Let t
0
R
be any
accumulation point of I and let f(x) := lim
tt
0
f
t
(x) exist for a.e. x().
Then f L
1
() and
_
fd = lim
tt
0
_
f
t
(x)d(x).
Further, if x the function t f
t
(x) is continuous, then so is
the function h(t) :=
_
f
t
(x) d, t I.
(Hint: Apply theorem 4.4.5 to every sequence t
n
t
0
.)
(4.39) Let f
n
n1
and g
n
n1
be sequences of measurable functions such
that [f
n
[ g
n
n. Let f and g be measurable functions such that
lim
n
f
n
(x) = f(x) for a.e. x() and lim
n
g
n
(x) = g(x) for a.e. x().
If
lim
n
_
g
n
d =
_
g d < +,
show that
lim
n
_
f
n
d =
_
fd.
(Hint: Apply Fatous lemma to (g
n
f
n
) and (g
n
+f
n
).)
4.5. The Lebesgue integral and its relation with the Riemann integral 55
(4.40) Let f
n
n0
be a sequence in L
1
(X, o, ). Show that
__
[f
n
[d
_
n1
converges to
_
[f
0
[d i
__
[f
n
f
0
[d
_
n1
converges to zero.
(Hint: Use exercise 4.38.)
(4.41) Let (X, o) be a measurable space and f : X R be o-measurable.
Prove the following:
(i) o
0
:= f
1
(E) [ E B
R
is the -algebra of subsets of X, and
o
0
o.
(ii) If : R R is Borel measurable, i.e.,
1
(E) B
R
E
B
R
, then f is an o
0
-measurable function on X.
(iii) If : X R is any o
0
-measurable function, then there exists
a Borel measurable function : R R such that = f.
(Hint: Use the simple function technique and note that if is
a simple o
0
-measurable function, then
=
n
i=1
a
n
f
1
(E
i
)
for some positive integer n, a
i
R for each i, and E
i
B
R
,
then
=
n
i=1
a
i
(
E
i
f).
(4.42) Let f
n
n1
be a decreasing sequence of nonnegative functions in
L
1
() such that f
n
(x) f(x). Show that
lim
x
_
f
n
d = 0 i f(x) = 0 a.e. x().
(4.43) Let , be as in proposition 4.4.9. Let o
.
(ii) There exist examples such that o is a proper subclass of o
.
Show that o = o
if
x X[ f(x) = 0 = 0.
(4.44) Let (X, o, ) be a nite measure space and f
n
n1
be a sequence
in L
1
() such that f
n
f uniformly. Show that f L
1
() and
lim
n
_
[f
n
f [ d = 0.
Can the condition of (X) < + be dropped?
4.5. The Lebesgue integral and its relation with
the Riemann integral
Concepts and examples:
56 4. Integration
In this section we analyze the integral, as constructed in the previous section,
for the particular situation when X = R, o = / (the -algebra of Lebesgue
measurable sets) and = , the Lebesgue measure. The space L
1
(R, /, ),
also denoted by L
1
(R) or L
1
(), is called the space of Lebesgue integrable
functions on R, and
_
fd is called the Lebesgue integral of f. For any
set E /, we write L
1
(E) for the space of integrable functions on the
measure space (E, / E, ), where is restricted to / E. In the special
case when E = [a, b], we would like to show that the new notion of integral
for f L
1
[a, b] indeed extends the notion of Riemann integral. To be precise,
we have the following theorem:
4.5.1. Theorem:
Let f : [a, b] R be a Riemann integrable function. Then f L
1
[a, b] and
_
fd =
_
b
a
f(x)dx.
4.5.2. Remark:
In fact, the proof of the above theorem includes a proof of the following: If
f 1[a, b], then f is continuous a.e. x(). This is because
f(x) = lim
n
n
(x) = lim
n
n
(x) a.e. x().
Thus, if we put
E := x [a, b] [ f(x) = lim
n
n
(x) = lim
n
n
(x),
then E is a Lebesgue measurable set and ([a, b] E) = 0. For x E, given
an arbitrary > 0, we can choose n
0
such that
n
0
(x)
n
0
(x) < . (4.2)
Further, if x is not a point in any partition P
n
, then we can choose >
0 such that whenever y [a, b] and [x y[ < , then y belongs to the
same subinterval of the partition P
n
0
to which x belongs. Thus by (4.2),
[f(x) f(y)[ < , showing that x is a point of continuity of f. Thus the
set of discontinuity points of f forms a subset of ([a, b]) E) P, where
P is the set of partition points of P
n
, n = 1, 2, . . . . Hence f is continuous
almost everywhere. Another proof of the converse is given in the one of the
exercises.
Exercises:
(4.45) Let f : [a, b] R be bounded and continuous for a.e. x().
4.5. The Lebesgue integral and its relation with the Riemann integral 57
(i) Let P
n
n1
be any sequence of partitions of [a, b] such that
each P
n+1
is a renement of P
n
and |P
n
| 0 as n .
Let
n
,
n
be as constructed in theorem 4.5.1. Let x (a, b)
be a point of continuity of f. Show that
lim
n
n
(x) = f(x) = lim
n
n
(x).
(ii) Using (i) and the dominated convergence theorem, deduce that
f L
1
([a, b]) and
_
fd = lim
n
_
n
d = lim
n
_
n
d.
(iii) Show that f 1[a, b] and
_
fd =
_
b
a
f(x)dx.
(4.46) Let f : [0, 1] [0, ) be Riemann integrable on [, 1] for all > 0.
Show that f L
1
[0, 1] i lim
0
_
1
f(x)dx.
(4.47) Let f(x) = 1/x
p
if 0 < x 1, and f(0) = 0. Find necessary and suf-
cient condition on p such that f L
1
[0, 1]. Compute
_
1
0
f(x)d(x)
in that case.
(Hint: Use exercise 4.46.)
(4.48) (Mean value property):Let f : [a, b] R be a continuous
function and let E [a, b], E /, be such that (E) > 0. Show
that there exists a real number such that
_
E
f(x)(x) = (E).
(4.49) Let f L
1
(R), and let g : R R be a measurable function such
that g(x) for a.e. x(). Show that fg L
1
(R) and there
exists [, ] such that
_
[f[gd =
_
[f[d.
(4.50) Let f L
1
(R, /, ) and let a R be xed. Dene
F(x) :=
_
_
_
[a,x]
f(t) d(t) for x a,
_
[x,a]
f(t)d(t) for x a.
Show that F is continuous.
58 4. Integration
(Hint: Without loss of generality take f 0 and show that F is
continuous from the left and right. In fact, F is actually uniformly
continuous.)
(4.51) Let f L
1
(R, /, ) and let c R be a point of continuity of f.
Show that
lim
n
n
_
[c, c + 1/n]
f(x) d(x) = f(c).
(4.52) (Arzelas theorem):
Let f
n
n1
be a sequence of Riemann integrable functions on [a, b]
such that for some M > 0, [f
n
(x)[ < M x [a, b] and n =
1, 2, . . . . Let f
n
(x) f(x) x [a, b] and let f be Riemann
integrable on [a, b]. Then
lim
n
_
b
a
f
n
(x)dx =
_
b
a
f(x)dx.
(4.53) Let f : [a, b] R be any constant function. Show that f
L
1
[a, b].
(4.54) Let f : [a, b] R be any bounded measurable function. Show
that f L
1
[a, b].
(4.55) Let f : [a, b] R be any continuous function. Show that f
L
1
[a, b].
(4.56) Let f L
1
(R) be such that
_
K
fd = 0 for every compact set
K R. Show that f(x) = 0 for a.e. x().
(4.57) Let f
n
n1
be a decreasing sequence of nonnegative functions in
C(a, b) and let f
1
L
1
(a, b). If
n=1
(1)
n1
f
n
C(a, b), show
that
_
b
a
_
n=1
(1)
n1
f
n
(x)
_
dx =
n=1
_
(1)
n1
__
b
a
f
n
(x)dx
__
.
(Hint: For every n,
n
k=1
(1)
k
f
k
f L
1
(a, b)).
(4.58) Give examples to show that analogues of the monotone convergence
theorem and the dominated convergence theorem do not hold for
the Riemann integral.
(4.59) Let f L
1
(R) and, t [0, ),
g(t) = sup
__
[f(x +y) f(x)[ d(x) t y t
_
.
Show that g is continuous at t = 0.
(Hint: Use the simple function technique).
4.6. L
1
[a, b] as the completion of 1[a, b] 59
4.6. L
1
[a, b] as the completion of 1[a, b]
Concepts and examples:
Recall that, 1[a, b] is not a complete metric space under the L
1
-metric
d(f, g) :=
_
b
a
[f(x) g(x)[dx,
for f, g 1[a, b]. Since every metric space has a completion, the question
arises: what is the completion of 1[a, b] under this metric? You might have
seen the abstract construction of the completion of a metric space. We shall
show that for 1[a, b] this completion is nothing but L
1
[a, b]. Thus L
1
[a, b]
is the concrete realization of the completion of 1[a, b]. We shall rst show
that L
1
[a, b] is a complete metric space, and then show that 1[a, b] is dense
in L
1
[a, b].
4.6.1. Denition:
For f, g L
1
[a, b], we say f is equivalent to g, and write f g, if the set
x [a, b] [ f(x) ,= g(x) has Lebesgue measure zero.
It is easy to see that the relation is an equivalence relation on L
1
[a, b].
We denote the set of equivalence classes again by L
1
[a, b]. In other words,
we identify two functions f, g with each other if f g. Thus, for g, f
L
1
[a, b], f = g i f(x) = g(x) for a.e. x(), as functions.
4.6.2. Denition:
For f L
1
[a, b], we dene
|f|
1
:=
_
[f(x)[d(x).
Clearly |f|
1
is well-dened, and it is easy to check that the function f
|f|
1
, f L
1
[a, b], has the following properties:
(i) |f|
1
0 f L
1
[a, b].
(ii) |f|
1
= 0 i f = 0.
(iii) |af|
1
= [a[ |f|
1
a R and f L
1
[a, b].
(iv) |f +g|
1
|f|
1
+|g|
1
f, g L
1
[a, b].
The function | |
1
is called a norm on L
1
[a, b]. (Geometrically it is the
distance of the vector f L
1
[a, b] from the vector 0 L
1
[a, b].) For
f, g L
1
[a, b], if we dene
d(f, g) := |f g|
1
,
60 4. Integration
then d is a metric on L
1
[a, b], called the L
1
-metric . The most important
property of this metric is given by the next theorem.
