Dietmar A. Salamon
Dietmar A. Salamon Dietmar A. Salamon
Measure and Integration
Measure
ISBN 978-3-03719-159-0
www.ems-ph.org
Salamon Cover | Font: Frutiger_Helvetica Neue | Farben: Pantone 116, Pantone 287 | RB 31 mm
EMS Textbooks in Mathematics
Measure
and
Integration
Author:
Dietmar A. Salamon
Departement Mathematik
ETH Zürich
Rämistrasse 101
8092 Zürich
Switzerland
E-mail: dietmar.salamon@math.ethz.ch
2010 Mathematics Subject Classification: Primary: 28-01; Secondary: 28A05, 28A10, 28A12,
28A15, 28A20, 28A25, 28A30, 28A33, 28A35, 28C05, 28C10, 28C15, 35J05, 43A05, 44A35,
46B22, 46C05, 46E27, 46E30
ISBN 978-3-03719-159-0
The Swiss National Library lists this publication in The Swiss Book, the Swiss national bibliography,
and the detailed bibliographic data are available on the Internet at http://www.helveticat.ch.
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broad-
casting, reproduction on microfilms or in other ways, and storage in data banks. For any kind of use
permission of the copyright owner must be obtained.
Contact address:
This book is based on notes for the lecture course Measure and Integration held
at ETH Zürich in the spring semester 2014. Prerequisites are the first-year courses
on analysis and linear algebra, including the Riemann integral [9, 18, 19, 21],
as well as some basic knowledge of metric and topological spaces. The course
material is based in large parts on Chapters 1–8 of the textbook real and complex
analysis by Walter Rudin [17]. In addition to Rudin’s book, the lecture notes by
Urs Lang [10, 11], the five volumes on measure theory by David H. Fremlin [4],
the paper by Heinz König [8] on the generalized Radon–Nikodým theorem,
the lecture notes by C. E. Heil [7] on absolutely continuous functions, Dan Ma’s
topology blog [12] on exotic examples of topological spaces, and the paper by
Gert K. Pedersen [16] on the Haar measure were very helpful in preparing this
manuscript.
The text also contains some material that was not covered in the lecture course,
namely some of the results in Sections 4.5 and 5.2 (concerning the dual space of
Lp ./ in the non- -finite case), Section 5.4 on the generalized Radon–Nikodým
theorem, Sections 7.6 and 7.7 on Marcinkiewicz interpolation and the Calderón–
Zygmund inequality, and Chapter 8 on the Haar measure.
I am grateful to many people who helped to improve this manuscript. Thanks
to the students at ETH who pointed out typos or errors in earlier drafts. Thanks to
Andreas Leiser for his careful proofreading. Thanks to Theo Buehler for many en-
lightening discussions and for pointing out the book by Fremlin, Dan Ma’s topology
blog, and the paper by Pedersen. Thanks to Urs Lang for his insightful comments
on the construction of the Haar measure.
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
3 Borel measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.1 Regular Borel measures . . . . . . . . . . . . . . . . . . . . . . . 95
3.2 Borel outer measures . . . . . . . . . . . . . . . . . . . . . . . . 107
3.3 The Riesz representation theorem . . . . . . . . . . . . . . . . . . 113
3.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4 Lp spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.1 Hölder and Minkowski . . . . . . . . . . . . . . . . . . . . . . . . 133
4.2 The Banach space Lp ./ . . . . . . . . . . . . . . . . . . . . . . . 136
4.3 Separability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.4 Hilbert spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
4.5 The dual space of Lp ./ . . . . . . . . . . . . . . . . . . . . . . . 152
4.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
viii Contents
6 Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
6.1 Weakly integrable functions . . . . . . . . . . . . . . . . . . . . . 221
6.2 Maximal functions . . . . . . . . . . . . . . . . . . . . . . . . . . 227
6.3 Lebesgue points . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6.4 Absolutely continuous functions . . . . . . . . . . . . . . . . . . . 239
6.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Appendices
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Introduction
We learn already in high school that integration plays a central role in mathemat-
ics and physics. One encounters integrals in the notions of area or volume, when
solving a differential equation, in the fundamental theorem of calculus, in Stokes’
theorem, or in classical and quantum mechanics. The first year analysis course
at ETH includes an introduction to the Riemann integral, which is satisfactory for
many applications. However, it has certain drawbacks, in that some very basic func-
tions are not Riemann integrable, that the pointwise limit of a sequence of Riemann
integrable functions need not be Riemann integrable, and that the space of Riemann
integrable functions is not complete with respect to the L1 -norm. One purpose of
this book is to introduce the Lebesgue integral, which does not suffer from these
drawbacks and agrees with the Riemann integral whenever the latter is defined.
Chapter 1 introduces abstract integration theory for functions on measure spaces.
It includes proofs of the Lebesgue monotone convergence theorem, the Fatou lemma,
and the Lebesgue dominated convergence theorem. In Chapter 2 we move on to
outer measures and introduce the Lebesgue measure on Euclidean space. Borel
measures on locally compact Hausdorff spaces are the subject of Chapter 3. Here
the central result is the Riesz representation theorem. In Chapter 4 we encounter Lp
spaces and show that the compactly supported continuous functions form a dense
subspace of Lp for a regular Borel measure on a locally compact Hausdorff space
when p < 1. Chapter 5 is devoted to the proof of the Radon–Nikodým theorem
about absolutely continuous measures and to the proof that Lq is naturally isomor-
phic to the dual space of Lp when 1=p C1=q D 1 and 1 < p < 1. Chapter 6 deals
with differentiation. Chapter 7 introduces product measures and contains a proof of
Fubini’s Theorem, an introduction to the convolution product on L1 .Rn /, and a
proof of the Calderón–Zygmund inequality. Chapter 8 constructs Haar measures on
locally compact Hausdorff groups.
Despite the overlap with the book of Rudin [17], there are some differences in
exposition and content. A small expository difference is that in Chapter 1 measur-
able functions are defined in terms of pre-images of (Borel) measurable sets rather
than pre-images of open sets. The Lebesgue measure in Chapter 2 is introduced in
terms of the Lebesgue outer measure instead of as a corollary of the Riesz repre-
2 Introduction
One way out of this dilemma is to give up on the idea of countable additivity and
replace it by the weaker requirement of countable subadditivity. This leads to the
notion of an outer measure which will be discussed in Chapter 2. Another way out is
to retain the requirement of countable additivity, but give up on the idea of assigning
a measure to every subset of a given domain. Instead, one assigns a measure only
to some subsets which are then called measurable. This idea will be pursued in
the present chapter. A subtlety of this approach is that in some important cases it
is not possible to give an explicit description of those subsets of a given domain
that one wants to measure, and instead one can only impose certain axioms that the
collection of all measurable sets must satisfy. By contrast, in topology the open sets
can often be described explicitly. For example, the open subsets of the real line are
countable unions of open intervals, while there is no such explicit description for
the Borel measurable subsets of the real line.
The precise formulation of this approach leads to the notion of a -algebra
which is discussed in Section 1.1. Section 1.2 introduces measurable functions and
examines their basic properties. Measures and the integrals of positive measurable
functions are the subject of Section 1.3. Here the nontrivial part is to establish addi-
tivity of the integral and the proof is based on the Lebesgue monotone convergence
theorem. An important inequality is the Fatou lemma. It is needed to prove the
Lebesgue dominated convergence theorem in Section 1.4 for real valued integrable
functions. Section 1.5 deals with sets of measure zero, which are negligible for many
purposes. For example, it is often convenient to identify two measurable functions if
they agree almost everywhere, i.e., on the complement of a set of measure zero. This
defines an equivalence relation. The quotient of the space of integrable functions by
this equivalence relation is a Banach space and is denoted by L1 . Section 1.6 dis-
cusses the completion of a measure space. Here the idea is to declare every subset
of a set of measure zero to be measurable as well.
1.1 -algebras
For any fixed set X denote by 2X the set of all subsets of X and, for any subset
A X , denote by
Ac WD X n A
its complement.
1.1. -algebras 5
is a -algebra.
Exercise 1.6. (i) Let X be a set and let A; B X be subsets such that the four sets
A n B; B n A; A \ B; X n .A [ B/ are nonempty. What is the cardinality of the
smallest -algebra A X containing A and B?
(ii) How many -algebras on X are there when #X D k for k D 0; 1; 2; 3; 4?
(iii) Is there an infinite -algebra with countable cardinality?
6 1. Abstract measure theory
Exercise 1.7. Let X be any set and let I be any nonempty index set. Suppose
that for every i 2 I a -algebra Ai 2X is given. Prove that the intersection
A WD i 2I Ai D ¹A X j A 2 Ai for all i 2 I º is a -algebra.
T
Lemma 1.8. Let X be a set and E 2X be any set of subsets of X. Then there is
a unique smallest -algebra A 2X containing E (i.e., A is a -algebra, E A,
and if B is any other -algebra with E B, then A B).
Proof. Uniqueness follows directly from the definition. Namely, if A and B are
two smallest -algebras containing E, we have both B A and A B and hence
X
A D B. To prove existence, denote by S 22 the collection of all -algebras
B 2X that contain E and define
ˇ if B 2X is a -algebra
\ ² ˇ ³
A WD BD AX ˇ :
such that E B; then A 2 B
ˇ
B2S
Thus A is a -algebra by Exercise 1.7. Moreover, it follows directly from the defi-
nition of A that E A and that every -algebra B that contains E also contains A.
This proves Lemma 1.8.
Lemma 1.8 is a useful tool to construct nontrivial -algebras. Before doing that,
let us first take a closer look at Definition 1.1. The letter “ ” stands for countable
and the crucial observation is that axiom (c) allows for countable unions. On the one
hand, this is a lot more general than only allowing for finite unions, which would
be the subject of Boolean algebra. On the other hand, it is a lot more restrictive
than allowing for arbitrary unions, which one encounters in the subject of topology.
Topological spaces will play a central role in this book and we recall here the formal
definition.
Definition 1.9 (topological space). Let X be a set. A collection U 2X of subsets
of X is called a topology on X if it satisfies the following axioms:
(a) ;; X 2 U;
Tn
(b) if n 2 N and U1 ; : : : ; Un 2 U, then Ui 2 U;
iD1
Conditions (a) and (b) in Definition 1.9 are also properties of every -algebra.
However, condition (c) in Definition 1.9 is not shared by -algebras because it per-
mits arbitrary unions. On the other hand, complements of open sets are typically
not open. Many of the topologies used in this book arise from metric spaces and are
familiar from first-year analysis courses. Here is a recollection of the definition.
Definition 1.10 (metric space). A metric space is a pair .X; d / consisting of a set
X and a function
dW X X ! R
satisfying the following axioms:
(a) d.x; y/ 0 for all x; y 2 X, with equality if and only if x D y;
(b) d.x; y/ D d.y; x/ for all x; y 2 X ;
(c) d.x; z/ d.x; y/ C d.y; z/ for all x; y; z 2 X .
A function d W X X ! R that satisfies these axioms is called a distance function
and the inequality in (c) is called the triangle inequality. A subset U X of a metric
space .X; d / is called open (or d -open) if for every x 2 U there exists a constant
" > 0 such that the open ball
B" .x/ WD B" .x; d / WD ¹y 2 X j d.x; y/ < "º
(centered at x and of radius ") is contained in U . The collection of d -open subsets
of X will be denoted by
U.X; d / WD ¹U X j U is d -openº:
It follows directly from the definitions that the collection U.X; d / 2X
of d -open sets in a metric space .X; d / satisfies the axioms of a topology in Defini-
tion 1.9. A subset F of a metric space .X; d / is closed if and only if the limit point
of every convergent sequence in F is itself contained in F .
Example 1.11. A normed vector space is a pair .X; kk/ consisting of a real vector
space X and a function
X ! RW x 7 ! kxk
satisfying the following:
(a) kxk 0 for all x 2 X , with equality if and only if x D 0;
(b) kxk D jj kxk for all x 2 X and 2 R;
(c) kx C yk kxk C kyk for all x; y 2 X.
8 1. Abstract measure theory
d.x; y/ WD kx yk
Example 1.12. The set X D R of real numbers is a metric space with the standard
distance function
d.x; y/ WD jx yj:
The topology on R induced by this distance function is called the standard topology
on R. The open sets in the standard topology are unions of open intervals.
Exercise. Every union of open intervals is a countable union of open intervals.
x WD Œ 1; 1 WD R [ ¹ 1; 1º:
R
For a; b 2 R define
x R
dW R x !R
2je x y e y x j
d.x; y/ WD ;
e xCy C ex y C ey x C e x y
2e x
d.x; 1/ WD d.1; x/ WD ;
ex C e x
2e x
d.x; 1/ WD d. 1; x/ WD ;
ex C e x
d. 1; 1/ WD d.1; 1/ WD 2:
1.1. -algebras 9
f .˙1/ WD ˙1;
is a homeomorphism. Prove that it is an isometry with respect to the metric
x and the standard metric on the interval Œ 1; 1. Deduce that .R
in (ii) on R x; d/
is a compact metric space.
x by
Exercise 1.14. Extend the total ordering of R to R x.
1 a 1 for a 2 R
Extend addition by
1 C a WD 1 for 1 < a 1,
and by
1 C a WD 1 for 1 a < 1.
(i) Define lim sup an and lim inf an and show that they always exist.
n!1 n!1
whenever the right-hand side exists. Find an example where the inequality is
strict.
(iv) If an bn for all n 2 N, show that
Definition 1.15. Let .X; U/ be a topological space and let B 2X be the smallest
-algebra containing U. Then B is called the Borel -algebra of .X; U/ and the
elements of B are called Borel (measurable) sets.
10 1. Abstract measure theory
Lemma 1.16. Let .X; U/ be a topological space. Then the following holds.
Proof. Part (i) follows from the definition of Borel sets and condition (b) in
Definition 1.1, part (ii) follows from (i) and (c), and part (iii) follows from (ii)
and (b), because the complement of an F -set is a Gı -set.
Consider for example the Borel -algebra on the real axis R with its standard
topology. In view of Lemma 1.16 it is a legitimate question whether there is any
subset of R at all that is not a Borel set. The answer to this question is positive,
which may not be surprising; however, the proof of the existence of subsets that are
not Borel sets is surprisingly nontrivial. It will only appear much later in this book,
after we have introduced the Lebesgue measure (see Lemma 2.15). For now it is
useful to note that, roughly speaking, every set that one can construct in terms of
some explicit formula, will be a Borel set, and one can only prove with the Axiom
of Choice that subsets of R must exist that are not Borel sets.
The set K is called compact if every open cover of K has a finite subcover, i.e.,
if for every open cover ¹Ui ºi2I of K there exist finitely many indices i1 ; : : : ; in 2 I
such that
K Ui1 [ [ Uin :
1.2. Measurable functions 11
When .X; d / is a metric space and U D U.X; d / is the topology induced by the
distance function (Definition 1.10), the two notions of compactness agree. Thus,
for every subset K X, every sequence in K has a subsequence converging to an
element of K if and only if every open cover of K has a finite subcover. For a proof
see, for example, Munkres [14] or [20, Appendix C.1]. We emphasize that when K
is a compact subset of a general topological space .X; U/ it does not follow that K is
closed. For example, a finite subset of X is always compact, but need not be closed
or, if U D ¹;; Xº, then every subset of X is compact, but only the empty set and
X itself are closed subsets of X. If, however, .X; U/ is a Hausdorff space (i.e., for
any two distinct points x; y 2 X there exist open sets U; V 2 U such that x 2 U ,
y 2 V , and U \ V D ;), then every compact subset of X is closed (Lemma A.2).
Next, recall that a map f W X ! Y between two metric spaces .X; dX / and
.Y; dY / is continuous (i.e., for every x 2 X and every " > 0 there is a ı > 0 such
that f .Bı .x; dX // B" .f .x/; dY /) if and only if the pre-image
1
f .V / WD ¹x 2 X j f .x/ 2 V º
of every open subset of Y is an open subset of X . This second notion carries over
to general topological spaces, i.e., a map f W X ! Y between topological spaces
.X; UX / and .Y; UY / is called continuous if
V 2 UY H) f 1
.V / 2 UX :
It follows directly from the definition that topological spaces form a category, in that
the composition g ı f W X ! Z of two continuous maps f W X ! Y and gW Y ! Z
between topological spaces is again continuous. Another basic observation is that
if f W X ! Y is a continuous map between topological spaces and K is a compact
subset of X, then its image f .K/ is a compact subset of Y .
Definition 1.17 (measurable function). (i) Let .X; AX / and .Y; AY / be measurable
spaces. A map f W X ! Y is called measurable if the pre-image of every measur-
able subset of Y under f is a measurable subset of X, i.e.,
B 2 AY H) f 1
.B/ 2 AX :
defined by
´
1 if x 2 A;
A .x/ WD (1.4)
0 if x … A:
Now assume .X; A/ is a measurable space, consider the Borel -algebra on R, and
let A X be any subset. Then A is a measurable function if and only if A is a
measurable set.
1.2. Measurable functions 13
Part (iii) in Definition 1.17 is the special case of part (i), where AX 2X and
AY 2Y are the -algebras of Borel sets (see Definition 1.15). Theorem 1.20
below shows that every continuous function between topological spaces is Borel
measurable. It also shows that a function from a measurable space to a topological
space is measurable with respect to the Borel -algebra on the target space if and
only if the pre-image of every open set is measurable. Since the collection of Borel
sets is in general much larger than the collection of open sets, the collection of
measurable functions is then also much larger than the collection of continuous
functions.
Theorem 1.19 (measurable maps). Let .X; AX /, .Y; AY /, and .Z; AZ / be measur-
able spaces.
(i) The identity map idX W X ! X is measurable.
(ii) If f W X ! Y and gW Y ! Z are measurable maps, then so is the composition
g ı f W X ! Z.
(iii) Let f W X ! Y be any map. Then the set
f AX WD ¹B Y j f 1
.B/ 2 AX º (1.5)
V 2 UY H) f 1
.V / 2 AX :
V 2 UY H) f 1
.V / 2 AX :
This proves part (i). If, in addition, AX is the Borel -algebra of a topology UX
on X and f W .X; UX / ! .Y; UY / is a continuous map, then the pre-image of every
open subset V Y under f is an open subset of X and hence is a Borel subset of
X ; thus it follows from part (i) that f is Borel measurable. This proves part (ii) and
Theorem 1.20.
(i) f is measurable;
1
(ii) f ..a; 1/ is a measurable subset of X for every a 2 R;
1
(iii) f .Œa; 1/ is a measurable subset of X for every a 2 R;
1
(iv) f .Œ 1; b// is a measurable subset of X for every b 2 R;
1
(v) f .Œ 1; b/ is a measurable subset of X for every b 2 R.
Proof. That (i) implies (ii), (iii), (iv), and (v) follows directly from the definitions.
We prove that (ii) implies (i). Thus let f W X ! R x be such that f 1 ..a; 1/ 2 AX
for every a 2 R and define
x jf x
B WD f AX D ¹B R 1
.B/ 2 AX º 2R :
by axiom (c) in Definition 1.1. Hence it follows from (f) in Lemma 1.2 that
.a; b/ D Œ 1; b/ \ .a; 1 2 B
1.2. Measurable functions 15
for every pair of real numbers a < b. Since every open subset of R x is a count-
able union of sets of the form .a; b/, .a; 1, Œ 1; b/, it follows from axiom (c) in
Definition 1.1 that every open subset of Rx is an element of B. Hence it follows from
Theorem 1.20 that f is measurable. This shows that (ii) implies (i). That either of
the conditions (iii), (iv), and (v) also implies (i) is shown by a similar argument
which is left as an exercise for the reader. This proves Theorem 1.21.
Our next goal is to show that sums, products, and limits of measurable func-
tions are again measurable. The next two results are useful for the proofs of these
fundamental facts.
Then
n
\
1 1
f .Q.a; b// D fi ..ai ; bi // 2 A
iD1
by property (f) in Lemma 1.2. Now every open subset of Rn can be expressed as
a countable union of sets of the form Q.a; b/. (Prove this!) Hence it follows from
axiom (c) in Definition 1.1 that f 1 .U / 2 A for every open set U Rn , and so f
is measurable. This proves Theorem 1.22.
Proof. The map f WD .u; v/W X ! R2 is measurable (with respect to the Borel
-algebra on R2 ) by Theorem 1.22 and the map W R2 ! R is Borel measurable
by Theorem 1.20. Hence the composition h D ı f W X ! R is measurable by
Theorem 1.19. This proves Lemma 1.23.
16 1. Abstract measure theory
is measurable. Hence it follows from Theorem 1.21 that g is measurable. It also fol-
lows from part (i) (already proved) that fk is measurable, hence so is supk . fk /
by what we have just proved, and then so is the function inf k fk D supk . fk /.
With this understood, we conclude that the functions
Step functions
We close this section with a brief discussion of measurable step functions. Such
functions will play a central role throughout this book. In particular, they are used
in the definition of the Lebesgue integral.
and define
1
Ai WD s .˛i / D ¹x 2 X j s.x/ D ˛i º
`
[
XD Ai ; Ai \ Aj D ; for i ¤ j: (1.6)
iD1
`
X
sD ˛i A ; (1.7)
i
i D1
f W X ! Œ0; 1
by
´
n n n
k2 if k2 t < .k C 1/2 ; k D 0; 1; : : : ; n2n 1;
n .t/ WD (1.8)
n if t n:
These functions are Borel measurable and satisfy n .0/ D 0 and n .1/ D n for
all n, as well as t 2 n n .t/ nC1 .t/ t whenever n t > 0. Thus
Our next goal is to define the integral of a measurable step function and then the inte-
gral of a general nonnegative measurable function via approximation. This requires
the notion of volume or measure of a measurable set. The definitions of measure
and integral will require some arithmetic on the space Œ0; 1. Addition to 1 and
multiplication by 1 are defined by
´
1 if a ¤ 0;
a C 1 WD 1 C a WD 1; a 1 WD 1 a WD
0 if a D 0:
With this convention, addition and multiplication are commutative, associative, and
distributive. Moreover, if ai and bi are nondecreasing sequences in Œ0; 1, then the
limits
These rules must be treated with caution. The product rule does not hold when
the sequences are not nondecreasing. For example ai WD i converges to a D 1,
bi WD 1=i converges to b D 0, but ai bi D 1 does not converge to ab D 0.
(Exercise. Show that the sum of two convergent sequences in Œ0; 1 always con-
verges to the sum of the limits.) Also, for all a; b; c 2 Œ0; 1,
a C b D a C c; a < 1 H) b D c;
(i) .;/ D 0.
.A/ .B/:
20 1. Abstract measure theory
Proof. We prove (i). Choose A1 2 A such that .A1 / < 1 and define Ai WD ; for
i > 1. Then it follows from -additivity that
X
.A1 / D .A1 / C .;/;
i>1
Here the last equality follows from part (ii). This proves part (iv).
We prove (v). Assume Ai Ai C1 for all i and define Ci WD Ai n AiC1 . Then
Ci is measurable for all i and, for n 2 N,
1
[ 1
\
An D A [ Ci ; A WD Ai :
iDn i D1
1.3. Integration of nonnegative functions 21
W A ! Œ0; 1
is defined by
.A/ WD #A for A 2 A.
As an example, consider the counting measure W 2N ! Œ0; 1 on the natural num-
bers. Then the sets An WD ¹n; nC1; º all have infinite measure and their intersec-
tion is the empty set and hence has measure zero. Thus the hypothesis .A1 / < 1
cannot be removed in part (v) of Theorem 1.28.
Example 1.32. Let X be an uncountable set and let A be the -algebra of all subsets
of X that are either countable or have countable complements (Example 1.4). Then
the function W A ! Œ0; 1 defined by .A/ WD 0 when A is countable and by
.A/ WD 1 when Ac is countable is a measure.
With these preparations in place we are now ready to introduce the Lebesgue
integral of a nonnegative measurable function
Definition 1.34 (Lebesgue integral). Let .X; A; / be a measure space and let
E 2 A be a measurable set.
(i) Let sW X ! Œ0; 1/ be a measurable step function of the form
n
X
sD ˛i A (1.9)
i
iD1
where the supremum is taken over all measurable step function sW X ! Œ0; 1/
that satisfy s.x/ f .x/ for all x 2 X .
The reader may verify that the right-hand side of (1.10) depends only on s and
not on the choice of ˛i and Ai . The same definition can be used if f is only defined
on the measurable set E X. Then AE WD ¹A 2 A j A Eº is a -algebra on E
Rand E WD jAE is a measure. So .E; AE ; E / is a measure space and the integral
E f dE is well defined. It agrees with the integral of the extended function on X
defined by f .x/ WD 0 for x 2 X n E.
Theorem 1.35 (basic properties of the Lebesgue integral). Let .X; A; / be a mea-
sure space, f; gW X ! Œ0; 1 be measurable functions, and E 2 A. Then the
following holds:
R R
(i) if f g on E, then E f d E g d;
R R
(ii) E f d D X f E d;
R
(iii) if f .x/ D 0 for all x 2 E, then E f d D 0;
R
(iv) if .E/ D 0, then E f d D 0;
(v) if A 2 A and E A, then E f d A f d;
R R
R R
(vi) if c 2 Œ0; 1/, then E cf d D c E f d.
1.3. Integration of nonnegative functions 23
by the definition of the integral of g. Now take the supremum over all measurable
step functions s f to obtain
Z Z
f d g d:
E E
Here the supremum is over all measurable step functions sW X ! Œ0; 1/, respec-
tively tW X ! Œ0; 1/, that satisfy s f , respectively t f E . The second
equality Rfollows fromR the fact that every measurable step function sW X ! Œ0; 1/
satisfies E s d D X sE d by the definition of the integral. The third equa-
tion follows from the fact that a measurable step function t W X ! Œ0; 1/ satisfies
t f E if and only if it has the form t D sE for some measurable step function
sW X ! Œ0; 1/ such that s f .
R
Part (iii) is a consequence of (i) with g D 0 Rand the fact that E f d 0
by definition. Part (iv) follows from the fact that E s d D 0 for every measur-
able step function s when .E/ D 0. Part (v) follows from parts (i) and (ii) and
R fact that fRE f A whenever E A. Part (vi) follows from the fact that
the
E cs d D c E s d for every c 2 Œ0; 1/ and every measurable step function s,
by the commutativity, associativity, and distributivity rules for calculations with
numbers in Œ0; 1. This proves Theorem 1.35.
24 1. Abstract measure theory
Notably absent from the statements of Theorem 1.35 is the assertion that the
integral of a sum is the sum of the integrals. This is a fundamental property that any
integral should have. The proof that the integral in Definition 1.34 indeed satisfies
this crucial condition requires some preparation. First one verifies this property for
integrals of step functions and then one uses the Lebesgue monotone convergence
Theorem 1.37.
Lemma 1.36 (additivity for step functions). Let .X; A; / be a measure space and
let s; tW X ! Œ0; 1/ be measurable step functions.
(i) For every measurable set E 2 A
Z Z Z
.s C t/ d D s d C t d:
E E E
and hence
Z m X
X n
.s C t/ d D .˛i C ˇj /.Ai \ Bj \ E/
E iD1 j D1
m
X n
X n
X m
X
D ˛i .Ai \ Bj \ E/ C ˇj .Ai \ Bj \ E/
iD1 j D1 j D1 i D1
m
X n
X
D ˛i .Ai \ E/ C ˇj .Bj \ E/
iD1 j D1
Z Z
D s d C t d:
E E
1.3. Integration of nonnegative functions 25
Z m
X
s d D ˛i .E \ Ai /
E iD1
m
X 1
X
D ˛i .Ek \ Ai /
iD1 kD1
m
X n
X
D ˛i lim .Ek \ Ai /
n!1
iD1 kD1
m
X n
X
D lim ˛i .Ek \ Ai /
n!1
iD1 kD1
n X
X m
D lim ˛i .Ek \ Ai /
n!1
kD1 iD1
n Z
X
D lim s d
n!1 Ek
kD1
1 Z
X
D s d:
Ek
kD1
Define f W X ! Œ0; 1 by
and hence Z
˛ f d: (1.12)
X
Now fix a measurable step function sW X ! Œ0; 1/ such that s f . Define
s W A ! Œ0; 1
by Z
s .E/ WD s d for E 2 A: (1.13)
E
This function is a measure by part (ii) of Lemma 1.36 (which asserts that s is
-additive) and by part (iv) of Theorem 1.35 (which asserts that s .;/ D 0).
Now fix a constant 0 < c < 1 and define
Here the last inequality follows from the definition of ˛ in (1.11). Hence
˛
s .En / for all n 2 N: (1.15)
c
Since s W A ! Œ0; 1 is a measure, by part (i) of Theorem 1.35, it follows
from (1.14) and part (iv) of Theorem 1.28 that
˛
Z
s d D s .X/ D lim s .En / : (1.16)
X n!1 c
Here the last inequality
R follows from (1.15). Since (1.16) holds for every constant
0 < c < 1, we have X s d ˛ for every measurable step function sW X ! Œ0; 1/
such that s f . Take the supremum over all such s to obtain
Z Z
f d D sup s d ˛:
X sf X
R
Combining this with (1.12) we obtain X f d D ˛ and hence the assertion of
Theorem 1.37 follows from the definition of ˛ in (1.11).
Proof. (i) By Theorem 1.26, there exist sequences of measurable step functions
sn ; tn W X ! Œ0; 1/ such that
sn snC1 and tn tnC1
for all n 2 N and
f .x/ D lim sn .x/ and g.x/ D lim tn .x/
n!1 n!1
Here the second equality follows from the definition of gn while the third one fol-
lows from part (i) of the present theorem (already proved). This proves (ii) for
E D X. To prove it in general, replace f; fn by f E ; fn E and use part (ii) of
Theorem 1.35.
(iii) Let f W X ! Œ0; 1 be a measurable function and let Ek 2 A be a sequence
of pairwise disjoint measurable sets. Define
1
[ n
X
E WD Ek ; fn WD f E :
k
kD1 kD1
Then it follows from part (i) of the present theorem (already proved) and part (ii) of
Theorem 1.35 that
Z n
Z X
fn d D f E d
k
X X
kD1
n Z
X
D f E d
k
X
kD1
n Z
X
D f d:
Ek
kD1
Exercise 1.39. Let W 2N ! Œ0; 1 be the counting measure on the natural num-
bers. Show that in this case equation (1.18) in part (ii) of Theorem 1.38 is equivalent
to the formula
X1 X 1 X 1 X 1
aij D aij (1.20)
iD1 j D1 j D1 iD1
The next theorem shows that every measurable function f W X ! Œ0; 1 induces
another measure f on .X; A/.
f W A ! Œ0; 1;
defined by Z
f .E/ WD f d for E 2 A (1.21)
E
is a measure, and Z Z
g df D fg d (1.22)
E E
Proof. f is -additive by part (iii) of Theorem 1.38 and f .;/ D 0 by part (iv)
of Theorem 1.35. Hence f is a measure (see Definition 1.27). Now let g WD A
be the characteristic function of a measurable set A 2 A. Then
Z Z Z
A df D f .A/ D f d D f A d:
X A X
Here the first equality follows from the definition of the integral for measurable step
functions in Definition 1.34, the second equality from the definition of f , and the
last equality from part (ii) of Theorem 1.35. Thus equation (1.22) (with E D X )
holds for characteristic functions of measurable sets. Taking finite sums and using
part (vi) of Theorem 1.35 and part (i) of Theorem 1.38 we find that (1.22) (with
E D X) continues to hold for all measurable step functions g D sW X ! Œ0; 1/.
Now approximate an arbitrary measurable function gW X ! Œ0; 1 by a sequence
of measurable step functions via Theorem 1.26 and use the Lebesgue monotone
convergence theorem (Theorem 1.37) to deduce that (1.22) holds with E D X for
all measurable functions gW X ! Œ0; 1. Now replace g by gE and use part (ii) of
Theorem 1.35 to obtain equation (1.22) in general. This proves Theorem 1.40.
1.3. Integration of nonnegative functions 31
for all x 2 X . Moreover, gn fi for all i n. By part (i) of Theorem 1.35 this
implies Z Z
gn d fi d
X X
for all i n, and hence
Z Z
gn d inf fi d
X i n X
Here the second equation follows from part (vi) of Theorem 1.35. If c < 0,
then .cf /C D . c/f and .cf / D . c/f C and hence, again using part (iv)
of Theorem 1.35, we obtain
Z Z Z
cf d D . c/f d . c/f C d
E E E
Z Z
D . c/ f d . c/ f C d
E E
Z
Dc f d:
E
Therefore,
Z Z Z
C
h d D h d h d
E E E
Z Z Z Z
C C
D f d C g d f d g d
E E E E
Z Z
D f d C g d
E E
Proof. f is measurable by part (ii) of Theorem 1.24 and jf .x/j g.x/ for all
x 2 X by (1.30) and (1.31). Hence it follows from part (i) of Theorem 1.35 that
Z Z
jf j d g d < 1;
X X
Here the penultimate step follows from part (i) of Theorem 1.44. This implies
Z
lim sup jfn f j d 0:
n!1 X
Hence
Z
lim jfn f j d D 0:
n!1 X
36 1. Abstract measure theory
Since
ˇZ Z ˇ Z Z
ˇ ˇ
ˇ
ˇ fn d f dˇˇ jfn f j d jfn f j d;
E E E X
part (iii) of Theorem 1.44 yields
ˇZ Z ˇ
ˇ ˇ
lim ˇˇ fn d f dˇˇ D 0;
n!1 E E
which is equivalent to (1.32). This completes the proof of Theorem 1.45.
Lemma 1.48. Assume either that f; gW X ! Œ0; 1 are measurable functions that
agree almost everywhere, or that f; gW X ! R are -integrable functions that agree
almost everywhere. Then
Z Z
f d D g d for all A 2 A: (1.33)
A A
Here the first equality follows from part (iii) of Theorem 1.38 in the nonnegative
case and from part (iv) of Theorem 1.44 in the integrable case. The second equality
follows from part (iv) of Theorem 1.35 in the nonnegative case and from part (vi)
of Theorem 1.44 in the integrable case. The third equality follows from part (ii)
of Theorem 1.35 in the nonnegative case and from part (v) of Theorem 1.44 in the
integrable case. Since f AnN D gAnN , it follows that the integrals of f and g
over A agree. This proves Lemma 1.48.
The converse of Lemma 1.48 fails for nonnegative measurable functions. For
example, if X is a singleton and .X/ D 1, then the integrals of any two positive
functions agree over every measurable set. However, the converse of Lemma 1.48
does hold for integrable functions. Since the difference of two integrable functions
is again integrable, it suffices to assume g D 0, and in this case the converse also
holds for nonnegative measurable functions. This is the content of the next lemma.
38 1. Abstract measure theory
is a set of measure zero. In the integrable case apply this argument to the function
jf jW X ! Œ0; 1/. This proves Lemma 1.49.
Lemma 1.50. Let f 2 L1 ./. Then
ˇZ ˇ Z
ˇ ˇ
ˇ f dˇ D jf j d (1.34)
ˇ ˇ
X X
Definition 1.51 (the Banach space L1 ./). Define an equivalence relation on the
real vector space of all measurable function from X to R by
def
f g () the set ¹x 2 X j f .x/ ¤ g.x/º (1.35)
has measure zero:
Thus two functions are equivalent if and only if they agree almost everywhere.
(Verify that this is an equivalence relation!) By Lemma 1.48 the subspace L1 ./
is invariant under this equivalence relation, i.e., if f; gW X ! R are measurable,
f 2 L1 ./, and f g, then g 2 L1 ./. Moreover, the set ¹f 2 L1 ./ j f 0º
is a linear subspace of L1 ./ and hence the quotient space
L1 ./ WD L1 ./=
is again a real vector space. It is the space of all equivalence classes in L1 ./ under
the equivalence relation (1.35). Thus an element of L1 ./ is not a function on X,
but a set of functions on X . By Lemma 1.48, the map
Z
L ./ ! RW f 7 ! jf j d DW kf kL1
1
X
takes on the same value on all the elements in a given equivalence class and so
descends to the quotient space L1 ./. By Lemma 1.49 it defines a norm on L1 ./
and Theorem 1.53 below shows that L1 ./ is a Banach space with this norm (i.e.,
a complete normed vector space).
