MEASURE THEORY
Volume 4
D.H.Fremlin
By the same author:
Topological Riesz Spaces and Measure Theory, Cambridge University Press, 1974.
Consequences of Martin’s Axiom, Cambridge University Press, 1982.
Companions to the present volume:
Measure Theory, vol. 1, Torres Fremlin, 2000.
Measure Theory, vol. 2, Torres Fremlin, 2001.
Measure Theory, vol. 3, Torres Fremlin, 2002.
First printing November 2003
MEASURE THEORY
Volume 4
Topological Measure Spaces
D.H.Fremlin
Research Professor in Mathematics, University of Essex
Dedicated by the Author
to the Publisher
This book may be ordered from the publisher at the address below. For price and means of pay
ment see the author’s Web page http://www.essex.ac.uk/maths/staff/fremlin/mtsales.htm,
or enquire from fremdh@essex.ac.uk.
First published in 2003
by Torres Fremlin, 25 Ireton Road, Colchester CO3 3AT, England
c _ D.H.Fremlin 2003
The right of D.H.Fremlin to be identiﬁed as author of this work has been asserted in accordance with
the Copyright, Designs and Patents Act 1988. This work is issued under the terms of the Design Science
License as published in http://dsl.org/copyleft/dsl.txt. For the source ﬁles see http://www.essex.
ac.uk/maths/staff/fremlin/mt4.2003/index.htm.
Library of Congress classiﬁcation QA312.F72
AMS 2000 classiﬁcation 28A99
ISBN 0953812944
Typeset by /
/
oT
E
X
Printed in England by Biddles Short Run Books, King’s Lynn
5
Contents
General Introduction 10
Introduction to Volume 4 11
Chapter 41: Topologies and measures I
Introduction 13
411 Deﬁnitions 13
Topological, inner regular, τadditive, outer regular, locally ﬁnite, eﬀectively locally ﬁnite, quasiRadon,
Radon, completion regular, Baire, Borel and strictly positive measures; measurable and almost contin
uous functions; selfsupporting sets and supports of measures; Stone spaces; Dieudonn´e’s measure.
412 Inner regularity 19
Exhaustion; Baire measures; Borel measures on metrizable spaces; completions and c.l.d. versions; com
plete locally determined spaces; inversemeasurepreserving functions; subspaces; indeﬁniteintegral mea
sures; products; outer regularity.
413 Inner measure constructions 31
Inner measures; constructing a measure from an inner measure; the inner measure deﬁned by a measure;
complete locally determined spaces; extension of functionals to measures; countably compact classes;
constructing measures dominating given functionals.
414 τadditivity 50
Semicontinuous functions; supports; strict localizability; subspace measures; regular topologies; density
topologies; lifting topologies.
415 QuasiRadon measure spaces 58
Strict localizability; subspaces; regular topologies; hereditarily Lindel¨of spaces; products of separable
metrizable spaces; comparison and speciﬁcation of quasiRadon measures; construction of quasiRadon
measures extending given functionals; indeﬁniteintegral measures; L
p
spaces; Stone spaces.
416 Radon measure spaces 73
Radon and quasiRadon measures; speciﬁcation of Radon measures; c.l.d. versions of Borel measures;
locally compact topologies; constructions of Radon measures extending or dominating given functionals;
additive functionals on Boolean algebras and Radon measures on Stone spaces; subspaces; products;
Stone spaces of measure algebras; compact and perfect measures; representation of homomorphisms of
measure algebras; the split interval.
417 τadditive product measures 88
The product of two eﬀectively locally ﬁnite τadditive measures; the product of many τadditive proba
bility measures; Fubini’s theorem; generalized associative law; measures on subproducts as image mea
sures; products of strictly positive measures; quasiRadon and Radon product measures; when ‘ordinary’
product measures are τadditive; continuous functions and Baire σalgebras in product spaces.
418 Measurable functions and almost continuous functions 110
Measurable functions; into (separable) metrizable spaces; and image measures; almost continuous func
tions; continuity, measurability, image measures; expressing Radon measures as images of Radon mea
sures; Prokhorov’s theorem on projective limits of Radon measures; representing measurable functions
into L
0
spaces.
419 Examples 126
A nearly quasiRadon measure; a Radon measure space in which the Borel sets are inadequate; a nearly
Radon measure; the Stone space of the Lebesgue measure algebra; measures with domain 1ω
1
; notes on
Lebesgue measure.
Chapter 42: Descriptive set theory
Introduction 138
421 Souslin’s operation 138
Souslin’s operation; is idempotent; as a projection operator; SouslinF sets; *constituents.
422 Kanalytic spaces 148
Uscocompact relations; Kanalytic sets; and SouslinF sets; *First Separation Theorem.
423 Analytic spaces 155
Analytic spaces; are Kanalytic spaces with countable networks; SouslinF sets; Borel measurable func
tions; injective images of Polish spaces; nonBorel analytic sets; von NeumannJankow selection theorem;
*constituents of coanalytic sets.
424 Standard Borel spaces 165
Elementary properties; isomorphism types; subspaces; Borel measurable actions of Polish groups.
6
Chapter 43: Topologies and measures II
Introduction 172
431 Souslin’s operation 172
The domain of a complete locally determined measure is closed under Souslin’s operation; the kernel of
a Souslin scheme is approximable from within.
432 Kanalytic spaces 176
Topological measures on Kanalytic spaces; extensions to Radon measures; expressing Radon measures
as images of Radon measures.
433 Analytic spaces 180
Measures on spaces with countable networks; inner regularity of Borel measures; expressing Radon
measures as images of Radon measures; measurable and almost continuous functions; the von Neumann
Jankow selection theorem; products; extension of measures on σsubalgebras; standard Borel spaces.
434 Borel measures 184
Classiﬁcation of Borel measures; Radon spaces; universally measurable sets and functions; Borelmeasure
compact, Borelmeasurecomplete and preRadon spaces; countable compactness and countable tight
ness; quasidyadic spaces and completion regular measures; ﬁrstcountable spaces and Borel product
measures.
435 Baire measures 203
Classiﬁcation of Baire measures; extension of Baire measures to Borel measures (Maˇrik’s theorem);
measurecompact spaces; sequential spaces and Baire product measures.
436 Representation of linear functionals 209
Smooth and sequentially smooth linear functionals; measures and sequentially smooth functionals; Baire
measures; products of Baire measures; quasiRadon measures and smooth functionals; locally compact
spaces and Radon measures.
437 Spaces of measures 219
Smooth and sequentially smooth duals; signed measures; embedding spaces of measurable functions in
the bidual of C
b
(X); vague and narrow topologies; product measures; extreme points; uniform tightness;
Prokhorov spaces.
438 Measurefree cardinals 240
Measurefree cardinals; pointﬁnite families of sets with measurable unions; measurable functions into
metrizable spaces; Radon and measurecompact metric spaces; metacompact spaces; hereditarily weakly
θreﬁnable spaces; when c is measurefree.
439 Examples 254
Measures with no extensions to Borel measures; universally negligible sets; Hausdorﬀ measures are rarely
semiﬁnite; a smooth linear functional not expressible as an integral; a ﬁrstcountable nonRadon space;
Baire measures not extending to Borel measures; N
c
is not Borelmeasurecompact; the Sorgenfrey line;
Q is not a Prokhorov space.
Chapter 44: Topological groups
Introduction 271
441 Invariant measures on locally compact spaces 271
Measures invariant under homeomorphisms; Haar measures; measures invariant under isometries.
442 Uniqueness of Haar measure 280
Two (left) Haar measures are multiples of each other; left and right Haar measures; Haar measurable
and Haar negligible sets; the modular function of a group; formulae for
_
f(x
−1
)dx,
_
f(xy)dx.
443 Further properties of Haar measure 287
The Haar measure algebra of a group carrying Haar measures; actions of the group on the Haar measure
algebra; locally compact groups; actions of the group on L
0
and L
p
; the bilateral uniformity; Borel sets
are adequate; completing the group; expressing an arbitrary Haar measure in terms of a Haar measure
on a locally compact group; completion regularity of Haar measure; invariant measures on the set of left
cosets of a closed subgroup of a locally compact group; modular functions of subgroups and quotient
groups; transitive actions of compact groups on compact spaces.
444 Convolutions 311
Convolutions of quasiRadon measures; the Banach algebra of signed τadditive measures; continuous
actions and corresponding actions on L
0
(ν) for an arbitrary quasiRadon measure ν; convolutions of
measures and functions; indeﬁniteintegral measures over a Haar measure µ; convolutions of functions;
L
p
(µ); approximate identities.
7
445 The duality theorem 334
Dual groups; FourierStieltjes transforms; Fourier transforms; identifying the dual group with the max
imal ideal space of L
1
; the topology of the dual group; positive deﬁnite functions; Bochner’s theorem;
the Inversion Theorem; the Plancherel Theorem; the Duality Theorem.
446 The structure of locally compact groups 357
Finitedimensional representations separate the points of a compact group; groups with no small sub
groups have Bsequences; chains of subgroups.
447 Translationinvariant liftings 372
Translationinvariant liftings and lower densities; Vitali’s theorem and a density theorem for groups with
Bsequences; Haar measures have translationinvariant liftings.
448 Invariant measures on Polish spaces 383
Countably full local semigroups of Aut A; σequidecomposability; countably nonparadoxical groups;
Ginvariant additive functions from A to L
∞
(C); measures invariant under Polish group actions (the
NadkarniBeckerKechris theorem).
449 Amenable groups 393
Amenable groups; permanence properties; locally compact amenable groups; Tarski’s theorem; discrete
amenable groups.
Chapter 45: Perfect measures, disintegrations and processes
Introduction 413
451 Perfect, compact and countably compact measures 414
Basic properties of the three classes; subspaces, completions, c.l.d. versions, products; measurable func
tions from compact measure spaces to metrizable spaces; *weakly αfavourable spaces.
452 Integration and disintegration of measures 427
Integrating families of probability measures; τadditive and Radon measures; disintegrations and regular
conditional probabilities; disintegrating countably compact measures; disintegrating Radon measures;
*images of countably compact measures.
453 Strong liftings 441
Strong and almost strong liftings; existence; on product spaces; disintegrations of Radon measures over
spaces with almost strong liftings; Stone spaces; Losert’s example.
454 Measures on product spaces 454
Perfect, compact and countably compact measures on product spaces; extension of ﬁnitely additive
functions with perfect countably additive marginals; Kolmogorov’s extension theorem; measures deﬁned
from conditional distributions; distributions of random processes; measures on C(T) for Polish T.
455 Markov process and Brownian motion 464
Deﬁnition of Markov process from conditional distributions; existence of a measure representing Brown
ian motion; continuous sample paths.
456 Gaussian distributions 472
Gaussian distributions; supports; universal Gaussian distributions; cluster sets of ndimensional pro
cesses; τadditivity; Gaussian processes.
457 Simultaneous extension of measures 488
Extending families of ﬁnitely additive functionals; Strassen’s theorem; extending families of measures;
examples.
458 Relative independence and relative products 497
Relatively independent families of σalgebras and random variables; relative distributions; relatively
independent families of closed subalgebras of a probability algebra; relative free products of probability
algebras; relative products of probability spaces; existence of relative products.
459 Symmetric measures and exchangeable random variables 510
Exchangeable families of inversemeasurepreserving functions; de Finetti’s theorem; countably compact
symmetric measures on product spaces disintegrate into product measures; symmetric quasiRadon
measures.
Chapter 46: Pointwise compact sets of measurable functions
Introduction 522
461 Barycenters and Choquet’s theorem 522
Barycenters; elementary properties; suﬃcient conditions for existence; closed convex hulls of compact
sets; Kreˇın’s theorem; measures on sets of extreme points.
462 Pointwise compact sets of continuous functions 532
Angelic spaces; the topology of pointwise convergence on C(X); weak convergence and weakly compact
sets in C
0
(X); Radon measures on C(X); separately continuous functions; convex hulls.
8
463 T
p
and T
m
538
Pointwise convergence and convergence in measure on spaces of measurable functions; compact and
sequentially compact sets; perfect measures and Fremlin’s Alternative; separately continuous functions.
464 Talagrand’s measure 550
The usual measure on 1I; the intersection of a sequence of nonmeasurable ﬁlters; Talagrand’s measure;
the Lspace of additive functionals on 1I; measurable and purely nonmeasurable functionals.
465 Stable sets 563
Stable sets of functions; elementary properties; pointwise compactness; pointwise convergence and con
vergence in measure; a law of large numbers; stable sets and uniform convergence in the strong law of
large numbers; stable sets in L
0
and L
1
; *Rstable sets.
466 Measures on linear topological spaces 590
QuasiRadon measures for weak and strong topologies; Kadec norms; constructing weakBorel measures;
characteristic functions of measures on locally convex spaces; universally measurable linear operators.
*467 Locally uniformly rotund norms 598
Locally uniformly rotund norms; separable normed spaces; long sequences of projections; Kcountably
determined spaces; weakly compactly generated spaces; Banach lattices with ordercontinuous norms;
Eberlein compacta.
Chapter 47: Geometric measure theory
Introduction 610
471 Hausdorﬀ measures 610
Metric outer measures; Increasing Sets Lemma; analytic spaces; inner regularity; Vitali’s theorem and a
density theorem; Howroyd’s theorem.
472 Besicovitch’s Density Theorem 626
Besicovitch’s Covering Lemma; Besicovitch’s Density Theorem; *a maximal theorem.
473 Poincar´e’s inequality 633
Diﬀerentiable and Lipschitz functions; smoothing by convolution; the GagliardoNirenbergSobolev in
equality; Poincar´e’s inequality for balls.
474 The distributional perimeter 647
The divergence of a vector ﬁeld; sets with locally ﬁnite perimeter, perimeter measures and outward
normal functions; the reduced boundary; invariance under isometries; isoperimetric inequalities; Federer
exterior normals; the Compactness Theorem.
475 The essential boundary 670
Essential interior, closure and boundary; the reduced boundary; perimeter measures; characterizing sets
with locally ﬁnite perimeter; the Divergence Theorem; calculating perimeters from crosssectional counts;
Cauchy’s Perimeter Theorem; the Isoperimetric Theorem for convex sets.
476 Concentration of measure 690
Hausdorﬀ metrics; Vietoris topologies; concentration by partial reﬂection; concentration of measure in
R
r
; the Isoperimetric Theorem; concentration of measure on spheres.
Chapter 48: Gauge integrals
Introduction 704
481 Tagged partitions 704
Tagged partitions and Riemann sums; gauge integrals; gauges; residual sets; subdivisions; examples
(the Riemann integral, the Henstock integral, the symmetric Riemanncomplete integral, the McShane
integral, box products, the approximately continuous Henstock integral).
482 General theory 714
SaksHenstock lemma; when gaugeintegrable functions are measurable; when integrable functions are
gaugeintegrable; I
ν
(f χH); integrating derivatives; B.Levi’s theorem; Fubini’s theorem.
483 The Henstock integral 729
The Henstock and Lebesgue integrals; indeﬁnite Henstock integrals; SaksHenstock lemma; fundamental
theorem of calculus; the Perron integral; ACG
∗
functions.
484 The Pfeﬀer integral 746
The TamaniniGiacomelli theorem; a family of taggedpartition structures; the Pfeﬀer integral; the Saks
Henstock indeﬁnite integral of a Pfeﬀer integrable function; Pfeﬀer’s Divergence Theorem; diﬀerentiating
the indeﬁnite integral; invariance under lipeomorphisms.
9
Chapter 49: Further topics
Introduction 765
491 Equidistributed sequences 765
The asymptotic density ideal ?; equidistributed sequences; when equidistributed sequences exist; Z =
1N/?; eﬀectively regular measures; equidistributed sequences and induced embeddings of measure al
gebras in Z.
492 Combinatorial concentration of measure 782
Concentration of measure in product spaces; concentration of measure in permutation groups.
493 Extremely amenable groups 789
Extremely amenable groups; concentrating additive functionals; measure algebras under .; L
0
; au
tomorphism groups of measure algebras; isometry groups of spheres in inner product spaces; locally
compact groups.
494 Cubes in product spaces 800
Subsets of measure algebras with nonzero inﬁma; product sets included in given sets of positive measure.
495 Poisson point processes 802
Poisson distributions; Poisson point processes; disintegrations; transforming disjointness into stochastic
independence; representing Poisson point processes by Radon measures; exponential distributions and
Poisson point processes on [0, ∞[.
Appendix to Volume 4
Introduction 823
4A1 Set theory 823
Cardinals; closed coﬁnal sets and stationary sets; ∆system lemma; free sets; Ramsey’s theorem; the
Marriage Lemma; ﬁlters; normal ultraﬁlters; Ostaszewski’s ♣; cardinals of σalgebras.
4A2 General topology 827
Glossary; general constructions; F
σ
, G
δ
, zero and cozero sets; countable chain condition; separation
axioms; compact and locally compact spaces; Lindel¨of spaces; Stone
ˇ
Cech compactiﬁcations; uniform
spaces; ﬁrstcountable, sequential, countably tight, metrizable spaces; countable networks; second
countable spaces; separable metrizable spaces; Polish spaces; order topologies.
4A3 Topological σalgebras 848
Borel σalgebras; measurable functions; hereditarily Lindel¨of spaces; secondcountable spaces; Polish
spaces; ω
1
; Baire σalgebras; product spaces; compact spaces; Baire property algebras; cylindrical σ
algebras.
4A4 Locally convex spaces 857
Linear topological spaces; locally convex spaces; HahnBanach theorem; normed spaces; inner product
spaces; maxﬂow mincut theorem.
4A5 Topological groups 863
Group actions; topological groups; uniformities; quotient groups; metrizable groups.
4A6 Banach algebras 869
StoneWeierstrass theorem (fourth form); multiplicative linear functionals; spectral radius; invertible
elements; exponentiation.
Concordance 873
References for Volume 4 874
Index to Volumes 14
Principal topics and results 881
General index 893
10
General introduction In this treatise I aim to give a comprehensive description of modern abstract measure
theory, with some indication of its principal applications. The ﬁrst two volumes are set at an introductory
level; they are intended for students with a solid grounding in the concepts of real analysis, but possibly with
rather limited detailed knowledge. As the book proceeds, the level of sophistication and expertise demanded
will increase; thus for the volume on topological measure spaces, familiarity with general topology will be
assumed. The emphasis throughout is on the mathematical ideas involved, which in this subject are mostly
to be found in the details of the proofs.
My intention is that the book should be usable both as a ﬁrst introduction to the subject and as a reference
work. For the sake of the ﬁrst aim, I try to limit the ideas of the early volumes to those which are really
essential to the development of the basic theorems. For the sake of the second aim, I try to express these ideas
in their full natural generality, and in particular I take care to avoid suggesting any unnecessary restrictions
in their applicability. Of course these principles are to to some extent contradictory. Nevertheless, I ﬁnd that
most of the time they are very nearly reconcilable, provided that I indulge in a certain degree of repetition.
For instance, right at the beginning, the puzzle arises: should one develop Lebesgue measure ﬁrst on the
real line, and then in spaces of higher dimension, or should one go straight to the multidimensional case? I
believe that there is no single correct answer to this question. Most students will ﬁnd the onedimensional
case easier, and it therefore seems more appropriate for a ﬁrst introduction, since even in that case the
technical problems can be daunting. But certainly every student of measure theory must at a fairly early
stage come to terms with Lebesgue area and volume as well as length; and with the correct formulations, the
multidimensional case diﬀers from the onedimensional case only in a deﬁnition and a (substantial) lemma.
So what I have done is to write them both out (¸¸114115). In the same spirit, I have been uninhibited,
when setting out exercises, by the fact that many of the results I invite students to look for will appear in
later chapters; I believe that throughout mathematics one has a better chance of understanding a theorem
if one has previously attempted something similar alone.
As I write this Introduction (September 2003), the plan of the work is as follows:
Volume 1: The Irreducible Minimum
Volume 2: Broad Foundations
Volume 3: Measure Algebras
Volume 4: Topological Measure Spaces
Volume 5: Settheoretic Measure Theory.
Volume 1 is intended for those with no prior knowledge of measure theory, but competent in the elementary
techniques of real analysis. I hope that it will be found useful by undergraduates meeting Lebesgue measure
for the ﬁrst time. Volume 2 aims to lay out some of the fundamental results of pure measure theory
(the RadonNikod´ ym theorem, Fubini’s theorem), but also gives short introductions to some of the most
important applications of measure theory (probability theory, Fourier analysis). While I should like to
believe that most of it is written at a level accessible to anyone who has mastered the contents of Volume 1,
I should not myself have the courage to try to cover it in an undergraduate course, though I would certainly
attempt to include some parts of it. Volumes 3 and 4 are set at a rather higher level, suitable to postgraduate
courses; while Volume 5 will assume a wideranging competence over large parts of analysis and set theory.
There is a disclaimer which I ought to make in a place where you might see it in time to avoid paying for
this book. I make no attempt to describe the history of the subject. This is not because I think the history
uninteresting or unimportant; rather, it is because I have no conﬁdence of saying anything which would not
be seriously misleading. Indeed I have very little conﬁdence in anything I have ever read concerning the
history of ideas. So while I am happy to honour the names of Lebesgue and Kolmogorov and Maharam in
more or less appropriate places, and I try to include in the bibliographies the works which I have myself
consulted, I leave any consideration of the details to those bolder and better qualiﬁed than myself.
The work as a whole is not yet complete; and when it is ﬁnished, it will undoubtedly be too long
to be printed as a single volume in any reasonable format. I am therefore publishing it one part at a
time. However, drafts of most of the rest are available on the Internet; see http://www.essex.ac.uk/
maths/staff/fremlin/mt.htm for detailed instructions. For the time being, at least, printing will be in
short runs. I hope that readers will be energetic in commenting on errors and omissions, since it should be
possible to correct these relatively promptly. An inevitable consequence of this is that paragraph references
may go out of date rather quickly. I shall be most ﬂattered if anyone chooses to rely on this book as a source
Introduction to Volume 4 11
for basic material; and I am willing to attempt to maintain a concordance to such references, indicating
where migratory results have come to rest for the moment, if authors will supply me with copies of papers
which use them.
I mention some minor points concerning the layout of the material. Most sections conclude with lists of
‘basic exercises’ and ‘further exercises’, which I hope will be generally instructive and occasionally enter
taining. How many of these you should attempt must be for you and your teacher, if any, to decide, as no
two students will have quite the same needs. I mark with a >>> those which seem to me to be particularly
important. But while you may not need to write out solutions to all the ‘basic exercises’, if you are in any
doubt as to your capacity to do so you should take this as a warning to slow down a bit. The ‘further
exercises’ are unbounded in diﬃculty, and are uniﬁed only by a presumption that each has at least one
solution based on ideas already introduced. Occasionally I add a ﬁnal ‘problem’, a question to which I do
not know the answer and which seems to arise naturally in the course of the work.
The impulse to write this book is in large part a desire to present a uniﬁed account of the subject.
Crossreferences are correspondingly abundant and wideranging. In order to be able to refer freely across
the whole text, I have chosen a reference system which gives the same code name to a paragraph wherever
it is being called from. Thus 132E is the ﬁfth paragraph in the second section of the third chapter of
Volume 1, and is referred to by that name throughout. Let me emphasize that crossreferences are supposed
to help the reader, not distract her. Do not take the interpolation ‘(121A)’ as an instruction, or even a
recommendation, to lift Volume 1 oﬀ the shelf and hunt for ¸121. If you are happy with an argument as it
stands, independently of the reference, then carry on. If, however, I seem to have made rather a large jump,
or the notation has suddenly become opaque, local crossreferences may help you to ﬁll in the gaps.
Each volume will have an appendix of ‘useful facts’, in which I set out material which is called on
somewhere in that volume, and which I do not feel I can take for granted. Typically the arrangement of
material in these appendices is directed very narrowly at the particular applications I have in mind, and is
unlikely to be a satisfactory substitute for conventional treatments of the topics touched on. Moreover, the
ideas may well be needed only on rare and isolated occasions. So as a rule I recommend you to ignore the
appendices until you have some direct reason to suppose that a fragment may be useful to you.
During the extended gestation of this project I have been helped by many people, and I hope that my
friends and colleagues will be pleased when they recognise their ideas scattered through the pages below.
But I am especially grateful to those who have taken the trouble to read through earlier drafts and comment
on obscurities and errors.
Introduction to Volume 4
I return in this volume to the study of measure spaces rather than measure algebras. For ﬁfty years now
measure theory has been intimately connected with general topology. Not only do a very large proportion
of the measure spaces arising in applications carry topologies related in interesting ways to their measures,
but many questions in abstract measure theory can be eﬀectively studied by introducing suitable topologies.
Consequently any course in measure theory at this level must be frankly dependent on a substantial knowl
edge of topology. With this proviso, I hope that the present volume will be accessible to graduate students,
and will lead them to the most important ideas of modern abstract measure theory.
The ﬁrst and third chapters of the volume seek to provide a thorough introduction into the ways in which
topologies and measures can interact. They are divided by a short chapter on descriptive set theory, on
the borderline between set theory, logic, real analysis and general topology, which I single out for detailed
exposition because I believe that it forms an indispensable part of the background of any measure theorist.
Chapter 41 is dominated by the concepts of inner regularity and τadditivity, coming together in Radon
measures (¸416). Chapter 43 concentrates rather on questions concerning properties of a topological space
which force particular relationships with measures on that space. But plenty of sideissues are treated in
both, such as Lusin measurability (¸418), the deﬁnition of measures from linear functionals (¸436) and
measurefree cardinals (¸438). Chapters 45 and 46 continue some of the same themes, with particular
investigations into ‘disintegrations’ or regular conditional probabilities (¸¸452453), the abstract theory of
stochastic processes (¸¸454455), Talagrand’s theory of GlivenkoCantelli classes (¸465) and the theory of
measures on normed spaces (¸¸466467).
In contrast with the relatively amorphous structure of Chapters 41, 43, 45 and 46, four chapters of this
volume have deﬁnite topics. I have already said that Chapter 42 is an introduction to descriptive set theory;
12 Introduction to Volume 4
like Chapters 31 and 35 in the last volume, it is a kind of appendix brought into the main stream of the
argument. Chapter 44 deals with topological groups. Most of it is of course devoted to Haar measure,
giving the Pontryaginvan Kampen duality theorem (¸445) and the Ionescu Tulcea theorem on the existence
of translationinvariant liftings (¸447). But there are also sections on Polish groups (¸448) and amenable
groups (¸449), and some of the general theory of measures on measurable groups (¸444). Chapter 47 is a
second excursion, after Chapter 26, into geometric measure theory. It starts with Hausdorﬀ measures (¸471),
gives a proof of the Di GiorgioFederer Divergence Theorem (¸475), and then examines a number of examples
of ‘concentration of measure’ (¸476). In Chapter 48, I set out the elementary theory of gauge integrals, with
sections on the Henstock and Pfeﬀer integrals (¸¸483484). Finally, in Chapter 49, I give notes on ﬁve special
topics: equidistributed sequences (¸491), combinatorial forms of concentration of measure (¸492), extremely
amenable groups (¸493), subproducts in product spaces (¸494) and Poisson point processes (¸495).
I had better mention prerequisites, as usual. To embark on this material you will certainly need a solid
foundation in measure theory. Since I do of course use my own exposition as my principal source of references
to the elementary ideas, I advise readers to ensure that they have easy access to all three previous volumes
before starting serious work on this one. But you may not need to read very much of them. It might be
prudent to glance through the detailed contents of Volume 1 and the ﬁrst ﬁve chapters of Volume 2 to check
that most of the material there is more or less familiar. But Volume 3, and the last three chapters of Volume
2, can probably be left on one side for the moment. Of course you will need the Lifting Theorem (Chapter
34) for ¸¸447, 452 and 453, and Chapter 26 is essential background for Chapter 47, while Chapter 28 (on
Fourier analysis) may help to make sense of Chapter 44, and parts of Chapter 27 (on probability theory)
are necessary for ¸¸455456. And measure algebras are mentioned in every chapter except (I think) Chapter
48; but I hope that the crossreferences are precise enough to lead you to just what you need to know at
any particular point. Even Maharam’s theorem is hardly used in this volume.
What you will need, apart from any knowledge of measure theory, is a sound background in general
topology. This volume calls on a great many miscellaneous facts from general topology, and the list in
¸4A2 is not a good place to start if continuity and compactness and the separation axioms are unfamiliar.
My primary reference for topology is Engelking 89. I do not insist that you should have read this book
(though of course I hope you will do so sometime); but I do think you should make sure that you can use it.
In the general introduction to this treatise, I wrote ‘I make no attempt to describe the history of the
subject’, and I have generally been casual – some would say negligent – in my attributions of results to
their discoverers. Through much of the ﬁrst three volumes I did at least have the excuse that the history
exists in print in far more detail than I am qualiﬁed to describe. In the present volume I ﬁnd my position
more uncomfortable, in that I have been watching the evolution of the subject relatively closely over the
last thirty years, and ought to be able to say something about it. Nevertheless I remain reluctant to make
deﬁnite statements crediting one person rather than another with originating an idea. My more intimate
knowledge of the topic makes me even more conscious than elsewhere of the danger of error and of the
breadth of reading that would be necessary to produce a balanced account. In some cases I do attach a
result to a speciﬁc published paper, but these attributions should never be regarded as an assertion that
any particular author has priority; at most, they declare that a historian should examine the source cited
before coming to any decision. I assure my friends and colleagues that my omissions are not intended to
slight either them or those we all honour. What I have tried to do is to include in the bibliography to this
volume all the published work which (as far as I am consciously aware) has inﬂuenced me while writing it,
so that those who wish to go into the matter will have somewhere to start their investigations.
411B Deﬁnitions 13
Chapter 41
Topologies and Measures I
I begin this volume with an introduction to some of the most important ways in which topologies and
measures can interact, and with a description of the forms which such constructions as subspaces and product
spaces take in such contexts. By far the most important concept is that of Radon measure (411H, ¸416). In
Radon measure spaces we ﬁnd both the richest combinations of ideas and the most important applications.
But, as usual, we are led both by analysis of these ideas and by other interesting examples to consider
wider classes of topological measure space, and the greater part of the chapter, by volume, is taken up by a
description of the many properties of Radon measures individually and in partial combinations.
I begin the chapter with a short section of deﬁnitions (¸411), including a handful of more or less elementary
examples. The two central properties of a Radon measure are ‘inner regularity’ (411B) and ‘τadditivity’
(411C). The former is an idea of great versatility which I look at in an abstract setting in ¸412. I take a
section (¸413) to describe some methods of constructing measure spaces, extending the rather limited range
of constructions oﬀered in earlier volumes. There are two sections on τadditive measures, ¸¸414 and 417;
the former covers the elementary ideas, and the latter looks at product measures, where it turns out that
we need a new technique to supplement the purely measuretheoretic constructions of Chapter 25. On the
way to Radon measures in ¸416, I pause over ‘quasiRadon’ measures (411H, ¸415), where inner regularity
and τadditivity ﬁrst come eﬀectively together.
The possible interactions of a topology and a measure on the same space are so varied that even a brief
account makes a long chapter; and this is with hardly any mention of results associated with particular types
of topological space, most of which must wait for later chapters. But I include one section on the two most
important classes of functions acting between topological measure spaces (¸418), and another describing
some examples to demonstrate special phenomena (¸419).
411 Deﬁnitions
In something of the spirit of ¸211, but this time without apologising, I start this volume with a list of
deﬁnitions. The rest of Chapter 41 will be devoted to discussing these deﬁnitions and relationships between
them, and integrating the new ideas into the concepts and constructions of earlier volumes; I hope that by
presenting the terminology now I can give you a sense of the directions the following sections will take. I
ought to remark immediately that there are many cases in which the exact phrasing of the deﬁnitions is
important in ways which may not be immediately apparent.
411A I begin with a phrase which will be a useful shorthand for the context in which most, but not all,
of the theory here will be developed.
Deﬁnition A topological measure space is a quadruple (X, T, Σ, µ) where (X, Σ, µ) is a measure space
and T is a topology on X such that T ⊆ Σ, that is, every open set (and therefore every Borel set) is
measurable.
411B Now I come to what are in my view the two most important concepts to master; jointly they will
dominate the chapter.
Deﬁnition Let (X, Σ, µ) be a measure space and / a family of sets. I say that µ is inner regular with
respect to / if
µE = sup¦µK : K ∈ Σ ∩ /, K ⊆ E¦
for every E ∈ Σ. (Cf. 256Ac, 342Aa.)
Remark Note that in this deﬁnition I do not assume that / ⊆ Σ, nor even that / ⊆ TX. But of course µ
will be inner regular with respect to / iﬀ it is inner regular with respect to / ∩ Σ.
It is convenient in this context to interpret sup ∅ as 0, so that we have to check the deﬁnition only when
µE > 0, and need not insist that ∅ ∈ /.
14 Topologies and measures I 411C
411C Deﬁnition Let (X, Σ, µ) be a measure space and T a topology on X. I say that µ is τadditive
(the phrase τregular has also been used) if whenever ( is a nonempty upwardsdirected family of open
sets such that ( ⊆ Σ and
( ∈ Σ then µ(
() = sup
G∈Ç
µG.
Remark Note that in this deﬁnition I do not assume that every open set is measurable. Consequently we
cannot take it for granted that an extension of a τadditive measure will be τadditive; on the other hand,
the restriction of a τadditive measure to any σsubalgebra will be τadditive.
411D Complementary to 411B we have the following.
Deﬁnition Let (X, Σ, µ) be a measure space and H a family of subsets of X. Then µ is outer regular
with respect to H if
µE = inf¦µH : H ∈ Σ ∩ H, H ⊇ E¦
for every E ∈ Σ.
411E I delay discussion of most of the relationships between the concepts here to later in the chapter.
But it will be useful to have a basic fact set out immediately.
Proposition Let (X, Σ, µ) be a measure space and T a topology on X. If µ is inner regular with respect to
the compact sets, it is τadditive.
proof Let ( be a nonempty upwardsdirected family of measurable open sets such that H =
( ∈ Σ. If
γ < µH, there is a compact set K ⊆ H such that µK ≥ γ; now there must be a G ∈ ( which includes K,
so that µG ≥ γ. As γ is arbitrary, sup
G∈Ç
µG = µH.
411F In order to deal eﬃciently with measures which are not totally ﬁnite, I think we need the following
ideas.
Deﬁnitions Let (X, Σ, µ) be a measure space and T a topology on X.
(a) I say that µ is locally ﬁnite if every point of X has a neighbourhood of ﬁnite measure, that is, if
the open sets of ﬁnite outer measure cover X.
(b) I say that µ is eﬀectively locally ﬁnite if for every nonnegligible measurable set E ⊆ X there is
a measurable open set G ⊆ X such that µG < ∞ and E ∩ G is not negligible.
Note that an eﬀectively locally ﬁnite measure must measure many open sets, while a locally ﬁnite measure
need not.
(c) This seems a convenient moment at which to introduce the following term. A realvalued function f
deﬁned on a subset of X is locally integrable if for every x ∈ X there is an open set G containing x such
that
_
G
f is deﬁned (in the sense of 214D) and ﬁnite.
411G Elementary facts (a) If µ is a locally ﬁnite measure on a topological space X, then µ
∗
K < ∞
for every compact set K ⊆ X. PPP The family ( of open sets of ﬁnite outer measure is upwardsdirected and
covers X, so there must be some G ∈ ( including K, in which case µ
∗
K ≤ µ
∗
G is ﬁnite. QQQ
(b) A measure µ on R
r
is locally ﬁnite iﬀ every bounded set has ﬁnite outer measure (cf. 256Ab). PPP (i)
If every bounded set has ﬁnite outer measure then, in particular, every open ball has ﬁnite outer measure,
so that µ is locally ﬁnite. (ii) If µ is locally ﬁnite and A ⊆ R
r
is bounded, then its closure A is compact
(2A2F), so that µ
∗
A ≤ µ
∗
A is ﬁnite, by (a) above. QQQ
(c) I should perhaps remark immediately that a locally ﬁnite topological measure need not be eﬀectively
locally ﬁnite (419A), and an eﬀectively locally ﬁnite measure need not be locally ﬁnite (411P).
(d) An eﬀectively locally ﬁnite measure must be semiﬁnite.
411K Deﬁnitions 15
(e) A locally ﬁnite measure on a Lindel¨of space X (deﬁnition: 4A2A) is σﬁnite. PPP Let ( be the family
of open sets of ﬁnite outer measure. Because µ is locally ﬁnite, ( is a cover of X. Because X is Lindel¨of,
there is a sequence ¸G
n
)
n∈N
in ( covering X. For each n ∈ N, there is a measurable set E
n
⊇ G
n
of ﬁnite
measure, and now ¸E
n
)
n∈N
is a sequence of sets of ﬁnite measure covering X. QQQ
(f ) Let (X, T, Σ, µ) be a topological measure space such that µ is locally ﬁnite and inner regular with
respect to the compact sets. Then µ is eﬀectively locally ﬁnite. PPP Suppose that µE > 0. Then there is a
measurable compact set K ⊆ E such that µK > 0. As in the argument for (a) above, there is an open set
G of ﬁnite measure including K, so that µ(E ∩ G) > 0. QQQ
(g) Corresponding to (a) above, we have the following fact. If µ is a measure on a topological space and
f ∈ L
0
(µ) is locally integrable, then
_
K
fdµ is ﬁnite for every compact K ⊆ X, because K can be covered
by a ﬁnite family of open sets G such that
_
G
[f[dµ < ∞.
(h) If µ is a locally ﬁnite measure on a topological space X, and f ∈ L
p
(µ) for some p ∈ [1, ∞], then f is
locally integrable; this is because
_
G
[f[ ≤
_
E
f ≤ f
p
χE
q
is ﬁnite whenever G ⊆ E and µE < ∞, where
1
p
+
1
q
= 1, by H¨older’s inequality (244Eb).
(i) If (X, T) is a completely regular space and µ is a locally ﬁnite topological measure on X, then the
set of open sets with negligible boundaries is a base for T. PPP If x ∈ G ∈ T, let H ⊆ G be an open set of
ﬁnite measure containing x, and f : X → [0, 1] a continuous function such that f(x) = 1 and f(y) = 0 for
y ∈ X ¸ H. Then ¦f
−1
[¦α¦] : 0 < α < 1¦ is an uncountable disjoint family of measurable subsets of H, so
there must be some α ∈ ]0, 1[ such that f
−1
[¦α¦] is negligible. Set U = ¦y : f(y) > α¦; then U is an open
neighbourhood of x included in G and ∂U ⊆ f
−1
[¦α¦] is negligible. QQQ
411H Two particularly important combinations of the properties above are the following.
Deﬁnitions (a) A quasiRadon measure space is a topological measure space (X, T, Σ, µ) such that (i)
(X, Σ, µ) is complete and locally determined (ii) µ is τadditive, inner regular with respect to the closed sets
and eﬀectively locally ﬁnite.
(b) A Radon measure space is a topological measure space (X, T, Σ, µ) such that (i) (X, Σ, µ) is
complete and locally determined (ii) T is Hausdorﬀ (iii) µ is locally ﬁnite and inner regular with respect to
the compact sets.
411I Remarks(a) You may like to seek your own proof that a Radon measure space is always quasi
Radon, before looking it up in ¸416 below.
(b) Note that a measure on Euclidean space R
r
is a Radon measure on the deﬁnition above iﬀ it is a
Radon measure as described in 256Ad. PPP In 256Ad, I said that a measure µ on R
r
is ‘Radon’ if it is a
locally ﬁnite complete topological measure, inner regular with respect to the compact sets. (The deﬁnition
of ‘locally ﬁnite’ in 256A was not the same as the one above, but I have already covered this point in 411Gb.)
So the only thing to add is that µ is necessarily locally determined, because it is σﬁnite (256Ba). QQQ
411J The following special types of inner regularity are of suﬃcient importance to have earned separate
names.
Deﬁnitions (a) If (X, T) is a topological space, I will say that a measure µ on X is tight if it is inner
regular with respect to the closed compact sets.
(b) If (X, T, Σ, µ) is a topological measure space, I will say that µ is completion regular if it is inner
regular with respect to the zero sets (deﬁnition: 3A3Pa).
411K Borel and Baire measures If (X, T) is a topological space, I will call a measure with domain
(exactly) the Borel σalgebra of X (4A3A) a Borel measure on X, and a measure with domain (exactly)
the Baire σalgebra of X (4A3K) a Baire measure on X.
Of course a Borel measure is a topological measure in the sense of 411A. On a metric space, the Borel
and Baire measures coincide (4A3Kb). The most important measures in this chapter will be c.l.d. versions
of Borel measures.
16 Topologies and measures I 411L
411L When we come to look at functions deﬁned on a topological measure space, we shall have to relate
ideas of continuity and measurability. Two basic concepts are the following.
Deﬁnition Let X be a set, Σ a σalgebra of subsets of X and (Y, S) a topological space. I will say that a
function f : X → Y is measurable if f
−1
[G] ∈ Σ for every open set G ⊆ Y .
Remarks (a) Note that a function f : X → R is measurable on this deﬁnition (when R is given its usual
topology) iﬀ it is measurable according to the familiar deﬁnition in 121C, which asks only that sets of the
form ¦x : f(x) < α¦ should be measurable (121Ef).
(b) For any topological space (Y, S), a function f : X → Y is measurable iﬀ f is (Σ, B(Y ))measurable,
where B(Y ) is the Borel σalgebra of Y (4A3Cb).
411M Deﬁnition Let (X, Σ, µ) be a measure space, T a topology on X, and (Y, S) another topological
space. I will say that a function f : X → Y is almost continuous or Lusin measurable if µ is inner
regular with respect to the family of subsets A of X such that fA is continuous.
411N Finally, I introduce some terminology to describe ways in which (sometimes) measures can be
located in one part of a topological space rather than another.
Deﬁnitions Let (X, Σ, µ) be a measure space and T a topology on X.
(a) I will call a set A ⊆ X selfsupporting if µ
∗
(A ∩ G) > 0 for every open set G such that A ∩ G is
nonempty. (Such sets are sometimes called of positive measure everywhere.)
(b) A support of µ is a closed selfsupporting set F such that X ¸ F is negligible.
(c) Note that µ can have at most one support. PPP If F
1
, F
2
are supports then µ
∗
(F
1
¸F
2
) ≤ µ
∗
(X¸F
2
) = 0
so F
1
¸ F
2
must be empty. Similarly, F
2
¸ F
1
= ∅, so F
1
= F
2
. QQQ
(d) If µ is a τadditive topological measure it has a support. PPP Let ( be the family of negligible open
sets, and F the closed set X ¸
(. Then ( is an upwardsdirected family in T ∩ Σ and
( ∈ T ∩ Σ, so
µ(X ¸ F) = µ(
() = sup
G∈Ç
µG = 0.
If G is open and G∩ F ,= ∅ then G / ∈ ( so µ
∗
(G∩ F) = µ(G∩ F) = µG > 0; thus F is selfsupporting and
is the support of µ. QQQ
(e) Let X and Y be topological spaces with topological measures µ, ν respectively and a continuous
inversemeasurepreserving function f : X → Y . Suppose that µ has a support E. Then f[E] is the support
of ν. PPP We have only to observe that for an open set H ⊆ Y
νH > 0 ⇐⇒ µf
−1
[H] > 0 ⇐⇒ f
−1
[H] ∩ E ,= ∅
⇐⇒ H ∩ f[E] ,= ∅ ⇐⇒ H ∩ f[E] ,= ∅. QQQ
(f ) µ is strictly positive (with respect to T) if µ
∗
G > 0 for every nonempty open set G ⊆ X, that is,
X itself is the support of µ.
*(g) If (X, T) is a topological space, and µ is a strictly positive σﬁnite measure on X such that the domain
Σ of µ includes a πbase  for T, then X is ccc. PPPLet ¸E
n
)
n∈N
be a sequence of sets of ﬁnite measure covering
X. Let ( be a disjoint family of nonempty open sets. For each G ∈ G, take U
G
∈  ¸¦∅¦ such that U
G
⊆ G;
then µU
G
> 0, so there is an n(G) such that µ(E
n(G)
∩ U
G
) > 0. Now
G∈G,n(G)=k
µ(E
k
∩ U
G
) ≤ µE
k
is
ﬁnite for every k, so ¦G : n(G) = k¦ must be countable and ( is countable. QQQ
411O Example Lebesgue measure on R
r
is a Radon measure (256Ha); in particular, it is locally ﬁnite
and tight. It is therefore τadditive and eﬀectively locally ﬁnite (411E, 411Gf). It is completion regular
(because every compact set is a zero set, see 4A2Lc), outer regular with respect to the open sets (134F) and
strictly positive.
411Q Deﬁnitions 17
411P Example: Stone spaces (a) Let (Z, T, Σ, µ) be the Stone space of a semiﬁnite measure algebra
(A, ¯ µ), so that (Z, T) is a zerodimensional compact Hausdorﬀ space, (Z, Σ, µ) is complete and semiﬁnite,
the openandclosed sets are measurable, the negligible sets are the nowhere dense sets, and every measurable
set diﬀers by a nowhere dense set from an openandclosed set (311I, 321K, 322Bd, 322Qa).
(b) µ is inner regular with respect to the openandclosed sets (322Qa); in particular, it is completion
regular and tight. Consequently it is τadditive (411E).
(c) µ is strictly positive, because the openandclosed sets form a base for T (311I) and a nonempty open
andclosed set has nonzero measure. µ is eﬀectively locally ﬁnite. PPP Suppose that E ∈ Σ is not negligible.
There is a measurable set F ⊆ E such that 0 < µF < ∞; now there is a nonempty openandclosed set G
included in F, in which case µG < ∞ and µ(E ∩ G) > 0. QQQ
(d) The following are equiveridical, that is, if one is true so are the others:
(i) (A, ¯ µ) is localizable;
(ii) µ is strictly localizable;
(iii) µ is locally determined;
(iv) µ is a quasiRadon measure.
PPP The equivalence of (i)(iii) is Theorem 322N. (iv)⇒(iii) is trivial. If one, therefore all, of (i)(iii) are true,
then µ is a topological measure, because if G ⊆ Z is open, then G is openandclosed, by 314S, therefore
measurable, and G ¸ G is nowhere dense, therefore also measurable. We know already that µ is complete,
eﬀectively locally ﬁnite and τadditive, so that if it is also locally determined it is a quasiRadon measure.
QQQ
(e) The following are equiveridical:
(i) µ is a Radon measure;
(ii) µ is totally ﬁnite;
(iii) µ is locally ﬁnite;
(iv) µ is outer regular with respect to the open sets.
PPP (ii)⇒(iv) If µ is totally ﬁnite and E ∈ Σ, then for any > 0 there is a closed set F ⊆ Z ¸ E such that
µF ≥ µ(Z ¸ E) −, and now G = Z ¸ F is an open set including E with µG ≤ µE + . (iv)⇒(iii) Suppose
that µ is outer regular with respect to the open sets, and z ∈ Z. Because Z is Hausdorﬀ, ¦z¦ is closed. If it
is open it is measurable, and because µ is semiﬁnite it must have ﬁnite measure. Otherwise it is nowhere
dense, therefore negligible, and must be included in open sets of arbitrarily small measure. Thus in both
cases z belongs to an open set of ﬁnite measure; as z is arbitrary, µ is locally ﬁnite. (iii)⇒(ii) Becasue Z
is compact, this is a consequence of 411Ga. (i)⇒(iii) is part of the deﬁnition of ‘Radon measure’. Finally,
(ii)+(iii)⇒(i), again directly from the deﬁnition and the facts set out in (a)(b) above. QQQ
411Q Example: Dieudonn´e’s measure Recall that a set E ⊆ ω
1
is a Borel set iﬀ either E or its
complement includes a coﬁnal closed set (4A3J). So we may deﬁne a Borel measure µ on ω
1
by saying that
µE = 1 if E includes a coﬁnal closed set and µE = 0 if E is disjoint from a coﬁnal closed set. If E is disjoint
from some coﬁnal closed set, so is any subset of E, so µ is complete. Since µ takes only the values 0 and 1,
it is a purely atomic probability measure.
µ is a topological measure; being totally ﬁnite, it is surely locally ﬁnite and eﬀectively locally ﬁnite. It
is inner regular with respect to the closed sets (because if µE > 0, there is a coﬁnal closed set F ⊆ E,
and now F is a closed set with µF = µE), therefore outer regular with respect to the open sets. It is not
τadditive (because ξ = [0, ξ[ is an open set of zero measure for every ξ < ω
1
, and the union of these sets is
a measurable open set of measure 1).