4.6.3. Theorem (Riesz-Fischer):
L
1
[a, b] is a complete metric space in the L
1
-metric.
We show next that C[a, b], the space of continuous function on [a, b],
(and hence 1[a, b]) is dense in L
1
[a, b].
4.6.4. Theorem:
The space C[a, b] is a dense subset of L
1
[a, b].
Since C[a, b] 1[a, b], theorems 4.6.3 and 4.6.4 together prove the fol-
lowing theorem:
4.6.5. Theorem:
L
1
[a, b] is the completion of 1[a, b].
4.6.6. Notes:
(i) In the proof of theorem 4.6.3, one does not use require anywhere the
fact that the functions are dened on an interval. One can dene | |
1
for functions dened on (E, / E, ), where E / is arbitrary, and the
proof of theorem 5.6.1 will show that L
1
(E) := L
1
(E, / E, ) is also a
complete metric space under the metric
|f g|
1
:=
_
E
[f(x) g(x)[d(x).
A closer look will show that the metric d makes sense on L
1
(X, o, ),
where (X, o, ) is any complete measure space and we identify functions
which agree for a.e. (). Further, theorem 4.6.3 also remains true for
L
1
(X, o, ).
(ii) Parts of the proof of theorem 4.6.4 also exhibit the following facts which
are of independent interest. Let f L
1
[a, b] and > 0 be given. Then
there exists a simple function such that |f |
1
< and a step function
h such that |f h|
1
< . Thus simple functions in L
1
[a, b] are dense in it.
Exercises:
(4.60) Let f L
1
(R) and let > 0 be given. Prove the following:
(i) There exists a positive integer n such that
|f
[n,n]
f|
1
< .
4.6. L
1
[a, b] as the completion of 1[a, b] 61
(ii) There exists a continuous function g on R such that g is zero
outside some nite interval and |f
[n,n]
g|
1
< .
(iii) For f : R R, let
supp (f) := closure x R [ f(x) ,= 0.
The set supp(f) is called the support of f and is the smallest
closed subset of R outside which f vanishes. Let the space of
continuous functions on R be denoted by C(R) and let
C
c
(R) := f : R R[ f C(R) and supp(f) is compact.
Then C
c
(R) is a dense subset of L
1
(R).
(4.61) Let f L
1
(R) and
f(x) := f(x) x R. Prove the following:
(i)
_
f(x)d(x) =
_
f(x)d(x).
(ii) fg L
1
(R) for any bounded measurable function g : R R.
Dense subspaces of L
1
(R) are very useful in proving results
about L
1
-functions. As an application of exercise 4.59 we have the
following:
(4.62) Let f L
1
(R) and for every h, k R, let
f
h
(x) := f(x +h) and (x) := f(kx +h), x R.
Prove the following:
(i) f
h
, L
1
(R) with
_
(x)d(x) = [k[
_
f(x)d(x) and
_
f
h
(x)d(x) =
_
f(x)d(x).
(ii) For every h R, f
h
L
1
(R) and lim
h0
|f
h
f|
1
= 0.
(Hint : Use exercise 4.60 to get a function g C
c
(R) with
|g f|
1
< , and note that |g
h
g|
1
< 2(b a + 1) when
supp(g) [a, b] and [h[ is suciently small.)
(iii) The function h |f
h
|
1
is continuous.
(4.63) (Riemann-Lebesgue lemma): Let f L
1
(R), and let g : R
R be any bounded measurable function such that g(x+p)g(x) = 0
for every x R and p R xed. Show that t ,= 0,
_
f(x)g(tx)dx
M |f
p/t
f|
1
,
where M = sup [g(x)[/2. Hence deduce (using exercise 4.60) that
lim
|t|
_
f(x)g(tx)dx = 0.
62 4. Integration
(In the special case when f L
1
[0, 2] and g(x) = cos x or sin x,
we have
lim
n
_
2
0
f(x) cos nxdx = 0 = lim
n
_
2
0
f(x) sin nxdx.
This nds applications in the theory of Fourier series.)
Chapter 5
Measure and
integration on product
spaces
5.1. Introduction
Concepts and examples:
The intuitive notion of length, originally dened for intervals in R, leads one
to the class of Lebesgue measurable sets which included not only the intervals
and all the topologically nice subsets of R, but also B
R
the -algebra of
Borel subsets of R. In a similar manner, one would like to extend the notion
of area in R
2
(volume in R
3
, and so on) to a larger class of subsets which
includes B
R
2 (B
R
3) the -algebra generated by open subsets of R
2
(R
3
).
In the abstract setting, given measure spaces (X, /, ) and (Y, B, ), one
would like to dene a measure on the -algebra generated by sets of the
form A B[ A /, B B in the natural way: (A B) = (A)(B).
We call this natural for, when X = Y = R, / = B = B
R
and = = , the
Lebesgue measure, then for intervals I and J, (I J) = (I)(J) is the area
of the rectangle with sides the intervals I and J. Thus will automatically
be an extension of the notion of area in R
2
. We note that the collection
1 := A B[ A /, B B is only a semi-algebra of subsets of A B in
general. Let /B denote the -algebra of subsets of X Y generated by
1.
63
64 5. Measure and integration on product spaces
5.1.1. Denition:
Let (X, /) and (Y, B) be measurable spaces. A subset E XY is called
a measurable rectangle if E = AB for some A / and B B. We
denote by 1 the class of all measurable rectangles. The -algebra of subsets
of XY generated by the semi-algebra 1 is called the product -algebra
and is denoted by /B.
5.1.2. Proposition:
Let p
X
: X Y X and p
Y
: X Y Y be dened by
p
X
(x, y) = x and p
Y
(x, y) = y,
x X, y Y . Then the following hold:
(i) The maps p
X
and p
Y
are measurable, i.e., A /, B B we
have p
1
X
(A) /B and p
1
Y
(B) /B.
(ii) The -algebra /B is the smallest -algebra of subsets of X Y
such that (i) holds.
5.1.3. Proposition:
Let X and Y be nonempty sets and let c, T be families of subsets of X and
Y, respectively. Let c T := C D[ C c, D T. Then the following
hold:
(i) o(c T) o(c) o(T).
(ii) Let c and T have the property that there exist increasing sequences
C
i
i1
and D
i
i1
in c and T respectively such that
_
i=1
C
i
= X and
_
i=1
D
i
= Y .
Then
o(c T) = o(c) o(T).
Exercises:
(5.1) Let (X, /) be a measurable space. Let , R and E /B
R
.
Show that (x, t) X R[ (x, t +) E /B
R
.
(Hint: Use the -algebra technique.)
(5.2) Let E B
R
. Show that (x, y) R
2
[ x + y E and (x, y)
R
2
[ x y E are elements of B
R
B
R
.
(5.3) Let X and Y be nonempty sets and c, T be nonempty families of
subsets of X and Y , respectively, as in proposition 5.1.3. Is it true
that o(cT) = o(c)o(T) in general? Check in the case when
5.2. Product of measure spaces 65
c = and T is a -algebra of subsets of Y containing at least
four elements.
(5.4) Let B
R
2 denote the -algebra of Borel subsets of R
2
, i.e., the -
algebra generated by the open subsets of R
2
. Show that
B
R
2 = B
R
B
R
.
(Hint: Use proposition 5.1.3.)
5.2. Product of measure spaces
Concepts and examples:
For the rest of the chapter, let (X, /, ) and (Y, B, ) be xed measure
spaces. Subsets of X Y of the form A B, A /, B B, are called
measurable rectangles. As before, let 1denote the class of all measurable
rectangles. The -algebra of subsets of X Y generated by 1, denoted by
/B, is called the product -algebra.
The problem we want to analyze in this section is the following: how
can we construct a measure : / B [0, +] such that (A B) =
(A)(B) for every A /, B B? The answer is given by the next theorem.
5.2.1. Theorem:
Let : 1 [0, ] be dened by
(AB) := (A)(B), A /, B B.
Then is a well-dened measure on 1. Further, if , are -nite, then
there exists a unique measure : /B [0, +] such that
(AB) = (AB) for every AB 1.
5.2.2. Denition:
The measure on / B given by theorem 5.2.1 is called the product
of the measures and and is denoted by . The measure space
(X Y, / B, ) is called the product of the measure space (X, /, )
and (Y, B, ), or just the product measure space.
We note that is uniquely dened on /B when , are -nite.
So from now on we shall assume that and are -nite. As is clear,
is dened on /B via the extension theory . So, the natural question arises:
how to compute ( )(E) for a general element E /B?
66 5. Measure and integration on product spaces
5.2.3. Denition:
Let E X Y, x X and y Y . Let
E
x
:= y Y [ (x, y) E and E
y
:= x X[ (x, y) E.
The set E
x
is called the section of E at x (or x-section of E) and the
set E
y
is called the section of E at y (or y-section of E).
5.2.4. Examples:
(i) Let E = A B, where A / and B B. Clearly, E
x
= B if x A and
E
x
= if x , A. Similarly, E
y
= A if y B and E
y
= if y , B.
(ii) Let (X, /) be a measurable space and let A /. Let
E = (x, t) X R [ 0 t <
A
(x).
It is easy to see that
E = (A[0, 1)) (A
c
0).
Thus
E
x
=
_
[0, 1) if x A,
0 if x , A.
Similarly,
E
y
=
_
_
_
X if y = 0,
A if y (0, 1),
if y , [0, 1).
5.2.5. Proposition:
For E, F, E
i
/B and i I, any indexing set, the following hold x
X, y Y :
(i) (
iI
E
i
)
x
=
iI
(E
i
)
x
and (
iI
E
i
)
y
=
iI
(E
i
)
y
.
(ii) (
iI
E
i
)
x
=
iI
(E
i
)
x
and (
iI
E
i
)
y
=
iI
(E
i
)
y
.
(iii) (E F)
x
= E
x
F
x
and (E F)
y
= E
y
F
y
.
(iv) If E F, then E
x
F
x
and E
y
F
y
.
5.2.6. Theorem:
Let E /B. Then the following hold:
(i) E
x
B and E
y
/ for every x X, y Y.
(ii) The functions x (E
x
) and y (E
y
) are measurable func-
tions on X and Y, respectively.
(iii)
_
X
(E
x
)d(x) = ( )(E) =
_
Y
(E
y
)d(y).