Theorem 1.52 (convergent series of integrable functions). Let .X; A; / be a mea-
sure space and let fn W X ! R be a sequence of -integrable functions such that
X1 Z
jfn j d < 1: (1.36)
nD1 X
Then there are a set N of measure zero and a function f 2 L1 ./ such that
1
X 1
X
jfn .x/j < 1 and f .x/ D fn .x/ for all x 2 X n N; (1.37)
nD1 nD1
Z 1 Z
X
f d D fn d for all A 2 A; (1.38)
A nD1 A
and Z ˇ n
X
ˇ
ˇ ˇ
lim ˇf fk ˇˇ d D 0: (1.39)
n!1 X ˇ
kD1
40 1. Abstract measure theory
Proof. Define
1
X
.x/ WD jfk .x/j
kD1
< 1:
by Lemma 1.48. Since jgn .x/j g.x/ for all n 2 N and gn converges pointwise to
f , it follows from the Lebesgue dominated convergence theorem (Theorem 1.45)
that f 2 L1 ./ and, for all A 2 A,
Z Z n
Z X 1 Z
X
f d D lim gn d D lim fk d D fn d:
A n!1 A n!1 A
kD1 nD1 A
Here the second step follows from Lemma 1.48 because gn D nkD1 fk almost
P
everywhere. The last step follows by interchanging sum and integral, using part (i)
of Theorem 1.44. This proves (1.38). To prove (1.39) note that
n
X
f fk D f gn
kD1
almost everywhere, that f .x/ gn .x/ converges to zero for all x 2 X , and that
jf gn j jf j C g, where jf j C g is integrable. Hence, by Lemma 1.48 and the
Lebesgue dominated convergence theorem (Theorem 1.45),
Z ˇ n
X ˇ Z
ˇ ˇ
lim ˇf fk ˇˇ d D lim jf gn j d D 0;
n!1 X ˇ n!1 X
kD1
satisfies (1.36). Hence, by Theorem 1.52, there exists a function g 2 L1 ./ such
that
X1 1
X
gD gi D .fniC1 fni /
i D1 iD1
Z ˇ kX1 ˇ Z
ˇ ˇ
0 D lim ˇ gi g ˇˇ d D lim jfnk fn1 gj d: (1.42)
k!1 k!1 X
ˇ
X i D1
Define
f WD fn1 C g:
Then
i 1
X
fni D fn1 C gj
j D1
converges almost everywhere to f . We prove (1.41). Let " > 0. By (1.42), there is
an ` 2 N such that
Z
jfnk f j d < "=2 for all k `.
X
Then Z Z Z
jfn f j d jfn fn` j d C jfn` f j d < "
X X X
N 2 A; .N / D 0; E N H) E 2 A:
jA D :
(iii) The triple .X; A ; / is a complete measure space. It is called the completion
of .X; A; /.
N WD ¹x 2 X j f .x/ ¤ f .x/º 2 A
This is the only possibility for defining a measure W A ! Œ0; 1 that agrees
with on A because .A/ D .B/ whenever A; B 2 A such that A B
and .B n A/ D 0. To prove that is well defined, let E 2 A and A; B 2 A as
in (1.44). If A0 ; B 0 2 A is another pair such that A0 E B 0 and .B 0 n A0 / D 0,
then A n A0 E n A0 B 0 n A0 and hence .A n A0 / D 0. This implies
.A/ D .A \ A0 / D .A0 /, where the last equality follows by interchanging
the roles of the pairs .A; B/ and .A0 ; B 0 /. Thus the map W A ! Œ0; 1 in (1.44)
is well defined.
We prove that is a measure. Let Ei 2 A be a sequence of pairwise disjoint
sets and choose sequences Ai ; Bi 2 A such that Ai Ei Bi for all i . Then the
Ai are pairwise disjoint and .Ei / D .Ai / for all i . Moreover, A WD i Ai 2 A,
S
part (i). Hence .E/ D .A/ D i .Ai / D i .Ei /. This proves (ii).
P P
0 s1 s2 f
and
f .x/ D lim sn .x/ for all x 2 X .
n!1
Since jA D we have
Z Z
sn d D sn d
X X
for all n and hence it follows from the Lebesgue monotone convergence theorem
(Theorem 1.37) for both and that
Z Z Z Z
f d D lim sn d D lim sn d D f d :
X n!1 X n!1 X X
This proves (1.43) for E D X and all A-measurable functions f W X ! Œ0; 1.
To prove it for all E replace f by f E and use part (ii) of Theorem 1.35. This
proves equation (1.43) for all A-measurable functions f W X ! Œ0; 1. That it
continues to hold for all f 2 L1 ./ follows directly from Definition 1.43. This
proves (iv).
(v) If f D E for E 2 A , choose A; B 2 A such that
A E B; .B n A/ D 0;
N D ¹x 2 X j f .x/ ¤ f .x/º D E n A B n A:
Corollary 1.56. Let .X; A; / be a measure space and let .X; A ; / be its com-
pletion. Denote the equivalence class of a -integrable function f 2 L1 ./ under
the equivalence relation (1.35) in Definition 1.51 by
As noted in Section 1.5, sets of measure zero can be neglected when integrating
functions. Hence it is sometimes convenient to enlarge the notion of integrability.
It is not even necessary that the function be defined on all of X , as long as it is
defined on the complement of a set of measure zero.
Thus, let .X; A; / be a measure space and call a function f W E ! R, defined on
a measurable subset E X, measurable if .X nE/ D 0 and the set f 1 .B/ E
is measurable for every Borel set B R. Call f integrable if the function on all
of X , obtained by setting f jX nE D 0, is integrable.
If .X; A; / is complete our integrable function f W E ! R can be extended in
any manner whatsoever to all of X , and the extended function on X is then inte-
grable in the original sense, regardless of the choice of the extension. Moreover,
its integral over any measurable set A 2 A is unaffected by the choice of the exten-
sion (see Lemma 1.48).
With this extended notion of integrability we see that the Lebesgue dominated
convergence theorem (Theorem 1.45) continues to hold if (1.31) is replaced by the
weaker assumption that fn only converges to f almost everywhere.
That such an extended terminology might be useful can also be seen in Theo-
rem 1.52, where the series 1
P
nD1 fn only converges on the complement of a set N
of measure zero, and the function f can only be naturally defined on E WD X n N .
Our choice in the proof of Theorem 1.52 was to define f jN WD 0, but this choice
1.7. Exercises 47
does not affect any of the statements of the theorem. Moreover, when working with
the quotient space L1 ./ D L1 ./= we are only interested in the equivalence
class of f under the equivalence relation (1.35) rather than a specific choice of an
element of this equivalence class.
1.7 Exercises
Exercise 1.57. Let X be an uncountable set and let A 2X be the set of all subsets
A X such either A or Ac is countable. Define
´
0 if A is countable;
.A/ WD
1 if Ac is countable;
Exercise 1.58. Let .X; A; / be a measure space such that .X/ < 1 and let
fn W X ! Œ0; 1/ be a sequence of bounded measurable functions that converges
uniformly to f W X ! Œ0; 1/. Prove that
Z Z
f d D lim fn d: (1.46)
X n!1 X
Prove that Z 1
lim fn .x/ dx D 0;
n!1 0
Exercise 1.63. Let .X; A; / be a measure space such that .X/ < 1 and let
f W X ! R be a measurable function. Show that f is integrable if and only if
1
X
j.¹x 2 X j jf .x/j > nº/j < 1:
nD1
(i) Prove that for every " > 0 there exists a ı > 0 such that, for any A 2 A,
ˇZ ˇ
ˇ ˇ
.A/ < ı H) ˇˇ f dˇˇ < ":
A
(ii) Prove that for every " > 0 there exists a measurable set A 2 A such that, for
any B 2 A, ˇZ ˇ
Z
ˇ ˇ
B A H) ˇ f d
ˇ f dˇˇ < ":
X B
Exercise 1.66. Let .X; A; / be a measure space such that D ıx0 is the
Dirac measure at some point x0 2 X (Example 1.31). Determine the completion
.X; A ; / and the space L1 ./.
Exercise 1.67. Let .X; A; / be a complete measure space. Prove that .X; A; / is
equal to its own completion.
Exercise 1.68. Let .X; A; / and .X; A0 ; 0 / be two measure spaces with A A0
and 0 jA D . Prove that L1 ./ L1 .0 / and
Z Z
f d D f d0
X X
(ii) Let 0 < c < 1, let f W X ! Œ0; c be A-measurable, and assume that
Then Z Z
f d D f d0 :
X X
(Consider also the function c f .)
A WD ¹B Y j 1
.B/ 2 Aº 2Y : (1.47)
W A ! Œ0; 1
defined by
1
. /.B/ WD . .B//; for B 2 A: (1.48)
(ii) Let .X; A ; / be the completion of .X; A; / and let .Y; . A/ ; . / /
be the completion of .Y; A; /. Prove that
(iii) Fix a function f W Y ! Œ0; 1. Prove that f is A-measurable if and only if
f ı is A-measurable. If f is A-measurable, prove that
Z Z
f d. / D .f ı / d: (1.50)
Y X
The following extended remark contains a brief introduction to some of the basic
concepts and terminology in probability theory. It will not be used elsewhere in this
book and can be skipped at first reading.
Remark 1.70 (probability theory). A probability space is a measure space .; F;P /
such that
P ./ D 1:
The underlying set is called the sample space, the -algebra F 2 is called the
set of events, and the measure
P W F ! Œ0; 1
Examples of infinite sample spaces are the set D N [ ¹1º for repeatedly tossing
a coin until the first tail shows up, a compact interval of real numbers for random
arrival times, and a disc in the plane for throwing a dart.
A random variable is an integrable function
XW ! R:
and Z
V.X/ WD .X E.X//2 dP D E.X 2 / E.X/2 ;
respectively.
Given a random variable X W ! R one is interested in the value of the prob-
ability measure on the set X 1 .B/ for a Borel set B R. This value is the prob-
ability of the event that the random variable X takes its value in the set B and is
denoted by
P .X 2 B/ WD P .X 1 .B// D .X P /.B/:
Here X P denotes the pushforward of the probability measure P to the Borel
-algebra B 2R (Exercise 1.69). By (1.50) the expectation and variance of X
are given by
Z
E.X/ D x d.X P /.x/
R
and Z
V.X/ D .x E.X//2 d.X P /.x/;
R
respectively.
The (cumulative) distribution function of a random variable X is the function
FX W R ! Œ0; 1
defined by
and the integral of a continuous function on R with respect to the pushforward mea-
sure X P agrees with the Riemann–Stieltjes integral (Exercise 6.20) with respect
to FX . Moreover,
FX .x/ lim FX .t/ D P .X 1 .x//
t !x
• converge in probability to X if
• converge in distribution to X if
Let E be the set of all ! 2 such that the sequence Xi .!/ does not con-
verge to X.!/. This set is measurable by Example 1.46 and has measure zero by
convergence almost everywhere. Moreover,
\ [
Aj E
i2N j i
and so [
lim P Aj D P .E/ D 0
i!1
j i
Then
FX .x ı/ P .jXi Xj ı/ FXi .x/
FX .x C ı/ C P .jXi X j ı/
and hence
FX .x/ " < FXi .x/ < FX .x/ C " for all i i0 .
This shows that
lim FXi .x/ D FX .x/
i !1
as claimed.
A finite collection of random variables X1 ; : : : ; Xn is called independent if,
for every collection of Borel sets B1 ; : : : ; Bn R,
n
\ Yn
P Xi 1 .Bi / D P .Xi 1 .Bi //:
iD1 i D1
In Chapter 7 we shall see that this condition asserts that the pushforward of P under
the map
X WD .X1 ; : : : ; Xn /W ! Rn
agrees with the product of the measures .Xi / P . Two foundational theorems in
probability theory are the law of large numbers and the central limit theorem. These
are results about sequences of random variables
Xk W ! R
For n 2 N define
Sn WD X1 C C Xn :
Kolmogorov’s strong law of large numbers asserts that, under these assumptions,
the sequence Sn =n converges almost everywhere to zero. (This continues to hold
when (c) is replaced by the assumption 1 1
P
kD1 k 2 V.Xk / < 1.) If, in addition,
2
V.Xk / D for all k and some positive real number , then the central limit the-
orem of Lindeberg and Lévy asserts that the sequence
p
Tn WD Sn = n
1.7. Exercises 55
This chapter introduces the most important example, namely the Lebesgue mea-
n
sure on Euclidean space. Let n 2 N and denote by B 2R the -algebra of all
Borel sets in Rn , i.e., the smallest -algebra on Rn that contains all open sets in the
standard topology (Definition 1.15). Then
B C x WD ¹y C x j y 2 Bº 2 B for all B 2 B and all x 2 Rn ;
because the translation Rn ! Rn W y 7! y C x is a homeomorphism. A measure
W B ! Œ0; 1 is called translation-invariant if it satisfies
.B C x/ D .B/ for all B 2 B and all x 2 Rn : (2.1)
Theorem 2.1. There exists a unique measure W B ! Œ0; 1 that is translation-
invariant and satisfies the normalization condition .Œ0; 1n / D 1.
Proof. See page 74.
Definition 2.2. Let .Rn ; B; / be the measure space in Theorem 2.1 and denote by
.Rn ; A; m/ its completion as in Theorem 1.55. Thus
n ˇ there exist Borel sets B0 ; B1 2 B
² ˇ ³
A WD A R ˇ
ˇ
(2.2)
such that B0 A B1 and .B1 n B0 / D 0
and
m.A/ WD .B0 / for A 2 A,
where B0 ; B1 2 B are chosen such that B0 A B1 and .B1 n B0 / D 0.
The elements of A are called Lebesgue measurable subsets of Rn , the function
mW A ! Œ0; 1
is called the Lebesgue measure, and the triple .Rn ; A; m/ is called the Lebesgue
measure space. A function
f W Rn ! R
is called Lebesgue measurable if it is measurable with respect to the Lebesgue
-algebra A on Rn (and the Borel -algebra on the target space R).
58 2. The Lebesgue measure
Step 1. X 2 A.
Step 2. If A 2 A, then Ac 2 A.
Let A 2 A. Since
D \ Ac D D n A; D n Ac D D \ A;
Step 3. If A; B 2 A, then A [ B 2 A.
Here the inequality follows from axioms (a) and (c) in Definition 2.3. Using ax-
ioms (a) and (c) again, we obtain
For k 2 N define
Bk WD A1 [ A2 [ [ Ak :
Then Bk 2 A for all k 2 N by Step 3. Now let D X . Then, for all k 2,
.D \ Bk / D .D \ Bk \ Ak / C ..D \ Bk / n Ak /
D .D \ Ak / C .D \ Bk 1/
Here the last inequality follows from axiom (b) in Definition 2.3. Since this holds
for all k 2 N and D \ A D 1
S
iD1 .D \ Ai /, it follows that
1
X
.D/ .D \ Ai / C .D n A/
i D1
.D \ A/ C .D n A/
.D/:
Here the last two inequalities follow from axiom (c). Hence
1
X
.D/ D .D \ Ai / C .D n A/ D .D \ A/ C .D n A/ (2.5)
iD1
Define ´
A1 for i D 1,
Bi D
Ai n .A1 [ [ Ai 1/ for i 2.
Then Bi \ Bj D ; for i ¤ j and Bi D .A1 [ [ Ai 1 [ Aci /c 2 A for all i by
Steps 2 and 3. Hence A D 1i D1 Bi 2 A by Step 4.
S
Proof. We prove that (i) implies (ii). Thus assume that satisfies (i). Let A; B X
such that " WD d.A; B/ > 0. Define
[
U WD ¹x 2 X j there exists a 2 A such that d.a; x/ < "º D B" .a/:
a2A
We prove that (ii) implies (i). Thus assume that satisfies (ii). We show that
every closed set A X is -measurable, i.e., .D/ D .D \ A/ C .D n A/ for all
D X . Since .D/ .D \ A/ C .D n A/, by definition of an outer measure,
it suffices to prove the following.
Claim 1. Fix a closed set A X and a set D X such that .D/ < 1. Then
.D/ .D \ A/ C .D n A/:
To see this, replace the set D n A by the smaller set D n Uk , where
[
Uk WD ¹x 2 X j there exists a 2 A such that d.a; x/ < 1=kº D B1=k .a/:
a2A
For each k 2 N, the set Uk is open and d.x; y/ 1=k for all x 2 D \ A and all
y 2 D n Uk . Hence
1
d.D \ A; D n Uk / :
k
By (ii) and axiom (b) this implies
.D \ A/ C .D n Uk / D ..D \ A/ [ .D n Uk // .D/ (2.6)
for every subset D X and every k 2 N. We will prove the following.
Claim 2. lim .D n Uk / D .D n A/.
k!1
Claim 1 follows directly from Claim 2 and (2.6). To prove Claim 2 note that
1
\
AD Ui ;
iD1
and hence
D n A D .D n Uk / [ .D \ .Uk n A//
1
[
D .D n Uk / [ .D \ .Ui n UiC1 //:
i Dk
Thus
1
[
D n A D .D n Uk / [ Ei ; Ei WD .D \ Ui / n Ui C1 : (2.7)
i Dk
2.1. Outer measures 63
converges to zero. Moreover, it follows from equation (2.7) and axiom (c) in Defi-
nition 2.3 that
1
X
.D n A/ .D n Uk / C .Ei / D .D n Uk / C "k :
i Dk
for every k 2 N. Since "k converges to zero, this implies Claim 2. The proof of
Claim 3 relies on the next assertion.
Claim 4 H) Claim 3. It follows from Claim 4, axiom (b), and (ii) that
n
X n
[
.E2i / D E2i .D/
i D1 iD1
and
n
X n
[
.E2i 1/ D E2i 1 .D/
i D1 iD1
is defined by
Vol.U / WD Vol.Q/:
The set of all closed cuboids in Rn will be denoted by
² ˇ ³
ˇ a ; : : : ; an ; b1 ; : : : ; bn 2 R;
Qn WD Q.a; b/ ˇˇ 1 :
aj < bj for j D 1; : : : ; n
Definition 2.7. A subset A Rn is called a Jordan null set if, for every " > 0,
there exist finitely many closed cuboids Q1 ; : : : ; Q` 2 Qn such that
`
[ `
X
A Qi ; Vol.Qi / < ":
iD1 iD1
Definition 2.8. A subset A Rn is called a Lebesgue null set if, for every " > 0,
there is a sequence of closed cuboids Qi 2 Qn , i 2 N, such that
1
[ 1
X
A Qi ; Vol.Qi / < ":
iD1 iD1
(iii) if A; B Rn are such that d.A; B/ > 0, then .A [ B/ D .A/ C .B/;
(iv) .int.Q// D .Q/ D Vol.Q/ for all Q 2 Qn .
66 2. The Lebesgue measure
and fix a constant " > 0. Then it follows from Definition 2.9 that, for i 2 N, there
exists a sequence of cuboids Qij 2 Qn , j 2 N, such that
1 1
[ X "
Ai Qij ; Vol.Qij / < C .Ai /:
2i
j D1 j D1
Hence
1 1
[ X X " X
A Qij ; Vol.Qij / < C .A i / D " C .Ai /:
2i
i;j 2N i;j 2N i D1 iD1
This implies
1
X
.A/ < " C .Ai /
iD1
Here jj denotes the Euclidean norm on Rn . Then, for every i 2 N, we have either
Qi \ A D ; or Qi \ B D ;. This implies
I \ J D ;; I WD ¹i 2 N j Qi \ A ¤ ;º; J WD ¹i 2 N j Qi \ B ¤ ;º:
2.2. The Lebesgue outer measure 67
Hence
X X
.A/ C .B/ Vol.Qi / C Vol.Qi /
i 2I i 2J
1
X
Vol.Qi /
iD1
Thus .A/ C .B/ < .A [ B/ C " for all " > 0, so .A/ C .B/ .A [ B/,
and hence .A/ C .B/ D .A [ B/, by axioms (a) and (c) in Definition 2.3.
This proves part (iii).
We prove (iv) by an argument due to von Neumann. Fix a closed cuboid
Q D I1 In ; Ii D Œai ; bi :
We claim that
Vol.Q/ .Q/: (2.11)
Equivalently, if Qi 2 Qn , i 2 N, is a sequence of closed cuboids, then
1
[ 1
X
Q Qi H) Vol.Q/ Vol.Qi /: (2.12)
iD1 i D1
jI j WD b a:
Then
jI j 1 #.I \ Z/ jI j C 1:
Hence
N jI j 1 #.NI \ Z/ N jI j C 1;
and thus
1 1 1 1
jI j # I \ Z jI j C
N N N N
for every integer N 2 N. Take the limit N ! 1 to obtain
1 1
jI j D lim # I\ Z :
N !1 N N
68 2. The Lebesgue measure
Thus
n
Y 1 1
Vol.Q/ D lim # Ij \ Z
N !1 N N
j D1 (2.13)
1 1 n
D lim # Q \ Z :
N !1 N n N
Now suppose Qi 2 Qn , i 2 N, is a sequence of closed cuboids such that
1
[
Q Qi :
iD1
Fix a constant " > 0 and choose a sequence of open cuboids Ui Rn such that
"
Qi Ui ; Vol.Ui / < Vol.Qi / C i :
2
Since Q is compact and the Ui form an open cover of Q, there exists a constant
k 2 N such that
[k
Q Ui :
iD1
This implies
k
1 1 n X 1 1 n
# Q \ Z # U i \ Z
Nn N Nn N
i D1
k
X 1 x 1 n
# U i \ Z :
Nn N
i D1
Since " > 0 can be chosen arbitrarily small, this proves (2.12) and (2.11). Thus we
have proved that .Q/ Vol.Q/ .Q/, and so .Q/ D Vol.Q/. To prove that
.int.Q// D Vol.Q/, fix a constant " > 0 and choose a closed cuboid P 2 Qn such
that
P int.Q/; Vol.Q/ " < Vol.P /:
Then
Vol.Q/ " < Vol.P / D .P / .int.Q//:
Thus Vol.Q/ " < .int.Q// for all " > 0. Therefore, by axiom (b),
and hence .int.Q// D Vol.Q/. This proves part (iv) and Theorem 2.10.
Rn
Definition 2.11. Let W 2 ! Œ0; 1 be the Lebesgue outer measure. A subset
A Rn is called Lebesgue measurable if A is -measurable, i.e.,
A WD ¹A Rn j A is Lebesgue measurableº:
The function
m WD jA W A ! Œ0; 1
is called the Lebesgue measure on Rn . A function f W Rn ! R is called Lebesgue
measurable if it is measurable with respect to the Lebesgue -algebra A on Rn
(and the Borel -algebra on the target space R).
Corollary 2.12. (i) .Rn ; A; m/ is a complete measure space.
(ii) m is translation-invariant, i.e., if A 2 A and x 2 Rn , then
Proof. Assertion (i) follows from Theorem 2.4 and part (i) of Theorem 2.10. Asser-
tion (ii) follows from the definitions and part (ii) of Theorem 2.10. Assertion (iii)
follows from Theorem 2.5 and part (iii) of Theorem 2.10. Finally, assertion (iv)
follows from (iii) and part (iv) of Theorem 2.10.
70 2. The Lebesgue measure
Proof. (i) Fix a subset A Rn and a constant " > 0. The assertion is obvious when
.A/ D 1. Hence assume .A/ < 1 and choose a sequence of closed cuboids
Qi 2 Qn such that
1 1
[ X "
A Qi ; Vol.Qi / < .A/ C :
2
iD1 iD1
(iii) If .Ai / D 1 for some i , then the assertion is obvious. Hence assume
.Ai / < 1 for all i and fix a constant " > 0. By (i), there is a sequence of open
sets Ui Rn such that .Ui / < .Ai / C "2 i for all i. Since Ai Ui \ UiC1 ,
this yields
.UiC1 n Ui / D .UiC1 / .Ui \ Ui C1 /
i 1
< .AiC1 / .Ai / C "2
for all i 2 N. It follows that
[ k k
X1
Ui D .U1 / C .UiC1 n Ui / < .Ak / C ":
i D1 iD1
Thus
.A/ lim .Ak / C "; for all " > 0,
k!1
and so
.A/ lim .Ak /:
k!1
The converse inequality is obvious. This proves Theorem 2.13.
72 2. The Lebesgue measure
n
Theorem 2.14 (the Lebesgue measure as a completion). Let W 2R ! Œ0; 1 be
the Lebesgue outer measure in Definition 2.9, let
m D jA W A ! Œ0; 1
(I) A 2 A, i.e.,
B0 A B1 and .B1 n B0 / D 0:
and hence
m.A/ D .A/ D .B0 / D .A/:
This shows that A D B and m D . Thus it remains to prove the claim. Fix a
subset A Rn .
We show that (II) implies (I). Thus assume that A 2 B and choose Borel mea-
surable sets B0 ; B1 2 B such that
B0 A B1 ; .B1 n B0 / D 0:
Define
1
[ 1
\
B0 WD Ki ; B1 WD Ui :
i D1 i D1
1 1
.A/ .Ki / .B0 / .B1 / .Ui / .A/ C :
i i
Take the limit i ! 1 to obtain
hence
.B0 / D .B1 / D .A/ < 1;
and hence
.B1 n B0 / D .B1 / .B0 / D 0:
This shows that A 2 B for every A 2 A with .A/ < 1.
Now suppose that our set A 2 A satisfies .A/ D 1 and define
Then Ak 2 A and .Ak / .2k/n for all k. Hence Ak 2 B for all k and so there
exist sequences of Borel sets Bk ; Bk0 2 B such that
Define
1
[ 1
[
B WD Bk and B 0 WD Bk0 :
kD1 kD1
Then B; B 0 2 B, B A B 0 , and
1
X 1
X
0
.B n B/ .Bk0 n B/ .Bk0 n Bk / D 0:
kD1 kD1
This shows that A 2 B for every A 2 A. Thus we have proved that (I) implies (II)
and this completes the proof of Theorem 2.14.
74 2. The Lebesgue measure
this implies
X
D 0 .R.x; 0// D 0 .R.x C 2 k
`; k// D 2nk ck :
`2Zn ; 0`j 2k 1
Hence ck D 2 nk D .R.x; k//. Here the last equation follows from the fact
that .0; 2 k /n R.0; k/ Œ0; 2 k n and hence .R.x; k// D .R.0; k// D 2 k
by part (iv) of Corollary 2.12.
Step 2. 0 .U / D .U / for every open set U Rn .
Let U Rn be open. We prove that U can be expressed as a countable union of
sets Ri D R.xi ; ki / as in Step 1. To see this, define
R0 WD ¹R.x; 0/ j x 2 Zn ; R.x; 0/ U º;
ˇ x 2 2 1 Zn ; R.x; 1/ U;
² ˇ ³
R1 WD R.x; 1/ ˇ ;
R.x; 1/ 6 R, for all R 2 R0
ˇ
ˇ x 2 2 k Zn ; R.x; k/ U;
² ˇ ³
Rk WD R.x; k/ ˇˇ for k 2,
R.x; k/ 6 R, for all R 2 R0 [ R1 [ [ Rk 1
and denote
1
[
R WD Rk :
kD0
2.2. The Lebesgue outer measure 75
Step 5. D 1 and 0 D .
Lemma 2.15. Let A R be a Lebesgue measurable set such that m.A/ > 0.
Then there exists a set B A that is not Lebesgue measurable.
Proof. Consider the equivalence relation on R defined by
def
x y () x y2Q
Bq WD A \ .E C q/ D ¹x 2 A j x q 2 Eº:
S
Then it follows from (I) that A D q2Q Bq :
Fix a rational number q 2 Q. We prove that if Bq is Lebesgue measurable, then
m.Bq / D 0. Indeed, let n 2 N, and define
This sum is infinite and all summands agree, so m.Bq \ Œ n; n/ D 0. This holds
for all n 2 N, and hence m.Bq / D 0, as claimed.
S
If Bq is Lebesgue measurable for all q 2 Q, it follows that A D q2Q Bq
is a Lebesgue null set, a contradiction. Thus one of the sets Bq is not Lebesgue
measurable and this proves Lemma 2.15.
Remark 2.16. (i) Using Lemma 2.15 one can construct a continuous function
f W R ! R and a Lebesgue measurable function gW R ! R such that the com-
position g ı f is not Lebesgue measurable (see Example 6.24).
(ii) Let E R be the set constructed in the proof of Lemma 2.15. Then the set
E R R2 is not Lebesgue measurable. This follows from a similar argument
as in Lemma 2.15 using the sets ..E \ Œ n; n/ C q/ Œ0; 1. On the other hand,
the set E ¹0º R2 is Lebesgue measurable and has Lebesgue measure zero.
However, it is not a Borel set, because its pre-image in R under the continuous map
R ! R2 W x 7! .x; 0/ is the original set E and hence is not a Borel set.
2.3. The transformation formula 77
AX WD ¹A 2 A j A X º
mX WD mjAX W AX ! Œ0; 1
The proof of Theorem 2.17 relies on the next two lemmas. The first lemma is
the special case where is linear and f is the characteristic function of a Lebesgue
measurable set. The second lemma is a basic estimate that follows from the linear
case and implies formula (2.14) for the characteristic functions of open sets.
WD mjB
the restriction of the Lebesgue measure to the Borel -algebra. Thus is the unique
translation-invariant Borel measure on Rn that satisfies the normalization condi-
tion .Œ0; 1n / D 1 (Theorem 2.1) and .Rn ; A; m/ is the completion of .Rn ; B; /
(Theorem 2.14).
Step 1. There exists a unique map
W GL.n; R/ ! .0; 1/
such that
.ˆ.B// D .ˆ/.B/ (2.17)
for every ˆ 2 GL.n; R/ and every Borel set B 2 B.
Fix a vector space isomorphism ˆW Rn ! Rn . Since ˆ is a homeomorphism of Rn
with its standard topology, it follows that ˆ.B/ 2 B for every B 2 B. Define the
number .ˆ/ 2 Œ0; 1 by
Since ˆ.Œ0; 1/n / has nonempty interior, it follows that .ˆ/ > 0 and since ˆ.Œ0; 1/n /
is contained in the compact set ˆ.Œ0; 1n /, it follows that .ˆ/ < 1. Now define
the map ˆ W B ! Œ0; 1 by
.ˆ.B//
ˆ .B/ WD for B 2 B:
.ˆ/
Then ˆ is a normalized translation-invariant Borel measure. The -additivity
follows directly from the -additivity of , the formula ˆ .;/ D 0 is obvious
from the definition, that compact sets have finite measure follows from the fact
that ˆ.K/ is compact if and only if K Rn is compact, the translation invari-
ance follows immediately from the translation invariance of and the fact that
ˆ.B C x/ D ˆ.B/ C ˆ.x/ for all B 2 B and all x 2 Rn , and the normaliza-
tion condition ˆ .Œ0; 1/n / D 1 follows directly from the definition of ˆ . Hence
ˆ D by Theorem 2.1.
2.3. The transformation formula 79
Step 2. Let be as in Step 1 and let A 2 A and ˆ 2 GL.n; R/. Then ˆ.A/ 2 A
and m.ˆ.A// D .ˆ/m.A/.
By Theorem 2.14, there exist Borel sets B0 ; B1 2 B such that B0 A B1 and
.B1 n B0 / D 0. Then ˆ.B0 / ˆ.A/ ˆ.B1 / and, by Step 1,
.ˆ.B1 / n ˆ.B0 // D .ˆ.B1 n B0 // D .ˆ/.B1 n B0 / D 0:
Hence ˆ.A/ is a Lebesgue measurable set and
m.ˆ.A// D .ˆ.B0 // D .ˆ/.B0 / D .ˆ/m.A/
by Theorem 2.14 and Step 1.
Step 3. Let be as in Step 1 and let ˆ D diag.1 ; : : : ; n / be a diagonal matrix
with nonzero diagonal entries i 2 R n ¹0º. Then .ˆ/ D j1 j jn j.
Define I WD Œ 1; 1 and Ii WD Œ ji j; ji j for i D 1; : : : ; n. Then Q WD I n has
Lebesgue measure m.Q/ D 2n and the cuboid ˆ.Q/ D I1 In has Lebesgue
measure m.ˆ.Q// D 2n j1 j jn j by part (iv) of Corollary 2.12. Hence Step 3
follows from Step 2.
Step 4. The map W GL.n; R/ ! .0; 1/ in Step 1 is a group homomorphism from
the general linear group of automorphisms of Rn to the multiplicative group of
positive real numbers.
Let ˆ; ‰ 2 GL.n; R/. Then it follows from (2.17) with B WD ‰.Œ0; 1/n / and from
the definition of .‰/ in (2.18) that
.ˆ‰/ D .ˆ‰.Œ0; 1/n // D .ˆ/.‰.Œ0; 1/n // D .ˆ/.‰/:
Thus is a group homomorphism as claimed and this proves Step 4.
Step 5. The map W GL.n; R/ ! .0; 1/ in Step 1 is continuous with respect to the
standard topologies on GL.n; R/ and .0; 1/.
It suffices to prove continuity at the identity. Define the norms
kˆxk1
kxk1 WD max jxi j ; kˆk1 WD sup (2.19)
iD1;:::;n 0¤x2Rn kxk1
for x 2 Rn and a linear map ˆW Rn ! Rn . Denote the closed unit ball in Rn by
Q WD ¹x 2 Rn j kxk1 1º D Œ 1; 1n . Fix a constant 0 < ı < 1 and a linear
map ˆW Rn ! Rn such that kˆ 1k1 < ı. Then ˆ 2 GL.n; R/ and
1
1
X 1
ˆ D .1 ˆ/k ; kˆ 1
k1 < :
1 ı
kD0
80 2. The Lebesgue measure
1
Thus ˆ.Q/ .1 C ı/Q and .1 ı/ˆ .Q/ Q. Hence
by
d1 .x; y/ WD kx yk1 for x; y 2 Rn ,
where kk1 is as in (2.19). The open ball of radius r > 0 about a point
a D .a1 ; : : : ; an / 2 Rn with respect to this metric is the open cube
Lemma 2.19. Let U Rn be an open set and let K U be a compact subset. Let
W U ! Rn be a continuously differentiable map such that det.d.x// ¤ 0 for all
x 2 K. For every " > 0 there exists a constant ı > 0 such that the following holds:
if 0 < r < ı, a 2 Rn , and R Rn satisfy Br .a/ R Bxr .a/ K, then
Let " > 0 and choose a constant 0 < ˛ < 1 so small that
" "
1 < .1 ˛/n < .1 C ˛/n < 1 C : (2.22)
c c
Choose ı > 0 so small that, for all x; y 2 Rn ,
˛
x; y 2 K; kx yk1 < ı H) kd.x/ d.y/k1 < : (2.23)
c
Such a constant exists because the map dW U ! Rnn is uniformly continuous on
the compact set K U . We prove that the assertion of Lemma 2.19 holds with this
constant ı.
Choose a 2 Rn and 0 < r < ı such that Bxr .a/ K. Then ka xk1 < ı for
all x 2 Bxr .a/. By (2.23) with ˆ WD d.a/ this implies
˛ ˛
kd.x/ ˆk1 < for all x 2 Bxr .a/:
c kˆ 1 k1
Here the first step follows from (2.23) and the second step follows from (2.21).
Define the map
W U ! Rn
by
1
.x/ WD ˆ ..x/ .a//:
1
Then .a/ D 0 and d .x/ D ˆ d.x/ and hence, by (2.23),
1 1
kd .x/ 1k1 D
ˆ .d.x/ ˆ/
1
ˆ 1
kd.x/ ˆk1 ˛
B.1 ˛/s .0/ .Br .a// .Bxr .a// Bx.1C˛/s .0/ (2.24)
Now fix a subset R Rn such that Br .a/ R Bxr .a/. Then by (2.24)
Since m.R/ D m.Br .0// D m.Bxr .0// by part (iv) of Corollary 2.12, it follows
from Lemma 2.18 and the inequalities (2.21) and (2.22) that
"
jdet.ˆ/j m.R/ " m.R/ 1 jdet.ˆ/j m.R/
c
< .1 ˛/n jdet.ˆ/j m.R/
D m..1 ˛/ˆ.Br .0///
m..R//
m..1 C ˛/ˆ.Bxr .0///
D .1 C ˛/n jdet.ˆ/j m.R/
"
< 1C jdet.ˆ/j m.R/
c
jdet.ˆ/j m.R/ C " m.R/:
Proof of Theorem 2.17. The proof has seven steps. The first four steps establish
equation (2.14) for the characteristic functions of open sets, compact sets, Borel
sets, and Lebesgue measurable sets with compact closure in U .
Fix a constant " > 0. Then there exists a constant ı > 0 that satisfies the following
two conditions.
" m.R/
jm..R// jdet.d.a//j m.R/j < :
2 m.W /
S
(b) For all x; y 2 W
"
kx yk1 < ı H) jdet.d.x// det.d.y//j < :
2 m.W /
2.3. The transformation formula 83
That ı > 0 can be chosen so small that (a) holds follows from Lemma 2.19, and
that it can be chosen so small that (b) holds follows from the fact that the function
det.d/W U ! R is uniformly continuous on the compact set W S . Now write W as a
countable union of pairwise disjoint half-open cubes Ri Rn centered at ai 2 Rn
with side lengths 2ri such that 0 < ri < ı. (See page 74.) Then
Bri .ai / Ri Bxri .ai / W
S for all i
and X X
m.W / D m.Ri /; m..W // D m..Ri //: (2.25)
i i
It follows from (2.25) and (a) that
ˇ X ˇ "
m..W // jdet.d.a //j m.Ri ˇ< :
/
ˇ ˇ
i (2.26)
2
ˇ
i
S U . Then
Choose an open set W K with compact closure W
m..K// D m..W // m..W n K//
Z Z
D jdet.d/j d m jdet.d/j d m
W W nK
Z
D jdet.d/j d m:
K
That .B/ is a Borel set follows from the fact that it is the pre-image of the Borel
set B under the continuous map 1 W V ! U (Theorem 1.20). Abbreviate
b WD m..B//:
Assume first that b < 1 and fix a constant " > 0. Then it follows from Theo-
rem 2.13 that there exist an open set W 0 Rn with compact closure W 0 V
such that .B/ W 0 and m.W 0 / < b C " and a compact set K 0 B such that
.K 0 / > b ". Define K WD 1 .K 0 / and W WD 1 .W 0 /. Then K is compact,
W is open, WS U is compact, and
Thus Z
b "< jdet.d/j d m < b C " for every " > 0,
B
and so Z
jdet.d/j d m D b D m..B//:
B
If b D 1, then by Theorem 2.13, there exists a sequence of compact sets
Ki0 .B/ such that .Ki0 / > i. Hence
1
Ki WD .Ki0 /
is compact and Z
jdet.d/j d m D ..Ki // > i
Ki
and hence Z
jdet.d/j d m D 1 D m..B//:
B
2.3. The transformation formula 85
Now Theorem 2.14 shows that .A/ is a Lebesgue measurable set and
Z Z
m..A// D m..B0 // D jdet.d/j d m D jdet.d/j d m:
B0 A
Here the last equation follows from the fact that the set A n B0 is Lebesgue measur-
able and has Lebesgue measure zero.