µ is not completion regular, because the set of countable limit ordinals is a closed set (4A1Bb) which
does not include any uncountable zero set (see 411Ra below).
The only selfsupporting subset of ω
1
is the empty set (because there is a cover of ω
1
by negligible open
sets). In particular, µ does not have a support.
Remark There is a measure of this type on any ordinal of uncountable coﬁnality; see 411Xj.
18 Topologies and measures I 411R
411R Example: The Baire σalgebra of ω
1
The Baire σalgebra Ba(ω
1
) of ω
1
is the countable
cocountable algebra (4A3P). The countablecocountable measure µ on ω
1
is therefore a Baire measure on
the deﬁnition of 411K. Since all sets of the form ]ξ, ω
1
[ are zero sets, µ is inner regular with respect to the
zero sets and outer regular with respect to the cozero sets. Since sets of the form [0, ξ[ (= ξ) form a cover
of ω
1
by measurable open sets of zero measure, µ is not τadditive.
411X Basic exercises (a) Let (X, Σ, µ) be a totally ﬁnite measure space and T a topology on X. Show
that µ is inner regular with respect to the closed sets iﬀ it is outer regular with respect to the open sets,
and is inner regular with respect to the zero sets iﬀ it is outer regular with respect to the cozero sets.
(b) Let µ be a Radon measure on R
r
, where r ≥ 1, and f ∈ L
0
(µ). Show that f is locally integrable in
the sense of 411Fc iﬀ it is locally integrable in the sense of 256E, that is,
_
E
fdν < ∞ for every bounded set
E ⊆ R
r
.
(c) Let µ be a measure on a topological space, ˆ µ its completion and ˜ µ its c.l.d. version. Show that µ is
locally ﬁnite iﬀ ˆ µ is locally ﬁnite, and in this case ˜ µ is locally ﬁnite.
>>>(d) Let µ be an eﬀectively locally ﬁnite measure on a topological space X. (i) Show that the completion
and c.l.d. version of µ are eﬀectively locally ﬁnite. (ii) Show that if µ is complete and locally determined, then
the union of the measurable open sets of ﬁnite measure is conegligible. (iii) Show that if X is hereditarily
Lindel¨of then µ must be σﬁnite.
(e) Let X be a topological space and µ a measure on X. Let U ⊆ L
0
(µ) be the set of equivalence classes
of locally integrable functions in L
0
(µ). Show that U is a solid linear subspace of L
0
(µ). Show that if µ is
locally ﬁnite then U includes L
p
(µ) for every p ∈ [0, ∞].
(f ) Let X be a topological space. (i) Let µ, ν be two totally ﬁnite Borel measures which agree on the
closed sets. Show that they are equal. (Hint: 136C.) (ii) Let µ, ν be two totally ﬁnite Baire measures which
agree on the zero sets. Show that they are equal.
(g) Let (X, T) be a topological space, µ a measure on X, and Y a subset of X; let T
Y
, µ
Y
be the subspace
topology and measure. Show that if µ is a topological measure, or locally ﬁnite, or a Borel measure, so is
µ
Y
.
(h) Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of measure spaces, with direct sum (X, Σ, µ); suppose that we are
given a topology T
i
on each X
i
, and let T be the disjoint union topology on X. Show that µ is a topological
measure, or locally ﬁnite, or eﬀectively locally ﬁnite, or a Borel measure, or a Baire measure, or strictly
positive, iﬀ every µ
i
is.
(i) Let (X, Σ, µ) and (Y, T, ν) be two measure spaces, with c.l.d. product measure λ on X Y . Suppose
we are given topologies T, S on X, Y respectively, and give X Y the product topology. Show that λ is
locally ﬁnite, or eﬀectively locally ﬁnite, if µ and ν are.
(j) Let κ be any cardinal of uncountable coﬁnality (deﬁnition: 3A1Fb). Show that there is a complete
topological probability measure µ on κ deﬁned by saying that µE = 1 if E includes a coﬁnal closed set in κ,
0 if E is disjoint from some coﬁnal closed set. Show that µ is inner regular with respect to the closed sets
but is not completion regular.
411Y Further exercises (a) Show that a function f : R
r
→R
s
is measurable iﬀ it is almost continuous
(where R
r
is endowed with Lebesgue measure and its usual topology, of course). (Hint: 256F.)
(b) Let (X, ρ) be a metric space, r ≥ 0, and write µ
Hr
for rdimensional Hausdorﬀ measure on X (264K,
¸471). (i) Show that µ
Hr
is a topological measure, outer regular with respect to the Borel sets. (ii) Show
that if X is complete then the c.l.d. version of µ
Hr
is tight, therefore completion regular.
412A Inner regularity 19
(c) Let (X, T, Σ, µ) be a topological measure space. Set c = ¦E : E ⊆ X, µ(∂E) = 0¦, where ∂E is
the boundary of A. (i) Show that c is a subalgebra of TX, and that every member of c is measured by
the completion of µ. (c is sometimes called the Jordan algebra of (X, T, Σ, µ). Do not confuse with the
‘Jordan algebras’ of abstract algebra.) (ii) Suppose that µ is complete and totally ﬁnite and inner regular
with respect to the closed sets, and that T is normal. Show that ¦E
•
: E ∈ c¦ is dense in the measure
algebra of µ endowed with its usual topology. (Hint: if f : X → R is continuous, then ¦x : f(x) ≤ α¦ ∈ c
for all but countably many α.) (iii) Suppose that µ is a quasiRadon measure and T is completely regular.
Show that ¦E
•
: E ∈ c¦ is dense in the measure algebra of µ. (Hint: 414Aa.)
411 Notes and comments Of course the list above can give only a rough idea of the ways in which
topologies and measures can interact. In particular I have rather arbitrarily given a sort of priority to three
particular relationships between the domain Σ of a measure and the topology: ‘topological measure space’
(in which Σ includes the Borel σalgebra), ‘Borel measure’ (in which Σ is precisely the Borel σalgebra) and
‘Baire measure’ (in which Σ is the Baire σalgebra).
Abstract topological measure theory is a relatively new subject, and there are many technical questions
on which diﬀerent authors take diﬀerent views. For instance, the phrase ‘Radon measure’ is commonly used
to mean what I would call a ‘tight locally ﬁnite Borel measure’ (cf. 416F); and some writers enlarge the
deﬁnition of ‘topological measure’ to include Baire measures as deﬁned above.
I give very few examples at this stage, two drawn from the constructions of Volumes 13 (Lebesgue
measure and Stone spaces, 411O411P) and one new one (‘Dieudonn´e’s measure’, 411Q), with a glance at
the countablecocountable measure of ω
1
(411R). The most glaring omission is that of the product measures
on ¦0, 1¦
I
and [0, 1]
I
. I pass these by at the moment because a proper study of them requires rather more
preparation than can be slipped into a parenthesis. (I return to them in 416U.) I have also omitted any
discussion of ‘measurable’ and ‘almost continuous’ functions, except for a reference to a theorem in Volume 2
(411Ya), which will have to be repeated and ampliﬁed later on (¸418). There is an obvious complementarity
between the notions of ‘inner’ and ‘outer’ regularity (411B, 411D), but it works well only for totally ﬁnite
spaces (411Xa); in other cases it may not be obvious what will happen (411O, 411Pe, 412W).
412 Inner regularity
As will become apparent as the chapter progresses, the concepts introduced in ¸411 are synergic; their
most interesting manifestations are in combinations of various kinds. Any linear account of their properties
will be more than usually like a spaceﬁlling curve. But I have to start somewhere, and enough results can
be expressed in terms of inner regularity, more or less by itself, to be a useful beginning.
After a handful of elementary basic facts (412A) and a list of standard applications (412B), I give some
useful suﬃcient conditions for inner regularity of topological and Baire measures (412D, 412E, 412G), based
on an important general construction (412C). The rest of the section amounts to a review of ideas from Vol
ume 2 and Chapter 32 in the light of the new concept here. I touch on completions (412H), c.l.d. versions and
complete locally determined spaces (412H, 412J, 412L), strictly localizable spaces (412I), inversemeasure
preserving functions (412K, 412M), measure algebras (412N), subspaces (412O, 412P), indeﬁniteintegral
measures (412Q) and product measures (412R412V), with a brief mention of outer regularity (412W); most
of the hard work has already been done in Chapters 21 and 25.
412A I begin by repeating a lemma from Chapter 34, with some further straightforward facts.
Lemma (a) Let (X, Σ, µ) be a measure space and / a family of sets such that
whenever E ∈ Σ and µE > 0 there is a K ∈ / ∩ Σ such that K ⊆ E and µK > 0.
Then whenever E ∈ Σ there is a countable disjoint family ¸K
i
)
i∈I
in / ∩ Σ such that K
i
⊆ E for every i
and
i∈I
µK
i
= µE. If moreover
(†) K ∪ K
t
∈ / whenever K, K
t
are disjoint members of /,
then µ is inner regular with respect to /. If
i∈I
K
i
∈ / for every countable disjoint family ¸K
i
)
i∈I
in /,
then for every E ∈ Σ there is a K ∈ / ∩ Σ such that K ⊆ E and µK = µE.
20 Topologies and measures 412A
(b) Let (X, Σ, µ) be a measure space, T a σsubalgebra of Σ, and / a family of sets. If µ is inner regular
with respect to T and µT is inner regular with respect to /, then µ is inner regular with respect to /.
(c) Let (X, Σ, µ) be a semiﬁnite measure space and ¸/
n
)
n∈N
a sequence of families of sets such that µ is
inner regular with respect to /
n
and
(‡) if ¸K
i
)
i∈N
is a nonincreasing sequence in /
n
, then
i∈N
K
i
∈ /
n
for every n ∈ N. Then µ is inner regular with respect to
n∈N
/
n
.
proof (a) This is 342B342C.
(b) If E ∈ Σ and γ < µE, there are an F ∈ T such that F ⊆ E and µF > γ, and a K ∈ /∩T such that
K ⊆ F and µK ≥ γ.
(c) Suppose that E ∈ Σ and that 0 ≤ γ < µE. Because µ is semiﬁnite, there is an F ∈ Σ such that
F ⊆ E and γ < µF < ∞ (213A). Choose ¸K
i
)
i∈N
inductively, as follows. Start with K
0
= F. Given that
K
i
∈ Σ and γ < µK
i
, then let n
i
∈ N be such that 2
−n
i
(i + 1) is an odd integer, and choose K
i+1
∈ /
n
i
such that K
i+1
⊆ K
i
and µK
i+1
> γ; this will be possible because µ is inner regular with respect to /
n
i
.
Consider K =
i∈N
K
i
. Then K ⊆ E and µK = lim
i→∞
µK
i
≥ γ. But also
K =
j∈N
K
2
n
(2j+1)
∈ /
n
because ¸K
2
n
(2j+1)
)
j∈N
is a nonincreasing sequence in /
n
, for each n. So K ∈
n∈N
/
n
. As E and γ are
arbitrary, µ is inner regular with respect to
n∈N
/
n
.
412B Corollary Let (X, Σ, µ) be a measure space and T a topology on X. Suppose that / is
either the family of Borel subsets of X
or the family of closed subsets of X
or the family of compact subsets of X
or the family of zero sets in X,
and suppose that whenever E ∈ Σ and µE > 0 there is a K ∈ / ∩ Σ such that K ⊆ E and µK > 0. Then
µ is inner regular with respect to /.
proof In every case, / satisﬁes the condition (†) of 412Aa.
412C The next lemma provides a particularly useful method of proving that measures are inner regular
with respect to ‘wellbehaved’ families of sets.
Lemma Let (X, Σ, µ) be a semiﬁnite measure space, and suppose that / ⊆ Σ and / are such that
∅ ∈ / ⊆ Σ,
(†) K ∪ K
t
∈ / whenever K, K
t
∈ /,
(‡)
n∈N
K
n
∈ / for every sequence ¸K
n
)
n∈N
in /,
X ¸ A ∈ / for every A ∈ /,
whenever A ∈ /, F ∈ Σ and µ(A ∩ F) > 0, there is a K ∈ / ∩ / such that K ⊆ A and
µ(K ∩ F) > 0.
Let T be the σsubalgebra of Σ generated by /. Then µT is inner regular with respect to /.
proof (a) Write A for the measure algebra of (X, Σ, µ), and L = /∩T, so that L is also closed under ﬁnite
unions and countable intersections. Set
H = ¦E : E ∈ Σ, sup
L∈1,L⊆E
L
•
= E
•
¦ in A,
T
t
= ¦E : E ∈ H, X ¸ E ∈ H¦,
so that the last two conditions tell us that / ⊆ T
t
.
(b) The intersection of any sequence in H belongs to H. PPP Let ¸H
n
)
n∈N
be a sequence in H with
intersection H. Write A
n
for ¦L
•
: L ∈ L, L ⊆ H
n
¦ ⊆ A for each n ∈ N. Since A is weakly (σ, ∞)
distributive (322F), A
n
is upwardsdirected, and sup A
n
= H
•
n
for each n ∈ N,
412F Inner regularity 21
H
•
= inf
n∈N
H
•
n
(because F → F
•
: Σ → A is sequentially ordercontinuous, by 321H)
= inf
n∈N
sup A
n
= sup¦ inf
n∈N
a
n
: a
n
∈ A
n
for every n ∈ N¦
(316J)
= sup¦(
n∈N
L
n
)
•
: L
n
∈ L, L
n
⊆ H
n
for every n ∈ N¦
⊆ ¦L
•
: L ∈ L, L ⊆ H¦
(by (‡))
⊆ H
•
,
and H ∈ H. QQQ
(c) The union of any sequence in H belongs to H. PPP If ¸H
n
)
n∈N
is a sequence in H with union H then
sup
L∈1,L⊆H
L
•
⊇ sup
n∈N
sup
L∈1,L⊆E
n
L
•
= sup
n∈N
H
•
n
= H
•
,
so H ∈ H. QQQ
(d) T
t
is a σsubalgebra of Σ. PPP (i) ∅ and X belong to / ⊆ H, so ∅ ∈ T
t
. (ii) Obviously X ¸ E ∈ T
t
whenever E ∈ T
t
. (iii) If ¸E
n
)
n∈N
is a sequence in T
t
with union E then E ∈ H, by (c); but also
X ¸ E =
n∈N
(X ¸ E
n
) belongs to H, by (b). So E ∈ T
t
. QQQ
(e) Accordingly T ⊆ T
t
, and E
•
= sup
L∈1,L⊆E
L
•
for every E ∈ T. It follows at once that if E ∈ T and
µE > 0, there must be an L ∈ L such that L ⊆ E and µL > 0; since (†) is true, and L ⊆ T, we can apply
412Aa to see that µT is inner regular with respect to L, therefore with respect to /.
412D As corollaries of the last lemma I give twoandahalf basic theorems.
Theorem Let (X, T) be a topological space and µ a semiﬁnite Baire measure on X. Then µ is inner regular
with respect to the zero sets.
proof Write Σ for the Baire σalgebra of X, the domain of µ, / for the family of zero sets, and / for
/ ∪ ¦X ¸ K : K ∈ /¦. Since the union of two zero sets is a zero set (4A2C(bii)), the intersection of a
sequence of zero sets is a zero set (4A2C(biii)), and the complement of a zero set is the union of a sequence
of zero sets (4A2C(bvi)), the conditions of 412C are satisﬁed; and as the σalgebra generated by / is just
Σ, µ is inner regular with respect to /.
412E Theorem Let (X, T) be a perfectly normal topological space (e.g., any metrizable space). Then
any semiﬁnite Borel measure on X is inner regular with respect to the closed sets.
proof Because the Baire and Borel σalgebras are the same (4A3Kb), this is a special case of 412D.
412F Lemma Let (X, Σ, µ) be a measure space and T a topology on X such that µ is eﬀectively locally
ﬁnite with respect to T. Then
µE = sup¦µ(E ∩ G) : G is a measurable open set of ﬁnite measure¦
for every E ∈ Σ.
proof Apply 412Aa with / the family of subsets of measurable open sets of ﬁnite measure.
22 Topologies and measures 412G
412G Theorem Let (X, Σ, µ) be a measure space with a topology T such that µ is eﬀectively locally
ﬁnite with respect to T and Σ is the σalgebra generated by T ∩ Σ. If
µG = sup¦µF : F ∈ Σ is closed, F ⊆ G¦
for every measurable open set G of ﬁnite measure, then µ is inner regular with respect to the closed sets.
proof In 412C, take / to be the family of measurable closed subsets of X, and / to be the family of
measurable sets which are either open or closed. If G ∈ Σ ∩ T, F ∈ Σ and µ(G ∩ F) > 0, then there is an
open set H of ﬁnite measure such that µ(H ∩ G∩ F) > 0, because µ is eﬀectively locally ﬁnite; now there
is a K ∈ / such that K ⊆ H ∩ G and µK > µ(H ∩ G) −µ(H ∩ G∩ F), so that µ(K ∩ F) > 0. This is the
only nontrivial item in the list of hypotheses in 412C, so we can conclude that µT is inner regular with
respect to /, where T is the σalgebra generated by /; but of course this is just Σ.
Remark There is a similar result in 416F(iii) below.
412H Proposition Let (X, Σ, µ) be a measure space and / a family of sets.
(a) If µ is inner regular with respect to /, so are its completion ˆ µ (212C) and c.l.d. version ˜ µ (213E).
(b) Now suppose that
(‡)
n∈N
K
n
∈ / whenever ¸K
n
)
n∈N
is a nonincreasing sequence in /.
If either ˆ µ is inner regular with respect to / or µ is semiﬁnite and ˜ µ is inner regular with respect to /,
then µ is inner regular with respect to /.
proof (a) If F belongs to the domain of ˆ µ, then there is an E ∈ Σ such that E ⊆ F and ˆ µ(F ¸ E) = 0. So
if 0 ≤ γ < ˆ µF = µE, there is a K ∈ / ∩ Σ such that K ⊆ E ⊆ F and ˆ µK = µK ≥ γ.
If H belongs to the domain of ˜ µ and 0 ≤ γ < ˜ µH, there is an E ∈ Σ such that µE < ∞ and ˆ µ(E∩H) > γ
(213D). Now there is a K ∈ / ∩ Σ such that K ⊆ E ∩ H and µK ≥ γ. As µK < ∞, ˜ µK = µK ≥ γ.
(b) Write ˇ µ for whichever of ˆ µ, ˜ µ is supposed to be inner regular with respect to /. Then ˇ µ is inner
regular with respect to Σ (212Ca, 213Fc), so is inner regular with respect to /∩Σ (412Ac). Also ˇ µ extends
µ (212D, 213Hc). Take E ∈ Σ and γ < µE = ˇ µE. Then there is a K ∈ / ∩ Σ such that K ⊆ E and
γ < ˇ µK = µK. As E and γ are arbitrary, µ is inner regular with respect to /.
412I Lemma Let (X, Σ, µ) be a strictly localizable measure space and / a family of sets such that
whenever E ∈ Σ and µE > 0 there is a K ∈ / ∩ Σ such that K ⊆ E and µK > 0.
(a) There is a decomposition ¸X
i
)
i∈I
of X such that at most one X
i
does not belong to /, and that
exceptional one, if any, is negligible.
(b) There is a disjoint family L ⊆ / ∩ Σ such that µ
∗
A =
L∈1
µ
∗
(A∩ L) for every A ⊆ X.
(c) If µ is σﬁnite then the family ¸X
i
)
i∈I
of (a) and the set L of (b) can be taken to be countable.
proof (a) Let ¸E
j
)
j∈J
be any decomposition of X. For each j ∈ J, let /
j
be a maximal disjoint subset of
¦K : K ∈ / ∩ Σ, K ⊆ E
j
, µK > 0¦.
Because µE
j
< ∞, /
j
must be countable. Set E
t
j
= E
j
¸
/
j
. By the maximality of /
j
, E
t
j
cannot include
any nonnegligible set in / ∩ Σ; but this means that µE
t
j
= 0. Set X
t
=
j∈J
E
t
j
. Then
µX
t
=
j∈J
µ(X
t
∩ E
j
) =
j∈J
µE
t
j
= 0.
Note that if j, j
t
∈ J are distinct, and K ∈ /
j
, K
t
∈ /
j
, then K∩K
t
= ∅; thus L =
j∈J
/
j
is disjoint.
Let ¸X
i
)
i∈I
be any indexing of ¦X
t
¦ ∪ L. This is a partition (that is, disjoint cover) of X into sets of ﬁnite
measure. If E ⊆ X and E ∩ X
i
∈ Σ for every i ∈ I, then for every j ∈ J
E ∩ E
j
= (E ∩ X
t
∩ E
j
) ∪
K∈K
j
E ∩ K
belongs to Σ, so that E ∈ Σ and
µE =
j∈J
K∈K
j
µ(E ∩ K) =
i∈I
µ(E ∩ X
i
).
Thus ¸X
i
)
i∈I
is a decomposition of X, and it is of the right type because every X
i
but one belongs to L ⊆ /.
412K Inner regularity 23
(b) If now A ⊆ X is any set,
µ
∗
A = µ
A
A =
i∈I
µ
A
(A∩ X
i
) =
i∈I
µ
∗
(A∩ X
i
)
by 214Ia, writing µ
A
for the subspace measure on A. So we have
µ
∗
A = µ
∗
(A∩ X
t
) +
L∈1
µ
∗
(A∩ L) =
L∈1
µ
∗
(A∩ L),
while L ⊆ / is disjoint.
(c) If µ is σﬁnite we can take J to be countable, so that I and L will also be countable.
412J Proposition Let (X, Σ, µ) be a complete locally determined measure space, and / a family of sets
such that µ is inner regular with respect to /.
(a) If E ⊆ X is such that E ∩ K ∈ Σ for every K ∈ / ∩ Σ, then E ∈ Σ.
(b) If E ⊆ X is such that E ∩ K is negligible for every K ∈ / ∩ Σ, then E is negligible.
(c) For any A ⊆ X, µ
∗
A = sup
K∈K∩Σ
µ
∗
(A∩ K).
(d) Let f be a nonnegative [0, ∞]valued function deﬁned on a subset of X. If
_
K
f is deﬁned in [0, ∞]
for every K ∈ /, then
_
f is deﬁned and equal to sup
K∈K
_
K
f.
(e) If f is a µintegrable function and > 0, there is a K ∈ / such that
_
X\K
[f[ ≤ .
Remark In (c), we must interpret sup ∅ as 0 if / ∩ Σ = ∅.
proof (a) If F ∈ Σ and µF < ∞, then E ∩ F ∈ Σ. PPP If µF = 0, this is trivial, because µ is complete and
E ∩ F is negligible. Otherwise, there is a sequence ¸K
n
)
n∈N
in / ∩ Σ such that K
n
⊆ F for each n and
sup
n∈N
µK
n
= µF. Now E ∩ F ¸
n∈N
K
n
is negligible, therefore measurable, while E ∩ K
n
is measurable
for every n ∈ N, by hypothesis; so E ∩ F is measurable. QQQ As µ is locally determined, E ∈ Σ, as claimed.
(b) By (a), E ∈ Σ; and because µ is inner regular with respect to /, µE must be 0.
(c) Let µ
A
be the subspace measure on A. Because µ is complete and locally determined, µ
A
is semiﬁnite
(214Ic). So if 0 ≤ γ < µ
∗
A = µ
A
A, there is an H ⊆ A such that µ
A
H is deﬁned, ﬁnite and greater than γ.
Let E ∈ Σ be a measurable envelope of H (132Ee), so that µE = µ
∗
H > γ. Then there is a K ∈ / ∩ Σ
such that K ⊆ E and µK ≥ γ. In this case
µ
∗
(A∩ K) ≥ µ
∗
(H ∩ K) = µ(E ∩ K) = µK ≥ γ.
As γ is arbitrary,
µ
∗
A ≤ sup
K∈K∩Σ
µ
∗
(A∩ K);
but the reverse inequality is trivial, so we have the result.
(d) Applying (b) with E = X ¸ domf, we see that f is deﬁned almost everywhere in X. Applying (a)
with E = ¦x : x ∈ domf, f(x) ≥ α¦ for each α ∈ R, we see that f is measurable. So
_
f is deﬁned in [0, ∞],
and of course
_
f ≥ sup
K∈K
_
K
f. If γ <
_
f, there is a nonnegative simple function g such that g ≤
a.e.
f
and
_
g > γ; taking E = ¦x : g(x) > 0¦, there is a K ∈ / such that K ⊆ E and µ(E ¸ K)g
∞
≤ γ −
_
g,
so that
_
K
f ≥
_
K
g ≥ γ. As γ is arbitrary,
_
f = sup
K∈K
_
K
f.
(e) By (d), there is a K ∈ / such that
_
K
[f[ ≥
_
[f[ −.
Remark See also 413F below.
412K Proposition Let (X, Σ, µ) be a complete locally determined measure space, (Y, T, ν) a measure
space and f : X → Y a function. Suppose that / ⊆ T is such that
(i) ν is inner regular with respect to /;
(ii) f
−1
[K] ∈ Σ and µf
−1
[K] = νK for every K ∈ /;
(iii) whenever E ∈ Σ and µE > 0 there is a K ∈ / such that νK < ∞ and µ(E ∩ f
−1
[K]) > 0.
Then f is inversemeasurepreserving for µ and ν.
proof (a) If F ∈ T, E ∈ Σ and µE < ∞, then E∩f
−1
[F] ∈ Σ. PPP Let H
1
, H
2
∈ Σ be measurable envelopes
for E ∩ f
−1
[F] and E ¸ f
−1
[F] respectively. ??? If µ(H
1
∩ H
2
) > 0, there is a K ∈ / such that νK is ﬁnite
24 Topologies and measures 412K
and µ(H
1
∩ H
2
∩ f
−1
[K]) > 0. Because ν is inner regular with respect to /, there are K
1
, K
2
∈ / such
that K
1
⊆ K ∩ F, K
2
⊆ K ¸ F and
νK
1
+νK
2
> ν(K ∩ F) +ν(K ¸ F) −µ(H
1
∩ H
2
∩ f
−1
[K]) = νK −µ(H
1
∩ H
2
∩ f
−1
[K]).
Now
µ(H
1
∩ f
−1
[K
2
]) = µ
∗
(E ∩ f
−1
[F] ∩ f
−1
[K
2
]) = 0,
µ(H
2
∩ f
−1
[K
1
]) = µ
∗
(E ∩ f
−1
[K
1
] ¸ f
−1
[F]) = 0,
so µ(H
1
∩ H
2
∩ f
−1
[K
1
∪ K
2
]) = 0 and
µ(H
1
∩ H
2
∩ f
−1
[K]) ≤ µ(f
−1
[K] ¸ f
−1
[K
1
∪ K
2
])
= µf
−1
[K] −µf
−1
[K
1
] −µf
−1
[K
2
]
= νK −νK
1
−νK
2
< µ(H
1
∩ H
2
∩ f
−1
[K]),
which is absurd. XXX
Now (E ∩H
1
) ¸ (E ∩f
−1
[F]) ⊆ H
1
∩H
2
is negligible, therefore measurable (because µ is complete), and
E ∩ f
−1
[F] ∈ Σ, as claimed. QQQ
(b) It follows (because µ is locally determined) that f
−1
[F] ∈ Σ for every F ∈ T.
(c) If F ∈ T and νF = 0 then µf
−1
[F] = 0. PPP??? Otherwise, there is a K ∈ / such that νK < ∞ and
0 < µ(f
−1
[F] ∩ f
−1
[K]) = µf
−1
[F ∩ K].
Let K
t
∈ / be such that K
t
⊆ K ¸ F and νK
t
> νK −µf
−1
[F ∩ K]. Then f
−1
[K
t
] ∩ f
−1
[F ∩ K] = ∅, so
νK = µf
−1
[K] ≥ µf
−1
[K
t
] +µf
−1
[F ∩ K] > νK
t
+νK −νK
t
= νK,
which is absurd. XXXQQQ
(d) Finally, µf
−1
[F] = νF for every F ∈ T. PPP Let ¸K
i
)
i∈I
be a countable disjoint family in / such that
K
i
⊆ F for every i and
i∈I
νK
i
= νF (412Aa). Set F
t
= F ¸
i∈I
K
i
. Then
µf
−1
[F] = µf
−1
[F
t
] +
i∈I
µf
−1
[K
i
] = µf
−1
[F
t
] +
i∈I
νK
i
= µf
−1
[F
t
] +νF.
If νF = ∞ then surely µf
−1
[F] = ∞ = νF. Otherwise, νF
t
= 0 so µf
−1
[F
t
] = 0 (by (c)) and again
µf
−1
[F] = νF. QQQ
Thus f is inversemeasurepreserving.
412L Corollary Let X be a set and / a family of subsets of X. Suppose that µ, ν are two complete
locally determined measures on X, with domains including /, agreeing on /, and both inner regular with
respect to /. Then they are identical (and, in particular, have the same domain).
proof Apply 412K with X = Y and f the identity function to see that µ extends ν; similarly, ν extends µ
and the two measures are the same.
412M Corollary Let (X, Σ, µ) be a complete probability space, (Y, T, ν) a probability space and f :
X → Y a function. Suppose that whenever F ∈ T and νF > 0 there is a K ∈ T such that K ⊆ F, νK > 0,
f
−1
[K] ∈ Σ and µf
−1
[K] ≥ νK. Then f is inversemeasurepreserving.
proof Set /
∗
= ¦K : K ∈ T, f
−1
[K] ∈ Σ, µf
−1
[K] ≥ νK¦. Then /
∗
is closed under countable disjoint
unions and includes /, so for every F ∈ T there is a K ∈ /
∗
such that K ⊆ F and νK = νF, by 412Aa.
But this means that µf
−1
[K] = νK for every K ∈ /
∗
. PPP There is a K
t
∈ /
∗
such that K
t
⊆ Y ¸ K and
νK
t
= 1 −νK; but in this case
µf
−1
[K
t
] +µf
−1
[K] ≤ 1 = νK
t
+νK,
so µf
−1
[K] must be equal to νK. QQQ Moreover, there is a K
∗
∈ /
∗
such that νK
∗
= νY = 1, so
µf
−1
[K
∗
] = µX = 1 and µ(E ∩ f
−1
[K
∗
]) > 0 whenever µE > 0. Applying 412K to /
∗
we have the result.
412Q Inner regularity 25
412N Lemma Let (X, Σ, µ) be a measure space and / a family of subsets of X such that µ is inner
regular with respect to /. Then
E
•
= sup¦K
•
: K ∈ / ∩ Σ, K ⊆ E¦
in the measure algebra A of µ, for every E ∈ Σ. In particular, ¦K
•
: K ∈ /∩Σ¦ is orderdense in A; and if
/ is closed under ﬁnite unions, then ¦K
•
: K ∈ / ∩ Σ¦ is topologically dense in A for the measurealgebra
topology.
proof ??? If E
•
,= sup¦K
•
: K ∈ /∩ Σ, K ⊆ E¦, there is a nonzero a ∈ A such that a ⊆ E
•
\ K
•
whenever
K ∈ / ∩ Σ and K ⊆ E. Express a as F
•
where F ⊆ E. Then µF > 0, so there is a K ∈ / ∩ Σ such that
K ⊆ F and µK > 0. But in this case 0 ,= K
•
⊆ a, while K ⊆ E. XXX
It follows at once that D = ¦K
•
: K ∈ / ∩ Σ¦ is orderdense. If / is closed under ﬁnite unions, and
a ∈ A, then D
a
= ¦d : d ∈ D, d ⊆ a¦ is upwardsdirected and has supremum a, so a ∈ D
a
⊆ D (323D(aii)).
412O Lemma Let (X, Σ, µ) be a measure space and / a family of subsets of X such that µ is inner
regular with respect to /.
(a) If E ∈ Σ, then the subspace measure µ
E
(131B) is inner regular with respect to /.
(b) Let Y ⊆ X be any set such that the subspace measure µ
Y
(214A214B) is semiﬁnite. Then µ
Y
is
inner regular with respect to /
Y
= ¦K ∩ Y : K ∈ /¦.
proof (a) This is elementary.
(b) Suppose that F belongs to the domain Σ
Y
of µ
Y
and 0 ≤ γ < µ
Y
F. Because µ
Y
is semiﬁnite there
is an F
t
∈ Σ
Y
such that F
t
⊆ F and γ < µ
Y
F
t
< ∞. Let E be a measurable envelope for F
t
with respect
to µ, so that
µE = µ
∗
F
t
= µ
Y
F
t
> γ.
There is a K ∈ / ∩ Σ such that K ⊆ E and µK ≥ γ, in which case K ∩ Y ∈ /
Y
∩ Σ
Y
and
µ
Y
(K ∩ Y ) = µ
∗
(K ∩ Y ) = µ
∗
(K ∩ F
t
) = µ(K ∩ E) = µK ≥ γ.
As F and γ are arbitrary, µ
Y
is inner regular with respect to /
Y
.
Remark Recall from 214I that if (X, Σ, µ) has locally determined negligible sets (in particular, is either
strictly localizable or complete and locally determined), then all its subspaces are semiﬁnite.
412P Proposition Let (X, Σ, µ) be a measure space, T a topology on X and Y a subset of X; write T
Y
for the subspace topology of Y and µ
Y
for the subspace measure on Y . Suppose that either Y ∈ Σ or µ
Y
is semiﬁnite.
(a) If µ is a topological measure, so is µ
Y
.
(b) If µ is inner regular with respect to the Borel sets, so is µ
Y
.
(c) If µ is inner regular with respect to the closed sets, so is µ
Y
.
(d) If µ is inner regular with respect to the zero sets, so is µ
Y
.
(e) If µ is eﬀectively locally ﬁnite, so is µ
Y
.
proof (a) is an immediate consequence of the deﬁnitions of ‘subspace measure’, ‘subspace topology’ and
‘topological measure’. The other parts follow directly from 412O if we recall that
(i) a subset of Y is Borel for T
Y
whenever it is expressible as Y ∩ E for some Borel set E ⊆ X (4A3Ca);
(ii) a subset of Y is closed in Y whenever it is expressible as Y ∩ F for some closed set F ⊆ X;
(iii) a subset of Y is a zero set in Y whenever it is expressible as Y ∩ F for some zero set F ⊆ X
(4A2C(bv));
(iv) µ is eﬀectively locally ﬁnite iﬀ it is inner regular with respect to subsets of open sets of ﬁnite measure.
412Q Proposition Let (X, Σ, µ) be a measure space, and ν an indeﬁniteintegral measure over µ (deﬁ
nition: 234B). If µ is inner regular with respect to a family / of sets, so is ν.
proof Because µ and its completion ˆ µ give the same integrals, ν is an indeﬁniteintegral measure over ˆ µ;
and as ˆ µ is still inner regular with respect to / (412H), we may suppose that µ itself is complete. Let f
26 Topologies and measures 412Q
be a RadonNikod´ ym derivative of ν with respect to µ; by 234Ca, we may suppose that f : X → [0, ∞[ is
Σmeasurable.
Suppose that F ∈ dom(ν) and that γ < νF. Set G = ¦x : f(x) > 0¦, so that F ∩ G ∈ Σ (234D). For
n ∈ N, set H
n
= ¦x : x ∈ F, 2
−n
≤ f(x) ≤ 2
n
¦, so that H
n
∈ Σ and
νF =
_
f χFdµ = lim
n→∞
_
f χH
n
dµ.
Let n ∈ N be such that
_
f χH
n
dµ > γ.
If µH
n
= ∞, there is a K ∈ / such that K ⊆ H
n
and µK ≥ 2
n
γ, so that νK ≥ γ. If µH
n
is ﬁnite, there
is a K ∈ / such that 2
n
(µH
n
−µK) ≤
_
f χH
n
dµ −γ, so that
_
f χ(H
n
¸ K)dµ +γ ≤
_
f χH
n
and
νK =
_
f χK dµ ≥ γ. Thus in either case we have a K ∈ / such that K ⊆ F and νK ≥ γ; as F and γ
are arbitrary, ν is inner regular with respect to /.
412R Lemma Let (X, Σ, µ) and (Y, T, ν) be measure spaces, with c.l.d. product space (X Y, Λ, λ)
(251F). Suppose that / ⊆ TX, L ⊆ TY , / ⊆ T(X Y ) are such that
(i) µ is inner regular with respect to /;
(ii) ν is inner regular with respect to L;
(iii) K L ∈ / for all K ∈ /, L ∈ L;
(iv) M ∪ M
t
∈ / whenever M, M
t
∈ /;
(v)
n∈N
M
n
∈ / for every sequence ¸M
n
)
n∈N
in /.
Then λ is inner regular with respect to /.
proof Write / = ¦E Y : E ∈ Σ¦ ∪¦X F : F ∈ T¦. Then if V ∈ /, W ∈ Λ and λ(W ∩V ) > 0, there is
an M ∈ /∩ / such that M ⊆ W and λ(M ∩ V ) > 0. PPP Suppose that V = E Y where E ∈ Σ. There
must be E
0
∈ Σ and F
0
∈ T, both of ﬁnite measure, such that λ(W ∩V ∩(E
0
F
0
)) > 0 (251F). Now there
is a K ∈ / such that K ⊆ E ∩ E
0
and µ((E ∩ E
0
) ¸ K) νF
0
< λ(W ∩ V ∩ (E
0
F
0
)); but this means that
M = KY is included in V and µ(W ∩M) > 0. Reversing the roles of the coordinates, the same argument
deals with the case in which V = X F for some F ∈ T. QQQ
By 412C, λ
0
= λΣ
´
⊗T is inner regular with respect to /. But λ is inner regular with respect to Σ
´
⊗T
(251Ib) so is also inner regular with respect to / (412Ab).
412S Proposition Let (X, Σ, µ) and (Y, T, ν) be measure spaces, with c.l.d. product space (XY, Λ, λ).
Let T, S be topologies on X and Y respectively, and give X Y the product topology.
(a) If µ and ν are inner regular with respect to the closed sets, so is λ.
(b) If µ and ν are tight (that is, inner regular with respect to the closed compact sets), so is λ.
(c) If µ and ν are inner regular with respect to the zero sets, so is λ.
(d) If µ and ν are inner regular with respect to the Borel sets, so is λ.
(e) If µ and ν are eﬀectively locally ﬁnite, so is λ.
proof We have only to read the conditions (i)(v) of 412R carefully and check that they apply in each
case. (In part (e), recall that ‘eﬀectively locally ﬁnite’ is the same thing as ‘inner regular with respect to
the subsets of open sets of ﬁnite measure’.)
412T Lemma Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of probability spaces, with product probability space
(X, Λ, λ) (¸254). Suppose that /
i
⊆ TX
i
, / ⊆ TX are such that
(i) µ
i
is inner regular with respect to /
i
for each i ∈ I;
(ii) π
−1
i
[K] ∈ / for every i ∈ I, K ∈ /
i
, writing π
i
(x) = x(i) for x ∈ X;
(iii) M ∪ M
t
∈ / whenever M, M
t
∈ /;
(iv)
n∈N
M
n
∈ / for every sequence ¸M
n
)
n∈N
in /.
Then λ is inner regular with respect to /.
proof The argument is nearly identical to that of 412R. Write / = ¦π
−1
i
[E] : i ∈ I, E ∈ Σ
i
¦. Then if
V ∈ /, W ∈ Λ and λ(W ∩ V ) > 0, express V as π
−1
i
[E], where i ∈ I and E ∈ Σ
i
, and take K ∈ /
i
such
that K ⊆ E and µ
i
(E ¸ K) < λ(W ∩V ); then M = π
−1
i
[K] belongs to /∩/, is included in W, and meets
V in a nonnegligible set. So, just as in 412R, the conditions of 412C are met.
It follows that λ
0
= λ
´
i∈I
Σ
i
is inner regular with respect to /. But λ is the completion of λ
0
(254Fd,
254Ff), so is also inner regular with respect to / (412Ha).
*412W Inner regularity 27
412U Proposition Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of probability spaces, with product probability space
(X, Λ, λ). Suppose that we are given a topology T
i
on each X
i
, and let T be the product topology on X.
(a) If every µ
i
is inner regular with respect to the closed sets, so is λ.
(b) If every µ
i
is inner regular with respect to the zero sets, so is λ.
(c) If every µ
i
is inner regular with respect to the Borel sets, so is λ.
proof This follows from 412T just as 412S follows from 412R.
412V Corollary Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of probability spaces, with product probability space
(X, Λ, λ). Suppose that we are given a Hausdorﬀ topology T
i
on each X
i
, and let T be the product topology
on X. Suppose that every µ
i
is tight, and that X
i
is compact for all but countably many i ∈ I. Then λ is
tight.
proof By 412Ua, λ is inner regular with respect to the closed sets. If W ∈ Λ and γ < λW, let V ⊆ W
be a measurable closed set such that λV > γ. Let J be the set of those i ∈ I such that X
i
is not
compact; we are supposing that J is countable. Let ¸
i
)
i∈J
be a family of strictly positive real numbers
such that
i∈J
j
≤ λV − γ (4A1P). For each i ∈ J, let K
i
⊆ X
i
be a compact measurable set such that
µ
i
(X
i
¸ K
i
) ≤
i
; and for i ∈ I ¸ J, set K
i
= X
i
. Then K =
i∈I
K
i
is a compact measurable subset of X,
and
λ(X ¸ K) ≤
i∈J
µ(X
i
¸ K
i
) ≤ λV −γ,
so λ(K ∩ V ) ≥ γ; while K ∩ V is a compact measurable subset of W. As W and γ are arbitrary, λ is tight.
*412W Outer regularity I have already mentioned the complementary notion of ‘outer regularity’
(411D). In this book it will not be given much prominence. It is however a useful tool when dealing with
Lebesgue measure (see, for instance, the proof of 225K), for reasons which the next proposition will make
clear.
Proposition Let (X, Σ, µ) be a measure space and T a topology on X.
(a) Suppose that µ is outer regular with respect to the open sets. Then for any integrable function
f : X → [0, ∞] and > 0, there is a lower semicontinuous measurable function g : X → [0, ∞] such that
f ≤ g and
_
g ≤ +
_
f.
(b) Now suppose that there is a sequence of measurable open sets of ﬁnite measure covering X. Then
the following are equiveridical:
(i) µ is inner regular with respect to the closed sets;
(ii) µ is outer regular with respect to the open sets;
(iii) for any measurable set E ⊆ X and > 0, there are a measurable closed set F ⊆ E and a measurable
open set H ⊇ E such that µ(H ¸ F) ≤ ;
(iv) for every measurable function f : X → [0, ∞[ and > 0, there is a lower semicontinuous measurable
function g : X → [0, ∞] such that f ≤ g and
_
g −f ≤ ;
(v) for every measurable function f : X → R and > 0, there is a lower semicontinuous measurable
function g : X → ]−∞, ∞] such that f ≤ g and µ¦x : g(x) ≥ f(x) +¦ ≤ .
proof (a) Let η ∈ ]0, 1] be such that η(7 +
_
fdµ) ≤ . For n ∈ Z, set E
n
= ¦x : (1 + η)
n
≤ f(x) <
(1 + η)
n+1
¦, and let E
t
n
∈ Σ be a measurable cover of E
n
; let G
n
⊇ E
t
n
be a measurable open set such
that µG
n
≤ 3
−]n]
η + µE
t
n
. Set g =
∞
n=−∞
(1 + η)
n+1
χG
n
. Then g is lower semicontinuous (4A2B(diii),
4A2B(dv)), f ≤ g and
_
gdµ =
∞
n=−∞
(1 +η)
n+1
µG
n
≤ (1 +η)
∞
n=−∞
(1 +η)
n
µE
t
n
+
∞
n=−∞
(1 +η)
n+1
3
−]n]
η
≤ (1 +η)
_
fdµ + 7η ≤
_
fdµ +,
28 Topologies and measures *412W
as required.
(b) Let ¸G
n
)
n∈N
be a sequence of open sets of ﬁnite measure covering X; replacing it by ¸
i<n
G
i
)
n∈N
if necessary, we may suppose that ¸G
n
)
n∈N
is nondecreasing and that G
0
= ∅.
(i)⇒(iii) Suppose that µ is inner regular with respect to the closed sets, and that E ∈ Σ, > 0.
For each n ∈ N let F
n
⊆ G
n
¸ E be a measurable closed set such that µF
n
≥ µ(G
n
¸ E) − 2
−n−2
. Then
H =
n∈N
(G
n
¸F
n
) is a measurable open set including E and µ(H¸ E) ≤
1
2
. Applying the same argument
to X ¸ E, we get a closed set F ⊆ E such that µ(E ¸ F) ≤
1
2
, so that µ(H ¸ F) ≤ .
(ii)⇒(iii) The same idea works. Suppose that µ is outer regular with respect to the open sets, and
that E ∈ Σ, > 0. For each n ∈ N, let H
n
⊇ G
n
∩ E
n
be an open set such that µ(H
n
¸ E
n
) ≤ 2
−n−2
;
then H =
n∈N
H
n
is a measurable open set including E, and µ(H ¸ E) ≤
1
2
. Now repeat the argument
on X ¸ E to ﬁnd a measurable closed set F ⊆ E such that µ(E ¸ F) ≤
1
2
.
(iii)⇒(iv) Assume (iii), and let f : X → [0, ∞[ be a measurable function, > 0. Set η
n
= 2
−n
/(16 +
4µG
n
) for each n ∈ N. For k ∈ N set E
k
=
n∈N
¦x : x ∈ G
n
, kη
n
≤ f(x) < (k +1)η
n
¦, and choose an open
set H
k
⊇ E
k
such that µ(H
k
¸ E
k
) ≤ 2
−k
. Set
g = sup
k,n∈N
(k + 1)η
n
χ(G
n
∩ H
k
).
Then g : X → [0, ∞] is lower semicontinuous (4A2B(dv) again). Since
sup
k,n∈N
kη
n
χ(G
n
∩ E
k
) ≤ f ≤ sup
k,n∈N
(k + 1)η
n
χ(G
n
∩ E
k
),
f ≤ g and
g −f ≤ sup
k,n∈N
(k + 1)η
n
χ(G
n
∩ H
k
¸ E
k
) + sup
k,n∈N
η
n
χ(G
n
∩ E
k
)
has integral at most
∞
k=0
∞
n=0
(k + 1)η
n
2
−k
+
∞
n=0
η
n
µG
n
≤ .
(i)⇒(v) Assume (i), and suppose that f : X →R is measurable and > 0. For each n ∈ N, let α
n
≥ 0
be such that µE
n
< 2
−n−1
, where E
n
= ¦x : x ∈ G
n+1
¸G
n
, f(x) ≤ −α
n
¦. Let F
n
⊆ (G
n+1
¸G
n
) ¸E
n
be a
measurable closed set such that µ((G
n+1
¸G
n
)¸F
n
) ≤ 2
−n−2
. Because ¸F
n
)
n∈N
is disjoint, h =
∞
n=0
α
n
χF
n
is deﬁned as a function from X to R. ¦F
n
: n ∈ N¦ is locally ﬁnite, so ¦x : h(x) ≥ α¦ =
n∈N,α
n
≥α
F
n
is closed for every α > 0 (4A2B(hii)), and h is upper semicontinuous. Now f
1
= f + h is a measurable
function. Since (i)⇒(iii)⇒(iv), there is a measurable lower semicontinuous function g
1
: X → [0, ∞] such
that f
+
1
≤ g
1
and
_
g
1
− f
+
1
≤
1
2
2
, where f
+
1
= max(0, f
1
). But if we now set g = g
1
− h, g is lower
semicontinuous, f ≤ g and
¦x : f(x) + ≤ g(x)¦ ⊆ ¦x : f
+
1
(x) + ≤ g
1
(x)¦ ∪ ¦x : f
1
(x) < 0¦
⊆ ¦x : f
+
1
(x) + ≤ g
1
(x)¦ ∪
_
n∈N
(G
n+1
¸ G
n
) ¸ F
n
has measure at most , as required.
(iv)⇒(ii) and (v)⇒(ii) Suppose that either (iv) or (v) is true, and that E ∈ Σ, > 0. Then there is a
measurable lower semicontinuous function g : X → ]0, ∞] such that χE ≤ g and µ¦x : χE(x)+
1
2
≤ g(x)¦ ≤
, since this is certainly true if
_
g −χE ≤
1
2
. Set G = ¦x : g(x) >
1
2
¦; then E ⊆ G and µ(G¸ E) ≤ .
(iii)⇒(i) is trivial. Assembling these fragments, the proof is complete.