5.2. Product of measure spaces 67
5.2.7. Remark:
Given -nite measure spaces (X, /, ) and (Y, B, ), we showed that is
the unique measure on /B such that ()(AB) = (A)(B). Note that
the measure space (XY, /B, ) need not be a complete measure space
even if the measure spaces (X, /, ) and (Y, B, ) are complete. For example,
if A X, A , / and , = B B with (B) = 0, then ( )
(A B) = 0,
but A B , / B. In fact, theorem 5.2.1 itself gives a complete measure
space (X Y, /B, ), where /B is the -algebra of
-measurable
subsets of X Y , being as in theorem 5.2.1, and is the restriction
of
n1
is an increasing sequence of nonnegative functions
on X increasing to f(x), show that E
(f
n
)
n1
is an increas-
ing sequence of sets in /B
R
with
_
n=1
E
(f
n
) = E
(f).
(ii) Using (i) and the simple function technique, show that E
(f)
/ B
R
, whenever f : X R is a nonnegative measurable
function. Deduce that E
(f) /B
R
.
(Hint: E
(f) =
n=1
E
(f + 1/n).)
(iii) Let f : X R be a nonnegative function such that E
(f)
/B
R
. Using exercise (5.1) and the following equality
A := (x, t) X R [ f(x) > c, t > 0
=
_
n=1
(x, t) X R [ (x, t/n +c) E
(f), t > 0,
show that A /B
R
, and deduce that f is measurable.
68 5. Measure and integration on product spaces
(iv) Let f : X R be any measurable function. Show that
G(f) /B
R
, where
G(f) := (x, t) X R [ f(x) = t.
The set G(f) is called the graph of the function.
(v) Let f L
1
(X, /, ), where (X) < +. Show that
( )(E
[f[) =
_
[f[d = ( )(E
[f[). (5.1)
(Hint: First prove this for the case when f is bounded. For the
general case, consider f
n
= [f[ n and note that E
(f
n
)
n1
increases to E
[f[.)
(vi) Extend (iv) to the case when is -nite.
The identity (5.1) shows that for nonnegative functions
_
fd
represents the area below the curve y = f(x).
(5.6) Let X be any nonempty set and T(X) be its power set. Let / :=
T(X) T(X) and let D := (x, y) X X[ x = y. Suppose
D /. Prove the following statements:
(i) There exist sets A
i
, B
i
T(X) such that D belongs to the
-algebra generated by (A
i
B
i
), i = 1, 2, . . . .
(Hint: See exercise (1.19).)
(ii) Let B = o(A
i
[ i = 1, 2, . . .). Then card (B) c. For every
x, y X, D
x
,= D
y
if x ,= y and D
x
B x X. Deduce
that card (X) c.
(iii) If card (X) > c, then D , T(X) T(X), even though D
x
T(X) and D
y
T(X), x, y X. Hence in general,
T(X) T(X) ,= T(X X).
(5.7) Let E /B. Then the functions x (E
x
) and y (E
y
)
are measurable and
_
X
(E
x
)d(x) = ( )(E) =
_
Y
(E
y
)d(y).
(Hint: E = F N, F / B and ( )
(N) = 0 by theorem
3.4.7.)
(5.8) Let E /B be such that (E
y
) = 0 for a.e. ()y Y . Show that
(E
x
) = 0 for a.e. ()x X. What can you say about ()(E)?
5.3. Integration on product spaces: Fubinis theorems 69
5.3. Integration on product spaces: Fubinis
theorems
Concepts and examples:
Let (X, /, ) and (Y, B, ) be -nite measure spaces and (XY, /B, )
the product measure space. Theorem 5.2.6 can be interpreted as follows: for
every E /B,
_
XY
E
(x, y)d( )(x, y) =
_
X
__
Y
E
(x, y)d(y)
_
d(x)
=
_
Y
__
X
E
(x, y)d(x)
_
d(y).
This allows us to compute the integral of the function
E
(x, y) by integrating
one variable at a time. So, the natural question arises: does the above hold
when
E
is replaced by a nonnegative measurable function on X Y ? The
answer is yes, and is made precise in the next theorem.
5.3.1. Theorem (Fubini):
Let f : X Y R be a nonnegative / B-measurable function. Then
the following statements hold:
(i) For x
0
X and y
0
Y xed, the functions x f(x, y
0
) and
y f(x
0
, y) are measurable on X and Y, respectively.
(ii) The functions y
_
X
f(x, y)d(x) and x
_
Y
f(x, y)d(y) are
well-dened nonnegative measurable functions on Y and X, respec-
tively.
(iii)
_
X
__
Y
f(x, y)d(y)
_
d(x) =
_
Y
__
X
f(x, y)d(x)
_
d(y)
=
_
XY
f(x, y)d( )(x, y).
In view of theorem 5.3.1, it is natural to expect a similar result for
f L
1
( ), and being -nite measures. This is given by the
following theorem.
5.3.2. Theorem (Fubini):
Let f L
1
( ). Then the following statements are true:
(i) The functions x f(x, y) and y f(x, y) are integrable for
a.e. y() and for a.e. x(), respectively.
70 5. Measure and integration on product spaces
(ii) The functions
y
_
X
f(x, y)d(x) and x
_
Y
f(x, y)d(y)
are dened for a.e. y() and a.e. x() , and are , -integrable,
respectively.
(iii)
_
Y
__
X
f(x, y)d(x)
_
d(y) =
_
XY
f(x, y) d( )
=
_
X
__
Y
f(x, y)d(y)
_
d(x).
Theorem 5.3.1, exercise (5.9) and theorem 5.3.2 together give us the
following theorem, which enables us to check the integrability of a function
of two variables and compute its integral.
5.3.4. Theorem:
Let (X, /, ) and (X, B, ) be -nite measure spaces. Let f : X Y R
be an /B-measurable function such that f satises any one of the following
statements:
(i) f is nonnegative.
(ii) f L
1
( ).
(iii)
_
X
__
Y
[f(x, y)[d(y)
_
d(x) < +.
(iv)
_
Y
__
X
[f(x, y)[d(x)
_
d(y) < +.
Then
_
XY
f(x, y)d( ) =
_
X
__
Y
f(x, y)d(y)
_
d(x)
=
_
Y
__
X
f(x, y)d(x)
_
d(y),
in the sense that all the integrals exist and are equal.
We give next some examples which illustrate the necessity of the condi-
tions on , and f for the conclusions of theorem 5.3.4 to hold.
5.3.5. Example:
Let X = Y = [0, 1] and / = B = B
[0,1]
, the -algebra of Borel subsets of
[0, 1]. Let be the Lebesgue measure on / and be the counting measure
on B, i.e., (E) := number of elements in E if E is nite and (E) := +
5.3. Integration on product spaces: Fubinis theorems 71
otherwise. Let D := (x, y) [ x = y. Let, for n 1,
D
n
:=
n
_
j=1
( [(j 1)/n, j/n] [(j 1)/n, j/n] ).
Then D =
n=1
D
n
. Thus D / B. In fact, D is a closed subset of
[0, 1] [0, 1]. Further,
_
X
D
(x, y)d(x) = 0 y Y and
_
Y
D
(x, y)d(y) = 1 x X.
Hence
_
Y
__
X
D
(x, y)d(x)
_
d(y) = 0 and
_
X
__
Y
D
(x, y)d(y)
_
d(x) = 1.
This does not contradict theorem 5.3.1, since is not -nite.
5.3.6. Example:
Let X = Y = [0, 1], / = B = B
[0,1]
, and let = be the Lebesgue measure
on [0,1]. Let
f(x, y) :=
_
_
_
x
2
y
2
(x
2
+y
2
)
2
if (x, y) ,= (0, 0),
0 if x = y otherwise.
Noting that for xed x, f(x, y) is a Riemann integrable function on [0, 1] and
y
_
y
x
2
+y
2
_
= f(x, y),
it is easy to see that
_
1
0
__
1
0
f(x, y)d(x)
_
d(y) =
_
1
0
__
1
0
f(x, y)d(y)
_
d(x) = /4.
This does not contradict theorem 5.3.2, because f , L
1
(X Y ) as
_
[0,1][0,1]
[f(x, y)[d( ) =
_
1
0
__
1
0
[f(x, y)[d(y)
_
d(x)
_
1
0
__
x
0
[f(x, y)[d(y)
_
d(x)
=
_
1
0
1
x
_
_
/4
0
cos 2d
_
d(x)
=
_
1
0
1
2x
d(x) = +.
72 5. Measure and integration on product spaces
*5.3.7. Example:
Assume the continuum hypothesis and let < be the well-order on [0, 1] such
that x [0, 1], y [0, 1] [ y < x is at most a countable set. Let
E := (x, y) [0, 1] [0, 1] [ x < y.
Then y xed and x [0, 1],
E
(x, y) = 1 if x < y and 0 otherwise.
Hence
E
(x, y) is the indicator function of a countable set for y [0, 1]
xed. Thus it is a Borel measurable function. Similarly, for x [0, 1] xed,
y
E
(x, y) is Borel measurable, for it is the indicator function of [0, 1]C,
a countable set. However,
E
is not Borel measurable on [0, 1][0, 1]. To see
this, note that B
[0,1][0,1]
= B
[0,1]
B
[0,1]
. Thus if
E
were Borel measurable,
then by theorem 5.3.1 we should have
_
[0,1][0,1]
E
(x, y)
d( )(x, y)
=
_
[0,1]
_
_
[0,1]
E
(x, y)d(x)
_
d(y)
=
_
[0,1]
_
_
[0,1]
E
(x, y)d(y)
_
d(x). (5.2)
However, it is easy to see that for x, y [0, 1],
_
[0,1]
E
(x, y)d(y) = 1 and
_
[0,1]
E
(x, y)d(x) = 0.
This contradicts (5.2). Hence
E
is not Borel measurable.
Exercises:
(5.9) Let f : XY R be /B-measurable. Show that the following
statements are equivalent:
(i) f L
1
( ) := L
1
(X Y, /B, ).
(ii)
_
Y
__
X
[f(x, y)[ d(x)
_
d(y) < +.
(iii)
_
X
__
Y
[f(x, y)[d(y)
_
d(x) < +.
(5.10) Let (X, /, ) and (Y, B, ) be complete -nite measure spaces and
let (XY, /B, ) be the completion of (XY, /B, ).
Let f : X Y R be any nonnegative extended real valued
/B-measurable function. Show that:
(i) The function x f(x, y) is /-measurable for a.e. y() and
the function y f(x, y) is B-measurable for a.e. x().
5.3. Integration on product spaces: Fubinis theorems 73
(ii) The function y
_
X
f(x, y)d(x) is B-measurable and the
function x
_
Y
f(x, y)d(y) is /-measurable.