Step 5. Assertion (i) of Theorem 2.17 holds for every Lebesgue measurable step
function f D sW V ! R whose support is a compact subset of V .
Write
`
X
sD ˛i A
i
iD1
Z `
X Z
.s ı /jdet.d/j d m D ˛i jdet.d/j d m
U 1 .A
iD1 i/
`
X
D ˛i m.Ai /
iD1
Z
D s d m:
V
si W V ! Œ0; 1/
such that
0 s1 s2
and
f .x/ D lim si .x/ for every x 2 V .
i!1
For E D U part (ii) follows from part (i) and the fact that .f ı /˙ D f ˙ ı .
If F 2 AV , then f WD F jV is Lebesgue measurable, hence f ı D 1 .F / jU
is Lebesgue measurable by part (i), and so 1 .F / 2 AU . Replace by 1 to
deduce that if E 2 AU , then .E/ 2 AV . Then (ii) follows for all E 2 AU by
replacing f with f .E / . This proves Step 7 and Theorem 2.17.
which was used in the proof of Theorem 2.1 on page 74. The closure of R.x; k/ is
the closed cube R.x; k/ D x C Œ0; 2 k n : The sets R.`; k/, with ` ranging over the
countable set 2 k Zn , form a partition of the Euclidean space Rn .
2.4. Lebesgue equals Riemann 87
supp.f / WD ¹x 2 Rn j f .x/ ¤ 0º
is a bounded subset of Rn . For k 2 N define the lower sum S.f; k/ 2 R and the
x k/ 2 R by
upper sum S.f; x
X X
S.f; k/ WD x k/ WD
inf f 2 nk ; S.f; sup f 2 nk : (2.28)
x k n R.`;k/ k n R.`;k/
`22 Z `22 Z
Remark 2.21. The Riemann integral can also be defined by allowing for arbitrary
partitions of Rn into cuboids (see [19, Definition 2.3]) or in terms of convergence of
the so-called Riemann sums (see [21, Definition 7.1.2]). That all three definitions
agree is proved in [19, Satz 2.8] and [21, Theorem 7.1.8]).
Definition 2.22. A bounded set A Rn is called Jordan measurable if its char-
acteristic function A W Rn ! R is Riemann integrable. The Jordan measure of a
Jordan measurable set A Rn is the real number
J .A/ WD R.A /
Z
D A .x/ dx (2.30)
Rn
nk k
D lim 2 #¹` 2 2 Zn j R.`; k/ \ A ¤ ;º:
k!1
m.A/ D J .A/:
f k ; fNk W Rn ! R
N
by
f ; fNW Rn ! R
N
by
f .x/ WD lim f k .x/; fN.x/ WD lim fNk .x/ for x 2 Rn :
N k!1 N k!1
Then
f .x/ f .x/ fN.x/
N
for every x 2 Rn . Moreover, jfNk j and jf k j are bounded above by the Lebesgue
integrable function cA , where N
c WD sup jf .x/j
x2Rn
2.4. Lebesgue equals Riemann 89
and
A WD Œ N; N n ;
with N 2 N chosen such that supp.f / Œ N; N n . Hence it follows from
the Lebesgue dominated convergence theorem (Theorem 1.45) that f and fN are
Lebesgue integrable and N
Z Z
f d m D lim f dm
Rn N k!1 Rn N k
D lim S.f; k/
k!1 x
D R.f /
x k/
D lim S.f;
k!1
Z
D lim fNk d m
k!1 Rn
Z
D fN d m:
Rn
provided that the limit exists. Here Br Rn denotes the ball of radius r centered
at the origin. There are
R many examples where the limit (2.32) exists even though
the Lebesgue integral Rn jf j d m is infinite and so the Lebesgue integral of f does
not exist. An example is the function f W R ! R given by f .x/ WD x 1 sin.x/
for x 2 R n ¹0º and f .0/ WD 1. This function is continuous and is not Lebesgue
integrable, but the improper Riemann integral exists and is equal to (see Exam-
ple 7.49). Improper integrals play a central role in Fourier analysis, probability
theory, and partial differential equations. However, this topic will not be pursued
any further in this book.
90 2. The Lebesgue measure
2.5 Exercises
Exercise 2.25. Show that the Cantor set in R is a Jordan null set. Show that Q\Œ0; 1
is a Lebesgue null set, but not a Jordan null set. Show that A Rn is a Lebesgue
null set if and only if .A/ D 0. Find an open set U R whose boundary has
positive Lebesgue measure.
Exercise 2.26. Prove that every subset of a proper linear subspace of Rn is Lebesgue
measurable and has Lebesgue measure zero. Find a Jordan measurable subset of
Rn that is not a Borel set. Find a bounded Lebesgue measurable subset of Rn with
positive Lebesgue measure that is neither a Borel set, nor Jordan measurable.
Exercise 2.27. Let A; B Rn be Lebesgue null sets. Prove that their sum
A C B WD ¹x C y j x 2 A; y 2 Bº
is a Lebesgue null set.
Exercise 2.28. Let .X; A; / be a measure space and define the function
W 2X ! Œ0; 1
by
.B/ WD inf¹.A/ j A 2 A; B Aº: (2.33)
(i) Prove that is an outer measure and that A A./.
(ii) Assume .X/ < 1. Prove that the measure space .X; A./; jA./ / is the
completion of .X; A; /. Hint. Show that for every subset B X there exists
a set A 2 A such that B A and .B/ D .A/.
(iii) Let X be a set and A ¨ X be a nonempty subset. Define
A WD ¹;; A; Ac ; Xº; .;/ WD .A/ WD 0; .Ac / WD .X/ WD 1:
Prove that .X; A; / is a measure space. Given B X , prove that .B/ D 0
whenever B A and .B/ D 1 whenever B 6 A. Prove that A./ D 2X
and that the completion of .X; A; / is the measure space .X; A ; / with
A D ¹B X j B A or Ac Bº and D jA . (Thus the hypothesis
.X/ < 1 cannot be removed in part (ii).)
Exercise 2.29. Let f W R ! R be continuously differentiable and define
A WD ¹x 2 R j f 0 .x/ D 0º:
Prove that f .A/ is a Lebesgue null set. Hint. Consider the sets
An;" WD ¹x 2 R j jxj < n; jf 0 .x/j < 2 n
"º:
2.5. Exercises 91
Exercise 2.32. Let .R; A; m/ be the Lebesgue measure space. Construct a Borel
set E R such that
m.E \ I /
0< <1
m.I /
for every nonempty bounded open interval I R. (See also Exercise 6.21.)
exist for every Lebesgue integrable function f W Œ0; 1 ! R? Determine the limit
when it does exist.
Exercise 2.34. Let .Rn ; A; m/ be the Lebesgue measure space and let W Rn ! Rn
be a C 1 -diffeomorphism. Prove that A D A and that
1
Z
. m/.A/ D d m for all A 2 A:
A jdet.d/ ı 1j
Exercise 2.35 (Hausdorff measure). Let .X; / be a metric space and fix a real
number d 0. The diameter of a subset A X is defined by
d;" W 2X ! Œ0; 1
by
8 ˇ 9
<X ˇ I is finite or countably infinite; =
diam.Di /d ˇˇ Di X; diam.Di / < " for i 2 I; :
ˇ
d;" .A/ WD inf (2.35)
:
i2I ˇ and A S Di ;
i2I
for A X. Thus d;" .;/ D 0 and d;" .A/ D 1 whenever A does not admit a
countable cover by subsets of diameter less than ". Moreover, the function
" 7! d;" .A/ is nonincreasing for every subset A X . The d -dimensional Haus-
dorff outer measure is the function
d W 2X ! Œ0; 1
defined by
d .A/ WD sup d;" .A/ D lim d;" .A/ for A X: (2.36)
">0 "!0
then
d .A [ B/ D d .A/ C d .B/:
Hence, by Theorems 2.4 and 2.5, the set
Ad WD ¹A X j A is d -measurableº
d WD d jAd W Ad ! Œ0; 1
The second equality follows from the fact that d .A/ > 0 implies r .A/ D 1
for 0 r < d , and d .A/ < 1 implies s .A/ D 0 for s > d .
The regularity properties established for the Lebesgue (outer) measure in Theo-
rem 2.13 play an important role in much greater generality. The present chapter is
devoted to the study of Borel measures on locally compact Hausdorff spaces that
satisfy similar regularity properties. The main result is the Riesz representation the-
orem (Theorem 3.15). We begin with some further recollections about topological
spaces.
Let .X; U/ be a topological space (see Definition 1.9). A neighborhood of a
point x 2 X is a subset A X that contains x in its interior, i.e., x 2 U A
for some open set U . X is called a Hausdorff space if any two distinct points in X
have disjoint neighborhoods, i.e., for all x; y 2 X with x ¤ y there exist open sets
U; V X such that x 2 U , y 2 V , and U \ V D ;. X is called locally compact if
every point in X has a compact neighborhood. It is called -compact if there exists
a sequence of compact sets Ki X, i 2 N, such that Ki Ki C1 for all i and
XD 1
S
iD1 Ki .
.K/ < 1
for every Borel set B 2 B, and is called regular if it is both outer and inner regular.
A Radon measure is an inner regular Borel measure.
Example 3.2. The restriction of the Lebesgue measure on X D Rn to the Borel
-algebra is a regular Borel measure by Theorem 2.13.
Example 3.3. The counting measure on X D N with the discrete topology
U D B D 2N is a regular Borel measure.
Example 3.4. Let .X; U/ be any locally compact Hausdorff space and fix a point
x0 2 X . Then the Dirac measure D ıx0 at x0 is a regular Borel measure (Exam-
ple 1.31).
Example 3.5. Let X be an uncountable set equipped with the discrete topology
U D B D 2X . Define
W B ! Œ0; 1
by
´
0 if B is countable;
.B/ WD
1 if B is uncountable:
This is a Borel measure. Moreover, a subset K X is compact if and only if it is
finite. Hence .X/ D 1 and .K/ D 0 for every compact set K X . Thus is
not a Radon measure.
The next example occupies four pages and illustrates the subtlety of the subject
(see also Exercise 18 in Rudin [17, page 59]). It constructs a compact Hausdorff
space .X; U/ and a Borel measure on X that is not a Radon measure. More
precisely, there is a point 2 X such that the open set
U WD X n ¹º
is not -compact and satisfies .U / D 1 and .K/ D 0 for every compact sub-
set K U . This example is a kind of refinement of Example 3.5. It is due to
Dieoudonné.
Example 3.6 (Dieudonné’s measure). (i) Let .X; 4/ be an uncountable well-ordered
set with a maximal element 2 X such that every element x 2 X n ¹º has only
countably many predecessors. Here a set is called countable if it is finite or count-
ably infinite. (Think of this as the uncountable Mount Everest: no sequence reaches
the mountain peak .) Thus the relation 4 on X satisfies the following axioms.
3.1. Regular Borel measures 97
(c) If x; y 2 X , then x 4 y or y 4 x.
Sx WD ¹y 2 X j x yº; Px WD ¹y 2 X j y xº:
V WD ¹x 2 X j K Px º:
(iv) We prove that .X; U/ is compact. Let ¹Ui ºi2I be an open cover of X.
We prove by induction that there exist finite sequences
x1 ; : : : ; x ` 2 X and i1 ; : : : ; i` 2 I
such that
xk 2 Uik n Uik 1
and
Sxk Ui1 [ [ Uik 1
S`
for k 2, and X D j D1 Uij . Define
x1 WD
and choose i1 2 I such that 2 Ui1 . If Ui1 D X the assertion holds with ` D 1.
Now suppose, by induction, that x1 ; : : : ; xk and i1 ; : : : ; ik have been constructed
such that xj 2 Uij for j D 1; : : : ; k and Sxk Ui1 [ [ Uik 1 . If
Ui1 [ [ Uik D X;
Ck WD X n Ui1 [ [ Uik
xkC1 WD max.Ck /
by part (iii). Then xkC1 2 Ck and Ck \SxkC1 D ;. Hence SxkC1 Ui1 [ [Uik .
Choose ikC1 2 I such that xkC1 2 UikC1 . This completes the induction argument.
The induction must stop because xkC1 xk for all k and every strictly decreasing
sequence in X is finite by the well ordering axiom (d). This shows that .X; U/ is
compact.
(v) Let Ki X, i 2 N, be a sequence of uncountable compact sets. We prove
that the compact set \
K WD Ki
i2N
K n ¹º ¤ ;: (3.3)
3.1. Regular Borel measures 99
Choose a sequence xn 2 X n¹º such that xn xnC1 for all n 2 N and x2k Ci 2 Ki
for 1 i 2k 1 and k 2 N. That such a sequence exists follows by induction
from the fact that the set X n Sxn D Ps.xn / is countable for each n, while the sets
Ki are uncountable for all i. Now the set
[
P WD Pxn
n2N
is countable and hence the set
S WD X n P
[
DXn Pxn
n2N
[
DXn Ps.xn /
n2N
\
D .X n Ps.xn / /
n2N
\
D Sxn
n2N
is uncountable. Hence
x WD min.S/ :
We prove that x 2 Ki for all i 2 N. Assume by contradiction that x … Ki for
some i . Then there are elements a; b 2 X such that a x b and
U WD Pb \ Sa X n Ki :
If xn 4 a for all n 2 N, then P Pa and so a 2 X n P D S , which is impossible
because a x D min.S/. Thus there must be an integer n0 2 N such that a xn0 .
This implies a xn x b and hence xn 2 U X n Ki for all n n0 ,
contradicting the fact that x2k Ci 2 Ki for all k 2 N. This contradiction shows that
our assumption that x … Ki for some i 2 N must have been wrong. Thus x 2 K
and this proves (3.3).
We prove that K is uncountable. Assume by contradiction that K is countable
and choose a sequence xi 2 K such that K n ¹º D ¹xi j i 2 Nº. Then s.xi /
and
Ki0 WD Ki \ Sxi D Ki n Ps.xi /
is an uncountable compact set for every i 2 N. Moreover,
\
K 0 WD Ki0 K n ¹xi j i 2 Nº D ¹º;
i2N
contradicting the fact that K 0 n ¹º ¤ ; by (3.3). This contradiction shows that K
is uncountable as claimed.
100 3. Borel measures
(vi) Define A 2X by
² ˇ ³
ˇ A [ ¹º contains an uncountable compact set;
A WD A X ˇ ˇ :
or Ac [ ¹º contains an uncountable compact set:
We prove that this is a -algebra. To see this, note first that X 2 A and that A 2 A
implies Ac 2 A by definition. Now choose a sequence Ai 2 A and denote
[
A WD Ai :
i2N
If one of the sets Ai [ ¹º contains an uncountable compact set, then so does the
set A [ ¹º. If none of the sets Ai [ ¹º contains an uncountable compact set, then
the set Aci [ ¹º contains an uncountable compact set for all i 2 N and hence so
T
does the set i2N .Aci [ ¹º/ D Ac [ ¹º by part (v). In both cases it follows that
A 2 A.
(vii) Define the map
W A ! Œ0; 1
by
´
1 if A [ ¹º contains an uncountable compact set,
.A/ WD
0 if Ac [ ¹º contains an uncountable compact set.
This map is well defined because the sets A [ ¹º and Ac [ ¹º cannot both contain
uncountable compact sets by part (v). It satisfies .;/ D 0. Moreover, if Ai 2 A is
a sequence of pairwise disjoint measurable sets, then at most one of the sets Ai [¹º
S P
can contain an uncountable compact set and hence . i 2N Ai / D i2N .Ai /.
Therefore, is a measure.
(viii) The -algebra B 2X of all Borel sets in X is contained in A. To see
this, let U X be open. If U c is uncountable, then U c [ ¹º is an uncountable
compact set and hence U 2 A. If U c is countable choose a sequence xi 2 U c such
that U c n ¹º D ¹xi j i 2 Nº and define
\
S WD Sxi :
i2N
S S
Then X n S D i2N .X n Sxi / D i2N Ps.xi / is a countable set and hence
s WD min.S/ :
The next lemma and theorem are included here in preparation for the Riesz rep-
resentation theorem (Theorem 3.15). They explain the relation between the various
regularity properties of Borel measures.
Lemma 3.7. Let W B ! Œ0; 1 be an outer regular Borel measure that is inner
regular on open sets, i.e.,
Proof. (i) Fix a Borel set B X with .B/ < 1 and a constant " > 0. Since is
outer regular, there exists an open set U X such that
"
B U; .U / < .B/ C :
2
Thus U n B is a Borel set and .U n B/ D .U / .B/ < "=2. Use the outer
regularity of again to obtain an open set V X such that
"
U n B V; .V / < :
2
Now it follows from (3.4) that there exists a compact set K X such that
"
K U; .K/ > .U / :
2
Define C WD K n V . Since X is a Hausdorff space, K is closed, hence C is a closed
subset of K, and so C is compact (see Lemma A.2). Moreover,
C U n V B; B n C .B n K/ [ V .U n K/ [ V;
satisfies (3.2) by (i). Hence assume .B/ D 1. Then it follows from part (iv) of
Theorem 1.28 that lim .B \ Ki / D 1. For each integer n 2 N choose in 2 N
i!1
such that
.B \ Kin / > n:
Since .B \ Kin / .Kin / < 1, it follows from (i) that (3.2) holds with B
replaced by B \ Kin . Hence there exists a compact set Cn B \ Kin such that
.Cn / > n. This proves Lemma 3.7.
Theorem 3.8. Let 1 W B ! Œ0; 1 be an outer regular Borel measure that is inner
regular on open sets. Define
0 W B ! Œ0; 1
by
(i) 0 is a Radon measure, it agrees with 1 on all compact sets and all open
sets, and 0 .B/ 1 .B/ for all B 2 B.
(iv) Let W B ! Œ0; 1 be a Borel measure that is inner regular on open sets. Then
Z Z
f d D f d1
X X
Proof. We prove that 0 is a measure. It follows directly from the definition that
0 .;/ D 0. Now assume that Bi 2 B is a sequence of pairwise disjoint Borel sets
and define
[1
B WD Bi :
i D1
Choose any compact set K B. Then 1 .Bi \ K/ < 1, and so, from part (i) of
Lemma 3.7,
0 .Bi \ K/ D 1 .Bi \ K/
for all i 2 N. Consequently,
1
X 1
X 1
X
1 .K/ D 1 .Bi \ K/ D 0 .Bi \ K/ 0 .Bi /:
iD1 iD1 iD1
To prove the converse inequality, it suffices to assume that 0 .B/ < 1. Then
0 .Bi / 0 .B/ < 1 for all i 2 N. Fix a constant " > 0 and choose a sequence
of compact sets Ki Bi such that 1 .Ki / > 0 .Bi / 2 i " for all i. Then, for
every n 2 N, the union K1 [ [ Kn is a compact subset of B and
n
X n
X
0 .B/ 1 .K1 [ [ Kn / D 1 .Ki / > 0 .Bi / ":
iD1 iD1
Since this holds for all " > 0, it follows that 0 .B/ 1
P
P1 i D1 0 .Bi / and hence
0 .B/ D i D1 0 .Bi / by (3.7). This shows that 0 is a measure. Moreover,
it follows directly from the definition of 0 that 0 .K/ D 1 .K/ for every compact
set K X . Since 1 is inner regular on open sets it follows that 0 .U / D 1 .U /
for every open set U X. Since 0 .K/ D 1 .K/ for every compact set K X ,
the definition of 0 in (3.5) shows that 0 is inner regular and hence is a Radon
measure. The inequality 0 .B/ 1 .B/ for B 2 B follows directly from the
definition of 0 . This proves part (i). Part (ii) follows directly from the definition
of 0 and part (ii) of Lemma 3.7.
104 3. Borel measures
`
X
sD ˛i B ;
i
iD1
where ˛i 2 R and Bi 2 B with 1 .Bi / < 1. Hence 0 .Bi / D 1 .Bi / by part (i)
of Lemma 3.7 and hence
Z `
X Z
s d0 D ˛i 0 .Bi / D s d1 :
X i D1 X
for all x 2 X. Thus sn has compact support for each n. By the Lebesgue monotone
convergence theorem (Theorem 1.37) this implies
Z Z Z Z
f d0 D lim sn d0 D lim sn d1 D f d1 :
X n!1 X n!1 X X
f jK 1; supp.f / U; 0 f 1:
and likewise Z Z
1 .K/ f d1 D f d .U /:
X X
Since 1 .K/ .U / for every compact set K U and 1 is inner regular on
open sets, we obtain
Step 2. Let be as in Step 1 and assume in addition that is inner regular on open
sets. Then .K/ D 1 .K/ for every compact set K X and .U / D 1 .U / for
every open set U X.
Here the two inequalities follow from Step 1. It follows that .U / D 1 .U /.
Now let K be a compact set. Then 1 .K/ < 1. Since 1 is outer regular, there
exists an open set U X such that K U and 1 .U / < 1. Since and 1
agree on open sets it follows that
for every compactly supported continuous function f W X ! Œ0; 1/, and hence also
for every compactly supported continuous function f W X ! R. This proves Step 4
and Theorem 3.8.
Example 3.9. Let .X; U/ be the compact Hausdorff space in Example 3.6 and let
W B ! Œ0; 1 be Dieudonné’s measure.
Moreover, the compact set K D ¹º satisfies .K/ D 0 < 1 D 1 .K/ and
the open set U WD X n ¹º satisfies 1 .U / D 0 < 1 D .U /. This shows that
the inequalities in Step 1 in the proof of Theorem 3.8 can be strict and that the
hypothesis that is inner regular on open sets cannot be removed in part (iv)
of Theorem 3.8.
Remark 3.10. As Example 3.6 shows, it may sometimes be convenient to define a
Borel measure first on a -algebra that contains the -algebra of all Borel measur-
able sets and then restrict it to B. Thus let A 2X be a -algebra containing B
and let W A ! Œ0; 1 be a measure. Call outer regular if it satisfies (3.1) for all
B 2 A, call it inner regular if it satisfies (3.2) for all B 2 A, and call it regular if
it is both outer and inner regular. If is regular and .X; B ; / denotes the com-
pletion of .X; B; jB /, it turns out that the completion is also regular (exercise).
If in addition .X; A; / is -finite (see Definition 4.29 below), then
A B ; D jA : (3.10)
To see this, let A 2 A such that .A/ < 1. Choose a sequence of compact sets
Ki X and a sequence of open sets Ui X such that Ki A Ui and
.A/ 2 i .Ki / .Ui / .A/ C 2 i for all i 2 N. Then B0 WD 1
S
T1 iD1 Ki
and B1 WD i D1 Ui are Borel sets such that B0 A B1 and .B1 n B0 / D 0.
Thus every set A 2 A with .A/ < 1 belongs to B and .A/ D .A/. This
proves (3.10) because every A-measurable set is a countable union of A-measurable
sets with finite measure. Note that if X is -compact and .K/ < 1 for every
compact set K X, then .X; A; / is -finite.
W 2X ! Œ0; 1
Theorem 3.12. Let W 2X ! Œ0; 1 be a Borel outer measure. Then jB is an outer
regular Borel measure and is inner regular on open sets.
One can deduce Theorem 3.12 from Carathéodory’s theorem (Theorem 2.4) and
use axioms (a) and (b) (instead of the Carathéodory criterion in Theorem 2.5) to
show that the -algebra of -measurable sets contains the Borel -algebra. That the
resulting Borel measure has the required regularity properties is then obvious from
axioms (c) and (d). We choose a different route, following Rudin [17], and give
a direct proof of Theorem 3.12 which does not rely on Theorem 2.4. The former
approach is left to the reader as well as the verification that both proofs give rise to
the the same -algebra, i.e., the -algebra A in (3.11) agrees with the -algebra of
-measurable subsets of X .
and
Here the subscript “e” stands for “endlich” and indicates that the elements of Ae are
sets of finite measure. We prove in seven steps that A is a -algebra containing B,
that
WD jA W A ! Œ0; 1
is a measure, and that .X; A; / is a complete measure space. That is outer regular
and is inner regular on open sets follows immediately from conditions (c) and (d)
in Definition 3.11.
3.2. Borel outer measures 109
and hence
1
X
.Ei / D .E/: (3.13)
i D1
Now (3.12) and (3.13) yield
.E/ .K1 [ [ Kn /
n
X
.Ei / "
i D1
1
X
D .E/ .Ei / "
iDnC1
for all n 2 N.
110 3. Borel measures
K" WD K1 [ [ Kn" E
satisfies
.E/ .K" / .E/ 2":
Since this holds for all " > 0, we obtain
Step 2. If E0 ; E1 2 Ae , then
E0 [ E1 2 A e ; E0 \ E1 2 Ae ; E0 n E1 2 Ae :
Ki ; Ui ; Ui n Ki 2 Ae
for i D 0; 1. Define
K WD K0 n U1 E0 n E1 : (3.15)
3.2. Borel outer measures 111
Here the last inequality follows from the definition of K in (3.15) and the inequali-
ties in (3.14). Since " > 0 was chosen arbitrarily, it follows that
E0 [ E1 D .E0 n E1 / [ E1 2 Ae ; E0 \ E1 D E0 n .E0 n E1 / 2 Ae :
Ac \ K D K n .A \ K/ 2 Ae :
Ai \ K 2 Ae
Bi WD Ai \ K 2 Ae
Ei WD Bi n .B1 [ [ Bi 1/ 2 Ae
for all i .
112 3. Borel measures
Since .A \ K/ .K/ < 1 by condition (a) in Definition 3.11, it follows from
Step 1 that A\K 2 Ae . This holds for every compact set K X and hence A 2 A.
This proves Step 3.
Step 4. B A.
Let F X be closed. If K X is compact, then F \ K is a closed subset of a
compact set and hence is compact (see Lemma A.2). Thus F \ K 2 Ae for every
compact subset K X and so F 2 A. Therefore we have proved that A contains
all closed subsets of X . Since A is a -algebra by Step 3, it also contains all open
subsets of X and thus B A. This proves Step 4.
Step 5. Let A X. Then A 2 Ae if and only if A 2 A and .A/ < 1.
If A 2 Ae , then A \ K 2 Ae for every compact set K X by Step 2 and hence
A 2 A. Conversely, let A 2 A such that .A/ < 1. Fix a constant " > 0.
By condition (c) in Definition 3.11, there exists an open set U X such that A U
and .U / < 1. By condition (d) in Definition 3.11, there exists a compact set
K X such that
K U; .K/ > .U / ":
Since K; U 2 Ae and U D .U n K/ [ K, it follows from Step 1 that
.U n K/ D .U / .K/ < ":
Moreover, A \ K 2 Ae because A 2 A. Hence it follows from the definition of Ae
that there exists a compact set H A \ K such that
.H / .A \ K/ "
D .A n .A n K// "
.A/ .A n K/ "
.A/ .U n K/ "
.A/ 2":
Since " > 0 was chosen arbitrarily, we conclude that
.A/ D sup¹.K/ j K A; K is compactº
and hence A 2 Ae . This proves Step 5.
3.3. The Riesz representation theorem 113
Step 6. WD jA is an outer regular extended Borel measure and is inner regular
on open sets.
We prove that is a measure. By definition, .;/ D 0. Now let Ai 2 A be a
sequence of pairwise disjoint measurable sets and define
1
[
A WD Ai :
iD1
supp.f / WD ¹x 2 X j f .x/ ¤ 0º
f 0 H) ƒ.f / 0
The next lemma shows that every positive linear functional on Cc .X/ is con-
tinuous with respect to the topology of uniform convergence when restricted to the
subspace of functions with support contained in a fixed compact subset of X.
then
ƒ.f / D lim ƒ.fi /:
i !1
Hence
"i ƒ./ ƒ.fi / ƒ.f / "i ƒ./;
because ƒ is positive, and hence jƒ.fi / ƒ.f /j "i ƒ./ for all i . Since "i
converges to zero, so does jƒ.fi / ƒ.f /j and this proves Lemma 3.14.
ƒ W Cc .X/ ! R
by Z
ƒ .f / WD f d: (3.16)
X
(iii) The Borel measures 0 and 1 in (i) and (ii) agree on all compact sets and
on all open sets. Moreover,
0 .B/ 1 .B/ for all B 2 B.
(iv) Let W B ! Œ0; 1 be a Borel measure that is inner regular on open sets. Then
ƒ D ƒ () 0 .B/ .B/ 1 .B/ for all B 2 B.
Proof. The proof has nine steps. Step 1 defines a function W 2X ! Œ0; 1, Step 2
shows that it is an outer measure, and Steps 3, 4, and 5 show that it satisfies the
axioms of Definition 3.11. Step 6 defines 1 and Step 7 shows that ƒ1 D ƒ.
Step 8 defines 0 and Step 9 proves uniqueness.
Step 1. Define the function
U W U ! Œ0; 1
by
U .U / WD sup¹ƒ.f / j f 2 Cc .X/; 0 f 1; supp.f / U º (3.17)
for every open set U X and define
W 2X ! Œ0; 1
.A/ WD inf¹U .U / j A U X; U is openº (3.18)
for every subset A X . Then
.U / D U .U /
for every open set U X.
116 3. Borel measures
U .U [ V / U .U / C U .V /: (3.19)
K WD supp.f / U [ V:
Hence f D f C f and so
Define
1
[
U WD Ui :
iD1
3.3. The Riesz representation theorem 117
ƒ.f / U .U1 [ [ Uk /
U .U1 / C C U .Uk /
< .A1 / C C .Ak / C ":
Hence
1
X
ƒ.f / .Ai / C "
iD1
for every f 2 Cc .X/ such that 0 f 1 and supp.f / U . Consequently
1
X
.A/ U .U / .Ai / C "
i D1
and hence
1
X
.A/ .Ai /:
iD1
This proves (3.20) and Step 2.
Step 3. Let U X be an open set. Then
0 f 1; K WD supp.f / U:
Then it follows from the definition of U in (3.17) that ƒ.f / U .V / for every
open set V X with K V . Hence it follows from the definition of in (3.18)
that
ƒ.f / .K/:
118 3. Borel measures
Hence
0 f 1; supp.f / U; f jK 1:
Hence,
a ƒ.f / U .U /:
This shows that a U .U / for every open set U X containing K. Take the infi-
mum over all open sets containing K and use the definition of in equation (3.18)
to obtain a .K/.
We prove that .K/ a. Choose a function f 2 Cc .X/ such that f 0 and
f .x/ D 1 for all x 2 K. Fix a constant 0 < ˛ < 1 and define
.K/ U .U˛ /:
This shows that .K/ ˛1 ƒ.f / for all ˛ 2 .0; 1/, and hence
.K/ ƒ.f /:
Since this holds for every function f 2 Cc .X/ with f 0 and f jK 1, it follows
that .K/ a. This proves Step 4.
0 f 1; f jK0 0; f jK1 1:
That such a function exists follows from Urysohn’s lemma (Theorem A.1) with
K WD K1 and U WD X n K0 :
Now fix a constant " > 0. Then it follows from Step 4 that there exists a function
g 2 Cc .X/ such that
Hence
.K0 / C .K1 / < .K0 C K1 / C " for every " > 0,
and therefore
.K0 / C .K1 / .K0 C K1 /:
and hence
Z
f d1 ƒ.f / for all f 2 Cc .X/.
X
Thus Z
ƒ.f / D f d1 for all f 2 Cc .X/,
X
for all i . (For each i , choose first an open set that satisfies the first two conditions
in (3.24) and then intersect it with the open set f 1 .. 1; yi C "//.) By Theo-
rem A.4, there exist functions 1 ; : : : ; n 2 Cc .X/ such that
n
X n
X
i 0; supp.i / Ui ; i 1; i jK 1: (3.25)
iD1 iD1
Hence,
n
X
ƒ.f / D ƒ.i f /
i D1
X n
.yi C "/ƒ.i /
i D1
X n n
X
D .yi C jaj C "/ƒ.i / jaj ƒ.i /
i D1 iD1
n
X "
.yi C jaj C "/ 1 .Ei / C jaj1 .K/
n
i D1
n n
X "X
D .yi C "/1 .Ei / C .yi C jaj C "/
n
i D1 iD1
n
X
.yi "/1 .Ei / C ".21 .K/ C b C jaj C "/
ZiD1
f d1 C ".21 .K/ C b C jaj C "/:
X
Here we have used the inequality yi C Rjaj C " 0. Since " > 0 can be chosen
arbitrarily small it follows that ƒ.f / X f d1 . This proves (3.23).
122 3. Borel measures
Step 8. Define
0 W B ! Œ0; 1
by
0 .B/ WD sup¹.K/ j K B; K is compactº
Then 0 is a Radon measure, ƒ0 D ƒ, and 0 and 1 satisfy (iii) and (iv).
It follows from Step 6 and part (i) of Theorem 3.8 that 0 is a Radon measure and
it follows from Step 7 and part (iii) of Theorem 3.8 that ƒ0 D ƒ1 D ƒ. That the
measures 0 and 1 satisfy assertions (iii) and (iv) follows from parts (i) and (iv)
of Theorem 3.8.
Step 9. We prove uniqueness in (i) and (ii).
By definition, 0 .K/ D .K/ D 1 .K/ for every compact set K X. Second,
it follows from and Steps 1 and 3 that 0 .U / D U .U / D .U / D 1 .U / for
every open set U X. Third, Steps 7 and 8 assert that ƒ0 D ƒ1 D ƒ. Hence it
follows from part (iv) of Theorem 3.8 that every Borel measure W B ! Œ0; 1 that
is inner regular on open sets and satisfies ƒ D ƒ agrees with on all compact
sets and on all open sets. Hence, every Radon measure W B ! Œ0; 1 that satisfies
ƒ D ƒ is given by
.B/ D sup¹.K/ j K B; K is compactº D 0 .B/
for every B 2 B. Likewise, every outer regular Borel measure W B ! Œ0; 1 that
is inner regular on open sets and satisfies ƒ D ƒ is given by
.B/ D inf¹.U / j B U X; U is openº D .B/ D 1 .B/
for every B 2 B. This proves Step 9 and Theorem 3.15.
The following corollary is the converse of Theorem 3.8.
Corollary 3.16. Let 0 W B ! Œ0; 1 be a Radon measure and define
1 .B/ WD inf¹0 .U / j B U X; U is openº for all B 2 B: (3.26)
Then 1 is an outer regular Borel measure, is inner regular on open sets, and
0 .B/ D sup¹1 .K/ j K B; K is compactº for all B 2 B: (3.27)
Proof. Let 1 be the unique outer regular Borel measure on X that is inner regular
on open sets and satisfies ƒ1 D ƒ0 . Then Theorem 3.15 asserts that 0 and 1
agree on all compact sets and all open sets. Since 1 is outer regular, it follows that
1 is given by (3.26). Since 0 is inner regular it follows that 0 satisfies (3.27).
This proves Corollary 3.16.
3.3. The Riesz representation theorem 123
Proof. This follows from Theorem 3.15 and part (ii) of Theorem 3.18.
Remark 3.20. Let X be a compact Hausdorff space and let C.X/ D Cc .X/ be
the space of continuous real valued functions on X. From a functional analytic
viewpoint it is interesting to understand the dual space of C.X/, i.e., the space
of all bounded linear functionals on C.X/ (Definition 4.23). Exercise 5.35 below
shows that every bounded linear functional on C.X/ is the difference of two positive
linear functionals. If every open subset of X is -compact, it then follows from
Corollary 3.19 that every bounded linear functional on C.X/ can be represented
uniquely by a signed Borel measure. (See Definition 5.10 in Section 5.3 below.)
An important class of locally compact Hausdorff spaces that satisfy the hypothe-
ses of Theorem 3.18 and Corollary 3.19 are the second countable ones. Here are
the definitions. A basis of a topological space .X; U/ is a collection V U of open
sets such that every open set U X is a union of elements of V. A topological
space .X; U/ is called second countable if it admits a countable basis. It is called
first countable if, for every x 2 X, there is a sequence of open sets Wi , i 2 N, such
that x 2 Wi for all i and every open set that contains x contains one of the sets Wi .