412X Basic exercises (a) Let (X, Σ, µ) be a measure space and T a topology on X such that µ is inner
regular with respect to the closed sets and with respect to the compact sets. Show that it is tight.
(b) Explain how 213A is a special case of 412Aa.
>>>(c) Let (X, Σ, µ) be a measure space, and Σ
0
a σsubalgebra of Σ such that µ is inner regular with
respect to Σ
0
. Show that if 1 ≤ p < ∞then every member of L
p
(µ) is of the form f
•
for some Σ
0
measurable
f : X →R.
412Xp Inner regularity 29
>>>(d) Let (X, Σ, µ) be a semiﬁnite measure space and / ⊆ Σ an algebra of sets such that the σalgebra
generated by / is Σ. Write / for ¦
n∈N
E
n
: E
n
∈ / for every n ∈ N¦. Show that µ is inner regular with
respect to /.
(e) Let (X, T, Σ, µ) be an eﬀectively locally ﬁnite Hausdorﬀ topological measure space such that µ is
inner regular with respect to the Borel sets. Suppose that µG = sup¦µK : K ⊆ G is compact¦ for every
open set G ⊆ X. Show that µ is tight.
(f ) Let (X, T) be a topological space such that every open set is an F
σ
set. Show that any eﬀectively
locally ﬁnite Borel measure on X is inner regular with respect to the closed sets.
(g) Let (X, T) be a normal topological space and µ a topological measure on X which is inner regular
with respect to the closed sets. Show that µG = sup¦µH : H ⊆ G is a cozero set¦ for every open set G ⊆ X.
Show that if µ is totally ﬁnite, then µF = inf¦µH : H ⊇ F is a zero set¦ for every closed set F ⊆ X.
(h) Let (X, Σ, µ) be a complete locally determined measure space, and suppose that µ is inner regular
with respect to a family / of sets. Let Σ
0
be the σalgebra of subsets of X generated by / ∩ Σ. (i) Show
that µ is the c.l.d. version of µΣ
0
. (Hint: 412J412L.) (ii) Show that if µ is σﬁnite, it is the completion of
µΣ
0
.
>>>(i) (i) Let (X, Σ, µ) be a σﬁnite measure space and T a σsubalgebra of Σ. Show that if µ is inner
regular with respect to T then the completion of µT extends µ, so that µ and µT have the same negligible
sets. (ii) Show that if µ is a σﬁnite topological measure which is inner regular with respect to the Borel
sets, then every µnegligible set is included in a µnegligible Borel set.
(j) Devise a direct proof of 412L, not using 412K, by (i) showing that µ
∗
(A∩K) = ν
∗
(A∩K) whenever
A ⊆ X, K ∈ / (ii) showing that µ
∗
= ν
∗
(iii) quoting 213C.
(k) Let (X, Σ, µ) be a complete locally determined measure space, Y a set and f : X → Y a function.
Show that the following are equiveridical: (i) µ is inner regular with respect to ¦f
−1
[B] : B ⊆ Y ¦ (ii)
f
−1
[f[E]] ¸ E is negligible for every E ∈ Σ.
(l) Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of measure spaces, with direct sum (X, Σ, µ). Suppose that for each
i ∈ I we are given a topology T
i
on X
i
, and let T be the corresponding disjoint union topology on X. Show
that (i) µ is inner regular with respect to the closed sets iﬀ every µ
i
is (ii) µ is inner regular with respect
to the compact sets iﬀ every µ
i
is (iii) µ is inner regular with respect to the zero sets iﬀ every µ
i
is (iv) µ is
inner regular with respect to the Borel sets iﬀ every µ
i
is.
(m) Use 412L and 412Q to shorten the proof of 253I.
(n) Let ¸X
i
)
i∈I
be a family of sets, and suppose that we are given, for each i ∈ I, a σalgebra Σ
i
of subsets
of X
i
and a topology T
i
on X
i
. Let T be the product topology on X =
i∈I
X
i
, and Σ =
´
i∈I
Σ
i
. Let µ
be a totally ﬁnite measure with domain Σ, and set µ
i
= µπ
−1
i
for each i ∈ I, where π
i
(x) = x(i) for i ∈ I,
x ∈ X. (i) Show that µ is inner regular with respect to the family / of sets expressible as X¸
n∈N
i∈I
E
ni
where E
ni
∈ Σ
i
for every n, i and ¦i : E
ni
,= X
i
¦ is ﬁnite for each n. (ii) Show that if every µ
i
is inner
regular with respect to the closed sets, so is µ. (iii) Show that if every µ
i
is inner regular with respect to
the zero sets, so is µ. (iv) Show that if every µ
i
is inner regular with respect to the Borel sets, so is µ. (v)
Show that if every µ
i
is tight, and all but countably many of the X
i
are compact, then µ is tight.
(o) Let (X, Σ, µ) be a measure space and T a Lindel¨of topology on X such that µ is locally ﬁnite. (i)
Show that µ is σﬁnite. (ii) Show that µ is inner regular with respect to the closed sets iﬀ it is outer regular
with respect to the open sets.
(p) Let X be a topological space and µ a measure on X which is outer regular with respect to the open
sets. Show that for any Y ⊆ X the subspace measure on Y is outer regular with respect to the open sets.
30 Topologies and measures 412Xq
(q) Let X be a topological space and µ a measure on X which is outer regular with respect to the open
sets. Show that if f : X →R is integrable and > 0 then there is a lower semicontinuous g : X → ]−∞, ∞]
such that f ≤ g and
_
g −f ≤ .
(r) Let (X, Σ, µ) be a semiﬁnite measure space and ¸f
n
)
n∈N
a sequence in L
0
(µ) which converges almost
everywhere to f ∈ L
0
(µ). Show that µ is inner regular with respect to ¦E : ¸f
n
E)
n∈N
is uniformly
convergent¦. (Cf. 215Yb.)
(s) In 216E, give ¦0, 1¦
I
its usual compact Hausdorﬀ topology. Show that the measure µ described there
is inner regular with respect to the zero sets.
412Y Further exercises (a) Let / be the family of subsets of R which are homeomorphic to the Cantor
set. Show that Lebesgue measure is inner regular with respect to /. (Hint: show that if F ⊆ R ¸ Q is an
uncountable compact set, then ¦x : [x −δ, x +δ] ∩ F is uncountable for every δ > 0¦ belongs to /.)
(b) (i) Show that if X is a perfectly normal space then any semiﬁnite topological measure on X which
is inner regular with respect to the Borel sets is inner regular with respect to the closed sets. (ii) Show that
any subspace of a perfectly normal space is perfectly normal. (iii) Show that the split interval I

(343J,
343Yc, 419L) is perfectly normal. (iv) Show that ω
1
, with its order topology, is completely regular, normal
and Hausdorﬀ, but not perfectly normal. (v) Show that I

I

is not perfectly normal. (vi) Show that
[0, 1]
I
is perfectly normal iﬀ I is countable.
(c) Let (X, Σ, µ) be a measure space, and suppose that µ is inner regular with respect to a family / ⊆ Σ
such that K∪K
t
∈ / for all K, K
t
∈ /. Write Σ
f
for ¦E : E ∈ Σ, µE < ∞¦. Show that ¦E
•
: E ∈ /∩Σ
f
¦
is dense in ¦E
•
: E ∈ Σ
f
¦ for the strong measurealgebra topology.
(d) Let X be a normal topological space and Y a closed subset of X. Show that every Baire subset of Y
is the intersection of Y with a Baire subset of X. (Hint: use Tietze’s theorem.)
(e) Let (X, Σ, µ) be [0, 1] with Lebesgue measure, and Y = [0, 1] with counting measure ν; give X its
usual topology and Y the discrete topology, and let λ be the c.l.d. product measure on X Y . (i) Show
that µ, ν and λ are all tight (for the appropriate topologies) and therefore completion regular. (ii) Let λ
0
be the primitive product measure on X Y (deﬁnition: 251C). Show that λ
0
is not tight. (Hint: 252Yf.)
Remark: it is undecidable in ZFC whether λ
0
is inner regular with respect to the closed sets.
(f ) Give an example of a Hausdorﬀ topological measure space (X, T, Σ, µ) such that µ is complete, strictly
localizable and outer regular with respect to the open sets, but not inner regular with respect to the closed
sets.
412 Notes and comments In this volume we are returning to considerations which have been left on one
side for almost the whole of Volume 3 – the exceptions being in Chapter 34, where I looked at realization of
homomorphisms of measure algebras by functions between measure spaces, and was necessarily dragged into
an investigation of measure spaces which had enough points to be adequate codomains (343B). The idea
of ‘inner regularity’ is to distinguish families / of sets which will be large enough to describe the measure
entirely, but whose members will be of recognisable types. For an example of this principle see 412Ya. Of
course we cannot always ﬁnd a single type of set adequate to ﬁll a suitable family /, though this happens
oftener than one might expect, but it is surely easier to think about an arbitrary zero set (for instance) than
an arbitrary measurable set, and whenever a measure is inner regular with respect to a recognisable class it
is worth knowing about it.
I have tried to use the symbols † and ‡ (412A, 412C) consistently enough for them to act as a guide
to some of the ideas which will be used repeatedly in this chapter. Note the emphasis on disjoint unions
and countable intersections; I mentioned similar conditions in 136Xi136Xj. You will recognise 412Aa as
an exhaustion principle; note that it is enough to use disjoint unions, as in 313K. In the examples of this
section this disjointness is not important. Of course inner regularity has implications for the measure algebra
(412N), but it is important to recognise that ‘µ is inner regular with respect to /’ is saying much more than
§413 intro. Inner measure constructions 31
‘¦K
•
: K ∈ /¦ is orderdense in the measure algebra’; the latter formulation tells us only that whenever
µE > 0 there is a K ∈ / such that K ¸ E is negligible and µK > 0, while the former tells us that we can
take K to be actually a subset of E.
412D, 412E and 412G are all of great importance. 412D looks striking, but of course the reason it works is
just that the Baire σalgebra is very small. In 412E the Baire and Borel σalgebras coincide, so it is nothing
but a special case of 412D; but as metric spaces are particularly important it is worth having it spelt out
explicitly. In 412D and 412E the hypothesis ‘semiﬁnite’ is suﬃcient, while in 412G we need ‘eﬀectively
locally ﬁnite’; this is because in both 412D and 412E the open sets we are looking at are countable unions
of measurable closed sets. There are interesting nonmetrizable spaces in which the same thing happens
(412Yb). As you know, I am strongly prejudiced in favour of complete and locally determined measures,
and the Baire and Borel measures dealt with in these three results are rarely complete; but they can still be
applied to completions and c.l.d. versions of these measures, using 412Ab or 412H.
412O412V are essentially routine. For subspace measures, the only problem we need to come to terms
with is the fact that subspaces of semiﬁnite measure spaces need not be semiﬁnite (216Xa). For product
measures the point is that the c.l.d. product of two measure spaces, and the product of any family of
probability spaces, as I deﬁned them in Chapter 25, are inner regular with respect to the σalgebra of sets
generated by the cylinder sets. This is not in general true of the ‘primitive’ product measure (412Ye), which
is one of my reasons for being prejudiced against it. I should perhaps warn you of a trap in the language
I use here. I say that if the factor measures are inner regular with respect to the closed sets, so is the
c.l.d. product measure. But I do not say that all closed sets in the product are measurable for the product
measure, even if closed sets in the factors are measurable for the factor measures. So the path is open
for a diﬀerent product measure to exist, still inner regular with respect to the closed sets; and indeed I
shall be going down that path in ¸417. The uniqueness result in 412L speciﬁcally refers to complete locally
determined measures deﬁned on all sets of the family /.
There is one special diﬃculty in 412V: in order to ensure that there are enough compact measurable sets
in X =
i∈I
X
i
, we need to know that all but countably many of the X
i
are actually compact. When we
come to look more closely at products of Radon probability spaces we shall need to consider this point again
(417Q, 417Xq).
In fact some of the ideas of 412U412V are not restricted to the product measures considered there.
Other measures on the product space will have inner regularity properties if their images on the factors,
their ‘marginals’ in the language of probability theory, are inner regular; see 412Xn. I will return to this in
¸454.
This section is almost exclusively concerned with inner regularity. The complementary notion of outer
regularity is not much use except in σﬁnite spaces (415Xh), and not always then (416Yd). In totally
ﬁnite spaces, of course, and some others, any version of inner regularity corresponds to a version of outer
regularity, as in 412Wb(i)(ii); and when we have something as strong as 412Wb(iii) available it is worth
knowing about it.
413 Inner measure constructions
I now turn in a diﬀerent direction, giving some basic results on the construction of inner regular measures.
The ﬁrst step is to describe ‘inner measures’ (413A) and a construction corresponding to the Carath´eodory
construction of measures from outer measures (413C). Just as every measure gives rise to an outer measure,
it gives rise to an inner measure (413D). Inner measures form an eﬀective tool for studying complete locally
determined measures (413F).
The most substantial results of the section concern the construction of measures as extensions of function
als deﬁned on various classes / of sets. Typically, / is closed under ﬁnite unions and countable intersections,
though it is sometimes possible to relax the hypotheses to some extent. The methods here make it possible
to distinguish arguments which produce ﬁnitely additive functionals (413H, 413N, 413P, 413Q) from the
succeeding steps to countably additive measures (413I, 413O, 413S). 413H413M investigate conditions on
a functional φ : / → [0, ∞[ suﬃcient to produce a measure extending φ, necessarily unique, which is inner
regular with respect to / or /
δ
, the set of intersections of sequences in /. 413N413O look instead at func
tionals deﬁned on sublattices of the class / of interest, and at suﬃcient conditions to ensure the existence of
a measure, not normally unique, deﬁned on the whole of /, inner regular with respect to / and extending
32 Topologies and measures I §413 intro.
the given functional. Finally, 413P413S are concerned with majorizations rather than extensions; we seek
a measure µ such that µK ≥ λK for K ∈ /, while µX is as small as possible.
413A I begin with some material from the exercises of earlier volumes.
Deﬁnition Let X be a set. An inner measure on X is a function φ : TX → [0, ∞] such that
φ∅ = 0;
(α) φ(A∪ B) ≥ φA+φB for all disjoint A, B ⊆ X;
(β) if ¸A
n
)
n∈N
is a nonincreasing sequence of subsets of X and φA
0
< ∞ then φ(
n∈N
A
n
) =
inf
n∈N
φA
n
;
(∗) φA = sup¦φB : B ⊆ A, φB < ∞¦ for every A ⊆ X.
413B The following fact will be recognised as an element of Carath´eodory’s method. There will be an
application later in which it will be useful to know that it is not conﬁned to proving countable additivity.
Lemma Let X be a set and φ : X → [0, ∞] any function such that φ∅ = 0. Then
Σ = ¦E : E ⊆ X, φA = φ(A∩ E) +φ(A¸ E) for every A ⊆ X¦
is an algebra of subsets of X, and φ(E ∪ F) = φE +φF for all disjoint E, F ∈ Σ.
proof The symmetry of the deﬁnition of Σ ensures that X ¸ E ∈ Σ whenever E ∈ Σ. If E, F ∈ Σ and
A ⊆ X, then
φ(A∩ (E∪F)) +φ(A¸ (E ∪ F))
= φ(A∩ (E ∪ F) ∩ E) +φ(A∩ (E ∪ F) ¸ E) +φ(A¸ (E ∪ F))
= φ(A∩ E) +φ((A¸ E) ∩ F) +φ((A¸ E) ¸ F)
= φ(A∩ E) +φ(A¸ E) = φA.
As A is arbitrary, E ∪ F ∈ Σ. Finally, if A ⊆ X,
φ(A∩ ∅) +φ(A¸ ∅) = φ∅ +φA = φA
because φ∅ = 0; so ∅ ∈ Σ.
Thus Σ is an algebra of sets. If E, F ∈ Σ and E ∩ F = ∅, then
φ(E ∪ F) = φ((E ∪ F) ∩ E) +φ((E ∪ F) ¸ E) = φE +φF.
413C Measures from inner measures I come now to a construction corresponding to Carath´eodory’s
method of deﬁning measures from outer measures.
Theorem Let X be a set and φ : X → [0, ∞] an inner measure. Set
Σ = ¦E : E ⊆ X, φ(A∩ E) +φ(A¸ E) = φA for every A ⊆ X¦.
Then (X, Σ, φΣ) is a complete measure space.
proof (Compare 113C.)
(a) The ﬁrst step is to note that if A ⊆ B ⊆ X then
φB ≥ φA+φ(B ¸ A) ≥ φA.
Next, a subset E of X belongs to Σ iﬀ φA ≤ φ(A ∩ E) + φ(A ¸ E) whenever A ⊆ X and µA < ∞. PPP Of
course any element of Σ satisﬁes the condition. If E satisﬁes the condition and A ⊆ X, then
φA = sup¦φB : B ⊆ A, φB < ∞¦
≤ sup¦φ(B ∩ E) +φ(B ¸ E) : B ⊆ A¦
= φ(A∩ E) +φ(A¸ E) ≤ φA,
413D Inner measure constructions 33
so E ∈ Σ. QQQ
(b) By 413B, Σ is an algebra of subsets of X. Now suppose that ¸E
n
)
n∈N
is a nondecreasing sequence
in Σ, with union E. If A ⊆ X and φA < ∞, then
φ(A¸ E) = inf
n∈N
φ(A¸ E
n
) = lim
n→∞
φ(A¸ E
n
)
because ¸A¸ E
n
)
n∈N
is nonincreasing and φ(A¸ E
0
) is ﬁnite; so
φ(A∩ E) +φ(A¸ E) ≥ lim
n→∞
φ(A∩ E
n
) +φ(A¸ E
n
) = φA.
By (a), E ∈ Σ. So Σ is a σalgebra.
(c) If E, F ∈ Σ and E ∩ F = ∅ then φ(E ∪ F) = φE +φF, by 413B. If ¸E
n
)
n∈N
is a disjoint sequence in
Σ with union E, then
µE ≥ µ(
i≤n
E
i
) =
n
i=0
µE
i
for every n, so µE ≥
∞
i=0
µE
i
. ??? If µE >
∞
i=0
µE
i
, there is an A ⊆ E such that
∞
i=0
µE
i
< φA < ∞.
But now, setting F
n
=
i≤n
E
i
for each n, we have lim
n→∞
φ(A¸ F
n
) = 0, so that
φA = lim
n→∞
φ(A∩ F
n
) +φ(A¸ F
n
) =
∞
i=0
φ(A∩ E
i
) < φA,
which is absurd. XXX Thus µE =
∞
i=0
µE
i
. As ¸E
n
)
n∈N
is arbitrary, µ is a measure.
(d) Finally, suppose that B ⊆ E ∈ Σ and µE = 0. Then for any A ⊆ X we must have
φ(A∩ B) +φ(A¸ B) ≥ φ(A¸ E) = φ(A∩ E) +φ(A¸ E) = φA,
so B ∈ Σ. Thus µ is complete.
Remark For a simple example see 213Yc.
413D The inner measure deﬁned by a measure Let (X, Σ, µ) be any measure space. Just as µ
has an associated outer measure µ
∗
deﬁned by the formula
µ
∗
A = inf¦µE : A ⊆ E ∈ Σ¦
(132A132B), it gives rise to an inner measure µ
∗
deﬁned by the formula
µ
∗
A = sup¦µE : E ∈ Σ
f
, E ⊆ A¦,
where I write Σ
f
for ¦E : E ∈ Σ, µE < ∞¦. PPP µ
∗
∅ = µ∅ = 0. (†) If A∩ B = ∅, and E ⊆ A, F ⊆ B belong
to Σ
f
, then E ∪ F ⊆ A∪ B also has ﬁnite measure, so
µ
∗
(A∪ B) ≥ µ(E ∪ F) = µE +µF;
taking the supremum over E and F, µ
∗
(A ∪ B) ≥ µ
∗
A + µ
∗
B. (‡) If ¸A
n
)
n∈N
is a nonincreasing sequence
of sets with intersection A and µ
∗
A
0
< ∞, then for each n ∈ N we can ﬁnd an E
n
⊆ A
n
such that
µE
n
≥ µ
∗
A
n
−2
−n
. In this case,
µ(
m∈N
E
m
) = sup
n∈N
µ(
m≤n
E
m
) ≤ µ
∗
A
0
< ∞.
Set
E =
n∈N
m≥n
E
m
⊆ A.
Then E ∈ Σ
f
, so
µ
∗
A ≥ µE ≥ limsup
n→∞
µE
n
= lim
n→∞
µ
∗
A
n
≥ µ
∗
A.
(*) If A ⊆ X and µ
∗
A = ∞ then
sup¦µ
∗
B : B ⊆ A, µ
∗
B < ∞¦ ≥ sup¦µE : E ∈ Σ
f
, E ⊆ A¦ = ∞. QQQ
Warning Many authors use the formula
µ
∗
A = sup¦µE : A ⊇ E ∈ Σ¦.
In ‘ordinary’ cases, when (X, Σ, µ) is semiﬁnite, this agrees with my usage (413Ed); but for nonsemiﬁnite
spaces there is a diﬀerence.
34 Topologies and measures I 413D
For elementary properties of the construction here see 213Xe.
413E I note the following elementary facts concerning inner measures deﬁned from measures.
Proposition Let (X, Σ, µ) be a measure space. Write Σ
f
for ¦E : E ∈ Σ, µE < ∞¦.
(a) For every A ⊆ X there is an E ∈ Σ such that E ⊆ A and µE = µ
∗
A.
(b) µ
∗
A ≤ µ
∗
A for every A ⊆ X.
(c) If E ∈ Σ and A ⊆ X, then µ
∗
(E ∩ A) +µ
∗
(E ¸ A) ≤ µE, with equality if either (i) µE < ∞ or (ii) µ
is semiﬁnite.
(d) In particular, µ
∗
E ≤ µE for every E ∈ Σ, with equality if either µE < ∞ or µ is semiﬁnite.
(e) If µ is inner regular with respect to /, then µ
∗
A = sup¦µK : K ∈ /∩ Σ
f
, K ⊆ A¦ for every A ⊆ X.
(f) If A ⊆ X is such that µ
∗
A = µ
∗
A < ∞, then A is measured by the completion of µ.
(g) If ˆ µ, ˜ µ are the completion and c.l.d. version of µ, then ˆ µ
∗
= ˜ µ
∗
= µ
∗
.
(h) If (Y, T, ν) is another measure space, and f : X → Y is an inversemeasurepreserving function, then
µ
∗
(f
−1
[B]) ≤ ν
∗
B, µ
∗
(f
−1
[B]) ≥ ν
∗
B
for every B ⊆ Y , and
ν
∗
(f[A]) ≥ µ
∗
A
for every A ⊆ X.
(i) Suppose that µ is semiﬁnite. If A ⊆ E ∈ Σ, then E is a measurable envelope of A iﬀ µ
∗
(E ¸ A) = 0.
proof (a) There is a sequence ¸E
n
)
n∈N
in Σ
f
such that E
n
⊆ A for each n and lim
n→∞
µE
n
= µ
∗
A; now
set E =
n∈N
E
n
.
(b) If E ⊆ A ⊆ F we must have µE ≤ µF.
(c) If F ⊆ E ∩ A and F ∈ Σ
f
, then
µF +µ
∗
(E ¸ A) ≤ µF +µ(E ¸ F) = µE;
taking the supremum over F, µ
∗
(E ∩ A) +µ
∗
(E ¸ A) ≤ µE. If µE < ∞, then
µ
∗
(E ∩ A) = sup¦µF : F ∈ Σ, F ⊆ E ∩ A¦
= µE −inf¦µ(E ¸ F) : F ∈ Σ, F ⊆ E ∩ A¦
= µE −inf¦µF : F ∈ Σ, E ¸ A ⊆ F ⊆ E¦ = µE −µ
∗
(E ¸ A).
If µ is semiﬁnite, then
µ
∗
(E ∩ A) +µ
∗
(E ¸ A) ≥ sup¦µ
∗
(F ∩ A) +µ
∗
(F ¸ A) : F ∈ Σ
f
, F ⊆ E¦
= sup¦µF : F ∈ Σ
f
, F ⊆ E¦ = µE.
(d) Take A = E in (c).
(e)
µ
∗
A = sup¦µE : E ∈ Σ
f
, E ⊆ A¦
= sup¦µK : K ∈ / ∩ Σ, ∃ E ∈ Σ
f
, K ⊆ E ⊆ A¦
= sup¦µK : K ∈ / ∩ Σ
f
, K ⊆ A¦.
(f ) By (a) above and 132Aa, there are E, F ∈ Σ such that E ⊆ A ⊆ F and
µE = µ
∗
A = µ
∗
A = µF < ∞;
now µ(F ¸ E) = 0, so F ¸ A and A are measured by the completion of µ.
(g) Write ˇ µ for either ˆ µ or ˜ µ, and
ˇ
Σ for its domain, and let A ⊆ X. (i) If γ < µ
∗
A, there is an E ∈ Σ such
that E ⊆ A and γ ≤ µE < ∞; now ˇ µE = µE (212D, 213Fa), so ˇ µ
∗
A ≥ γ. As γ is arbitrary, µ
∗
A ≤ ˇ µ
∗
A.
413F Inner measure constructions 35
(ii) If γ < ˇ µ
∗
A, there is an E ∈
ˇ
Σ such that E ⊆ A and γ ≤ ˇ µE < ∞. Now there is an F ∈ Σ such that
F ⊆ E and µF = ˇ µE (212C, 213Fc), so that µ
∗
A ≥ γ. As γ is arbitrary, µ
∗
A ≥ ˇ µ
∗
A.
(h) This is elementary; all we have to note is that if F, F
t
∈ T and F ⊆ B ⊆ F
t
, then f
−1
[F] ⊆ f
−1
[B] ⊆
f
−1
[F
t
], so that
νF = µf
−1
[F] ≤ µ
∗
f
−1
[B] ≤ µ
∗
f
−1
[B] ≤ µf
−1
[F
t
] = νF
t
.
Now, for A ⊆ X,
µ
∗
A ≤ µ
∗
(f
−1
[f[A]]) ≤ ν
∗
(f[A]).
(i)(i) If E is a measurable envelope of A and F ∈ Σ is included in E ¸ A, then
µF = µ(F ∩ E) = µ
∗
(F ∩ A) = 0;
as F is arbitrary, µ
∗
(E ¸ A) = 0. (ii) If E is not a measurable envelope of A, there is an F ∈ Σ such that
µ
∗
(F ∩ A) < µ(F ∩ E). Let G ∈ Σ be such that F ∩ A ⊆ G and µG = µ
∗
(F ∩ A). Then µ(F ∩ E ¸ G) > 0;
because µ is semiﬁnite, µ
∗
(E ¸ A) ≥ µ
∗
(F ∩ E ¸ G) > 0.
413F The language of 413D makes it easy to express some useful facts about complete locally determined
measure spaces, complementing 412J.
Lemma Let (X, Σ, µ) be a complete locally determined measure space and / a family of subsets of X such
that µ is inner regular with respect to /. Then for E ⊆ X the following are equiveridical:
(i) E ∈ Σ;
(ii) E ∩ K ∈ Σ whenever K ∈ Σ ∩ /;
(iii) µ
∗
(K ∩ E) +µ
∗
(K ¸ E) = µ
∗
K for every K ∈ /;
(iv) µ
∗
(K ∩ E) +µ
∗
(K ¸ E) = µ
∗
K for every K ∈ /;
(v) µ
∗
(E ∩ K) = µ
∗
(E ∩ K) for every K ∈ / ∩ Σ;
(vi) min(µ
∗
(K ∩ E), µ
∗
(K ¸ E)) < µK whenever K ∈ / ∩ Σ and 0 < µK < ∞;
(vii) max(µ
∗
(K ∩ E), µ
∗
(K ¸ E)) > 0 whenever K ∈ / ∩ Σ and µK > 0.
proof (a) Assume (i). Then of course E ∩ K ∈ Σ for every K ∈ Σ ∩ /, and (ii) is true. For any K ∈ /
there is an F ∈ Σ such that F ⊇ K and µF = µ
∗
K (132Aa); now
µ
∗
K ≤ µ
∗
(K ∩ E) +µ
∗
(K ¸ E) ≤ µ(F ∩ E) +µ(F ¸ E) = µF = µ
∗
K,
so (iii) is true. Next, for any K ∈ /,
µ
∗
(K ∩ E) +µ
∗
(K ¸ E) ≤ µ
∗
K = sup¦µF : F ∈ Σ
f
, F ⊆ K¦
(writing Σ
f
for ¦F : F ∈ Σ, µF < ∞¦)
= sup¦µ(F ∩ E) +µ(F ¸ E) : F ∈ Σ
f
, F ⊆ K¦
≤ µ
∗
(K ∩ E) +µ
∗
(K ¸ E).
So (iv) is true. If K ∈ / ∩ Σ, then
µ
∗
(E ∩ K) = sup¦µF : F ∈ Σ
f
, F ⊆ E ∩ K¦ = µ(E ∩ K) = µ
∗
(E ∩ K)
because µ is semiﬁnite. So (v) is true. Since (iii)⇒(vi) and (iv)⇒(vii), we see that all the conditions are
satisﬁed.
(b) Now suppose that E / ∈ Σ; I have to show that (ii)(vii) are all false. Because µ is locally determined,
there is an F ∈ Σ
f
such that E ∩ F / ∈ Σ. Take measurable envelopes H, H
t
of F ∩ E, F ¸ E respectively
(132Ee). Then F ¸ H
t
⊆ F ∩ E ⊆ F ∩ H, so
G = (F ∩ H) ¸ (F ¸ H
t
) = F ∩ H ∩ H
t
cannot be negligible. Take K ∈ / ∩ Σ such that K ⊆ G and µK > 0. As G ⊆ F, µK < ∞. Now
µ
∗
(K ∩ E) = µ
∗
(K ∩ F ∩ E) = µ(K ∩ H) = µK,
36 Topologies and measures I 413F
µ
∗
(K ¸ E) = µ
∗
(K ∩ F ¸ E) = µ(K ∩ H
t
) = µK.
But this means that
µ
∗
(K ∩ E) = µK −µ
∗
(K ¸ E) = 0, µ
∗
(K ¸ E) = µK −µ
∗
(K ∩ E) = 0,
by 413Eb. Now we see that this K witnesses that (ii)(vii) are all false.
413G The ideas of 413F can be used to give criteria for measurability of realvalued functions. I spell
out one which is particularly useful.
Lemma Let (X, Σ, µ) be a complete locally determined measure space and suppose that µ is inner regular
with respect to / ⊆ Σ. Suppose that f : X → R is a function, and for α ∈ R set E
α
= ¦x : f(x) ≤ α¦,
F
α
= ¦x : f(x) ≥ β¦. Then f is Σmeasurable iﬀ
min(µ
∗
(E
α
∩ K), µ
∗
(F
β
∩ K)) < µK
whenever K ∈ /, 0 < µK < ∞ and α < β.
proof (a) If f is measurable, then
µ
∗
(E
α
∩ K) +µ
∗
(F
β
∩ K) = µ(E
α
∩ K) +µ(F
β
∩ K) ≤ µK
whenever K ∈ Σ and α < β, so if 0 < µK < ∞ then we must have min(µ
∗
(E
α
∩ K), µ
∗
(F
β
∩ K)) < µK.
(b) If f is not measurable, then there is some α ∈ R such that E
α
is not measurable. 413F(iii) tells us
that there is a K ∈ / such that 0 < µK < ∞ and µ
∗
(E
α
∩ K) = µ
∗
(K ¸ E
α
) = µK. Note that K is a
measurable envelope of K ∩ E
α
(132Eb). Now ¸K ∩ F
α+2
−n)
n∈N
is a nondecreasing sequence with union
K ¸ E
α
, so there is some β > α such that K ∩F
β
is not negligible. Let H ⊆ K be a measurable envelope of
K ∩ F
β
, and K
t
∈ / such that K
t
⊆ H and µK
t
> 0; then
µ
∗
(K
t
∩ E
α
) = µ
∗
(K
t
∩ K ∩ E
α
) = µ(K
t
∩ K) = µK
t
,
µ
∗
(K
t
∩ F
β
) = µ
∗
(K
t
∩ H ∩ F
β
) = µ(K
t
∩ H) = µK
t
,
so K
t
, α and β witness that the condition is not satisﬁed.
413H Inner measure constructions based on 413C are important because they oﬀer an eﬃcient way of
setting up measures which are inner regular with respect to given families of sets. Two of the fundamental
results are 413I and 413J. I proceed by means of a lemma on ﬁnitely additive functionals.
Lemma Let X be a set and / a family of subsets of X such that
∅ ∈ /,
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡) K ∩ K
t
∈ / for all K, K
t
∈ /.
Let φ
0
: / → [0, ∞[ be a functional such that
(α) φ
0
K = φ
0
L + sup¦φ
0
K
t
: K ∈ /, K
t
⊆ K ¸ L¦ whenever K, L ∈ / and L ⊆ K.
Set
φA = sup¦φ
0
K : K ∈ /, K ⊆ A¦ for A ⊆ X,
Σ = ¦E : E ⊆ X, φA = φ(A∩ E) +φ(A¸ E) for every A ⊆ X¦.
Then Σ is an algebra of subsets of X, including /, and φΣ : Σ → [0, ∞] is an additive functional extending
φ
0
.
proof (a) To see that Σ is an algebra of subsets and φΣ is additive, all we need to know is that φ∅ = 0
(413B); and this is because, applying hypothesis (α) with K = L = ∅, φ
0
∅ = φ
0
∅ + φ
0
∅, so φ
0
∅ = 0. (α)
also assures us that φ
0
L ≤ φ
0
K whenever K, L ∈ / and L ⊆ K, so φK = φ
0
K for every K ∈ /.
(b) To check that / ⊆ Σ, we have a little more work to do. First, observe that (†) and (α) together tell
us that φ
0
(K ∪ K
t
) = φ
0
K +φ
0
K
t
for all disjoint K, K
t
∈ /. So if A, B ⊆ X and A∩ B = ∅ then
413I Inner measure constructions 37
φA+φB = sup
K∈K,K⊆A
φ
0
K + sup
L∈K,L⊆A
φ
0
L
= sup
K,L∈K,K⊆A,L⊆B
φ
0
(K ∪ L) ≤ φ(A∪ B).
(c) / ⊆ Σ. PPP Take K ∈ /, A ⊆ X. If L ∈ / and L ⊆ A, then
φ
0
L = φ
0
(K ∩ L) + sup¦φ
0
L
t
: L
t
∈ /, L
t
⊆ L ¸ K¦ ≤ φ(A∩ K) +φ(A¸ K).
(Note the use of the hypothesis (‡).) As L is arbitrary, φA ≤ φ(A∩ K) +φ(A¸ K). We already know that
φ(A∩ K) +φ(A¸ K) ≤ φA; as A is arbitrary, K ∈ Σ. QQQ
This completes the proof.
413I Theorem Let X be a set and / a family of subsets of X such that
∅ ∈ /,
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡)
n∈N
K
n
∈ / whenever ¸K
n
)
n∈N
is a sequence in /.
Let φ
0
: / → [0, ∞[ be a functional such that
(α) φ
0
K = φ
0
L + sup¦φ
0
K
t
: K ∈ /, K
t
⊆ K ¸ L¦ whenever K, L ∈ / and L ⊆ K,
(β) inf
n∈N
φ
0
K
n
= 0 whenever ¸K
n
)
n∈N
is a nonincreasing sequence in / with empty inter
section.
Then there is a unique complete locally determined measure µ on X extending φ
0
and inner regular with
respect to /.
proof (a) Set
φA = sup¦φ
0
K : K ∈ /, K ⊆ A¦ for A ⊆ X,
Σ = ¦E : E ⊆ X, φA = φ(A∩ E) +φ(A¸ E) for every A ⊆ X¦.
Then 413H tells us that Σ is an algebra of subsets of X, including /, and µ = φΣ is an additive functional
extending φ
0
.
(b) Now µ(
n∈N
K
n
) = inf
n∈N
µK
n
whenever ¸K
n
)
n∈N
is a nonincreasing sequence in /. PPP Set L =
n∈N
K
n
. Of course µL ≤ inf
n∈N
µK
n
. For the reverse inequality, take > 0. Then (α) tells us that there
is a K
t
∈ / such that K
t
⊆ K
0
¸ L and µK
0
≤ µL + µK
t
+ . Since ¸K
n
∩ K
t
)
n∈N
is a nonincreasing
sequence in / with empty intersection, (β) tells us that there is an n ∈ N such that µ(K
n
∩ K
t
) ≤ . Now
µK
0
−µL = µ(K
0
¸ L) = µ(K
0
¸ (K
t
∪ L)) +µK
t
≤ +µ(K
n
∩ K
t
) +µ(K
t
¸ K
n
) ≤ 2 +µ(K
0
¸ K
n
) = 2 +µK
0
−µK
n
.
(These calculations depend, of course, on the additivity of µ and the ﬁniteness of µK
0
.) So µL ≥ µK
n
−2.
As is arbitrary, µL = inf
n∈N
µK
n
. QQQ
(c) If ¸A
n
)
n∈N
is a nonincreasing sequence of subsets of X, with intersection A, and φA
0
< ∞, then
φA = inf
n∈N
φA
n
. PPP Of course φA ≤ φA
n
for every n. Given > 0, then for each n ∈ N choose K
n
∈ /
such that K
n
⊆ A
n
and φ
0
K
n
≥ φA
n
− 2
−n
(this is where I use the hypothesis that φA
0
is ﬁnite); set
L
n
=
i≤n
K
i
for each n, L =
n∈N
L
n
. Then we have
φA
n+1
−µL
n+1
= φA
n+1
−µ(K
n+1
∩ L
n
)
= φA
n+1
−µK
n+1
−µL
n
+µ(K
n+1
∪ L
n
)
≤ 2
−n−1
−µL
n
+φA
n
because K
n+1
⊆ A
n+1
⊆ A
n
and L
n
⊆ K
n
⊆ A
n
. Inducing on n, we see that µL
n
≥ φA
n
− 2 + 2
−n
for
every n. So
φA ≥ µL = inf
n∈N
µL
n
≥ inf
n∈N
φA
n
−2,
38 Topologies and measures I 413I
using (b) above for the middle equality. As is arbitrary, φA = inf
n∈N
φA
n
. QQQ
(d) It follows that φ is an inner measure. PPP The arguments of parts (a) and (b) of the proof of 413H
tell us that φ∅ = 0 and φ(A ∪ B) ≤ φA + φB whenever A, B ⊆ X are disjoint. We have just seen that
φ(
n∈N
A
n
) = inf
n∈N
φA
n
whenever ¸A
n
)
n∈N
is a nonincreasing sequence of sets and φA
0
< ∞. Finally,
φK = φ
0
K is ﬁnite for every K ∈ /, so φA = sup¦φB : B ⊆ A, φB < ∞¦ for every A ⊆ X. Putting these
together, φ is an inner measure. QQQ
(e) So 413C tells us that µ is a complete measure, and of course it is inner regular with respect to /,
by the deﬁnition of φ. It is semiﬁnite because µK = φ
0
K is ﬁnite for every K ∈ /. Now suppose that
E ⊆ X and that E ∩ F ∈ Σ whenever µF < ∞. Take any A ⊆ X. If L ∈ / and L ⊆ A, we have L ∈ Σ and
µL < ∞, so
φ
0
L = µL = µ(L ∩ E) +µ(L ¸ E) = φ(L ∩ E) +φ(L ¸ E) ≤ φ(A∩ E) +φ(A¸ E);
taking the supremum over L, φA ≤ φ(A ∩ E) + φ(A ¸ E). As A is arbitrary, E ∈ Σ; as E is arbitrary, µ is
locally determined.
(f ) Finally, µ is unique by 412L.
413J Theorem Let X be a set and / a family of subsets of X such that
∅ ∈ /,
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡) K ∩ K
t
∈ / whenever K, K
t
∈ /.
Let φ
0
: / → [0, ∞[ be a functional such that
(α) φ
0
K = φ
0
L + sup¦φ
0
K
t
: K ∈ /, K
t
⊆ K ¸ L¦ whenever K, L ∈ / and L ⊆ K,
(β) inf
n∈N
φ
0
K
n
= 0 whenever ¸K
n
)
n∈N
is a nonincreasing sequence in / with empty inter
section.
Then there is a unique complete locally determined measure µ on X extending φ
0
and inner regular with
respect to /
δ
, the family of sets expressible as intersections of sequences in /.
proof (a) Set
ψA = sup¦φ
0
K : K ∈ /, K ⊆ A¦ for A ⊆ X,
T = ¦E : E ⊆ X, ψA = ψ(A∩ E) +ψ(A¸ E) for every A ⊆ X¦.
Then 413H tells us that T is an algebra of subsets of X, including /, and ν = ψT is an additive functional
extending φ
0
.
(b) Write T
f
for ¦E : E ∈ T, νE < ∞¦. If ¸E
n
)
n∈N
is a nonincreasing sequence in T
f
with empty
intersection, lim
n→∞
νE
n
= 0. PPP Given > 0, we can choose a sequence ¸K
n
)
n∈N
in / such that K
n
⊆ E
n
and
νK
n
= φ
0
K
n
≥ νE
n
−2
−n
for each n. Set L
n
=
i≤n
K
i
for each n; then
lim
n→∞
νL
n
= lim
n→∞
φ
0
L
n
= 0
by hypothesis (β). But also, for each n,
νE
n
≤ νL
n
+
n
i=0
ν(E
i
¸ K
i
) ≤ νL
n
+ 2,
because ν is additive and nonnegative and E
n
⊆ L
n
∪
i≤n
(E
i
¸ K
i
). So limsup
n→∞
νE
n
≤ 2; as is
arbitrary, lim
n→∞
νE
n
= 0. QQQ
(c) Write T
f
δ
for the family of sets expressible as intersections of sequences in T
f
, and for H ∈ T
f
δ
set
φ
1
H = inf¦νE : H ⊆ E ∈ T¦. Note that because E ∩ F ∈ T
f
for every E, F ∈ T
f
, every member of T
f
δ
can be expressed as the intersection of a nonincreasing sequence in T
f
.
(i) If ¸E
n
)
n∈N
is a nonincreasing sequence in T
f
with intersection H ∈ T
f
δ
, φ
1
H = lim
n→∞
νE
n
. PPP
Of course
413J Inner measure constructions 39
φ
1
H ≤ inf
n∈N
νE
n
= lim
n→∞
νE
n
.
On the other hand, if H ⊆ E ∈ T, then ¸E
n
¸ E)
n∈N
is a nonincreasing sequence in T
f
with empty
intersection, and
νE ≥ lim
n→∞
ν(E
n
∩ E) = lim
n→∞
νE
n
−lim
n→∞
ν(E
n
¸ E) = lim
n→∞
νE
n
by (b) above. As E is arbitrary, φ
1
(
n∈N
E
n
) = lim
n→∞
νE
n
. QQQ
(ii) Because / ⊆ T
f
, /
δ
⊆ T
f
δ
. Now for any H ∈ T
f
δ
, φ
1
H = sup¦φ
1
L : L ∈ /
δ
, L ⊆ H¦. PPP Express
H as
n∈N
E
n
where ¸E
n
)
n∈N
is a nonincreasing sequence in T
f
. Given > 0, we can choose a sequence
¸K
n
)
n∈N
in / such that K
n
⊆ E
n
and νK
n
≥ νE
n
−2
−n
for each n. Setting L
n
=
i≤n
K
i
for each n and
L =
n∈N
L
n
, we have L ∈ /
δ
, L ⊆ H and
φ
1
H = lim
n→∞
νE
n
≤ lim
n→∞
(νL
n
+
n
i=0
ν(E
i
¸ K
i
)) ≤ φ
1
L + 2.
As is arbitrary, this gives the result. QQQ
(d) We ﬁnd that T
f
δ
, φ
1
satisfy the conditions of 413I. PPP Of course ∅ ∈ T
f
δ
. If G, H ∈ T
f
δ
and G∩H = ∅,
express them as
n∈N
E
n
,
n∈N
F
n
where ¸E
n
)
n∈N
, ¸F
n
)
n∈N
are nonincreasing sequences in T
f
. Then
G∪ H =
n∈N
E
n
∪ F
n
belongs to T
f
δ
, and
φ
1
(G∪ H) = lim
n→∞
ν(E
n
∪ F
n
) = lim
n→∞
νE
n
+νF
n
−ν(E
n
∩ F
n
)
= lim
n→∞
νE
n
+νF
n
(by (b))
= φ
1
G+φ
1
H.
The deﬁnition of T
f
δ
as the set of intersections of sequences in T
f
ensures that the intersection of any
sequence in T
f
δ
will belong to T
f
δ
.
Now suppose that G, H ∈ T
f
δ
and that G ⊆ H. Express them as intersections
n∈N
E
n
,
n∈N
F
n
of nonincreasing sequences in T
f
, so that φ
1
G = lim
n→∞
νE
n
, φ
1
H = lim
n→∞
νF
n
. For each n, set
H
n
=
m∈N
F
m
¸ E
n
, so that H
n
∈ T
f
δ
, H
n
⊆ H ¸ G, and
φ
1
H
n
= lim
m→∞
ν(F
m
¸ E
n
) = lim
m→∞
νF
m
−ν(F
m
∩ E
n
)
≥ lim
m→∞
νF
m
−νE
n
= φ
1
H −νE
n
.
Accordingly
sup¦φ
1
G
t
: G
t
∈ T
f
δ
, G
t
⊆ H ¸ G¦ ≥ sup
n∈N
φ
1
H −νE
n
= φ
1
H −φ
1
G.
On the other hand, if G
t
∈ T
f
δ
and G
t
⊆ H ¸ G, then
φ
1
G+φ
1
G
t
= φ
1
(G∪ G
t
) ≤ φ
1
H
because of course φ
1
is nondecreasing, as well as being additive on disjoint sets. So
sup¦φ
1
G
t
: G
t
∈ T
f
δ
, G
t
⊆ H ¸ G¦ = φ
1
H −φ
1
G
as required by condition (α) of 413I. Finally, suppose that ¸H
n
)
n∈N
is a nonincreasing sequence in T
f
δ
with
empty intersection. For each n ∈ N, let ¸E
ni
)
i∈N
be a nonincreasing sequence in T
f
with intersection
H
n
, and set F
m
=
n≤m
E
nn
for each m. Then ¸F
m
)
m∈N
is a nonincreasing sequence in T
f
with empty
intersection, while H
m
⊆ F
m
for each m, so
lim
m→∞
φ
1
H
m
≤ lim
m→∞
νF
m
= 0.
Thus condition 413I(β) is satisﬁed, and we have the full list. QQQ
40 Topologies and measures I 413J
(e) By 413I, we have a complete locally determined measure µ, extending φ
1
, and inner regular with
respect to T
f
δ
. Since φ
1
K = νK = φ
0
K for K ∈ /, µ extends φ
0
. If G belongs to the domain of µ, and
γ < µG, there is an H ∈ T
f
δ
such that H ⊆ G and γ < µH = φ
1
H; by (cii), there is an L ∈ /
δ
such that
L ⊆ H and γ ≤ φ
1
L = µL. Thus µ is inner regular with respect to /
δ
. To see that µ is unique, observe
that if µ
t
is any other measure with these properties, and L ∈ /
δ
, then L is expressible as
n∈N
K
n
where
¸K
n
)
n∈N
is a sequence in /. Now
µL = lim
n→∞
µ(
i≤n
K
i
) = lim
n→∞
φ
0
(
i≤n
K
i
) = µ
t
L.
So µ and µ
t
must agree on /
δ
, and by 412L they are identical.
413K Corollary (a) Let X be a set, Σ a subring of TX, and ν : Σ → [0, ∞[ a nonnegative ﬁnitely
additive functional such that lim
n→∞
νE
n
= 0 whenever ¸E
n
)
n∈N
is a nonincreasing sequence in Σ with
empty intersection. Then ν has a unique extension to a complete locally determined measure on X which
is inner regular with respect to the family Σ
δ
of intersections of sequences in Σ.
(b) Let X be a set, Σ a subalgebra of TX, and ν : Σ → [0, ∞[ a nonnegative ﬁnitely additive functional
such that lim
n→∞
νE
n
= 0 whenever ¸E
n
)
n∈N
is a nonincreasing sequence in Σ with empty intersection.
Then ν has a unique extension to a measure deﬁned on the σalgebra of subsets of X generated by Σ.
proof (a) Take Σ, ν in place of /, φ
0
in 413J.