(iii)
_
XY
f(x, y)d( ) =
_
X
__
Y
f(x, y)d(y)
_
d(x)
=
_
Y
__
X
f(x, y)d(x)
_
d(y).
(5.11) Let X = Y = [1, 1], / = B = B
[1,1]
, and let = be the
Lebesgue measure on [1, 1]. Let
f(x, y) :=
_
xy
(x
2
+y
2
)
2
if (x, y) ,= (0, 0),
0 otherwise.
Show that
_
1
1
__
1
1
f(x, y)d(y)
_
d(x) = 0 =
_
1
1
__
1
1
f(x, y)d(x)
_
d(y).
Can you conclude that
_
1
1
__
1
1
f(x, y)d(y)
_
d(x) =
_
XY
f(x, y)d( )(x, y)?
(5.12) Let f L
1
(X, /, ) and g L
1
(Y, B, ). Let
(x, y) := f(x)g(y), x X and y Y.
Show that L
1
(X Y, /B, ) and
_
XY
(x, y)d( ) =
__
X
fd
___
Y
gd
_
.
(5.13) Let f L
1
(0, a) and let
g(x) :=
_
a
x
(f(t)/t)d(t),0 < x a.
Show that g L
1
(0, a), and compute
_
a
0
g(x)d(x).
*(5.14) Let (X, /, ), and (X, B, ) be as in example 5.3.7. Dene, for
x, y [0, 1],
f(x, y) :=
_
1 if x is rational,
2y if y is irrational.
Compute
_
1
0
__
1
0
f(x, y)d(y)
_
d(x) and
_
1
0
__
1
0
f(x, y)d(x)
_
d(y).
Is f in L
1
( )?
74 5. Measure and integration on product spaces
(5.15) Let (X, /, ) be as in example 5.3.7. Let Y = [1, ), B = /
R
[1, ), and let be the Lebesgue measure restricted to [1, ).
Dene, for (x, y) X Y,
f(x, y) := e
xy
2e
2xy
.
Show that f , L
1
( ).
(5.16) Let X be a topological space and let B
X
be the -algebra of Borel
subsets of X. A function f : X R is said to be Borel measur-
able if f
1
(E) B
X
E B
R
. Prove the following:
(i) f is Borel measurable i f
1
(U) B
X
for every open set
U R.
(Hint: Use the -algebra technique.)
(ii) Let f : X R be continuous. Show that f is Borel measur-
able.
(iii) Let f
n
n1
be a sequence of Borel measurable functions on
X such that f(x) := lim
n
f
n
(x) exists x X. Show that f
also is Borel measurable.
(iv) Consider R
2
with the product topology and let f, g be Borel
measurable functions on R. Show that the function on R
2
dened by
(x, y) := f(x)g(y), x X, y Y,
is Borel measurable.
(5.17) Let f : R
2
R be Borel measurable. Show that for x X xed,
y f(x, y) is a Borel measurable function on R. Is the function
x f(x, y), for y Y xed, also Borel measurable?
(5.18) Let f : R
2
R be such that for x X xed, y f(x, y) is
Borel measurable and for y Y xed, x f(x, y) is continuous.
(i) For every n 1 and x, y R, dene
f
n
(x, y) := (i nx)f((i 1)/n, y) + (nx i + 1)f(i/n, y),
whenever x [(i 1)/n, i/n), i Z. Show that each f
n
: R
2
1
2
:= I J [ I, J
1. Then the following hold:
(i)
1
2
is a semi-algebra of subsets of R
2
, and o(
1
2
) = B
R
2 .
(ii)
R
2(I J) = (I)(J), I, J
1.
(iii) The measure space (R
2
, /
R
2,
R
2) is the completion of the measure
spaces (R
2
, /
R
/
R
, ) and (R
2
, B
R
2,
R
2).
We describe next some properties of the Lebesgue measure
R
2. For
E R
2
and x R
2
, let E +x := y +x [ y E.
5.4.2. Theorem:
The Lebesgue measure
R
2 has the following properties:
(i) Let E B
R
2 and x R
2
. Then E +x B
R
2 and
R
2(E) =
R
2(E +x).
(This property of
R
2 is called translation invariance.)
(ii) For every nonnegative Borel measurable function f on R
2
and y
R
2
,
_
f(x +y) d
R
2(x) =
_
f(x) d
R
2(x) =
_
f(x) d
R
2(x).
76 5. Measure and integration on product spaces
(iii) Let be any -nite measure on B
R
2 such that (E + x) = (E)
E B
R
2, x R
2
. Suppose
0 < (E
0
) = C
R
2(E
0
) < +,
for some E
0
B and for some C 0. Then
(E) = C
R
2(E), E B
R
.
5.4.3. Theorem:
Let T : R
2
R
2
be a linear transformation and E /
R
2. Then T(E)
/
R
2 and
R
2(T(E)) = [ det T[
R
2(E).
5.4.4. Theorem (Integration of radial functions):
Let f : [0, ) (0, ) be a nonnegative measurable function. Then
_
R
2
f([x[)d
R
2(x) = 2
_
0
f(r)rd(r).
Proof: The proof is once again an application of the simple function tech-
nique. We outline the steps
Step 1: The theorem holds for f =
(a,b)
, 0 a < b < +.
Step 2: Let E
n
n1
be a sequence of sets from [0, ) /
R
such that
either the E
n
s are pairwise disjoint, or the E
n
s are increasing. If the
theorem holds for each
E
n
, then the theorem holds for
E
also, where
E =
n=1
E
n
.
Step 3: The theorem holds for f =
U
, U being any open subset of [0, ).
Step 4: The theorem holds for f =
N
, where N [0, ) and (N) = 0.
Step 5: The theorem holds when f =
E
, E /
R
and E [0, ).
Step 6: The theorem holds for any nonnegative measurable function.
Exercises:
(5.19) (Regularity of
R
2): Prove the following:
(i)
R
2(U) > 0 for every nonempty open subset U of R
2
.
(ii) A set E /
R
2 i > 0, there exists an open set U such
that E U and (U E) < .
(Hint: Proceed as in theorem 3.5.2.)
(iii)
R
2(K) < + for every compact subset K of R.
5.4. Lebesgue measure on R
2
and its properties 77
(iv)
R
2(U) = sup
R
2(K) [ K compact , K U.
(Hint: Given U open, there exists a sequence of compact sets
K
n
such that K
n
U.)
(v) If E B
R
2 is such that
R
2(E) < +, then
R
2(E) = sup
R
2(k) [ K E, K compact.
(5.19) Show that for f L
1
(R
2
, /
R
2,
R
2), x R
2
, the function y
f(x +y) is integrable and
_
f(x +y)d
R
2(x) =
_
f(x)d
R
2(x).
(Hint: Use exercise (4.30) and theorem 5.4.2.)
(5.20) Let E /
R
2 and x = (x, y) R
2
. Let
xE := (xt, yr) [ (t, r) E.
Prove the following:
(i) xE /
R
2 for every x R, E /
R
2, and
R
2(xE) = [xy[
R
2(E).
(ii) For every nonnegative Borel measurable function f : R
2
R,
_
f(xt)d
R
2(t) = [xy[
_
f(t)d
R
2(t),
where for x = (x, y) and t = (s, r), xt := (xs, yr).
(iii) Let
R
2x R
2
[ [x[ 1 =: . Then
R
2x R
2
[ [x[ < 1 = and
R
2x R
2
[ [x[ < r = r
2
.
(iv) Let E be a vector subspace of R
2
. Then
R
2(E) = 0 if E has
dimension less than 2.
(5.21) Let T : R
2
R be a linear map.
(i) If N R
2
is such that
R
2
(N) = 0, show that
R
2
(T(N)) = 0.
(ii) Use (i) above and proposition 5.4.1(ii) to complete the proof
of theorem 5.4.3 for sets E /
R
2.
(5.14) Consider the vectors (a
1
, b
1
), (a
2
, b
2
) R
2
and let
P := (
1
a
1
+
2
a
2
,
1
b
1
+
2
b
2
) R
2
[
1
,
2
R, 0
i
1,
called the parallelogram determined by these vectors. Show that
R
2(P) = [a
1
b
2
a
2
b
1
[.
(5.15) Let E /
R
2 with 0 < (E) < . Show that there exists a unique
point (c
1
, c
2
) R
2
such that
_
x
E
(x +c
1
, y +c
2
)d
R
2(x, y) = 0
and
_
y
E
(x +c
1
, y +c
2
)d
R
2(x, y) = 0.
78 5. Measure and integration on product spaces
In fact,
c
1
=
1
(E)
_
x
E
(x, y)d
R
2(x, y)
and
c
2
=
1
(E)
_
y
E
(x, y)d
R
2(x, y).
(c = (c
1
, c
2
) is called the centroid of E.)
(5.16) Let f : [0, ) R be a measurable function. Show that, if either
of the two integrals in theorem 5.4.4 exists, then so does the other
and the equality holds.
(5.17) Let f : (a, b) (c, d) R be such that
(i) f is continuous;
(ii)
f
x
exists and is continuous;
(iii) for some t (a, b),
d
dy
f(t, y) exists y;
(iv)
y
_
f
x
_
exists and is continuous.
Show that
f
y
and
x
_
f
y
_
exist. Moreover,
x
_
f
y
_
=
y
_
f
x
_
.
(Hint: Use Fubinis theorem and the fundamental theorem of cal-
culus to show that (a, b), (c, d) and s < y
f(, ) f(t, ) f(, s) +f(t, s) =
_
s
_
t
y
_
t
x
_
dxdy,
and note that the inner integral on the right is a continuous function
of .)
Chapter 6
L
p
-spaces
Throughout this chapter, we shall work with a xed -nite measure space
(X, o, ), which is assumed to be complete.
The main aim of this chapter is to analyze the convergence of sequences
of measurable functions. Given a sequence f
n
n1
of measurable functions
on (X, o, ), we have already come across concepts like pointwise conver-
gence of f
n
n1
to a measurable function f, i.e., when f
n
(x) converges
to f(x) x X. When a sequence does not converge pointwise, one would
like to nd other methods of analyzing the behavior of the sequence f
n
n1
for large n. Analysis of these methods, and the relations between them, is
the main aim of this chapter. But before we do that, we extend the notion
of measurability and integration to complex-valued functions.
6.1. Integration of complex-valued functions
Concepts and examples:
Let (X, o, ) be a measure space and let C denote the eld of complex
numbers. For a function f : X C, consider the functions Re (f) and
Im(f) dened by: x X,
Re (f)(x) := Real part of f(x)
and
Im(f)(x) := Imaginary part of f(x).