Lemma 3.21. Let X be a locally compact Hausdorff space:
Proof. (i) Let V be a countable basis of the topology and let U X be an open
set. Denote by V.U / the collection of all sets V 2 V such that Vx U and Vx is
compact. Let x 2 U . By Lemma A.3 there is an open set W X with compact
closure such that x 2 W W S U . Since V is a basis of the topology, there is an
element V 2 V such that x 2 V W . Hence Vx is a closed subset of the compact
set WS and so is compact by Lemma A.2. Thus V 2 V.U / and x 2 V . This shows
that [
U D V:
V 2V.U /
Ki WD Vx1 [ [ Vxi
Ki X n ¹xº
3.3. The Riesz representation theorem 125
such that
Ki KiC1 for all i 2 N
and
1
[
Ki D X n ¹xº:
iD1
is open and contains x. By Lemma A.3, there exists a sequence of open sets Vi X
with compact closure such that
Define
Wi WD V1 \ \ Vi for i 2 N.
Then
i
\
Si
W .X n Kj / D X n Ki
j D1
and hence
1
\
Si D ¹xº:
W
iD1
This implies that each open set U X that contains x also contains one of the
Si . Namely, if x 2 U and U is open, then W
sets W S1 n U is a compact set contained
in
[1
X n ¹xº D Si /;
.X n W
iD1
j
[
S1 n U
W Si / D X n W
.X n W Sj ;
iD1
and so WSj U . Thus the sets Wj form a countable neighborhood basis of x and
this proves Lemma 3.21.
126 3. Borel measures
X WD Œ0; 1 ¹0; 1º
Sa WD ¹x 2 X j a xº; Pb WD ¹x 2 X j x bº; a; b 2 X:
are both homeomorphic to the Sorgenfrey line, defined as the real axis with the
(nonstandard) topology in which the open sets are the unions of half open intervals
Œa; b/. Since the Sorgenfrey line is not second countable, neither is the double arrow
space .X; U/. (The Sorgenfrey line is Hausdorff and perfectly normal, but is not
locally compact because every compact subset of it is countable.)
3.4 Exercises
Exercise 3.23. This exercise shows that the measures 0 ; 1 in Theorem 3.15 need
not agree. Let .X; d / be the metric space given by X WD R2 and
´
0 if x1 D x2 ;
d..x1 ; y1 /; .x2 ; y2 // WD jy1 y2 j C
1 if x1 ¤ x2 :
(Here the integrals on the right are understood as the Riemann integrals or,
equivalently by Theorem 2.24, as the Lebesgue integrals.) Let W B ! Œ0; 1
be a Borel measure such that
Z
f d D ƒ.f / for all f 2 Cc .X/:
X
Exercise 3.24. This exercise shows that the Borel assumption cannot be removed
in Theorem 3.18. (The measure in part (ii) is not a Borel measure.) Let .X; U/ be
the topological space defined by X WD N [ ¹1º and
U WD ¹U X j U N or #U c < 1º:
(i) Prove that .X; U/ is a compact Hausdorff space and that every subset of X is
-compact. Prove that the Borel -algebra of X is B D 2X .
(ii) Let W 2X ! Œ0; 1 be the counting measure. Prove that is inner regular,
but not outer regular.
Exercise 3.25. Let .X; UX / and .Y; UY / be locally compact Hausdorff spaces and
denote their Borel -algebras by BX 2X and BY 2Y . Let W X ! Y be a
continuous map and let X W BX ! Œ0; 1 be a measure.
(iii) Find an example where X is outer regular and X jBY is not outer regular.
Hint. Consider the inclusion of N into its one-point compactification and use
Exercise 3.24. (In this example X is a Borel measure, however, X is not
a Borel measure.)
Exercise 3.26. Let .X; d / be a metric space. Prove that .X; d / is perfectly normal,
i.e., if F0 ; F1 X are disjoint closed subsets, then there is a continuous function
f W X ! Œ0; 1 such that F0 D f 1 .0/ and F1 D f 1 .1/. Compare this with
Urysohn’s lemma (Theorem A.1). Hint. An explicit formula for f is given by
d.x; F0 /
f .x/ WD ;
d.x; F0 / C d.x; F1 /
where
d.x; F / WD inf d.x; y/
y2F
for x 2 X and F X.
Exercise 3.27. Recall that the Sorgenfrey line is the topological space .R; U/, where
U 2R is the smallest topology that contains all half open intervals Œa; b/ with
a < b. Prove that the Borel -algebra of .R; U/ agrees with the Borel -algebra of
the standard topology on R.
Exercise 3.28. Recall from Example 3.22 that the double arrow space is
X WD Œ0; 1 ¹0; 1º
Hint 1. Show that the projection f W X ! Œ0; 1 onto the first factor is continuous
with respect to the standard topology on the unit interval.
Hint 2. Denote by B 2X the set of all sets of the form (3.28) with E Œ0; 1
a Borel set and F; G X countable. Prove that B is a -algebra.
Exercise 3.29 (the Baire -algebra). Let .X; U/ be a locally compact Hausdorff
space and define
² ˇ ³
ˇ K is compact and there is a sequence of open sets
Ka WD K X ˇ :
Ui such that UiC1 Ui for all i and K D 1
ˇ T
iD1 Ui
3.4. Exercises 129
Let
Ba 2X
be the smallest -algebra that contains Ka . It is contained in the Borel -algebra
B 2X and is called the Baire -algebra of .X; U/. The elements of Ba are called
Baire sets. A function f W X ! R is called Baire measurable if f 1 .U / 2 Ba for
every open set U R. A Baire measure is a measure W Ba ! Œ0; 1 such that
.K/ < 1 for all K 2 Ka .
(ii) Prove that Ba is the smallest -algebra such that every continuous function
f W X ! R with compact support is Ba -measurable.
Exercise 3.30. (i) Let X be an uncountable set and let U WD 2X be the discrete
topology. Prove that B X is a Baire set if and only if B is countable or has a
countable complement. Define
W Ba ! Œ0; 1
by
´
0 if B is countable;
.B/ WD
1 if B c is countable:
R
Show that X f d D 0 for every f 2 Cc .X/. Thus positive linear functionals
ƒW Cc .X/ ! R need not be uniquely represented by Baire measures.
(ii) Let X be the compact Hausdorff space of Example 3.6. Prove that the Baire
sets in X are the countable subsets of X n ¹º and their complements.
(iii) Let X be the Stone–Čech compactification of N in Example 4.60 below.
Prove that the Baire sets in X are the subsets of N and their complements.
(iv) Let X D R2 be the locally compact Hausdorff space in Exercise 3.23
(with a nonstandard topology). Show that B X is a Baire set if and only if
the set Bx WD ¹y 2 R j .x; y/ 2 Bº is a Borel set in R for every x 2 R and one of
the sets S0 WD ¹x 2 R j Bx ¤ ;º and S1 WD ¹x 2 R j Bx ¤ Rº is countable.
130 3. Borel measures
Exercise 3.31. Let .X; U/ be a locally compact Hausdorff space and let
Ba B 2X
be the Baire and Borel -algebras. Let F.X/ denote the real vector space of all
functions f W X ! R. For F F.X/ consider the following conditions:
(a) Cc .X/ F;
Let Fa F.X/ be the intersection of all subsets F F.X/ that satisfy condi-
tions (a) and (b). Prove the following.
(ii) Every element of Fa is Baire measurable. Hint. The set of Baire measurable
functions on X satisfies (a) and (b).
(vi) If f 2 Fa and g 2 Cc .X/, then fg 2 Fa . Hint. Fix a real number c such that
c C g.x/ > 0 for all x 2 R. Then the set .c C g/ 1 Fa satisfy (a) and (b) and
hence contains Fa . Now use (iv) and (v).
A; B 2 A H) A [ B 2 A:
3.4. Exercises 131
(x) For every f 2 Fa there exists a sequence of compact sets Ki 2 Ka such that
S
Ki KiC1 for all i and supp.f / i2N Ki . Hint. The set of functions
f W X ! R with this property satisfies conditions (a) and (b).
Exercise 3.32. Show that, for every locally compact Hausdorff space X and any
two Borel measures 0 ; 1 as in Theorem 3.8, there is a Baire set N X such that
0 .N / D 0 and 0 .B/ D 1 .B/ for every Baire set B X n N .
Hint 1. Show first that
where Ka is as in Exercise 3.29. To see this, prove that the right-hand side of equa-
tion (3.29) defines a Borel measure on X that is inner regular on open sets and
satisfies 0 and ƒ D ƒ0 .
Hint 2. Suppose there exists a Baire set N X such that 0 .N / < 1 .N /.
Show that 1 .N / D 1 and that N can be chosen such that 0 .N / D 0. Next
show that XnN 2 Fa , where Fa is as in Exercise 3.31, and deduce that X n N is
contained in a countable union of compact sets.
Exercise 3.35. Let Z be the disjoint union of the locally compact Hausdorff spaces
X0 in Example 3.33 and X D R2 in Example 3.34. Find Baire sets B0 X0 and
B X whose (disjoint) union is not a Baire set in Z.
Chapter 4
Lp spaces
This chapter discusses the Banach space Lp ./ associated to a measure space
.X; A; / and a number 1 p 1. Section 4.1 introduces the inequalities of
Hölder and Minkowski and Section 4.2 shows that Lp ./ is complete. In Sec-
tion 4.3 we prove that, when X is a locally compact Hausdorff space, is a Radon
measure, and 1 p < 1, the subspace of continuous functions with compact sup-
port is dense in Lp ./. If in addition X is second countable, it follows that Lp ./
is separable. When 1 < p < 1 (or p D 1 and the measure space .X; A; / is
localizable) the dual space of Lp ./ is isomorphic to Lq ./, where 1=pC1=q D 1.
For p D 2 this follows from elementary Hilbert space theory and is proved in Sec-
tion 4.4. For general p the proof requires the Radon–Nikodým theorem and is de-
ferred to Chapter 5. Some preparatory results are proved in Section 4.5.
Assume throughout that .X; A; / is a measure space and that p; q are real numbers
such that
1 1
C D 1; 1 < p < 1; 1 < q < 1: (4.1)
p q
Then any two nonnegative real numbers a and b satisfy Young’s inequality
1 p 1 q
ab a C b ; (4.2)
p q
1 p
.0; 1/ ! RW x 7 ! x xb:
p
134 4. Lp spaces
hence
f C g 2.f p C g p /1=p ;
therefore
.f C g/p 2p .f p C g p /;
and integrating this inequality and raising the integral to the power 1=p we obtain
Exercise 4.2. (i) Assume 0 < X f p d < 1 and 0 < X g q d < 1. Prove
R R
that equality holds in (4.3) if and only if there exists a constant ˛ > 0 such that
g q D ˛f p almost everywhere. Hint. Use the proof of the Hölder inequality and the
fact that equality holds in (4.2) if and only ap D b q .
(ii) Assume 0 < X f p d < 1 and 0 < X g p d < 1. Prove that equality
R R
holds in (4.4) if and only if there is a real number > 0 such that g D f almost
everywhere. Hint. Use part (i) and the proof of the Minkowski inequality.
136 4. Lp spaces
Definition 4.3. Let .X; A; / be a measure space and let 1 p < 1. Consider a
measurable function f W X ! R x . The Lp -norm of f is the number
Z 1=p
kf kp WD jf jp d : (4.5)
X
It follows from the Minkowski inequality (4.4) that the sum of two Lp -functions
is again an Lp -function, and hence Lp ./ is a real vector space. Moreover, the
function
Lp ./ ! Œ0; 1/W f 7 ! kf kp
kf C gkp kf kp C kgkp
kf kp D jj kf kp ;
f g () f D g -almost everywhere: (4.7)
Lp ./ WD Lp ./= : (4.8)
4.2. The Banach space Lp ./ 137
This is again a real vector space. (For p D 1 see Definition 1.51.) The Lp -norm
in (4.5) depends only on the equivalence class of f and so the map
is well defined. It is a norm on Lp ./ by Lemma 1.49. Thus we have defined the
normed vector space Lp ./ for 1 p < 1. It is often convenient to abuse notation
and write f 2 Lp ./ instead of Œf 2 Lp ./, always bearing in mind that then
f denotes an equivalence class of p-integrable functions. If .X; A ; / denotes
the completion of .X; A; /, it follows as in Corollary 1.56 that Lp ./ is naturally
isomorphic to Lp . /.
Remark 4.4. Assume 1 < p < 1 and let f; g 2 Lp ./ such that
kf C gkp D kf kp C kgkp ; kf kp ¤ 0:
Then it follows from part (ii) of Exercise 4.2 that there exists a real number 0
such that g D f almost everywhere.
Example 4.5. If .Rn ; A; m/ is the Lebesgue measure space we write
Lp .Rn / WD Lp .m/:
`p WD Lp ./:
Thus the elements of `p are sequences .xn /n2N of real numbers such that
1
X 1=p
k.xn /kp WD jxn jp < 1:
pD1
If we define
f WN ! R
by
f .n/ WD xn for n 2 N,
then Z 1
X
jf jp d D jxn jp :
N pD1
Definition 4.7. Let .X; A; / be a measure space and let f W X ! Œ0; 1 be a mea-
surable function. The essential supremum of f is the number ess sup f 2 Œ0; 1
defined by
This is a normed vector space with the norm defined by (4.10), which depends only
on the equivalence class of f .
Lemma 4.8. For every f 2 L1 ./ there exists a measurable set E 2 A such that
for all i 2 N.
4.2. The Banach space Lp ./ 139
Define
k
X 1
X
gk WD jfniC1 fni j; g WD jfniC1 fni j D lim gk :
k!1
iD1 iD1
gkp gkC1
p
for all k 2 N
and the sequence of functions gkp W X ! Œ0; 1 converges pointwise to the integrable
function g p . Hence it follows from the Lebesgue monotone convergence theorem
(Theorem 1.37) that
kgkp D lim kgk kp 1:
k!1
lim kf fn kp D 0:
n!1
To see this, fix a constant " > 0. Then there exists an integer n0 2 N such that
"p
for all m; n 2 N. To see this, use Lemma 4.8 to find null sets En ; Em;n 2 A such
that
Then the union E of the sets En and Em;n is measurable and satisfies (4.12). Since
Œfn is a Cauchy sequence in L1 ./, we have
Since jfm .x/ fn .x/j "n for all m n and all x 2 X n E, it follows that
.fn .x//n2N is Cauchy sequence in R and hence converges for every x 2 X n E.
Define
f WX ! R
by 8
< lim fn .x/ for x 2 X n E,
n!1
f .x/ WD
:0 for x 2 E.
4.3. Separability 141
Then
lim kfn f kp D 0:
n!1
4.3 Separability
Definition 4.11. Let X be a topological space. A subset S X is said to be dense
(in X ) if its closure is equal to X or, equivalently, U \ S ¤ ; for every nonempty
open set U X. A subset S X of a metric space is dense if and only if every
element of X is the limit of a sequence in S . The topological space X is called
separable if it admits a countable dense subset.
Every second countable topological space is separable and first countable
(see Lemma 3.21). The Sorgenfrey line is separable and first countable, but is not
second countable (see Example 3.22). A metric space is separable if and only if it
is second countable. (If S is a countable dense subset, then the balls with rational
radii centered at the points of S form a basis of the topology.) The Euclidean space
X D Rn with its standard topology is separable (Qn is a countable dense subset) and
hence is second countable. The next lemma gives a condition for a linear subspace
to be dense in Lp ./.
142 4. Lp spaces
Lemma 4.12. Let .X; A; / be a measure space and let 1 p < 1. Let X be
a linear subspace of Lp ./ such that ŒA 2 X for every measurable set A 2 A
with .A/ < 1. Then X is dense in Lp ./.
Proof. Let Y denote the closure of X in Lp ./. Then Y is a closed linear subspace
of Lp ./. We prove in three steps that Y D Lp ./.
Z Z
p p
j˛i j .Ai / D j˛i A j d jsjp d < 1
i
X X
and hence .Ai / < 1 for all i . This implies ŒA 2 Y for all i . Since Y is a
i
linear subspace of Lp ./, it follows that Œs 2 Y.
Step 3. Y D Lp ./.
Theorem 4.15. Assume is outer regular and is inner regular on open sets. Define
² ˇ ³
ˇ s is a Borel measurable step function
Sc .X/ WD sW X ! R ˇˇ : (4.13)
and supp.s/ is a compact subset of X
Then the linear subspaces Sc .X/= and Cc .X/= are dense in Lp ./. This con-
tinues to hold when is a Radon measure.
Lemma 4.17. Assume D 1 is outer regular and is inner regular on open sets.
Let 0 W B ! Œ0; 1 be the unique Radon measure such that ƒ1 D ƒ0 . Then
Lp .1 / Lp .0 / and the linear map
Proof. Since 0 .B/ 1 .B/ for all B 2 B by Theorem 3.15, it follows that
Z Z
p
jf j d0 jf jp d1
X X
Thus the map (4.14) is injective and has a closed image. To prove (4.15), define
Then 1 .E" / < 1 and hence 1 and 0 agree on all Borel subsets of E" by
Lemma 3.7. This implies
Z Z
jf jp d0 D jf jp d1 ;
E" E"
Then
1 .A/ D 0 .A/ D lim 0 .Ki / D 0 .B/:
i !1
This implies A 2 Lp .1 / and ŒB 0 D ŒA 0 2 X. By Lemma 4.12, it follows
that X D Lp .0 / and this proves Lemma 4.17.
Proof of Theorem 4.13. Let V U be a countable basis for the topology. Assume
without loss of generality that Vx is compact for all V 2 V. (If W U is any
countable basis for the topology, then the set V WD ¹V 2 W j Vx is compactº is also
a countable basis for the topology by Lemma A.3.) Choose a bijection
N ! VWi 7 ! Vi
and let
I WD ¹I N j #I < 1º
i2I
Define
I WD ¹i 2 N j Vi U º:
Since V is a basis of the topology, we have K U D i 2I Vi . Since K is compact,
S
K VI U:
x
Since V 2 V and " > 0 was chosen arbitrary, it follows that ŒB 2 X D V.
I
This proves Theorem 4.13.
Proof of Theorem 4.15. By Corollary 3.16 and Lemma 4.17, it suffices to consider
the case where is outer regular and is inner regular on open sets. Define
We must prove that Lp ./ D S D C. Since S and C are closed linear subspaces
of Lp ./, it suffices to prove that ŒB 2 S \ C for every Borel set B 2 B
with .B/ < 1 by Lemma 4.12. Let B 2 B with .B/ < 1 and let " > 0.
By Lemma 3.7, there is a compact set K X and an open set U X such that
K B U and .U n K/ < "p . By Urysohn’s lemma (Theorem A.1), there
146 4. Lp spaces
Remark 4.18. The reader may wonder whether Theorem 4.15 continues to hold for
all Borel measures W B ! Œ0; 1 that are inner regular on open sets. To answer
this question one can try to proceed as follows. Let 0 ; 1 be the Borel measures on
X in Theorem 3.15 that satisfy ƒ0 D ƒ1 D ƒ . Then 0 is a Radon measure,
1 is outer regular and is inner regular on open sets, and 0 .B/ .B/ 1 .B/
for all B 2 B. Thus Lp .1 / Lp ./ Lp .0 / and one can consider the maps
Question. Do there exist a locally compact Hausdorff space .X; U/ and Borel mea-
sures
0 ; 1 ; W B ! Œ0; 1
on its Borel -algebra B 2X such that all three measures are inner regular on
open sets, 1 is outer regular, 0 is given by (3.5),
Lemma 4.20. Let H be a real vector space equipped with an inner product (4.16)
and the associated norm (4.17). The inner product and norm satisfy the Cauchy–
Schwarz inequality
jhx; yij kxk kyk (4.18)
and the triangle inequality
This implies jh; ij 1 and hence jhx; yij kxk kyk. In turn it follows from the
Cauchy–Schwarz inequality that
Definition 4.21. An inner product space .H; h; i/ is called a Hilbert space if the
norm (4.17) is complete, i.e., every Cauchy sequence in H converges.
by Lemma 1.48 which is positive definite by Lemma 1.49. Hence (4.21) is an inner
product on L2 ./. It is called the L2 inner product. The norm associated to this
inner product is
Z 1=2 p
2 2
L ./ ! RW Œf 7 ! kf k2 D f d D hf; f i: (4.22)
X
This is the L2 -norm in (4.5) with p D 2. By Theorem 4.9, L2 ./ is complete with
the norm (4.22) and hence is a Hilbert space.
4.4. Hilbert spaces 149
Definition 4.23. Let .V; kk/ be a normed vector space. A linear functional
ƒW V ! R is called bounded if there exists a constant c 0 such that
The set of bounded linear functionals on V is denoted by V and is called the dual
space of V .
Exercise 4.24. Prove that a linear functional on a normed vector space is bounded
if and only if it is continuous.
Exercise 4.25. Let .V; kk/ be a normed vector space. Prove that the dual space V
with the norm (4.23) is a Banach space. (See Example 1.11.)
E WD ¹x 2 H j ƒ.x/ D 1º:
Then E ¤ ; because there exists an element 2 H such that ƒ./ ¤ 0 and hence
x WD ƒ./ 1 2 E. The set E is a closed because ƒW H ! R is continuous,
150 4. Lp spaces
Now put
x0
y WD :
kx0 k2
Fix an element x 2 H and put WD ƒ.x/. Then ƒ.x x0 / D ƒ.x/ D 0.
Hence it follows from (4.26) that
This implies
hx0 ; xi
hy; xi D D D ƒ.x/:
kx0 k2
Thus y satisfies (4.24).
We prove (4.25). Assume y 2 H satisfies (4.24). If y D 0, then ƒ D 0,
and so kyk D 0 D kƒk. Hence assume y ¤ 0. Then
We claim that xi is a Cauchy sequence. Indeed, fix a constant " > 0. Then there
exists an integer i0 2 N such that
"
i 2 N; i i0 H) kxi k2 < ı 2 C :
4
Let i; j 2 N such that i i0 and j i0 . Then 12 .xi C xj / 2 E because E is
convex, and hence kxi C xj k 2ı. This implies
kxi xj k2 D 2kxi k2 C 2kxj k2 kxi C xj k2
"
< 4 ı2 C 4ı 2
4
D ":
Thus xi is a Cauchy sequence. Since H is complete, the limit
x0 WD lim xi
i !1
It is also true that every -finite measure space is localizable. This can be derived
as a consequence of Theorem 4.35 (see Corollary 5.9 below). A more direct proof
is outlined in Exercise 4.58.
X WD ¹a; bº;
A WD 2X ;
.¹aº/ WD 1; .¹bº/ WD 1:
This measure space is not semi-finite. Thus the linear map L1 ./ ! L1 ./ in
Theorem 4.33 below is not injective. In fact, L1 ./ has dimension two and L1 ./
has dimension one.
ƒW L1 ./ ! R
The next theorem is the first step towards understanding the dual space of Lp ./
and is a fairly easy consequence of the Hölder inequality. It asserts that for
1=p C 1=q D 1 every element of Lq ./ determines a bounded linear functional
on Lp ./ and that the resulting map Lq ./ ! Lp ./ is an isometric embedding
(for p D 1 under the semi-finite hypothesis). The key question is then whether
every bounded linear functional on Lp ./ is of that form. That this is indeed the
case for 1 < p < 1 (and for p D 1 under the localizable hypothesis) is the con-
tent of Theorem 4.35 below. This is a much deeper theorem whose proof for p ¤ 2
requires the Radon–Nikodým theorem and will be carried out in Chapter 5.
4.5. The dual space of Lp ./ 155
1 1
1 p 1; 1 q 1; C D 1: (4.30)
p q
ƒg W Lp ./ ! R
and ˇZ ˇ
ˇ ˇ
ˇ fg dˇ
ˇ ˇ
X
kƒg k D sup kgkq : (4.32)
f 2Lp ./; kf kp ¤0 kf kp
(iv) Assume p D 1. Then the map L1 ./ ! L1 ./ in (4.33) is injective if and
only if it is an isometric embedding if and only if .X; A; / is semi-finite.
The heart of the proof is the next lemma. It is slightly stronger than what is
required to prove Theorem 4.33 in that the hypothesis on g to be q-integrable is
dropped in part (iii) and replaced by the assumption that the measure space is semi-
finite. In this form Lemma 4.34 is needed in the proof of Theorem 4.35 and will
also be useful for proving the Minkowski and Calderón–Zygmund inequalities in
Theorems 7.19 and 7.43 below.
156 4. Lp spaces
Lemma 4.34. Let .X; A; / be a measure space and let p; q be as in (4.30). Let
gW X ! Œ0; 1 be a measurable function and suppose that there exists a constant
c 0 such that
Z
f 2 L ./; f 0 H)
p
fg d c kf kp : (4.34)
X
hW X ! Œ0; 1/
by
´
g.x/ for x 2 X n A,
h.x/ WD
0 for x 2 A.
Then h is measurable and
Z Z
khkq D kgkq < 1; f h d D fg d c kf kp
X X
f W X ! Œ0; 1/
by
f .x/ WD h.x/q 1
for x 2 X .
Then
f p D hp.q 1/
D hq D f h
4.5. The dual space of Lp ./ 157
and hence
Z 1 1=q Z
kf kp D q
h d D khkqq 1 ; f h d D khkqq :
X X
Since khkq < 1, it follows that kgkq D khkq c, and this proves part (ii).
(iii) Assume .X; A; / is semi-finite and 1 < q < 1. Suppose, by contra-
diction, that kgkq > c. We will prove that there exists a measurable function
hW X ! Œ0; 1/ such that
for all f 2 L ./ with f 0. Since khkq < 1 it follows from part (ii) that
p
A WD ¹x 2 X j g.x/ c C ıº
The next theorem asserts that, for 1 < p < 1, every bounded linear functional
on Lp ./ has the form (4.31) for some g 2 Lq ./. For p ¤ 2 this is a much
deeper result than Corollary 4.28. The proof requires the Radon–Nikodým theorem
and will be deferred to the next chapter.
Theorem 4.35 (the dual space of Lp ). Let .X; A; / be a measure space and fix
constants
1 1
1 p < 1; 1 < q 1; C D 1:
p q
Then the following holds.
(i) Assume 1 < p < 1. Then the map Lq ./ ! Lp ./ W Œg 7! ƒg ; defined
by (4.31) is bijective and hence is a Banach space isometry.
This next example shows that, in general, Theorem 4.35 does not extend to the
case p D 1 (regardless of whether or not the measure space .X; A; / is -finite).
By Theorem 4.33, the Banach space L1 ./ is equipped with an isometric inclusion
L1 ./ ! L1 ./ , however, the dual space of L1 ./ is typically much larger
than L1 ./.
Example 4.36. Let W 2N ! Œ0; 1 be the counting measure on the positive inte-
gers. Then
`1 WD L1 ./ D L1 ./
is the Banach space of bounded sequences x D .xn /n2N of real numbers, equipped
with the supremum norm
kxk1 WD sup jxn j:
n2N
1
An interesting closed subspace of ` is
ƒy W `1 ! R
by
1
X
ƒy .x/ WD xn y n for x D .xn /n2N 2 `1 :
nD1
P1
Choose N 2 N such that nDN jyn j DW ˛ < 1 and define x D .xn /n2N 2 c by
xn WD 0 for n < N and xn WD 1 for n N . Then ƒy .x/ ˛ < 1 D ƒ.x/ and
hence ƒy ¤ ƒ. This shows that ƒ does not belong to the image of the isometric
inclusion `1 ,! .`1 / .
In the language of functional analysis, this means that the Banach spaces c0 and `1
are not reflexive, and neither is `1 .
We close this section with two results that will be needed in the proof of The-
orem 4.35. The first asserts that every bounded linear functional on Lp ./ can be
written as the difference of two positive bounded linear functionals (Theorem 4.39).
The second asserts that every positive bounded linear functional on Lp ./ is sup-
ported on a -finite subset of X (Theorem 4.40). When ƒW Lp ./ ! R is a
bounded linear functional it will be convenient to abuse notation and write
Definition 4.38. Let .X; A; / be a measure space and let 1 p < 1. A bounded
linear functional ƒW Lp ./ ! R is called positive if
f 0 H) ƒ.f / 0
Theorem 4.39. Let .X; A; / be a measure space and let 1 p < 1. Consider a
bounded linear functional ƒW Lp ./ ! R. Define
˙ W A ! Œ0; 1
by
˙ .A/ WD sup¹ƒ.˙E / j E 2 A; E A; .E/ < 1º (4.37)
Then the maps ˙ are measures, Lp ./ L1 .C / \ L1 . /, and the formulas
Z
˙
ƒ .f / WD f d˙ for f 2 Lp ./ (4.38)
X
It follows directly from the definition that ˙ .;/ D 0. We must prove that C is
-additive. That is then also -additive follows by reversing the sign of ƒ.
Thus, let Ai 2 A be a sequence of pairwise disjoint measurable sets and set
1
[
A WD Ai :
iD1
Let E 2 A such that E A and .E/ < 1. Then it follows from the definition
of C that
ƒ.E \A / C .Ai / for all i 2 N: (4.40)
i
converges pointwise to zero and satisfies 0 fnp E for all n. Since .E/ < 1,
the function E is integrable, and so it follows from the Lebesgue dominated con-
vergence theorem (Theorem 1.45) that
Z
lim fnp d D 0;
n!1 X
i.e.,
n
X
lim
E E \A
D 0:
n!1 i p
i D1
Take the supremum over all E 2 A with E A and .E/ < 1 to obtain
1
X
C
.A/ C .Ai /:
iD1
To prove the converse inequality, assume first that C .Ai / D 1 for some i ; since
Ai A, this implies C .A/ D 1 D 1 C
P
i D1 .Ai /. Hence it suffices to assume
C
.Ai / < 1 for all i . Fix a constant " > 0 and choose a sequence of measurable
sets Ei 2 A such that Ei Ai and ƒ.E / > C .Ai / 2 i " for all i . Since
i
E1 [ [ En A, it follows from the definition of C that
n
X n
X
C .A/ ƒ.E [[En
/D ƒ.E / > C .Ai / ":
1 i
i D1 iD1
so
1
X
C .A/ C .Ai /;
i D1
Then
Z Z `
X
C
s d C s d D ˛i .C .Ai / C .Ai //
X X iD1
`
X
˛i .ƒ.E C / ƒ.E / C 2"i /
i i
iD1
`
X
Dƒ ˛i .E C E / C "
i i
iD1
X`
c
˛i .E C E /
C "
i i p
iD1
`
X 1=p
Dc ˛ip ..EiC n Ei / C .Ei n EiC // C"
iD1
`
X 1=p
c ˛ip .Ai / C"
iD1
D c kskp C ":
164 4. Lp spaces
Take the limit " ! 0 to obtain (4.41) for f D s. To prove (4.41) in general it suffices
to assume that f 2 Lp ./ is nonnegative. By Theorem 1.26, there is a sequence
of measurable step functions 0 s1 s2 that converges pointwise to f .
Then .f sn /p converges pointwise to zero and is bounded above by f p 2 L1 ./.
Hence
lim kf sn kp D 0
n!1
Now let " > 0 and choose E 2 A such that E A and ƒ.E / > C .A/ ".
Since ƒ.AnE / .A/, this implies
Since this holds for all " > 0, we obtain ƒ.A / C .A/ .A/. Reversing the
sign of ƒ we also obtain ƒ.A / .A/ C .A/ and this proves Step 3.
Step 4. If f 2 Lp ./, then
Z Z
ƒ.f / D f dC f d : (4.43)
X X
`
X `
X Z Z
C C
ƒ.s/ D ˛i ƒ.A / D ˛i . .Ai / .Ai // D s d s d :
i
iD1 iD1 X X
4.5. The dual space of Lp ./ 165
This proves (4.43) for p-integrable step functions. Now let f 2 Lp ./ and as-
sume f 0. By Theorem 1.26, there is a sequence of measurable step functions
0 s1 s2 that converges pointwise to f . Then .f sn /p converges
pointwise to zero and is bounded above by f p 2 L1 ./. Hence
lim kf sn kp D 0
n!1
Moreover, Z
f d˙ c kf kp < 1
X
by Step 2, and Z Z
lim sn d˙ D f d˙
n!1 X X
by the Lebesgue monotone convergence theorem. Thus every nonnegative Lp -func-
tion f W X ! Œ0; 1/ satisfies (4.43). If f 2 Lp ./, then f ˙ 2 Lp ./
satisfy (4.43) by what we have just proved, and hence so does f D f C f .
This proves Step 4.
It follows from Steps 2 and 4 that the linear functionals ƒ˙ W Lp ./ ! R
in (4.38) are bounded and satisfy (4.39). This proves Theorem 4.39.
Theorem 4.40. Let .X; A; / be a measure space and let 1 < p < 1. Consider a
positive bounded linear functional ƒW Lp ./ ! R. Define
.A/ WD sup¹ƒ.E / j E 2 A; E A; .E/ < 1º (4.44)
for A 2 A. Then the map W A ! Œ0; 1 is a measure, Lp ./ L1 ./, and
Z
ƒ.f / D f d for all f 2 Lp ./: (4.45)
X
for all n 2 N.
Proof. That is a measure satisfying Lp ./ L1 ./ and (4.45) follows from
Theorem 4.39 and the fact that C D and D 0 because ƒ is positive.
Now define c WD kƒk. We prove in three steps that there exist measurable sets
N 2 A and Xn 2 A for n 2 N satisfying (4.46).
166 4. Lp spaces
Step 1. For every " > 0 there exist a measurable set A 2 A and a measurable
function
f W X ! Œ0; 1/
such that
In particular,
p
.A/ .inf f / < 1:
A
Choose h 2 Lp ./ such that khkp D 1 and ƒ.h/ > c ". Assume without loss of
generality that h 0. (Otherwise replace h by jhj.) Define
Ai WD ¹x 2 X j h.x/ > 2 i º:
lim kh hA kp D 0;
i !1 i
and therefore
lim ƒ.hA / D ƒ.h/ > c ":
i !1 i
and define
hA
i
A WD Ai ; f WD :
khA kp
i
Define
" 1=p
g WD f C E :
.E/
Then
Z 1=p
p
kgkp D f d C " D .1 C "/1=p
X
and, by (4.47),
" 1=p ƒ.E /
ƒ.g/ D ƒ.f / C ƒ.E / > c " C "1=p :
.E/ .E/1=p
ƒ.E /
c " C "1=p < c.1 C "/1=p :
.E/1=p
ƒ.E / c
"1=p < c..1 C "/1=p 1/ C " " C1 :
.E/1=p p
ƒ.E / 1 c
< C 1
.E/1=p n1=q p
for all n 2 N, and hence ƒ.E / D 0. This implies .N / D 0 by (4.44). More-
over, .Xn / niD1 .Ai / < 1 for every n by Step 1. This proves Step 3 and
P
Theorem 4.40.
168 4. Lp spaces
4.6 Exercises
Many of the exercises in this section are taken from Rudin [17, pages 71–75].
Exercise 4.41. Let .X; A; / be a measure space and let
f D .f1 ; : : : ; fn /W X ! Rn
be a measurable function such that
Z
jfi j d < 1
X
for i D 1; : : : ; n. Define
Z Z Z
f d WD f1 d; : : : ; fn d 2 Rn :
X X X
n n
Let R ! Œ0; 1/W v 7! kvk ; be any norm on R . Prove that the function
X ! Œ0; 1/W x 7 ! kf .x/k
is integrable and
Z
Z
f d
kf k d: (4.49)
X X
Hint. First prove the inequality first for vector-valued integrable step functions
sW X ! Rn . Then show that forR all " > 0 there is a vector-valued integrable step
n
R
function sW X ! R such that k X .f s/ dk < " and X kf sk d < ".
Exercise 4.42. Let .X; A; / be a measure space such that .X/ D 1. Let
f 2 L1 ./, and let W R ! R be convex. Prove Jensen’s inequality
Z Z
f d . ı f / d: (4.50)
X X
Exercise 4.43. Let .X; A; / be a measure space, choose p; q; r 2 Œ1; 1 such that
1 1 1
C D ;
p q r
and let f 2 Lp ./ and g 2 Lq ./. Prove that fg 2 Lr ./ and
kf kp kf kr kf k1s
: (4.54)
Prove that If is an interval. Assume f does not vanish almost everywhere and
define the function f W .1; 1/ ! R by
lim kf kp D kf k1 :
p!1
Exercise 4.46. For each of the following three conditions find an example of mea-
sure space .X; A; / that satisfies it for all p; q 2 Œ1; 1.
Exercise 4.47. Let .X; U/ be a locally compact Hausdorff space and define
8 ˇ 9
< ˇ f is continuous and =
ˇ
C0 .X/ WD f W X ! R ˇ for all " > 0, there exists K X such that :
ˇ
XnK jf j < "
: ˇ K is compact and sup ;
Prove that X is a Banach space with respect to the sup-norm. Prove that Cc .X/ is
dense in C0 .X/.
Exercise 4.48. Let .X; A; / be a measure space such that .X/ D 1 and let
f; gW X ! Œ0; 1 be measurable functions such that fg 1. Prove that
kf k1 kgk1 1:
Exercise 4.49. Let .X; A; / be a measure space such that .X/ D 1 and let
f W X ! Œ0; 1 be a measurable function. Prove that
q Z q
1 C kf k21 1 C f 2 d 1 C kf k1 : (4.55)
X
and Z
f p d .E/1 p
for 0 < p < 1: (4.57)
E
(i) Let f W .0; 1/ ! R be Lebesgue measurable and suppose that the function
.0; 1/ ! RW x 7! x p 1 a jf .x/jp ; is integrable. Show that the restriction of
f to each interval .0; x is integrable and prove Hardy’s inequality
Z 1
ˇZ x
ˇp 1=p Z 1 1=p
1 a p
xp 1 a p
ˇ ˇ
x ˇ f .t/ dt ˇ dx
ˇ jf .x/j dx :
0
ˇ
0 a 0
(4.59)
Show that equality holds in (4.59) if and only if f D 0 almost everywhere.