(b) Let ν
1
be the complete extension as in (a), and let ν
t
1
be the restriction of ν
1
to the σalgebra Σ
t
generated by Σ; this is the extension required here. To see that ν
t
1
is unique, use the Monotone Class
Theorem (136C).
Remark You will sometimes see (b) above stated as ‘an additive functional on an algebra of sets extends
to a measure iﬀ it is countably additive’. But this formulation depends on a diﬀerent interpretation of the
phrase ‘countably additive’ from the one used in this book; see the note after the deﬁnition in 326E.
413L It will be useful to have a deﬁnition extending an idea in ¸342.
Deﬁnition A countably compact class (or semicompact paving) is a family / of sets such that
n∈N
K
n
,= ∅ whenever ¸K
n
)
n∈N
is a sequence in / such that
i≤n
K
i
,= ∅ for every n ∈ N.
413M Corollary Let X be a set and / a countably compact class of subsets of X such that
∅ ∈ /,
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡) K ∩ K
t
∈ / whenever K, K
t
∈ /.
Let φ
0
: / → [0, ∞[ be a functional such that
(α) φ
0
K = φ
0
L + sup¦φ
0
K
t
: K ∈ /, K
t
⊆ K ¸ L¦ whenever K, L ∈ / and L ⊆ K.
Then there is a unique complete locally determined measure µ on X extending φ
0
and inner regular with
respect to /
δ
, the family of sets expressible as intersections of sequences in /.
proof The point is that the hypothesis (β) of 413J is necessarily satisﬁed: if ¸K
n
)
n∈N
is a nonincreasing
sequence in / with empty intersection, then, because / is countably compact, there must be some n such
that K
n
= ∅. Since hypothesis (α) here is already enough to ensure that φ
0
∅ = 0, φ
0
K ≥ 0 for every K ∈ /,
we must have inf
n∈N
φ
0
K
n
= 0. So we apply 413J to get the result.
413N I now turn to constructions of a diﬀerent kind, being extension theorems in which the extension
is not uniquely deﬁned. Again I start with a theorem on ﬁnitely additive functionals.
Theorem Let X be a set, T
0
a subring of TX, and ν
0
: T
0
→ [0, ∞[ a ﬁnitely additive functional. Suppose
that / ⊆ TX is a family of sets such that
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡) K ∩ K
t
∈ / for all K, K
t
∈ /,
every member of / is included in some member of T
0
,
413N Inner measure constructions 41
and ν
0
is inner regular with respect to / in the sense that
(α) ν
0
E = sup¦ν
0
K : K ∈ / ∩ T
0
, K ⊆ E¦ for every E ∈ T
0
.
Then ν
0
has an extension to a nonnegative ﬁnitely additive functional ν
1
, deﬁned on a subring T
1
of TX
including T
0
∪/, inner regular with respect to /, and such that whenever E ∈ T
1
, > 0 there is an E
0
∈ T
0
such that ν
1
(E´E
0
) ≤ .
proof (a) Let P be the set of all nonnegative additive realvalued functionals ν, deﬁned on subrings of
TX, inner regular with respect to /, and such that
(∗) whenever E ∈ domν, > 0 there is an E
0
∈ T
0
such that ν(E´E
0
) ≤ .
Order P by extension of functions, so that P is a partially ordered set.
(b) It will be convenient to borrow some notation from the theory of countably additive functionals. If
T is a subring of TX and ν : T → [0, ∞[ is a nonnegative additive functional, set
ν
∗
A = inf¦νE : A ⊆ E ∈ T¦, ν
∗
A = sup¦νE : A ⊇ E ∈ T¦
for every A ⊆ X (interpreting inf ∅ as ∞ if necessary). Now if A ⊆ X and E, F ∈ T are disjoint,
ν
∗
(A∩ (E ∪ F)) = ν
∗
(A∩ E) +ν
∗
(A∩ F),
ν
∗
(A∩ (E ∪ F)) = ν
∗
(A∩ E) +ν
∗
(A∩ F).
PPP ν
∗
(A∩ (E ∪ F)) = inf¦νG : G ∈ T, A∩ (E ∪ F) ⊆ G¦
= inf¦νG : G ∈ T, A∩ (E ∪ F) ⊆ G ⊆ E ∪ F¦
= inf¦ν(G∩ E) +ν(G∩ F) : G ∈ T, A∩ (E ∪ F) ⊆ G ⊆ E ∪ F¦
= inf¦νG
1
+νG
2
: G
1
, G
2
∈ T, A∩ E ⊆ G
1
⊆ E, A∩ F ⊆ G
2
⊆ F¦
= inf¦νG
1
: G
1
∈ T, A∩ E ⊆ G
1
⊆ E¦
+ inf¦νG
2
: G
2
∈ T, A∩ F ⊆ G
2
⊆ F¦
= ν
∗
(E ∩ A) +ν
∗
(F ∩ A),
ν
∗
(A∩ (E ∪ F)) = sup¦νG : G ∈ T, A∩ (E ∪ F) ⊇ G¦
= sup¦ν(G∩ E) +ν(G∩ F) : G ∈ T, A∩ (E ∪ F) ⊇ G¦
= sup¦νG
1
+νG
2
: G
1
, G
2
∈ T, A∩ E ⊇ G
1
, A∩ F ⊇ G
2
¦
= sup¦νG
1
: G
1
∈ T, A∩ E ⊇ G
1
¦
+ sup¦νG
2
: G
2
∈ T, A∩ F ⊇ G
2
¦
= ν
∗
(E ∩ A) +ν
∗
(F ∩ A). QQQ
(c) The key to the proof is the following fact: if ν ∈ P and M ∈ /, there is a ν
t
∈ P such that ν
t
extends
ν and M ∈ domν
t
. PPP Set T = domν, T
t
= ¦(E ∩ M) ∪ (F ¸ M) : E, F ∈ T¦. For H ∈ T
t
, set
ν
t
H = ν
∗
(H ∩ M) +ν
∗
(H ¸ M).
Now we have to check the following.
(i) T
t
is a subring of TX, because if E, F, E
t
, F
t
∈ T then
((E ∩ M) ∪ (F ¸ M)) ∗ ((E
t
∩ M) ∪ (F
t
¸ M)) = ((E ∗ E
t
) ∩ M) ∪ ((F ∗ F
t
) ¸ M)
for both the Boolean operations ∗ = ´ and ∗ = ∩. T
t
⊇ T because E = (E∩M) ∪(E¸M) for every E ∈ T.
(Cf. 312M.) M ∈ T
t
because there is some E ∈ T
0
such that M ⊆ E, so that M = (E ∩M) ∪(∅ ¸ M) ∈ T
t
.
(ii) ν
t
is ﬁnitevalued because if H = (E∩M) ∪(F ¸ M), where E, F ∈ T, then ν
t
H ≤ νE+νF. If H,
H
t
∈ T are disjoint, they can be expressed as (E ∩ M) ∪ (F ¸ M), (E
t
∩ M) ∪ (F
t
¸ M) where E, F, E
t
, F
t
belong to T; replacing E
t
, F
t
by E
t
¸ E and F
t
¸ F if necessary, we may suppose that E ∩E
t
= F ∩F
t
= ∅.
Now
42 Topologies and measures I 413N
ν
t
(H ∪ H
t
) = ν
∗
((E ∪ E
t
) ∩ M) +ν
∗
((F ∪ F
t
) ∩ (X ¸ M))
= ν
∗
(E ∩ M) +ν
∗
(E
t
∩ M) +ν
∗
(F ∩ (X ¸ M)) +ν
∗
(F
t
∩ (X ¸ M))
(by (b) above)
= ν
t
H +ν
t
H
t
.
Thus ν
t
is additive.
(iii) If E ∈ T, then
ν
∗
(E ¸ M) = sup¦νF : F ∈ T, F ⊆ E ¸ M¦
= sup¦νE −ν(E ¸ F) : F ∈ T, F ⊆ E ¸ M¦
= sup¦νE −νF : F ∈ T, E ∩ M ⊆ F ⊆ E¦
= νE −inf¦νF : F ∈ T, E ∩ M ⊆ F ⊆ E¦ = νE −ν
∗
(E ∩ M).
So
ν
t
E = ν
∗
(E ∩ M) +ν
∗
(E ¸ M) = νE.
Thus ν
t
extends ν.
(iv) If H ∈ T
t
and > 0, express H as (E ∩ M) ∪ (F ¸ M), where E, F ∈ T. Then we can ﬁnd (α) a
K ∈ /∩T such that K ⊆ E and ν(E¸K) ≤ (β) an F
t
∈ T such that F
t
⊆ F ¸M and νF
t
≥ ν
∗
(F ¸M) −
(γ) a K
t
∈ / ∩ T such that K
t
⊆ F
t
and νK
t
≥ νF
t
− . Set L = (K ∩ M) ∪ K
t
∈ T
t
; by the hypotheses
(†) and (‡), L ∈ /. Now L ⊆ H and
ν
t
L = ν
t
(K ∩ M) +ν
t
K
t
= ν
t
(E ∩ M) −ν
t
((E ¸ K) ∩ M) +νK
t
= ν
∗
(H ∩ M) −ν
∗
((E ¸ K) ∩ M) +νK
t
≥ ν
∗
(H ∩ M) −ν(E ¸ K) +νF
t
−
≥ ν
∗
(H ∩ M) +ν
∗
(F ¸ M) −3 = ν
t
H −3.
As H and are arbitrary, ν is inner regular with respect to /.
(v) Finally, given H ∈ T
t
and > 0, take E, F ∈ T such that H ∩ M ⊆ E, F ⊆ H ¸ M, νE ≤
ν
∗
(H ∩ M) + and νF ≥ ν
∗
(H ¸ M) −. In this case,
ν
t
(E ¸ (H ∩ M)) = ν
t
E −ν
t
(H ∩ M) = νE −ν
∗
(H ∩ M) ≤ ,
ν
t
((H ¸ M) ¸ F) = ν
t
(H ¸ M) −ν
t
F = ν
∗
(H ¸ M) −νF ≤ .
But as
H´(E ∪ F) ⊆ (E ¸ (H ∩ M)) ∪ ((H ¸ M) ¸ F),
ν
t
(H´(E∪F)) ≤ 2. Now ν satisﬁes the condition (∗), so there is an E
0
∈ T
0
such that ν((E∪F)´E
0
) ≤ ,
and ν
t
(H´E
0
) ≤ 3. As H and are arbitrary, ν
t
satisﬁes (∗).
This completes the proof that ν
t
is a member of P extending ν. QQQ
(d) It is easy to check that if Q ⊆ P is a nonempty totally ordered subset, the smallest common extension
ν
t
of the functions in Q belongs to P. (To see that ν
t
is inner regular with respect to /, observe that if
E ∈ domν
t
and γ < ν
t
E, there is some ν ∈ Q such that E ∈ domν; now there is a K ∈ /∩domν such that
K ⊆ E and νK ≥ γ, so that K ∈ /∩domν
t
and ν
t
K ≥ γ.) And of course P is not empty, because ν
0
∈ P.
So by Zorn’s Lemma P has a maximal element ν
1
say; write T
1
for the domain of ν
1
. If M ∈ / there is
an element of P, with a domain containing M, extending ν
1
; as ν
1
is maximal, this must be ν
1
itself, so
M ∈ T
1
. Thus / ⊆ T
1
, and ν
1
has all the required properties.
413P Inner measure constructions 43
413O Corollary Let (X, Σ
0
, µ
0
) be a measure space and / a countably compact class of subsets of X
such that
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡)
n∈N
K
n
∈ / for every sequence ¸K
n
)
n∈N
in /,
µ
∗
0
K < ∞ for every K ∈ /,
µ
0
is inner regular with respect to /.
Then µ
0
has an extension to a complete locally determined measure µ, deﬁned on every member of /, inner
regular with respect to /, and such that whenever E ∈ domµ and µE < ∞ there is an E
0
∈ Σ
0
such that
µ(E´E
0
) = 0.
proof (a) Set T
0
= ¦E : E ∈ Σ
0
, µ
0
E < ∞¦, ν
0
= µ
0
T
0
. Then ν
0
, T
0
satisfy the conditions of 413N; take
ν
1
, T
1
as in 413N. If K, L ∈ / and L ⊆ K, then
ν
1
L + sup¦ν
1
K
t
: K
t
∈ /, K
t
⊆ K ¸ L¦ = ν
1
L +ν
1
(K ¸ L) = ν
1
K.
So ν
1
/ satisﬁes the conditions of 413M and there is a complete locally determined measure µ, extending
ν
1
/, and inner regular with respect to /.
(b) Write Σ for the domain of µ. Then T
1
⊆ Σ. PPP If E ∈ T
1
and K ∈ /,
µ
∗
(K ∩ E) +µ
∗
(K ¸ E)
≥ sup¦µK
t
: K
t
∈ /, K
t
⊆ K ∩ E¦ + sup¦µK
t
: K
t
∈ /, K
t
⊆ K ¸ E¦
= sup¦ν
1
K
t
: K
t
∈ /, K
t
⊆ K ∩ E¦ + sup¦ν
1
K
t
: K
t
∈ /, K
t
⊆ K ¸ E¦
= ν
1
(K ∩ E) +ν
1
(K ¸ E) = ν
1
K = µK.
By 413F(iv), E ∈ Σ. QQQ It follows at once that µ extends ν
1
, since if E ∈ T
1
ν
1
E = sup¦ν
1
K : K ∈ /, K ⊆ E¦ = sup¦µK : K ∈ /, K ⊆ E¦ = µE.
(c) In particular, µ agrees with µ
0
on T
0
. Now in fact µ extends µ
0
. PPP Take E ∈ Σ
0
. If K ∈ /, there is
an E
0
∈ Σ
0
such that K ⊆ E
0
and µ
0
E
0
< ∞. Since E ∩ E
0
∈ T
0
⊆ Σ, E ∩ K = E ∩ E
0
∩ K ∈ Σ. As K is
arbitrary, E ∈ Σ, by 413F(ii). Next, because every member of / is included in a member of T
0
,
µ
0
E = sup¦µ
0
K : K ∈ / ∩ Σ
0
, K ⊆ E¦ = sup¦µ
0
(E ∩ E
0
) : E
0
∈ T
0
¦
= sup¦µ(E ∩ E
0
) : E
0
∈ T
0
¦ = sup¦µK : K ∈ /, K ⊆ E¦ = µE. QQQ
(d) Finally, suppose that E ∈ Σ and µE < ∞. For each n ∈ N we can ﬁnd K
n
∈ /, E
n
∈ Σ
0
such that
K
n
⊆ E, µ(E ¸ K
n
) ≤ 2
−n
and ν
1
(K
n
´E
n
) ≤ 2
−n
. In this case
∞
n=0
µ(E
n
´E) < ∞, so µ(E´E
t
) = 0,
where E
t
=
n∈N
m≥n
E
m
∈ Σ
0
.
Thus µ has all the required properties.
413P I now describe an alternative route to some of the applications of 413N. As before, I do as much
as possible in the context of ﬁnitely additive functionals.
Lemma Let X be a set and / a sublattice of TX containing ∅. Let λ : / → [0, ∞[ be a bounded functional
such that
λ∅ = 0, λK ≤ λK
t
whenever K, K
t
∈ / and K ⊆ K
t
,
λ(K ∪ K
t
) +λ(K ∩ K
t
) ≥ λK +λK
t
for all K, K
t
∈ /.
Then there is a ﬁnitely additive functional ν : TX → [0, ∞[ such that
νX = sup
K∈K
λK, νK ≥ λK for every K ∈ /.
proof (a) The key to the proof is the following fact: if K
0
, . . . , K
n
∈ /, then there are K
t
0
, . . . , K
t
n
∈ /
such that K
t
0
⊆ K
t
1
⊆ . . . ⊆ K
t
n
,
n
i=0
χK
t
i
=
n
i=0
χK
i
and
n
i=0
λK
t
i
≥
n
i=0
λK
i
. PPP Induce on n. If
44 Topologies and measures I 413P
n = 0 the result is trivial. For the inductive step to n + 1, given K
0
, . . . , K
n+1
∈ /, set L
n+1
= K
n+1
and
use the inductive hypothesis to ﬁnd L
0
, . . . , L
n
∈ / such that
L
0
⊆ . . . ⊆ L
n
,
n
i=0
χL
i
=
n
i=0
χK
i
and
n
i=0
λL
i
≥
n
i=0
λK
i
.
Now set L
t
n
= L
n+1
∩L
n
, L
t
i
= L
i
for i < n, and use the inductive hypothesis again to ﬁnd K
t
0
, . . . , K
t
n
∈ /
such that
K
t
0
⊆ . . . ⊆ K
t
n
,
n
i=0
χK
t
i
=
n
i=0
χL
t
i
and
n
i=0
λK
t
i
≥
n
i=0
λL
t
i
.
Set
K
t
n+1
= L
t
n+1
= L
n
∪ L
n+1
.
Then all the L
i
, L
t
i
and K
t
i
belong to /. We know that χK
t
n
≤
n
i=0
χL
t
i
, so
K
t
n
⊆
i≤n
L
t
i
⊆ L
n
⊆ K
t
n+1
;
accordingly we have K
t
0
⊆ K
t
1
⊆ . . . ⊆ K
t
n
⊆ K
t
n+1
. Next,
n+1
i=0
χK
t
i
=
n
i=0
χL
t
i
+χL
t
n+1
=
n−1
i=0
χL
i
+χ(L
n
∩ L
n+1
) +χ(L
n
∪ L
n+1
)
=
n−1
i=0
χL
i
+χL
n
+χL
n+1
=
n
i=0
χL
i
+χK
n+1
=
n+1
i=0
χK
i
,
n+1
i=0
λK
t
i
≥
n
i=0
λL
t
i
+λL
t
n+1
=
n−1
i=0
λL
i
+λ(L
n
∩ L
n+1
) +λ(L
n
∪ L
n+1
)
≥
n−1
i=0
λL
i
+λL
n
+λL
n+1
(using the hypothesis on λ)
=
n
i=0
λL
i
+λK
n+1
≥
n+1
i=0
λK
i
,
so we have an appropriate family K
t
0
, . . . , K
t
n+1
, and the induction continues. QQQ
(b) For the moment, suppose that sup
K∈K
λK = 1. In this case, if ¸K
i
)
i∈I
is a ﬁnite indexed family in
/, there is a set J ⊆ I such that #(J) ≥
i∈I
λK
i
and
i∈J
K
i
,= ∅. PPP If I = ∅ this is trivial, so we may
suppose that I = ¦0, . . . , n¦ for some n. Take K
t
0
, . . . , K
t
n
as in (a). If K
t
n
= ∅ then every K
i
is empty so
(because λ∅ = 0)
n
i=0
λK
i
= 0 and we may take J = ∅. Otherwise, let m be the ﬁrst number such that
K
t
m
,= ∅, and take x ∈ K
t
m
, J = ¦i : i ≤ n, x ∈ K
i
¦. Then
i∈J
K
i
,= ∅ and
#(J) =
n
i=0
χK
i
(x) =
n
i=0
χK
t
i
(x) = n −m+ 1
≥
n
i=m
λK
t
i
=
n
i=0
λK
t
i
≥
n
i=0
λK
i
,
as claimed. QQQ
By 391F, there is an additive functional ν : TX → [0, 1] such that νX = 1 and νK ≥ λK for every
K ∈ /, as required.
(c) For the general case, set γ = sup
K∈K
λK. If γ = 0, take ν to be the zero functional; if γ > 0, apply
(a)(b) to the functional γ
−1
λ.
Remark If P is a lattice, a function f : P →R such that f(p ∨ q) +f(p ∧ q) ≥ f(p) +f(q) for all p, q ∈ P
is called supermodular.
413Q Inner measure constructions 45
413Q Theorem Let X be a set and / a sublattice of TX containing ∅. Let Σ be the algebra of subsets
of X generated by /, and ν
0
: Σ → [0, ∞[ a ﬁnitely additive functional. Then there is a ﬁnitely additive
functional ν : Σ → [0, ∞[ such that (i) νX = sup
K∈K
ν
0
K (ii) νK ≥ ν
0
K for every K ∈ / (iii) ν is inner
regular with respect to / in the sense that νE = sup¦νK : K ∈ /, K ⊆ E¦ for every E ∈ Σ.
proof (a) Set γ = sup
K∈K
ν
0
K. Let P be the set of all functionals λ : / → [0, γ] such that
λK +λK
t
≤ λ(K ∪ K
t
) +λ(K ∩ K
t
)
for every K, K
t
∈ /. Give P the natural partial ordering inherited from R
K
. Note that ν
0
/ belongs
to P. If Q ⊆ P is nonempty and upwardsdirected, then sup Q, taken in R
K
, belongs to P; so there is a
maximal λ ∈ P such that ν
0
/ ≤ λ. By 413P, there is a nonnegative additive functional ν on TX such
that νK ≥ λK for every K ∈ / and νX = γ. Since ν/ also belongs to P, we must have νK = λK for
every K ∈ /.
(b) Now for any K
0
∈ /,
νK
0
+ sup¦νL : L ∈ /, L ⊆ X ¸ K
0
¦ = γ.
PPP (i) Set L = ¦L : L ∈ /, L ⊆ X ¸ K
0
¦. For A ⊆ X, set θ
0
A = sup
L∈1
ν(A ∩ L). Because L is
upwardsdirected, θ
0
: TX →R is additive, and of course 0 ≤ θ
0
≤ ν. Set θ
1
= ν −θ
0
, so that θ
1
is another
additive functional, and write
λ
t
K = θ
0
K + sup¦θ
1
M : M ∈ /, M ∩ K
0
⊆ K¦
for K ∈ /.
(ii) If K, K
t
∈ / and > 0, there are M, M
t
∈ / such that M ∩ K
0
⊆ K, M
t
∩ K
0
⊆ K
t
and
θ
0
K +θ
1
M ≥ λ
t
K −, θ
0
K
t
+θ
1
M
t
≥ λ
t
K
t
−.
Now
M ∪ M
t
, M ∩ M
t
∈ /,
(M ∪ M
t
) ∩ K
0
⊆ K ∪ K
t
, (M ∩ M
t
) ∩ K
0
⊆ K ∩ K
t
,
so
λ
t
(K ∪ K
t
) +λ
t
(K ∩ K
t
) ≥ θ
0
(K ∪ K
t
) +θ
1
(M ∪ M
t
) +θ
0
(K ∩ K
t
) +θ
1
(M ∩ M
t
)
= θ
0
K +θ
1
M +θ
0
K
t
+θ
1
M
t
≥ λ
t
K +λ
t
K
t
−2.
As is arbitrary, λ
t
(K ∪ K
t
) +λ
t
(K ∩ K
t
) ≥ λK +λK
t
.
(iii) Suppose that K, M ∈ / are such that M ∩ K
0
⊆ K. If L ∈ L, then
ν(K ∩ L) +θ
1
M = ν(K ∩ L) +νM −θ
0
M
= ν(M ∩ K ∩ L) +ν(M ∪ (K ∩ L)) −θ
0
M ≤ γ
because K ∩ L ∈ L; taking the supremum over L and M, λ
t
K ≤ γ. As K is arbitrary, λ
t
∈ P.
(iv) If K ∈ /, then of course K ∩ K
0
⊆ K, so
λ
t
K ≥ θ
0
K +θ
1
K = νK = λK.
Thus λ
t
≥ λ. Because λ is maximal, λ
t
= λ. But this means that
λK
0
= λ
t
K
0
= θ
0
K
0
+ sup¦θ
1
M : M ∈ /, M ∩ K
0
⊆ K
0
¦ = sup
M∈K
θ
1
M.
Now given > 0 there is an M ∈ / such that
γ − ≤ ν
0
M ≤ λM = νM,
so that
νK
0
= λK
0
≥ θ
1
M = νM −θ
0
M ≥ γ − −θ
0
M ≥ γ − −sup
L∈1
νL,
46 Topologies and measures I 413Q
and νK
0
+sup
L∈1
νL ≥ γ −. As is arbitrary, νK
0
+sup
L∈1
νL ≥ γ. But of course νK
0
+νL ≤ νX = γ
for every L ∈ L, so νK
0
+ sup
L∈1
νL = γ, as claimed. QQQ
(c) It follows that if K, L ∈ / and L ⊆ K,
νK = νL + sup¦νK
t
: K
t
∈ /, K
t
⊆ K ¸ L¦.
PPP Because ν is additive and nonnegative, we surely have
νK ≥ νL + sup¦νK
t
: K
t
∈ /, K
t
⊆ K ¸ L¦.
On the other hand, given > 0, there is an M ∈ / such that M ⊆ X ¸ L and νL + νM ≥ γ − , so that
M ∩ K ∈ /, M ∩ K ⊆ K ¸ L and
νL +ν(M ∩ K) = νL +νK +νM −ν(M ∪ K) ≥ νK +γ − −γ = νK −.
As is arbitrary,
νK ≤ νL + sup¦νK
t
: K
t
∈ /, K
t
⊆ K ¸ L¦
and we have equality. QQQ
(d) By 413H, we have an additive functional ν
t
: Σ → [0, ∞[ such that ν
t
E = sup¦νK : K ∈ /, K ⊆ E¦
for every E ∈ Σ. It is easy to show that ν
t
and ν must agree on Σ, but even without doing so we can see
that ν
t
has the properties (i)(iii) required in the theorem.
413R The following lemma on countably compact classes, corresponding to 342Db, will be useful.
Lemma (Marczewski 53) Let / be a countably compact class of sets. Then there is a countably compact
class /
∗
⊇ / such that K ∪ L ∈ /
∗
and
n∈N
K
n
∈ /
∗
whenever K, L ∈ /
∗
and ¸K
n
)
n∈N
is a sequence in
/
∗
.
proof (a) Write /
s
for ¦K
0
∪. . . ∪K
n
: K
0
, . . . , K
n
∈ /¦. Then /
s
is countably compact. PPP Let ¸L
n
)
n∈N
be a sequence in /
s
such that
i≤n
L
i
, = ∅ for each n ∈ N. Then there is an ultraﬁlter T on X =
/
containing every L
n
. For each n, L
n
is a ﬁnite union of members of /, so there must be a K
n
∈ / such
that K
n
⊆ L
n
and K
n
∈ T. Now
i≤n
K
i
,= ∅ for every n, so
n∈N
K
n
,= ∅ and
n∈N
L
n
,= ∅. As ¸L
n
)
n∈N
is arbitrary, /
s
is countably compact. QQQ
Note that L ∪ L
t
∈ /
s
for all L, L
t
∈ /
s
.
(b) Write /
∗
for
¦
L
0
: L
0
⊆ /
s
is nonempty and countable¦.
Then /
∗
is countably compact. PPP If ¸M
n
)
n∈N
is any sequence in /
∗
such that
i≤n
M
i
,= ∅ for every n ∈ N,
then for each n ∈ N let L
n
⊆ /
s
be a countable nonempty set such that M
n
=
L
n
. Let ¸L
n
)
n∈N
be a
sequence running over
n∈N
L
n
; then
i≤n
L
i
,= ∅ for every n, so
n∈N
L
n
=
n∈N
M
n
is nonempty. As
¸M
n
)
n∈N
is arbitrary, /
∗
is countably compact. QQQ
(c) Of course / ⊆ /
s
⊆ /
∗
. It is immediate from the deﬁnition of /
∗
that it is closed under countable
intersections. Finally, if M
1
, M
2
∈ /
∗
, let L
1
, L
2
⊆ /
s
be countable sets such that M
1
=
L
1
and
M
2
=
L
2
; then L = ¦L
1
∪L
2
: L
1
∈ L
1
, L
2
∈ L
2
¦ is a countable subset of /
s
, so M
1
∪M
2
=
L belongs
to /
∗
.
413S Corollary Let X be a set and / a countably compact class of subsets of X. Let T be a subalgebra
of TX and ν : T →R a nonnegative ﬁnitely additive functional.
(a) There is a complete measure µ on X such that µX ≤ νX, / ⊆ domµ and µK ≥ νK for every
K ∈ / ∩ T.
(b) If
(†) K ∪ K
t
∈ / whenever K, K
t
∈ /,
(‡)
n∈N
K
n
∈ / for every sequence ¸K
n
)
n∈N
in /,
we may arrange that µ is inner regular with respect to /.
413Xj Inner measure constructions 47
proof By 413R, there is always a countably compact class /
∗
⊇ / satisfying (†) and (‡); for case (b), take
/
∗
= /. By 391G, there is an extension of ν to a ﬁnitely additive functional ν
t
: TX → R. Let T
1
be the
subalgebra of TX generated by /
∗
. By 413Q, there is a nonnegative additive functional ν
1
: T
1
→R such
that ν
1
X ≤ ν
t
X = νX, ν
1
K ≥ ν
t
K = νK for every K ∈ /
∗
∩ T and ν
1
E = sup¦ν
1
K : K ∈ /
∗
, K ⊆ E¦
for every E ∈ T
1
. In particular, if K, L ∈ /
∗
,
ν
1
L + sup¦ν
1
K
t
: K
t
∈ /
∗
, K
t
⊆ K ¸ L¦ = ν
1
L +ν
1
(K ¸ L) = ν
1
K.
So /
∗
and ν
1
/
∗
satisfy the hypotheses of 413M. Accordingly we have a complete measure µ extending
ν
1
/
∗
and inner regular with respect to /
∗
= /
∗
δ
; in which case
µK = ν
1
K ≥ νK
for every K ∈ / ∩ T,
µX = sup
K∈K
∗ µK = sup
K∈K
∗ ν
1
K ≤ ν
1
X ≤ νX,
as required.
413X Basic exercises (a) Deﬁne φ : TN → [0, ∞[ by setting φA = 0 if A is ﬁnite, ∞ otherwise. Check
that φ satisﬁes conditions (α) and (β) of 413A, but that if we attempt to reproduce the construction of
413C then we obtain Σ = TN, µ = φ, so that µ is not countably additive.
(b) Let φ
1
, φ
2
be two inner measures on a set X, inducing measures µ
1
, µ
2
by the method of 413C. (i)
Show that φ = φ
1
+φ
2
is an inner measure. (ii) Show that the measure µ induced by φ extends the measure
µ
1
+µ
2
deﬁned on domµ
1
∩ domµ
2
.
>>>(c) Let X be a set, φ an inner measure on X, and µ the measure constructed from it by the method
of 413C. (i) Let Y be a subset of X. Show that φTY is an inner measure on Y , and that the measure on
Y deﬁned from it extends the subspace measure µ
Y
induced on Y by µ. (ii) Let Y be a set and f : X → Y
a function. Show that B → φf
−1
[B] is an inner measure on Y , and that it deﬁnes a measure on Y which
extends the image measure µf
−1
.
(d) Let (X, Σ, µ) be any measure space. Set θA =
1
2
(µ
∗
A + µ
∗
A) for every A ⊆ X. Show that θ is an
outer measure on X, and that the measure deﬁned from θ by Carath´eodory’s method extends µ.
>>>(e) Show that there is a partition ¸A
n
)
n∈N
of [0, 1] such that µ
∗
(
i≤n
A
i
) = 0 for every n, where µ
∗
is
Lebesgue inner measure. (Hint: set A
n
= (A+q
n
) ∩[0, 1] where ¸q
n
)
n∈N
is an enumeration of Q and A is a
suitable set; cf. 134B.)
(f ) Let (X, Σ, µ) be a measure space. (i) Show that µ
∗
Σ is the semiﬁnite version µ
sf
of µ as constructed
in 213Xc. (ii) Show that if A is any subset of X, and Σ
A
the subspace σalgebra, then µ
∗
Σ
A
is a semiﬁnite
measure on A.
>>>(g) Let (X, Σ, µ) and (Y, T, ν) be two measure spaces, and λ the c.l.d. product measure on X Y .
Show that λ
∗
(A B) = µ
∗
A ν
∗
B for all A ⊆ X and B ⊆ Y . (Hint: use Fubini’s theorem to show that
λ
∗
(AB) ≤ µ
∗
A ν
∗
B.)
(h) Let X be a set and µ, ν two complete locally determined measures on X with domains Σ, T
respectively, both inner regular with respect to / ⊆ Σ ∩ T. Suppose that, for K ∈ /, µK = 0 iﬀ νK = 0.
Show that Σ = T and that µ and ν give rise to the same negligible sets.
>>>(i) Let (X, T, ν) be a measure space. (i) Show that the measure constructed by the method of 413C
from the inner measure ν
∗
is the c.l.d. version of ν. (ii) Set / = ¦E : E ∈ T, νE < ∞¦, φ
0
= ν/. Show
that /, φ
0
satisfy the conditions of 413I, and that the measure constructed by the method there is again
the c.l.d. version of ν.
(j) Let (X, Σ, µ) be a complete locally determined measure space and L a family of subsets of X such
that µ is inner regular with respect to L. Set / = ¦K : K ∈ L∩Σ, µK < ∞¦ and φ
0
= µ/. Show that /,
φ
0
satisfy the conditions of 413I and that the measure constructed from /, φ
0
by the method there is just
µ.
48 Topologies and measures I 413Xk
>>>(k) Let / be the family of subsets of R expressible as ﬁnite unions of bounded closed intervals. (i) Show
from ﬁrst principles that there is a unique functional φ
0
: / → [0, ∞[ such that φ
0
[α, β] = β −α whenever
α ≤ β and φ
0
satisﬁes the conditions of 413J. (ii) Show that the measure on R constructed from φ
0
by the
method of 413J is Lebesgue measure.
(l) Let X be a set, Σ a subring of TX, and ν : Σ → [0, ∞[ a nonnegative additive functional such that
lim
n→∞
νE
n
= 0 whenever ¸E
n
)
n∈N
is a nonincreasing sequence in Σ with empty intersection, as in 413K.
Deﬁne θ : TX → [0, ∞] by setting
θA = inf¦
∞
n=0
νE
n
: ¸E
n
)
n∈N
is a sequence in Σ covering A¦
for A ⊆ X, interpreting inf ∅ as ∞ if necessary. Show that θ is an outer measure. Let µ
θ
be the measure
deﬁned from θ by Carath´eodory’s method. Show that the measure deﬁned from ν by the process of 413K is
the c.l.d. version of µ
θ
. (Hint: the c.l.d. version of µ
θ
is inner regular with respect to Σ
δ
.)
>>>(m) Let X be a set, Σ a subring of TX, and ν : Σ → [0, ∞[ a nonnegative additive functional. Show
that the following are equiveridical: (i) ν has an extension to a measure on X; (ii) lim
n→∞
νE
n
= 0 whenever
¸E
n
)
n∈N
is a nonincreasing sequence in Σ with empty intersection; (iii) ν(
n∈N
E
n
) =
∞
n=0
νE
n
whenever
¸E
n
)
n∈N
is a disjoint sequence in Σ such that
n∈N
E
n
∈ Σ.
(n) Let ¸(X
n
, Σ
n
, µ
n
))
n∈N
be a sequence of probability spaces, and T a nonprincipal ultraﬁlter on N.
For x, y ∈
n∈N
X
n
, write x ∼ y if ¦n : x(n) = y(n)¦ ∈ T. (i) Show that ∼ is an equivalence relation;
write X for the set of equivalence classes, and x
•
∈ X for the equivalence class of x ∈
n∈N
X
n
. (Compare
351M.) (ii) Let Σ be the set of subsets of X expressible in the form Q(¸E
n
)
n∈N
) = ¦x
•
: x ∈
n∈N
E
n
¦,
where E
n
∈ Σ
n
for each n ∈ N. Show that Σ is an algebra of subsets of X, and that there is a welldeﬁned
additive functional ν : Σ → [0, 1] deﬁned by setting ν(Q(¸E
n
)
n∈N
)) = lim
n→J
µ
n
E
n
. (iii) Show that for any
nondecreasing sequence ¸H
i
)
i∈N
in Σ there is an H ∈ Σ such that H ⊆
i∈N
H
i
and νH = lim
n→∞
νH
n
.
(Hint: express each H
i
as Q(¸E
in
)
n∈N
). Do this in such a way that E
i+1,n
⊆ E
in
for all i, n. Take a
decreasing sequence ¸J
i
)
i∈N
in T, with empty intersection, such that νH
i
≤ µE
in
+ 2
−i
for n ∈ J
i
. Set
E
n
= E
in
for n ∈ J
i
¸ J
i+1
.) (iv) Show that there is a unique extension of ν to a complete probability
measure µ on X which is inner regular with respect to Σ. (This is a kind of Loeb measure.)
(o) Let A be a Boolean algebra and K ⊆ A a sublattice containing 0. Suppose that λ : K → [0, ∞[ is a
bounded functional such that λ0 = 0, λa ≤ λa
t
whenever a, a
t
∈ K and a ⊆ a
t
, and λ(a ∪ a
t
) + λ(a ∩ a
t
) ≥
λa + λa
t
for all a, a
t
∈ K. Show that there is a nonnegative additive functional ν : A → R such that
νa ≥ λa for every a ∈ K and ν1 = sup
a∈K
λa.
(p) Let X be a set and / a sublattice of TX containing ∅. Let λ : / →R be a bounded orderpreserving
function such that λ∅ = 0 and λ(K ∪ K
t
) + λ(K ∩ K
t
) = λK + λK
t
for all K, K
t
∈ /. Show that there is
a nonnegative additive functional ν : TX → R extending λ. (Hint: for ﬁnite /, induce on #(/). For the
general case, recall that if γ = sup
K∈K
λK, the additive functionals form a compact subset of [0, γ]
1X
.) (If
P is a lattice, a functional f : P → R such that f(p ∨ q) + f(p ∧ q) = f(p) + f(q) for all p, q ∈ P is called
modular.)
(q) Let X be a set and / a sublattice of TX containing K. Let λ : / → [0, 1] be a functional such that
λK ≤ λK
t
whenever K, K
t
∈ / and K ⊆ K
t
, inf
K∈K
λK = 0,
λ(K ∪ K
t
) +λ(K ∩ K
t
) ≤ λK +λK
t
for all K, K
t
∈ /
(λ is submodular). Show that there is a ﬁnitely additive functional ν : TX → [0, 1] such that
νX = sup
K∈K
λK, νK ≤ λK for every K ∈ /.
413Y Further exercises (a) Give an example of two inner measures φ
1
, φ
2
on a set X such that the
measure deﬁned by φ
1
+φ
2
strictly extends the sum of the measures deﬁned by φ
1
, φ
2
.
413 Notes Inner measure constructions 49
(b) Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be any family of probability spaces, and λ the product measure on X =
i∈I
X
i
.
Show that λ
∗
(
i∈I
A
i
) ≤
i∈I
(µ
i
)
∗
A
i
whenever A
i
⊆ X
i
for every i, with equality if I is countable.
(c) Let (X, Σ, µ) be a totally ﬁnite measure space, and Z the Stone space of the Boolean algebra Σ. For
E ∈ Σ write
´
E for the corresponding openandclosed subset of Z. Show that there is a unique function
f : X → Z such that f
−1
[
´
E] = E for every E ∈ Σ. Show that there is a measure ν on Z, inner regular with
respect to the openandclosed sets, such that f is inversemeasurepreserving with respect to µ and ν, and
that f represents an isomorphism between the measure algebras of µ and ν. Use this construction to prove
(vi)⇒(i) in Theorem 343B without appealing to the Lifting Theorem.
(d) Let X be a set, T a subalgebra of TX, and ν : T → [0, ∞[ a ﬁnitely additive functional. Suppose
that there is a set / ⊆ T, containing ∅, such that (i) µF = sup¦µK : K ∈ /, K ⊆ F¦ for every F ∈ T (ii)
/ is monocompact, that is,
n∈N
K
n
,= ∅ for every nondecreasing sequence in /. Show that ν extends
to a measure on X.
(e) (i) Let X be a topological space. Show that the family of closed countably compact subsets of X is
a countably compact class. (ii) Let X be a Hausdorﬀ space. Show that the family of sequentially compact
subsets of X is a countably compact class.
413 Notes and comments I gave rather few methods of constructing measures in the ﬁrst three volumes
of this treatise; in the present volume I shall have to make up for lost time. In particular I used Cara
th´eodory’s construction for Lebesgue measure (Chapter 11), product measures (Chapter 25) and Hausdorﬀ
measures (Chapter 26). The ﬁrst two, at least, can be tackled in quite diﬀerent ways if we choose. The ﬁrst
alternative approach I oﬀer is the ‘inner measure’ method of 413C. Note the exact deﬁnition in 413A; I do
not think it is an obvious one. In particular, while (α) seems to have something to do with subadditivity,
and (β) is a kind of sequential ordercontinuity, there is no straightforward way in which to associate an
outer measure with an inner measure, unless they both happen to be derived from measures (132B, 413D),
even when they are ﬁnitevalued; and for an inner measure which is allowed to take the value ∞ we have to
add the semiﬁniteness condition (∗) of 413A (see 413Xa).
Once we have got these points right, however, we have a method which rivals Carath´eodory’s in scope,
and in particular is especially well adapted to the construction of inner regular measures. As an almost
trivial example, we have a route to the c.l.d. version of a measure µ (413Xi(i)), which can be derived from the
inner measure µ
∗
deﬁned from µ (413D). Henceforth µ
∗
will be a companion to the familiar outer measure
µ
∗
, and many calculations will be a little easier with both available, as in 413D413F.
The intention behind 413I413J is to ﬁnd a minimal set of properties of a functional φ
0
which will ensure
that it has an extension to a measure. Indeed it is easy to see that, in the context of 413I, given a family
/ with the properties (†) and (‡) there, a functional φ
0
on / can have an extension to an inner regular
measure iﬀ it satisﬁes the conditions (α) and (β), so in this sense 413I is the best possible result. Note that
while Carath´eodory’s construction is liable to produce wildly inﬁnite measures (like Hausdorﬀ measures, or
primitive product measures), the construction here always gives us locally determined measures, provided
only that φ
0
is ﬁnitevalued.
We have to work rather hard for the step from 413I to 413J. Of course 413I is a special case of 413J, and I
could have saved a little space by giving a direct proof of the latter result. But I do not think that this would
have made it easier; 413J really does require an extra step, because somehow we have to extend the functional
φ
0
from / to /
δ
. The method I have chosen uses 413B and 413H to cast as much of the argument as possible
into the context of algebras of sets with additive functionals, where I hope the required manipulations will
seem natural. (But perhaps I should insist that you must not take them too much for granted, as some of the
time we have a ﬁnitely additive functional taking inﬁnite values, and must take care not to subtract illegally,
as well as not to take limits in the wrong places.) Note that the progression φ
0
→ φ
1
→ µ in the proof of
413J involves ﬁrst an approximation from outside (if K ∈ /
δ
, then φ
1
K will be inf¦φ
0
K
t
: K ⊆ K
t
∈ /¦)
and then an approximation from inside (if E ∈ Σ, then µE = sup¦φ
1
K : K ∈ /
δ
, K ⊆ E¦). The essential
diﬃculty in the proof is just that we have to take successive nonexchangeable limits. I have slipped 413K
in as a corollary of 413J; but it can be regarded as one of the fundamental results of measure theory. A
nonnegative ﬁnitely additive functional ν on an algebra Σ of sets can be extended to a countably additive
50 Topologies and measures 413 Notes
measure iﬀ it is ‘relatively countably additive’ in the sense that ν(
n∈N
E
n
) =
∞
n=0
µE
n
whenever ¸E
n
)
n∈N
is a disjoint sequence in Σ such that
n∈N
E
n
∈ Σ (413Xm). Of course the same result can easily be got
from an outer measure construction (413Xl). Note that the outer measure construction also has repeated
limits, albeit simpler ones: in the formula
θA = inf¦
∞
n=0
νE
n
: ¸E
n
)
n∈N
is a sequence in Σ covering A¦
the sum
∞
n=0
νE
n
= sup
n∈N
n
i=0
νE
i
can be regarded as a crude approximation from inside, while the
inﬁmum is an approximation from outside. To get the result as stated in 413K, of course, the outer measure
construction needs a third limiting process, to obtain the c.l.d. version automatically provided by the inner
measure method, and the inner regularity with respect to Σ
δ
, while easily checked, also demands a few
words of argument.
Many applications of the method of 413I413J pass through 413M; if the family / is a countably compact
class then the sequential ordercontinuity hypothesis (β) of 413I or 413J becomes a consequence of the other
hypotheses. The essence of the method is the inner regularity hypothesis (α). I have tried to use the labels
†, ‡, α and β consistently enough to suggest the currents which I think are ﬂowing in this material.
In 413N we strike out in a new direction. The object here is to build an extension which is not going to be
unique, and for which choices will have to be made. As with any such argument, the trick is to specify the
allowable intermediate stages, that is, the partially ordered set P to which we shall apply Zorn’s Lemma. But
here the form of the theorem makes it easy to guess what P should be: it is the set of functionals satisfying
the hypotheses of the theorem which have not wandered outside the boundary set by the conclusion, that
is, which satisfy the condition (∗) of part (a) of the proof of 413N. The ﬁnitistic nature of the hypotheses
makes it easy to check that totally ordered subsets of P have upper bounds (that is to say, if we did this by
transﬁnite induction there would be no problem at limit stages), and all we have to prove is that maximal
elements of P are deﬁned on adequately large domains; which amounts to showing that a member of P not
deﬁned on every element of / has a proper extension, that is, setting up a construction for the step to a
successor ordinal in the parallel transﬁnite induction (part (c) of the proof).
Of course the principal applications of 413N in this book will be in the context of countably additive
functionals, as in 413O.
It is clear that 413N and 413Q overlap to some extent. I include both because they have diﬀerent virtues.
413N can be applied to inﬁnite measures in a way that 413Q, as given, cannot; but its chief advantage, from
the point of view of the work to come, is the approximation of members of T
1
, in measure, by members of
T
0
. This will eventually enable us to retain control of the Maharam types of measures constructed by the
method of 413O. In 413S we have a diﬀerent kind of control; we can specify a lower bound for the measure
of each member of our basic class /, provided only that our speciﬁcations are consistent with some ﬁnitely
additive functional.
414 τadditivity
The second topic I wish to treat is that of ‘τadditivity’. Here I collect results which do not depend
on any strong kind of inner regularity. I begin with what I think of as the most characteristic feature of
τadditivity, its eﬀect on the properties of semicontinuous functions (414A), with a variety of corollaries, up
to the behaviour of subspace measures (414K). A very important property of τadditive topological measures
is that they are often strictly localizable (414J).
The theory of inner regular τadditive measures belongs to the next section, but here I give two intro
ductory results: conditions under which a τadditive measure will be inner regular with respect to closed
sets (414M) and conditions under which a measure which is inner regular with respect to closed sets will be
τadditive (414N). I end the section with notes on ‘density’ and ‘lifting’ topologies (414P414R).
414A Theorem Let (X, T) be a topological space and µ an eﬀectively locally ﬁnite τadditive measure
on X with domain Σ.
(a) Suppose that ( is a nonempty family in Σ ∩ T such that H =
( also belongs to Σ. Then
sup
G∈Ç
G
•
= H
•
in the measure algebra A of µ.
414D τadditivity 51
(b) Write L for the family of Σmeasurable lower semicontinuous functions from X to R. Suppose that
∅ ,= A ⊆ L and set g(x) = sup
f∈A
f(x) for every x ∈ X. If g is Σmeasurable and ﬁnite almost everywhere,
then ˜ g
•
= sup
f∈A
f
•
in L
0
(µ), where ˜ g(x) = g(x) whenever g(x) is ﬁnite.
(c) Suppose that T is a nonempty family of measurable closed sets such that
T ∈ Σ. Then inf
F∈J
F
•
=
(
T)
•
in A.
(d) Write U for the family of Σmeasurable upper semicontinuous functions from X to R. Suppose that
A ⊆ U is nonempty and set g(x) = inf
f∈A
f(x) for every x ∈ X. If g is Σmeasurable and ﬁnite almost
everywhere, then ˜ g
•
= inf
f∈A
f
•
in L
0
(µ), where ˜ g(x) = g(x) whenever g(x) is ﬁnite.
proof (a) ??? If H
•
,= sup
G∈Ç
G
•
, there is a nonzero a ∈ A such that a ⊆ H
•
but a ∩ G
•
= 0 for every G ∈ (.