The functions Re (f) and Im(f) are called, respectively, the real part and
the imaginary part of the function f. Note that Re (f) and Im(f) are
real-valued functions on X.
79
80 6. L
p
-spaces
6.1.1. Denition:
A complex-valued function f : X C is said to be measurable if both
Re (f) and Im(f) are measurable functions. We say f is -integrable if
both Re (f) and Im(f) are integrable. In that case we dene the integral
of f, denoted by
_
fd, to be
_
fd :=
_
Re (f)d +i
_
Im(f)d.
We denote the set of all complex-valued -integrable functions on X
by L
1
(X, o, ) itself. Whenever we restrict ourselves to only real-valued
-integrable functions on X, we shall specify it by L
r
1
(X, o, ). Our next
theorem tells us that L
1
(X, o, ) is as nice a space as L
r
1
(X, o, ) is.
6.1.2. Theorem:
(i) Let f, g L
1
(X, o, ) and , C. Then f + g L
1
(X, o, )
and
_
(f +g)d =
_
fd +
_
gd.
(ii) Let f : X C be a measurable function. Then f L
1
(X, o, )
i [f[ L
r
1
(X, o, ). Further, in either case,
_
fd
_
[f[d.
(iii) Let f L
1
(X, o, ) and E o. Then
E
f L
1
(X, o, ). We write
_
E
fd :=
_
E
fd.
If E
1
, E
2
o and E
1
E
2
= , then
_
E
1
E
2
fd =
_
E
1
fd +
_
E
2
fd.
(iv) Let f L
1
(X, o, ), and let E
n
n1
be a sequence of pairwise
disjoint sets from o. Then the series
n=1
(
_
E
n
fd) is absolutely
convergent. Also, (
E
f) L
1
(X, o, ), where E :=
n=1
E
n
, and
_
E
fd =
n=1
_
E
n
fd.
That the integral of complex-valued functions behaves as nicely with
respect to limiting operations as the integral of real-valued functions is the
content of the next theorem.
6.1. Integration of complex-valued functions 81
6.1.3. Theorem (Complex form of Lebesgues dominated conver-
gence):
Let f
n
n1
be a sequence in L
1
(X, o, ). Let f(x) := lim
n
f
n
(x) exist for
a.e. x X and let there exist a function g L
r
1
(X, o, ) such that n,
[f
n
(x)[ g(x) for a.e. x(). Then f L
1
(X, o, ) and
lim
n
_
f
n
d =
_
f d.
6.1.4. Theorem:
Let f L
1
(X, o, ), where (X) < . Let S be a closed subset of C such
that E o with (E) > 0,
_
1
(E)
_
E
f d
_
S.
Then f(x) S for a.e. x X.
Exercises:
(6.1 ) Let f L
1
(X, o, ) be such that
_
f d
=
_
[f[ d.
What can you conclude about f?
(6.2) Let f : X C. Show that f is measurable i f
1
(E) o for every
Borel set E X.
(6.3) Extend the claim of exercise (4.30) to complex-valued measurable
functions.
(6.4) Let f
n
n1
be a sequence of o-measurable complex-valued func-
tions on (X, o, ). Show that the following statements are equiva-
lent:
(i) (
n=1
[f
n
[) L
r
1
(X, o, ).
(ii)
n=1
_
[f
n
[d < +.
Further, if either of the above is true, then (
n=1
f
n
) L
1
(X, o, )
and
_
_
n=1
f
n
_
d =
n=1
_
f
n
d.
(6.5) State and prove the Riesz-Fischer theorem (theorem 4.6.3) for complex-
valued integrable functions.
82 6. L
p
-spaces
(6.6) Let f L
1
(X, o, ) and
_
E
fd = 0 for every E o.
Show that f(x) = 0 for a.e. x() (see proposition 5.4.6).
6.2. L
p
-spaces
Concepts and examples:
In section 4.6, we analyzed L
1
[a, b], the space of Lebesgue integrable func-
tions on [a, b]. We saw that L
1
[a, b] is a vector space over R, and that for
f L
1
[a, b], we can dene the notion of absolute value of f, i.e., |f|
1
. Fur-
ther, this notion of absolute value can be used to dene the L
1
-metric on
L
1
[a, b] so that L
1
[a, b] becomes a complete metric space. In this section
we look at examples of a family of spaces of this type, called L
p
-spaces.
Throughout this section, we x a measure space (X, o, ) which is -nite
and complete.
6.2.1. Denition:
Let 0 < p < . Let L
p
() := L
p
(X, o, ) denote the space of all complex-
valued o-measurable functions on X such that
_
[f[
p
d < +.
The space L
p
() is called the space of p
th
-power integrable functions.
6.2.2. Proposition:
The space L
p
(X, o, ) is a vector space over C.
6.2.3. Denition:
Let f L
p
(X, B, ). Dene |f|
p
, called the p
th
-norm of f, as follows:
|f|
p
:=
__
[f[
p
d
_
1/p
.
Since |f|
p
= |g|
p
if f(x) = g(x) for a.e. x(), we treat such f and g as
the same element of L
p
(X, B, ). To show that |f|
p
has the properties of a
metric (as was the case for p = 1 in section 4.6), we need some inequalities.
6.2.4. Lemma:
For nonnegative real numbers a, b and 0 < t < 1, the following inequalities
hold:
(i) a
t
b
1t
ta + (1 t)b.
6.2. L
p
-spaces 83
(ii) (
a +b
2
)
1/t
1
2
(a
1/t
+b
1/t
).
6.2.5. Theorem (H olders inequality):
Let p > 1 and q > 1 be such that 1/p + 1/q = 1. Let f L
p
() and
g L
q
(). Then fg L
1
() and
_
[fg[d
__
[f[
p
d
_
1/p
__
[g[
q
d
_
1/q
.
6.2.6. Corollary:
Let p, q > 1 be real numbers with 1/p + 1/q = 1. Let a
n
n1
and b
n
n1
be sequences of complex numbers such that
n=1
[a
n
[
p
< and
n=1
[b
n
[
q
< .
Then
n=1
[a
n
b
n
[ < + and
_
n=1
[a
n
b
n
[
_
_
n=1
[a
n
[
p
_
1/p
_
n=1
[b
n
[
q
_
1/q
.
In particular, if a
n
= b
n
= 0 n k, then
k
n=1
[a
n
b
n
[
_
k
n=1
[a
n
[
p
_
1/p
_
k
n=1
[b
n
[
q
_
1/q
.
6.2.7. Note:
In the special case when p = q = 2, H olders inequality is known as the
Cauchy-Schwarz inequality.
6.2.8. Theorem (Minkowskis inequality):
Let 1 p < and f, g L
p
(). Then f +g L
p
() and
|f +g|
p
|f|
p
+|g|
p
.
6.2.9. Theorem (Riesz-Fischer):
The space L
p
(X, o, ), for 1 p < , is a complete metric space with the
metric dened by
d
p
(f, g) := |f g|
p
f, g L
p
(X, o, ).
84 6. L
p
-spaces
Of course, here we identify f, g L
p
(X, o, ) if f(x) = g(x) for a.e.
x().
6.2.10. Note:
In theorem 6.2.9, we showed that L
p
(), 1 p < , is a complete metric
space under the metric
d
p
(f, g) := |f g|
p
:=
__
[f g[
p
d
_
1/p
.
It is natural to ask the same question for 0 < p < 1. If 0 < p < 1, dene for
f, g L
p
()
d
p
(f, g) :=
_
[f g[
p
d.
Using the inequality (see exercise (6.7))
1 +t
p
(1 +t)
p
t 0,
it is easy to see that d
p
is a metric on L
p
(). Also, proceeding as in theorem
6.2.9, we can show that L
p
() is a complete metric space for 0 < p < 1.
In the case 1 p < , f |f|
p
is a real-valued map on L
p
() with the
properties: f, g L
p
() and C
(i) |f|
p
0, and |f|
p
= 0 i f = 0.
(ii) |f|
p
= [[|f|
p
.
(iii) |f +g|
p
|f|
p
+|g|
p
, called the triangle inequality.
Such a function can be dened on any vector space, and is called a norm.
A vector space with a norm is called a normed linear space. Thus for
1 p < , the L
p
-spaces are examples of normed linear spaces which
are also complete under the metric |f g|
p
, the metric induced by the
norm. Such normed linear spaces are called Banach spaces. However,
when 0 < p < 1, f |f|
p
is no longer a norm, as it fails to satisfy the
triangle inequality. To see this, consider a measure space (X, o, ) such that
A, B o with A B = and 0 < (A) < , 0 < (B) < . Let , be
positive real numbers. Then
|
A
|
p
=
__
[[
p
A
(x)d(x)
_
1/p
= ((A))
1/p
.
6.2. L
p
-spaces 85
Similarly, |
B
|
p
= ((B))
1/p
. Further,
|
A
+
B
|
p
=
_ _
(
A
+
B
)
p
d
_
1/p
=
_ _
(
p
A
+
p
B
) d
_
1/p
= (
p
(A) +
p
(B))
1/p
.
Now, using exercise 6.7(ii) with t = (/)
p
((B)/(A)), we have
_
1 +
_
_
p
(B)
(A)
_
1/p
> 1 +
_
__
(B)
(A)
_
1/p
,
i.e.,
(
1/p
(A) +
1/p
(B))
1/p
> ((A))
1/p
+((B))
1/p
,
i.e.,
|
A
+
B
|
p
> |
A
|
p
+|
B
|
p
.
This is the reason that L
p
-spaces for 0 < p < 1 are not very interesting to
study.
Exercises:
(6.7) Let 0 < t < and 0 < p < 1. Show that
(i) (1 +t)
p
< 1 +t
p
.
(ii) (1 +t)
1/p
> 1 +t
1/p
.
(6.8) Let 0 < p < 1 and < q < 0 be such that 1/p + 1/q = 1. Let
f, g be positive functions such that f L
p
() and g L
q
(). Show
that
|f|
p
|g|
q
<
_
fgd.
(Hint: Apply H olders inequality to (fg)
p
L
1/p
and g
p
L
p/q
.)
(6.9) If (X) < and 1 p q < , show that L
q
() L
p
().
(Hint: Apply H olders inequality to [f[
p
L
q/p
() and 1 L
(qp)/q
().)
(6.10) Let f L
p
(X, o, ), where 1 p < . Show that > 0,
(x X[ [f(x)[ 0) |f|
p
/
p
.