Hint. Assume first that f is nonnegative with compact support and define
1 x
Z
F .x/ WD f .t/ dt for x > 0.
x 0
Use integration by parts to obtain
Z 1
p 1 p
Z
p 1 a p 1 a
x F .x/ dx D x F .x/p 1
f .x/ dx:
0 a 0
Use Hölder’s inequality.
(ii) Show that the constant p=a in Hardy’s inequality is sharp. Hint. Choose
< 1 a=p and take f .x/ WD x for x 1 and f .x/ WD 0 for x > 1.
(iii) Prove that every sequence .an /n2N of positive real numbers satisfies
1 N 1
X 1 X p p p X p
an a : (4.60)
N nD1 p 1 nD1 n
N D1
(iv) Let f W .0; 1/ ! R be Lebesgue measurable and suppose that the function
.0; 1/ ! RW x 7! x p 1Ca jf .x/jp ; is integrable. Show that the restriction
of f to each interval Œx; 1/ is integrable and prove the inequality
Z 1
ˇZ 1
ˇp 1=p Z 1 1=p
a 1 p
xp 1Ca p
ˇ ˇ
x ˇ f .t/ dt ˇ dx
ˇ jf .x/j dx :
0
ˇ
x a 0
(4.61)
2 1
Hint. Apply the inequality (4.59) to the function g.x/ WD x f .x /.
172 4. Lp spaces
Exercise 4.53. Let .X; U/ be a locally compact Hausdorff space and let
W B ! Œ0; 1 be an outer regular Borel measure on X that is inner regular on
open sets. Let g 2 L1 ./. Prove that the following are equivalent:
Hint. Assume (ii). Let K X be compact. Use Urysohn’s lemma (Theorem A.1)
to show that there is a sequence fn 2 CcR.X/ such that 0 fn 1 and Rfn converges
almost everywhere to K . DeduceR that K g d D 0. Then prove that U g d D 0
for every open set U X and B g d D 0 for all B 2 B.
Warning! The regularity hypotheses on cannot be removed. Find an example of
a Borel measure where (ii) does not imply (i). (See Example 4.16.)
S.k; n/ WD ¹x 2 X j jfi .x/ fj .x/j < 1=k, for all i; j > nº for k; n 2 N:
Prove that
lim .S.k; n// D .X / for all k 2 N:
n!1
T
Deduce that there is a sequence nk 2 N such that E WD k2N S.k; nk / satisfies
the required conditions. Show that Egoroff’s theorem does not extend to -finite
measure spaces.
Exercise 4.55. Let .X; A; / be a measure space and let 1 < p < 1. Let
f 2 Lp ./ and let fn 2 Lp ./ be a sequence such that
lim kfn kp D kf kp
n!1
lim kf fn kp D 0:
n!1
4.6. Exercises 173
cannot be removed.
Hint 1. Fix a constant " > 0. Use Egoroff’s
R theorem to construct disjoint mea-
surable sets A; B 2 A such that X D A [ B, A jf jp d < ", .B/ < 1, and fn
R to fp uniformly on B. Use Fatou’s lemma (Theorem 1.41) to prove that
converges
lim sup A jfn j d < ".
n!1
Hint 2. Let gn WD 2p 1 .jfn jp C jf jp / jf fn jp and use Fatou’s lemma
(Theorem 1.41) as in the proof of the Lebesgue dominated convergence theorem
(Theorem 1.45).
for all " > 0. (On page 53 this is called convergence in probability.) Assume
.X/ < 1 and prove the following.
lim kfn f kp D 0;
n!1
(b) there exists a Borel set E X and an element x0 2 X such that every open
neighborhood U of x0 satisfies .U \ E/ > 0 and .U n E/ > 0;
Ea WD ¹Afa j f 2 Fº:
(v) Define \ [
E0 WD Xr [ X n Xs :
r2R s2R
Then .Ef / D 0.
(vii) Define gW X ! R by
´
0 if x 2 E0 ;
g.x/ WD (4.62)
a if x 2 X s for all s > a and x … X r for all r < a:
Example 4.60. This example is closely related to Exercise 3.24, however, it requires
a considerable knowledge of functional analysis and the details go much beyond
the scope of the present manuscript. It introduces the Stone–Čech compactification
X of the natural numbers. This is a compact Hausdorff space containing N and
satisfying the universality property that every continuous map from N to another
compact Hausdorff space Y extends uniquely to a continuous map from X to Y .
176 4. Lp spaces
The space C.X/ of continuous functions on X can be naturally identified with the
space `1 . Hence the space of positive bounded linear functionals on `1 is isomor-
phic to the space of Radon measures on X by Theorem 3.15. Thus the Stone–Čech
compactification of N can be used to understand the dual space of `1 . Moreover, it
gives rise to an interesting example of a Radon measure which is not outer regular
(explained to me by Theo Buehler).
Consider the inclusion
N ! .`1 / W n 7 ! ƒn
ƒn W `1 ! R
defined by
ƒn ./ WD n for D .i /i 2N 2 `1 .
This functional has norm one. Now the space of all bounded linear functionals on
`1 of norm at most one, i.e., the unit ball in .`1 / , is compact with respect to the
weak- topology by the Banach–Alaoglu theorem. Define X to be the closure of
the set ¹ƒn j n 2 Nº in .`1 / with respect to the weak- topology. Thus
ˇ For all finite sequences c 1 ; : : : ; c ` 2 R
8 ˇ 9
ˆ >
ˇ and 1 D .i1 /i2N ; : : : ; ` D .i` /i2N 2 `1>
ˆ
ˆ ˇ >
ˆ
< >
=
X WD ƒ 2 .`1 / ˇˇ satisfying ƒ. j / < c j for j D 1; : : : ; `;
ˇ
:
ˇ there exists an n 2 N such that
ˆ
ˆ ˇ >
>
ˆ
ˆ >
>
: ˇ j < c j for j D 1; : : : ; ` ;
n
The weak- topology U 2X is the smallest topology such that the map
f W X ! R; f .ƒ/ WD ƒ./;
`1 ! C.X/W 7 ! f
is a Banach space isometry. (Verify that kf k WD sup jf .ƒ/j D kk1 for all
ƒ2X
2 `1 .) Thus the dual space of `1 can be understood in terms of the Borel
measures on X.
4.6. Exercises 177
for every Borel set B X. This measure is -finite and inner regular, but not
outer regular. (The set U WD ¹ƒn j n 2 Nº is open, its complement K WD X n U
is compact and has measure zero, and every open set containing K misses only a
finite subset of U and hence has infinite measure.) Now let X0 X be the union of
all open sets in X with finite measure. Then X0 is not -compact and the restriction
of to the Borel -algebra of X0 is a Radon measure, but is not outer regular.
Chapter 5
The Radon–Nikodým theorem
En WD ¹x 2 An j f .x/ ¤ g.x/º
Define
ƒW L2 . C / ! R
by Z
ƒ.f / WD f d:
X
for f 2 L2 .C/. (Here we abuse notation and use the same letter f for a function
in L2 . C / and its equivalence class in L2 . C /.) Then
Z Z
jƒ.f /j jf j d jf j d. C / c kf kL2 .C/
X X
Z Z
0 .E0 / D E d D E g d. C / 0:
0 0
X X
Hence
Z
E g d. C / D 0
0
X
Z Z
.E1 / D E d D .1 g/ d. C / 0:
1
X E1
By Theorem 1.40 this implies that (5.5) continues to hold for every measurable
function f W X ! Œ0; 1/, whether or not it belongs to L2 . C /. Now define the
measurable function
hW X ! Œ0; 1/
by
g.x/
h.x/ WD for x 2 X:
1 g.x/
184 5. The Radon–Nikodým theorem
Since and are -finite measures, there exist sequences of measurable sets
An ; Bn 2 A such that An AnC1 , .An / < 1, Bn BnC1 , .Bn / < 1
for all n, and X D 1
S S1
nD1 An D nD1 Bn . Define Xn WD An \ Bn . Then
It follows from Step 1 that the restriction of fnC1 to Xn agrees with fn -almost
everywhere. Thus, modifying fnC1 on a set of measure zero if necessary, we may
assume without loss of generality that fnC1 jXn D fn for all n 2 N. With this
understood, define
f W X ! Œ0; 1/
by
f jXn WD fn for n 2 N:
5.1. Absolutely continuous measures 185
Here the last but one equality follows from the Lebesgue monotone convergence
theorem (Theorem 1.37). This proves Step 3 and Theorem 5.4.
Example 5.5. Let X be a one-element set and let A WD 2X . Define the measure
W 2X ! Œ0; 1 by .;/ WD 0 and .X/ WD 1.
(i) Choose .;/ WD 0 and .X/ WD 1. Then , R but there does not exist a
(measurable) function f W X ! Œ0; 1 such that X f d D .X/. Thus the
hypothesis that .X; A; / is -finite cannot be removed in Theorem 5.4.
R
(ii) Choose WD . Then we get .A/ D A f d for every nonzero func-
tion f W X ! Œ0; 1/. Thus the hypothesis that .X; A; / is -finite cannot be
removed in Step 1 in the proof of Theorem 5.4.
Example 5.6. Let X be an uncountable set and denote by A 2X the set of all
subsets A X such that either A or Ac is countable. Choose an uncountable subset
H X with an uncountable complement and define ; ; W A ! Œ0; 1 by
´
0 if A is countable;
.A/ WD .A/ WD #.A \ H /; .A/ WD #A:
1 if Ac is countable;
186 5. The Radon–Nikodým theorem
.X n .A \ A0 // D 0:
Let E 2 A. Then
s .E \ A \ A0 / D 0 D 0s .E \ A \ A0 /
and hence
a .E \ A \ A0 / D .E \ A \ A0 / D 0a .E \ A \ A0 /:
a .E n .A \ A0 // D 0 D 0a .E n .A \ A0 //
and so
s .E n .A \ A0 // D .E n .A \ A0 // D 0s .E n .A \ A0 //:
This implies
a .E/ D a .E \ A \ A0 / D 0a .E \ A \ A0 / D 0a .E/
s .E/ D s .E n .A \ A0 // D 0s .E n .A \ A0 // D 0s .E/:
This proves uniqueness.
We prove existence. The measure
WD C W A ! Œ0; 1
is -finite. Hence it follows from the Radon–Nikodým theorem (Theorem 5.4) that
there exist measurable functions f; gW X ! Œ0; 1/ such that
Z Z
.E/ D f d; .E/ D g d for all E 2 A: (5.7)
E E
5.2. The dual space of Lp ./ revisited 187
Define
A WD ¹x 2 X j g.x/ > 0º (5.8)
and
a .E/ WD .E \ A/; s .E/ WD .E \ Ac / for E 2 A: (5.9)
Then it follows directly from (5.9) that the maps a ; s W A ! Œ0; 1 are measures
and satisfy a C s D . Moreover, by (5.9),
Hence Z
a .E/ D .E \ A/ D f d D 0:
E \A
This section is devoted to the proof of Theorem 4.35. Assume throughout that
.X; A; / is a measure space and fix two constants
1 1
1 p < 1; 1 < q 1; C D 1: (5.10)
p q
188 5. The Radon–Nikodým theorem
ƒ.f / WD ƒ.Œf /
ƒg W Lp ./ ! R
via Z
ƒg .f / WD fg d for f 2 Lp ./:
X
The next result proves Theorem 4.35 in -finite case.
Theorem 5.7. Assume .X; A; / is -finite and let ƒW Lp ./ ! R be a bounded
linear functional. Then there exists a function g 2 Lq ./ such that
ƒg D ƒ:
Proof. Assume first that ƒ is positive. We prove in six steps that there exists a
function g 2 Lq ./ such that g 0 and ƒg D ƒ.
Step 1. Define
gW X ! Œ0; 1/
such that Z
.A/ D g d for all A 2 A.
A
5.2. The dual space of Lp ./ revisited 189
ƒ D ƒg :
ƒg .f / D ƒ.f / for all f 2 Lp ./ with f 0. Apply this identity to the functions
f ˙ W X ! Œ0; 1/ for all f 2 Lp ./ to obtain ƒ D ƒg . This proves the assertion
of Theorem 5.7 for every positive bounded linear functional ƒW Lp ./ ! R.
Let ƒW Lp ./ ! R be any bounded linear functional. By Theorem 4.39 there
exist positive bounded linear functionals ƒ˙ W Lp ./ ! R such that ƒ D ƒC ƒ .
Hence, by what we have just proved, there exist functions g ˙ 2 Lq ./ such that
g ˙ 0 and ƒ˙ D ƒg ˙ . Define
g WD g C g :
ƒg D ƒg C ƒg D ƒC ƒ D ƒ:
E WD ¹E 2 A j .E/ < 1º
E W L1 .E / ! L1 ./
defined by
´
f .x/ for x 2 E;
E .f /.x/ WD (5.15)
0 for x 2 X n E:
It descends to a bounded linear operator from L1 .E / to L1 ./ which will still be
denoted by E . Define
ƒE D ƒ ı E W L1 .E / ! R:
such that
Z
ƒE .f / D fgE dE for all E 2 E and all f 2 L1 .E /:
E
F WD ¹gE j E 2 Eº
satisfies the hypotheses of condition (F) on page 174. Thus it follows from (ii) that
there exists a measurable function gW X ! R such that, for all E 2 E, the restriction
gjE agrees with gE almost everywhere on E.
We claim that g 0 almost everywhere. Suppose on the contrary that the set
A WD ¹x 2 X j g.x/ < 0º
has positive measure. Since .X; A; / is semi-finite, there exists a set E 2 E such
that E A and .E/ > 0. Since g.x/ < 0 gE .x/ for all x 2 E it follows that
gjE does not agree with gE almost everywhere, a contradiction. This shows that
g 0 almost everywhere.
We prove that g kƒk almost everywhere. Suppose otherwise that the set
has positive measure. Since .X; A; / is semi-finite, there exists a set E 2 E such
that E AC and .E/ > 0. Since kgE k1 D kƒE k kƒk, it follows from
Lemma 4.8 that gE .x/ kƒk < g.x/ for almost every x 2 E. Hence gjE does
not agree with gE almost everywhere, a contradiction. This shows that g kƒk
almost everywhere and we may assume without loss of generality that
0 g.x/ kƒk
for all x 2 X.
We prove that ƒg D ƒ. Fix a function f 2 L1 ./ such that f 0. Then there
exists a sequence Ei 2 E such that
E1 E2 E3
si W X ! Œ0; 1/
192 5. The Radon–Nikodým theorem
such that
0 s1 s2
the sets
Ei WD ¹x 2 X j si .x/ > 0º
0 si E f f for all i .
i
lim kf E f k1 D 0:
i!1 i
Hence
Here the last step follows from the Lebesgue monotone convergence theorem
(Theorem 1.37). This shows that ƒ.f / D ƒg .f / for every nonnegative integrable
function f W X ! Œ0; 1/. Consequently,
This shows that every positive bounded linear functional on L1 ./ belongs to
the image of the map (5.13). Since every bounded linear functional on L1 ./ is the
difference of two positive bounded linear functionals by Theorem 4.39, it follows
that the map (5.13) is surjective. Thus we have proved that (ii) H) (iii).
We prove that (iii) H) (i). Assume that the map (5.13) is bijective. Then
.X; A; / is semi-finite by part (iv) of Theorem 4.33. Now let E A be any col-
lection of measurable sets. Assume without loss of generality that
E1 ; : : : ; E` 2 E H) E1 [ [ E` 2 E:
ƒE W L1 .E / ! R
by Z
ƒE .f / WD f dE for f 2 L1 .E /: (5.16)
E
E F; f 0 H) ƒE .f / ƒF .f /: (5.17)
Define
ƒW L1 ./ ! R
by
ƒ.f / WD sup ƒE .f C jE / sup ƒE .f jE /: (5.18)
E 2E E 2E
We prove that this is a well defined bounded linear functional with kƒk 1. To see
this, note that Z
ƒE .f jE / f d
X
Moreover, it follows directly from the definition that ƒ.cf / D cƒ.f / for all c 0
and ƒ. f / D ƒ.f /. Now let f; g 2 L1 ./ be nonnegative integrable functions.
194 5. The Radon–Nikodým theorem
Then
D ƒ.f / C ƒ.g/:
To prove the converse inequality, let " > 0 and choose E; F 2 E such that
Hence
ƒ.f C g/ > ƒ.f / C ƒ.g/ 2" for all " > 0
and so
ƒ.f / C ƒ.g/ ƒ.f C g/ ƒ.f / C ƒ.g/:
This shows that
ƒ.f C g/ D ƒ.f / C ƒ.g/
for all f; g 2 L1 ./ such that f; g 0. If f; g 2 L1 ./, then
.f C g/C C f C g D .f C g/ C f C C g C ;
and hence
With this understood, it follows from (iii) that there exists a function g 2 L1 ./
such that ƒ D ƒg . Define
a contradiction. This shows that our assumption .E n H / > 0 was wrong. Hence
.E n H / D 0 for all E 2 E, as claimed.
Now let G 2 A be any measurable set such that .E n G/ D 0 for all E 2 E.
We must prove that .H n G/ D 0. Suppose, by contradiction, that .H n G/ > 0.
Since .X; A; / is semi-finite there exists a measurable set A 2 A such that
0 < .A/ < 1 and A H n G. Then
Z
g d D ƒ.A / D sup ƒE .A jE / D sup .E \ A/ D 0:
A E 2E E 2E
Here the second equality follows from (5.18), the third follows from (5.16), and the
last follows from the fact that E \ A E n G for all E 2 E. Since g > 0 on A,
it follows from Lemma 1.49 that .A/ D 0, a contradiction. This shows that our
assumption that .H n G/ > 0 was wrong and so .H n G/ D 0 as claimed. Thus
we have proved that every collection of measurable sets E A has a measurable
envelope, and this completes the proof of Theorem 5.8.
Proof of Theorem 4.35. For p D 1 the assertion of Theorem 4.35 follows from the
equivalence of (i) and (iii) in Theorem 5.8. Hence assume p > 1. We must prove
that the linear map Lq ./ ! Lp ./ W g 7! ƒg is surjective. Let ƒW Lp ./ ! R
be a positive bounded linear functional and define
Theorem 4.40 also asserts that there exists a measurable set N 2 A such that
.N / D 0 and the restriction of to X n N is -finite. Define
X0 WD X n N; A0 WD ¹A 2 A j A X0 º; 0 WD jA0
Since .X0 ; A0 ; 0 / is -finite, Theorem 5.7 shows that there exists a function
g0 2 Lq .0 / such that g0 0 and
Z
ƒ0 .f0 / D f0 g0 d0 for all f0 2 Lp .0 /:
X0
Define
gW X ! Œ0; 1/
by
´
g0 .x/ for x 2 X0 D X n N ,
g.x/ WD
0 for x 2 N .
Then, by Theorem 4.33,
This proves the assertion for positive bounded linear functionals. Since every
bounded linear functional ƒW Lp ./ ! R is the difference of two positive bounded
linear functionals by Theorem 4.39, this proves Theorem 4.35.
Corollary 5.9. Every -finite measure space is localizable.
Proof. Let .X; A; / be a -finite measure space. Then .X; A; / is semi-finite by
Lemma 4.30. Hence the map L1 ./ ! L1 ./ W g 7! ƒg in (4.31) is injective by
Theorem 4.33 and is surjective by Theorem 5.7. Hence it follows from Theorem 5.8
that .X; A; / is localizable.
5.3. Signed measures 197
Proof. To prove (i) take Ei WD ; in (5.19). To prove (ii) take Ei WD ; for all i > `.
Given a signed measure W A ! R, it is a natural question to ask whether it can
be written as the difference of two measures ˙ W A ! Œ0; 1/. Closely related to
this is the question whether there exists a measure W A ! Œ0; 1/ that satisfies
j.A/j .A/ for all A 2 A: (5.20)
If such a measure exists, it must satisfy
.E; F 2 A; E \ F D ;/ H) .E/ .F / .E [ F /
Thus a lower bound for .A/ is the supremum of the numbers .E/ .F / over
all decompositions of A into pairwise disjoint measurable sets E and F . The next
theorem shows that this supremum defines the smallest measure that satisfies (5.20).
Theorem 5.12. Let W A ! R be a signed measure and define
ˇ E; F 2 A; =
8 ˇ 9
<
for A 2 A:
ˇ
jj.A/ WD sup .E/ .F / ˇˇ E \ F D ;; (5.21)
: ˇ E [F DA;
Then j.A/j jj.A/ < 1 for all A 2 A and jjW A ! Œ0; 1/ is a measure,
called the total variation of .
198 5. The Radon–Nikodým theorem
Proof. We prove that jj is a measure. If follows directly from the definition that
jj.;/ D 0 and jj.A/ j.A/j 0 for all A 2 A. We must prove that the
function jjW A ! Œ0; 1 is -additive. Let Ai 2 A be a sequence of pairwise
disjoint measurable sets and put
1
[
A WD Ai :
i D1
E \ F D ;; E [ F D A: (5.22)
Then
1
[ 1
[
ED .E \ Ai /; F D .F \ Ai /:
iD1 i D1
Hence
1
X 1
X
.E/ .F / D .E \ Ai / .F \ Ai /
iD1 i D1
1
X
D ..E \ Ai / .F \ Ai //
iD1
1
X
jj.Ai /:
iD1
Take the supremum over all pairs of measurable sets E; F satisfying (5.22) to obtain
1
X
jj.A/ jj.Ai / (5.23)
i D1
To prove the converse inequality, fix a constant " > 0. Then there are sequences of
measurable sets Ei ; Fi 2 A such that
"
Ei \ Fi D ;; Ei [ Fi D Ai ; .Ei / .Fi / > jj.Ai /
2i
for all i 2 N. The sets
1
[ 1
[
E WD Ei and F WD Fi
iD1 i D1
5.3. Signed measures 199
Hence
1
X
jj.A/ > jj.Ai / " for all " > 0.
iD1
Thus
1
X
jj.A/ jj.Ai /
iD1
and so
1
X
jj.A/ D jj.Ai /
iD1
by (5.23). This shows that jj is a measure.
It remains to prove that jj.X/ < 1. Suppose, by contradiction, that
jj.X/ D 1:
Claim. Let A 2 A such that jj.X n A/ D 1. Then there exists a measurable set
B 2 A such that A B, j.B n A/j 1, and jj.X n B/ D 1.
There exist measurable sets E; F such that E \ F D ;, E [ F D X n A, and
.E/ .F / 2 C j.X n A/j;
Definition 5.13. Let W A ! R be a signed measure and let jjW A ! Œ0; 1/ the
measure in Theorem 5.12. The Jordan decomposition of is the representation of
as the difference of two measures ˙ whose sum is equal to jj. The measures
˙ W A ! Œ0; 1/
are defined by
jj.A/ ˙ .A/
˙ .A/ WD D sup¹˙.E/ j E 2 A; E Aº (5.24)
2
for A 2 A and they satisfy
C D ; C C D jj: (5.25)
Exercise 5.14. Let .X; A; / be a measure space, let f 2 L1 ./, and define
Z
.A/ WD f d for A 2 A.
A
Step 1. Let A 2 A. Then jj.A/ D 0 if and only if .E/ D 0 for all measurable
sets E A.
The signed measure is concentrated on A if and only if .E/ D .E \ A/ for all
E 2 A, or equivalently .E n A/ D 0 for all E 2 A. By Step 1, this holds if and
only if jj.X n A/ D 0.
such that
D a C s ; a ; s ? : (5.27)
a WD C
a a and s WD C
s s
satisfy (5.27).
202 5. The Radon–Nikodým theorem
Hence ja 0a j and ja 0a j ? by part (iii) of Lemma 5.2. Thus, by
part (iv) of Lemma 5.2, ja 0a j D 0, and therefore a D 0a and s D 0s . This
proves Theorem 5.17.
We prove that h.x/ 2 ¹1; 1º for -almost every x 2 X. To see this, fix a real
number 0 < r < 1 and set
Ar WD ¹x 2 X j jh.x/j rº:
r.Ar /:
P WD ¹x 2 X j h.x/ 1º;
N WD ¹x 2 X j h.x/ 1º:
Then P \ N D ;, P [ N D X, and
Z
.P / h d D .P / .P /;
P
Z
.N / .N / D h d .N /:
N
Hence
Z
.h 1/ d D .P / .P / D 0;
P
Z
.h C 1/ d D .N / C .N / D 0:
N
(i) C C D jj;
(ii) C ? ;
(iii) there exists a measurable set P 2 A such that C .A/ D .A \ P / and
.A/ D .A n P / for all A 2 A.
hW X ! ¹˙1º
such that Z
.A/ D h d jj
A
Hence C ? .
(ii) H) (iii). By (ii), there exists a measurable set P 2 A such that
C .P c / D 0 and .P / D 0:
Therefore,
C .A/ D C .A \ P / D C .A \ P / .A \ P / D .A \ P /;
.A/ D .A n P / D .A n P / C .A n P / D .A n P /
for all A 2 A.
5.4. Radon–Nikodým generalized 205
Take the supremum over all such pairs E; F 2 A to obtain the inequality
Proof. That (ii) implies (i) is obvious. Conversely, assume (i). Then jj by
Lemma 5.16. Assume by contradiction that (ii) does not hold. Then there exist a
constant " > 0 and a sequence of measurable sets Ai 2 A such that
.Ai / 2 i ; j.Ai /j " for all i 2 N:
For n 2 N define
1
[ 1
\
Bn WD Ai ; B WD Bn :
iDn nD1
Then
1
Bn BnC1 ; .Bn / ; jj.Bn / jj.An / j.An /j "
2n 1
for all n 2 N. Hence .B/ D 0 and jj.B/ D limn!1 jj.Bn / " by part (v) of
Theorem 1.28. This contradicts the fact that jj and shows that our assumption
that (ii) does not hold was wrong. Thus (i) implies (ii) and this proves Lemma 5.21.
Definition 5.22. The signed measure is called truly continuous with respect to
if, for every " > 0, there exist a constant ı > 0 and a measurable set E 2 A such
that .E/ < 1 and
A 2 A; .A \ E/ < ı H) j.A/j < ": (5.31)
Lemma 5.23. The following are equivalent:
(i) is truly continuous with respect to ;
(ii) jj is truly continuous with respect to ;
(iii) C and are truly continuous with respect to .
Proof. Assume (i), fix a constant " > 0, and choose ı > 0 and E 2 A such
that .E/ < 1 and (5.31) holds. Let A 2 A such that .A \ E/ < ı. Then
.B/ .A n B/ < 2" for every measurable set B A, and hence jj.A/ 2"
by Theorem 5.12. This shows that (i) H) (ii). That (ii) H) (iii) and (iii) H) (i)
follows directly from the definitions.
5.4. Radon–Nikodým generalized 207
Definition 5.22 is due to Fremlin [4, Chapter 23]. If the measure space .X; A; /
is -finite, then is truly continuous with respect to if and only if it is absolutely
continuous with respect to by Theorem 5.26 below. However, for general mea-
sure spaces the condition of true continuity is stronger than absolute continuity.
The reader may verify that, when .X; A; / and are as in part (i) of Example 5.5
or Example 5.6, the finite measure is absolutely continuous with respect to ,
but is not truly continuous with respect to . Fremlin’s condition was reformulated
by König [8] in terms of inner regularity of with respect to . This notion can be
defined in several equivalent ways. To formulate the conditions it is convenient to
introduce the notation
for all A 2 A. Such a measurable set P will be fixed throughout the proof.
We prove that (i) implies (ii). Fix a set A 2 A such that jj.A \ E/ D 0 for all
E 2 E. Then it follows from (5.32) that .A \ E \ P / D .A \ E n P / D 0 for
all E 2 E. By (i), this implies .A \ P / D .A n P / D 0 and hence jj.A/ D 0
by (5.32). This shows that (i) implies (ii).
We prove that (ii) implies (i). Fix a set A 2 A such that .A \ E/ D 0
for all E 2 E. Since E \ P 2 E and E n P 2 E for all E 2 E, this implies
.A \ E \ P / D .A \ E n P / D 0 for all E 2 E. Hence it follows from (5.32)
that jj.A \ E/ D 0 for all E 2 E. By (ii), this implies jj.A/ D 0 and hence
.A/ D 0 because j.A/j jj.A/. This shows that (ii) implies (i).
208 5. The Radon–Nikodým theorem
lim jj.Ei / D c:
i!1
For i 2 N define
Fi WD E1 [ E2 [ [ Ei :
Then
Fi 2 E; Fi Fi C1 A; jj.Ei / jj.Fi / c (5.34)
for all i , and hence
lim jj.Fi / D c: (5.35)
i !1
Put
1
[
B WD A n F; F WD Fi : (5.36)
iD1
Then
jj.F / D lim jj.Fi / D c
i!1
by (5.35). Hence jj.E/ D 0 for all E 2 E with E B and it follows from (ii)
that jj.B/ D 0. Therefore, (5.37) yields jj.A/ D c. This shows that (ii) H) (iii).
That (iii) H) (ii) is obvious and this proves Lemma 5.24.
Definition 5.25. The signed measure is called inner regular with respect to if
it satisfies the equivalent conditions of Lemma 5.24.
5.4. Radon–Nikodým generalized 209
If these equivalent conditions are satisfied, then the function f in (iii) is uniquely
determined by (5.28) up to equality -almost everywhere.
First proof of Theorem 5.26. This proof is due to König [8]. It has the advantage
that it reduces the proof of the generalized Radon–Nikodým theorem to the standard
Radon–Nikodým theorem (Theorem 5.18) for -finite measure spaces.
(i) H) (ii). To see that is absolutely continuous with respect to , fix a
measurable set A 2 A such that .A/ D 0 and fix a constant " > 0. Choose
ı > 0 and E 2 A such that .E/ < 1 and (5.31) holds. Then we obtain
.A \ E/ .A/ D 0 < ı, and hence j.A/j < " by (5.31). Thus j.A/j < " for
all " > 0, and hence .A/ D 0. This shows that .
We prove that is inner regular with respect to by verifying that satisfies
condition (i) in Lemma 5.24. Fix a set A 2 A such that
We must prove that .A/ D 0. Let " > 0 and choose ı > 0 and E 2 A such that
.E/ < 1 and (5.31) holds. Then ..A n E/ \ E/ D 0 < ı, hence j.A n E/j < "
by (5.31), so
This shows that j.A/j < " for all " > 0 and so .A/ D 0, as claimed. Thus we
have proved that (i) H) (ii).
(ii) H) (iii). Since is inner regular with respect to , there exists a sequence
of measurable sets Ei 2 A such that Ei Ei C1 and .Ei / < 1 for all i 2 N and
jj.X/ D lim jj.Ei /. Define
i !1
1
[
X0 WD Ei ; A0 WD ¹A 2 A j A X0 º; 0 WD jA0 ; 0 WD jA0 :
i D1
Moreover,
c D jj.X/ D jj.E1 / D lim jj.En /:
n!1
Now fix a constant " > 0. Choose n 2 N such that jj.En / > c "=2 and put
n 1
ı WD 2 ":
If A 2 A such that .A \ En / < ı, then
jj.A/ D jj.A n En / C jj.A \ En /
jj.X n En / C 2n .A \ En /
"
< C 2n ı D ":
2
This shows that is truly continuous with respect to . This completes the first
proof of Theorem 5.26.
5.4. Radon–Nikodým generalized 211
Second proof of Theorem 5.26. This proof is due to Fremlin [4, Chapter 23].
It shows directly that (i) implies (iii) and has the advantage that it only uses the Hahn
decomposition theorem (Theorem 5.19). It thus also provides an alternative proof
of Theorem 5.18 (assuming the Hahn decomposition theorem) which is of interest
on its own. By Lemma 5.23, it suffices to consider the case where W A ! Œ0; 1/
is a finite measure that is truly continuous with respect to .
Consider the set
² ˇ ³
ˇ f is measurable and
F WD f W X ! Œ0; 1/ ˇˇ R :
A f d .A/ for all A 2 A
and1
f; g 2 F H) max¹f; gº 2 F: (5.38)
Now put Z
c WD sup f d .X/
f 2F X
fi WD max¹g1 ; g2 ; : : : ; gi º 2 F
and Z Z
gi d fi d c for all i 2 N.
X X
1
Let f; g 2 F and A 2 A and define the sets
Af WD ¹x 2 A j f .x/ > g.x/º
and
Ag WD ¹x 2 A j g.x/ f .x/ºI
then Af ; Ag 2 A, Af \ Ag D ;, and Af [ Ag D A; hence
Z Z Z
max¹f; gº d D f d C g d .Af / C .Ag / D .A/:
A Af Ag
212 5. The Radon–Nikodým theorem
Hence f < 1 -almost everywhere by Lemma 1.47 and we may assume without
loss of generality that f .x/ < 1 for all x 2 X . Thus f 2 F.
We claim that Z
f d D .A/ for all A 2 A.
A
Suppose on the contrary that there exists a set A0 2 A such that
Z
f d < .A0 /:
A0
Define
0 .A0 /
" WD > 0: (5.41)
3
Since is truly continuous with respect to , so is 0 . Hence there are a ı > 0 and
a set E 2 A such that .E/ < 1 and
Then take A WD A0 . Since 0 .A0 / D 3" " by (5.41), it follows from (5.42) that
.A \ P \ E/
Z
0
.A \ P / " D h d:
.E/ A
Thus Z
h d 0 .A/ for all A 2 A,
A
R
this contradicts the definition of c. Thus we have proved that A f d D .A/
for all A 2 A, and hence f satisfies (5.28). This completes the second proof of
Theorem 5.26.
5.5 Exercises
Exercise 5.27. Let .X; A; / be a measure space such that .X/ < 1. Define
(denoted by the same letter) and that the pair .A=; / is a complete metric space.
Prove that the function
Z
A ! RW A 7 ! f d
A
Exercise 5.28 (Rudin [17, page 133]). Let .X; A; / be a measure space. A subset
F L1 ./ is called uniformly integrable if, for every " > 0, there is a constant
ı > 0 such that, for all E 2 A and all f 2 F,
ˇZ ˇ
ˇ ˇ
.E/ < ı H) ˇ f dˇˇ < ":
ˇ
E
5.5. Exercises 215
(i) Every finite subset of L1 ./ is uniformly integrable. Hint. Lemma 5.21.
(v) Find an example where Vitali’s theorem applies although the hypotheses of
the Lebesgue dominated convergence theorem are not satisfied.
(vi) Find an example of a measure space .X; A; / with .X/ < 1 and a sequence
fn 2 L1 ./ that is not uniformly integrable, converges pointwise to zero, and
satisfies Z
lim fn d D 0:
n!1 X
E WD E0 n A and E WD E0 [ A
both satisfy .E; E0 / < ı. Deduce that, for all A 2 A and all n 2 N,
ˇZ ˇ
ˇ ˇ
.A/ < ı; n n0 H) ˇ .fn fn0 / dˇˇ < 2":
ˇ (5.47)
A
Now use part (i) to find a constant ı 0 > 0 such that, for all A 2 A,
ˇZ ˇ
0
ˇ ˇ
.A/ < ı H) sup ˇ fn dˇˇ < 3":
ˇ (5.48)
n2N A
Exercise 5.29 (Rudin [17, page 134]). Let .X; A; / be a measure space such that
.X/ < 1 and fix a real number p > 1. Let f W X ! R be a measurable function
and let fn 2 L1 ./ be a sequence that converges pointwise to f and satisfies
Z
sup jfn jp d < 1:
n2N X
Prove that Z
f 2 L1 ./; lim jf fn j d D 0:
n!1 X
(1) If B 2 B and 0 < c < .B/, then there exists a Borel set A B such that
.A/ D c. Hint. Show that the function f .t/ WD .B \Œ t; t/ is continuous.
.B/ D c H) .B/ D c:
(i) .
(ii) is not inner regular with respect to .
(iii) There does not exist any measurable function f W X ! Œ0; 1 such that
Z
.B/ D f d for all B 2 B.
B
Exercise 5.32. Let X WD Œ1; 1/, denote by B 2X the Borel -algebra, and let
W B ! Œ0; 1 be the restriction of the Lebesgue measure to B. Let W B ! Œ0; 1
be a Borel measure such that
1
Z
2 .B/ WD dx
B\Œ1;1/ x2
for B 2 B.
2
Here we denote by Z Z
f .x/ dx WD f d
B B
the Lebesgue integral of a Borel measurable function f W Œ0; 1/ ! Œ0; 1/ over a Borel set B 2 B.
218 5. The Radon–Nikodým theorem
1 ; 2 ; 1 2 ; 2 1 ;
and
6 1 ; 6 2 :
Exercise 5.34. Let .X; A; / be a measure space. Show that the signed measures
W A ! R form a Banach space M D M.X; A/ with the norm
kk WD jj.X/:
C.X/ WD Cc .X/
Let M.X/ denote the space of signed Borel measures as in Exercise 5.34.