Express a as E
•
where E ∈ Σ and E ⊆ H. Because µ is eﬀectively locally ﬁnite, there is a measurable
open set H
0
of ﬁnite measure such that µ(H
0
∩ E) > 0. Now ¦H
0
∩ G : G ∈ (¦ is an upwardsdirected
family of measurable open sets with union H
0
∩ H ⊇ H
0
∩ E; as µ is τadditive, there is a G ∈ ( such that
µ(H
0
∩G) > µH
0
−µ(H
0
∩E). But in this case µ(G∩E) > 0, which is impossible, because G
•
∩ E
•
= 0. XXX
(b) For any α ∈ R,
¦x : g(x) > α¦ =
f∈A
¦x : f(x) > α¦,
and these are all measurable open sets. Identifying ¦x : g(x) > α¦
•
∈ A with [[˜ g
•
> α]] (364Jb), we see from
(a) that [[˜ g
•
> α]] = sup
f∈A
[[f
•
> α]] for every α. But this means that ˜ g
•
= sup
f∈A
f
•
, by 364Mb.
(c) Apply (a) to ( = ¦X ¸ F : F ∈ T¦.
(d) Apply (b) to ¦−f : f ∈ A¦.
414B Corollary Let X be a topological space and µ an eﬀectively locally ﬁnite τadditive topological
measure on X.
(a) Suppose that A is a nonempty upwardsdirected family of lower semicontinuous functions from X
to [0, ∞]. Set g(x) = sup
f∈A
f(x) in [0, ∞] for every x ∈ X. Then
_
g = sup
f∈A
_
f in [0, ∞].
(b) Suppose that A is a nonempty downwardsdirected family of nonnegative continuous realvalued
functions on X, and that g(x) = inf
x∈A
f(x) for every x ∈ X. If any member of A is integrable, then
_
g = inf
f∈A
_
f.
proof (a) Of course all the f ∈ A, and also g, are measurable functions. Set g
n
= g ∧nχX for every n ∈ N.
Then
g
n
(x) = sup
f∈A
(f ∧ nχX)(x)
for every x ∈ X, so g
•
n
= sup
f∈A
(f ∧ nχX)
•
, by 414Ab, and
_
g
n
=
_
g
•
n
= sup
f∈A
_
(f ∧ nχX)
•
= sup
f∈A
_
f ∧ nχX
by 365Dh. But now, of course,
_
g = sup
n∈N
_
g
n
= sup
n∈N,f∈A
_
f ∧ nχX = sup
f∈A
_
f,
as claimed.
(b) Take an integrable f
0
∈ A, and apply (a) to ¦(f
0
−f)
+
: f ∈ A¦.
414C Corollary Let (X, T, Σ, µ) be an eﬀectively locally ﬁnite τadditive topological measure space and
T a nonempty downwardsdirected family of closed sets. If inf
F∈J
µF is ﬁnite, this is the measure of
T.
proof Setting F
0
=
T, then F
•
0
= inf
F∈J
F
•
, by 414Ac; now
µF
0
= ¯ µF
•
0
= inf
F∈J
¯ µF
•
= inf
F∈J
µF
by 321F.
414D Corollary Let µ be an eﬀectively locally ﬁnite τadditive measure on a topological space X. If ν is
a totally ﬁnite measure with the same domain as µ, truly continuous with respect to µ, then ν is τadditive.
In particular, if µ is σﬁnite and ν is absolutely continuous with respect to µ, then ν is τadditive.
52 Topologies and measures 414D
proof We have a functional ¯ ν : A → [0, ∞[, where A is the measure algebra of µ, such that ¯ νE
•
= νE
for every E in the common domain Σ of µ and ν. Now ¯ ν is continuous for the measurealgebra topology
of A (327Cd), therefore completely additive (327Ba), therefore ordercontinuous (326Kc). So if ( is an
upwardsdirected family of open sets belonging to Σ with union G
0
∈ Σ,
sup
G∈Ç
νG = sup
G∈Ç
¯ νG
•
= ¯ νG
•
0
= νG
0
because G
•
0
= sup
G∈Ç
G
•
.
The last sentence follows at once, because on a σﬁnite space an absolutely continuous countably additive
functional is truly continuous (232Bc).
414E Corollary Let (X, T, Σ, µ) be an eﬀectively locally ﬁnite τadditive topological measure space.
Suppose that ( ⊆ T is nonempty and upwardsdirected, and H =
(. Then
(a) µ(E ∩ H) = sup
G∈Ç
µ(E ∩ G) for every E ∈ Σ;
(b) if f is a nonnegative virtually measurable realvalued function deﬁned almost everywhere on X, then
_
H
f = sup
G∈Ç
_
f in [0, ∞].
proof (a) In the measure algebra (A, ¯ µ) of µ,
(E ∩ H)
•
= E
•
∩ H
•
= E
•
∩ sup
G∈Ç
G
•
= sup
G∈Ç
E
•
∩ G
•
= sup
G∈Ç
(E ∩ G)
•
,
using 414Aa and the distributive law 313Ba. So
µ(E ∩ H) = ¯ µ(E ∩ H)
•
= sup
G∈Ç
¯ µ(E ∩ G)
•
= sup
G∈Ç
µ(E ∩ G)
by 321D, because ( and ¦(E ∩ G)
•
: G ∈ (¦ are upwardsdirected.
(b) For each G ∈ (,
_
G
f =
_
f χG =
_
(f χG)
•
=
_
f
•
χG
•
,
where χG
•
can be interpreted either as (χG)
•
(in L
0
(µ)) or as χ(G
•
) (in L
0
(A), where A is the measure
algebra of µ); see 364K. Now H
•
= sup
G∈Ç
G
•
(414Aa); since χ and are ordercontinuous (364Kc, 364P),
f
•
χH
•
= sup
G∈Ç
f
•
χG
•
; so
_
H
f =
_
f
•
χH
•
= sup
G∈Ç
_
f
•
χG
•
= sup
G∈Ç
_
G
f
by 365Dh.
414F Corollary Let (X, T, Σ, µ) be an eﬀectively locally ﬁnite τadditive topological measure space.
Then for every E ∈ Σ there is a unique relatively closed selfsupporting set F ⊆ E such that µ(E ¸ F) = 0.
proof Let ( be the set ¦G : G ∈ T, µ(G ∩ E) = 0¦. Then ( is upwardsdirected, so µ(E ∩ G
∗
) =
sup
G∈Ç
µ(E∩G) = 0, where G
∗
=
(. Set F = E¸G
∗
. Then F ⊆ E is relatively closed, and µ(E¸F) = 0.
If H ∈ T and H ∩ F ,= ∅, then H / ∈ ( so µ(F ∩ H) = µ(E ∩ H) > 0; thus F is selfsupporting. If F
t
⊆ E is
another selfsupporting relatively closed set such that µ(E¸ F
t
) = 0, then µ(F ¸ F
t
) = µ(F
t
¸ F) = 0; but as
F ¸ F
t
is relatively open in F, and F
t
¸ F is relatively open in F
t
, these must both be empty, and F = F
t
.
414G Corollary If (X, T, Σ, µ) is a Hausdorﬀ eﬀectively locally ﬁnite τadditive topological measure
space and E ∈ Σ is an atom for µ (deﬁnition: 211I), then there is an x ∈ E such that E ¸ ¦x¦ is negligible.
proof Let F ⊆ E be a selfsupporting set such that µ(E ¸ F) = 0. Since µF = µE > 0, F is not empty;
take x ∈ F. ??? If F ,= ¦x¦, let y ∈ F ¸ ¦x¦. Because T is Hausdorﬀ, there are disjoint open sets G, H
containing x, y respectively; and in this case µ(E ∩ G) = µ(F ∩ G) and µ(E ∩ H) = µ(F ∩ H) are both
nonzero, which is impossible, since E is an atom. XXX
So F = ¦x¦ and E ¸ ¦x¦ is negligible.
414H Corollary If (X, T, Σ, µ) is an eﬀectively locally ﬁnite τadditive topological measure space and
ν is an indeﬁniteintegral measure over µ (deﬁnition: 234B), then ν is a τadditive topological measure.
414K τadditivity 53
proof Because ν measures every set in Σ (234Da), it is a topological measure. To see that it is τadditive,
apply 414Eb to a RadonNikod´ ym derivative of ν.
414I Proposition Let (X, T, Σ, µ) be a complete locally determined eﬀectively locally ﬁnite τadditive
topological measure space. If E ⊆ X, ( ⊆ T are such that E ⊆
( and E ∩ G ∈ Σ for every G ∈ (, then
E ∈ Σ.
proof Set / = ¦K : K ∈ Σ, E ∩ K ∈ Σ¦. Then whenever F ∈ Σ and µF > 0 there is a K ∈ / included in
F with µK > 0. PPP Set K
1
= F ¸
(. Then K
1
is a member of / included in F. If µK
1
> 0 then we can
stop. Otherwise, (
∗
= ¦G
0
∪ . . . ∪ G
n
: G
0
, . . . , G
n
∈ (¦ is an upwardsdirected family of open sets, and
sup
G∈Ç
∗ µ(F ∩ G) = µ(F ∩
(
∗
) = µF > 0,
by 414E. So there is a G ∈ (
∗
such that µ(F ∩ G) > 0; but now E ∩ G ∈ Σ so F ∩ G ∈ /. QQQ
By 412Aa, µ is inner regular with respect to /; by 412Ja, E ∈ Σ.
414J Theorem Let (X, T, Σ, µ) be a complete locally determined eﬀectively locally ﬁnite τadditive
topological measure space. Then µ is strictly localizable.
proof Let T be a maximal disjoint family of selfsupporting measurable sets of ﬁnite measure. Then
whenever E ∈ Σ and µE > 0, there is an F ∈ T such that µ(E ∩ F) > 0. PPP??? Otherwise, let G be an open
set of ﬁnite measure such that µ(G ∩ E) > 0, and set T
0
= ¦F : F ∈ T, F ∩ G ,= ∅¦. Then µ(F ∩ G) > 0
for every F ∈ T
0
, while µG < ∞ and T
0
is disjoint, so T
0
is countable and
T
0
∈ Σ. Set E
t
= E ¸
T
0
;
then E ¸ E
t
= E ∩
T
0
is negligible, so µ(G∩E
t
) > 0. By 414F, there is a selfsupporting set F
t
⊆ G∩E
t
such that µF
t
> 0. But in this case F
t
∩ F = ∅ for every F ∈ T, so we ought to have added F
t
to T. XXXQQQ
This means that T satisﬁes the criterion of 213O. Because (X, Σ, µ) is complete and locally determined,
it is strictly localizable.
414K Proposition Let (X, Σ, µ) be a measure space and T a topology on X, and Y ⊆ X a subset such
that the subspace measure µ
Y
is semiﬁnite (see the remark following 412O). If µ is an eﬀectively locally
ﬁnite τadditive topological measure, so is the subspace measure µ
Y
.
proof By 412P, µ
Y
is an eﬀectively locally ﬁnite topological measure. Now suppose that H is a nonempty
upwardsdirected family in T
Y
with union H
∗
. Set
( = ¦G : G ∈ T, G∩ Y ∈ H¦, G
∗
=
(,
so that ( is upwardsdirected and H
∗
= Y ∩G
∗
. Let / be the family of sets K ⊆ X such that K∩G
∗
¸G = ∅
for some G ∈ (. If E ∈ Σ,
µE = µ(E ¸ G
∗
) +µ(E ∩ G
∗
) = µ(E ¸ G
∗
) + sup
G∈Ç
µ(E ∩ G)
(414Ea)
= sup
G∈Ç
µ(E ¸ (G
∗
¸ G)),
so µ is inner regular with respect to /. By 412Ob, µ
Y
is inner regular with respect to ¦K ∩ Y : K ∈ /¦.
So if γ < µ
Y
H
∗
, there is a K ∈ / such that K ∩ Y ⊆ H
∗
and µ
Y
(K ∩ Y ) ≥ γ. But now there is a G ∈ (
such that K ∩ G
∗
¸ G = ∅, so that K ∩ Y ⊆ G∩ Y ∈ H and sup
H∈1
µH ≥ γ. As H and γ are arbitrary, µ
is τadditive.
Remarks Recall from 214I that if (X, Σ, µ) has locally determined negligible sets (in particular, is either
strictly localizable or complete and locally determined), then all its subspaces are semiﬁnite. In 419C below
I describe a tight locally ﬁnite Borel measure with a subset on which the subspace measure is not semiﬁnite,
therefore not eﬀectively locally ﬁnite or τadditive. In 419A I describe a σﬁnite locally ﬁnite τadditive
topological measure, inner regular with respect to the closed sets, with a closed subset on which the subspace
measure is totally ﬁnite but not τadditive.
54 Topologies and measures 414L
414L Lemma Let (X, T) be a topological space, and µ, ν two eﬀectively locally ﬁnite Borel measures
on X which agree on the open sets. Then they are equal.
proof Write T
f
for the family of open sets of ﬁnite measure. (I do not need to specify which measure I am
using here.) For G ∈ T
f
, set µ
G
E = µ(G∩E), ν
G
E = ν(G∩E) for every Borel set E. Then µ
G
and ν
G
are
totally ﬁnite Borel measures which agree on T. By the Monotone Class Theorem (136C), µ
G
and ν
G
agree
on the σalgebra generated by T, that is, the Borel σalgebra B. Now, for any E ∈ B,
µE = sup
G∈T
f µ
G
E = sup
G∈T
f ν
G
E = νE,
by 412F. So µ = ν.
414M Proposition Let (X, Σ, µ) be a measure space with a regular topology T such that µ is eﬀectively
locally ﬁnite and τadditive and Σ includes a base for T.
(a) µG = sup¦µF : F ∈ Σ is closed, F ⊆ G¦ for every open set G ∈ Σ.
(b) If µ is inner regular with respect to the σalgebra generated by T∩Σ, it is inner regular with respect
to the closed sets.
proof (a) For U ∈ Σ ∩ T, the set
H
U
= ¦H : H ∈ Σ ∩ T, H ⊆ U¦
is an upwardsdirected family of open sets, and
H
U
= U because T is regular and Σ includes a base for
T. Because µ is τadditive, µU = sup¦µH : H ∈ H
U
¦. Now, given γ < µG, we can choose ¸U
n
)
n∈N
in Σ∩T
inductively, as follows. Start by taking U
0
⊆ G such that γ < µU
0
< ∞ (using the hypothesis that µ is
eﬀectively locally ﬁnite). Given U
n
∈ Σ ∩ T and µU
n
> γ, take U
n+1
∈ Σ ∩ T such that U
n+1
⊆ U
n
and
µU
n+1
> γ. On completing the induction, set
F =
n∈N
U
n
=
n∈N
U
n
;
then F is a closed set belonging to Σ, F ⊆ G and µF ≥ γ. As γ is arbitrary, we have the result.
(b) Let Σ
0
be the σalgebra generated by Σ ∩ T and set µ
0
= µΣ
0
. Then Σ
0
∩ T = Σ ∩ T is still a
base for T and µ
0
is still τadditive and eﬀectively locally ﬁnite, so by (a) and 412G it is inner regular with
respect to the closed sets. Now we are supposing that µ is inner regular with respect to Σ
0
, so µ is inner
regular with respect to the closed sets, by 412Ab.
414N Proposition Let (X, Σ, µ) be a measure space and T a topology on X. Suppose that (i) µ is semi
ﬁnite and inner regular with respect to the closed sets (ii) whenever T is a nonempty downwardsdirected
family of measurable closed sets with empty intersection and inf
F∈J
µF < ∞, then inf
F∈J
µF = 0. Then
µ is τadditive.
proof Let ( be a nonempty upwardsdirected family of measurable open sets with measurable union H.
Take any γ < µH. Because µ is semiﬁnite, there is a measurable set E ⊆ H such that γ < µE < ∞.
Now there is a measurable closed set F ⊆ E such that µF ≥ γ. Consider T = ¦F ¸ G : G ∈ (¦. This is a
downwardsdirected family of closed sets of ﬁnite measure with empty intersection. So inf
G∈Ç
µ(F ¸ G) = 0,
that is,
γ ≤ µF = sup
G∈Ç
µ(F ∩ G) ≤ sup
G∈Ç
µG.
As γ is arbitrary, µH = sup
G∈Ç
µG; as ( is arbitrary, µ is τadditive.
414O The following elementary result is worth noting.
Proposition If X is a hereditarily Lindel¨of space (e.g., if it is separable and metrizable) then every measure
on X is τadditive.
proof If µ is a measure on X, with domain Σ, and ( ⊆ Σ is a nonempty upwardsdirected family of
measurable open sets, then there is a sequence ¸G
n
)
n∈N
in ( such that
( =
n∈N
G
n
. Now
µ(
() = lim
n→∞
µ(
i≤n
G
i
) ≤ sup
G∈Ç
µG.
As ( is arbitrary, µ is τadditive.
414Xb τadditivity 55
414P Density topologies Recall that a lower density of a measure space (X, Σ, µ) is a function φ :
Σ → Σ such that φE = φF whenever E, F ∈ Σ and µ(E´F) = 0, µ(E´φE) = 0 for every E ∈ Σ, φ∅ = ∅
and φ(E ∩ F) = φE ∩ φF for all E, F ∈ Σ (341C).
Proposition Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lower density
such that φX = X. Set
T = ¦E : E ∈ Σ, E ⊆ φE¦.
Then T is a topology on X, the density topology associated with φ, and (X, T, Σ, µ) is an eﬀectively
locally ﬁnite τadditive topological measure space; µ is strictly positive and inner regular with respect to
the open sets.
proof (a)(i) For any E ∈ Σ, φ(E ∩ φE) = φE because E ¸ φE is negligible; consequently E ∩ φE ∈ T. In
particular, ∅ = ∅ ∩ φ∅ and X = X ∩ φX belong to T. If E, F ∈ T then
φ(E ∩ F) = φE ∩ φF ⊇ E ∩ F,
so E ∩ F ∈ T.
(ii) Suppose that ( ⊆ T and H =
(. By 341M, µ is (strictly) localizable, so ( has an essential
supremum F ∈ Σ such that F
•
= sup
G∈Ç
G
•
in the measure algebra A of µ; that is, for E ∈ Σ, µ(G¸E) = 0
for every G ∈ ( iﬀ µ(F ¸ E) = 0. Now F ¸ H is negligible, by 213K. On the other hand,
G ⊆ φG = φ(G∩ F) ⊆ φF
for every G ∈ (, so H ⊆ φF, and H ¸ F ⊆ φF ¸ F is negligible. But as µ is complete, this means that
H ∈ Σ. Also φH = φF ⊇ H, so H ∈ T. Thus T is closed under arbitrary unions and is a topology.
(b) By its deﬁnition, T is included in Σ, so µ is a topological measure. If E ∈ Σ then E ∩ φE belongs
to T, is included in E and has the same measure as E; so µ is inner regular with respect to the open sets.
If E ∈ T is nonempty, then φE ⊇ E is nonempty, so µE > 0; thus µ is strictly positive. Finally, if ( is a
nonempty upwardsdirected family in T, then the argument of (aii) shows that (
()
•
= sup
G∈Ç
G
•
in A,
so that µ(
() = sup
G∈Ç
µG. Thus µ is τadditive. If E ∈ Σ and µE > 0 then there is an F ⊆ E such that
0 < µF < ∞, and now E ∩ φF is an open set of nonzero ﬁnite measure included in E; so µ is eﬀectively
locally ﬁnite.
414Q Lifting topologies Let (X, Σ, µ) be a measure space and φ : Σ → Σ a lifting, that is, a Boolean
homomorphism such that φE = ∅ whenever µE = 0 and µ(E´φE) = 0 for every E ∈ Σ (341A). The lifting
topology associated with φ is the topology generated by ¦φE : E ∈ Σ¦. Note that ¦φE : E ∈ Σ¦ is a
topology base, so is a base for the lifting topology.
414R Proposition Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a
lifting with lifting topology S and density topology T. Then S ⊆ T ⊆ Σ, and µ is τadditive, eﬀectively
locally ﬁnite and strictly positive with respect to S. Moreover, S is zerodimensional.
proof Of course φ is a lower density, so we can talk of its density topology, and since φ
2
E = φE, φE ∈ T
for every E ∈ Σ, so S ⊆ T. Because µ is τadditive and strictly positive with respect to T, it must also
be τadditive and strictly positive with respect to S. If E ∈ Σ and µE > 0 there is an F ⊆ E such that
0 < µF < ∞, and now φF is an Sopen set of ﬁnite measure meeting E in a nonnegligible set; so µ is
eﬀectively locally ﬁnite with respect to S. Of course S is zerodimensional because φ[Σ] is a base for S
consisting of openandclosed sets.
414X Basic exercises (a) Let (X, Σ, µ) and (Y, T, ν) be measure spaces with topologies T and S, and
f : X → Y a continuous inversemeasurepreserving function. Show that if µ is τadditive with respect to
T then ν is τadditive with respect to S. Show that if ν is locally ﬁnite, so is µ.
(b) Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of measure spaces, with direct sum (X, Σ, µ); suppose that we are
given a topology T
i
on X
i
for each i, and let T be the disjoint union topology on X. Show that µ is τadditive
iﬀ every µ
i
is.
56 Topologies and measures 414Xc
>>>(c) Let (X, T) be a topological space and µ a totally ﬁnite measure on X which is inner regular with
respect to the closed sets. Suppose that µX = sup
G∈Ç
µG whenever ( is an upwardsdirected family of
measurable open sets covering X. Show that µ is τadditive.
(d) Let µ be an eﬀectively locally ﬁnite τadditive σﬁnite measure on a topological space X, and
ν : Σ → [0, ∞[ a countably additive functional which is absolutely continuous with respect to µ. Show from
ﬁrst principles that ν is τadditive.
(e) Give an example of an indeﬁniteintegral measure over Lebesgue measure on R which is not eﬀectively
locally ﬁnite. (Hint: arrange for every nontrivial interval to have inﬁnite measure.)
(f ) Let (X, T) be a topological space and µ a complete locally determined eﬀectively locally ﬁnite τ
additive topological measure on X. Show that if f is a realvalued function, deﬁned on a subset of X, which
is locally integrable in the sense of 411Fc, then f is measurable.
(g) Let (X, T) be a topological space and µ an eﬀectively locally ﬁnite τadditive measure on X. Let (
be a cover of X consisting of measurable open sets, and / the ideal of subsets of X generated by (. Show
that µ is inner regular with respect to /.
(h) Let (X, T, Σ, µ) be a complete locally determined eﬀectively locally ﬁnite τadditive topological
measure space, and A a subset of X. Suppose that for every x ∈ X there is an open set G containing x such
that A∩ G is negligible. Show that A is negligible.
(i) Give an alternative proof of 414K based on the fact that the canonical map from the measure algebra
of µ to the measure algebra of µ
Y
is ordercontinuous (322Yd).
>>>(j) (i) If µ is an eﬀectively locally ﬁnite τadditive Borel measure on a regular topological space, show
that the c.l.d. version of µ is a quasiRadon measure. (ii) If µ is a locally ﬁnite, eﬀectively locally ﬁnite
τadditive Borel measure on a locally compact Hausdorﬀ space, show that µ is tight, so that the c.l.d. version
of µ is a Radon measure.
>>>(k) Let (X, Σ, µ) be a complete locally determined measure space and φ a lower density on X such that
φX = X; let T be the corresponding density topology. (i) Show that a dense open subset of X must be
conegligible. (ii) Show that a subset of X is nowhere dense for T iﬀ it is negligible iﬀ it is meager for T. (iii)
Show that a function f : X → R is Σmeasurable iﬀ it is Tcontinuous at almost every point of X. (Hint:
if f is measurable, set E
q
= ¦x : f(x) > q¦, F
q
= ¦x : f(x) < q¦; show that f is continuous at every point
of X ¸
q∈Q
((E
q
¸ φE
q
) ∪ (F
q
¸ φF
q
)).)
(l) Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lower density such
that φX = X, with density topology T. Show that if A ⊆ X and E is a measurable envelope of A then the
Tclosure of A is just A∪ (X ¸ φ(X ¸ E)).
(m) Let µ be Lebesgue measure on R
r
, Σ its domain, φ : Σ → Σ lower Lebesgue density (341E)
and T the corresponding density topology. Show that for any A ⊆ R, the closure of A for T is just
A∪ ¦x : limsup
δ↓0
µ
∗
(A∩B(x,δ))
µB(x,δ)
> 0¦, and the interior is A∩ ¦x : lim
δ↓0
µ
∗
(A∩B(x,δ))
µB(x,δ)
= 1¦.
(n) Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lower density such
that φX = X; let T be the associated density topology. Let A be a subset of X and E a measurable envelope
of A; let Σ
A
be the subspace σalgebra and µ
A
the subspace measure on A. (i) Show that we have a lower
density φ
A
: Σ
A
→ Σ
A
deﬁned by setting φ
A
(F ∩ A) = A ∩ φ(E ∩ F) for every F ∈ Σ. (ii) Show that
φ
A
A = A iﬀ A ⊆ φE, and that in this case the density topology on A derived from φ
A
is just the subspace
topology.
(o) Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lifting, with density
topology T and lifting topology S. (i) Show that
T = ¦H ∩ G : G ∈ S, H is conegligible¦ = ¦H ∩ φE : E ∈ Σ, H is conegligible¦.
(ii) Show that if A ⊆ X and E is a measurable envelope of A then the Tclosure of A is A∪ φE.
414Yf τadditivity 57
(p) Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lifting; let S be its
lifting topology. Let A be a subset of X such that A ⊆ φE for some (therefore any) measurable envelope E
of A. Let Σ
A
be the subspace σalgebra and µ
A
the subspace measure on A. (i) Show that we have a lifting
φ
A
: Σ
A
→ Σ
A
deﬁned by setting φ
A
(F ∩ A) = A∩ φF for every F ∈ Σ. (ii) Show that the lifting topology
on A derived from φ
A
is just the subspace topology.
(q) Let (X, Σ, µ) and (Y, T, ν) be complete locally determined measure spaces and f : X → Y an inverse
measurepreserving function. (i) Suppose that we have lower densities φ : Σ → Σ and ψ : T → T such that
φX = X, ψY = Y and φf
−1
[F] = f
−1
[ψF] for every F ∈ T. Show that f is continuous for the density
topologies of φ and ψ. (ii) Show that if φ and ψ are liftings then f is continuous for the lifting topologies.
(r) Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lifting, with associated
lifting topology S. Show that a function f : X → R is Σmeasurable iﬀ there is a conegligible set H such
that fH is Scontinuous. (Compare 414Xk.)
(s) Let (X, Σ, µ) be a complete locally determined measure space and φ : Σ → Σ a lifting. Let (Z, T, ν)
be the Stone space of the measure algebra of µ, and f : X → Z the inversemeasurepreserving function
associated with φ (341P). Show that the lifting topology on X is just ¦f
−1
[G] : G ⊆ Z is open¦.
(t) Let (X, Σ, µ) be a strictly localizable measure space and φ : Σ → Σ a lifting. Write L
∞
for the
Banach lattice of bounded Σmeasurable realvalued functions on X, identiﬁed with L
∞
(Σ) (363H); let
T : L
∞
→L
∞
be the Riesz homomorphism associated with φ (363F). (i) Show that T
2
= T. (ii) Show that
if X is given the lifting topology S deﬁned by φ, then T[L
∞
] is precisely the space of bounded continuous
realvalued functions on X. (iii) Show that if f ∈ L
∞
, x ∈ X and > 0 there is an Sopen set U containing
x such that [(Tf)(x) −
1
µV
_
V
fdµ[ ≤ for every nonnegligible measurable set V included in U.
414Y Further exercises (a) Let (X, T, Σ, µ) be a totally ﬁnite topological measure space. For E ∈ Σ
set
µ
τ
E = inf¦µ(E ¸
() + sup
G∈Ç
µ(E ∩ G) : ( ⊆ T is an upwardsdirected set¦.
Suppose either that µ is inner regular with respect to the closed sets or that T is regular. Show that µ
τ
is
a τadditive measure, the largest τadditive measure with domain Σ which is dominated by µ.
(b) Let X be a set, Σ an algebra of subsets of X, and T a topology on X. Let M be the Lspace of
bounded ﬁnitely additive realvalued functionals on Σ (362B). Let N ⊆ M be the set of those functionals
ν such that inf
G∈Ç
[ν[(H ¸ G) = 0 whenever ( ⊆ T ∩ Σ is a nonempty upwardsdirected family with union
H ∈ Σ. Show that N is a band in M. (Cf. 362Xi.)
(c) Find a probability space (X, Σ, µ) and a topology T on X such that Σ includes a base for T and µ is
τadditive, but there is a set E ∈ Σ such that the subspace measure µ
E
is not τadditive.
(d) Let (X, Σ, µ) be a complete locally determined measure space and φ a lower density on X such that
φX = X; let T be the density topology. Show that Σ is precisely the Baire property algebra for T, so that
(X, T) is a Baire space.
(e) Let φ be lower Lebesgue density on R
r
, and T the associated density topology. Show that every
TBorel set is an F
σ
set for T.
(f ) Let (X, ρ) be a metric space and µ a strictly positive locally ﬁnite quasiRadon measure on X; write T
for the topology of X and Σ for the domain of µ. For E ∈ Σ set φ(E) = ¦x : x ∈ X, lim
δ↓0
µ(E∩B(x,δ))
µB(x,δ)
= 1¦.
Suppose that E ¸ φ(E) is negligible for every E ∈ Σ (cf. 261D, 472D). (i) Show that φ is a lower density for
µ, with φ(X) = X. Let T
d
be the associated density topology. (ii) Suppose that H ∈ T
d
and that K ⊆ H
is Tclosed and ρtotally bounded. Show that there is a Tclosed, ρtotally bounded K
t
⊆ H such that K is
included in the T
d
interior of K
t
. (iii) Show that T
d
is completely regular. (Hint: Lukeˇs Mal´ y & Zaj´ıˇ cek
86.)
58 Topologies and measures 414Yg
(g) Show that the density topology on R associated with lower Lebesgue density is not normal.
(h) Let µ be Lebesgue measure on R
r
, Σ its domain, φ : Σ → Σ lower Lebesgue density and T the
corresponding density topology. (i) Show that if f : R
r
→ R
r
is a bijection such that f and f
−1
are both
diﬀerentiable everywhere, with continuous derivatives, then f is a homeomorphism for T. (Hint: 263D.)
(ii) Show that if φ : Σ → Σ is a lifting and S the corresponding lifting topology, then x → −x is not a
homeomorphism for S. (Hint: 345Xc.)
414 Notes and comments I have remarked before that it is one of the abiding frustrations of measure
theory, at least for anyone ambitious to apply the power of modern general topology to measuretheoretic
problems, that the basic convergence theorems are irredeemably conﬁned to sequences. In Volume 3 I
showed that if we move to measure algebras and function spaces, we can hope that the countable chain
condition or the countable sup property will enable us to replace arbitrary directed sets with monotonic
sequences, thereby giving theorems which apply to apparently more general types of convergence. In 414A
and its corollaries we come to a quite diﬀerent context in which a measure, or integral, behaves like an
ordercontinuous functional. Of course the theorems here depend directly on the hypothesis of τadditivity,
which rather begs the question; but we shall see in the rest of the chapter that this property does indeed
often appear. For the moment, I remark only that as Lebesgue measure is τadditive we certainly have a
nontrivial example to work with.
The hypotheses of the results above move a touch awkwardly between those with the magic phrase
‘topological measure’ and those without. The point is that (as in 412G, for instance) it is sometimes useful
to be able to apply these ideas to Baire measures on completely regular spaces, which are deﬁned on a base
for the topology but may not be deﬁned on every open set.
I hope that no confusion will arise between the two topologies associated with a lifting on a complete
locally determined space. I have called them the ‘density topology’ and the ‘lifting topology’ because the
former can be deﬁned directly from a lower density; but it would be equally reasonable to call them the
‘ﬁne’ and ‘coarse’ lifting topologies. The density topology has the apparent advantage of giving us a measure
which is inner regular with respect to the Borel sets, but at the cost of being wildly irregular regarded as a
topological space. It has the important advantage that there are densities (like the Lebesgue lower density)
which have some claim to be called canonical, and others with useful special properties, as in ¸346, while
liftings are always arbitrary and invariance properties for them sometimes unachievable. So, for instance,
the Lebesgue density topology on R
r
is invariant under diﬀeomorphisms, which no lifting topology can be
(414Yh). The lifting topology is wellbehaved as a topology, but only in special circumstances (as in 453Xd)
is the measure inner regular with respect to its Borel sets, and even the closure of a set can be diﬃcult to
determine.
As with inner regularity, τadditivity can be associated with the band structure of the space of bounded
additive functionals on an algebra (414Yb); there will therefore be corresponding decompositions of measures
into τadditive and ‘purely nonτadditive’ parts (cf. 414Ya).
415 QuasiRadon measure spaces
We are now I think ready to draw together the properties of inner regularity and τadditivity. Indeed
this section will unite several of the themes which have been running through the treatise so far: (strict)
localizability, subspaces and products as well as the new concepts of this chapter. In these terms, the principal
results are that a quasiRadon space is strictly localizable (415A), any subspace of a quasiRadon space is
quasiRadon (415B), and the product of a family of strictly positive quasiRadon probability measures on
separable metrizable spaces is quasiRadon (415E). I describe a basic method of constructing quasiRadon
measures (415K), with details of one of the standard ways of applying it (415L, 415N) and some notes on
how to specify a quasiRadon measure uniquely (415H415I). I spell out useful results on indeﬁniteintegral
measures (415O) and L
p
spaces (415P), and end the section with a discussion of the Stone space Z of a
localizable measure algebra A and an important relation in Z X when A is the measure algebra of a
quasiRadon measure space X (415Q415R).
It would be fair to say that the study of quasiRadon spaces for their own sake is a minority interest.
If you are not already well acquainted with Radon measure spaces, it would make good sense to read this
415D QuasiRadon measure spaces 59
section in parallel with the next. In particular, the constructions of 415K and 415L derive much of their
importance from the corresponding constructions in ¸416.
415A Theorem A quasiRadon measure space is strictly localizable.
proof This is a special case of 414J.
415B Theorem Any subspace of a quasiRadon measure space is quasiRadon.
proof Let (X, T, Σ, µ) be a quasiRadon measure space and (Y, T
Y
, Σ
Y
, µ
Y
) a subspace with the induced
topology and measure. Because µ is complete, locally determined and localizable (by 415A), so is µ
Y
(214Id). Because µ
Y
is semiﬁnite and µ is an eﬀectively locally ﬁnite τadditive topological measure, so
is µ
Y
(414K). Because µ is inner regular with respect to the closed sets and µ
Y
is semiﬁnite, µ
Y
is inner
regular with respect to the relatively closed subsets of Y (412Pc). So µ
Y
is a quasiRadon measure.
415C In regular topological spaces, the condition ‘inner regular with respect to the closed sets’ in the
deﬁnition of ‘quasiRadon measure’ can be weakened or omitted.
Proposition Let (X, T) be a regular topological space.
(a) If µ is a complete locally determined eﬀectively locally ﬁnite τadditive topological measure on X,
inner regular with respect to the Borel sets, then it is a quasiRadon measure.
(b) If µ is an eﬀectively locally ﬁnite τadditive Borel measure on X, its c.l.d. version is a quasiRadon
measure.
proof (a) By 414Mb, µ is inner regular with respect to the closed sets, which is the only feature missing
from the given hypotheses.
(b) The c.l.d. version of µ satisﬁes the hypotheses of (a).
415D In separable metric spaces, among others, we can even omit τadditivity.
Proposition Let (X, T) be a hereditarily Lindel¨of topological space; e.g., a separable metrizable space
(4A2P(aiii)).
(i) If µ is a complete eﬀectively locally ﬁnite measure on X, inner regular with respect to the Borel sets,
and its domain includes a base for T, then it is a quasiRadon measure.
(ii) If µ is an eﬀectively locally ﬁnite Borel measure on X, then its completion is a quasiRadon measure.
(iii) Any quasiRadon measure on X is σﬁnite.
(iv) If X is regular, any quasiRadon measure on X is completion regular.
proof (a) The basic fact we need is that if ( is any family of open sets in X, then there is a countable
(
0
⊆ ( such that
(
0
=
( (4A2H(ci)). Consequently any eﬀectively locally ﬁnite measure µ on X is
σﬁnite. PPP Let ( be the family of measurable open sets of ﬁnite measure. Let (
0
⊆ ( be a countable set
with the same union as (. Then E = X ¸
(
0
is measurable, and E ∩ G = ∅ for every G ∈ (, so µE = 0;
accordingly (
0
∪ ¦E¦ is a countable cover of X by sets of ﬁnite measure, and µ is σﬁnite. QQQ
Moreover, any measure on X is τadditive. PPP If ( is a nonempty upwardsdirected family of open
measurable sets, there is a sequence ¸G
n
)
n∈N
in ( with union
(. If n ∈ N there is a G ∈ ( such that
i≤n
G
i
⊆ G, so
µ(
() = µ(
n∈N
G
n
) = sup
n∈N
µ(
i≤n
G
i
) ≤ sup
G∈Ç
µG.
As ( is arbitrary, µ is τadditive. QQQ
(b)(i) Now let µ be a complete eﬀectively locally ﬁnite measure on X, inner regular with respect to the
Borel sets, and with domain Σ including a base for the topology of X. If H ∈ T, then ( = ¦G : G ∈
Σ ∩ T, G ⊆ H¦ has union H, because Σ ∩ T is a base for T; but in this case there is a countable (
0
⊆ (
such that H =
(
0
, so that H ∈ Σ. Thus µ is a topological measure. We know also from (a) that it is
τadditive and σﬁnite, therefore locally determined. By 415Ca, it is a quasiRadon measure.
(ii) If µ is an eﬀectively locally ﬁnite Borel measure on X, then its completion ˆ µ satisﬁes the conditions
of (i), so is a quasiRadon measure.
60 Topologies and measures 415D
(iii) If µ is a quasiRadon measure on X, it is surely eﬀectively locally ﬁnite, therefore σﬁnite.
(iv) If X is regular, then every closed set is a zero set (4A2H(ciii)), so any measure which is inner
regular with respect to the closed sets is completion regular.
415E I am delaying most of the theory of products of (quasi)Radon measures to ¸417. However, there
is one result which is so important that I should like to present it here, even though some of the ideas will
have to be repeated later.
Theorem Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of separable metrizable quasiRadon probability spaces such
that every µ
i
is strictly positive, and λ the product measure on X =
i∈I
X
i
. Then
(i) λ is a completion regular quasiRadon measure;
(ii) if F ⊆ X is a closed selfsupporting set, there is a countable set J ⊆ I such that F is determined by
coordinates in J, so F is a zero set.
proof (a) Write Λ for the domain of λ, and  for the family of subsets of X of the form
i∈I
G
i
where
G
i
∈ T
i
for every i ∈ I and ¦i : G
i
,= X
i
¦ is ﬁnite. Then  is a base for the topology of X, included in Λ.
For J ⊆ I let λ
J
be the product measure on X
J
=
i∈J
X
i
and Λ
J
its domain. Write 
(J)
for the family
of subsets of X
J
of the form
i∈J
G
i
where G
i
∈ T
i
for each i ∈ J and ¦i : G
i
,= X
i
¦ is ﬁnite.
(b) Consider ﬁrst the case in which I is countable. In this case X also is separable and metrizable
(4A2P(av)), while Λ includes a base for its topology. Also λ is a complete probability measure and inner
regular with respect to the closed sets (412Ua), so must be a quasiRadon measure, by 415D(i).
(c) Now consider uncountable I. The key to the proof is the following fact: if 1 ⊆  has union W, then
W ∈ Λ and λW = sup
V ∈\
∗ λV , where 1
∗
is the set of unions of ﬁnite subsets of 1.
PPP (i) By 215B(iv), there is a countable set 1
1
⊆ 1 such that λ(U ¸ W
1
) = 0 for every U ∈ 1, where
W
1
=
1
1
. Every member of  is determined by coordinates in some ﬁnite set (see 254M for this concept),
so there is a countable set J ⊆ I such that every member of 1
1
is determined by coordinates in J, and W
1
also is determined by coordinates in J. Let π
J
: X → X
J
be the canonical map. Because it is an open map
(4A2B(fi)), π
J
[W] and π
J
[W
1
] are open in X
J
, and belong to Λ
J
, by (b).
(ii) ??? Suppose, if possible, that λ
J
π
J
[W] > λ
J
π
J
[W
1
]. Since π
J
[W] =
¦π
J
[U] : U ∈ 1¦, while λ
J
is
quasiRadon and all the sets π
J
[U] are open, there must be some U ∈ 1 such that λ
J
(π
J
[U] ¸ π
J
[W
1
]) > 0
(414Ea). Now π
J
is inversemeasurepreserving (254Oa), so
0 < λπ
−1
J
[π
J
[U] ¸ π
J
[W
1
]] = λ(π
−1
J
[π
J
[U]] ¸ π
−1
J
[π
J
[W
1
]]) = λ(π
−1
J
[π
J
[U]] ¸ W
1
),
because W
1
is determined by coordinates in J.
At this point note that U is of the form
i∈I
G
i
, where G
i
∈ T
i
for each I, so we can express U as
U
t
∩ U
tt
, where U
t
= π
−1
J
[π
J
[U]] and U
tt
= π
−1
I\J
[π
I\J
[U]]. U
t
depends on coordinates in J and U
tt
depends
on coordinates in I ¸ J. In this case
λ(U ¸ W
1
) = λ(U
tt
∩ U
t
¸ W
1
) = λU
tt
λ(U
t
¸ W
1
),
because U
tt
depends on coordinates in I ¸ J and U
t
¸ W
1
depends on coordinates in J, and we can identify
λ with the product λ
I\J
λ
J
(254N). But now recall that every µ
i
is strictly positive. Since U is surely not
empty, no G
i
can be empty and no µ
i
G
i
can be 0. Consequently
i∈I
µ
i
G
i
> 0 (because only ﬁnitely many
terms in the product are less than 1) and λU > 0; more to the point, λU
tt
> 0. Since we already know that
λ(U
t
¸ W
1
) > 0, we have λ(U ¸ W
1
) > 0. But this contradicts the ﬁrst sentence of (i) just above. XXX
(iii) Thus λ
J
π
J
[W] = λ
J
π
J
[W
1
]. But this means that λπ
−1
J
[π
J
[W]] = λW
1
. Since λ is complete and
W
1
⊆ W ⊆ π
−1
J
[π
J
[W]], λW is deﬁned and equal to λW
1
.
Taking ¸V
n
)
n∈N
to be a sequence running over 1
1
∪ ¦∅¦, we have
λW = λW
1
= λ(
n∈N
V
n
) = sup
n∈N
λ(
i≤n
V
i
) ≤ sup
V ∈\
∗ λV ≤ λW,
so λW = sup
V ∈\
∗ λV , as required. QQQ
(d) Thus we see that λ is a topological measure. But it is also τadditive. PPP If J is an upwardsdirected
family of open sets in X with union W
∗
, set
415G QuasiRadon measure spaces 61
1 = ¦U : U ∈ , ∃ W ∈ J, U ⊆ W¦.
Then W
∗
=
1, so λW
∗
= sup
V ∈\
∗ λV , where 1
∗
is the set of ﬁnite unions of members of 1. But because
J is upwardsdirected, every member of 1
∗
is included in some member of J, so
λW
∗
= sup
V ∈\
∗ λV ≤ sup
W∈V
λW ≤ λW
∗
.
As J is arbitrary, λ is τadditive. QQQ
(e) As in (b) above, we know that λ is a complete probability measure and is inner regular with respect
to the closed sets, so it is a quasiRadon measure. Because λ is inner regular with respect to the zero sets
(412Ub), it is completion regular.
(f ) Now suppose that F ⊆ X is a closed selfsupporting set. By 254Oc, there is a set W ⊆ X, determined
by coordinates in some countable set J ⊆ I, such that W´F is negligible. ??? Suppose, if possible, that
x ∈ F and y ∈ X ¸ F are such that xJ = yJ. Then there is a U ∈  such that y ∈ U ⊆ X ¸ F. As in
(bii) above, we can express U as U
t
∩ U
tt
where U
t
, U
tt
∈  are determined by coordinates in J and I ¸ J
respectively. In this case,
λ(F ∩ U) = λ(W ∩ U) = λ(W ∩ U
t
) λU
tt
= λ(F ∩ U
t
) λU
tt
> 0,
because x ∈ F ∩U
t
and F is selfsupporting, while U
tt
,= ∅ and λ is strictly positive. But F ∩U = ∅, so this
is impossible. XXX
Thus F is determined by coordinates in the countable set J. Consequently it is of the form π
−1
J
[π
J
[F]].
But π
J
[X ¸ F] is open (4A2B(fi)), so its complement π
J
[F] is closed. Now X
J
is metrizable (4A2P(av)),
so π
J
[F] is a zero set (4A2Lc) and F is a zero set (4A2C(biv)).
415F Corollary(a) If Y is either [0, 1[ or ]0, 1[, endowed with Lebesgue measure, and I is any set, then
Y
I
, with the product topology and measure, is a quasiRadon measure space.
(b) If ¸ν
i
)
i∈I
is a family of probability distributions on R, in the sense of ¸271 (that is, Radon probability
measures), and every ν
i
is strictly positive, then the product measure on R
I
is a quasiRadon measure.
Remark See also 416U below, and 453I, where there is an alternative proof of the main step in 415E,
applicable to some further cases. Yet another approach, most immediately applicable to [0, 1[
I
, is in 443Xq.
For further facts about these product measures, see ¸417, particularly 417G and 417M.
415G Comparing quasiRadon measures: Proposition Let X be a topological space, and µ, ν two
quasiRadon measures on X. Then the following are equiveridical:
(i) µF ≤ νF for every closed set F ⊆ X;
(ii) domν ⊆ domµ and µE ≤ νE for every E ∈ domν.
If ν is locally ﬁnite, we can add
(iii) µG ≤ νG for every open set G ⊆ X;
(iv) there is a base  for the topology of X such that G ∪ H ∈  for all G, H ∈  and µG ≤ νG for
G ∈ .
proof (a) Of course (ii)⇒(i). Suppose that (i) is true. Observe that if E ∈ domµ ∩ domν (for instance, if
E ⊆ X is Borel), then
µE = sup¦µF : F ⊆ E is closed¦ ≤ sup¦νF : F ⊆ E is closed¦ = νE.
Set H = ¦H : H ⊆ X is open, νH < ∞¦, and W =
H. Then ν(X ¸ W) = 0, because ν is eﬀectively
locally ﬁnite, so µ(X ¸ W) = 0. Set T = ¦F : F ⊆ X is closed, µF < ∞¦.
Take any E ∈ domν. If F ∈ T and > 0, then µ(F ∩ W) = µF, so there is an H ∈ H such
that µ(F ∩ H) ≥ µF − . Now there are closed sets F
1
⊆ F ∩ H ∩ E, F
2
⊆ F ∩ H ¸ E such that
νF
1
+ νF
2
≥ ν(F ∩ H) − , that is, ν((F ∩ H) ¸ (F
1
∪ F
2
)) ≤ , so that µ((F ∩ H) ¸ (F
1
∪ F
2
)) ≤ and
µF
1
+µF
2
≥ µ(F ∩ H) −. This means that
µ
∗
(F ∩ E) +µ
∗
(F ¸ E) ≥ µ(F ∩ H) − ≥ µF −.
62 Topologies and measures 415G
As is arbitrary, µ
∗
(F ∩ E) + µ
∗
(F ¸ E) ≥ µF; as µ is inner regular with respect to T, µ measures E, by
413F(vii).
Thus domν ⊆ domµ; and we have already observed that µE ≤ νE whenever E is measured by both.
(b) The ﬁrst sentence in the proof of (a) shows that (i)⇒(iii), and (iii)⇒(iv) is trivial. If (iv) is true and
G ⊆ X is open, then 1 = ¦V : V ∈ , V ⊆ G¦ is upwardsdirected and has union G, so
µG = sup
V ∈\
µV ≤ sup
V ∈\
νV = νG.
Thus (iv)⇒(iii).
Now assume that ν is locally ﬁnite and that (iii) is true. ??? Suppose, if possible, that F ⊆ X is a closed
set such that νF < µF. Then H, as deﬁned in part (a) of the proof, is upwardsdirected and has union X,
so there is an H ∈ H such that νF < µ(F ∩ H). Now there is a closed set F
t
⊆ H ¸ F such that
νF
t
> ν(H ¸ F) −µ(F ∩ H) +νF ≥ νH −µ(F ∩ H).
Set G = H ¸ F
t
, so that F ∩ H ⊆ G and
νG = νH −νF
t
< µ(F ∩ H) ≤ µG,
which is impossible. XXX
This shows that (provided that ν is locally ﬁnite) (iii)⇒(i).