This is called Chebyshevs inequality for L
p
-functions.
86 6. L
p
-spaces
6.3. L
(X, o, )
Concepts and examples:
6.3.1. Denition:
A measurable function f : X R
(X, o, ) dene
|f|
(X, o, ) and
|f|
= [[ |f|
.
(ii) (f +g) L
(X, B, ) and
|f +g|
|f|
+|g|
.
(iii) |f|
= sup
xX\E
[f(x)[.
Thus (x X[ [f(x)[ > |f|
) = 0.
(v) |f|
(X, o, ), let
d
(f, g) := |f g|
.
Then for f, g, h L
(f, g) d
(f, h) +d
(h, g).
(iii) If f
n
n1
is a sequence in L
(X, o, ), then d
(f
n
, f) 0 as
n i f
n
f uniformly a.e.
(iv) If f
n
n1
is a Cauchy sequence in L
(X, o, ), i.e., d
(f
n
, f
m
)
0 as n, m , then h L
(f
n
, h)
0 as n .
6.4. L
2
(X, o, ) 87
6.3.4. Remark:
For f, g L
is well-dened, and L
(X, o, ) is a Banach
space under this norm.
Exercises:
(6.11 ) Let f L
((X))
1/p
.
(ii) If N > 0 is such that (x X[ [f(x)[ > N) > 0, show that
N((X))
1/p
|f|
p
.
(iii) Let p
n
n1
be any monotonically increasing sequence of real
numbers such that p
n
1 n and p
n
n1
is not bounded
above. Using (i), (ii) above and theorem 6.3.2(v), show that
|f|
= lim
n
|f|
p
n
.
6.4. L
2
(X, o, )
Concepts and examples:
The reason for analyzing the space L
2
(X, o, ) in detail is the following.
We have already seen that for 1 p , the spaces L
p
(X, o, ) are linear
spaces and have a notion of distance (norm) under which they are complete.
Let us recall, for f L
2
(X, o, ), that
|f|
2
:=
__
[f(x)[
2
d(x)
_
1/2
.
If we treat f L
2
(X, o, ) as a vector with uncountable components, f(x)
being the x
th
component, x X, then |f|
2
can be viewed as a generalization
of the Euclidean magnitude of a vector in R
n
with summation being replaced
by integration. We recall that on R
n
, we have the notion of dot product of
vectors, which is related to the magnitude of vectors and helps us to dene
the notion of angles on R
n
. That such a dot product can also be dened on
L
2
(X, o, ) and enables one to do geometry on L
2
(X, o, ), is the reason we
discuss this space in detail.
88 6. L
p
-spaces
6.4.1. Denition:
For f, g L
2
(X, o, ), we dene
f, g) :=
_
f(x) g(x) d(x),
whenever it exists, where g denote the complex conjugate function: g(x) :=
g(x) x X.
6.4.2. Proposition (Cauchy-Schwarz inequality):
For every f, g L
2
(X, o, ), f, g) is a well-dened scalar and
[f, g)[ |f|
2
|g|
2
.
The scalar f, g) is called the inner product of f and g.
6.4.3. Proposition:
For f, g, h L
2
(X, o, ) and , C, the following hold:
(i) f, f) 0, and equality holds i f = 0.
(ii) f, g) = g, f).
(iii) f +g, h) = f, h) +g, h).
(iv) f, g +h) = f, g) +f, h).
(v) |f|
2
= f, f)
1/2
.
6.4.4.Note:
In the Cauchy-Schwarz inequality, the equality holds i f and g are linearly
dependent.
6.4.5. Note:
Given an arbitrary vector space H over the eld R (or C), if there is a map
, ) : H H R (or C) having the properties (i) to (iv) as given in
proposition 6.4.3, then it is called an inner product space. On every
inner product space H, it is easy to show that |u| := u, u)
1/2
, u H,
is indeed a norm on H, called the norm induced by the inner product.
Further, the Cauchy-Schwarz inequality holds:
[u, v)[ |u| |v|, u, v H.
This can be proved as follows: For u H, let |u| = (u, u))
1/2
. If either
u = 0 or v = 0, then clearly [u, v)[ = 0 = |u| |v|. So, let u, v H be
such that u ,= 0 and v ,= 0. Then |u| > 0 and |v| > 0. Let u
= u/|u| and
6.4. L
2
(X, o, ) 89
v
= v/|v|. Then u
, u
) = 1 = v
, v
) and
0 u
, v
)v
, u
, v
)v
)
= u
, u
) +[u
, v
)[
2
v
, v
) 2[u
, v
)[
2
= 1 [u
, v
)[
2
.
Hence
[u
, v
)[ 1 = |u
||v
|,
i.e.,
[u, v)[ |u||v|.
If H is also complete under the norm induced by the inner product, then
H is called a Hilbert space. Thus L
2
(X, o, ) is an example of a Hilbert
space. We shall not go into the general theory of Hilbert spaces. We discuss
some results for L
2
(X, o, ) which are in fact true, without any change in
the arguments, for general Hilbert spaces also.
6.4.6. Denition:
Let f, g L
2
(X, o, ). We say that f and g are orthogonal, and write
f g, if f, g) = 0. For a subset S of L
2
(X, B, ), we write f S if
f, h) = 0 h S.
6.4.7. Denition:
Let S be a nonempty subset of L
2
(X, o, ).
(i) We say S is a subspace of L
2
(X, o, ) if , C and f, g
L
2
(X, B, ), we have f +g S.
(ii) We say S is a closed subspace if it is closed under the ||
2
metric,
i.e., for every sequence f
n
n1
in S with lim
n
|f
n
f|
2
= 0 for
some f L
2
(X, o, ), we have f S.
6.4.8. Proposition:
Let S be any nonempty subset of L
2
(X, o, ) and let
S
:= g L
2
(X, o, ) [ f, g) = 0 f S.
Then S
is a closed subspace of L
2
(X, o, ).
The set S
,= 0.
Next we give two applications of theorem 6.4.9. Our rst application
says that if S is any closed subspace of L
2
(X, o, ), then L
2
(X, o, ) =
S +S
:= f +g[f S, g S
with S S
= 0.
6.4.12. Theorem:
Let S
1
, S
2
be subsets of L
2
(X, o, ). Then the following hold:
(i) S
1
is a closed subspace of L
2
(X, o, ) and S
1
S
1
0. If S
1
is
also a subspace, then S
1
S
1
= 0.
(ii) S
1
S
2
if S
2
S
1
.
(iii) S
1
(S
1
)
, with equality i S
1
is a closed subspace of L
2
(X, o, ).
(iv) If S
1
and S
2
are closed subspaces and f g f S
1
and g S
2
,
then S
1
+S
2
:= f +g [ f S
1
, g S
2
is also a closed subspace.
(v) If S
1
is a closed subspace, then S
1
S
1
= 0 and L
2
(X, o, ) =
S
1
+ S
1
. Thus every f L
2
(X, o, ) can be uniquely expressed as
f = g +h, where g S
1
and h S
1
.
(This is also expressed as L
2
(X, o, ) = S
1
S
1
, and is called the
projection theorem.)
As our second application of theorem 6.4.9, we characterize all bounded
linear functions on L
2
, as dened next.
6.4.12. Denition:
A map T : L
2
(X, o, ) C is called a bounded linear functional if it
has the following properties:
(i) For every f, g L
2
(X, o, ) and , C,
T(f +g) = T(f) +T(g).
(ii) There exists a real number M such that
[T(f)[ M|f|
2
, f L
2
(X, o, ).
6.4. L
2
(X, o, ) 91
6.4.13. Proposition:
Let T : L
2
(X, o, ) C be such that f, g L
2
(X, o, ) and , C
T(f +g) = T(f) +T(g).
Then the following are equivalent:
(i) T is bounded, i.e., M R such that
[T(f)[ M|f|
2
, f L
2
(X, o, ).
(ii) T is continuous.
(iii) T is continuous at 0 L
2
(X, o, ).
6.4.14. Example:
Let g L
2
(X, o, ) be xed. Dene the map
T
g
: L
2
(X, B, ) C
as follows:
T
g
(f) = f, g) =
_
f g d f L
2
(X, o, ).
It is easy to see that T
g
is linear, i.e.,
T
g
(f
1
+f
2
) = T
g
(f
1
) +T
g
(f
2
)
, C and f
1
, f
2
L
2
(X, o, ). Also, by the Cauchy-Schwarz inequal-
ity,
[T
g
(f)[ = [f, g)[ |g|
2
|f|
2
.
Hence T
g
is a bounded linear functional on L
2
(X, o, ). We note that g
(Ker(T
g
))
, where
Ker(T
g
) := f L
2
(X, B, ) [ T
g
(f) = 0.
Ker(T
g
) is called the kernel of T
g
. Our next theorem tell us that every
bounded linear functional on L
2
(X, o, ) is of this type.
6.4.15. Theorem (Riesz representation):
Let T : L
2
(X, B, ) C be a bounded linear functional. Then there is a
unique g
0
L
2
(X, o, ) such that
T(f) = f, g
0
) f L
2
(X, o, ).
6.4.16. Note:
In theorem 6.4.15, we showed that, given a closed subspace M of L
2
(X, o, ),
every element f L
2
(X, o, ) can be written uniquely as f = g + h, where
g M and h M
. Let us denote g by P
M
(f), called the projection of f
onto M. Geometrically, P
M
(f) is the unique best approximator of f in M.
92 6. L
p
-spaces
The properties of the map P
M
: L
2
(X, o, ) L
2
(X, o, ) are given in the
next theorem.
6.4.17. Theorem:
P
M
: L
2
(X, o, ) L
2
(X, o, ), as dened in note 6.4.16 above, has the
following properties: f, g L
2
(X, o, )
(i) P
M
is linear and P
M
(f) = f whenever f M.
(ii) f P
M
(f) M
.
(iii) P
M
(f), g) = f, P
M
(g)) = P
M
(f), P
M
(g)).
(iv) P
M
(P
M
(f)) = f.
(v) |f|
2
2
= |P
M
(f)|
2
2
+ |f P
M
(f)|
2
2
.
(vi) P
M
is continuous.
Exercises:
(6.12) Let f L
2
(X, o, ) be such that f g g L
2
(X, o, ). What
can you conclude about f?
(6.13) Pythagoras identity: Let f, g L
2
(X, o, ) and f g. Show
that
|f +g|
2
2
= |f|
2
2
+|g|
2
2
.