For 2 M.X/ define the linear functional
ƒ W C.X/ ! R
by Z
ƒ .f / WD f d:
X
Prove the following.
(i) kƒ k D kk. Hint. Use the Hahn decomposition theorem (Theorem 5.19)
and the fact that every Borel measure on X is regular by Theorem 3.18.
(ii) Every bounded linear functional on C.X/ is the difference of two positive
linear functionals. Hint. For f 2 C.X/ with f 0 prove that
ƒC .f / WD sup¹ƒ.hf / j h 2 C.X/; 0 h 1º
(5.52)
D sup¹ƒ.g/ j g 2 C.X/; 0 g f º:
5.5. Exercises 219
Here the second supremum is obviously greater than or equal to the first.
To prove the converse inequality, show that, for all g 2 C.X/ with
0 g f and all " > 0 there is an h 2 C.X/ such that 0 h 1 and
jƒ.g hf /j < ". Namely, find 2 C.X/ such that 0 1, .x/ D 0
when f .x/ "=2 kƒk and .x/ D 1 when f .x/ "= kƒk; then define
h WD g=f . Once (5.52) is established, show that ƒC extends to a positive
linear functional on C.X/.
(iii) The map M.X/ ! C.X/ W 7! ƒ is bijective. Hint. Use the Riesz repre-
sentation theorem (Theorem 3.15).
(iv) The hypothesis that every open subset of X is -compact cannot be removed
in part (i). Hint. Consider Example 3.6.
Exercise 5.36. Let .X; A; / be a measure space and let f W X ! Œ0; 1/ be a
measurable function. Define the measure f W A ! Œ0; 1 by
Z
f .A/ WD f d for A 2 A: (5.53)
A
Lemma 6.1. Let 0 ˛ < 1 and 0 t < 1. Then the following holds:
(ii) f .˛/ D t if and only if f .t/ ˛ and f .s/ > ˛ for 0 s < t;
Proof. It follows directly from the definition of f in (6.2) that f .˛/ D 1 if and
only if .s; f / > ˛ for all s 2 Œ0; 1/, and this proves (i).
To prove (ii), fix a constant 0 t < 1. Assume first that
f .t/ D .A.t; f //
D lim .A.t C 1=n; f //
n!1
˛:
kf k1;1 kgk1;1
kf C gk1;1 C for 0 < < 1; (6.5)
1
q q q
kf C gk1;1 kf k1;1 C kgk1;1 : (6.6)
Proof. For 0 < t; c < 1 it follows from part (iii) of Lemma 6.1 that
1
t.t; f / c () .t; f / ct
() f .ct 1
/t
1
() ct f .ct 1
/ c:
Moreover,
Z Z
t.t; f / D t.A.t; f // jf j d jf j d for all t > 0.
A.t;f / X
and so
kf k1;1 kgk1;1
t.t; f C g/ t.t; f / C t..1 /t; g/ C
1
for all t > 0. Take the supremum over all t > 0 to obtain (6.5).
224 6. Differentiation
f WX ! R
A subset of L1;1 ./ is open in the topology determined by the metric (6.9) if
and only if it is a union of sets of the form ¹Œg 2 L1;1 ./ j kf gk1;1 < rº
with f 2 L1;1 ./ and r > 0. A sequence Œfi 2 L1;1 ./ converges to Œf
in this topology if and only if lim kfi f k1;1 D 0. The inequality (6.3) in
i!1
Lemma 6.2 shows that
L1 ./ L1;1 ./
for every measure space .X; A; /. In general, L1;1 ./ is not equal to L1 ./.
For example, the function
f WR ! R
defined by
81
< for x > 0,
f .x/ WD x
:
0 for x 0,
is weakly integrable, but not integrable.
fi W X ! R
whose equivalence classes Œfi form a Cauchy sequence in L1;1 ./ with respect
to the metric (6.9). Then there is a subsequence
such that
2k
kfik fikC1 k1;1 < 2 for all k 2 N.
For k; ` 2 N put
k
Ak WD A.2 ; fik fikC1 /
and
1
[ 1
\
E` WD Ak ; E WD E` :
kD` `D1
Then
k 2k
2 .Ak /
fik fikC1
1;1 < 2 for all k 2 N,
hence
1
X 1
X
k
.E` / .Ak / 2 D 21 `
kD` kD`
f .x/ WD 0 for x 2 E.
We prove that
lim kfi f k1;1 D 0
i!1
and hence also f 2 L1;1 ./. To see this, fix a constant " > 0 and choose an integer
i0 2 N such that
.i; j 2 N; i; j i0 / H) 4
fi fj
1;1 < ":
i` i0 ; 22 ` t "; 22 `
t:
6.2. Maximal functions 227
and so
1
X 1
X
k
jfi` .x/ f .x/j jfik .x/ fikC1 .x/j 2 D 21 `
t=2:
kD` kD`
tfi f .t/ tfi fi` .t=2/ C tfi f .t=2/ 2kfi fi` k1;1 C "=2 < "
`
f WR ! R
by
f .x/ WD .. 1; x// for x 2 R: (6.10)
Fix two real numbers x; A 2 R. Then the following are equivalent:
(ii) For every " > 0 there is a ı > 0 such that, for every open interval U R,
ˇ ˇ
ˇ .U / ˇ
x 2 U; m.U / < ı H) ˇ ˇ Aˇˇ ": (6.11)
m.U /
228 6. Differentiation
Proof. (i) H) (ii). Fix a constant " > 0. Since f is differentiable at x and we have
f 0 .x/ D A, there exists a constant ı > 0 such that, for all y 2 R,
ˇ ˇ
ˇ f .x/ f .y/ ˇ
0 < jx yj < ı H) ˇ ˇ Aˇˇ ": (6.12)
x y
Let a; b 2 R such that a < x < b and b a < ı. Then, by (6.12),
ˇ ˇ ˇ ˇ
ˇ f .x/ f .a/ ˇ ˇ f .b/ f .x/ ˇ
ˇ Aˇˇ "; ˇ Aˇˇ ";
ˇ x a ˇ b x
or, equivalently,
".x a/ f .x/ f .a/ A.x a/ ".x a/;
Thus (6.12) holds for x < y < x C ı and an analogous argument proves the
inequality for x ı < y < x. Thus (ii) implies (i) and this proves Theorem 6.6.
6.2. Maximal functions 229
The main theorem of this chapter will imply that, when is absolutely con-
tinuous with respect to m, the derivative of the function f in (6.10) exists almost
everywhere, defines a Lebesgue integrable function f 0 W R ! R, and that
Z
.A/ D f 0 dm
A
for all Lebesgue measurable sets A 2 A. It will then follow that an absolutely
continuous function on R can be written as the integral of its derivative. This is the
fundamental theorem of calculus in measure theory (Theorem 6.19).
The starting point for this program is the assertion of Theorem 6.6. It suggests
the definition of the derivative of a signed measure
W A ! R
at a point x 2 R as the limit of the quotients .U /=m.U / over all open intervals U
containing x as m.U / tends to zero, provided that the limit exists. This idea carries
over to all dimensions and leads to the concept of a maximal function, which we
explain next.
Notation. Fix a natural number n 2 N. Let .Rn ; A; m/ denote the Lebesgue mea-
sure space and let
n
B 2R
denote the Borel -algebra of Rn with the standard topology. Thus L1 .Rn / denotes
the space of Lebesgue integrable functions f W Rn ! R. An element of L1 .Rn /
need not be Borel measurable, but differs from a Borel measurable function on a
Lebesgue null set by Theorem 2.14 and part (v) of Theorem 1.55. For x 2 Rn and
r > 0, denote the open ball of radius r, centered at x, by
Here q
jj WD 12 C C n2
denotes the Euclidean norm of D .1 ; : : : ; n / 2 Rn .
Definition 6.7 (Hardy–Littlewood maximal function). Let W B ! Œ0; 1/ be a
finite Borel measure. The maximal function of is the function
MW Rn ! R
defined by
.Br .x//
.M/.x/ WD sup : (6.13)
r>0 m.Br .x//
230 6. Differentiation
Lemma 6.9. Let W B ! Œ0; 1/ be a finite Borel measure. Then the maximal
function MW Rn ! R is lower semi-continuous and hence is Borel measurable.
We prove that U t is open. Fix an element x 2 U t . Since .M/.x/ > t, there exists
a number r > 0 such that
.Br .x//
t< :
m.Br .x//
Choose ı > 0 such that
.r C ı/n .Br .x//
t n
< :
r m.Br .x//
Further, choose y 2 Rn such that jy xj < ı. Then Br .x/ BrCı .y/, and hence
This implies
.BrCı .y//
.M/.y/ > t;
m.BrCı .y//
and hence y 2 U t . This shows that U t is open for all t > 0. It follows that
n
S
U0 D t >0 U t is open and U t D R is open for t < 0. Thus M is lower
semi-continuous, as claimed. This proves Lemma 6.9.
Proof. Abbreviate Bi WD Bri .xi / and choose the ordering such that
r1 r2 r` :
Choose i1 WD 1 and let i2 > 1 be the smallest index such that Bi2 \ Bi1 D ;.
Continue by induction to obtain a sequence
such that
Then,
and hence
`
[ k
[
W D Bi B3rij .xij /:
iD1 j D1
U t WD A.t; M /
x1 ; : : : ; x ` 2 K
such that
`
[
K Bri .xi /;
iD1
Here the second step follows from (6.18) with ri D r.xi / and the last step fol-
lows from the fact that the balls Bri .xi / for i 2 S are pairwise disjoint. Take the
supremum over all compact sets K U t to obtain
3n
m.A.t; M // D m.U t / jj.Rn /: (6.19)
t
(See Theorem 2.13.) Multiply the inequality (6.19) by t and take the supremum over
all real numbers t > 0 to obtain kM k1;1 3n jj.Rn /. This proves Theorem 6.8.
6.3. Lebesgue points 233
Mf W Rn ! Œ0; 1/
defined by
1
Z
.Mf /.x/ WD sup jf j d m for x 2 Rn : (6.20)
r>0 m.Br .x// Br .x/
Corollary 6.12. Let f 2 L1 .Rn / and define the signed Borel measure f on Rn by
Z
f .B/ WD f dm
B
is continuous (because jMf Mgj M.f g/). Note that it is not linear.
By Theorem 6.8, M extends naturally to an operator 7! M from the Banach
space of signed Borel measures on Rn to L1;1 .Rn /. (See Exercise 5.34.)
Tr f W Rn ! Œ0; 1/
by
1
Z
.Tr f /.x/ WD jf f .x/j d m for x 2 Rn : (6.22)
m.Br .x// Br .x/
Tf W Rn ! Œ0; 1
by
.Tf /.x/ WD lim sup.Tr f /.x/ for x 2 Rn : (6.23)
r!0
hi WD f gi
Thus
Tr hi M hi C jhi j for all i and all r > 0.
This implies
A."; Tf / A."=2; M hi / [ A."=2; hi / for all i and all " > 0. (6.24)
(See (6.1) for the notation A."; Tf /, etc.) Since hi and M hi are Borel measurable
(see Theorem 6.8) the set
2 2 1
m.A."=2; hi // khi k1;1 khi k1 i 1
" " 2 "
and, by Theorem 6.8,
2 2 3n 3n
m.A."=2; M hi // kM hi k1;1 khi k1 i 1 :
" " 2 "
Thus
3n C 1
m.Ei ."// :
2i 1 "
Since this holds for all i 2 N, it follows that the Borel set
1
\
E."/ WD Ei ."/
iD1
A.1=k; Tf / E.1=k/
D E:
The first consequence of Theorem 6.14 discussed here concerns a signed Borel
measure on Rn that is absolutely continuous with respect to the Lebesgue mea-
sure. The following theorem provides a formula for the function f in Theorem 5.18
(also called the Radon–Nikodým derivative of ).
.Br .x//
f .x/ D lim for all x 2 Rn n E: (6.26)
r!0 m.Br .x//
6.3. Lebesgue points 237
1
Z
jf f .x/j d m
m.Br .x// Br .x/
for all r > 0 and all x 2 Rn , it follows that (6.26) holds for all x 2 Rn n E.
This proves Theorem 6.15.
Proof. Since x is a Lebesgue point of f , there exists a constant ı > 0 such that
Z
1
m.Br .x// jf f .x/j d m < ˛" for 0 < r < ı.
Br .x/
Assume 0 < r < ı and let E Br .x/ be a Borel set such that m.E/ ˛m.Br .x//.
Then
ˇ ˇ
ˇ 1 1
Z Z
ˇ
ˇ
ˇ m.E/ f d m f .x/ˇ
ˇ jf f .x/j d m
E m.E/ E
1
Z
jf f .x/j d m
˛m.Br .x// Br .x/
< ":
Proof. The proof follows an argument in Heil [7, Section 3.4]. By assumption and
Lemma 5.16, there exists a Borel set A Rn such that
m.A/ D 0; jj.Rn n A/ D 0:
For " > 0, define the set
° ˇ jj.Br .x// ±
A" WD x 2 Rn n A ˇ lim sup >" :
ˇ
r!0 m.Br .x//
We claim that A" is a Lebesgue null set for every " > 0. To see this, fix two constants
" > 0 and ı > 0. Since the Borel measure jj is regular by Theorem 3.18, there
exists an open set Uı Rn such that
A Uı ; jj.Uı / < ı:
For x 2 A" choose a radius r D r.x/ > 0 such that
jj.Br.x/ .x//
> "; Br.x/ .x/ Uı ;
m.Br.x/ .x//
and consider the open set
[
Wı WD Br.x/ .x/ Uı :
x2A"
Fix a compact subset K Wı and cover K by finitely many of the balls Br.x/ .x/
with x 2 A" . By Vitali’s covering Lemma 6.10, there are elements x1 ; : : : ; xN 2 A"
such that the balls Br.xi / .xi / are pairwise disjoint and K N
S
i D1 B3r.xi / .xi /. Thus
N
X
m.K/ 3n m.Br.xi / .xi //
i D1
N
3n X
< jj.Br.xi / .xi //
"
iD1
N
3n X
jj.Uı /
"
iD1
n
3 ı
:
"
6.4. Absolutely continuous functions 239
This holds for every compact set K Wı and hence m.Wı / 3n ı=". Since ı > 0
was arbitrary, the set A" is contained in an open set of arbitrarily small Lebesgue
measure and so is a Lebesgue null set as claimed. This implies that
1
[
E WD A [ A1=k
kD1
is a Lebesgue null set. It satisfies jj.Rn n E/ D 0 and (6.28) by definition, and this
proves Theorem 6.17.
Proof. We prove that (iii) implies the last assertion of the theorem. Thus assume that
there exists a function g 2 L1 .I / that satisfies (6.30) for all x; y 2 I with x < y.
Then Theorem 6.14 asserts that there exists a Borel set E I of Lebesgue measure
zero such that every element of I n E is a Lebesgue point of g. By Theorem 6.16
with ˛ D 1=2, every element x 2 I n E satisfies condition (ii) in Theorem 6.6 with
A WD g.x/. Hence Theorem 6.6 asserts that the function f is differentiable at every
point x 2 I n E and satisfies f 0 .x/ D g.x/ for x 2 I n E.
(iii) H) (i). Thus assume f satisfies (iii) and define the signed measure
W B ! R
by Z
.B/ WD g dm (6.31)
B
for every Borel set B I . Then is absolutely continuous with respect to the
Lebesgue measure and Z
jj.B/ D jgj d m
B
for ever Borel set B I . Now let " > 0. Since jj m, it follows from
Lemma 5.21 that there exists a constant ı > 0 such that jj.B/ < " for every
Borel set B I with m.B/ < ı. Choose a sequence
s1 t1 s` t`
in I such that
`
X
jti si j < ı
iD1
and define
Ui WD .si ; ti / for i D 1; : : : ; `.
Then the Borel set
`
[
B WD Ui
iD1
Hence
jj.B/ < ":
6.4. Absolutely continuous functions 241
Since ˇZ ˇ Z
ˇ ˇ
jf .ti / f .si /j D ˇˇ g d mˇˇ jgj d m D jj.Ui /
Ui Ui
f .y/ f .x/ D .Œx; y/ D ..x; y/ D .Œx; y// D ..x; y// (6.32)
D a C s ; a m; s ? m: (6.33)
fa ; fs W I ! R
242 6. Differentiation
by
Z x
fa .x/ WD f .a/ C a .Œa; x/ D f .a/ C g.t/ dt; fs .x/ WD s .Œa; x/:
a
Then f D fa C fs by (6.32) and (6.33). Since (iii) implies (i) and (i) implies (ii)
(already proved), both functions f and fa satisfy (ii) and fa is absolutely continu-
ous. Moreover fs D f fa is continuous.
It remains to prove that fs 0. The proof given below follows an argument in
Heil [7, Section 3.5.4]. By Theorem 6.17, there exists a Lebesgue null set Es I
such that a; b 2 Es (without loss of generality) and
js ..x r; x C r//j
js j.I n Es / D 0; lim D0 for all x 2 I n Es :
r!0 r
This implies that every element x 2 I n Es satisfies condition (ii) in Theorem 6.6
with A D 0 and f replaced by fs . Hence fs is differentiable at every point x 2 I nEs
and fs0 .x/ D 0 for x 2 I n Es .
Since Es is a Lebesgue null set, so are the sets f .Es / and fa .Es /, because the
functions f and fa satisfy (ii). Since the difference of two Lebesgue null sets in R
is a Lebesgue null set, by Exercise 2.27, it follows that
fs .Es / D ¹f .x/ fa .x/ j x 2 Es º ¹f .x/ fa .y/ j x; y 2 Es º
is a Lebesgue null set.
We claim that fs .I n Es / is also a Lebesgue null set. To see this, fix a constant
" > 0. For n 2 N define the set
° ˇ 1 ±
An WD x 2 I n Es ˇ y 2 I; jx yj < H) jfs .x/ fs .y/j "jx yj :
ˇ
n
Then
An AnC1 for all n 2 N
and [
I n Es D An :
n2N
Here the last assertion follows from the fact that fs is differentiable on I n Es with
derivative zero. We prove that the Lebesgue outer measure of the set fs .I n Es /
satisfies the estimate
.fs .I n Es // ".b a C "/: (6.35)
To see this, cover the set An by at most countably many open intervals Ui , each of
length less than 1=n, such that
X
m.Ui / .An / C "
i
6.5. Exercises 243
Since the Lebesgue outer measure is continuous from below by part (iii) of Theo-
rem 2.13 and [
fs .I n Es / D fs .An /;
n2N
it follows that
This proves (6.35). Since " > 0 was chosen arbitrary, this implies that fs .I n Es / is
a Lebesgue null set, as claimed. Since fs .Es / is also a Lebesgue null set, as noted
above, it follows that fs .I / is a Lebesgue null set. Finally, since fs is continuous
and fs .0/ D 0 by definition, this implies fs 0. Hence f D fa is absolutely
continuous and this proves Theorem 6.19.
6.5 Exercises
Exercise 6.20. Let I D Œa; b R be a compact interval and let B 2I be the
Borel -algebra. A function f W I ! R is said to be of bounded variation if
`
X
V .f / WD sup jf .ti / f .ti 1 /j < 1: (6.36)
aDt0 <t1 <<t` Db i D1
(This is the Riemann–Stieltjes integral. See Körner [9] and compare it with the
Riemann integral [9, 18, 21].) Prove that ƒf W C.I / ! R is a positive linear
functional. Use the Riesz representation theorem R(Theorem 3.15) to find a
Borel measure f W B ! Œ0; 1/ such that ƒf .h/ D I h df for all h 2 C.I /.
Use the fact that f is right continuous to prove that f satisfies (6.38).
Exercise 6.21. Let .R; A; m/ be the Lebesgue measure space and fix a constant
0 < " < 1=2. Prove that there does not exist a Lebesgue measurable set E R
such that
m.E \ I /
"< <1 "
m.I /
for every nonempty bounded open interval I R.
Hint. Consider the function
f WD E \Œ 1;1
Hint. Assume first that f 0. Use Theorem 1.26 to find a sequence of measur-
able sets Ei 2 A, not necessarily disjoint, and a sequence of real numbers ci > 0
such that .Ei / < 1 for all i and
1
X
f D ci E :
i
iD1
Thus
1
X Z
ci .Ei / D f d < 1:
iD1 X
Show that X .v u/ d < ", v is lower semi-continuous (i.e., v 1 ..t; 1// is open
R
for all t 2 R), and u is upper semi-continuous (i.e., u 1 .. 1; t// is open for all
t 2 R).
Exercise 6.23. Fix two real numbers a < b and prove the following.
(i) If f W Œa; b ! R is everywhere differentiable, then f 0 W Œa; b ! R is Borel
measurable.
Rb
(ii) If f W Œa; b ! R is everywhere differentiable and a jf 0 .t/j dt < 1, then f
is absolutely continuous.
Hint. Fix a constant " > 0. By the Vitali–Carathéodory theorem in Exer-
cise 6.22, there is a lower semi-continuous function gW Œa; b ! R such that
Z b Z b
0
g>f ; g.t/ dt < f 0 .t/ dt C ":
a a
246 6. Differentiation
F W Œa; b ! R
by
Z x
F .x/ WD g.t/ dt f .x/ C f .a/ C .x a/
a
for a x b. Consider a point a x < b. Since g.x/ > f 0 .x/ and g is lower
semi-continuous, find a number ıx > 0 such that
f .t/ f .x/
g.t/ > f 0 .x/; < f 0 .x/ C for x < t < x C ıx :
t x
Deduce that
F .t/ > F .x/ for x < t < x C ıx :
Since F .a/ D 0, there exists a maximal element x 2 Œa; b such that F .x/ D 0.
If x < b, it follows from the previous discussion that F .t/ > 0 for x < t b.
In either case F .b/ 0 and hence
Z b Z b
f .b/ f .a/ g.t/ dt C .b a/ < f 0 .t/ dt C " C .b a/:
a a
Since this holds for all > 0 and all " > 0, it follows that
Z b
f .b/ f .a/ f 0 .t/ dt:
a
Rb
Replace f by f to obtain the relation f .b/ f .a/ D a f 0 .t/ dt. Now deduce
that Z x
f .x/ f .a/ D f 0 .t/ dt
a
Example 6.24. (i) The Cantor function is the unique monotone function
f W Œ0; 1 ! Œ0; 1
that satisfies
1 1
X ai X ai
f 2 D
3i 2i
iD1 i D1
6.5. Exercises 247
for all sequences ai 2 ¹0; 1º. It is continuous and nonconstant, and its derivative
exists and vanishes on the complement of the standard Cantor set
1 n n
\ [ h X ai X ai 1i
C WD 2 ; 2 C :
nD1
3i 3i 3n
ai 2¹0;1º iD1 iD1
This Cantor set has Lebesgue measure zero. Hence f is almost everywhere differ-
entiable and its derivative is integrable. However, f is not equal to the integral of
its derivative, and therefore is not absolutely continuous.
(ii) The following construction was explained to me by Theo Buehler. Define
the homeomorphisms
gW Œ0; 1 ! Œ0; 2 and hW Œ0; 2 ! Œ0; 1
by
1
g.x/ WD f .x/ C x; h WD g :
The image g.Œ0; 1nC / is a countable union of disjoint open intervals of total length
one and hence has Lebesgue measure one. Thus its complement
K WD g.C / Œ0; 2
is a modified Cantor set of Lebesgue measure one. Hence, by Theorem 6.19, g is
not absolutely continuous. Moreover, by Lemma 2.15, there exists a set E K
which is not Lebesgue measurable. However, its image
F WD h.E/ Œ0; 1
under h is a subset of the Lebesgue null set C and hence is a Lebesgue null set.
Thus F is a Lebesgue measurable set and E D h 1 .F / is not Lebesgue measurable.
This shows that the function hW Œ0; 2 ! Œ0; 1 is not measurable with respect to the
Lebesgue -algebras on both domain and target (i.e., it is not Lebesgue-Lebesgue
measurable).
(iii) Let I; J R be intervals. Then it follows from Lemma 2.15 and The-
orem 6.19 that every Lebesgue-Lebesgue measurable homeomorphism hW I ! J
has an absolutely continuous inverse.
(iv) Let hW Œ0; 2 ! Œ0; 1 and F C Œ0; 1 be as in part (ii). Then the
characteristic function F W R ! R is Lebesgue measurable and hW Œ0; 2 ! R is
continuous. However, the composition F ı hW Œ0; 2 ! R is not Lebesgue measur-
able because the set .F ı h/ 1 .1/ D E is not Lebesgue measurable.
(v) By contrast, if I; J R are intervals, f W J ! R is Lebesgue measurable,
and hW I ! J is a C 1 diffeomorphism, then f ı hW I ! R is again Lebesgue
measurable by Theorem 2.17.
Chapter 7
Product measures
X Y WD ¹.x; y/ j x 2 X; y 2 Y º
Ex WD ¹y 2 Y j .x; y/ 2 Eº 2 B: (7.1)
E y WD ¹x 2 X j .x; y/ 2 Eº 2 A: (7.2)
250 7. Product measures
WD ¹E X Y j Ex 2 B for all x 2 X º:
.E c /x D ¹y 2 Y j .x; y/ … Eº D .Ex /c 2 B
Proof. Let E 2XY denote the collection of all elementary subsets and define
M 2XY as the smallest monotone class that contains E. This is well defined
because the intersection of any collection of monotone classes is again a monotone
class. Since every -algebra is a monotone class and every elementary set is an
element of A ˝ B, it follows that
M A ˝ B:
.P / WD ¹Q X Y j P n Q; Q n P; P [ Q 2 Mº
is a monotone class.
This follows immediately from the definition of monotone class.
Step 2. If P; Q X Y , then Q 2 .P / if and only if P 2 .Q/.
This follows immediately from the definition of .P / in Step 1.
Step 3. If P; Q 2 E, then P \ Q; P n Q; P [ Q 2 E.
For the intersection this follows from the fact that
and hence 1
S1
nD1 Qn 2 M because M is a monotone class. This proves
S
iD1 Pi D
Step 7 and Lemma 7.5.
252 7. Product measures
Lemma 7.6. Let .X; UX / and .Y; UY / be topological spaces, let UX Y be the prod-
uct topology on X Y (see Appendix B), and let BX , BY , BX Y be the associated
Borel -algebras. Then
BX ˝ BY BXY : (7.3)
If .X; UX / is a second countable locally compact Hausdorff space, then
BX ˝ BY D BX Y : (7.4)
V WD V .x1 / \ \ V .x` /:
Let .x0 ; y0 / 2 W . Then there exist open sets U 2 UX and V 2 UY such that
.x0 ; y0 / 2 U V W :
Since .X; UX / is a locally compact Hausdorff space, Lemma A.3 asserts that there
exists an open set U 0 X such that x0 2 U 0 U 0 U . Since the sets Ui form
a basis of the topology, there exists an integer i 2 N such that x0 2 Ui U 0 and
hence x0 2 Ux i U 0 U . Thus
Ux i ¹y0 º U V W ;
7.1. The product -algebra 253
UX Y BX ˝ BY
BX Y BX ˝ BY :
Lemma 7.7. Let .X; A/ be a measurable space such that the cardinality of X is
greater than that of 2N . Then the diagonal WD ¹.x; x/ j x 2 X º is not an element
of A ˝ A.
D 2 .¹Ei j i 2 Nº/:
The union of the sets .E0 / over all countable subsets E0 E is a -algebra that
contains E and is contained in .E/. Hence it is equal to .E/.
Step 2. Let Y be a set, let E 2Y , and let D 2 .E/. Then there are a sequence
Ei 2 E and a set I 2N such that
[\ \
DD Ei \ Eic :
I 2I i2I i2NnI
Step 3. … A ˝ A.
where \
\
EI WD .Ai Bi / \ .Ai Bi /c :
i 2I i2NnI
Thus
[
EI D AIJ BIJ ;
J NnI
where
\ \
AIJ WD Ai \ .X n Aj /
i2I j 2J
and
\ \
BIJ WD Bi \ .X n Bj / :
i2I j 2Nn.I [J /
If D , then for all I and J , we have AIJ BIJ and so AIJ BIJ is either
empty or a singleton. Thus the cardinality of D is at most the cardinality of the set
of pairs of disjoint subsets of N, which is equal to the cardinality of 2N . Since the
cardinality of the diagonal is bigger than that of 2N , it follows that … A ˝ A, as
claimed. This proves Lemma 7.7.
Theorem 7.9. Let .X; A; / and .Y; B; / be -finite measure spaces and let
Q 2 A ˝ B. Then the functions
Definition 7.10. Let .X; A; / and .Y; B; / be -finite measure spaces. The prod-
uct measure of and is the map
˝ W A ˝ B ! Œ0; 1
defined by
Z Z
. ˝ /.Q/ WD .Qx / d.x/ D .Qy / d.y/ (7.8)
X Y
By assumption
´ ´
B if x 2 A; y
A if y 2 B;
Qx D Q D
; if x … A; ; if y … B:
Thus Q 2 .
Define
and so Q 2 .
7.2. The product measure 257
S1
Step 3. If Qi 2 for i 2 N and Qi QiC1 for all i, then Q WD i D1 Qi 2 .
and .Qi /x 2 B and .Qi /y 2 A for all i , it follows from Theorem 1.28 (iv) that
Step 4. Let A 2 A and B 2 B such that .A/ < 1 and .B/ < 1. If Qi 2
for i 2 N such that A B Q1 Q2 , then Q WD 1
T
i D1 Qi 2 .
Step 5. D A ˝ B.
Since .X; A; / and .Y; B; / are -finite, there exist sequences of measurable sets
Xn 2 A and Yn 2 B such that
f WD W X Y !R
is given by
´
1 if x D y;
f .x; y/ WD
0 if x ¤ y:
7.2. The product measure 259
Hence
Z
y
. / D f .x; y/ d.x/ D 0 for 0 y 1;
X
Z
.x / D f .x; y/ d.y/ D 1 for 0 x 1;
Y
and so Z Z
y
. / d D 0 ¤ 1 D .x / d.x/:
X Y
Thus the hypothesis that .X; A; / and .Y; B; / are -finite cannot be removed in
Theorem 7.9.
Claim 1. Assume the continuum hypothesis. Then there is a set Q Œ0; 12 such
that Œ0; 1 n Qx is countable for all x and Qy is countable for all y.
f WD Q W Œ0; 12 ! R:
Then
Z
.Qy / D f .x; y/ d.x/ D 0 for 0 y 1;
X
Z
.Qx / D f .x; y/ d.y/ D 1 for 0 x 1;
Y
and hence Z Z
.Qy / d D 0 ¤ 1 D .Q/ d.x/:
X Y
The sets Qx and Qy are all measurable and the integrals are finite, but the set Q is
not A ˝ B-measurable. This shows that the hypothesis Q 2 A ˝ B in Theorem 7.9
cannot be replaced by the weaker hypothesis that the sets Qx and Qy are all mea-
surable, even when the integrals are finite. It also shows that Lemma 7.2 does not
have a converse. Namely, fx and f y are measurable for all x and y, but f is not
A ˝ B-measurable.
Then the set Qy D ¹x 2 Œ0; 1 j j.x/ j.y/º is countable for all y 2 Œ0; 1 and
the set Œ0; 1 n Qx D ¹y 2 Œ0; 1 j j.y/ 4 j.x/º is countable for all x 2 Œ0; 1.
Proof of Claim 2. By Zorn’s lemma, every set admits a well ordering. Choose any
well ordering on A WD Œ0; 1 and define
W WD B n A D ¹w 2 A j w b0 º
W X ! X; WY !Y
. / .A ˝ B/ D A ˝ B; . / . ˝ / D ˝ :
n W Bn ! Œ0; 1
Bk ˝ B` D Bn
k ˝ ` D n :
There are three versions of Fubini’s Theorem. The first concerns nonegative func-
tions that are measurable with respect to the product -algebra (Theorem 7.17),
the second concerns real-valued functions that are integrable with respect to the
product measure (Theorem 7.20), and the third concerns real valued functions that
are integrable with respect to the completion of the product measure (Theorem 7.23).
Theorem 7.17 (Fubini for positive functions). Let .X; A; /, .Y; B; / be -finite
measure spaces and let ˝ W A ˝ B ! Œ0; 1 be the product measure in Def-
R 1 be an A ˝ B-measurable function. Then
inition 7.10. Let f W X Y ! Œ0;
the function X ! RŒ0; 1W x 7! Y f .x; y/ d.y/ is A-measurable, the function
Y ! Œ0; 1W y 7! X f .x; y/ d.x/ is B-measurable, and
Z Z Z
f d. ˝ / D f .x; y/ d.y/ d.x/
X Y X Y
Z Z (7.11)
D f .x; y/ d.x/ d.y/:
Y X
Example 7.18. equation (1.20) is equivalent to equation (7.11) for the counting
measure on X D Y D N.
262 7. Product measures
by Z Z
.x/ WD fx d; .y/ WD f y d (7.12)
Y X
for x 2 X and y 2 Y . We prove in three steps that is A-measurable,
is B-measurable, and and satisfy (7.11).
Step 1. The assertion holds when f W X Y ! Œ0; 1/ is the characteristic function
of an A ˝ B-measurable set.
Let Q 2 A ˝ B and f D Q . Then fx D Qx and f y D Qy , and so
for all x 2 X and all y 2 Y . Hence it follows from Theorem 7.9 that
Z Z Z
d D d D . ˝ /.Q/ D f d. ˝ /:
X Y X
Here the third equality follows from the definition of the measure ˝. This proves
Step 1.
Step 2. The assertion holds when f W X Y ! Œ0; 1/ is an A ˝ B-measurable
step-function.
This follows immediately from Step 1 and the linearity of the integral.
Step 3. The assertion holds when f W X Y ! Œ0; 1 is A ˝ B-measurable.
By Theorem 1.26, there exists a sequence of A ˝ B-measurable step-functions
sn W X Y ! Œ0; 1/
Then
n nC1 ; n nC1 for all n 2 N
by part (i) of Theorem 1.35. Moreover, it follows from the Lebesgue monotone
convergence theorem (Theorem 1.37) that
.x/ D lim n .x/; .y/ D lim n .y/
n!1 n!1
for all x 2 X and all y 2 Y . Use the Lebesgue monotone convergence theorem
(Theorem 1.37) again as well as Step 2 to obtain
Z Z
d D lim n d
X n!1 X
Z
D lim sn d. ˝ /
n!1 X Y
Z
D f d. ˝ /
XY
Z
D lim n d
n!1 Y
Z
D d:
Y
This proves Step 3 and Theorem 7.17.
A first application of Fubini’s Theorem 7.17 is Minkowski’s inequality for a
measurable function on a product space that is p-integrable with respect to one
variable such that the resulting Lp norms define an integrable function of the other
variable.
Theorem 7.19 (Minkowski). Fix a constant 1 p < 1. Let .X; A; / and
.Y; B; / be -finite measure spaces and let
f WX Y ! Œ0; 1
be A ˝ B-measurable. Then
Z Z p 1=p Z Z 1=p
f .x; y/ d.y/ d.x/ f .x; y/p d.x/ d.y/:
X Y Y X
In the notation
fx .y/ WD f y .x/ WD f .x; y/;
Minkowski’s inequality has the form
Z 1=p Z
p
kfx kL1 ./ d.x/ kf y kLp ./ d.y/: (7.13)
X Y
264 7. Product measures
Y ! Œ0; 1W y 7! kf kLp ./ is B-measurable. Hence both sides of the in-
y
equality (7.13) are well defined. Theorem 7.17 also shows that for p D 1 equality
holds in (7.13). Hence assume 1 < p < 1 and a choose 1 < q < 1 such that
1=p C 1=q D 1. It suffices to assume
Z
c WD kf y kLp ./ d.y/ < 1:
Y
Define W X ! Œ0; 1 by
Z
.x/ WD fx d for x 2 X
Y
D c kgkLq ./ :
Here the third step follows from Hölder’s inequality in Theorem 4.1. Since .X; A; /
is semi-finite by part (ii) of Lemma 4.30, it follows from Lemma 4.34 that
kkLp ./ c:
Theorem 7.20 (Fubini for integrable functions). Let .X; A; / and .Y; B; / be
-finite measure spaces, let ˝ W A ˝ B ! Œ0; 1 be the product measure,
and let f 2 L1 . ˝ /. Define
W X ! R
defined by
8Z
< fx d if fx 2 L1 ./;
.x/ WD Y (7.14)
if fx … L1 ./;
:
0
is -integrable.
WY !R
defined by
8Z
< f y d if f y 2 L1 ./;
.y/ WD X (7.15)
y
… L ./;
1
:
0 if f
is -integrable.
Proof. We prove part (i) and the first equality in (7.16). The functions
Define
ˆ˙ W Y ! Œ0; 1
by Z
˙
ˆ .x/ WD fx˙ d for x 2 X:
Y
x 2 E () fx … L1 ./:
Define
˙ W X ! Œ0; 1/
by
´
˙
ˆ˙ .x/ if x … E;
.x/ WD for x 2 X:
0 if x 2 E;
Then it follows from (7.17) that ˙ 2 L1 ./ and
Z Z
˙
d D f ˙ d. ˝ /:
X X Y
This proves (i) and the first equality in (7.16). An analogous argument proves (ii)
and the second equality in (7.16). This proves Theorem 7.20.