415H Uniqueness of quasiRadon measures: Proposition Let (X, T) be a topological space and
µ, ν two quasiRadon measures on X. Then the following are equiveridical:
(i) µ = ν;
(ii) µF = νF for every closed set F ⊆ X;
(iii) µG = νG for every open set G ⊆ X;
(iv) there is a base  for the topology of X such that G∪ H ∈  for every G, H ∈  and µ = ν;
(v) there is a base  for the topology of X such that G∩ H ∈  for every G, H ∈  and µ = ν.
proof Of course (i) implies all the others. (ii)⇒(i) is immediate from 415G (see also 412L). If (iii) is true,
then, for any closed set F ⊆ X,
µF = sup¦µ(G∩ F) : G ∈ T, µG < ∞¦
= sup¦µG−µ(G¸ F) : G ∈ T, µG < ∞¦
= sup¦νG−ν(G¸ F) : G ∈ T, νG < ∞¦ = νF;
so (iii)⇒(ii). (iv)⇒(iii) by the argument of (iv)⇒(iii) in the proof of 415G.
Finally, suppose that (v) is true. Then µ(G
0
∪ . . . ∪ G
n
) = ν(G
0
∪ . . . ∪ G
n
) for all G
0
, . . . , G
n
∈ . PPP
Induce on n. For the inductive step to n ≥ 1, if any G
i
has inﬁnite measure (for either measure) the result
is trivial. Otherwise,
µ(G
0
∪ . . . ∪ G
n
) = µ(
_
i<n
G
i
) +µG
n
−µ(
_
i<n
(G
n
∩ G
i
))
= ν(
_
i<n
G
i
) +νG
n
−ν(
_
i<n
(G
n
∩ G
i
)) = ν(G
0
∪ . . . ∪ G
n
). QQQ
So µ and ν agree on the base ¦G
0
∪ . . . ∪ G
n
: G
0
, . . . , G
n
∈ ¦, and (iv) is true.
415I Proposition Let X be a completely regular topological space and µ, ν two quasiRadon measures
on X such that
_
fdµ =
_
fdν whenever f : X → R is a bounded continuous function integrable with
respect to both measures. Then µ = ν.
proof ??? Otherwise, there is an open set G ⊆ X such that µG ,= νG; suppose µG < νG. Because ν is
eﬀectively locally ﬁnite, there is an open set G
t
⊆ G such that µG < νG
t
< ∞. Now the cozero sets form a
base for the topology of X, so H = ¦H : H ⊆ G
t
is a cozero set¦ has union G
t
; as ν is τadditive, there is
an H ∈ H such that νH > µG. Express H as ¦x : g(x) > 0¦ where g : X → [0, ∞[ is continuous. For each
n ∈ N, set f
n
= ng ∧ χX; then ¸f
n
)
n∈N
is a nondecreasing sequence with limit χH, so there is an n ∈ N
such that
_
f
n
dν > µG ≥
_
f
n
dµ. But f
n
is both µintegrable and νintegrable because µG and νH are
both ﬁnite. XXX
415K QuasiRadon measure spaces 63
415J Proposition Let X be a regular topological space, Y a subspace of X, and ν a quasiRadon
measure on Y . Then there is a quasiRadon measure µ on X such that µE = ν(E∩Y ) whenever µ measures
E, that is, Y has full outer measure in X and ν is the subspace measure on Y .
proof Write B for the Borel σalgebra of X, and set µ
0
E = ν(E ∩ Y ) for every E ∈ B. Then it is easy to
see that µ
0
is a τadditive Borel measure on X. Moreover, µ
0
is eﬀectively locally ﬁnite. PPP If E ∈ B and
µ
0
E > 0, there is a relatively open set H ⊆ Y such that νH < ∞ and ν(H ∩ E ∩ Y ) > 0. Now H is of the
form G∩ Y where G ⊆ X is open, and we have µ
0
G = νH < ∞, µ
0
(E ∩ G) = ν(H ∩ E ∩ Y ) > 0. QQQ
By 415Cb, the c.l.d. version µ of µ
0
is a quasiRadon measure on X. If E ∈ domµ, then E ∩Y ∈ domν.
PPP Let T
Y
be the set of relatively closed subsets of Y of ﬁnite measure for ν. If F ∈ T
Y
, it is expressible as
F
t
∩Y where F
t
is a closed subset of X, and µF
t
= µ
0
F
t
= νF is ﬁnite. So there are E
1
, E
2
∈ B such that
E
1
⊆ E ∩ F
t
⊆ E
2
and µE
1
= µ(E ∩ F
t
) = µE
2
. Accordingly
E
1
∩ Y ⊆ E ∩ Y ∩ F ⊆ E
2
∩ Y
and
ν(E
1
∩ Y ) = ν(E
2
∩ Y ) = µ(E ∩ F
t
)
is ﬁnite. This means that E ∩ Y ∩ F ∈ domν; because ν is complete and locally determined and inner
regular with respect to T
Y
, E ∩ Y ∈ domν, by 412Ja. QQQ
If E ∈ domµ, then
µE = sup¦µF : F ⊆ E is closed¦
= sup¦ν(F ∩ Y ) : F ⊆ E is closed¦ ≤ ν(E ∩ Y ).
On the other hand, if γ < ν(E ∩ Y ), there is a relatively open set H ⊆ Y such that νH < ∞ and
ν(E ∩ Y ∩ H) ≥ γ (412F). Let G ⊆ X be an open set such that G∩ Y = H. Then
µE ≥ µG−µ(G¸ E) = νH −ν(H ¸ E) = ν(E ∩ Y ∩ H) ≥ γ.
As γ is arbitrary, µE = ν(E ∩ Y ).
Thus µE = ν(E ∩ Y ) whenever µ measures E. So if E, F ∈ domµ and E ∩ Y ⊆ F,
µE = ν(E ∩ Y ) ≤ ν(F ∩ Y ) = µF;
as F is arbitrary, µ
∗
(E ∩Y ) = µE; as E is arbitrary, Y has full outer measure in X. Moreover, if µ
Y
is the
subspace measure on Y , µ
Y
H = µ
∗
H = νH whenever H ∈ domµ
Y
, that is, H = E∩Y for some E ∈ domµ.
Now µ
Y
, like ν, is a quasiRadon measure on Y (415B), and they agree on the (relatively) closed subsets of
Y , so are equal, by 415H.
415K I come now to a couple of basic results on the construction of quasiRadon measures. The ﬁrst
follows 413J.
Theorem Let X be a topological space and / a family of closed subsets of X such that
∅ ∈ /,
(†) K ∪ K
t
∈ / whenever K, K
t
∈ / are disjoint,
(‡) F ∈ / whenever K ∈ / and F ⊆ K is closed.
Let φ
0
: / → [0, ∞[ be a functional such that
(α) φ
0
K = φ
0
L + sup¦φ
0
K
t
: K
t
∈ /, K
t
⊆ K ¸ L¦ whenever K, L ∈ / and L ⊆ K,
(β) inf
K∈K
φ
0
K = 0 whenever /
t
is a nonempty downwardsdirected subset of / with empty
intersection,
(γ) whenever K ∈ / and φ
0
K > 0, there is an open set G such that the supremum
sup
K
∈K,K
⊆G
φ
0
K
t
is ﬁnite, while φ
0
K
t
> 0 for some K
t
∈ / such that K
t
⊆ K ∩ G.
Then there is a unique quasiRadon measure on X extending φ
0
and inner regular with respect to /.
proof By 413J, there is a complete locally determined measure µ on X, inner regular with respect to /,
and extending φ
0
; write Σ for the domain of µ. If F ⊆ X is closed, then K ∩ F ∈ / ⊆ Σ for every K ∈ /,
so F ∈ Σ, by 413F(ii); accordingly every open set is measurable. Because µ is inner regular with respect to
64 Topologies and measures 415K
/ it is surely inner regular with respect to the closed sets. If E ∈ Σ and µE > 0, there is a K ∈ / such that
K ⊆ E and µK > 0; now (γ) tells us that there is an open set G such that µG < ∞ and µ(G ∩ K) > 0,
so that µ(G∩ E) > 0. As E is arbitrary, µ is eﬀectively locally ﬁnite. Now suppose that ( is a nonempty
upwardsdirected family of open sets with union H, and that γ < µH. Then there is a K ∈ / such that
K ⊆ H and µK > γ. Applying the hypothesis (β) to /
t
= ¦K¸G : G ∈ (¦, we see that inf
G∈Ç
µ(K¸G) = 0,
so that
sup
G∈Ç
µG ≥ sup
G∈Ç
µ(K ∩ G) = µK ≥ γ.
As ( and γ are arbitrary, µ is τadditive. So µ is a quasiRadon measure.
415L Proposition Let (X, Σ
0
, µ
0
) be a measure space and T a topology on X such that µ
0
is τadditive,
eﬀectively locally ﬁnite and inner regular with respect to the closed sets, and Σ
0
includes a base for T. Then
µ
0
has a unique extension to a quasiRadon measure µ on X such that
(i) µF = µ
∗
0
F whenever F ⊆ X is closed and µ
∗
0
F < ∞,
(ii) µG = (µ
0
)
∗
G whenever G ⊆ X is open,
(iii) the embedding Σ
0
⊆ Σ identiﬁes the measure algebra (A
0
, ¯ µ
0
) of µ
0
with an orderdense
subalgebra of the measure algebra (A, ¯ µ) of µ, so that the subrings A
f
0
, A
f
of elements of ﬁnite
measure coincide, and L
p
(µ
0
) may be identiﬁed with L
p
(µ) for 1 ≤ p < ∞,
(iv) whenever E ∈ Σ and µE < ∞, there is an E
0
∈ Σ
0
such that µ(E´E
0
) = 0,
(v) for every µintegrable realvalued function f there is a µ
0
integrable function g such that
f = g µa.e.
If µ
0
is complete and locally determined, then we have
(i)
t
µF = µ
∗
0
F for every closed F ⊆ X.
If µ
0
is localizable, then we have
(iii)
t
A
0
= A, so that L
0
(µ)
∼
= L
0
(µ
0
), L
∞
(µ)
∼
= L
∞
(µ
0
),
(iv)
t
for every E ∈ Σ there is an E
0
∈ Σ
0
such that µ(E´E
0
) = 0,
(v)
t
for every Σmeasurable realvalued function f there is a Σ
0
measurable realvalued function
g such that f = g µa.e.
proof (a) Let / be the set of closed subsets of X of ﬁnite outer measure for µ
0
. Note that µ
0
is inner
regular with respect to /, because it is inner regular with respect to the closed sets and also with respect
to the sets of ﬁnite measure.
It is obvious from its deﬁnition that / satisﬁes (†) and (‡) of 415K. For K ∈ /, set φ
0
K = µ
∗
0
K. Then
φ
0
satisﬁes (α)(γ) of 415K.
PPP (ααα) If K, L ∈ / and L ⊆ K, take measurable envelopes E
0
, E
1
∈ Σ
0
of K, L respectively. (i) Let
> 0. Because µ
0
is inner regular with respect to the closed sets, there is a closed set F ∈ Σ
0
such that
F ⊆ E
0
¸ E
1
and µF ≥ µ
0
(E
0
¸ E
1
) −. Set K
t
= F ∩ K. Then K
t
∈ / and
φ
0
K
t
= µ
∗
0
(F ∩ K) = µ
0
(F ∩ E
0
) = µ
0
F ≥ µ
0
E
0
−µ
0
E
1
− = φ
0
K −φ
0
L −.
As is arbitrary, we have
φ
0
K ≤ φ
0
L + sup¦φ
0
K
t
: K
t
∈ /, K
t
⊆ K ¸ L¦.
(ii) On the other hand, ??? suppose, if possible, that there is a closed set K
t
⊆ K¸L such that µ
∗
0
L+µ
∗
0
K
t
>
µ
∗
K. Let E
2
be a measurable envelope of K
t
, so that µ
0
E
1
+µ
0
E
2
> µ
0
E
0
; since
µ
0
(E
1
¸ E
0
) = µ
∗
0
(L ¸ E
0
) = µ
∗
0
∅ = 0, µ
0
(E
2
¸ E
0
) = µ
∗
0
(K
t
¸ E
0
) = 0,
µ
0
(E
1
∩E
2
) > 0. Because µ
0
is eﬀectively locally ﬁnite, there is a measurable open set G
0
, of ﬁnite measure,
such that µ
0
(G
0
∩ E
1
∩ E
2
) > 0. Set
( = ¦G∪ G
t
: G, G
t
∈ Σ
0
∩ T, G ⊆ G
0
¸ L, G
t
⊆ G
0
¸ K
t
¦.
Then ( is an upwardsdirected family of measurable open sets, and because Σ
0
includes a base for the
topology of X, its union is (G
0
¸L)∪(G
0
¸K
t
) = G
0
. So there is an H ∈ ( such that µ
0
H > µ
0
G
0
−µ
0
(E
1
∩E
2
),
415L QuasiRadon measure spaces 65
that is, there are open sets G, G
t
∈ Σ
0
such that G ⊆ G
0
¸L, G
t
⊆ G
0
¸K
t
and µ
0
((G∪G
t
) ∩E
1
∩E
2
)) > 0.
But we must have
µ
0
(G∩ E
1
) = µ
∗
0
(G∩ L) = 0, µ
0
(G
t
∩ E
2
) = µ
∗
0
(G
t
∩ K
t
) = 0,
so this is impossible. XXX
Accordingly
φ
0
K ≥ φ
0
L + sup¦φ
0
K
t
: K
t
∈ /, K
t
⊆ K ¸ L¦,
so that φ
0
satisﬁes condition (α) of 415K.
(βββ) Let /
t
⊆ / be a nonempty downwardsdirected family with empty intersection. Fix K
0
∈ /
t
and
> 0. Let E
0
be a measurable envelope of K
0
and G
0
a measurable open set of ﬁnite measure such that
µ
0
(G
0
∩ E
0
) ≥ µ
0
E
0
−. Then
( = ¦G : G ∈ Σ
0
∩ T, G ⊆ G
0
¸ K for some K ∈ /
t
such that K ⊆ K
0
¦
is an upwardsdirected family of measurable open sets, and its union is G
0
¸
/
t
= G
0
, again because Σ
0
includes a base for the topology T. So there is a G ∈ ( such that µ
0
G ≥ µ
0
G
0
−. Let K ∈ /
t
be such that
K ⊆ K
0
and G∩ K = ∅; then
φ
0
K = µ
∗
0
K ≤ µ
0
(E
0
¸ G) ≤ µ
0
(E
0
¸ G
0
) +µ
0
(G
0
¸ G) ≤ 2.
As is arbitrary, inf
K∈K
φ
0
K = 0.
(γγγ) If K ∈ / and φ
0
K > 0, let E
0
be a measurable envelope of K. Then there is a measurable open
set G of ﬁnite measure such that µ
0
(G∩E
0
) > 0. Of course sup
K
∈K,K
⊆G
φ
0
K
t
≤ µ
0
G < ∞; but also there
is a measurable closed set K
t
⊆ G∩ E
0
such that µ
0
K
t
> 0, in which case φ
0
(K ∩ K
t
) = µ
0
(E
0
∩ K
t
) > 0.
So φ
0
satisﬁes condition (γ). QQQ
(b) By 415K, φ
0
has an extension to a quasiRadon measure µ on X which is inner regular with respect
to /. Write Σ for the domain of µ. Note that, for K ∈ /,
µK = φ
0
K = µ
∗
0
K,
so we can already be sure that the conclusion (i) of the proposition is satisﬁed. Now µ extends µ
0
.
PPP(i) Take any K ∈ /. Let E
0
∈ Σ
0
be a measurable envelope of K for the measure µ
0
. If E ∈ Σ
0
, then
surely
µ
∗
(K ∩ E) = sup¦µK
t
: K
t
∈ /, K
t
⊆ K ∩ E¦
= sup¦µ
∗
0
K
t
: K
t
∈ /, K
t
⊆ K ∩ E¦ ≤ µ
∗
0
(K ∩ E).
On the other hand, given γ < µ
∗
0
(K ∩ E) = µ
0
(E
0
∩ E), there is a closed set F ∈ Σ
0
such that F ⊆ E
0
∩ E
and µ
0
F ≥ γ, so that
µ
∗
(K ∩ E) ≥ µ(K ∩ F) = µ
∗
0
(K ∩ F) = µ
0
(E
0
∩ F) ≥ γ.
Thus µ
∗
(K ∩ E) = µ
∗
0
(K ∩ E) for every K ∈ /, E ∈ Σ
0
.
(ii) If K ∈ /, E ∈ Σ
0
then
µ
∗
(K ∩ E) +µ
∗
(K ¸ E) = µ
∗
0
(K ∩ E) +µ
∗
0
(K ¸ E) = µ
∗
0
K = µK.
Because µ is complete and locally determined and inner regular with respect to /, E ∈ Σ (413F(iv)). Thus
Σ
0
⊆ Σ.
(iii) For any E ∈ Σ
0
, we now have
µE = sup¦µK : K ∈ /, K ⊆ E¦ = sup¦µ
∗
0
K : K ∈ /, K ⊆ E¦
≤ µ
0
E = sup¦µ
0
K : K ∈ / ∩ Σ
0
, K ⊆ E¦ ≤ µE.
As E is arbitrary, µ extends µ
0
. QQQ
66 Topologies and measures 415L
(c) Because Σ
0
∩ T is a base for T, closed under ﬁnite unions, µ is unique, by 415Hc.
(d) Now for the conditions (i)(v). I have already noted that (i) is guaranteed by the construction.
Concerning (ii), if G ⊆ X is open, we surely have (µ
0
)
∗
G ≤ µ
∗
G = µG because µ extends µ
0
. On the other
hand, writing ( = ¦G
t
: G
t
∈ Σ
0
∩ T, G
t
⊆ G¦, ( is upwardsdirected and has union G, so
µG = sup
G
∈Ç
µG
t
= sup
G
∈Ç
µ
0
G
t
≤ (µ
0
)
∗
G.
So (ii) is true.
Because µ extends µ
0
, the embedding Σ
0
⊆ Σ corresponds to a measurepreserving embedding of A
0
as
a σsubalgebra of A. To see that A
0
is orderdense in A, take any nonzero a ∈ A. This is expressible as E
•
for some E ∈ Σ with µE > 0. Now there is a K ∈ / such that K ⊆ E and µK > 0. There is an E
0
∈ Σ
0
which is a measurable envelope for K with respect to µ
0
, so that
µE
0
= µ
0
E
0
= µ
∗
0
K = µK.
But this means that
0 ,= E
•
0
= K
•
⊆ E
•
= a
in A, while E
•
0
∈ A
0
. As a is arbitrary, A
0
is orderdense in A.
If a ∈ A
f
, then B = ¦b : b ∈ A
0
, b ⊆ a¦ is upwardsdirected and sup
b∈B
¯ µ
0
b ≤ ¯ µa is ﬁnite; accordingly B
has a supremum in A
0
(321C), which must also be its supremum in A, which is a (313O, 313K). So a ∈ A
0
.
Thus A
f
can be identiﬁed with A
f
0
. But this means that, for any p ∈ [1, ∞[, L
p
(µ)
∼
= L
p
(A, ¯ µ) is identiﬁed
with L
p
(A
0
, ¯ µ
0
)
∼
= L
p
(µ) (366H). This proves (iii).
Of course (iv) and (v) are just translations of this. If E ∈ Σ and µE < ∞, then E
•
∈ A
f
⊆ A
0
, that is,
there is an E
0
∈ Σ
0
such that µ(E´E
0
) = 0. If f is µintegrable, then f
•
∈ L
1
(µ) = L
1
(µ
0
), that is, there
is a µ
0
integrable function f
0
such that f = f
0
µa.e.
(e) If µ
0
is complete and locally determined and F ⊆ X is an arbitrary closed set, then
µ
∗
0
F = sup
K∈K
µ
∗
0
(F ∩ K) = sup
K∈K
µ(F ∩ K) = sup
K∈K,K⊆F
µK = µF
by 412Jc, because µ and µ
0
are both inner regular with respect to /.
(f ) If µ
0
is localizable, A
0
is Dedekind complete; as it is orderdense in A, the two must coincide (314Ia).
Consequently
L
0
(µ)
∼
= L
0
(A) = L
0
(A
0
)
∼
= L
0
(µ
0
), L
∞
(µ)
∼
= L
∞
(A) = L
∞
(A
0
)
∼
= L
∞
(µ
0
).
So (iii)
t
is true; now (iv)
t
and (v)
t
follow at once.
415M Corollary Let (X, T) be a regular topological space and µ
0
an eﬀectively locally ﬁnite τadditive
measure on X, deﬁned on the σalgebra Σ
0
generated by a base for T. Then µ
0
has a unique extension to
a quasiRadon measure on X.
proof By 414Mb, µ
0
is inner regular with respect to the closed sets. So 415L gives the result.
415N Corollary Let (X, T) be a completely regular topological space, and µ
0
a τadditive eﬀectively
locally ﬁnite Baire measure on X. Then µ
0
has a unique extension to a quasiRadon measure on X.
proof This is a special case of 415M, because the domain Σ
0
of µ
0
, the Baire σalgebra, is generated by
the family of cozero sets, which form a base for T (4A2Fc).
415O Proposition (a) Let (X, T) be a topological space, and µ, ν two quasiRadon measures on X.
Then ν is an indeﬁniteintegral measure over µ iﬀ νF = 0 whenever F ⊆ X is closed and µF = 0.
(b) Let (X, T, Σ, µ) be a quasiRadon measure space, and ν an indeﬁniteintegral measure over µ (deﬁni
tion: 234B). If ν is eﬀectively locally ﬁnite it is a quasiRadon measure.
proof (a) If ν is an indeﬁniteintegral measure over µ, then of course it is zero on all µnegligible closed
sets. So let us suppose that the condition is satisﬁed. Write Σ = domµ and T = domν.
(i) If E ⊆ X is a µnegligible Borel set it is νnegligible, because every closed subset of E must be
µnegligible, therefore νnegligible, and ν is inner regular with respect to the closed sets. In particular,
415P QuasiRadon measure spaces 67
taking U
∗
to be the union of the family  = ¦U : U ∈ T, µU < ∞¦, ν(X ¸ U
∗
) = µ(X ¸ U
∗
) = 0 because
µ is eﬀectively locally ﬁnite. Also, of course, taking V
∗
to be the union of the family 1 = ¦V : V ∈ T,
νV < ∞¦, ν(X ¸ V
∗
) = 0 because ν is eﬀectively locally ﬁnite. Setting ( =  ∩ 1 and G
∗
=
(, we have
G
∗
= U
∗
∩ V
∗
, so G
∗
is νconegligible.
(ii) In fact, every µnegligible set E is νnegligible. PPP??? Otherwise, ν
∗
(E ∩ G
∗
) > 0. Because the
subspace measure ν
E
is quasiRadon (415B), there is a G ∈ ( such that ν
∗
(E ∩ G) > 0. But there is an F
σ
set H ⊆ G¸ E such that µH = µ(G¸ E), and now E ∩ G is included in the µnegligible Borel set G¸ H, so
that ν(E ∩ G) = ν(G¸ H) = 0. XXXQQQ
(iii) Let / be the family of closed subsets F of X such that either F is included in some member of
( or F ∩ G
∗
= ∅. If E ∈ domµ and µE > 0, then there is an F ∈ / such that F ⊆ E and µF > 0. PPP
If µ(E ¸ G
∗
) > 0 take any closed set F ⊆ E ¸ G
∗
with µF > 0. Otherwise, µ(E ∩ G
∗
) > 0. Because the
subspace measure µ
E
is quasiRadon, there is a G ∈ ( such that µ(E ∩ G) > 0; and now we can ﬁnd a
closed set F ⊆ E ∩ G with µF > 0, and F ∈ /. QQQ
(iv) By 412I, there is a decomposition ¸X
i
)
i∈I
for µ such that every X
i
except perhaps one belongs to
/ and that exceptional one, if any, is µnegligible. Now ¸X
i
)
i∈I
is a decomposition for ν. PPP Every X
i
is
measured by ν because it is either closed or µnegligible, and of ﬁnite measure for ν because it is included
in either a member of ( or the νnegligible set X ¸ G
∗
. If E ⊆ X and νE > 0, then ν(E ∩ G
∗
) > 0, so
there must be some G ∈ ( such that ν(E ∩ G) > 0. Now J = ¦i : i ∈ I, µ(X
i
∩ G) > 0¦ is countable, and
ν(G¸
i∈J
X
i
) = µ(G¸
i∈J
X
i
) = 0, so there is an i ∈ J such that ν(X
i
∩ E) > 0. By 213O, ¸X
i
)
i∈I
is a
decomposition for ν. QQQ
(v) It follows that Σ ⊆ T. PPP If E ∈ Σ, then for every i ∈ I there is an F
σ
set H ⊆ E ∩ X
i
such
that E ∩ X
i
¸ H is µnegligible, therefore νnegligible, and E ∩ X
i
∈ T. As i is arbitrary, E ∈ T. QQQ In
fact, ν is the completion of νΣ. PPP If F ∈ T, then for every i ∈ I there is an F
σ
set H
i
⊆ F ∩ X
i
such
that F ∩ X
i
¸ H
i
is νnegligible. Set H =
i∈I
H
i
; because H ∩ X
i
= H
i
belongs to Σ for every i, H ∈ Σ;
and ν(F ¸ H) =
i∈I
ν(F ∩ X
i
¸ H) = 0. Similarly, there is an H
t
∈ Σ such that H
t
⊆ X ¸ F and
ν((X¸ F) ¸ H
t
) = 0, so that H ⊆ F ⊆ X¸ H
t
and ν((X¸ H
t
) ¸ H) = 0. So F is measured by the completion
of νΣ. Since ν itself is complete, it must be the completion of νΣ. QQQ
(vi) By (iv), ν is inner regular with respect to ¦E : E ∈ Σ, µE < ∞¦. By 234G, ν is an indeﬁnite
integral measure over µ.
(b) Let f ∈ L
0
(µ) be a nonnegative function such that νF =
_
f χF dµ whenever this is deﬁned.
Because µ is complete and locally determined, so is ν (234Fb). Because µ is an eﬀectively locally ﬁnite
τadditive topological measure, ν is a τadditive topological measure (414H). Because µ is inner regular
with respect to the closed sets, so is ν (412Q). Since we are assuming in the hypotheses that ν is eﬀectively
locally ﬁnite, it is a quasiRadon measure.
415P Proposition Let (X, T, Σ, µ) be a quasiRadon measure space.
(a) Suppose that (X, T) is completely regular. If 1 ≤ p < ∞ and f ∈ L
p
(µ), then for any > 0 there is a
bounded continuous function g : X →R such that µ¦x : g(x) ,= 0¦ < ∞ and f −g
p
≤ .
(b) Suppose that (X, T) is regular and Lindel¨of. Let f ∈ L
0
(µ) be locally integrable. Then for any > 0
there is a continuous function g : X →R such that f −g
1
≤ .
proof (a) Write C for the set of bounded continuous functions g : X → R such that ¦x : g(x) ,= 0¦ has
ﬁnite measure. Then C is a linear subspace of R
X
included in L
p
= L
p
(µ). Let U be the closure of C in
L
p
, that is, the set of h ∈ L
p
such that for every > 0 there is a g ∈ C such that h − g
p
≤ . Then
U is closed under addition and scalar multiplication. Also χE ∈ U whenever µE < ∞. PPP Let > 0. Set
δ =
1
4
1/p
. Write ( for the family of open sets of ﬁnite measure. Because µ is eﬀectively locally ﬁnite, there
is a G ∈ ( such that µ(E ¸ G) ≤ δ. Let F ⊆ G ¸ E be a closed set such that µF ≥ µ(G ¸ E) − δ; then
µ(E´(G¸ F)) ≤ 2δ. Write H for the family of cozero sets. Because T is completely regular, H is a base for
T; because H is closed under ﬁnite unions (4A2C(biii)) and µ is τadditive, there is an H ∈ H such that
H ⊆ G¸ F and µH ≥ µ(G¸ F) −δ, so that µ(E´H) ≤ 3δ. Express H as ¦x : g(x) > 0¦ where g : X →R
is a continuous function. For each n ∈ N, set g
n
= ng ∧ χX ∈ C; then
68 Topologies and measures 415P
[χE −g
n
[
p
≤ χ(E´H) + (χH −g
n
)
p
for every n, so
_
[χE −g
n
[
p
≤ µ(E´H) +
_
(χH −g
n
)
p
→ µ(E´H)
as n → ∞, because g
n
→ χH. So there is an n ∈ N such that
_
[χE − g
n
[
p
≤ 4δ, that is, χE − g
n

p
≤ .
As is arbitrary, χE ∈ U. QQQ
Accordingly every simple function belongs to U. But if f ∈ L
p
and > 0, there is a simple function h
such that f −h
p
≤
1
2
(244Ha); now there is a g ∈ C such that h−g
p
≤
1
2
and f −g
p
≤ , as claimed.
(b) This time, write ( for the family of open subsets of X such that
_
G
f is ﬁnite, so that ( is an open
cover of X. As X is paracompact (4A2H(bi)), there is a locally ﬁnite family (
0
⊆ ( covering X, which
must be countable (4A2H(bii)).
Let ¸
G
)
G∈Ç
0
be a family of strictly positive real numbers such that
G∈Ç
0
G
≤ (4A1P). Since X
is completely regular (4A2H(bi)), we can apply (a) to see that, for each G ∈ (
0
, there is a continuous
function g
G
: X → R such that
_
[g
G
− f χG[ ≤
G
. Next, because X is normal (4A2H(bi)), there
is a family ¸h
G
)
G∈Ç
0
of continuous functions from X to [0, 1] such that h
G
≤ χG for every G ∈ (
0
and
G∈Ç
0
h
G
(x) = 1 for every x ∈ X (4A2F(dix)).
Set g(x) =
G∈Ç
0
g
G
(x)h
G
(x) for every x ∈ X. Because (
0
is locally ﬁnite, g : X → R is continuous
(4A2Bh). Now
_
[f −g[ =
_
[
G∈Ç
0
(f −g
G
) h
G
[ ≤
G∈Ç
0
_
[(f −g
G
) h
G
[
≤
G∈Ç
0
_
G
[f −g
G
[ ≤
G∈Ç
0
G
≤ ,
as required.
415Q Recall (411P) that if (A, ¯ µ) is a localizable measure algebra, with Stone space (Z, S, T, ν), then ν is
a strictly positive completion regular quasiRadon measure, inner regular with respect to the openandclosed
sets (which are all compact). The following construction is primarily important for Radon measure spaces
(see 416V), but is also of interest for general quasiRadon spaces.
Proposition Let (X, T, Σ, µ) be a quasiRadon measure space and (A, ¯ µ) its measure algebra. Let (Z, S, T, ν)
be the Stone space of (A, ¯ µ). For E ∈ Σ let E
∗
⊆ Z be the openandclosed set corresponding to the image
E
•
of E in A. Deﬁne R ⊆ Z X by saying that (z, x) ∈ R iﬀ x ∈ F whenever F ⊆ X is closed and z ∈ F
∗
.
Set Q = R
−1
[X].
(a) R is a closed subset of Z X.
(b) For any E ∈ Σ, R[E
∗
] is the smallest closed set such that µ(E ¸ R[E
∗
]) = 0. In particular, if F ⊆ X
is closed then R[F
∗
] is the selfsupporting closed set included in F such that µ(F ¸ R[F
∗
]) = 0; and R[Z] is
the support of µ.
(c) Q is of full outer measure in Z.
(d) For any E ∈ Σ, R
−1
[E]´(Q∩E
∗
) is negligible; consequently ν
∗
R
−1
[E] = µE and R
−1
[E]∩R
−1
[X¸E]
is negligible.
(e) For any A ⊆ X, ν
∗
R
−1
[A] = µ
∗
A.
(f) If (X, T) is regular, then R
−1
[G] is relatively open in Q for every open set G ⊆ X, R
−1
[F] is relatively
closed in Q for every closed set F ⊆ X and R
−1
[X ¸ E] = Q¸ R
−1
[E] for every Borel set E ⊆ X.
proof (a)
R =
F⊆X is closed
((Z ¸ F
∗
) X) ∪ (Z F)
is an intersection of closed sets, therefore closed.
(b) Let ( be the family of open sets G ⊆ X such that µ(E ∩ G) = 0, and G
0
=
(; then G
0
∈ (
(414Ea). Set F
0
= X ¸ G
0
, so that F
0
is the smallest closed set such that E ¸ F
0
is negligible, and F
∗
0
⊇ E
∗
.
415Q QuasiRadon measure spaces 69
If (z, x) ∈ R and z ∈ E
∗
we must have x ∈ F
0
. Thus R[E
∗
] ⊆ F
0
. On the other hand, if x ∈ F
0
, and G is an
open set containing x, then G / ∈ ( so µ(G∩E) > 0 and (E∩G)
∗
,= ∅. Accordingly ¦(G∩E)
∗
: x ∈ G ∈ T¦ is
a downwardsdirected family of nonempty openandclosed sets in the compact space Z and has nonempty
intersection, containing a point z say. If H ⊆ X is closed and z ∈ H
∗
, then X¸H is open and z / ∈ (X¸H)
∗
,
so x cannot belong to X ¸ H, that is, x ∈ H; as H is arbitrary, (z, x) ∈ R and x ∈ R[E
∗
]; as x is arbitrary,
R[E
∗
] = F
0
, as claimed.
Of course, when E is actually closed, R[E
∗
] = F
0
⊆ E. Taking E = X we see that R[Z] = R[X
∗
] is the
support of µ.
(c) If W ∈ T and νW > 0, there is a nonempty openandclosed set U ⊆ W, by 322Qa, which must
be of the form E
∗
for some E ∈ Σ. By (b), R[E
∗
] cannot be empty; but E
∗
⊆ W, so R[W] ,= ∅, that is,
W ∩ Q ,= ∅. As W is arbitrary, ν
∗
(Z ¸ Q) = 0, that is, Z is a measurable envelope of Q (413Ei).
(d)(i) Let T be the set of closed subsets of X included in E. Then sup
F∈J
F
•
= E
•
in A (412N), so
E
∗
¸
F∈J
F
∗
is nowhere dense and negligible. Now for each F ∈ T, R[F
∗
] ⊆ F, so Q ∩ F
∗
⊆ R
−1
[F] ⊆
R
−1
[E]. Accordingly
Q∩ E
∗
¸ R
−1
[E] ⊆ E
∗
¸
F∈J
F
∗
is nowhere dense and negligible. (ii) ??? Suppose, if possible, that ν
∗
(R
−1
[E] ¸ E
∗
) > 0. Then there is an
openandclosed set U of ﬁnite measure such that ν
∗
(R
−1
[E] ∩ U ¸ E
∗
) > 0 (use 412Jc). Express U as H
∗
,
where µH < ∞, and let F ⊆ H ¸ E be a closed set such that µ((H ¸ E) ¸ F) < ν
∗
(R
−1
[E] ∩H
∗
¸ E
∗
). Then
we must have ν
∗
(R
−1
[E] ∩ F
∗
) > 0. But R[F
∗
] ⊆ F ⊆ X ¸ E so F
∗
∩ R
−1
[E] = ∅, which is impossible.
XXX (iii) Putting these together, (Q∩ E
∗
)´R
−1
[E] is negligible. (iv) It follows at once that (because Z is a
measurable envelope for Q)
ν
∗
R
−1
[E] = ν
∗
(Q∩ E
∗
) = νE
∗
= µE.
Moreover, applying the result to X ¸ E,
R
−1
[X ¸ E] ∩ R
−1
[E] ⊆ (R
−1
[X ¸ E]´(Q∩ (X ¸ E)
∗
)) ∪ (R
−1
[E]´(Q∩ E
∗
))
is negligible.
(e)(i) Take E ∈ Σ such that A ⊆ E and µE = µ
∗
A; then R
−1
[A] ⊆ R
−1
[E], so
ν
∗
R
−1
[A] ≤ ν
∗
R
−1
[E] ≤ µE = µ
∗
A.
(ii) ??? Suppose, if possible, that ν
∗
R
−1
[A] < µ
∗
A. Let W ∈ T be such that R
−1
[A] ⊆ W and
νW = ν
∗
R
−1
[A]. Then there is an F ∈ Σ such that ν(W´F
∗
) = 0. Since
µF = νF
∗
= νW < µ
∗
A,
µ
∗
(A¸ F) > 0; let G be a measurable envelope of A¸ F disjoint from F. Then G
∗
∩ F
∗
= ∅ so
ν(G
∗
¸ W) = νG
∗
= µG > 0
and there is a nonempty openandclosed V ⊆ G
∗
¸ W; let H ∈ Σ be such that H ⊆ G and V = H
∗
. In
this case, R[V ] is closed and µ(H ¸ R[V ]) = 0, by (a), so that H ∩ R[V ] is measurable, not negligible, and
included in G. But H∩R[V ] ∩A is empty, because V ∩R
−1
[A] is empty, so µ
∗
(H∩R[V ] ∩A) < µ(H∩R[V ]),
and G cannot be a measurable envelope of A¸ F. XXX
Thus ν
∗
R
−1
[A] = µ
∗
A, as claimed.
(f ) Suppose now that (X, T) is regular.
(i) If G ⊆ X is open, R
−1
[G] ∩ R
−1
[X ¸ G] = ∅. PPP If z ∈ R
−1
[G], then there is an x ∈ G such that
(z, x) ∈ R. Let H be an open set containing x such that H ⊆ G. Then x / ∈ X ¸ H so z / ∈ (X ¸ H)
∗
, that is,
z ∈ H
∗
. But
R[H
∗
] ⊆ R[H
∗
] ⊆ H ⊆ G,
so H
∗
∩ R
−1
[X ¸ G] = ∅ and z / ∈ R
−1
[X ¸ G]. QQQ
(ii) It is easy to check that
70 Topologies and measures 415Q
c = ¦E : E ⊆ X, R
−1
[E] ∩ R
−1
[X ¸ E] = ∅¦ = ¦E : E ⊆ X, R
−1
[X ¸ E] = Q¸ R
−1
[E]¦
is a σalgebra of subsets of X (indeed, an algebra closed under arbitrary unions), just because R ⊆ Z X
and R
−1
[X] = Q. Because it contains all open sets, c must contain all Borel sets.
(iii) Now suppose once again that G ⊆ X is open and that z ∈ R
−1
[G]. As in (i) above, there is an
open set H ⊆ G such that z ∈ H
∗
⊆ Z ¸ R
−1
[X ¸ G], so that z ∈ H
∗
∩ Q ⊆ R
−1
[G]. As z is arbitrary,
R
−1
[G] is relatively open in Q.
(iv) Finally, if F ⊆ X is closed, R
−1
[F] = Q¸ R
−1
[X ¸ F] is relatively closed in Q.
415R Proposition Let (X, T, Σ, µ) be a Hausdorﬀ quasiRadon measure space and (Z, S, T, ν) the Stone
space of its measure algebra. Let R ⊆ Z X be the relation described in 415Q. Then
(a) R is (the graph of) a function f;
(b) f is inversemeasurepreserving for the subspace measure ν
Q
on Q = domf, and in fact µ is the image
measure ν
Q
f
−1
;
(c) if (X, T) is regular, then f is continuous.
proof (a) If z ∈ Z and x, y ∈ X are distinct, let G, H be disjoint open sets containing x, y respectively.
Then
(X ¸ G)
∗
∪ (X ¸ H)
∗
= ((X ¸ G) ∪ (X ¸ H))
∗
= Z,
deﬁning
∗
as in 415Q, so z must belong to at least one of (X ¸ G)
∗
, (X ¸ H)
∗
. In the former case (z, x) / ∈ R
and in the latter case (z, y) / ∈ R. This shows that R is a function; to remind us of its new status I will
henceforth call it f. The domain of f is just Q = R
−1
[X].
(b) By 415Qd, f is inversemeasurepreserving for ν
Q
and µ. Suppose that A ⊆ X and f
−1
[A] is in the
domain T
Q
of ν
Q
, that is, is of the form Q ∩ U for some U ∈ T. Take any E ∈ Σ such that µE > 0; then
either ν(E
∗
∩ U) > 0 or ν(E
∗
¸ U) > 0. (α) Suppose that ν(E
∗
∩ U) > 0. Because ν is inner regular with
respect to the openandclosed sets, there is an H ∈ Σ such that H
∗
⊆ E
∗
∩ U and µH = νH
∗
> 0. Now
there is a closed set F ⊆ E∩H with µF > 0. In this case, f[F
∗
] ⊆ F ⊆ E, by 415Qb, while F
∗
∩U ⊆ f
−1
[A],
so f[F
∗
] ⊆ E ∩ A. But this means that
µ
∗
(E ∩ A) ≥ µf[F
∗
] = µF > 0.
(β) If ν(E
∗
¸U) > 0, then the same arguments show that µ
∗
(E¸A) > 0. (γ) Thus µ
∗
(E∩A)+µ
∗
(E¸A) > 0
whenever µE > 0. Because µ is complete and locally determined, A ∈ Σ (413F(vii)).
Thus we see that ¦A : A ⊆ X, f
−1
[A] ∈ T
Q
¦ is included in Σ, and µ is the image measure ν
Q
f
−1
.
(c) If T is regular, then 415Qf tells us that f is continuous.
415X Basic exercises >>>(a) Let (X, T, Σ, µ) be a quasiRadon measure space and E ∈ Σ an atom for
the measure. Show that there is a closed set F ⊆ E such that µF > 0 and F is an atom of Σ, in the sense
that the only measurable subsets of F are ∅ and F. (Hint: 414G.) Show that µ is atomless iﬀ all countable
subsets of X are negligible.
(b) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be any family of quasiRadon measure spaces. Show that the direct sum
measure µ on X = ¦(x, i) : i ∈ I, x ∈ X
i
¦ is a quasiRadon measure when X is given its disjoint union
topology.
(c) Let S be the rightfacing Sorgenfrey topology or lower limit topology on R, that is, the
topology generated by the halfopen intervals of the form [α, β[. Show that Lebesgue measure is completion
regular and quasiRadon for S. (Hint: 114Yj or 221Yb, or 419L.)
(d) Let X be a topological space and µ a complete measure on X, and suppose that there is a conegligible
closed measurable set Y ⊆ X such that the subspace measure on Y is quasiRadon. Show that µ is quasi
Radon.
415Yb QuasiRadon measure spaces 71
(e) Let (X, T, Σ, µ) be a quasiRadon measure space. Show that µ is inner regular with respect to the
family of selfsupporting closed sets included in open sets of ﬁnite measure.
(f ) Let (X, T, Σ, µ) be a quasiRadon measure space. Show that for any E ∈ Σ, > 0 there is an open
set G such that µG ≤ µE + and E ¸ G is negligible.
(g) Find a compact Hausdorﬀ quasiRadon measure space which is not σﬁnite.
(h) Let (X, T, Σ, µ) be an atomless quasiRadon measure space which is outer regular with respect to the
open sets. Show that it is σﬁnite. (Hint: if not, take a decomposition ¸X
i
)
i∈I
in which every X
i
except
one is selfsupporting, and a set A meeting every X
i
in just one point.)
(i) Let (X, Σ, µ) be a σﬁnite measure space in which Σ is countably generated as a σalgebra. Show
that, for a suitable topology on X, the completion of µ is a quasiRadon measure. (Hint: take the topology
generated by a countable subalgebra of Σ, and use the arguments of 415D.)
(j) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of quasiRadon probability spaces such that every µ
i
is strictly
positive, and λ the product measure on X =
i∈I
X
i
. Show that if every T
i
has a countable network, λ is
a quasiRadon measure.
(k) Let ¸X
i
)
i∈I
be a family of separable metrizable spaces, and µ a quasiRadon measure on X =
i∈I
X
i
.
Show that µ is completion regular iﬀ every selfsupporting closed set in X is determined by coordinates in
a countable set. (Hint: 4A2Eb.)
(l) Find two quasiRadon measures µ, ν on the unit interval such that µG ≤ νG for every open set G
but there is a closed set F such that νF < µF.
(m) Let X be a topological space and µ, ν two quasiRadon measures on X. (i) Suppose that µF = νF
whenever F ⊆ X is closed and both µF and νF are ﬁnite. Show that µ = ν. (ii) Suppose that µG = νG
whenever G ⊆ X is open and both µG and νG are ﬁnite. Show that µ = ν.
(n) In 415L, write ˜ µ
0
for the c.l.d. version of µ
0
(213E). Show that µ extends ˜ µ
0
. Show that ˜ µ
0
is
τadditive and inner regular with respect to the closed sets.
(o) Let (X, T, Σ, µ) be a σﬁnite paracompact Hausdorﬀ quasiRadon measure space, and f ∈ L
0
(µ) a
locally integrable function. Show that for any > 0 there is a continuous function g : X → R such that
_
[f −g[ ≤ .
>>>(p) Let (X, T, Σ, µ) be a σﬁnite completely regular quasiRadon measure space. (i) Show that for
every E ∈ Σ there is an F in the Baire σalgebra Ba(X) of X such that µ(E´F) = 0. (Hint: start with
an open set E of ﬁnite measure.) (ii) Show that for every Σmeasurable function f : X → R there is a
Ba(X)measurable function equal almost everywhere to f.
(q) Let (X, Σ, µ) be a measure space and f a µintegrable realvalued function. Show that there is a
unique quasiRadon measure λ on R such that λ¦0¦ = 0 and λ[α, ∞[ = µ
∗
¦x : x ∈ domf, f(x) ≥ α¦,
λ]−∞, −α] = µ
∗
¦x : x ∈ domf, f(x) ≤ −α¦ whenever α > 0; and that
_
hdλ =
_
hfdµ whenever
h ∈ L
0
(λ) and h(0) = 0 and either integral is deﬁned in [−∞, ∞]. (Hint: set λE = µ
∗
f
−1
[E¸ ¦0¦] for Borel
sets E ⊆ R, and use 414Mb, 414O and 235Ib.)
415Y Further exercises (a) Give an example of two quasiRadon measures µ, ν on R such that their
sum, as deﬁned in 112Xe, is not eﬀectively locally ﬁnite, therefore not a quasiRadon measure.
(b) Show that any quasiRadon measure space is isomorphic, as topological measure space, to a subspace
of a compact quasiRadon measure space. (Hint: if X is a T
1
quasiRadon measure space, let
ˆ
X be its
Wallman compactiﬁcation (Engelking 89, 3.6.21).)
72 Topologies and measures 415Yc
(c) Let (X, T, Σ, µ) be a quasiRadon measure space. Show that the following are equiveridical: (i) µ is
outer regular with respect to the open sets; (ii) every negligible subset of X is included in an open set of
ﬁnite measure; (iii) ¦x : µ¦x¦ = 0¦ can be covered by a sequence of open sets of ﬁnite measure.
(d) Show that + : R R →R is continuous for the rightfacing Sorgenfrey topology.
(e) Let r ≥ 1. On R
r
let S be the topology generated by the halfopen intervals [a, b[ where a, b ∈ R
r
(as
deﬁned in ¸115). (i) Show that S is the product topology if each factor is given the rightfacing Sorgenfrey
topology (415Xc). (ii) Show that Lebesgue measure is quasiRadon for S. (Hint: induce on r. See also
417Yi.)
(f ) Let Y ⊆ [0, 1] be a set of full outer measure and zero inner measure for Lebesgue measure µ. Give [0, 1]
the topology T generated by S∪¦Y ¦ where S is the usual topology. Show that the subspace measure ν = µ
Y
is quasiRadon for the subspace topology T
Y
, but that there is no measure λ on X which is quasiRadon
for T and such that the subspace measure λ
Y
is equal to ν.
(g) Find a base  for the topology of X = ¦0, 1¦
N
and two totally ﬁnite (quasi)Radon measures µ, ν on
X such that G∩ H ∈  for all G, H ∈ , µG ≤ νG for every G ∈ , but νX < µX.
(h) Let X be a topological space and ( an open cover of X. Suppose that for each G ∈ ( we are given
a quasiRadon measure µ
G
on G such that µ
G
(U) = µ
H
(U) whenever G, H ∈ ( and U ⊆ G ∩ H is open.
Show that there is a unique quasiRadon measure on X such that each µ
G
is the subspace measure on G.
(Hint: if ¸µ
G
)
G∈Ç
is a maximal family with the given properties, then ( is upwardsdirected.)