(6.14) (Bessels inequality):Let f
1
, f
2
, . . . , f
n
be elements of L
2
(X, o, )
such that |f
i
|
2
= 1 i and f
i
f
j
for i ,= j. Show that
n
j=1
[f, f
j
)[
2
|f|
2
2
.
(Hint: Consider z, z), where z = f
n
j=1
f, f
j
)f
j
.)
(6.15) Let f
n
n1
, g
n
n1
be sequences in L
2
(X, o, ) such that lim
n
|f
n
f|
2
= 0, lim
n
|g
n
g|
2
= 0 for f, g L
2
(X, o, ). Show that
f
n
, h) f, h), h, g
n
) h, g), f
n
, g
n
) f, g)
for every h L
2
(X, o, ). In other words, the inner product map
, ) : L
2
L
2
C is continuous in each variable and jointly.
(6.16) (Parallelogram identity): Let f, g L
2
(X, o, ). Then
|f +g|
2
2
+|f g|
2
2
= 2|f|
2
2
+ 2|g|
2
2
.
6.5. L
2
-convergence of Fourier series 93
6.5. L
2
-convergence of Fourier series
Concepts and examples:
The problem of investigating the convergence of Fourier series motivated
mathematicians to look for an extension of the integral concept. It is natu-
ral to ask the question: did the extended integral, i.e., the Lebesgue integral,
achieve success in this direction? One can say without any doubt that one
of the pioneering applications of the Lebesgue integral lies in the study of
Fourier series. We shall not go into the general theory of Fourier series,
but only prove an important result (the Riesz-Fischer theorem) which has
applications in many branches of mathematics, physics and electrical engi-
neering. For the general theory of Fourier series, one can consult Bhatia [4]
and other references given there.
6.5.1. Denition:
For f L
r
1
[, ], let
a
n
:=
1
n=1
(a
n
cos nx +b
n
sin nx)
is called the Fourier series of f. Let
s
n
(x) :=
a
0
2
+
n
k=1
(a
k
cos kx +b
n
sin kx), x [, ].
The function s
n
(x) is called the n
th
-partial sum of the Fourier series of f at
x. The main problem analyzed in the theory of Fourier series is the following:
when does s
n
(x)
n1
converge to f(x), x [, ]? This is known as the
pointwise convergence problem of Fourier series. For f L
1
[, ] and
x [, ], if lim
n
s
n
(x) = f(x), we say that f has pointwise representation
by its Fourier series at x. The answer to the pointwise convergence problem
is not easy, and, given the nature and scope of this text, we shall not go into
it. For details one can consult any one of the texts listed above. We state
below an important relation between functions and their Fourier coecients.
94 6. L
p
-spaces
6.5.2. Theorem (Uniqueness of the Fourier series):
If f L
r
1
[, ] is such that all its Fourier coecients are zero, then f(x) =
0 for a.e. x, i.e., a function is uniquely determined by its Fourier coecients.
Though the pointwise representation of a function f L
r
1
[, ] by its
Fourier series is undeniably of great intrinsic interest, it has its limitations.
First of all, not every function f L
r
1
[, ] can have a pointwise represen-
tation one has to put extra conditions on f to get a pointwise representation.
Secondly, f L
r
1
[, ] need be dened only a.e. Thus pointwise represen-
tation makes sense only a.e. Finally, from the point of view of many applica-
tions, pointwise representations have very little utility. Such consideration
motivated mathematicians to look for some other methods of analyzing the
convergence problem. We consider below one such method, namely the con-
vergence of Fourier series in the L
2
-metric. For other methods we refer to
the texts cited earlier.
Let f L
r
2
[, ]. Using the Cauchy-Schwarz inequality, it follows that
f L
r
1
[, ] also. Let a
n
, b
n
be its Fourier coecients and let s
n
(x) be the
n
th
-partial sum of the Fourier-series of f. In the terminology of the previous
section , if we write
k
(x) := cos kx and
k
(x) := sin kx,
then
a
k
= f,
k
)/ and b
k
= f,
k
)/.
6.5.3. Lemma:
For nonnegative integers n and m, the following relations hold:
n
,
m
) = 0;
n
,
m
) = 0 if n ,= m and = if n = m;
n
,
m
) = 0 if n ,= m and = if n = m.
6.5.4. Theorem (Bessels inequality):
For f L
r
2
[, ],
[a
0
[
2
2
+
n=1
([a
n
[
2
+[b
n
[
2
)
1
|f|
2
2
.
The above theorem has a converse, as given in the next theorem.
6.5. L
2
-convergence of Fourier series 95
6.5.5. Theorem (Riesz-Fischer):
Let a
n
n0
and b
n
n1
be sequences of real numbers such that
[a
0
[
2
2
+
n=1
([a
n
[
2
+[b
n
[
2
) < +.
Then there exists a unique function f L
r
2
[, ] such that a
n
, b
n
are its
Fourier coecients.
6.5.6. Corollary ( Parsevals identity):
Let f L
r
2
[, ]. Then the Fourier series of f converges to f in the L
2
-
norm, i.e., |s
n
f|
2
0 as n . Further,
1
|f|
2
2
=
[a
0
[
2
2
+
n=1
([a
n
[
2
+[b
n
[
2
).
6.5.7 Note:
A careful observation of the above arguments will tell the reader that Fourier
coecients can be dened for Riemann integrable functions also (as was done
historically rst), and Bessels inequality remains valid. However, Riesz-
Fischer critically uses the fact that L
r
2
[, ] is complete under the L
2
-
metric, which is not true for Riemann integrable functions.
96 6. L
p
-spaces
Exercises:
(6.17) (Eulers identity):
Show that
=0
1
(2 + 1)
2
=
2
8
.
Appendix A
Extended real numbers
A.1. Extended real numbers
Let R denote the set of all real numbers and let
R
:= R + ,
where + and are two symbols read as plus innity and minus
innity. We extend the algebraic operations and the order relation of R to
R
as follows:
(1) For every x R, < x < +.
(2) For every x R,
() +x = and (+) +x = +;
(+) + (+) = + and () + () = .
(3) For every x R,
x(+) = (+)x = +
x() = ()x =
_
if x > 0,
x(+) = (+)x =
x() = ()x = +
_
if x < 0.
Further,
(+)0 = ()0 = 0, ()(+) = () and ()() = ().
Note that the relations + (+) and (+) + () are not dened.
97
98 Appendix A
The set R
.
A.3. Limits of sequences in R
For x
n
n1
any monotonically increasing sequence in R
which is not
bounded above, we say x
n
n1
is convergent to + and write
lim
n
x
n
= +.
Similarly, if x
n
n1
is a monotonically decreasing sequence which is not
bounded below, we say x
n
n1
is convergent to and write
lim
n
x
n
= .
Hence
every monotone sequence in R
is convergent.
Thus for any sequence x
n
n1
in R
j1
and
inf
kj
x
k
j1
always converge. We write
limsup
n
x
n
:= lim
j
(sup
kj
x
k
)
and
liminf
n
x
n
:= lim
j
(inf
kj
x
k
).
lim
n
sup x
n
is called the limit superior of the sequence x
n
n1
and
liminf
n
x
n
is called the limit inferior of the sequence x
n
n1
. Note
that
liminf
n
x
n
limsup
n
x
n
.
We say a sequence x
n
n1
is convergent to x R
if liminf
n
x
n
=
limsup
n
x
n
=: x, say. In that case we write lim
n
x
n
:= x.
A.4. Series in R
Let x
k
k1
be a sequence in R
n
k=1
x
k
is
well-dened. We say that the series
k=1
x
k
is convergent to x if s
n
n1
is convergent. We write this as x =
k=1
x
k
, x being called the sum of
Extended real numbers 99
the series
k=1
x
k
. For example, if each x
k
0, then the series
k=1
x
k
is always convergent.
Let x
k
k1
be a sequence in R
, x
n
0 for every n. Let : N N be
any bijective map. Then the series
k=1
x
(k)
is called an arrangement
of the series
k=1
x
k
.
For every rearrangement , the series
k=1
x
(k)
and
k=1
x
k
con-
verge to the same sum.
Clearly, both the series are convergent in R
. Let :=
k=1
x
k
and :=
k=1
x
(k)
. To prove = , in view of the symmetry, it is enough to prove
that . Now
=
n
k=1
x
(k)
m
k=1
x
k
k=1
x
k
= ,
where in the the middle term m := max(1), , (k). Hence ,
proving the required claim.
Similar arguments apply to x
n,m
n,m1
, any double-indexed sequence of
nonnegative elements of R
m=1
x
n,m
is con-
vergent in R
. Let y
n
:=
m=1
x
n,m
. Further,
n=1
y
n
is also convergent in
R
. Let y :=
n=1
y
n
. We can also dene for all xed m, z
m
:=
n=1
x
n,m
,
and z :=
m=1
z
m
in R
m=1
_
n=1
x
n,m
_
=
n=1
_
m=1
x
n,m
_
.
To see this, we note that r, s N,
r
n=1
s
m=1
x
n,m
s
m=1
n=1
x
n,m
m=1
n=1
x
n,m
.
Hence
y :=
n=1
m=1
x
n,m
m=1
n=1
x
n,m
=: z.
The reverse inequality follows from symmetry.
A similar result holds for rearrangement of double-indexed series of non-
negative terms in R
:
Let : N NN be any bijective map, and let x
n,m
n1,m1
be
a double-indexed sequence of nonnegative elements of R
. Then
r=1
x
(r)
=
n=1
m=0
x
n,m
=
m=1
n=1
x
n,m
.
Appendix B
Axiom of choice
Let A, B be two sets. One denes AB, the Cartesian product of A and
B, to be the empty set if either A or B or both are empty, and to be the set
of all ordered pairs (a, b), a A, b B, when both A and B are nonempty.
Similarly, for a nite family of nonempty sets A
1
, . . . , A
n
, we dene their
Cartesian product to be the set
A
1
A
n
:= (x
1
, x
2
, . . . x
n
) [ x
i
A
i
, i = 1, 2, . . . , n.
Obviously, A
1
A
n
is a nonempty set. We can think of A
1
A
n
as
the set of all functions f : 1, 2, . . . , n
n
i=1
A
i
with f(i) = x
i
A
i
for
each i. One can copy this to dene the Cartesian product of any arbitrary
family of sets, say A
I
, to be the set
I
A
:=
_
f : I
_
I
A
f() A
, I
_
.
However, there is no surety that the set
I
A
such
that f() A
I
is a non-empty family of sets such that each A
is non-
empty, then there exists a function f : I
I
A
such that
f() A
I.