7.3. Fubini’s Theorem 267
Example 7.21. Let .X; A; / D .Y; B; / be the Lebesgue measure space in the
unit interval Œ0; 1 as in Example 7.12. Let gn W Œ0; 1 ! Œ0; 1/ be a sequence of
smooth functions such that
Z 1
gn .x/ dx D 1;
0
and
n 1 n
gn .x/ D 0 for x 2 Œ0; 1 n Œ2 ;2
by
1
X
f .x; y/ WD .gn .x/ gnC1 .x//gn .y/:
nD1
The sum on the right is finite for every pair .x; y/ 2 Œ0; 12 . Then
Z
f .x; y/ dx D 0;
X
Z 1
X
f .x; y/ dy D .gn .x/ gnC1 .x// D g1 .x/;
Y nD1
and hence
Z 1 Z 1 Z 1 Z 1
f .x; y/ dx dy D 0 ¤ 1 D f .x; y/ dy dx:
0 0 0 0
Example 7.22. This example shows that the product measure is typically not com-
plete. Let .X; A; / and .Y; B; / be two complete -finite measure spaces. Sup-
pose .X; A; / admits a nonempty null set A 2 A and B ¤ 2Y . Choose B 2 2Y nB.
Then A B … A ˝ B. However, A B is contained in the ˝ -null set A Y
and so belongs to the completion .A ˝ B/ .
In the first version of Fubini’s Theorem integrability was not an issue. In the
second version integrability of fx was only guaranteed for almost all x. In the third
version the function fx may not even be measurable for all x.
268 7. Product measures
Theorem 7.23 (Fubini for the completion). Let .X; A; / and .Y; B; / be complete
-finite measure spaces, let .X Y; .A ˝ B/ ; . ˝ / / denote the completion of
the product space, and let f 2 L1 .. ˝ / /. Define
fx .y/ WD f y .x/ WD f .x; y/ for x 2 X and y 2 Y .
Then the following holds.
(i) fx 2 L1 ./ for -almost every x 2 X and the map
W X ! R
defined by 8Z
< fx d if fx 2 L1 ./;
.x/ WD Y (7.18)
if fx … L1 ./;
:
0
is -integrable.
(ii) f y 2 L1 ./ for -almost every y 2 Y and the map
WY !R
defined by 8Z
< f y d if f y 2 L1 ./;
.y/ WD X (7.19)
if f y … L1 ./;
:
0
is -integrable.
(iii) Let 2 L1 ./ and 2 L1 ./ be as in (i) and (ii). Then
Z Z Z
d D f d. ˝ / D d: (7.20)
X X Y Y
by
8Z
< fx d for x 2 X n .E [ E 0 /;
.x/ WD Y
0 for x 2 E [ E 0
:
and
WY !R
by
8Z
< f y d for y 2 Y n .F [ F 0 /;
.y/ WD X
0 for y 2 F [ F 0 :
:
Since Z
.x/ D gx d for all x 2 X n .E [ E 0 /,
Y
it follows from part (i) of Theorem 7.20 for g that 2 L1 ./. The same argument,
using part (ii) of Theorem 7.20 for g, shows that 2 L1 ./. Moreover, the three
integrals in (7.20) for f agree with the corresponding integrals for g because
Hence equation (7.20) for f follows from part (iii) of Theorem 7.20 for g. This
proves Theorem 7.23.
270 7. Product measures
f WD E Y 2 L1 .. ˝ / /:
Exercise 7.25. Continue the notation of Theorem 7.23 and suppose that the map
f W X Y ! Œ0; 1 is .A ˝ B/ -measurable. Prove that fx is B-measurable
for -almost all x 2 X, that f y is A-measurable for -almost all y 2 Y , and
that (7.11) continues to hold. Hint. The proof of Theorem 7.23 carries over verbatim
to nonnegative measurable functions.
We close this section with two remarks about the construction of product mea-
sures in the non -finite case, where the story is considerably more subtle. These
remarks are not used elsewhere in this book and can safely be ignored.
Remark 7.26. Let .X; A; / and .Y; B; / be two arbitrary measure spaces.
In [4, Chapter 251] Fremlin defines the function
W 2X Y ! Œ0; 1
by
1
ˇ A 2 A; Bn 2 B for n 2 N
²X ˇ ³
.W / WD inf .An / .Bn / ˇˇ n S1 (7.21)
and W nD1 .An Bn /
nD1
for W X Y and proves that it is an outer measure. He shows that the -algebra
C 2XY of -measurable sets contains the product -algebra A ˝ B and calls
the measure
1 WD jC W C ! Œ0; 1
the primitive product measure. By Carathéodory’s theorem (Theorem 2.4), the mea-
sure space .X Y; C; 1 / is complete. By definition,
1 .A B/ D .A/ .B/
for all A 2 A and all B 2 B. Fremlin then defines the complete locally determined
(CLD) product measure
0 W C ! Œ0; 1
7.3. Fubini’s Theorem 271
by
ˇ E 2 A; F 2 B;
² ˇ ³
0 .W / WD sup 1 .W \ .E F // ˇ
ˇ : (7.22)
.E/ < 1; .F / < 1
(See [4, Theorem 251I].) One can also prove that a measure W C ! Œ0; 1 satisfies
.E F / D .E/ .F / for all E 2 A and F 2 B with .E/ .F / < 1
if and only if 0 1 . With these definitions Fubini’s Theorem holds
for 0 whenever the factor .Y; B; / (over which the integral is performed first)
is -finite and the factor .X; A; / (over which the integral is performed second)
S
is either strictly localizable (i.e., there is a partition X D i 2I Xi into measur-
able sets with .Xi / < 1 such that a set A X is A-measurable if and only
if A \ Xi 2 A for all i 2 I and, moreover, .A/ D
P
i 2I .A \ Xi / for all
A 2 A), or is complete and locally determined (i.e., it is semi-finite and a set
A X is A-measurable if and only if A \ E 2 A for all E 2 A with .E/ < 1).
See Fremlin [4, Theorem 252B] for details.
If the measure spaces .X; A; / and .Y; B; / are both -finite, then the mea-
sures 0 and 1 agree and are equal to the completion of the product measure ˝
on A ˝ B (see [4, Proposition 251K]).
Remark 7.27. For topological spaces yet another approach to the product measure
is based on the Riesz representation theorem (Theorem 3.15). Let .X; UX / and
.Y; UY / be two locally compact Hausdorff spaces, denote by BX and BY their Borel
-algebras, and let X W BX ! Œ0; 1 and Y W BY ! Œ0; 1 be Borel measures.
Define
ƒW Cc .X Y / ! R
by
Z Z
ƒ.f / WD f .x; y/ d.y/ d.x/
X Y
Z Z (7.23)
D f .x; y/ d.x/ d.y/
Y X
for f 2 Cc .X Y /. That the two integrals agree for every continuous function with
compact support follows from Fubini’s Theorem (Theorem 7.20) for finite measure
spaces. (To see this, observe that every compact set K X Y is contained in
the product of the compact sets KX WD ¹x 2 X j .¹xº Y / \ K ¤ ;º and
272 7. Product measures
for all f 2 Cc .X Y /. It turns out that in this situation the Borel -algebra BX Y
is contained in the -algebra C 2X Y of Remark 7.26 and
0 D 0 jBXY ; 1 D 1 jBXY :
Recall from Lemma 7.6 that the product -algebra BX ˝ BY agrees with the Borel
-algebra BXY whenever one of the spaces X or Y is second countable. If they are
both second countable, then so is the product space .X Y; UXY / (Appendix B).
In this case
0 D 1 D X ˝ Y
is the product measure of Definition 7.10 and 0 D 1 W C ! Œ0; 1 is its comple-
tion. (See Theorem 3.15 and Remark 7.26.)
(See Exercise 7.16.) Thus Theorem 7.17 has the following consequence.
Theorem 7.28 (Fubini and Borel). Let k; ` 2 N and n WD kC`. Let f W Rn ! Œ0; 1
be Borel measurable. Then
are Borel measurable for all x 2 Rk and all y 2 R` . Moreover, the functions
Z
k
R ! Œ0; 1W x 7 ! f .x; y/ d m` .y/
R`
7.4. Fubini and Lebesgue 273
and
Z
R` ! Œ0; 1W y 7 ! f .x; y/ d mk .x/
Rk
Proof. The assertion follows directly from (7.24) and Theorem 7.17.
Proof. Define
By Lemma 7.6, we have Bn D Bk ˝B` Ak ˝A` . It then follows from the unique-
ness of a normalized translation-invariant Borel measure on Rn in Theorem 2.1 that
mn jBn D .mk ˝ m` /jBn . Here is a more direct proof.
274 7. Product measures
1
X 1
X
.mk ˝ m` /.U / D .mk ˝ m` /.Ei / D mn .Ei / D mn .U /:
i D1 iD1
and
the first k coordinates. This map is continuous and hence Borel measurable by
Theorem 1.20. Thus A R` D 1 .A/ and B R` D 1 .B/ are Borel sets in
Rn . Moreover, by Step 1,
mn ..B R` / n .A R` // D mn ..B n A/ R` /
D .mk ˝ m` /..B n A/ R` /
D mk .B n A/ m` .R` /
D 0:
F 2 A` H) Rk F 2 An :
W Rk ! R
defined by
8Z
< fx d m` for x 2 Rk n E;
.x/ WD R` (7.27)
0 for x 2 E;
:
is Lebesgue integrable.
W R` ! R
defined by
8Z
ˆ
< f y d mk for y 2 R` n F;
.y/ WD Rk (7.28)
ˆ
0 for y 2 F;
:
is Lebesgue integrable.
Proof. This follows directly from Theorem 7.23 and Theorem 7.29.
7.5. Convolution 277
7.5 Convolution
An application of Fubini’s Theorem is the convolution product on the space of
Lebesgue integrable functions on Euclidean space. Fix an integer n 2 N and
let .Rn ; A; m/ be the Lebesgue measure space. The convolution of two Lebesgue
integrable functions f; g 2 L1 .Rn / is defined by
Z
.f g/.x/ WD f .x y/g.y/ d m.y/ for almost all x 2 Rn :
Rn
This map is associative and endows L1 .Rn / with the structure of a Banach alge-
bra. Throughout we use the notation f g for two Lebesgue measurable func-
tions f; gW Rn ! R to mean that they agree almost everywhere with respect to the
Lebesgue measure.
f gW Rn ! R
defined by
8Z
< f .x y/g.y/ d m.y/ for x 2 Rn n E.f; g/;
.f g/.x/ WD Rn (7.31)
0 for x 2 E.f; g/.
:
The next theorem shows that the convolution is very robust in that f g is
always Borel measurable and depends only on the equivalence classes of f and g
under equality almost everywhere.
278 7. Product measures
and f .g h/ D .f g/ h on Rn n E.
fx W Rn ! R and x W Rn ! Rn
by
fx .y/ WD f .x y/ and x .y/ WD x y:
A WD ¹y 2 Rn j f .y/ ¤ f 0 .y/º
and
B WD ¹y 2 Rn j g.y/ ¤ g 0 .y/º
for all x 2 Rn . Hence the functions fx g and fx0 g 0 agree on the complement of a
Lebesgue null set for every x 2 Rn . Hence they are either both integrable or both
not integrable and when they are their integrals agree.
7.5. Convolution 279
(iii) By (ii) and Theorem 1.55, it suffices to assume that f and g are Borel mea-
surable. Now define
F; GW R2n ! R
and
W R2n ! R2n
by
F .x; y/ WD f .x y/g.y/;
G.x; y/ WD f .x/g.y/;
.x; y/ WD .x y; y/
F DGı
is Borel measurable, by Fubini’s Theorem (Theorem 7.28). Thus the set E.f; g/
where this function takes on the value 1 is a Borel set. Moreover, the functions
F ˙ WD max¹˙F; 0º
defined by
´
˙ F ˙ .x; y/ if x 2 Rn n E.f; g/;
Fz .x; y/ WD for .x; y/ 2 R2n :
0 if x 2 E.f; g/;
Since Z Z
C
.f g/.x/ D Fz .x; y/ d m.y/ Fz .x; y/ d m.y/
Rn Rn
and
Z
.f g/.x/ D fx g d m
Rn
Z
D .fx g/ ı x d m
Rn
Z
D gx f d m
Rn
D .g f /.x/
Fx W R2n ! R
by
Fx .y; z/ WD f .z/g.x y z/h.y/:
Thus Fx is the composition of the maps R2n ! R3n W .y; z/ 7! .z; x y z; y/ and
R3n ! R; .; ; / 7! .f ./; g./; h.//. Since the first map is continuous and the
second is Borel measurable, it follows that Fx is Borel measurable. We claim that
Z
x 2 E.jf j; jgj jhj/ () jFx j D 1 () x 2 E.jf j jgj; jhj/: (7.32)
R2n
D jf .z/j.jgj jhj/.x z/
for z 2 Rn n .x E.g; h//. Since E.g; h/ is a Lebesgue null set, it follows that
Z
kFx kL1 .R2n / D jf .z/j.jgj jhj/.x z/ d m.z/:
Rn
7.5. Convolution 281
The integral on the right is infinite if and only if x 2 E.jf j; jgj jhj/. This proves
the first equivalence in (7.32). The proof of the second equivalence is analogous,
with y and z interchanged.
Now let x 2 Rn n E. Then Fx 2 L1 .R2n / and x 2 Rn n E.f; g h/.
Moreover, for z 2 Rn , the function Rn ! RW y 7! Fx .y; z/ is integrable if and
only x z … E.g; h/, and in that case its integral is equal to
Z Z
Fx .y; z/ d m.y/ D f .z/ g.x y z/h.y/ d m.y/
Rn Rn
D .f .g h//.x/
The last equality holds because x … E.f; g h/. A similar argument with y and z
interchanged shows that
Z
Fx d m2n D ..f g/ h/.x/ for all x 2 Rn n E.
R2n
Theorem 7.33. Let 1 p; q; r 1 such that 1=p C 1=q D 1 C 1=r and let
f 2 Lp .Rn / and g 2 Lq .Rn /. Then m.E.f; g// D 0 and
It follows from Theorem 7.33 and part (ii) of Theorem 7.32 that the convolution
descends to a map
L1 .Rn / L1 .Rn / ! L1 .Rn /W .f; g/ 7 ! f g: (7.35)
This map is bilinear by Theorem 1.44, it is associative by part (v) of Theorem 7.32,
and satisfies kf gk1 kf k1 kgk1 by Young’s inequality in Theorem 7.33.
Hence L1 .Rn / is a Banach algebra. By part (iv) of Theorem 7.32, L1 .Rn / is
commutative, and, by Theorem 7.33 with q D 1 and r D p, it acts on Lp .Rn /.
(A Banach algebra is a Banach space .X; kk/ equipped with an associative bilinear
map X X ! XW .x; y/ 7! xy that satisfies the inequality kxyk kxk kyk for all
x; y 2 X.)
7.5. Convolution 283
Theorem 7.35. Let 1 p < 1 and 1 < q 1 such that 1=p C 1=q D 1.
Proof. (i) Assume first that f 2 Lp .Rn / and fix a constant " > 0.
By Theorem 4.15, there is a function g 2 Cc .Rn / such that kf gkp < "1=p =3.
Since g is uniformly continuous, there is a ı > 0 such that, for all 2 Rn ,
" 1=p
jj < ı H) sup jg.x C / g.x/j < :
x2Rn 3p m.supp.g//
284 7. Product measures
2"1=p
C .m.supp.g// sup jg.x C / g.x/jp /1=p
3 x2Rn
< "1=p :
This proves (i) for f 2 Lp .Rn /. To prove the result in general, choose a compact
set K Rn such that B1 .x/ K for all x 2 B and multiply f by a smooth
compactly supported cutoff function to obtain a function f 0 2 Lp .Rn / that agrees
with f on K. Then (i) holds for f 0 and hence also for f .
(ii) Assume first that f 2 Lp .Rn / and g 2 Lq .Rn / and fix a constant " > 0.
By part (i), there exists a ı > 0 such that, for all 2 Rn ,
Z " p
jj < ı H) jf .y C / f .y/jp d m.y/ <
Rn kgkq
fx .y/ WD f .x y/:
kfxC fx kp kgkq
Z 1=p
D jf .y C / f .y/jp d m.y/ kgkq
Rn
< ":
for all h 2 R with 0 < jhj < ı. Take h 2 Rn with 0 < jhj < ı. Then
ˇ ˇ
ˇ .f g/.x C hei / .f g/.x/ ˇ
ˇ .f @i g/.x/ˇ
ˇ h ˇ
ˇZ g.x C he ˇ
ˇ i y/ g.x y/ ˇ
D ˇˇ f .y/ @i g.x y/ d m.y/ˇˇ
Rn h
ˇ ˇ
ˇ g.x C hei y/ g.x y/
Z
ˇ
jf .y/j ˇˇ @i g.x y/ d m.y/ˇˇ
Rn h
< ":
Define
ı W R n ! R
by
1 x
ı .x/ WD
ın ı
for ı > 0 and x 2 Rn . Then
Z
supp.ı / Bı ; ı d m D 1
Rn
286 7. Product measures
fı WD ı f W Rn ! R
is smooth for all ı > 0. Now fix a constant " > 0. By part (i), there exists a constant
ı > 0 such that, for all y 2 Rn ,
Z
jyj < ı H) jf .x y/ f .x/jp d m.x/ < "p :
Rn
":
The method explained in the proof of part (iv) of Theorem 7.35 is called the
mollifier technique. The functions ı can be viewed as approximate Dirac delta
functions that concentrate near the origin as ı tends to zero.
for all t 0.
Proof. Inequality (7.36) was established in the proof of Lemma 6.2. We prove (7.37)
in four steps.
Step 1. For all t 0, Z
p
t f .t/ jf jp d:
X
Since t p A.t;f / jf jp , it follows that
Z Z
t p f .t/ D t p A.t;f / d jf jp d for all t 0.
X X
By Step 1, we have Z
jf jp d D 1:
X
Moreover, t p 1
f .t/ D 1 for t > 0 sufficiently small, and hence
Z 1
t p 1 f .t/ dt D 1:
0
Step 3. Assume .X; A; / is -finite and f .t/ < 1 for all t > 0. Then equa-
tion (7.37) holds.
Let B 2Œ0;1/ be the Borel -algebra and denote by mW B ! Œ0; 1 the restriction
of the Lebesgue measure to B. Let .X Œ0; 1/; A ˝ B; ˝ m/ be the product
measure space of Definition 7.10. We prove that
To see this, assume first that f is an A-measurable step-function. Then there exist
finitely many pairwise disjoint measurable sets A1 ; : : : ; A` 2 A and positive real
numbers ˛1 ; : : : ; ˛` such that
`
X
jf j D ˛i A :
i
iD1
In this case Q.f / D `iD1 Ai Œ0; ˛i / 2 A ˝ B. Now consider the general case.
S
Now define
hW X Œ0; 1/ ! Œ0; 1/
by
h.x; t/ WD pt p 1
:
Step 4. Assume f .t/ < 1 for all t > 0. Then (7.37) holds.
Define
X0 WD ¹x 2 X j f .x/ ¤ 0º;
A0 WD ¹A 2 A j A X0 º;
and
0 WD jA0 :
Then the measure space .X0 ; A0 ; 0 / is -finite because Xn WD A.1=n; f / is a
sequence of An -measurable sets such that 0 .Xn / D f .1=n/ < 1 for all n and
X0 D 1 nD1 Xn . Moreover, f0 WD f jX0 W X0 ! R is A0 -measurable and f D f0 .
S
c1 p Z
D jf jp d;
p 1 X
7.6. Marcinkiewicz interpolation 291
and
Z 1
tp q 1
kg t kqq dt
0
Z 1 Z 1
p q 1 q 1
D t qs g t .s/ ds dt
0 0
Z 1 Z ct
p q 1 q 1
D t qs .f .s/ f .ct// ds dt
0 0
Z 1 Z 1 Z 1
p q 1 q 1 q
Dq t dt s f .s/ ds c tp 1
f .ct/ dt
0 s=c 0
1 1
sp 1cq p
Z Z
Dq f .s/ ds cq p
tp 1
f .t/ dt
0 q p 0
cq p p 1 p 1
Z
D t f .t/ dt
q p 0
cq p
Z
D jf jp d:
q p X
Moreover, f D f t C g t for all t 0. Hence, by Lemma 7.36 and (7.39),
Here the last equality follows with the choice of c WD .2c1 /1=.q 1/
=.2cq /q=.q 1/
.
This proves Theorem 7.37.
292 7. Product measures
and, for i D 1; : : : ; n, denote the partial derivative with respect to the i th coordinate
by @i D @=@xi . Denote the open ball of radius r > 0 centered at the origin by
Br WD ¹x 2 Rn j jxj < rº. Call a function uW Rn ! R smooth if all its partial
derivatives exist and are continuous. Denote by C01 .Rn / the space of compactly
supported smooth functions on Rn .
!n =n WD m.B1 /
denotes the Lebesgue measure of the unit ball in Rn . The first and second partial
derivatives
Ki WD @i K and Kij WD @i @j K
of the fundamental solution are given by
Exercise 7.39. Prove that K D 0. Prove that K and Ki are locally integrable,
while Kij is not Lebesgue integrable over any neighborhood of the origin.
Hint. Use Fubini’s Theorem in polar coordinates (Exercise 7.47).
f D K f: (7.45)
u D f; @i u D Ki f for i D 1; : : : ; n: (7.47)
Then @ D @Br [ @B" ./ and the functions v and @v=@ vanish on @Br . Moreover,
K 0. Hence Green’s formula (7.48) asserts that
Z Z @K @f
K f d m D f K d: (7.49)
Rn nB" ./ @B" ./ @ @
Here the reversal of sign arises from the fact that the outward unit normal vector on
@B" ./ is inward pointing with respect to . Moreover,
1
.x/ D jx j .x / for x 2 @B" ./,
so
@K =@.x/ D !n 1 "1 n
u D .K f / D K f D f:
7.7. The Calderón–Zygmund inequality 295
To prove the second equation in (7.47), fix an index i 2 ¹1; : : : ; nº and a point
2 Rn . Then the divergence theorem on
WD Br n Bx" ./
asserts that
Z
.Ki . x/f .x/ K. x/@i f .x// d m.x/
Rn nB" ./
Z
D ..@i K /f C K @i f / d m
Rn nB" ./
Z
D @i .K f / d m
Rn nB" ./
Z
D i K f d
@B" ./
xi i
Z
D ."/ f .x/ d.x/:
@B" ./ "
Proof. See page 304. The proof is based on the exposition in Gilbarg and
Trudinger [5].
296 7. Product measures
By Theorem 7.35, the space C01 .Rn / is dense in L2 .Rn /. Thus Lemma 7.44
shows that the linear operator f 7! @k .Kj f / extends uniquely to a bounded
linear operator from L2 .Rn / to L2 .Rn /. The heart of the proof of the Calderón–
Zygmund inequality is the following delicate argument which shows that this oper-
ator also extends to a continuous linear operator from the Banach space L1 .Rn / to
the topological vector space L1;1 .Rn / of weakly integrable functions introduced
in Section 6.1. This argument occupies the next six pages. Recall the definition
where
f .t/ WD m.A.t; f //; A.t; f / WD ¹x 2 Rn j jf .x/j > tº:
(See equation (6.1).)
Lemma 7.45. Fix an integer n 2. Then there is a constant c D c.n/ > 0 such
that
k@k .Kj f /k1;1 ckf k1 (7.54)
for all f 2 C01 .Rn / and all indices j; k D 1; : : : ; n.
Proof. Fix two integers j; k 2 ¹1; : : : ; nº and let T W L2 .Rn / ! L2 .Rn / be the
unique bounded linear operator that satisfies
Tf D @k .Kj f / (7.55)
for f 2 C01 .Rn /. This operator is well defined by Lemma 7.44. We prove in three
steps that there is a constant c D c.n/ > 0 such that kTf k1;1 c kf k1 for all
f 2 L1 .Rn / \ L2 .Rn /. Throughout we abuse notation and use the same letter f
to denote a function in L2 .Rn / and its equivalence class in L2 .Rn /.
then
1
T h .t/ c m.B/ C khk1 (7.57)
t
for all t > 0.
298 7. Product measures
For i 2 N define
hi W Rn ! R
by
´
h.x/ if x 2 Qi ;
hi .x/ WD
0 if x … Qi :
Denote by qi 2 Qi the center of the cube Qi and by 2ri > 0 its side length. Then
p
jx qi j nri for all x 2 Qi . Fix an element x 2 Rn n Qi . Then Kj is smooth
on x Qi and so Theorem 7.35 asserts that
This identity is more delicate than it looks at first glance. To see this, note that
equation (7.55) only holds for compactly supported smooth functions, but is not
meaningful for all L2 functions f because Kj f may not be differentiable.
The function hi is not smooth, so care must be taken. Since x … Qi D supp.hi /
one can approximate hi in L2 .Rn / by a sequence of compactly supported smooth
functions that vanish near x (by using the mollifier method in the proof of Theo-
rem 7.35). For the approximating sequence part (iii) of Theorem 7.35 asserts that
the partial derivative with respect to the kth variable of the convolution with Kj is
equal to the convolution with @k Kj near x. Now the first equality in (7.58) follows
by taking the limit. The second equality follows from (7.56). It follows from (7.58)
that
Z
j.T hi /.x/j j@k Kj .x y/ @k Kj .x qi /jjhi .y/j d m.y/
Qi
1
c1 ri sup khi k1
y2Qi jx yjnC1
c1 ri
D khi k1 :
d.x; Qi /nC1
7.7. The Calderón–Zygmund inequality 299
n.n C 3/
c1 D c1 .n/ WD max sup jyjnC1 jr@k Kj .y/j :
j;k y2Rn n¹0º !n
Here the last inequality follows by differentiating equation (7.43).
Now define
p
Pi WD ¹x 2 Rn j jx qi j < 2 nri º Qi :
Then p
d.x; Qi / jx qi j nri for all x 2 Rn n Pi .
Hence
1
Z Z
jT hi j d m c1 ri p d m.x/ khi k1
Rn nPi Rn nPi .jx qi j nri /nC1
1
Z
D c1 ri p p d m.y/ khi k1
jyj>2 nri .jyj nri /nC1
1
!n s n 1 ds
Z
D c1 ri p p khi k1
2 nri .s nri /nC1
Z 1 p
.s C nri /n 1 ds
D c1 !n ri p khi k1
nri s nC1
Z 1
n 1 ds
c1 2 !n ri p 2
khi k1
nri s
D c2 khi k1 :
p
Here c2 D c2 .n/ WD c1 .n/2n 1 !n n 2n 1 n3=2 .n C 3/. The third step in
the above computation follows from Fubini’s Theorem in polar coordinates (Exer-
cise 7.47). Thus we have proved that
Z
jT hi j d m c2 khi k1 for all i 2 N: (7.59)
Rn nPi
Recall that T h and T hi are only equivalence classes in L2 .Rn /. Choose square
integrable functions on Rn representing these equivalence classes and denote them
by the same letters T h; T hi 2 L2 .Rn /. We prove that there is a Lebesgue null set
E Rn such that
1
X
jT h.x/j jT hi .x/j for all x 2 Rn n E: (7.60)
iD1
300 7. Product measures
To see this, note that the sequence `iD1 hi converges to h in L2 .Rn / as ` tends to
P
P`
infinity. So the sequence iD1 T hi converges to T h in L2 .Rn / by Lemma 7.44.
By Corollary 4.10, a subsequence converges almost everywhere. Hence there exist
a Lebesgue null set E Rn and a sequence of integers 0 < `1 < `2 < `3 <
such that the sequence `i D1
P
T hi .x/ converges to T h.x/ as tends to infinity for
n
P` P1 n
all x 2 R n E. Since j iD1 T hi .x/j iD1 jT hi .x/j for all x 2 R , this
proves (7.60).
Now put
1
[
A WD Pi :
i D1
Then it follows from (7.59), (7.60), and Theorem 1.38 that
Z Z 1
X
jT hj d m jT hi j d m
Rn nA Rn nA iD1
1 Z
X
D jT hi j d m
i D1 Rn nA
1 Z
X
jT hi j d m
i D1 Rn nPi
1
X
c2 khi k1
i D1
D c2 khk1 :
Moreover,
1
X 1
X
m.A/ m.Pi / D c3 m.Qi / D c3 m.B/;
i D1 iD1
where
m.B2pn /
c3 D c3 .n/ WD D m.Bpn / D !n nn=2 1
:
m.Œ 1; 1n /
Hence
tT h .t/ tm.A/ C tm.¹x 2 Rn n A j jT h.x/j > tº/
Z
tm.A/ C jT hj d m
Rn nA
c3 tm.B/ C c2 khk1
c4 .tm.B/ C khk1 /
7.7. The Calderón–Zygmund inequality 301
Let
Q WD ¹Q.; `/ j 2 Zn ; ` 2 Zº
and define the subset Q0 Q by
ˇ tm.Q/ < Q jf j d m and, for all Q0 2 Q;
² ˇ R ³
Q0 WD Q 2 Q ˇˇ :
Q ¨ Q0 H) Q0 jf j d m tm.Q0 /
R
Define
g; hW Rn ! R
by Z
f dm
Qi
X
g WD f Rn nB C Q ; h WD f g:
m.Qi / i
i
Then
kgk1 kf k1 ; khk1 2 kf k1 :
Further, h vanishes on Rn n B and Qi h d m D 0 for all i. Hence it follows from
R
Step 1 that
1 3c
T h .t/ c m.B/ C khk1 kf k1 : (7.65)
t t
Moreover, it follows from Step 2 that jg.x/j t for almost every x 2 Rn n B and
jg.x/j 2n t for every x 2 int.Qi /. Thus jgj 2n t almost everywhere. Hence it
follows from Lemma 7.36 that
1 2n 2n
Z Z
2
T g .t/ 2 jgj d m jgj d m kf k1 : (7.66)
t Rn t Rn t
Now combine (7.65) and (7.66) with the inequality (7.36) in Lemma 7.36 to obtain
the estimate
2nC1 C 6c
Tf .2t/ T g .t/ C T h .t/ kf k1 :
2t
Here the splitting f D g C h depends on t , but the constant c does not. Multiply
the inequality by 2t and take the supremum over all t to obtain (7.64). This proves
Step 3 and Lemma 7.45.
7.7. The Calderón–Zygmund inequality 303
Then kTij f k1;1 c1 .n/ kf k1 for all f 2 C01 .Rn / and all i; j by Lemma 7.45.
Since C01 .Rn / is dense in L2 .Rn / \ L1 .Rn / by Theorem 7.35, it follows that
kTij f k1;1 c1 .n/ kf k1 for all f 2 L2 .Rn / \ L1 .Rn /. Hence Theorem 7.37
(with q D 2) asserts that (7.67) holds with
p 1=p
c D c.n; p/ WD 2 c1 .n/2=p 1
:
.2 p/.p 1/
Third, suppose 2 < p < 1 and choose 1 < q < 2 such that 1=p C 1=q D 1. Then
it follows from Theorem 7.41, integration by parts, Hölder’s inequality, and from
what we have just proved that, for all f; g 2 C01 .Rn /,
Z Z
.@i .Kj f //g d m D .@i @j f /g d m
Rn Rn
Z
D f .@i @j g/ d m
Rn
Z
D f .@i .Kj g// d m
Rn
Since C01 .Rn / is dense in Lq .Rn / by Theorem 7.35, and the Lebesgue measure is
semi-finite, it follows from Lemma 4.34 that k@i .Kj f /kp c.n; q/ kf kp for all
f 2 C01 .Rn /. This proves Theorem 7.46.
304 7. Product measures
Proof of Theorem 7.43. Fix both an integer n 2 and a number 1 < p < 1.
Let c D c.n; p/ be the constant of Theorem 7.46 and let u 2 C01 .Rn /. Then
@j u D @j .K u/ D Kj u
7.8 Exercises
Exercise 7.47 (Lebesgue measure on the sphere). For n 2 N, let .Rn ; An ; mn / the
Lebesgue measure space, denote the open unit ball by B n WD ¹x 2 Rn j jnj < 1º,
and the unit sphere by
Sn 1
WD @B n D ¹x 2 Rn j jxj D 1º:
For A S n 1
define
p
A˙ WD ¹x 2 B n 1
j .x; ˙ 1 jxj2 / 2 Aº:
AS WD ¹A S n 1
j AC ; A 2 An 1º
W AS ! Œ0; 1
defined by
1 1
Z Z
.A/ WD p d mn 1 .x/ C p d mn 1 .x/
AC 1 jxj2 A 1 jxj2
!n WD .S n 1
/ < 1:
7.8. Exercises 305
Then
g 2 L1 ./; h 2 L1 .Œ0; 1//;
and Z Z Z
f d mn D g d D h d m1 : (7.69)
Rn Sn 1 Œ0;1/
Hint. Assume first that i D n. Use the fundamental theorem of calculus and Fubini’s
Theorem (Theorem 7.30) for Rn D Rn 1 R.
f W R2 ! R
by
8
ˆ
ˆ 1 if z > 0;
sign.xy/ <
f .x; y/ WD 2 ; sign.z/ WD 0 if z D 0; for .x; y/ ¤ 0,
x C y2 ˆ
ˆ
1 if z < 0;
:
and by
f .0; 0/ WD 0:
Prove that fx ; f y W R ! R
R are Lebesgue integrable for allR x; y 2 R. Prove that the
functions R ! RW x 7! R fx d m1 and R ! RW y 7! R f y d m1 are Lebesgue
integrable and
Z Z Z Z
f .x; y/ d m1 .x/ d m1 .y/ D f .x; y/ d m1 .y/ d m1 .x/:
R R R R
Exercise 7.51. Let .X; A; / and .Y; B; / be two -finite measure spaces and let
f 2 L1 ./ and g 2 L1 ./. Define
hW X Y !R
by
h.x; y/ WD f .x/g.y/; for x 2 X and y 2 Y:
Prove that h 2 L1 . ˝ / and
Z Z Z
h d. ˝ / D f d g d:
X Y X Y
Exercise 7.52. Let .X; A; / and .Y; B; / be two -finite measure spaces and let
W A ˝ B ! R be any measure such that .A B/ D .A/.B/ for all A 2 A
and all B 2 B. Prove that D ˝ .
Exercise 7.53. Define
W R ! R
by
´
1 cos.x/ for 0 x 2;
.x/ WD
0 otherwise:
Define the functions
f; g; hW R ! R
by Z x
0
f .x/ WD 1; g.x/ WD .x/; h.x/ WD .t/ dt
1
for x 2 R. Prove that .f g/ h D 0 and f .g h/ > 0. Thus the convolution
need not be associative on nonintegrable functions. Compare this with part (v) of
Theorem 7.32. Prove that
while
E.f g; h/ D E.f; g h/ D ;:
Exercise 7.54. Let .R; A; m/ be the Lebesgue measure space, let B A be the
Borel -algebra, and denote by M the Banach space of all signed Borel measures
W B ! Œ0; 1/ with the norm kk WD jj.R/. (See Exercise 5.34.) The convolu-
tion of two signed measures ; 2 M is the map
W B ! R
308 7. Product measures
defined by
for B 2 B, where
. ˝ / WD C ˝ C C ˝ C ˝ ˝ C:
k k kk kk :
ıD
for all 2 M.
(iii) The convolution product on M is commutative, associative, and distributive.
Thus M is a commutative Banach algebra with unit.
(iv) If f 2 L1 .R/ and
f W B ! R
is defined by Z
f .B/ WD f d m for B 2 B;
B
then f 2 M and kf k D kf k1 .
(v) If f; g 2 L1 .R/, then
f g D f g :
The purpose of this last chapter is to prove the existence and uniqueness of a nor-
malized invariant Radon measure on any compact Hausdorff group. In the case of a
locally compact Hausdorff group the theorem asserts the existence of a left-invariant
Radon measure that is unique up to a scaling factor. An example is the Lebesgue
measure on Rn . A useful exposition is the paper by Gert K. Pedersen [16], which
also discusses the original references.
G G ! GW .x; y/ 7 ! xy (8.1)
the unit 1 2 G, and the inverse map
1
G ! GW x 7 ! x : (8.2)
U 2G
such that the group multiplication (8.1) and the inverse map (8.2) are continuous.
Here the continuity of the group multiplication (8.1) is understood with respect to
the product topology on G G (see Appendix B). A locally compact Hausdorff
group is a topological group .G; U/ such that the topology is locally compact and
Hausdorff (see page 95).
Example 8.1. Let G be any group and define U WD ¹;; Gº. Then .G; U/ is a compact
topological group, but is not Hausdorff unless G D ¹1º.
Example 8.2 (discrete groups). Let G be any group. Then the pair .G; U/ with
the discrete topology U WD 2G is a locally compact Hausdorff group, called a
discrete group. Examples of discrete groups (where the discrete topology appears
naturally) are the additive group Zn , the multiplicative group SL.n; Z/ of integer
310 8. The Haar measure
n n-matrices with determinant one, the mapping class group of isotopy classes of
diffeomorphisms of any manifold, and every finite group.