(i) Let (X, Σ, µ) be a measure space and T a topology on X, and suppose that there is a family  ⊆ Σ∩T
such that (i) µU < ∞ for every U ∈  (ii) for every U ∈ , T∩Σ∩TU is a base for the subspace topology
of U (iii) if ( is an upwardsdirected family in T ∩ Σ and
( ∈ , then µ(
() = sup
G∈Ç
µG (iv) µ is
inner regular with respect to the closed sets (v) if E ∈ Σ and µE > 0 then there is a U ∈  such that
µ(E ∩ U) > 0. Show that µ has an extension to a quasiRadon measure on X.
(j) Let (X, T, Σ, µ) be a quasiRadon measure space such that T is normal (but not necessarily Hausdorﬀ
or regular). Show that if 1 ≤ p < ∞, f ∈ L
p
(µ) and > 0, there is a bounded continuous function g : X →R
such that f −g
p
≤ and ¦x : g(x) ,= 0¦ has ﬁnite measure.
(k) Let (X, T, Σ, µ) be a completely regular quasiRadon measure space and suppose that we are given
a uniformity deﬁning the topology T. Show that if 1 ≤ p < ∞, f ∈ L
p
(µ) and > 0, there is a bounded
uniformly continuous function g : X →R such that f −g
p
≤ and ¦x : g(x) ,= 0¦ has ﬁnite measure.
(l) Let (X, T, Σ, µ) be a completely regular quasiRadon measure space and τ an extended Fatou norm on
L
0
(µ) such that (i) τL
τ
is an ordercontinuous norm (ii) whenever E ∈ Σ and µE > 0 there is an open set
G such that µ(E ∩ G) > 0 and τ(χG
•
) < ∞. Show that L
τ
∩ ¦f
•
: f : X →R is continuous¦ is normdense
in L
τ
.
(m) Let (X, T, Σ, µ) be a quasiRadon measure space. Show that µ is a compact measure in the sense of
¸342 iﬀ there is a locally compact topology S on X such that (X, S, Σ, µ) is quasiRadon.
415 Notes and comments 415B is particularly important because a very high proportion of the quasi
Radon measure spaces we study are actually subspaces of Radon measure spaces. I would in fact go so far
as to say that when you have occasion to wonder whether all quasiRadon measure spaces have a property,
you should as a matter of habit look ﬁrst at subspaces of Radon measure spaces; if the answer is aﬃrmative
for them, you will have most of what you want, even if the generalization to arbitrary quasiRadon spaces
gives diﬃculties. Of course the reverse phenomenon can also occur. Stone spaces (411P) can be thought of
as quasiRadon compactiﬁcations of Radon measure spaces (416V). But this is relatively rare. Indeed the
reason why I give so few examples of quasiRadon spaces at this point is just that the natural ones arise
from Radon measure spaces. Note however that the quasiRadon product of an uncountable family of Radon
§416 intro. Radon measure spaces 73
probability spaces need not be Radon (see 417Xq), so that 415E here and 417O below are sources of non
Radon quasiRadon measure spaces. Density and lifting topologies can also provide us with quasiRadon
measure spaces (453Xd, 453Xg).
415K is the second in a series of innerregularextension theorems; there will be a third in 416J.
I have been saying since Volume 1 that the business of measure theory, since Lebesgue’s time, has been
to measure as many sets and integrate as many functions as possible. I therefore take seriously any theorem
oﬀering a canonical extension of a measure. 415L and its corollaries can all be regarded as improvement
theorems, showing that a good measure can be made even better. We have already had such improvement
theorems in Chapter 21: the completion and c.l.d. version of a measure (212C, 213E). In all such theorems
we need to know exactly what eﬀect our improvement is having on the other constructions we are interested
in; primarily, the measure algebra and the function spaces. The machinery of Chapter 36 shows that if
we understand the measure algebra(s) involved then the function spaces will give us no further surprises.
Completion of a measure does not aﬀect the measure algebra at all (322Da). Taking the c.l.d. version does
not change A
f
= ¦a : ¯ µa < ∞¦ or L
1
(213Fc, 213G, 322Db, 366H), but can aﬀect the rest of the measure
algebra and therefore L
0
and L
∞
. In this respect, what we might call the ‘quasiRadon version’ behaves
like the c.l.d. version (as could be expected, since the quasiRadon version must itself be complete and
locally determined; cf. 415Xn). The archetypal application of 415L is 415N. We shall see later how Baire
measures arise naturally when studying Banach spaces of continuous functions (436E). 415N will be one of
the keys to applying the general theory of topological measure spaces in such contexts. A virtue of Baire
measures is that inner regularity with respect to closed sets comes almost free (412D); but there can be
unsurmountable diﬃculties if we wish to extend them to Borel measures (439M), and it is important to
know that τadditivity, even in the relatively weak form allowed by the deﬁnition I use here (411C), is often
enough to give a canonical extension to a wellbehaved measure deﬁned on every Borel set. In 415C we have
inner regularity for a diﬀerent reason, and the measure is already known to be deﬁned on every Borel set,
so in fact the quasiRadon version of the measure is just the c.l.d. version (415Xd).
For a volume and a half I have neglected indeﬁniteintegral measures, though they are mentioned in the
exercises; but we shall need them later, and in 415O I spell out a result which it will be useful to be able to
quote. The exact hypotheses are not perhaps instantly predictable; see 414Xe.
One interpretation of the Lifting Theorem is that for a complete strictly localizable measure space
(X, Σ, µ) there is a function g : X → Z, where Z is the Stone space of the measure algebra of µ, such
that E´g
−1
[E
∗
] is negligible for every E ∈ Σ, where E
∗
⊆ Z is the openandclosed set corresponding
to the image of E in the measure algebra (341Q). For a Hausdorﬀ quasiRadon measure space we have a
function f : Q → X, where Q is a dense subset of Z, such that (Q ∩ E
∗
)´f
−1
[E] is negligible for every
E ∈ Σ (415Qd, 415R); moreover, there is a canonical construction for this function. For completeness’ sake,
I have given the result for general, not necessarily Hausdorﬀ, spaces X (415Q); but evidently it will be of
greatest interest for regular Hausdorﬀ spaces (415Rc), especially if they happen to be ‘compact’ in the sense
of ¸342. Perhaps I should remark that in the most important applications, Q is the whole of Z (416Xw).
Of course the question arises, whether fg can be the identity. (Z typically has larger cardinal than X, so
asking for gf to be the identity is a bit optimistic.) This is in fact an important question; I will return to
it in 453M.
416 Radon measure spaces
We come now to the results for which the chapter so far has been preparing. The centre of topological
measure theory is the theory of ‘Radon’ measures (411Hb), measures inner regular with respect to compact
sets. Most of the section is devoted to pulling the earlier work together, and in particular to restating
theorems on quasiRadon measures in the new context. Of course this has to begin with a check that Radon
measures are quasiRadon (416A). It follows immediately that Radon measures are (strictly) localizable
(416B). After presenting a miscellany of elementary facts, I turn to the constructions of ¸413, which take on
simpler and more dramatic forms in this context (416J416P). I proceed to investigate subspace measures
(416R416T) and some special product measures (416U). I end the section with further notes on the forms
which earlier theorems on Stone spaces (416V) and compact measure spaces (416W) take when applied to
Radon measure spaces.
74 Topologies and measures I 416A
416A Proposition A Radon measure space is quasiRadon.
proof Let (X, T, Σ, µ) be a Radon measure space. Because T is Hausdorﬀ, every compact set is closed, so
µ is inner regular with respect to the closed sets. By 411E, µ is τadditive; by 411Gf, it is eﬀectively locally
ﬁnite. Thus all parts of condition (ii) of 411Ha are satisﬁed, and µ is a quasiRadon measure.
416B Corollary A Radon measure space is strictly localizable.
proof Put 416A and 415A together.
416C In order to use the results of ¸415 eﬀectively, it will be helpful to spell out elementary conditions
ensuring that a quasiRadon measure is Radon.
Proposition Let (X, T, Σ, µ) be a locally ﬁnite Hausdorﬀ quasiRadon measure space. Then the following
are equiveridical:
(i) µ is a Radon measure;
(ii) whenever E ∈ Σ and µE > 0 there is a compact set K such that µ(E ∩ K) > 0;
(iii) sup¦K
•
: K ⊆ X is compact¦ = 1 in the measure algebra of µ.
If µ is totally ﬁnite we can add
(iv) sup¦µK : K ⊆ X is compact¦ = µX.
proof (i)⇒(ii) and (ii) ⇐⇒ (iii) are trivial. For (ii)⇒(i), observe that if E ∈ Σ and µE > 0 there is
a compact set K ⊆ E such that µK > 0. PPP There is a compact set K
t
such that µ(E ∩ K
t
) > 0, by
hypothesis; now there is a closed set K ⊆ E ∩K
t
such that µK > 0, because µ is inner regular with respect
to the closed sets, and K is compact. QQQ By 412B, µ is tight. Being a complete, locally determined, locally
ﬁnite topological measure, it is a Radon measure.
When µX < ∞, of course, we also have (ii) ⇐⇒ (iv).
416D Some further elementary facts are worth writing out plainly.
Lemma (a) In a Radon measure space, every compact set has ﬁnite measure.
(b) Let (X, T, Σ, µ) be a Radon measure space, and E ⊆ X a set such that E ∩K ∈ Σ for every compact
K ⊆ X. Then E ∈ Σ.
(c) A Radon measure is inner regular with respect to the selfsupporting compact sets.
(d) Let X be a Hausdorﬀ space and µ a locally ﬁnite complete locally determined measure on X which is
tight (that is, inner regular with respect to the compact sets). If every compact set belongs to the domain
of µ, µ is a Radon measure.
proof (a) 411Ga.
(b) We have only to remember that µ is complete, locally determined and tight, and apply 413F(ii).
(c) If (X, T, Σ, µ) is a Radon measure space, E ∈ Σ and γ < µE, there is a compact set K ⊆ E such
that µK ≥ γ. By 414F, there is a selfsupporting relatively closed set L ⊆ K such that µL = µK; but now
of course L is compact, while L ⊆ E and µL ≥ γ.
(d) Let / be the family of compact subsets of X; write Σ for the domain of µ. If F ⊆ X is closed, then
F ∩ K ∈ / ⊆ Σ for every K ∈ /; accordingly F ∈ Σ. But this means that every closed set, therefore every
open set, belongs to Σ, and µ is a Radon measure.
416E Speciﬁcation of Radon measures In 415H I described some conditions which enable us to
be sure that two quasiRadon measures on a given topological space are the same. In the case of Radon
measures we have a similar list. This time I include a note on the natural ordering of Radon measures.
Proposition Let X be a Hausdorﬀ space and µ, ν two Radon measures on X.
(a) The following are equiveridical:
(i) domν ⊆ domµ and µE ≤ νE for every E ∈ domν;
(ii) µK ≤ νK for every compact set K ⊆ X.
416E Radon measure spaces 75
(iii) µG ≤ νG for every open set G ⊆ X;
(iv) µF ≤ νF for every closed set F ⊆ X.
If X is locally compact, we can add
(v)
_
fdµ ≤
_
fdν for every nonnegative continuous function f : X → R with compact
support.
(b) The following are equiveridical:
(i) µ = ν;
(ii) µK = νK for every compact set K ⊆ X.
(iii) µG = νG for every open set G ⊆ X;
(iv) µF = νF for every closed set F ⊆ X.
If X is locally compact, we can add
(v)
_
fdµ =
_
fdν for every continuous function f : X →R with compact support.
proof (a)(i)⇒(iv)⇒(ii) and (i)⇒(iii) are trivial.
(ii)⇒(i) If (ii) is true, then
µE = sup
K⊆E is compact
µK ≤ sup
K⊆E is compact
νK = νE
for every set E measured by both µ and ν. Also domν ⊆ domµ. PPP Suppose that E ∈ domν and that
K ⊆ X is a compact set such that µK > 0. Then there are compact sets K
1
⊆ K ∩ E, K
2
⊆ K ¸ E such
that
νK
1
+νK
2
≥ ν(K ∩ E) +ν(K ¸ E) −µK = νK −µK.
So
µ(K ¸ (K
1
∪ K
2
)) ≤ ν(K ¸ (K
1
∪ K
2
)) < µK
and µK
1
+µK
2
> 0. This shows that µ
∗
(K∩E) +µ
∗
(K¸E) > 0. As K is arbitrary, E ∈ domµ (413F(vii)).
QQQ
So (i) is true.
(iii)⇒(ii) The point is that if K ⊆ X is compact, then µK = inf¦µG : G ⊆ X is open, K ⊆ G¦. PPP
Because X =
¦µG : G ⊆ X is open, µG < ∞¦, there is an open set G
0
of ﬁnite measure including K.
Now, for any γ > µK, there is a compact set L ⊆ G
0
¸ K such that µL ≥ µG
0
−γ, so that µG ≤ γ, where
G = G
0
¸ L is an open set including K. QQQ
The same is true for ν. So, if (iii) is true,
µK = inf
G⊇K is open
µG ≤ inf
G⊇K is open
νG = νK
for every compact K ⊆ X, and (ii) is true.
(iii)⇒(v) If (iii) is true and f : X → [0, ∞[ is a nonnegative continuous function, then
_
fdµ =
_
∞
0
µ¦x : f(x) > t¦dt
(252O)
≤
_
∞
0
ν¦x : f(x) > t¦dt =
_
fdν.
(v)⇒(iii) If X is locally compact and (v) is true, take any open set G ⊆ X, and consider
A = ¦f : f is a continuous function with compact support from X to [0, 1] and f ≤ χG¦.
Then A is upwardsdirected and sup
f∈A
f(x) = χG(x) for every x ∈ X, by 4A2G(ei). So
µG = sup
f∈A
_
fdµ ≤ sup
f∈A
_
fdν = νG
by 414Ba. As G is arbitrary, (iii) is true.
(b) now follows at once, or from 415H.
76 Topologies and measures I 416F
416F Proposition Let X be a Hausdorﬀ space and µ a Borel measure on X. Then the following are
equiveridical:
(i) µ has an extension to a Radon measure on X;
(ii) µ is locally ﬁnite and tight;
(iii) µ is locally ﬁnite and eﬀectively locally ﬁnite, and µG = sup¦µK : K ⊆ G is compact¦ for every
open set G ⊆ X;
(iv) µ is locally ﬁnite, eﬀectively locally ﬁnite and τadditive, and µG = sup¦µ(G ∩ K) : K ⊆ X is
compact¦ for every open set G ⊆ X.
In this case the extension is unique; it is the c.l.d. version of µ.
proof (a)(i)⇒(iv) If µ = ˜ µB where ˜ µ is a Radon measure and B is the Borel σalgebra of X, then of
course µ is locally ﬁnite and eﬀectively locally ﬁnite and τadditive because ˜ µ is (see 416A) and every open
set belongs to B. Also
µG = sup¦µK : K ⊆ G is compact¦ ≤ sup¦µ(G∩ K) : K ⊆ X is compact¦ ≤ µG
for every open set G ⊆ X, because ˜ µ is tight and compact sets belong to B.
(b)(iv)⇒(iii) Suppose that (iv) is true. Of course µ is locally ﬁnite and eﬀectively locally ﬁnite. Suppose
that G ⊆ X is open and that γ < µG. Then there is a compact K ⊆ X such that µ(G∩ K) > γ. By 414K,
the subspace measure µ
K
is τadditive. Now K is a compact Hausdorﬀ space, therefore regular. By 414Ma
there is a closed set F ⊆ G∩ K such that µ
K
F ≥ γ. Now F is compact, F ⊆ G and µF ≥ γ. As G and γ
are arbitrary, (iii) is true.
(c)(iii)⇒(ii) I have to show that if µ satisﬁes the conditions of (iii) it is tight. Let / be the family of
compact subsets of X and / the family of subsets of X which are either open or closed. Then whenever
A ∈ /, F ∈ Σ and µ(A ∩ F) > 0, there is a K ∈ / such that K ⊆ A and µ(K ∩ F) > 0. PPP Because µ
is eﬀectively locally ﬁnite, there is an open set G of ﬁnite measure such that µ(G ∩ A ∩ F) > 0. (α) If A
is open, then there will be a compact set K ⊆ G ∩ A such that µK > µ(G ∩ A) − µ(G ∩ A ∩ F), so that
µ(K ∩ F > 0. (β) If A is closed, then let L ⊆ G be a compact set such that µL > µG−µ(G∩ A∩ F); then
K = L ∩ A is compact and µ(K ∩ F) > 0. QQQ
By 412C, µ is inner regular with respect to /, as required.
(d)(ii)⇒(i) If µ is locally ﬁnite and tight, let ˜ µ be the c.l.d. version of µ. Then ˜ µ is complete, locally
determined, locally ﬁnite (because µ is), a topological measure (because µ is) and tight (because µ is, using
412Ha); so is a Radon measure. Every compact set has ﬁnite measure for µ, so µ is semiﬁnite and ˜ µ extends
µ (213Hc).
(e) By 416Eb there can be at most one Radon measure extending µ, and we have observed in (c) above
that in the present case it is the c.l.d. version of µ.
416G One of the themes of ¸434 will be the question: on which Hausdorﬀ spaces is every locally ﬁnite
quasiRadon measure a Radon measure? I do not think we are ready for a general investigation of this, but
I can give one easy special result.
Proposition Let (X, T) be a locally compact Hausdorﬀ space and µ a locally ﬁnite quasiRadon measure
on X. Then µ is a Radon measure.
proof µ satisﬁes condition (ii) of 416C. PPP Take E ∈ domµ such that µE > 0. Let ( be the family of
relatively compact open subsets of X; then ( is upwardsdirected and has union X. By 414Ea, there is a
G ∈ ( such that µ(E ∩ G) > 0. But now G is compact and µ(E ∩ G) > 0. QQQ By 416C, µ is a Radon
measure.
416H Corollary Let (X, T) be a locally compact Hausdorﬀ space, and µ a locally ﬁnite, eﬀectively
locally ﬁnite, τadditive Borel measure on X. Then µ is tight and its c.l.d. version is a Radon measure, the
unique Radon measure on X extending µ.
proof By 415Cb, the c.l.d. version ˜ µ of µ is a quasiRadon measure extending µ. Because µ is locally ﬁnite,
so is ˜ µ; by 416G, ˜ µ is a Radon measure. By 416Eb, the extension is unique. Now
416K Radon measure spaces 77
µE = ˜ µE = sup
K⊆E is compact
˜ µK = sup
K⊆E is compact
µK
for every Borel set E ⊆ X, so µ itself is tight.
416I While on the subject of locally compact spaces, I mention an important generalization of a result
from Chapter 24.
Proposition Let (X, T, Σ, µ) be a locally compact Radon measure space. Write C
k
for the space of con
tinuous realvalued functions on X with compact supports. If 1 ≤ p < ∞, f ∈ L
p
(µ) and > 0, there is a
g ∈ C
k
such that f −g
p
≤ .
proof By 415Pa, there is a bounded continuous function h
1
: X → R such that G = ¦x : h
1
(x) ,= 0¦ has
ﬁnite measure and f −h
1

p
≤
1
2
. Let K ⊆ G be a compact set such that h
1

∞
(µ(G¸ K))
1/p
≤
1
2
, and
let h
2
∈ C
k
be such that χK ≤ h
2
≤ χG (4A2G(ei)). Set g = h
1
h
2
. Then g ∈ C
k
and
_
[h
1
−g[
p
≤
_
G\K
[h
1
[
p
≤ µ(G¸ K)h
1

p
∞
,
so h
1
−g
p
≤
1
2
and f −g
p
≤ , as required.
416J I turn now to constructions of Radon measures based on ideas in ¸413.
Theorem Let X be a Hausdorﬀ space. Let / be the family of compact subsets of X and φ
0
: / → [0, ∞[
a functional such that
(α) φ
0
K = φ
0
L + sup¦φ
0
K
t
: K ∈ /, K
t
⊆ K ¸ L¦ whenever K, L ∈ / and L ⊆ K,
(γ) for every x ∈ X there is an open set G containing x such that sup¦φ
0
K : K ∈ /, K ⊆ G¦
is ﬁnite.
Then there is a unique Radon measure on X extending φ
0
.
proof By 413M, there is a unique complete locally determined measure µ on X, extending φ
0
, which is
inner regular with respect to /. By (γ), µ is locally ﬁnite; by 416Dd, it is a Radon measure.
416K Proposition Let X be a regular Hausdorﬀ space. Let / be the family of compact subsets of X,
and φ
0
: / → [0, ∞[ a functional such that
(α
1
) φ
0
K ≤ φ
0
(K ∪ L) ≤ φ
0
K +φ
0
L for all K, L ∈ /,
(α
2
) φ
0
(K ∪ L) = φ
0
K +φ
0
L whenever K, L ∈ / and K ∩ L = ∅,
(γ) for every x ∈ X there is an open set G containing X such that sup¦φ
0
K : K ∈ /, K ⊆
G¦ < ∞.
Then there is a unique Radon measure µ on X such that
µK = inf
G⊆X is open,K⊆G
sup
L⊆G is compact
φ
0
L
for every K ∈ /.
proof (a) For open sets G ⊆ X set
ψG = sup
L∈K,L⊆G
φ
0
L,
and for compact sets K ⊆ X set
φ
1
K = inf¦ψG : G ⊆ X is open, K ⊆ G¦.
Evidently ψG ≤ ψH whenever G ⊆ H. We need to know that ψ(G ∪ H) ≤ ψG + ψH for all open sets G,
H ⊆ X. PPP If L ⊆ G ∪ H is compact, then the disjoint compact sets L ¸ G, L ¸ H can be separated by
disjoint open sets H
t
, G
t
(4A2F(hi)); now L ¸ G
t
⊆ H, L ¸ H
t
⊆ G are compact and cover L, so
φ
0
L ≤ φ
0
(L ¸ G
t
) +φ
0
(L ¸ H
t
) ≤ ψH +ψG.
As L is arbitrary, ψ(G∪ H) ≤ ψG+ψH. QQQ
Moreover, ψ(G∪ H) = ψG+ψH if G∩ H = ∅. PPP If K ⊆ G, L ⊆ H are compact, then
φ
0
K +φ
0
L = φ
0
(K ∪ L) ≤ ψ(G∪ H).
As K and L are arbitrary, ψG+ψH ≤ ψ(G∪ H). QQQ
78 Topologies and measures I 416K
(b) It follows that φ
1
K is ﬁnite for every compact K ⊆ X. PPP Set ( = ¦G : G ⊆ X is open, ψG < ∞¦.
Then (a) tells us that ( is upwardsdirected. But also we are supposing that ( covers X, by (γ). So if
K ⊆ X is compact there is a member of ( including K and φ
1
K < ∞. QQQ
(c) Now φ
1
satisﬁes the conditions of 416J.
PPP(ααα) Suppose that K, L ∈ / and L ⊆ K. Set γ = sup¦φ
1
M : M ∈ /, M ⊆ K ¸ L¦. Take any > 0.
Let G be an open set such that K ⊆ G and ψG ≤ φ
1
K +. If M ∈ / and M ⊆ K ¸ L, there are disjoint
open sets U, V such that L ⊆ U and M ⊆ V (4A2F(hi) again); we may suppose that U ∪ V ⊆ G. In this
case,
φ
1
L +φ
1
M ≤ ψU +ψV = ψ(U ∪ V )
(by the second part of (a) above)
≤ ψG ≤ φ
1
K +.
As M is arbitrary, γ ≤ φ
1
K −φ
1
L +.
On the other hand, there is an open set H such that L ⊆ H and ψH ≤ φ
1
L+. Set F = K ¸ H, so that
F is a compact subset of K ¸ L. Then there is an open set V such that F ⊆ V and ψV ≤ φ
1
F +. In this
case K ⊆ H ∪ V , so
φ
1
K ≤ ψ(H ∪ V ) ≤ ψH +ψV
≤ φ
1
L +φ
1
F + 2 ≤ φ
1
L +γ + 2,
so γ ≥ φ
1
K −φ
1
L −2.
As is arbitrary, γ = φ
1
K −φ
1
L; as K and L are arbitrary, φ
1
satisﬁes condition (α) of 416J.
(γγγ) Any x ∈ X is contained in an open set G such that ψG < ∞; but now sup¦φ
1
K : K ∈ /, K ⊆
G¦ ≤ ψG is ﬁnite. So φ
1
satisﬁes condition (γ) of 416J. QQQ
(d) By 416J, there is a unique Radon measure on X extending φ
1
, as claimed.
416L Corollary Let X be a locally compact Hausdorﬀ space. Let / be the family of compact subsets
of X, and φ
0
: / → [0, ∞[ a functional such that
φ
0
K ≤ φ
0
(K ∪ L) ≤ φ
0
K +φ
0
L for all K, L ∈ /,
φ
0
(K ∪ L) = φ
0
K +φ
0
L whenever K, L ∈ / and K ∩ L = ∅.
Then there is a unique Radon measure µ on X such that
µK = inf¦φ
0
K
t
: K
t
∈ /, K ⊆ int K
t
¦
for every K ∈ /.
proof Observe that φ
0
satisﬁes the conditions of 416K; 416K(γ) is true because X is locally compact.
Deﬁne ψ, φ
1
as in the proof of 416K, and set
φ
t
1
K = inf¦φ
0
K
t
: K
t
∈ /, K ⊆ int K
t
¦
for every K ∈ /. Then φ
t
1
= φ
1
. PPP Let K ∈ /, > 0. (i) There is an open set G ⊆ X such that
K ⊆ G and ψG ≤ φ
1
K + . Now the relatively compact open subsets with closures included in G form an
upwardsdirected cover of K, so there is a K
t
∈ / such that K ⊆ int K
t
and K
t
⊆ G. Accordingly
φ
t
1
K ≤ φ
0
K
t
≤ ψG ≤ φK +.
(ii) There is an L ∈ / such that K ⊆ int L and φ
0
L ≤ φ
t
1
K +, so that
φ
1
K ≤ ψ(int L) ≤ φ
0
L ≤ φ
t
1
K +.
(iii) As K, are arbitrary, φ
t
1
= φ
1
. QQQ
Now 416K tells us that there is a unique Radon measure extending φ
1
, and this is the measure we seek.
416O Radon measure spaces 79
416M The extension theorems in the second half of ¸413 also have important applications to Radon
measures.
Henry’s Theorem (Henry 69) Let X be a Hausdorﬀ space and µ
0
a measure on X which is locally ﬁnite
and tight. Then µ
0
has an extension to a Radon measure µ on X; and the extension may be made in such
a way that whenever µE < ∞ there is an E
0
∈ Σ
0
such that µ(E´E
0
) = 0.
proof All the work has been done in ¸413; we need to check here only that the family / of compact subsets
of X and the measure µ
0
satisfy the hypotheses of 413O. But (†) and (‡) there are elementary, and µ
∗
0
K < ∞
for every K ∈ / by 411Ga.
Now take the measure µ from 413O. It is complete, locally determined and inner regular with respect
to /; also / ⊆ domµ. Because µ
0
is locally ﬁnite and µ extends µ
0
, µ is locally ﬁnite. By 416Dd, µ is a
Radon measure. And the construction of 413O ensures that every set of ﬁnite measure for µ diﬀers from a
member of Σ
0
by a µnegligible set.
416N Proposition Let X be a Hausdorﬀ space and T a subalgebra of TX. Let ν : T → R be a
nonnegative ﬁnitely additive functional. Then there is a Radon measure µ on X such that µX ≤ νX and
µK ≥ νK for every compact set K ∈ T.
proof Use 413S, with / the family of compact subsets of X, and 416Dd.
416O Theorem Let X be a Hausdorﬀ space and T a subalgebra of TX. Let ν : T →R be a nonnegative
ﬁnitely additive functional such that
νE = sup¦νF : F ∈ T, F ⊆ E, F is closed¦ for every E ∈ T,
νX = sup
K⊆X is compact
inf
F∈T,F⊇K
νF.
Then there is a Radon measure µ on X extending ν.
proof (a) For A ⊆ X, write
ν
∗
A = inf
F∈T,F⊇A
νF.
Let ¸K
n
)
n∈N
be a sequence of compact subsets of X such that lim
n→∞
ν
∗
K
n
= νX; replacing K
n
by
i<n
K
i
if necessary, we may suppose that ¸K
n
)
n∈N
is nondecreasing and that K
0
= ∅. For each n ∈ N, set
ν
t
n
E = ν
∗
(E ∩ K
n
)
for every E ∈ T. Then ν
t
n
is additive. PPP (I copy from the proof of 413N.) If E, F ∈ T and E ∩ F = ∅,
ν
t
n
(E ∪ F) = inf¦νG : G ∈ T, K
n
∩ (E ∪ F) ⊆ G¦
= inf¦νG : G ∈ T, K
n
∩ (E ∪ F) ⊆ G ⊆ E ∪ F¦
= inf¦ν(G∩ E) +ν(G∩ F) : G ∈ T, K
n
∩ (E ∪ F) ⊆ G ⊆ E ∪ F¦
= inf¦νG
1
+νG
2
: G
1
, G
2
∈ T, K
n
∩ E ⊆ G
1
⊆ E, K
n
∩ F ⊆ G
2
⊆ F¦
= inf¦νG
1
: G
1
∈ T, K
n
∩ E ⊆ G
1
⊆ E¦
+ inf¦νG
2
: G
2
∈ T, K
n
∩ F ⊆ G
2
⊆ F¦
= ν
t
n
E +ν
t
n
F.
As E and F are arbitrary, ν
t
n
is additive. QQQ
(b) For each n ∈ N, set ν
n
E = ν
t
n+1
E −ν
t
n
E for every E ∈ T; then ν
n
is additive. Because K
n+1
⊇ K
n
,
ν
n
is nonnegative.
If E ∈ T and E ∩ K
n+1
= ∅, then ν
n
E = ν
t
n+1
E = 0. So if we set T
n
= ¦E ∩ K
n+1
: E ∈ T¦, we
have an additive functional ˜ ν
n
: T
n
→ [0, ∞[ deﬁned by setting ˜ ν
n
(E ∩ K
n+1
) = ν
n
E for every E ∈ T.
Also ˜ ν
n
H = sup¦˜ ν
n
K : K ∈ T
n
, K ⊆ H, K is compact¦ for every H ∈ T
n
. PPP Express H as E ∩ K
n+1
where E ∈ T. Given > 0, there is a closed set F ∈ T such that F ⊆ E and νF ≥ νE − ; but now
K = F ∩ K
n+1
∈ T
n
is a compact subset of H, and
˜ ν
n
(H ¸ K) = ν
n
(E ¸ F) ≤ ν
t
n+1
(E ¸ F) ≤ ν(E ¸ F) ≤ ,
80 Topologies and measures I 416O
so ˜ ν
n
K ≥ ˜ ν
n
H −. QQQ
(c) For each n ∈ N, we have a Radon measure µ
n
on K
n+1
, with domain Σ
n
say, such that µ
n
K
n+1
≤
˜ ν
n
K
n+1
and µ
n
K ≥ ˜ ν
n
K for every compact set K ⊆ K
n+1
(416N). Since K
n+1
is itself compact, we must
have µ
n
K
n+1
= ˜ ν
n
K
n+1
. But this means that µ
n
extends ˜ ν
n
. PPP If H ∈ T
n
, > 0 there is a compact set
K ∈ T
n
such that ˜ ν
n
K ≥ ˜ ν
n
H − , so that (µ
n
)
∗
H ≥ µ
n
K ≥ ˜ ν
n
H − ; as is arbitrary, (µ
n
)
∗
H ≥ ˜ ν
n
H.
So there is an F
1
∈ Σ
n
such that F
1
⊆ H and µ
n
F
1
≥ ˜ ν
n
H. Similarly, there is an F
2
∈ Σ
n
such that
F
2
⊆ K
n+1
¸ H and µ
n
F
2
≥ ˜ ν
n
(K
n+1
¸ H). But in this case H ¸ F
1
⊆ K
n+1
¸ (F
1
∪ F
2
) is µ
n
negligible,
because
µ
n
F
1
+µ
n
F
2
≥ ˜ ν
n
H + ˜ ν
n
(K
n+1
¸ H) = ˜ ν
n
K
n+1
= µ
n
K
n+1
.
So H ¸ F
1
and H belong to Σ
n
and µ
n
H = µ
n
F
1
= ˜ ν
n
H. QQQ
(d) Set
Σ = ¦E : E ⊆ X, E ∩ K
n+1
∈ Σ
n
for every n ∈ N¦,
µE =
∞
n=0
µ
n
(E ∩ K
n+1
) for every E ∈ Σ.
Then µ is a Radon measure on X extending ν.
PPP (i) It is easy to check that Σ is a σalgebra of subsets of X including T, just because each Σ
n
is a
σalgebra of subsets of K
n+1
including T
n
; and that µ is a complete measure because every µ
n
is.
(ii) If E ∈ T, then
µE =
∞
n=0
µ
n
(E ∩ K
n+1
) =
∞
n=0
˜ ν
n
(E ∩ K
n+1
) =
∞
n=0
ν
n
E = lim
n→∞
n
i=0
ν
i
E
= lim
n→∞
n
i=0
ν
∗
(E ∩ K
i+1
) −ν
∗
(E ∩ K
i
) = lim
n→∞
ν
∗
(E ∩ K
n
) ≤ νE.
On the other hand,
µX = lim
n→∞
ν
∗
K
n+1
= νX,
so in fact µE = νE for every E ∈ T, that is, µ extends ν. In particular, µ is totally ﬁnite, therefore locally
determined and locally ﬁnite.
(iii) If G ⊆ X is open, then G∩ K
n+1
∈ Σ
n
for every n, so G ∈ Σ; thus µ is a topological measure. If
µE > 0, there is some n ∈ N such that µ
n
(E ∩ K
n+1
) > 0; now there is a compact set K ⊆ E ∩ K
n+1
such
that µ
n
K > 0, so that µK > 0. This shows that µ is tight, so is a Radon measure, as required. QQQ
Remark Observe that in this construction
µK
n+1
=
∞
i=0
µ
i
(K
n+1
∩ K
i+1
) =
∞
i=0
˜ ν
i
(K
n+1
∩ K
i+1
) =
∞
i=0
ν
i
(K
n+1
∩ K
i+1
)
=
∞
i=0
ν
t
i+1
(K
n+1
∩ K
i+1
) −ν
t
i
(K
n+1
∩ K
i+1
)
=
∞
i=0
ν
∗
(K
n+1
∩ K
i+1
) −ν
∗
(K
n+1
∩ K
i
)
=
n
i=0
ν
∗
(K
n+1
∩ K
i+1
) −ν
∗
(K
n+1
∩ K
i
) = ν
∗
K
n+1
for every n ∈ N. What this means is that if instead of the hypothesis
νX = sup
K⊆X is compact
inf
F∈T,F⊇K
νF
we are presented with a speciﬁed nondecreasing sequence ¸L
n
)
n∈N
of compact subsets of X such that
νX = sup
n∈N
ν
∗
L
n
, then we can take K
n+1
= L
n
in the argument above and we shall have µL
n
= ν
∗
L
n
for
every n.
416P Radon measure spaces 81
416P Theorem Let X be a Hausdorﬀ space and µ a locally ﬁnite measure on X which is inner regular
with respect to the closed sets. Then the following are equiveridical:
(i) µ has an extension to a Radon measure on X;
(ii) for every nonnegligible measurable set E ⊆ X there is a compact set K ⊆ E such that
µ
∗
K > 0.
If µ is totally ﬁnite, we can add
(iii) sup¦µ
∗
K : K ⊆ X is compact¦ = µX.
proof Write Σ for the domain of µ.
(a)(i)⇒(ii) If λ is a Radon measure extending µ, and µE > 0, then λE > 0, so there is a compact set
K ⊆ E such that λK > 0; but now, because λ is an extension of µ,
µ
∗
K ≥ λ
∗
K = λK > 0.
(b)(ii)⇒(i) & (iii)(ααα) Let c be the family of measurable envelopes of compact sets. Then µE < ∞
for every E ∈ c. PPP If E ∈ c, there is a compact set K such that E is a measurable envelope of K. Now
µE = µ
∗
K is ﬁnite by 411G. QQQ
Next, c is closed under ﬁnite unions, by 132Ed. The hypothesis (ii) tells us that if µE > 0 then there is
some F ∈ c such that F ⊆ E and µF > 0; for there is a compact set K ⊆ E such that µ
∗
K > 0, K has a
measurable envelope F
0
, and F = E ∩ F
0
is still a measurable envelope of K. So in fact µ is inner regular
with respect to c (412Aa). In particular, µ is semiﬁnite.
If γ < µX there is an F ∈ c such that µF ≥ , and now there is a compact set K such that F is a
measurable envelope of K, so that µ
∗
K = µF ≥ γ. As γ is arbitrary, (iii) is true.
(βββ) Because µ is inner regular with respect to c, D = ¦E
•
: E ∈ c¦ is orderdense in the measure
algebra (A, ¯ µ) of µ (412N), so there is a family ¸d
i
)
i∈I
in D which is a partition of unity in A (313K). For
each i ∈ I, take E
i
∈ c such that E
•
i
= d
i
. Then
i∈I
µ(E ∩ E
i
) =
i∈I
¯ µ(E
•
∩ d
i
) = ¯ µE
•
(321E)
= µE
for every E ∈ Σ.
(γγγ) For each i ∈ I, let µ
i
be the subspace measure on E
i
. Then there is a Radon measure λ
i
on E
i
extending µ
i
. PPP Because µ is inner regular with respect to the closed sets, µ
i
is inner regular with respect
to the relatively closed subsets of E
i
(412O). Also there is a compact subset K ⊆ E
i
such that
µ
i
E
i
= µE
i
= µ
∗
K = µ
∗
i
K,
so µ
i
satisﬁes the conditions of 416O and has an extension to a Radon measure. QQQ
(δδδ) Deﬁne
λE =
i∈I
λ
i
(E ∩ E
i
)
whenever E ⊆ X is such that λ
i
measures E ∩ E
i
for every i ∈ I. Then λ is a Radon measure on X
extending µ. PPP It is easy to check that it is a measure, just because every λ
i
is a measure, and it extends
µ by (β) above. If G ⊆ X is open, then G∩E
i
is relatively open for every i ∈ I, so λ measures G; thus λ is
a topological measure. If λE = 0 and A ⊆ E, then λ
i
(A∩ E
i
) ≤ λ(E ∩ E
i
) = 0 for every i, so λA = 0; thus
λ is complete. For all distinct i, j ∈ I,
λ
i
(E
i
∩ E
j
) = µ
i
(E
i
∩ E
j
) = µ(E
i
∩ E
j
) = ¯ µ(d
i
∩ d
j
) = 0,
so λE
i
= λ
i
E
i
= µ
i
E
i
is ﬁnite. This means that if E ⊆ X is such that λ measures E∩F whenever λF < ∞,
then λ must measure E ∩ E
i
for every i, and λ measures E; thus λ is locally determined. If λE > 0 there
are an i ∈ I such that λ
i
(E ∩E
i
) > 0 and a compact K ⊆ E ∩E
i
such that 0 < λ
i
K = λK; consequently λ
82 Topologies and measures I 416P
is tight. Finally, if x ∈ X, there is an E ∈ Σ such that x ∈ int E and λE = µE < ∞, so λ is locally ﬁnite.
Thus λ is a Radon measure. QQQ
So (i) is true.
(c) Finally, suppose that µ is totally ﬁnite and (iii) is true. Then we can appeal directly to 416O to see
that (i) is true.
416Q Proposition (a) Let X be a compact Hausdorﬀ space and c the algebra of openandclosed subsets
of X. Then any nonnegative ﬁnitely additive functional from c to R has an extension to a Radon measure
on X. If X is zerodimensional then the extension is unique.
(b) Let A be a Boolean algebra, and Z its Stone space. Then there is a onetoone correspondence between
nonnegative additive functionals ν on A and Radon measures µ on Z given by the formula
νa = µ´a for every a ∈ A,
where for a ∈ A I write ´a for the corresponding openandclosed subset of Z.
proof (a) Let ν : c → [0, ∞[ be a nonnegative additive functional. Then ν satisﬁes the conditions of 416O
(because every member of c is closed, while X is compact), so has an extension to a Radon measure µ. If
X is zerodimensional, c is a base for the topology of X closed under ﬁnite unions and intersections, so µ is
unique, by 415H(iv) or 415H(v).
(b) The map a →´a is a Boolean isomorphism between A and the algebra c of openandclosed subsets
of Z, so we have a onetoone correspondence between nonnegative additive functionals ν on A and non
negative additive functionals ν
t
on c deﬁned by the formula ν
t
´a = νa. Now Z is compact, Hausdorﬀ and
zerodimensional, so ν
t
has a unique extension to a Radon measure on Z, by part (a). And of course every
Radon measure µ on Z gives us a nonnegative additive functional µc on c, corresponding to a nonnegative
additive functional on A.
416R Theorem (a) Any subspace of a Radon measure space is a quasiRadon measure space.
(b) A measurable subspace of a Radon measure space is a Radon measure space.
(c) If (X, T, Σ, µ) is a Hausdorﬀ complete locally determined topological measure space, and Y ⊆ X is
such that the subspace measure µ
Y
on Y is a Radon measure, then Y ∈ Σ.
proof (a) Put 416A and 415B together.
(b) Let (X, T, Σ, µ) be a Radon measure space, and (E, T
E
, Σ
E
, µ
E
) a member of Σ with the induced
topology and measure. Because µ is complete and locally determined, so is µ
E
(214Ja). Because T is
Hausdorﬀ, so is T
E
(4A2F(ai)). Because µ is locally ﬁnite, so is µ
E
. Because µ is tight (and a subset of E
is compact for T
E
whenever it is compact for T), µ
E
is tight (412Oa).
(c) ??? If Y / ∈ Σ, then there is a set F ∈ Σ such that µ
∗
(Y ∩ F) < µ
∗
(Y ∩ F) (413F(v)). But now
µ
∗
(Y ∩F) = µ
Y
(Y ∩F), so there is a compact set K ⊆ Y ∩F such that µ
Y
K > µ
∗
(Y ∩F). When regarded
as a subset of X, K is still compact; because T is Hausdorﬀ, K is closed, so belongs to Σ, and
µ
∗
(Y ∩ F) ≥ µK = µ
Y
K > µ
∗
(Y ∩ F),
which is absurd. XXX
416S Corresponding to 415O, we have the following.
Proposition Let (X, T, Σ, µ) be a Radon measure space, and ν an indeﬁniteintegral measure over µ (deﬁ
nition: 234B). If ν is locally ﬁnite, it is a Radon measure.
proof Because µ is complete and locally determined, so is ν (234Fb). Because µ is tight (412Q). So if ν is
also locally ﬁnite, it is a Radon measure.
416T I said in the notes to ¸415 that the most important quasiRadon measure spaces are subspaces of
Radon measure spaces. I do not know of a useful necessary and suﬃcient condition, but the following deals
with completely regular spaces.
416V Radon measure spaces 83
Proposition Let (X, T, Σ, µ) be a locally ﬁnite completely regular Hausdorﬀ quasiRadon measure space.
Then it is isomorphic, as topological measure space, to a subspace of a locally compact Radon measure
space.
proof (a) Write βX for the Stone
ˇ
Cech compactiﬁcation of X (4A2I); I will take it that X is actually a
subspace of βX. Let  be the set of those open subsets U of βX such that µ(U ∩ X) < ∞; then  is
upwardsdirected and covers X, because µ is locally ﬁnite. Set W =
 ⊇ X. Then W is an open subset
of βX, so is locally compact.
(b) Let B be the Borel σalgebra of W. Then V ∩ X is a Borel subset of X for every V ∈ B (4A3Ca),
so we have a measure ν : B → [0, ∞] deﬁned by setting νV = µ(X ∩ V ) for every V ∈ B. Now ν satisﬁes
the conditions of 415Cb. PPP (α) If νV > 0, then, because µ is eﬀectively locally ﬁnite, there is an open set
G ⊆ X such that µ(G∩V ) > 0 and µG < ∞. There is an open set U ⊆ βX such that U ∩X = G, in which
case U ⊆ W, νU < ∞ and ν(U ∩V ) > 0. Thus ν is eﬀectively locally ﬁnite. (β) If  is an upwardsdirected
family of open subsets of W, then ¦U ∩ X : U ∈ ¦ is an upwardsdirected family of open subsets of X, so
ν(
_
) = µ(X ∩
_
) = µ(
_
¦U ∩ X : U ∈ ¦)
= sup
U∈/
µ(X ∩ U) = sup
U∈/
νU. QQQ
So the c.l.d. version ˜ ν of ν is a quasiRadon measure on W (415Cb).
(c) The construction of W ensures that ν and ˜ ν are locally ﬁnite. By 416G, ˜ ν is a Radon measure. So the
subspace measure ˜ ν
X
is a quasiRadon measure on X (416Ra). But ˜ ν
X
G = µG for every open set G ⊆ X.
PPP Note ﬁrst that as ν eﬀectively locally ﬁnite, therefore semiﬁnite, ˜ ν extends ν (213Hc). If K ⊆ W is a
compact set not meeting X, then
˜ νK = νK = µ(K ∩ X) = 0;
accordingly ˜ ν
∗
(W ¸ X) = 0, by 413Ee. Now there is an open set U ⊆ W such that G = X ∩ U, and
˜ ν
X
G = ˜ ν
∗
G ≤ ˜ νU = νU = µ(U ∩ X) = µG
= ˜ ν
∗
(U ∩ X) + ˜ ν
∗
(U ¸ X)
(by 413E(cii), because ˜ ν is semiﬁnite)
≤ ˜ ν
∗
(U ∩ X) + ˜ ν
∗
(W ¸ X) = ˜ ν
∗
G. QQQ
So 415H(iii) tells us that µ = ˜ ν
X
is the subspace measure induced by ν.
416U Theorem If ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
is a family of compact metrizable Radon probability spaces such
that every µ
i
is strictly positive, the product measure on X =
i∈I
X
i
is a completion regular Radon
measure. In particular, the usual measures on ¦0, 1¦
I
and [0, 1]
I
and TI are completion regular Radon
measures, for any set I.
proof By 415E, it is a completion regular quasiRadon probability measure; but X is a compact Hausdorﬀ
space, so it is a Radon measure, by 416G or otherwise.
Remark I suppose it is obvious that by the ‘usual measure on [0, 1]
I
’ I mean the product measure when
each copy of [0, 1] is given Lebesgue measure. Recall also that the ‘usual measure on TI’ is just the copy of
the usual measure on ¦0, 1¦
I
induced by the standard bijection A ↔ χA (254Jb), which is a homeomorphism
(4A2Ud).
416V Stone spaces The results of 415Q415R become simpler and more striking in the present context.
Theorem Let (X, T, Σ, µ) be a Radon measure space, and (Z, S, T, ν) the Stone space of its measure algebra
(A, ¯ µ). For E ∈ Σ let E
∗
be the openandclosed set in Z corresponding to the image E
•
of E in A. Deﬁne
R ⊆ Z X by saying that (z, x) ∈ R iﬀ x ∈ F whenever F ⊆ X is closed and z ∈ F
∗
.
84 Topologies and measures I 416V
(a) R is the graph of a function f : Q → X, where Q = R
−1
[X]. If we set W =
¦K
∗
: K ⊆ X
is compact¦, then W ⊆ Q is a νconegligible open set, and the subspace measure ν
W
on W is a Radon
measure.
(b) Setting g = fW, g is continuous and µ is the image measure ν
W
g
−1
.
(c) If X is compact, W = Q = Z and µ = νg
−1
.
proof (a) By 415Ra, R is the graph of a function. If K ⊆ X is compact and z ∈ K
∗
, then T = ¦F : F ⊆ X
is closed, z ∈ F
∗
¦ is a family of nonempty closed subsets of X, closed under ﬁnite intersections, and
containing the compact set K; so it has nonempty intersection, and there is an x ∈ K such that (z, x) ∈ R,
that is, z ∈ Q and f(z) ∈ K. Thus W ⊆ Q. Of course W is an open set, being the union of a family of
openandclosed sets; but it is also conegligible, because sup¦K
•
: K ⊆ X is compact¦ = 1 in A (412N), so
Z ¸ W must be nowhere dense, therefore negligible. Now the subspace measure ν
W
is quasiRadon because
ν is (411P(div), 415B); but W is a union of compact open sets of ﬁnite measure, so ν
W
is locally ﬁnite and
W is locally compact; by 416G, ν
W
is a Radon measure.