101
102 Appendix B
Such a function is called a choice function.
The axiom of choice has many equivalent formulations; a useful one is
the following:
If A
I
A
such that E A
0
and card(T(R)) := 2
c
.
103
104 Appendix C
For a nite set A, we know that
card(T(A)) = 2
card(A)
> card(A).
Can the same be said about arbitrary sets which are not necessarily nite?
For two sets A and B, we say card(A) _ card(B) if there exists a one-one
map from A into B. We write card(A) <card(B) if card (A) _card(B) but
card(A) ,=card(B). Now we ask the question:
Is card(A) < card(T(A)) for any nonempty set A?
The answer is in the armative and is due to George Cantor (see Rana
[30] for a proof). It can be shown that 2
0
= c, i.e., T(N) R. Thus we
have the following:
0 < 1 < 2 < < n < <
0
< 2
0
= c < 2
c
< 2
2
c
< .
This raises the following natural question:
Does there exist a cardinal number such that
0
< < 2
0
= c?
That is, does there exist a set A R such that A , R and N A but
A , N? The answer to this question is not known. The statement that the
answer to the above question is in the negative is called the continuum
hypothesis:
There does not exist any cardinal number between
0
and 2
0
= c.
An equivalent formulation of this is the following:
The set R can be well-ordered in such a way that each element of
R is preceded by only countably many elements.
We used this in section 3.4, to prove Ulams theorem. It is known that
the continuum hypothesis is independent of the Zermelo-Fraenkel axioms of
set theory.
References
[1] Aliprantis, C.D. and Burkinshaw, O. Principles of Real Analysis (3rd
Edition). Academic Press, Inc. New York, 1998.
[2] Apostol, T.M. Mathematical Analysis. Narosa Publishing House, New
Delhi (India), 1995.
[3] Bhatia, Rajendra Fourier Series. Hindustan Book Agency, New Delhi
(India), 1993.
[4] Billingsley, Patrick Probability and Measure. 3rd Edition, John Wiley
and Sons, New York, 1995.
[5] Friedman, A. Foundations of Modern Analysis. Holt, Rinehart and
Winston, Inc., New York, 1970.
[6] Halmos, P.R. Measure Theory. Van Nostrand, Princeton, 1950.
[7] Halmos, P.R. Naive Set Theory. Van Nostrand, Princeton, 1960.
[8] Hewitt, E. and Stromberg, K. Real and Abstract Analysis. Springer-
Verlag, Heidelberg, 1969.
[9] Kolmogorov, A.N. Foundations of Probability Theory. Chelsea Pub-
lishing Company, New York, 1950.
105
106 References
[10] Parthasarathy, K.R. Introduction to Probability and Measure.
Macmillan Company of India Ltd., Delhi, 1977.
[11] Parthasarathy, K. R. Probablity Measures on Metric Spaces, Academic
Press, New York, 1967.
[12] Rana, Inder K. An Introduction to Measure and Integration,
(First Edition) Narosa Publishers, Delhi, 1997.
(Second Edition) Graduate Seies in Mathematics Volume 45, Ameri-
can Mathematical Society, USA, 2001.
[13] Rana, Inder K. From Numbers to Analysis,
World Scientic Publishers, Singapore, 1998.
[14] Royden, H.L. Real Analysis (3rd Edition). Macmillan, New York,
1963.
Index
(X, S, ), 26
C(R), 61
C[a, b], 60
C
c
(R), 61
E
y
, 66
E
x
, 66
L
r
1
(X, S, ), 80
L
1
(X, S, ), 50
L
1
[a, b], 56
L
1
-metric, 60
L
1
(E), 56
L
p
(X, S, ), 82
L
p
(), 82
S
, 89
C, 79
L, 44
L
+
, 39
L
+
0
, 35
R, 10
I, 2
Im(f), 79
f, g, 88
_
fd, 39, 80
_
sd, 36
_
E
fd, 40
_
E
fd, 53
, 21
F
, 14
Re (f), 79
- algebra of Borel subsets of R, 27
- nite set function, 17
-algebra generated, 5
-algebra monotone class theorem, 7
-algebra of Borel subsets of X, 7
-algebra technique, 7
-algebra, product, 64
-set, 3
f g, 59
f
1
(C), 3
s
1
s
2
, 37
s
1
s
2
, 37
AB, 64
B
X
, 7
B
R
, 27
B
R
2, 65
C E, 3
F(C), 3
I, 10
I
0
, 32
I
d
, 7
I
r
, 7
L
F
, 24
L
R
, 27
M(C), 6
S(C), 5
S
, 23
p
th
-norm of f, 82
a.e., 39
a.e. ()x, 39
a.e. ()x Y , 39
a.e. x(), 39
aleph-nought, 103
algebra, 1
algebra generated, 3
almost everywhere, 39
analytic set, 28
arrangement, 99
Arzelas theorem, 58
107
108 Index
Banach spaces, 84
Bessels inequality, 92, 94
Binomial distribution, 10
Borel measurable function, 74
Borel subsets, 27, 33
Borel subsets of R
2
, 65
bounded convergence theorem, 52
bounded linear functional, 90
cardinal number, 103
cardinality of a set, 103
cardinality of the continuum, 103
Cartesian product, 101
Cauchy-Schwartz inequality, 83, 88
Chebyshevs inequality, 54, 85
choice function, 102
closed subspace, 89
complete measure space, 26
completion of a measure space, 25, 26
complex numbers, 79
continuity from above, 16
continuity from below, 16
continuum hypothesis, 104
convergent to +, 98
convergent to , 98
countable, 103
countably additive, 9
countably subadditive, 10
countably subadditivity, 12
counting measure, 70
cylinder set, 4
discrete measure, 10
discrete probability measure, 10
distribution function, 15, 32
distribution function, probability, 32
distribution of the measurable function, 48
distribution, Binomial, 10
distribution, discrete probability, 10
distribution, Poisson, 10
distribution, uniform, 10
equipotent, 103
equivalent, 59
essential supremum, 86
essentially bounded, 86
Eulers identity, 96
example, Vitalis, 34
extended real numbers, 10, 98
extension, 19
extension of a measure, 19, 24
Fatous lemma, 45
nite additive property, 11
nite set, 103
nitely additive, 9
Fourier coecients, 93
Fourier series, 93
Fubinis theorem, 69
function, choice, 102
function, imaginary part of a, 79
function, integrable, 50
function, length, 11
function, measurable, 44
function, nonnegative measurable, 35, 39
function, nonnegative simple measurable, 35
function, real part of a, 79
function, simple measurable, 44
function, support of a, 61
function,integrable, 80
generated, -algebra, 5
generated, algebra, 3
generated, monotone class, 6
graph of the function, 68
H olders inequality, 83
Haar measure, 31
Hilbert space, 89
inner product, 88
integrable, 50, 80
integral, 39, 50, 80
integral of nonnegative simple measurable
function, 36
integral over E , 40
integration of radial functions, 76
intervals with dyadic endpoints, 7
intervals with rational endpoints, 7
kernel, 91
Lebesgue integrable functions, 56
Lebesgue integral, 56
Lebesgue measurable sets, 27
Lebesgue measurable subsets of R
2
, 75
Lebesgue measure, 27
Lebesgue measure space, 27, 75
Lebesgue null, 32
Lebesgue outer measure, 27
Lebesgues dominated convergence, 81
Lebesgue-Stieltjes measure, 32
left open, right closed intervals, 2
length function, 11
length function, countable additivity of, 12
length function, countable subadditivity of,
12
length function, nite additivity of, 11
length function, monotonicity property of,
11
length function, translation invariance of, 12
limit inferior, 98
limit superior, 98
Index 109
mean value property, 57
measurable space, 25
measurable cover, 25
measurable function, 44
measurable function, nonnegative, 39
measurable kernel, 25
measurable partition, 41
measurable rectangle, 64, 65
measurable set, 23
measure, 10
measure space, 25
measure space, complete, 26
measure space, completion of a, 26
measure space,completeness of a, 25
measure, counting, 70
measure, discrete, 10
measure, Haar, 31
measure, Lebesgue, 27
measure, Lebesgue-Stieltjes, 24, 32
measure,outer, 21
measure,outer regular, 33
Minkowskis inequality, 83
monotone class, 5
monotone class generated, 6
monotone convergence theorem, 40
monotonicity property, 11
nonnegative simple measurable function, 35
norm, 59, 84
norm, induced by the inner product, 88
normed linear space, 84
null subset, 26
open intervals, 7, 32
orthogonal, 89
orthogonal complement, 89
outer measure, 21
outer measure induced, 21
outer measure, , 27
outer measure, Lebesgue, 27
outer regular, 33
outer-regularity of , 30
parallelogram identity, 92
Parsevals identity, 95
partial sum of the Fourier series, 93
partition, measurable, 41
Poisson distribution, 10
power set, 1
probability, 26
probability distribution function, 32
probability measure, discrete, 10
probability space, 26
product -algebra, 64, 65
product measure space, 65
product of the measures and , 65
projection theorem, 90
Pythagoras identity, 92
real part, 79
regularity of
R
2, 76
representation, 35
Riemann-Lebesgue lemma, 61
Riesz representation, 91
Riesz-Fischer theorem, 83, 95
section of E at x, 66
section of E at y, 66
semi-algebra, 1
set function, 9
set function, -nite, 17
set function, countably additive, 9
set function, countably subadditive, 10
set function, nite, 17
set function, nitely additive, 9
set function, induced, 14
set function, monotone, 9
sigma algebra, 5
simple function technique, 52
simple measurable function, 44
simple,nonnegative measurable function, 35
space, Banach, 84
space, Hilbert, 88
space, inner product, 88
space, Lebesgue measure, 27
space, measurable, 25
space, measure, 25
space, normed, 84
space, probability, 26
space, product measure, 65
standard representation, 35
subset,null, 26
subspace, 89
subspace, closed, 89
supp (f), 61
support, 61
Theorem, -algebra monotone class, 7
Theorem, Arzelas, 58
theorem, bounded convergence, 52
Theorem, Fubini, 69
Theorem, Lebesgues dominated convergence,
51
Theorem, Monotone Convergence, 40
Theorem, Riesz-Fischer, 60, 95
Theorem, Ulams, 20
topological group, 31
totally nite, 17
transition measure, 75
translation invariance, 12, 75
triangle-inequality, 84
truncation sequence, 48
Ulams theorem, 20
110 Index
ultralter, 13
uncountable, 103
Uniform distribution, 10
Vitalis example, 34