Example 8.3 (Lie groups). Let G GL.n; C/ be any subgroup of the general linear
group of invertible complex n n-matrices that is closed as a subset of GL.n; C/
with respect to the relative topology, i.e., GL.n; C/ n G is an open set in Cnn .
Let U 2G be the relative topology on G, i.e., U G is open if and only if
there is an open subset V Cnn such that U D G \ V . Then .G; U/ is a
locally compact Hausdorff group. In fact, it is a basic result in the theory of Lie
groups that every closed subgroup of GL.n; C/ is a smooth submanifold of Cnn
and hence is a Lie group. Specific examples of Lie groups are the general linear
groups GL.n; R/ and GL.n; C/, the special linear groups SL.n; R/ and SL.n; C/
of real and complex n n-matrices with determinant one, the orthogonal
group O.n/ of matrices x 2 Rnn such that x T x D 1, the special orthogonal group
SO.n/ WD O.n/ \ SL.n; R/, the unitary group U.n/ of matrices x 2 Cnn such
that x x D 1, the special unitary group SU.n/ WD U.n/ \ SL.n; C/, the group
Sp.1/ of the unit quaternions, the unit circle S 1 D U.1/ in the complex plane, the
torus Tn WD S 1 S 1 (n times), or, for any multi-linear form W .Cn /k ! C,
the group of all matrices x 2 GL.n; C/ that preserve . The additive groups Rn
and Cn are also Lie groups. Lie groups form an important class of locally compact
Hausdorff groups and play a central role in differential geometry.
Example 8.4. If .V; kk/ is a normed vector space (Example 1.11), then the additive
group V is a Hausdorff topological group. It is locally compact if and only if V is
finite-dimensional.
Example 8.5. The rational numbers Q with the additive structure form a Hausdorff
topological group with the relative topology as a subset of R. It is totally discon-
nected (every connected component is a single point), but does not have the discrete
topology. It is not locally compact.
N0 WD N [ ¹0º
k
dp .x; y/ WD jx yjp WD inf¹p j k 2 N0 ; x y 2 p k Zº: (8.3)
8.1. Topological groups 311
This set is a commutative ring with unit. Addition and multiplication are defined
term by term, i.e.,
for x D .xk /k2N0 2 Zp and y D .yk /k2N0 2 Zp . The ring of p-adic integers is a
compact metric space with
k
dp .x; y/ WD inf¹p j k 2 N0 ; xk D yk º: (8.4)
This is an isometric embedding with respect to the distance functions (8.3) and (8.4).
The additive p-adic integers form an uncountable compact Hausdorff group
(with the topology of a Cantor set) that is not a Lie group.
dp W Q Q ! Œ0; 1/
by
dp .x; y/ WD jx yjp
° ˇ a ± (8.5)
WD inf p k ˇ k 2 Z; x y D p k ; a 2 Z; b 2 N n p N :
ˇ
b
312 8. The Haar measure
Example 8.8 (infinite products). Let I be any index set and, for i 2 I , let Gi be a
compact Hausdorff group. Then the product
Y
G WD Gi
i2I
F
is a compact Hausdorff group. Its elements are maps I ! i 2I Gi W i 7! xi
such that xi 2 Gi for all i 2 I . Write such a map as x D .xi /i2I . The product
topology on G is defined as the smallest topology such that the obvious projections
Q
i W G ! Gi are continuous. Thus the (infinite) products U D i2I Ui of open
sets Ui Gi , such that Ui D Gi for all, but finitely many i , form a basis for the
topology of G. (See Appendix B for #I D 2.) The product topology is obviously
Hausdorff and Tychonoff’s theorem asserts that it is compact (see Munkres [14]).
An uncountable product of nontrivial groups Gi is not first countable.
Example 8.9. Let .X; kk/ be a Banach algebra with a unit 1 and the product
X X ! XW .x; y/ 7! xy. (See page 282.) Then the group of invertible ele-
ments G WD ¹x 2 X j there exists y 2 X such that xy D yx D 1º is a Hausdorff
topological group. Examples include the group of nonzero quaternions, the general
linear group of a finite-dimensional vector space, the group of bijective bounded
linear operators on a Banach space, and the multiplicative group of nowhere van-
ishing real-valued continuous functions on a compact topological space. In general
G is not locally compact.
Lemma 8.10. Let f 2 Cc .G/ and fix a constant " > 0. Then there exists an open
neighborhood U of 1 such that, for all x; y 2 G,
1
x y 2 U H) jf .x/ f .y/j < ": (8.6)
8.2. Haar measures 313
K WD ¹ab 1
j a 2 supp.f /; b 2 Ux0 º:
This set is compact because the maps (8.1) and (8.2) are continuous. Also,
1
x … K; x y 2 U0 H) f .x/ D f .y/ D 0 (8.7)
`
\
U WD U0 \ U.xi /
iD1
Lx ; Rx W G ! G
by
Lx .a/ WD xa; Rx .a/ WD ax for x 2 G: (8.10)
They satisfy
Theorem 8.12 (Haar). Let G be a locally compact Hausdorff group. Then the fol-
lowing holds.
(i) G admits a left Haar measure , unique up to a positive factor. Every such
measure satisfies .U / > 0 for every nonempty open set U G.
(ii) G admits a right Haar measure , unique up to a positive factor. Every such
measure satisfies .U / > 0 for every nonempty open set U G.
(iii) Assume G is compact. Then G admits a unique invariant Haar measure such
that .G/ D 1. This measure satisfies .B 1 / D .B/ for all B 2 B and
.U / > 0 for every nonempty open set U G.
Examples of Haar measures are the restriction to the Borel -algebra of the
Lebesgue measure on Rn (where the group structure is additive), the restriction to
the Borel -algebra of the measure on S 1 D U.1/ or on S 3 D Sp.1/ in Exer-
cise 7.47, the measure dt=t on the multiplicative group of the positive real numbers,
and the counting measure on any discrete group. The proof of Theorem 8.12 rests
on the Riesz representation theorem (Theorem 3.15) and the following result about
positive linear functionals.
Definition 8.13. Let G be a locally compact Hausdorff group. A linear functional
ƒW Cc .G/ ! R is called left-invariant if
ƒ.f ı Lx / D ƒ.f / (8.12)
for all f 2 Cc .G/ and all x 2 G. It is called right-invariant
ƒ.f ı Rx / D ƒ.f / (8.13)
for all f 2 Cc .G/ and all x 2 G. It is called invariant if it is both left- and right-
invariant. It is called a left Haar integral if it is left-invariant, positive, and nonzero.
It is called a right Haar integral if it is right-invariant, positive, and nonzero.
Theorem 8.14 (Haar). Every locally compact Hausdorff group G admits a left Haar
integral, unique up to a positive factor. Moreover, if ƒW Cc .G/ ! R is a left Haar
integral and f 2 Cc .G/ is a nonnegative function that does not vanish identically,
then ƒ.f / > 0.
Proof. See page 318.
The proof of Theorem 8.14 given below follows the notes of Pedersen [16],
which are based on a proof by Weil. Our exposition benefits from elegant sim-
plifications due to Urs Lang [11]. In preparation for the proof it is convenient to
introduce some notation. Let
CcC .G/ WD ¹f 2 Cc .G/ j f 0; f 6 0º (8.14)
be the space of nonnegative continuous functions with compact support that do not
vanish identically. A function ƒW CcC .G/ ! Œ0; 1/ is called
• additive if ƒ.f C g/ D ƒ.f / C ƒ.g/ for all f; g 2 CcC .G/,
• subadditive if ƒ.f C g/ ƒ.f / C ƒ.g/ for all f; g 2 CcC .G/,
• homogeneous if ƒ.cg/ D cƒ.f / for all f 2 CcC .G/ and all c > 0,
• monotone if f g implies ƒ.f / ƒ.g/ for all f; g 2 CcC .G/,
• left-invariant if ƒ.f ı Lx / D ƒ.f / for all f 2 CcC .G/ and all x 2 G.
316 8. The Haar measure
If WD j`D1 ˇj ıyj 2 P is any other such measure, define the convolution product
P
of and by
Xk X `
WD ˛i ˇj ıxi yj :
i D1 j D1
for f 2 Cc .G/ and a 2 G. The next two lemmas establish some basic properties of
the operators L and R . Denote by
kf k1 WD sup jf .x/j
x2G
Lemma 8.17. Let f; g 2 CcC .G/. Then there exists a 2 P such that
f L g:
This set is an open neighborhood of x. Since f has compact support, there exist
finitely many points x1 ; : : : ; xk 2 G such that supp.f / Ux1 [ [ Uxk . Define
k
X f .xi / C 1
WD ıyx 1 :
g.y/ i
i D1
Then
k
X f .xi / C 1
f .a/ g.yxi 1 a/ D .L g/.a/
g.y/
iD1
Proof of Theorem 8.14. The proof has five steps. Step 1 is the main construction
of a subadditive functional
(ii) For every g 2 CcC .G/, the functional Mg W CcC .G/ ! .0; 1/ is subadditive,
homogeneous, monotone, and left-invariant, and hence is an element of L.
We prove part (ii). Monotonicity follows directly from the definition. Homogeneity
follows from the identities Lc g D cL g and kck D c kk. To prove left-
invariance, observe that
Thus
M.f C f 0 I g/ < M.f I g/ C M.f 0 I g/ C " for all " > 0.
k
X
ƒ.L f / ƒ.˛i .f ı Lxi //
i D1
k
X
D ˛i ƒ.f ı Lxi /
i D1
k
X
D ˛i ƒ.f /
i D1
D kk ƒ.f /:
Here the first step follows from subadditivity, the second step follows from homo-
geneity, the third step follows from left-invariance, and the last step follows from
the definition of kk. This proves (8.21). Now let f; g 2 CcC .G/. By Lemma 8.17,
there is a 2 P such that f L g. Since ƒ is monotone this implies
ƒ.f / ƒ.L g/ kk ƒ.g/ by (8.21). Now take the infimum over all 2 P
such that f L g to obtain ƒ.f / M.f I g/ƒ.g/.
We prove parts (i) and (iv). Since the map CcC .G/ ! .0; 1/W f 7! kf k1 ; is
an element of L, it follows from part (iii) that
and hence M.f I g/ > 0 for all f; g 2 CcC .G/. Next observe that M.f I f / 1
by (8.22) and M.f I f / 1 because f D Lı1 f . This proves Step 1.
Step 2. Let f; f 0 2 CcC .G/ and let " > 0. Then there is an open neighborhood
U G of 1 such that every g 2 CcC .G/ with supp.g/ U satisfies
Define
h WD f C f 0 C ıkf C f 0 k1 :
` X f .x 1 / `
L g.a/ X f .a/
i
f .a/ f .a/ D ˛i g.xi a/ ˛i C ı g.xi a/:
h.a/ h.a/
iD1
h.xi 1 / iD1
Thus f L g, where
` f .x 1
X
i/
WD ˛i C ı ıx i :
iD1
h.xi 1 /
This implies
` f .x 1
X /i
Mg .f / ˛i C ı :
i D1
h.xi 1 /
` 1
0
X f .x
i / C f 0 .xi 1 /
Mg .f / C Mg .f / ˛i C 2ı kk .1 C 2ı/:
iD1
h.xi 1 /
Mg .f / C Mg .f 0 / .1 C 2ı/Mg .h/
.1 C 2ı/.Mg .f C f 0 / C ıkf C f 0 k1 Mg .//
.1 C 2ı C 2ıkf C f 0 k1 M.I f C f 0 //Mg .f C f 0 /
.1 C "/Mg .f C f 0 /:
Here we have used the inequalities 1 C 2ı 2 and (8.24). This proves Step 2.
322 8. The Haar measure
Step 3. Let f 2 CcC .G/ and let " > 0. Then there is an open neighborhood U G
of 1 with the following significance. For every g 2 CcC .G/ such that
1
supp.g/ U; g.x/ D g.x / for all x 2 G; (8.25)
there exists an open neighborhood W G of 1 such that every h 2 CcC .G/ with
supp.h/ W satisfies the inequality
M.f I g/Mh .g/ .1 C "/Mh .f /: (8.26)
This inequality continues to hold with Mh replaced by any left-invariant positive
linear functional ƒW Cc .G/ ! R.
By Urysohn’s lemma (Theorem A.1) there is a function 2 CcC .G/ such that
.x/ D 1 for all x 2 K WD supp.f /. Choose "0 and "1 such that
1 C "0 "0
0 < "0 < 1; 1 C "; "1 WD : (8.27)
1 "0 2M.I f /
By Lemma 8.10, there exists an open neighborhood U G of 1 such that
1
x y 2 U H) jf .x/ f .y/j < "1 (8.28)
for all x; y 2 G. We prove that the assertion of Step 3 holds with this neighbor-
hood U . Fix a function g 2 CcC .G/ that satisfies (8.25). Put
"1
"2 WD : (8.29)
2M.f I g/
Use Lemma 8.10 to find an open neighborhood V G of 1 such that
1
xy 2 V H) jg.x/ g.y/j < "2 (8.30)
for all x; y 2 G. Then the sets xV for x 2 K form an open cover of K. Hence
S`
there exist finitely many points x1 ; : : : ; x` 2 K such that K i D1 xi V .
By Theorem A.4, there exist functions 1 ; : : : ; ` 2 CcC .G/ such that
`
X
0 i 1; supp.i / xi V; i jK 1: (8.31)
i D1
It follows from (8.25) and (8.28) that f .x/g.y 1 x/ f .y/g.y 1 x/ "1 g.y 1 x/
for all x; y 2 G. Since g.y 1 x/ D g.x 1 y/ D .gıLx 1 /.y/ by (8.25), this implies
f .x/g ı Lx 1 Fx C "1 g ı Lx 1 : Hence, for all x 2 G and all h 2 CcC .G/,
Now fix a function h 2 CcC .G/ with supp.h/ W . By (8.30) and (8.31),
1
i .y/Fx .y/ D i .y/f .y/g.y x/ i .y/f .y/.g.xi 1 x/ C "2 /
P
for all x; y 2 G and all i D 1; : : : ; `. Since Fx D i i Fx , this implies
X
Mh .Fx / Mh .i Fx /
i
X
Mh .i f /.g.xi 1 x/ C "2 / (8.35)
i
X
Mh .i f /g.xi 1 x/ C 2"2 Mh .f /:
i
Thus
.f 2"1 /C Mh .g/ L g;
where X
WD Mh .i f /ıx 1 :
i
i
This implies
X
Mg ..f 2"1 /C /Mh .g/ Mh .i f / .1 C "0 /Mh .f /:
i
324 8. The Haar measure
Here the second step uses (8.32). Since f .f 2"1 /C C 2"1 , we have
and
.1 C "/ƒ.f0 / M.f0 I g/ƒ.g/ ƒ.f0 /:
Take the quotient of these inequalities to obtain
2 ƒ.f / ƒ0 .f / ƒ.f /
.1 C "/ 0 .1 C "/2 :
ƒ.f0 / ƒ .f0 / ƒ.f0 /
8.2. Haar measures 325
Since this holds for all f 2 CcC .G/, ƒ0 and cƒ agree on CcC .G/. Hence ƒ0 D cƒ
by Lemma 8.15. This proves Step 4.
Step 5. We prove existence.
The proof follows the elegant exposition [11] by Urs Lang. Fix a reference function
f0 2 CcC .G/ and, for g 2 CcC .G/, define
by
M.f I g/
ƒg .f / WD for f 2 CcC .G/: (8.36)
M.f0 I g/
Then ƒg 2 L for all g 2 CcC .G/ by Step 1. It follows also from Step 1 that
M.f0 I g/ M.f0 I f /M.f I g/ and M.f I g/ M.f I f0 /M.f0 I g/ and hence
1
M.f0 I f / ƒg .f / M.f I f0 / (8.37)
for all f; g 2 CcC .G/. Fix a function f 2 CcC .G/ and a number " > 0. Define
8 ˇ 9
< ˇ ƒ.f0 / D 1 and there exists a neighborhood =
L" .f / WD ƒ 2 L ˇˇ W G of 1 such that for all h 2 CcC .G/
ˇ
:
: ˇ supp.h/ W H) ƒ.f / .1 C "/ƒh .f / ;
We prove that the functional ƒ0 W CcC .G/ ! .0; 1/ is left-invariant. To see this,
fix a function f 2 CcC .G/ and an element x 2 G. Then
L" .f / D L" .f ı Lx /
for all " > 0. Namely, if W G is an open neighborhood of 1 such that
ƒ.f / .1 C "/ƒh .f / for all h 2 CcC .G/ with supp.h/ W , then the same in-
equality holds with f replaced by f ı Lx because both ƒ and ƒh are left-invariant.
Hence ƒ" .f / D ƒ" .f ı Lx / for all " > 0, and so ƒ0 .f / D ƒ0 .f ı Lx /.
We prove that the functional ƒ0 W CcC .G/ ! .0; 1/ is additive. To see this,
fix two functions f; f 0 2 CcC .G/. We show that
1
.1 C "/ ƒ" .f C f 0 / ƒ" .f / C ƒ" .f 0 / .1 C "/ƒ" .f C f 0 / (8.41)
for all " > 0. To establish the first inequality in (8.41), choose any functional
ƒ 2 L" .f C f 0 /. Then there exists an open neighborhood W G of 1 such that
ƒ.f Cf 0 / .1C"/ƒh .f Cf 0 / for all h 2 CcC .G/ with supp.h/ W . Moreover,
we have seen above that h 2 CcC .G/ can be chosen such that supp.h/ W and
also ƒh 2 L" .f / \ L" .f 0 /. Any such h satisfies
1
.1 C "/ ƒ.f C f 0 / ƒh .f C f 0 / ƒh .f / C h .f 0 / ƒ" .f / C ƒ" .f 0 /:
Take the supremum over all ƒ 2 L" .f C f 0 / to obtain the first inequality in (8.41).
To prove the second inequality in (8.41), fix a constant ˛ > 1 and choose two
functionals ƒ 2 L" .f / and ƒ0 2 L" .f 0 /. Then there exists an open neighborhood
W G of 1 such that ƒ.f / .1 C "/ƒh .f / and ƒ0 .f 0 / .1 C "/ƒh .f 0 / for
all h 2 CcC .G/ with supp.h/ W . By Step 2, the function h 2 CcC .G/ can be
chosen such that supp.h/ W and also ƒh .f / C ƒh .f 0 / ˛ƒh .f C f 0 / and
ƒh 2 L" .f C f 0 /. Any such h satisfies
1
.1 C "/ .ƒ.f / C ƒ0 .f 0 // ƒh .f / C ƒh .f 0 /
˛ƒh .f C f 0 /
˛ƒ" .f C f 0 /:
8.2. Haar measures 327
Take the supremum over all pairs of functionals ƒ 2 L" .f / and ƒ0 2 L" .f 0 /
to obtain .1 "/ 1 .ƒ" .f / C ƒ" .f 0 // ˛ƒ" .f C f 0 / for all ˛ > 1: This proves
the second inequality in (8.41). Take the limit " ! 0 in (8.41) to obtain that ƒ0 is
additive. Moreover, it follows directly from the definition that ƒ0 .f0 / D 1. Hence
by Lemma 8.15, ƒ0 extends to a nonzero left-invariant positive linear functional on
Cc .G/. This proves Step 5 and Theorem 8.14.
If one is prepared to use some abstract concepts from general topology, then the
existence proof in Theorem 8.14 is essentially complete after Step 2. This approach
is taken in Pedersen [16]. In this language the space
is a directed set equipped with a map g 7! ƒg that takes values in the space
Proof of Theorem 8.12. Existence and uniqueness in (i) follow directly from
Theorem 8.14 and the Riesz representation theorem (Theorem 3.15). That nonempty
open sets have positive measure follows from Urysohn’s lemma (Theorem A.1).
To prove (ii), consider the map
W G ! G
defined by
1
.x/ WD x for x 2 G.
328 8. The Haar measure
ı Rx D Lx 1 ı ;
W B ! Œ0; 1
defined by
1
.B/ WD ..B// D .B /
x W B ! Œ0; 1
by
x .B/ WD .Rx .B// for B 2 B.
Since Rx commutes with Ly for all y by (8.11), x is a left Haar measure. Since
x .G/ D .Rx .G// D .G/ D 1, it follows that x D for all x 2 G. Hence
is right-invariant. Therefore, the map
is a left Haar measure and, since .G/ D 1, it agrees with . This proves Theo-
rem 8.12.
In the noncompact case the left and right Haar measures need not agree.
The above argument then shows that the measure x differs from by a positive
factor. Thus there exists a unique map W G ! .0; 1/ such that
for all x 2 G and all B 2 B. The map W G ! .0; 1/ in (8.42) is a continuous group
homomorphism, called the modular character. It is independent of the choice of .
A locally compact Hausdorff group is called unimodular if its modular character is
trivial or, equivalently, if its left and right Haar measures agree. Thus every compact
Hausdorff group is unimodular.
8.2. Haar measures 329
Exercise 8.19. Prove that the group of all real 2 2-matrices of the form
a b
; a; b 2 R; a > 0;
0 1
is not unimodular. Prove that the additive group Rn is unimodular. Prove that every
discrete group is unimodular.
Exercise 8.20. Let W B ! Œ0; 1 be a right Haar measure. Show that the modular
character is characterized by the condition .Lx 1 .B// D .x/.B/ for all x 2 G
and all B 2 B.
Haar measures are extremely useful tools in geometry, especially when the group
in question is compact. For example, if a compact Hausdorff group G acts on a
topological space X continuously via
G X ! XW .g; x/ 7 ! g x (8.43)
one can use the Haar measure to produce G-invariant continuous functions by aver-
aging. Namely, if f W X ! R is any continuous function, and is the Haar measure
on G with .G/ D 1, then the function F W X ! R defined by
Z
F .x/ WD f .a x/ d.a/ (8.44)
G
F .g x/ D F .x/
Exercise 8.21. Give a precise definition of what it means for a topological group to
act continuously on a topological space.
Exercise 8.22. Show that the map F in (8.44) is continuous and G-invariant.
Exercise 8.24. Show that the Haar measure on a discrete group is a multiple of the
counting measure. Deduce that for a finite group the formula (8.44) defines F .x/
as the average (with multiplicities) of the values of f over the group orbit of x.
Exercise 8.25. Let G be a locally compact Hausdorff group and let be a left Haar
measure on G. Define the convolution product on L1 ./. Show that L1 ./ is a
Banach algebra. (See page 282.) Find conditions under which f g D g f .
Show that the convolution is not commutative in general. Hint. See Section 7.5 for
G D Rn . See also Step 3 in the proof of Theorem 8.14.
Appendices
Appendix A
Urysohn’s lemma
K U:
f W X ! Œ0; 1
such that
f jK 1; supp.f / D ¹x 2 X j f .x/ ¤ 0º U: (A.1)
Lemma A.2. Let X be a topological space and let K X be compact. Then the
following holds:
Proof. (i) Let F K be closed and let ¹Ui ºi2I be an open cover of F . Then the
sets ¹Ui ºi 2I together with V WD X n F form an open cover of K. Therefore, there
exist finitely many indices i1 ; : : : ; in such that the sets Ui1 ; : : : ; Uin ; V cover K.
Hence F Ui1 [ [ Uin . This shows that every open cover of F has a finite
subcover, and so F is compact.
(ii) Assume X is Hausdorff and let y 2 X n K. Define
U WD ¹U X j U is open and y … Ux º:
334 A. Urysohn’s lemma
K U WD U1 [ [ Un :
Lemma A.3. Let X be a locally compact Hausdorff space and let K; U be subsets
of X such that K is compact, U is open, and K U . Then there exists an open set
V X such that Vx is compact and
K V Vx U: (A.2)
Proof. We first prove the assertion in the case where K D ¹xº consist of a single
element. Choose a compact neighborhood B X of x. Then F WD B n U is a
closed subset of B and hence is compact by part (i) of Lemma A.2. Since x … F ,
it follows from part (ii) of Lemma A.2 that there exist open sets W; W 0 X such
that x 2 W , F W 0 and W \ W 0 D ;. Hence V WD int.B/ \ W is an open
neighborhood of x, its closure is a closed subset of B and hence compact, and
Vx B \ W
S B n W 0 B n F U:
Proof of Theorem A.1. The proof has three steps. The first step requires the Axiom
of Dependent Choice.1
1
See http://en.wikipedia.org/wiki/Axiom_of_dependent_choice.
335
Step 1. There exists a family of open sets Vr X with compact closure, para-
metrized by r 2 Q \ Œ0; 1, such that
K V1 Vx1 V0 Vx0 U (A.3)
and
s > r H) Vxs Vr (A.4)
for all r; s 2 Q \ Œ0; 1.
The existence of open sets V0 and V1 with compact closure satisfying (A.3) follows
from Lemma A.3. Now choose a bijective map N0 ! Q \ Œ0; 1W i 7! qi such that
q0 D 0 and q1 D 1. Suppose by induction that the open sets Vi D Vqi have been
constructed for i D 0; 1; : : : ; n such that (A.4) holds for r; s 2 ¹q0 ; q1 ; : : : ; qn º.
Choose k; ` 2 ¹0; 1; : : : ; nº such that
qk WD max¹qi j 0 i n; qi < qnC1 º;
This implies that the pre-image under f of every open interval in R is an open
subset of X. Hence also the pre-image under f of every union of open intervals is
open in X and so f is continuous. This proves Step 3 and Theorem A.1.
Theorem A.4 (partition of unity). Let X be a locally compact Hausdorff space, let
U1 ; : : : ; Un X be open sets, and let K U1 [ [ Un be a compact set. Then
there exist continuous functions f1 ; : : : ; fn W X ! R with compact support such that
n
X
fi 0; fi 1; supp.fi / Ui
i D1
Pn
for all i and iD1 fi .x/ D 1 for all x 2 K.
337
0 gi 1; supp.gi / Ui ; gi jKi 1:
Define
f1 WD g1 ;
f2 WD .1 g1 /g2 ;
f3 WD .1 g1 /.1 g2 /g3 ;
::
:
fn WD .1 g1 / .1 gn 1 /gn :
Let .X; UX / and .Y; UY / be topological spaces, denote the product space by
X Y WD ¹.x; y/ j x 2 X; y 2 Y º;
(F) Let .Z; UZ / be any topological space and let hW Z ! X Y be any map.
Then hW .Z; UZ / ! .X Y; U/ is continuous if and only if the maps
are continuous.
UXY 2X Y :
h WD idW .X Y; U/ ! .X Y; U0 /:
Let U be the topology in (i) and let U0 be any topology on X Y with respect
to which X and Y are continuous. If U 2 UX and V 2 UY , then U V D
X 1 .U / \ Y 1 .V / 2 U0 . Hence U U0 by (ii). Since X and Y are continuous
with respect to U, it follows that U is the smallest topology on X Y with respect
to which X and Y are continuous.
341
This appendix contains a proof of the inverse function theorem. The result is for-
mulated in the setting of continuously differentiable maps between open sets in a
Banach space. Readers who are unfamiliar with bounded linear operators on Banach
spaces may simply think of continuously differentiable maps between open sets in
finite-dimensional normed vector spaces. The inverse function theorem is used on
page 81 in the proof of Lemma 2.19, which is a key step in the proof of the transfor-
mation formula for the Lebesgue measure (Theorem 2.17). Assume throughout that
.X; kk/ is a Banach space. When ˆW X ! X is a bounded linear operator denote
its operator norm by
kˆxk
kˆk WD kˆkL.X/ WD sup :
x2X n¹0º kxk
Theorem C.1 (inverse function theorem). Fix an element x0 2 X and two real num-
bers r > 0 and 0 < ˛ < 1. Let W Br .x0 / ! X be a continuously differentiable
map such that
Then
B.1 ˛/r . .x0 // .Br .x0 // B.1C˛/r . .x0 //: (C.2)
1
Moreover, the map is injective, its image is open, the map is continuously
differentiable, and
1 1 1
d .y/ D d . .y// for all y 2 .Br .x0 //.
344 C. The inverse function theorem
The second inequality in (C.4) shows that .Br / B.1C˛/r and the first inequality
1
in (C.4) shows that is injective and is Lipschitz continuous.
Step 2. B.1 ˛/r .Br /.
Let 2 B.1 ˛/r and define " > 0 by kk DW .1 ˛/.r "/. Then, by (C.3) with
y D 0, we have k.x/k ˛kxk for all x 2 Br . If kxk r " this implies
k.x/ C k r ". Thus the map x 7! .x/ C is a contraction of the closed
ball Bxr " . By the contraction mapping principle, it has a unique fixed point x and
the fixed point satisfies .x/ D x .x/ D . Hence 2 .Br /.
Step 3. .Br / is open.
Let x 2 Br and define y WD .x/. Choose " > 0 such that B" .x/ Br . Then,
by Step 2, B.1 ˛/" . .x// .B" .x// .Br /.
1
Step 4. is continuously differentiable.
Let x0 2 Br and define y0 WD .x0 / and ‰ WD d .x0 /. Then k1 ‰k ˛, so ‰
is invertible, ‰ 1 D 1 ‰/k , and k‰ 1 k .1 ˛/ 1 . We prove that 1
P
kD0 .1
1 1
is differentiable at y0 and d .y0 / D ‰ . Let " > 0. Since is differentiable
at x0 and d .x0 / D ‰, there is a constant ı > 0 such that, for all x 2 Br with
kx x0 k < ı.1 ˛/ 1 , we have
and hence
1 1 1 1
k .y/ .y0 / ‰ .y y0 /k D k‰ .y y0 ‰.x x0 //k
1
k .x/ .x0 / ‰.x x0 /k
1 ˛
".1 ˛/ kx x0 k
"ky y0 k:
1
Hence is differentiable at y0 and d 1 .y0 / D ‰ 1 D d . 1
.y0 // 1
.
1
Thus d is continuous by Step 1. This proves Theorem C.1.
Bibliography
[1] R. B. Ash, Basic probability theory. John Wiley & Sons, New York, 1970.
Reprint, Dover Publications, Mineola, N,Y., 2008. John Wiley & Sons 1970,
Dover Publications, 2008.
MR 0268923 Zbl 1234.60001 Zbl 0206.18001 (reprint)
http://www.math.uiuc.edu/~r-ash/BPT/BPT.pdf
[2] V. I. Bogachev, Measure Theory. Volumes I and II. Springer, Berlin etc., 2006.
MR 2267655 Zbl 1120.28001
[9] T. W. Körner, A companion to analysis. A second first and first second course
in analysis. Graduate Studies in Mathematics, 62. American Mathematical
Society, Providence, R.I., 2004.
MR 2015825 Zbl 1061.26003
https://www.dpmms.cam.ac.uk/~twk/
[11] U. Lang, Existence and uniqueness of Haar integrals. Preprint. ETH Zürich,
January 2015.
http://www.math.ethz.ch/~lang/HaarIntegral.pdf
[12] D. Ma, The lexicographic order and the double arrow space. Online survey,
2009.
http://dantopology.wordpress.com/2009/10/07/
[14] J.R. Munkres, Topology. Second edition. Prentice Hall, Upper Saddle River,
N.J., 2000.
Zbl 0951.54001
[15] G. Nagy, Real Alalysis. Version 2.0. Section IV.6, Duals of Lp spaces. Lecture
notes. Kansas State University, Manhattan, KS.
http://www.math.ksu.edu/~nagy/real-an/4-06-dual-lp.pdf
[16] G. K. Pedersen, The existence and uniqueness of the Haar integral on a locally
compact topological group. Preprint, University of Copenhagen, Copenhagen,
2000.
http://www.math.ku.dk/kurser/2004-2/mat3re/haarintegral.pdf
Bibliography 349
[17] W. Rudin, Real and complex analysis. Third edition. McGraw-Hill, New York,
1987.
MR 0924157 Zbl 0925.00005
[18] D. A. Salamon, Das Riemannsche Integral. Lecture Notes, ETH, Zürich, 2014.
http://www.math.ethz.ch/~salamon/PREPRINTS/ana1-int.pdf
—A— —C—
absolutely continuous Calderón–Zygmund inequality, 295
function, 239, 245 Carathéodory criterion, 61
measure, 179 Cauchy–Schwarz inequality, 147
signed measure, 200, 206 characteristic function, 12
almost everywhere, 36 closed set, 6
compact set, 10
compactly supported function, 113
—B— complete measure space, 43
Baire measurable complete metric space, 8
completion of a measure space, 43
function, 129
continuous map, 11
set, 129
continuum hypothesis, 147
Baire measure, 129
convergence in measure, 173
Baire -algebra, 129
convolution, 277
Banach algebra, 282
of signed measures, 307
Banach space, 8 counting measure, 21
Banach–Zarecki theorem, 239 cuboid, 64
basis of a topological space, 124
bilinear form
positive definite, 147
symmetric, 147 —D—
Boolean algebra, 6 dense subset, 141
Borel measurable Dieudonné’s measure, 96
function, 12 Dirac measure, 21
set, 9 divergence theorem, 306
Borel measure, 95 double arrow space, 126
extended, 107 dual space
left-invariant, 314 of a Hilbert space, 149
right-invariant, 314 of a normed vector space, 149
Borel outer measure, 108 of C.X /, 124, 218
Borel set, 9 of L2 ./, 152
Borel -algebra, 9 of Lp ./, 159
bounded linear functional, 149 of `1 , 160, 176
bounded variation, 243
352 Index
—E— —I—
Egoroff’s Thorem, 172 inner product, 147
elementary set, 250 on L2 ./, 148
envelope, 152 inner regular, 96, 208
essential supremum, 138 on open sets, 101
integrable function
Lebesgue, 32
—F— locally, 283
first countable, 124 partially defined, 46
Fubini’s Theorem Riemann, 87
for integrable functions, 265 weakly, 224
for positive functions, 261 integral
for the completion, 268 Haar, 315
Lebesgue, 22, 32
for the Lebesgue measure, 276
Riemann, 87
in polar coordinates, 305
Riemann–Stieltjes, 244
fundamental solution
invariant
of Laplace’s equation, 292
linear functional, 315
fundamental theorem
measure, 314
of calculus, 239 inverse limit, 311
—G— —J—
Jensen’s inequality, 168
Green’s formula, 293
Jordan decomposition, 200, 204
group
Jordan measurable set, 87
discrete, 309
Jordan measure, 87
Lie, 310
Jordan null set, 65
topological, 309
unimodular, 328
—L—
—H— Laplace operator, 292
Lebesgue
Haar integral, 315
differentiation theorem, 234, 238
Haar measure, 314
Lebesgue decomposition, 180
Hahn decomposition, 202
for signed measures, 201
Hahn–Banach theorem, 160
Lebesgue dominated
Hardy’s inequality, 171
convergence theorem, 35
Hardy–Littlewood Lebesgue integrable, 32
maximal inequality, 230 Lebesgue integral, 22, 32
Hausdorff dimension, 93 Lebesgue measurable
Hausdorff measure, 92 function, 57, 69
Hausdorff space, 11, 95 set, 57, 69
Hilbert space, 148 Lebesgue measure, 57, 69
Hölder inequality, 134 on the sphere, 304
Index 353
—N— —R—
neighborhood, 95 Radon measure, 96
nonatomic measure, 147 Radon–Nikodým derivative, 236
norm of a bounded Radon–Nikodým theorem, 180
linear functional, 149 for signed measures, 202
normed vector space, 7 generalized, 209
null set, 36 regular measure, 96
Jordan, 65 Riemann integrable, 87
Lebesgue, 65 Riemann integral, 87
Riemann–Stieltjes integral, 244
Riesz representation theorem, 115
—O— right-invariant
one-point compactification, 127 linear functional, 315
open ball, 7 measure, 314
open set, 6
in a metric space, 7 —S—
outer measure, 58 second countable, 124
Borel, 108 semi-finite, 152
Lebesgue, 65 separability of Lp ./, 142
translation-invariant, 65 separable, 141
outer regular, 95 set of measure zero, 36
-additive
measure, 19
—P— signed measure, 197
p-adic integers, 311 -algebra, 5
p-adic rationals, 312 Baire, 129
partition of a set, 5 Borel, 9
partition of unity, 336 Lebesgue, 57, 69
perfectly normal, 126 product, 249
metric spaces are, 128 -compact, 95
Poisson identity, 293 -finite, 152
positive linear functional signed measure, 197
on Cc .X /, 113 absolutely continuous, 200, 206
on Lp ./, 161 concentrated, 200
pre-image, 11 Hahn decomposition, 202
probability theory, 51–55, 147 inner regular
product measure, 255 w.r.t. a measure, 208
complete locally determined, 270 Jordan decomposition, 200, 204
primitive, 270 Lebesgue decomposition, 201
product -algebra, 249 mutually singular, 200
product topology, 312, 340 Radon–Nikodým theorem, 202
pushforward total variation, 197
of a measure, 50 truly continuous, 206
of a -algebra, 13, 50 simple function, 17
singular measure, 179
Index 355
Dietmar A. Salamon
Dietmar A. Salamon Dietmar A. Salamon
Measure and Integration
Measure
ISBN 978-3-03719-159-0
www.ems-ph.org
Salamon Cover | Font: Frutiger_Helvetica Neue | Farben: Pantone 116, Pantone 287 | RB 31 mm