(b) g is continuous. PPP Let G ⊆ X be an open set and z ∈ g
−1
[G]. Let K ⊆ X be a compact set such
that z ∈ K
∗
. As remarked above, g(z) = f(z) belongs to K. K, being a compact Hausdorﬀ space, is regular
(3A3Bb), so there is an open set H containing g(z) such that L = H ∩ K ⊆ G. Note that L is compact,
so L
∗
⊆ W. Now g(z) does not belong to the closed set X ¸ H, so z / ∈ (X ¸ H)
∗
and z ∈ H
∗
; accordingly
z ∈ (H ∩K)
∗
⊆ L
∗
. If w ∈ L
∗
, g(w) ∈ L ⊆ G; so L
∗
⊆ g
−1
[G], and z ∈ int g
−1
[G]. As z is arbitrary, g
−1
[G]
is open; as G is arbitrary, g is continuous. QQQ
By 415Rb, we know that µ = ν
Q
f
−1
, where ν
Q
is the subspace measure on Q. But as ν is complete and
both Q and W are conegligible, we have
ν
Q
f
−1
[A] = νf
−1
[A] = νg
−1
[A] = ν
W
g
−1
[A]
whenever A ⊆ X and any of the four terms is deﬁned, so that µ = ν
Q
f
−1
= ν
W
g
−1
.
(c) If X is compact, then Z = X
∗
⊆ W, so W = Q = Z and νg
−1
= ν
W
g
−1
= µ.
416W Compact measure spaces Recall that a semiﬁnite measure space (X, Σ, µ) is ‘compact’ (as
a measure space) if there is a family / ⊆ Σ such that µ is inner regular with respect to / and
/
t
,= ∅
whenever /
t
⊆ / has the ﬁnite intersection property (342A); while (X, Σ, µ) is ‘perfect’ if whenever f :
X →R is measurable and µE > 0, there is a compact set K ⊆ f[E] such that µf
−1
[K] > 0 (342K). In ¸342
I introduced these concepts in order to study the realization of homomorphisms between measure algebras.
The following result is now very easy.
Proposition (a) Any Radon measure space is a compact measure space, therefore perfect.
(b) Let (X, T, Σ, µ) be a Radon measure space, with measure algebra (A, ¯ µ), and (Y, T, ν) a complete
strictly localizable measure space, with measure algebra (B, ¯ ν). If π : A → Bis an ordercontinuous Boolean
homomorphism, there is a function f : Y → X such that f
−1
[E] ∈ T and f
−1
[E]
•
= πE
•
for every E ∈ Σ.
If π is measurepreserving, f is inversemeasurepreserving.
proof (a) If (X, T, Σ, µ) is a Radon measure space, µ is inner regular with respect to the compact class
consisting of the compact subsets of X, so (X, Σ, µ) is a compact measure space. By 342L, it is perfect.
(b) Use (i)⇒(v) of Theorem 343B. (Of course f is inversemeasurepreserving iﬀ π is measurepreserving.)
416X Basic exercises >>>(a) Let (X, T, Σ, µ) be a Radon measure space, and E ∈ Σ an atom for the
measure. Show that there is a point x ∈ E such that µ¦x¦ = µE.
(b) Let X be a topological space and µ a pointsupported measure on X, as described in 112Bd. (i)
Show that µ is tight, so is a Radon measure iﬀ it is locally ﬁnite. In particular, show that if X has the
discrete topology then counting measure on X is a Radon measure. (ii) Show that every purely atomic
Radon measure is of this type.
(c) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of Radon measure spaces, with direct sum (X, Σ, µ) (214K). Give
X its disjoint union topology. Show that µ is a Radon measure.
416Xr Radon measure spaces 85
(d) Let (X, T) be a Hausdorﬀ space, and µ
1
, µ
2
two Radon measures on X, with domains Σ
1
, Σ
2
. (i)
Show that µ
1
+µ
2
, deﬁned on Σ
1
∩ Σ
2
(112Xe), is a Radon measure. (ii) Show that αµ
1
, deﬁned on Σ
1
, is
a Radon measure for any α > 0.
(e) Let (X, T, Σ, µ) be a Radon measure space. (i) Show that µ has a decomposition ¸X
i
)
i∈I
in which
every X
i
except at most one is a selfsupporting compact set, and the exceptional one, if any, is negligible. (ii)
Show that µ has a decomposition ¸X
i
)
i∈I
in which every X
i
is expressible as the intersection of a closed set
with an open set. (Hint: enumerate the open sets of ﬁnite measure as ¸G
ξ
)
ξ<κ
, and set X
ξ
= G
ξ
¸
η<ξ
G
η
.)
(f ) Let X be a Hausdorﬀ space and µ, ν two Radon measures on X such that νG = µG whenever G ⊆ X
is open and min(µG, νG) < ∞. Show that µ = ν.
(g) Explain how to prove 416H from 416F, without appealing to ¸415.
(h) Give a direct proof of 416I not relying on 415O.
(i) Let (X, T) be a completely regular topological space and µ a locally ﬁnite topological measure on X
which is inner regular with respect to the closed sets. Show that
µK = inf¦µG : G ⊇ K is a cozero set¦
= inf¦µF : F ⊇ K is a zero set¦ = inf¦
_
fdµ : χK ≤ f ∈ C(X)¦
for every compact set K ⊆ X.
(j) Let (X, T, Σ, µ) be a locally compact Radon measure space, and C
k
the space of continuous real
valued functions on X with compact supports. Show that ¦f
•
: f ∈ C
k
¦ is dense in L
0
(µ) for the topology
of convergence in measure.
(k) Let X be a completely regular Hausdorﬀ space and ν a locally ﬁnite Baire measure on X. (i) Show
that ν
∗
K = inf¦νG : G ⊆ X is a cozero set, K ⊆ G¦ for every compact set K ⊆ X. (ii) Show that there is
a Radon measure µ on X such that µK = ν
∗
K for every compact set K ⊆ X. (Hint: in the language of
the proof of 416K, φ
1
= ν
∗
/.)
(l) Let X be a Hausdorﬀ space and ν a nonnegative ﬁnitely additive functional deﬁned on some algebra
of subsets of X. Show that there is a Radon measure µ on X such that µX ≤ νX and µK ≥ νE whenever
E ∈ T, K ⊆ X is compact and E ⊆ K. (Hint: start by extending ν to TX.)
(m) Let (X, T) be a Hausdorﬀ space, Σ ⊇ T a σalgebra of subsets of X, and ν : Σ → [0, ∞[ a ﬁnitely
additive functional such that νE = sup¦νK : K ⊆ E is compact¦ for every E ∈ Σ. Show that ν is countably
additive and that its completion is a Radon measure on X.
(n) Explain how to prove 416Rb from 416C and 415B.
(o) Let X be a Hausdorﬀ space, µ a complete locally ﬁnite measure on X, and Y a conegligible subset
of X. Show that µ is a Radon measure iﬀ the subspace measure on Y is a Radon measure.
(p) Let X be a Hausdorﬀ space, Y a subset of X, and ν a Radon measure on Y . Deﬁne a measure µ on
X by setting µE = ν(E ∩ Y ) whenever ν measures E ∩ Y . Show that if either Y is closed or ν is totally
ﬁnite, µ is a Radon measure on X. (Cf. 418I.)
(q) Let (X, T, Σ, µ) be a Radon measure space and c ⊆ Σ a nonempty upwardsdirected family. Set
νF = sup
E∈L
µ(E ∩ F) whenever F ⊆ X is such that µ measures E ∩ F for every E ∈ c. Show that ν is a
Radon measure on X.
(r) Let ¸X
n
)
n∈N
be a sequence of Hausdorﬀ spaces with product X; write B(X
n
) for the Borel σalgebra
of X
n
. Let T be the σalgebra
´
n∈N
B(X
n
) (deﬁnition: 254E). Let ν : T → [0, ∞[ be a ﬁnitely additive
functional such that E → νπ
−1
n
[E] : B(X
n
) → [0, ∞[ is countably additive and tight for each n ∈ N, writing
π
n
(x) = x(n) for x ∈ X, n ∈ N. Show that there is a unique Radon measure on X extending ν.
86 Topologies and measures I 416Xs
(s) Set S
∗
2
=
n∈N
¦0, 1¦
n
, and let φ : S
∗
2
→ [0, ∞[ be a functional such that φ(σ) = φ(σ
0) + φ(σ
1)
for every σ ∈ S
∗
2
, writing σ
0 and σ
1 for the two members of ¦0, 1¦
n+1
extending any σ ∈ ¦0, 1¦
n
. Show
that there is a unique Radon measure µ on ¦0, 1¦
N
such that µ¦x : x¦0, . . . , n −1¦ = σ¦ = φ(σ) whenever
n ∈ N, σ ∈ ¦0, 1¦
N
. (Hint: use 416Xr or 416Q.)
(t) Let (X, T, Σ, µ) be a Radon measure space. Show that a measure ν on X is an indeﬁniteintegral
measure over µ iﬀ (α) ν is a complete, locally determined topological measure (β) ν is tight (γ) νK = 0
whenever K ⊆ X is compact and µK = 0.
(u) Let (X, T, Σ, µ) be a compact Hausdorﬀ quasiRadon measure space. Let W ⊆ X be the union of
the open subsets of X of ﬁnite measure. Show that the subspace measure on W is a Radon measure.
(v) Let (X, T, Σ, µ) be a completely regular Radon measure space. Show that it is isomorphic, as
topological measure space, to a measurable subspace of a locally compact Radon measure space.
(w) Let (X, T, Σ, µ) be a compact Radon measure space and (Z, S, T, ν) the Stone space of its measure
algebra. For E ∈ Σ let E
∗
be the corresponding openandclosed subset of Z, as in 416V. Show that the
function described in 416V is the unique continuous function h : Z → X such that ν(E
∗
´h
−1
[E]) = 0 for
every E ∈ Σ.
(x) Show that the rightfacing Sorgenfrey line (415Xc), with Lebesgue measure, is a quasiRadon measure
space which, regarded as a measure space, is compact, but, regarded as a topological measure space, is not
a Radon measure space.
416Y Further exercises (a) Let X be a Hausdorﬀ space and ν a countably additive realvalued
functional deﬁned on a σalgebra Σ of subsets of X. Show that the following are equiveridical: (i) [ν[ : Σ →
[0, ∞[, deﬁned as in 362B, is a Radon measure on X; (ii) ν is expressible as µ
1
−µ
2
, where µ
1
, µ
2
are Radon
measures on X and Σ = domµ
1
∩ domµ
2
.
(b) Let X be a topological space and µ
0
a semiﬁnite measure on X which is inner regular with respect
to the family /
ccc
of closed countably compact sets. Show that µ
0
has an extension to a complete locally
determined topological measure µ on X, still inner regular with respect to /
ccc
; and that the extension may
be done in such a way that whenever µE < ∞ there is an E
0
∈ Σ
0
such that µ(E´E
0
) = 0. (Hint: use the
argument of 416M, but with / = ¦K : K ∈ /
ccc
, µ
∗
0
K < ∞¦.)
(c) Let X be a topological space and µ
0
a semiﬁnite measure on X which is inner regular with respect to
the family /
sc
of sequentially compact sets. Show that µ
0
has an extension to a complete locally determined
topological measure µ on X, still inner regular with respect to /
sc
; and that the extension may be done in
such a way that whenever µE < ∞ there is an E
0
∈ Σ
0
such that µ(E´E
0
) = 0.
(d) Let X ⊆ βN be the union of all those open sets G ⊆ βN such that
n∈G∩N
1
n+1
is ﬁnite. For E ⊆ X
set µE =
n∈E∩N
1
n+1
. Show that µ is a σﬁnite Radon measure on the locally compact Hausdorﬀ space
X. Show that µ is not outer regular with respect to the open sets.
(e) Set S
∗
=
n∈N
N
n
, and let φ : S
∗
→ [0, ∞[ be a functional such that φ(σ) =
∞
i=0
φ(σ
i) for every
σ ∈ S
∗
, writing σ
i for the members of N
n+1
extending any σ ∈ N
n
. Show that there is a unique Radon
measure µ on N
N
such that µI
σ
= φ(σ) for every σ ∈ S
∗
, where I
σ
= ¦x : x¦0, . . . , n − 1¦ = σ¦ for any
n ∈ N, σ ∈ N
N
. (Hint: set θA = inf¦
σ∈R
φ(σ) : R ⊆ S
∗
, A ⊆
σ∈R
I
σ
¦ for every A ⊆ N
N
, and use
Carath´eodory’s method.)
(f ) Let (X, T, Σ, µ) be a Radon measure space. Show that a measure ν on X is an indeﬁniteintegral
measure over µ iﬀ (i) there is a topology S on X, including T, such that ν is a Radon measure with respect
to S (ii) νK = 0 whenever K is a Tcompact set and µK = 0.
(g) Let ¸x
n
)
n∈N
enumerate a dense subset of X = ¦0, 1¦
c
(4A2B(eii)). Let λ be the usual measure on
X, and set µE =
1
2
λE +
∞
n=0
2
−n−2
χE(x
n
) for E ∈ domλ. (i) Show that µ is a strictly positive Radon
416 Notes Radon measure spaces 87
probability measure on X of Maharam type c. (ii) Let I ∈ [c]
≤ω
be such that x
m
I ,= x
n
I whenever
m ,= n. Set Z = ¦0, 1¦
I
and let π : X → Z be the canonical map. Show that if f ∈ C(X) is such that
_
f gπ dµ = 0 for every g ∈ C(Z), then f = 0. (Hint: otherwise, take n ∈ N such that [f(x
n
)[ ≥
1
2
f
∞
,
and let g ≥ 0 be such that g(πx
n
) = 1 and
_
g d(µπ
−1
) < 3 2
−n−3
; show that
_
f gπ dµ > 0.) (iii) Show
that there is no orthonormal basis for L
2
(µ) in ¦f
•
: f ∈ C(X)¦. (See Hart & Kunen 99.)
416 Notes and comments The original measures studied by Radon (Radon 13) were, in eﬀect, what
I call diﬀerences of Radon measures on R
r
, as introduced in ¸256. Successive generalizations moved ﬁrst
to Radon measures on general compact Hausdorﬀ spaces, then to locally compact Hausdorﬀ spaces, and
ﬁnally to arbitrary Hausdorﬀ spaces, as presented in this section. I ought perhaps to remark that, following
Bourbaki 65, many authors use the term ‘Radon measure’ to describe a linear functional on a space
of continuous functions; I will discuss the relationship between such functionals and the measures of this
chapter in ¸436. For the moment, observe that by 415I a Radon measure on a completely regular space
can be determined from the integrals it assigns to continuous functions. It is also common for the phrase
‘Radon measure’ to be used for what I would call a tight Borel measure; you have to check each author to
see whether local ﬁniteness is also assumed. In my usage, a Radon measure is necessarily the c.l.d. version
of a Borel measure. The Borel measures which correspond to Radon measures are described in 416F.
In ¸256, I discussed Radon measures on R
r
as a preparation for a discussion of convolutions of measures.
It should now be becoming clear why I felt that it was impossible, in that context, to give you a proper idea
of what a Radon measure, in the modern form, ‘really’ is. In Euclidean space, too many concepts coincide.
As a trivial example, the simplest deﬁnition of ‘local ﬁniteness’ (256Ab, 411Fa) is not the right formulation
in other spaces. Next, because every closed set is a countable union of compact sets, there is no distinction
between ‘inner regular with respect to closed sets’ and ‘inner regular with respect to compact sets’, so one
cannot get any intuition for which is important in which arguments. (When we come to subspace measures
on nonmeasurable subsets, of course, this changes; quasiRadon measures on subsets of Euclidean space
are important and interesting.) Third, the fact that the c.l.d. product of two Radon measures on Euclidean
space is already a Radon measure (256K) leaves us with no idea of what to do with a general product of
Radon measures. (There are real diﬃculties at this point, which I will attack in the next section. For the
moment I oﬀer just 416U.) And ﬁnally, we simply cannot represent a product of uncountably many Radon
probability measures on Euclidean spaces as a measure on Euclidean space.
As you would expect, a very large proportion of the results of this chapter, and many theorems from
earlier volumes, were originally proved for compact Radon measure spaces. The theory of general totally
ﬁnite Radon measures is, in eﬀect, the theory of measurable subspaces of compact Radon measure spaces,
while the theory of quasiRadon measures is pretty much the theory of nonmeasurable subspaces of Radon
measure spaces. Thus the theorem that every quasiRadon measure space is strictly localizable is almost a
consequence of the facts that every Radon measure space is strictly localizable and any subspace of a strictly
localizable space is strictly localizable.
The cluster of results between 416J and 416P form only a sample, I hope a reasonably representative
sample, of the many theorems on construction of Radon measures from functionals on algebras or lattices of
sets. (See also 416Ye.) The essential simpliﬁcation, compared with the theorems in ¸413 and ¸415, is that
we do not need to mention any σ or τadditivity condition of the type 413I(β) or 415K(β), because we are
dealing with a ‘compact class’, the family of compact subsets of a Hausdorﬀ space. We can use this even at
some distance, as in 416O (where the hypotheses do not require any nonempty compact set to belong to
the domain of the original functional). The particular feature of 416O which makes it diﬃcult to prove from
such results as 413J and 413O above is that we have to retain control of the outer measures of a sequence
¸K
n
)
n∈N
of nonmeasurable sets. In general this is hard to do, and is possible here principally because the
sequence is nondecreasing, so that we can make sense of the functionals ν
n
E = ν
∗
(E∩K
n+1
) −ν
∗
(E∩K
n
).
88 Topologies and Measures §417 intro.
417 τadditive product measures
The ‘ordinary’ product measures introduced in Chapter 25 have served us well for a volume and a half.
But we come now to a fundamental obstacle. If we start with two Radon measure spaces, their product
measure, as deﬁned in ¸251, need not be a Radon measure (419E). Furthermore, the counterexample is one
of the basic compact measure spaces of the theory; and while it is dramatically nonmetrizable, there is no
other reason to set it aside. Consequently, if we wish (as we surely do) to create Radon measure spaces as
products of Radon measure spaces, we need a new construction. This is the object of the present section. It
turns out that the construction can be adapted to work well beyond the special context of Radon measure
spaces; the methods here apply to general eﬀectively locally ﬁnite τadditive topological measures (for the
product of ﬁnitely many factors) and to τadditive topological probability measures (for the product of
inﬁnitely many factors). Elsewhere in this volume I will show how these constructions are related to some
others which have been described for special spaces.
The fundamental theorems are 417C and 417E, listing the essential properties of what I call ‘τadditive
product measures’, which are extensions of the c.l.d. product measures and product probability measures of
Chapter 25. They depend on a straightforward lemma on the extension of a measure to make every element
of a given class of sets negligible (417A). It is relatively easy to prove that the extensions are more or less
canonical (417D, 417F). We still have Fubini’s theorem for the new product measures (417H), and the basic
operations from ¸254 still apply (417J, 417K, 417M).
It is easy to check that if we start with quasiRadon measures, then the τadditive product measure is
again quasiRadon (417N, 417O). The τadditive product of two Radon measures is Radon (417P), and the
τadditive product of Radon probability measures with compact supports is Radon (417Q).
In the last part of the section I look at continuous realvalued functions and Baire σalgebras; it turns
out that for these the ordinary product measures are adequate (417U, 417V).
417A Lemma Let (X, Σ, µ) be a semiﬁnite measure space, and / ⊆ TX a family of sets such that
µ
∗
(
n∈N
A
n
) = 0 for every sequence ¸A
n
)
n∈N
in /. Then there is a measure µ
t
on X, extending µ, such
that
(i) µ
t
A is deﬁned and zero for every A ∈ /,
(ii) µ
t
is complete if µ is,
(iii) for every F in the domain Σ
t
of µ
t
there is an E ∈ Σ such that µ
t
(F´E) = 0.
In particular, µ and µ
t
have isomorphic measure algebras, so that µ
t
is localizable if µ is.
proof Let /
∗
be the collection of subsets of X which can be covered by a countable subfamily of /. Then
/
∗
is a σideal of subsets of X and µ
∗
A = 0 for every A ∈ /
∗
. Set
Σ
t
= ¦E´A : E ∈ Σ, A ∈ /
∗
¦.
Then Σ
t
is a σalgebra of subsets of X. PPP (i) ∅ = ∅´∅ ∈ Σ
t
. (ii) If E ∈ Σ, A ∈ /
∗
then X ¸ (E´A) =
(X ¸ E)´A ∈ Σ
t
. (iii) If ¸E
n
)
n∈N
, ¸A
n
)
n∈N
are sequences in Σ, /
∗
respectively, then
E =
n∈N
E
n
∈ Σ, A = E´
n∈N
(E
n
´A
n
) ⊆
n∈N
A
n
∈ /
∗
,
so
n∈N
(E
n
´A
n
) = E´A ∈ Σ
t
. QQQ
If E, E
t
∈ Σ, A, A
t
∈ /
∗
and E´A = E
t
´A
t
, then E´E
t
= A´A
t
∈ /
∗
and µ
∗
(E´E
t
) = 0; because µ
is semiﬁnite, µ(E´E
t
) = 0 and µE = µE
t
. Accordingly we can deﬁne µ
t
: Σ
t
→ [0, ∞] by setting
µ
t
(E´A) = µE whenever E ∈ Σ, A ∈ /
∗
.
Evidently µ
t
extends µ and µ
t
A = 0 for every A ∈ /. Also µ
t
is a measure. PPP (i) µ
t
∅ = µ∅ = 0. (ii) If
¸F
n
)
n∈N
is a disjoint sequence in Σ
t
, with union F, express each F
n
as E
n
´A
n
where E
n
∈ Σ, A
n
∈ /
∗
; set
E =
n∈N
E
n
, so that F´E ∈ /
∗
and µ
t
F = µE. If m ,= n, then E
m
∩E
n
⊆ A
m
∪A
n
, so µ(E
m
∩E
n
) = 0;
accordingly
µ
t
F = µE =
∞
n=0
µE
n
=
∞
n=0
µ
t
F
n
. QQQ
A subset of X is µ
t
negligible iﬀ it can be included in a set of the form E´A where µE = 0 and A ∈ /
∗
,
so µ
t
is complete if µ is. The embedding Σ ⊆ Σ
t
induces a measurepreserving homomorphism from the
measure algebra of µ to the measure algebra of µ
t
which is an isomorphism just because every member of
Σ
t
is the symmetric diﬀerence of a member of Σ and a µ
t
negligible set.
417C τadditive product measures 89
417B Lemma Let X and Y be topological spaces, and ν a τadditive topological measure on Y .
(a) If W ⊆ X Y is open, then x → νW[¦x¦] : X → [0, ∞] is lower semicontinuous.
(b) If ν is eﬀectively locally ﬁnite and σﬁnite and W ⊆ XY is a Borel set, then x → νW[¦x¦] is Borel
measurable.
(c) If f : X Y → [0, ∞] is a lower semicontinuous function, then x →
_
f(x, y)ν(dy) : X → [0, ∞] is
lower semicontinuous.
(d) If ν is totally ﬁnite and f : X Y →R is a bounded continuous function, then x →
_
f(x, y)ν(dy) is
continuous.
(e) If ν is totally ﬁnite and W ⊆ X Y is a Baire set, then x → νW[¦x¦] is Baire measurable.
proof (a) If x ∈ X and νW[¦x¦] > α, then
H = ¦H : H ⊆ Y is open, there is an open set G containing x such that GH ⊆ W¦
is an upwardsdirected family of open sets with union W[¦x¦], so there is an H ∈ H such that νH ≥ α.
Now there is an open set G containing x such that GH ⊆ W, so that νW[¦x
t
¦] ≥ α for every x
t
∈ G.
(b)(i) Suppose to begin with that ν is totally ﬁnite. In this case, the set
¦W : W ⊆ X Y, x → νW[¦x¦] is a Borel measurable function
deﬁned everywhere on X¦
is a Dynkin class containing every open set, so contains every Borel set, by the Monotone Class Theorem
(136B).
(ii) For the general case, let ¸Y
n
)
n∈N
be a disjoint sequence of sets of ﬁnite measure covering Y , and
for n ∈ N let ν
n
be the subspace measure on Y
n
. Then ν
n
is eﬀectively locally ﬁnite and τadditive (414K).
If W ⊆ X Y is a Borel set, then W
n
= W ∩ (X Y
n
) is a relatively Borel set for each n, so that
x → ν
n
W
n
[¦x¦] is Borel measurable, by (i). Since νW[¦x¦] =
∞
n=0
ν
n
W
n
[¦x¦] for every x, x → νW[¦x¦]
is Borel measurable.
(c) For i, n ∈ N set W
ni
= ¦(x, y) : f(x, y) > 2
−n
i¦, so that W
ni
⊆ X Y is open. Set f
n
=
2
−n
4
n
i=1
χW
ni
; then ¸f
n
)
n∈N
is a nondecreasing sequence with supremum f. For n ∈ N and x ∈ X,
_
f
n
(x, y)ν(dy) = 2
−n
4
n
i=1
νW
ni
[¦x¦], so x →
_
f
n
(x, y)ν(dy) is lower semicontinuous, by (a) and 4A2B(d
iii). By 414Ba,
_
f(x, y)ν(dy) = sup
n∈N
_
f
n
(x, y)ν(dy) for every x, so x →
_
f(x, y)ν(dy) is lower semi
continuous (4A2B(dv)).
(d) Applying (c) to f+f
∞
χ(XY ), we see that x →
_
f(x, y)ν(dy) is lower semicontinuous. Similarly,
x → −
_
f(x, y)ν(dy) is lower semicontinuous, so x →
_
f(x, y)ν(dy) is continuous (4A2B(dvi)).
(e) Suppose ﬁrst that W is a cozero set; let f : X Y → [0, 1] be a continuous function such that
W = ¦(x, y) : f(x, y) > 0¦. For n ∈ N set f
n
= nf ∧ χ(X Y ). Then ¸f
n
)
n∈N
is a nondecreasing sequence
of continuous functions with supremum χW. By (d), all the functions x →
_
f
n
(x, y)ν(dy) are continuous,
so their limit x → νW[¦x¦] is Baire measurable.
Now
¦W : W ⊆ X Y, x → νW[¦x¦] is a Baire measurable function
deﬁned everywhere on X¦
is a Dynkin class containing every cozero set, so contains every Baire set, by the Monotone Class Theorem
again.
417C Theorem Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological mea
sure spaces, with c.l.d. product (X Y, Λ, λ). Then λ has an extension to a τadditive topological measure
˜
λ on X Y . Moreover, we can arrange that:
(i)
˜
λ is complete, locally determined and eﬀectively locally ﬁnite, therefore strictly localizable;
(ii) if Q belongs to the domain
˜
Λ of
˜
λ, there is a Q
1
∈ Λ such that
˜
λ(Q´Q
1
) = 0; that is to say, the
embedding Λ ⊆
˜
Λ induces an isomorphism between the measure algebras of λ and
˜
λ;
90 Topologies and Measures 417C
(iii) if Q ∈
˜
Λ, then
˜
λQ = sup¦
˜
λ(Q∩ (GH)) : G ∈ T, µG < ∞, H ∈ S, νH < ∞¦;
(iv) if W ⊆ X Y is open, then there is an open set W
0
∈ Λ such that W
0
⊆ W and λW
0
=
˜
λW, and
˜
λW = λ
∗
W =
_
νW[¦x¦]µ(dx) =
_
µW
−1
[¦y¦]ν(dy);
in particular,
˜
λ(GH) = µG νH for every G ∈ T, H ∈ S;
(v) the support of λ is the product of the supports of µ and ν;
(vi) if µ and ν are both inner regular with respect to the Borel sets, so is
˜
λ;
(vii) if µ and ν are both inner regular with respect to the closed sets, so is
˜
λ;
(viii) if µ and ν are both tight (that is, inner regular with respect to the closed compact sets), so is
˜
λ.
proof Write
Σ
f
= ¦E : E ∈ Σ, µE < ∞¦, T
f
= ¦F : F ∈ T, νF < ∞¦,
T
f
= T ∩ Σ
f
, S
f
= S∩ T
f
.
(a) Let  be ¦GH : G ∈ T
f
, H ∈ S
f
¦. Because T ⊆ Σ and S ⊆ T,  ⊆ Λ.  need not be a base for the
topology of XY , unless µ and ν are locally ﬁnite, but if an open subset of XY is included in a member
of  it is the union of the members of  it includes. Moreover, if Q ∈ Λ, then λQ = sup
U∈/
λ(Q∩ U). PPP
By 412R, λ is inner regular with respect to
U∈/
TU. QQQ
Write 
s
for the set of ﬁnite unions of members of , and V for the set of nonempty upwardsdirected
families 1 ⊆ 
s
such that sup
V ∈\
λV < ∞. For each 1 ∈ V, ﬁx on a countable 1
t
⊆ 1 such that
sup
V ∈\
λV = sup
V ∈\
λV ; because 1 is upwardsdirected, we may suppose that 1
t
= ¦V
n
: n ∈ N¦ for some
nondecreasing sequence ¸V
n
)
n∈N
in 1. Set A(1) =
1 ¸
1
t
.
(b)(i) For V ∈ 
s
, set f
V
(x) = νV [¦x¦] for every x ∈ X. This is always deﬁned because V is open;
moreover, f
V
is lower semicontinuous, by 417Ba. Because V is a ﬁnite union of products of sets of ﬁnite
measure,
_
f
V
dµ = λV .
(ii) The key to the proof is the following fact: for any 1 ∈ V, almost every vertical section of A(1) is
negligible. PPP ¸f
V
)
V ∈\
is a nonempty upwardsdirected set of lower semicontinuous functions. Set
g(x) = ν(
V ∈\
V [¦x¦]), h(x) = ν(
V ∈\
V [¦x¦])
for every x ∈ X. Because 1 is upwardsdirected and ν is τadditive,
g(x) = sup
V ∈\
νV [¦x¦] = sup
V ∈\
f
V
(x)
in [0, ∞] for each x, so, by 414Ba,
_
g dµ = sup
V ∈\
_
f
V
dµ = sup
V ∈\
λV = sup
V ∈\
λV =
_
hdµ.
Since h ≤ g and sup
V ∈\
λV is ﬁnite, g(x) = h(x) < ∞ for µalmost every x. But for any such x, we must
have
ν(
1)[¦x¦] = ν(
1
t
)[¦x¦] < ∞,
so that
A(1)[¦x¦] = (
1)[¦x¦] ¸ (
1
t
)[¦x¦]
is negligible. QQQ
(c) ??? Suppose, if possible, that there is a sequence ¸1
n
)
n∈N
in V such that λ
∗
(
n∈N
A(1
n
)) > 0. Take
W ∈ Λ such that W ⊆
n∈N
A(1
n
) and λW > 0. Because almost every vertical section of every A(1
n
) is
negligible, almost every vertical section of W is negligible. But this contradicts Fubini’s theorem (252F). XXX
(d) We may therefore apply the construction of 417A to form a measure λ
t
on the σalgebra Λ
t
= ¦W´A :
W ∈ Λ, A ∈ /
∗
¦, where /
∗
is the σideal generated by ¦A(1) : 1 ∈ V¦. Let
˜
λ be the c.l.d. version of λ
t
(213E), and
˜
Λ its domain.
417C τadditive product measures 91
(i) If W ∈ Λ, then W ∈ Λ
t
⊆
˜
Λ. Also, because λ is semiﬁnite,
λ
t
W = λW = sup¦λW
t
: W
t
⊆ W, W ∈ Λ, λW
t
< ∞¦
≤ sup¦λ
t
W
t
: W
t
⊆ W, W ∈ Λ
t
, λ
t
W
t
< ∞¦ =
˜
λW ≤ λ
t
W.
Thus λW =
˜
λW; as W is arbitrary,
˜
λ extends λ.
(ii) It will be useful if we go directly to one of the targets: if Q ∈
˜
Λ and γ <
˜
λQ, there is a U ∈  such
that
˜
λ(Q ∩ U) ≥ γ. PPP There is a Q
1
∈ Λ
t
such that Q
1
⊆ Q and γ < λ
t
Q
1
< ∞. There is a Q
2
∈ Λ such
that λ
t
(Q
1
´Q
2
) = 0, so that
λQ
2
= λ
t
Q
2
= λ
t
Q
1
> γ.
There is a U ∈  such that λ(Q
2
∩ U) ≥ γ, by (a). Now
˜
λ(Q∩ U) ≥ λ
t
(Q
1
∩ U) = λ
t
(Q
2
∩ U) = λ(Q
2
∩ U) ≥ γ. QQQ
(iii)
˜
λ is a topological measure. PPP Let W ⊆ X Y be an open set. Suppose that Q ∈
˜
Λ and
˜
λQ > 0.
By (ii), there is a U ∈  such that
˜
λ(Q ∩ U) > 0. Let 1 be ¦V : V ∈ 
s
, V ⊆ W ∩ U¦. Then 1 ∈ V, so
˜
λA(1) = λ
t
A(1) = 0; since
1
t
∈ Λ,
W ∩ U =
1 ∈ Λ
t
⊆
˜
Λ.
But this means that
˜
λ
∗
(Q∩ W) +
˜
λ
∗
(Q¸ W) ≥
˜
λ
∗
(Q∩ U ∩ W) +
˜
λ
∗
(Q∩ U ¸ W) =
˜
λ(Q∩ U) > 0.
Because
˜
λ is complete and locally determined, and Q is arbitrary, this is enough to ensure that W ∈
˜
Λ
(413F(vii)). QQQ
(iv)
˜
λ is τadditive. PPP??? Suppose, if possible, otherwise; that there is a nonempty upwardsdirected
family J of open sets in X Y such that
˜
λW
∗
> γ = sup
W∈V
˜
λW, where W
∗
=
J. In this case, we
can ﬁnd a Q
0
∈ Λ
t
such that Q
0
⊆ W
∗
and λ
t
Q
0
> γ, a Q
1
∈ Λ such that λ
t
(Q
0
´Q
1
) = 0, and a U ∈ 
such that λ(Q
1
∩ U) > γ (using (a) again). Let 1 ∈ V be the set of those V ∈ 
s
such that V ⊆ W ∩ U for
some W ∈ J. Then
1 = W
∗
∩ U, so
γ < λ(Q
1
∩ U) = λ
t
(Q
1
∩ U) = λ
t
(Q
0
∩ U)
≤
˜
λ(W
∗
∩ U) =
˜
λ(
_
1) =
˜
λ(
_
1
t
)
(because
˜
λA(1) = 0)
= sup
V ∈\
˜
λV
(because 1
t
is countable and upwardsdirected)
≤ sup
W∈V
˜
λW ≤ γ,
which is absurd. XXXQQQ
(e) Now for the supplementary properties (i)(viii), in order.
(i)
˜
λ was constructed to be complete and locally determined. If
˜
λQ > 0, then by (dii) there is a U ∈ 
such that
˜
λ(Q∩U) > 0; since U is open and
˜
λU = λU is ﬁnite, this shows that
˜
λ is eﬀectively locally ﬁnite.
By 414J, it is strictly localizable.
(ii) The point is that λ is also (strictly) localizable. PPP Let ˜ µ, ˜ ν be the c.l.d. versions of µ and ν.
These are τadditive topological measures (because µ and ν are), complete and locally determined (by
construction), and are still eﬀectively locally ﬁnite (cf. 412Ha), so are strictly localizable (414J again). Now
λ is the c.l.d. product of ˜ µ and ˜ ν (251S), therefore strictly localizable (251N). QQQ
92 Topologies and Measures 417C
As remarked in 417A, it follows that λ
t
is localizable, so that every member Q of
˜
Λ diﬀers by a
˜
λnegligible
set from a member Q
0
of Λ
t
(213Hb). Now there is a Q
1
∈ Λ such that λ
t
(Q
1
´Q
0
) = 0, in which case
˜
λ(Q
1
´Q) = 0.
(iii) This is just (dii) above.
(iv) Let W ⊆ X Y be an open set. Set 1 = ¦V : V ∈ 
s
, V ⊆ W¦. Then 1 is upwardsdirected and
has union W ∩ (X
∗
Y
∗
), where X
∗
and Y
∗
are the unions of the open sets of ﬁnite measure in X and Y
respectively. Because µ and ν are eﬀectively locally ﬁnite, X
∗
and Y
∗
are both conegligible. Now, as before,
˜
λW =
˜
λ(W ∩ (X
∗
Y
∗
)) = sup
V ∈\
˜
λV = sup
V ∈\
λV ≤ λ
∗
W ≤
˜
λW.
If we take a countable set 1
0
⊆ 1 such that sup
V ∈\
0
λV = sup
V ∈\
λV , and set W
0
=
1
0
, then W
0
is an
open set, belonging to λ and included in W, and λW
0
=
˜
λW.
Next, deﬁning the functions f
V
as in part (a) of this proof, we have
_
f
V
dµ = λV for every V ∈ 1; and
setting g(x) = νW[¦x¦], we have g(x) = sup
V ∈\
f
V
(x) for every x ∈ X
∗
. So 414Ba tells us that
_
νW[¦x¦]µ(dx) =
_
g dµ = sup
V ∈\
_
f
V
dµ = sup
V ∈\
λV =
˜
λW.
_
µW
−1
[¦y¦]ν(dy) =
˜
λW.
(The point here is that while the arguments of part (b) of this proof give diﬀerent roles to µ and ν, the
actual construction performed in part (d) is symmetric between them.)
Now, given G ∈ T and H ∈ S, set W = GH; then
˜
λ(GH) =
_
µW
−1
[¦y¦]ν(dy) =
_
H
µGν(dy) = µG νH.
(v) Let E and F be the supports of µ, ν respectively. (By 411Nd these are deﬁned.) Then E F is a
closed subset of X Y . Because X ¸ E and Y ¸ F are negligible, E F is conegligible. If W ⊆ X Y is an
open set and (x, y) ∈ W ∩(E F), there are open sets G ⊆ X, H ⊆ Y such that (x, y) ∈ GH ⊆ W. Now
λ(W ∩ (E F)) ≥ λ((GH) ∩ (E F)) = µ(G∩ E) ν(H ∩ F) > 0.
This shows that E F is selfsupporting, so is the support of λ.
(vi), (vii), (viii) The same method works for all of these. Take / to be either the family of Borel
subsets of XY (for (vi)) or the family of closed subsets of XY (for (vii)) or the family of closed compact
subsets of X Y (for (viii)); the essential features of /, valid for all three cases, are that
(α) K ¸ W ∈ / whenever K ∈ / and W ⊆ X Y is open,
(β) K ∪ K
t
∈ / for all K, K
t
∈ /,
(γ)
n∈N
K
n
∈ / for every sequence ¸K
n
)
n∈N
in /.
Now the hypotheses of each part are just what we need in order to be sure that λ is inner regular with
respect to / (412Sd, 412Sa, 412Sb), and I am trying to show that
˜
λ also is, without assuming that
˜
λ is
inner regular with respect to Λ, because this need not be true.
Take Q ∈
˜
Λ and γ <
˜
λQ; then there is a Q
0
∈ Λ
t
such that Q
0
⊆ Q and γ < λ
t
Q
0
< ∞. We have a
Q
1
∈ Λ and a sequence ¸1
n
)
n∈N
in V such that Q
0
´Q
1
⊆
n∈N
A(1
n
). Of course λQ
1
= λ
t
Q
0
> γ; set
=
1
4
(λQ
1
−γ) > 0. For each n ∈ N, take K
n
, L
n
∈ / ∩ Λ such that
K
n
⊆ Q
1
∩
1
t
n
, λ((Q
1
∩
1
t
n
) ¸ K
n
) ≤ 2
−n
,
L
n
⊆ Q
1
, λ(Q
1
¸ L
n
) ≤ 2
−n
.
Set K
t
n
= K
n
∪ (L
n
¸
1
n
), so that K
t
n
∈ / and
˜
λ(Q
1
¸ K
t
n
) ≤
˜
λ(Q
1
¸ L
n
) +
˜
λA(1
n
) +
˜
λ(Q
1
∩
_
1
t
n
¸ K
n
)
= λ(Q
1
¸ L
n
) +λ(Q
1
∩
_
1
t
n
¸ K
n
) ≤ 2
−n+1
.
Now K =
n∈N
K
t
n
belongs to / and
K ⊆ Q
1
¸
n∈N
A(1
n
) ⊆ Q
0
⊆ Q
417D τadditive product measures 93
because K
t
n
⊆ Q
1
¸ A(1
n
) for every n. And
λ(Q
1
¸ K) ≤
∞
n=0
˜
λ(Q
1
¸ K
t
n
) ≤ 4,
so
˜
λK = λK ≥ γ. As Q and γ are arbitrary,
˜
λ is inner regular with respect to /.
417D Multiple products Just as with the c.l.d. product measure (see 251W), we can apply the con
struction of 417C repeatedly to obtain measures on the products of ﬁnite families of τadditive measure
spaces.
Proposition (a) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a ﬁnite family of eﬀectively locally ﬁnite τadditive topological
measure spaces. Then there is a unique complete locally determined eﬀectively locally ﬁnite τadditive
topological measure
˜
λ on X =
i∈I
X
i
, inner regular with respect to the Borel sets, such that
˜
λ(
i∈I
G
i
) =
i∈I
µ
i
G
i
whenever G
i
∈ T
i
for every i ∈ I.
(b) If now ¸I
k
)
k∈K
is a partition of I, and
˜
λ
k
is the product measure deﬁned by the construction of (a)
on Z
k
=
i∈I
k
X
i
for each k ∈ K, then the natural bijection between X and
k∈K
Z
k
identiﬁes
˜
λ with the
product of the
˜
λ
k
deﬁned by the construction of (a).
proof (a)(i) Suppose ﬁrst that every µ
i
is inner regular with respect to the Borel sets. Then a direct
induction on #(I), using 417C for the inductive step, tells us that there is a measure
˜
λ with the required
properties. Note that, in 417C, (vi) ensures that (in the present context) all our product measures will be
inner regular with respect to the Borel sets.
(ii) For the general case, apply (a) to µ
i
B(X
i
), where B(X
i
) is the Borel σalgebra of X
i
for each i.
(iii) To see that
˜
λ is unique, suppose that λ
t
is another measure with the same properties. Let  be
the set ¦
i∈I
G
i
: G
i
∈ T
i
for every i ∈ I¦, and 
s
the set of ﬁnite unions of members of . Then
˜
λ and
λ
t
agree on 
s
. PPP Suppose that U =
j≤n
i∈I
G
ji
, where G
ji
∈ T
i
for j ≤ n, i ∈ I. If there is any j
such that
i∈I
µ
i
G
ji
= ∞, then
˜
λU = λ
t
U = ∞. Otherwise, set L = ¦j : j ≤ n,
i∈I
µ
i
G
ji
> 0¦ and
G
∗
i
=
j∈L
G
ji
for i ∈ I. Then
˜
λ(U ¸ (
i∈I
G
∗
i
)) = 0 = λ
t
(U ¸ (
i∈I
G
∗
i
)).
On the other hand, µ
i
G
ji
must be ﬁnite whenever j ∈ L and i ∈ I, so µ
i
G
∗
i
is ﬁnite for every i. Consider
1 = ¦
i∈I
G
i
: G
i
∈ T
i
, G
i
⊆ G
∗
i
for every i ∈ I¦. Then V ∩ V
t
∈ 1 for all V , V
t
∈ 1, and
˜
λ, λ
t
agree on
1. It follows from the Monotone Class Theorem (136C), or otherwise, that
˜
λ and λ
t
agree on the algebra of
subsets of
i∈I
G
∗
i
generated by 1. In particular,
˜
λ(U ∩(G
∗
H
∗
)) = λ
t
(U ∩(G
∗
H
∗
)), so that
˜
λU = λ
t
U.
QQQ
But now, because
˜
λ and λ
t
are τadditive,
˜
λW = sup¦
˜
λU : U ∈ 
s
, U ⊆ W¦ = sup¦λ
t
U : U ∈ 
s
, U ⊆ W¦ = λ
t
W
for every open set W ⊆ X. Writing B(X) for the Borel σalgebra of X,
˜
λB(X) and λ
t
B(X) are eﬀectively
locally ﬁnite Borel measures which agree on the open sets, so must be equal, by 414L. Since both
˜
λ and λ
t
are complete locally determined measures deﬁned on B and inner regular with respect to B, they also are
equal, by 412L.
(b) Let λ
t
be the measure on X corresponding to the τadditive product of the
˜
λ
k
on
k∈K
Z
k
. Then λ
t
is an eﬀectively locally ﬁnite complete locally determined τadditive topological measure inner regular with
respect to the zero sets, and if G
i
∈ T
i
for every i ∈ I then
λ
t
(
i∈I
G
i
) =
k∈K
˜
λ
k
(
i∈I
k
G
i
) =
k∈K
i∈I
k
µ
i
G
i
=
i∈I
µ
i
G
i
,
so λ
t
=
˜
λ.
94 Topologies and Measures 417E
417E Theorem Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces, with
product probability space (X, Λ, λ). Then λ is τadditive, and has an extension to a τadditive topological
measure
˜
λ on X. Moreover, we can arrange that:
(i)
˜
λ is complete;
(ii) if Q is measured by
˜
λ, there is a Q
1
∈ Λ such that
˜
λ(Q´Q
1
) = 0; that is to say, the embedding Λ ⊆
˜
Λ
induces an isomorphism between the measure algebras of λ and
˜
λ;
(iii)
˜
λW = λ
∗
W for every open set W ⊆ X, and
˜
λF = λ
∗
F for every closed set F ⊆ X;
(iv) the support of λ is the product of the supports of the µ
i
;
(v) if λ is inner regular with respect to the Borel sets, so is
˜
λ;
(vi) if λ is inner regular with respect to the closed sets, so is
˜
λ;
(vii) if λ is tight, so is
˜
λ.
proof The strategy of the proof is the same as in 417C, subject to some obviously necessary modiﬁcations.
The key step, showing that every union
n∈N
A(1
n
) has zero inner measure, is harder, but we do save a
little work because we no longer have to worry about sets of inﬁnite measure.
(a) I begin by setting up some machinery. Let ( be the family of subsets of X expressible in the form
i∈I
E
i
, where E
i
∈ Σ
i
for every i and ¦i : E
i
,= X
i
¦ is ﬁnite. Let  ⊆ ( be the standard basis for the
topology T of X, consisting of sets expressible as
i∈I
G
i
where G
i
∈ T
i
for every i ∈ I and ¦i : G
i
,= X
i
¦
is ﬁnite. Write 
s
for the set of ﬁnite unions of members of , and V for the set of nonempty upwards
directed families in 
s
. Note that every member of 
s
is determined by coordinates in some ﬁnite subset of
I (deﬁnition: 254M).
If J ⊆ I, write λ
J
for the product measure on
i∈J
X
i
; we shall need λ
∅
, which is the unique probability
measure on the singlepoint set ¦∅¦ =
i∈∅
X
i
. For J ⊆ I, v ∈
i∈J
X
i
, W ⊆ X set
f
W
(v) = λ
I\J
¦w : (v, w) ∈ W¦
if this is deﬁned, identifying
i∈J
X
i
i∈I\J
X
i
with X.
(b) We need two easy facts.
(i) f
W
(v) =
_
f
W
(v
t)µ
j
(dt) whenever W ∈
´
i∈I
Σ
i
, J ⊆ I, v ∈
i∈J
X
i
and j ∈ I ¸ J, writing v
t
for the member of
i∈J∪¦j]
X
i
extending v and taking the value t at the coordinate j. PPP Let / be the
family of sets W satisfying the property. Then / is a Dynkin class including (, so includes the σalgebra
generated by (, which is
´
i∈I
Σ
i
. QQQ
(ii) If J ⊆ I, v ∈
i∈J
X
i
, j ∈ I ¸ J and V ∈ 
s
, and we set g(t) = f
V
(v
t) for t ∈ X
j
, then g is lower
semicontinuous. PPP We can express V as
n≤m
i∈I
G
ni
, where G
ni
⊆ X
i
is open for every n ≤ m, i ∈ I.
Now if t ∈ X
j
, we shall have
¦w : (v
t, w) ∈ V ¦ ⊆ ¦w : (v
t
t
, w) ∈ V ¦
whenever
t
t
∈ H = X
j
∩
¦G
nj
: n ≤ m, t ∈ G
nj
¦.
So g(t
t
) ≥ g(t) for every t
t
∈ H, which is an open neighbourhood of t. As t is arbitrary, g is lower
semicontinuous. QQQ
(c) For each 1 ∈ V, ﬁx, for the remainder of this proof, a countable 1
t
⊆ 1 such that sup
V ∈\
λV =
sup
V ∈\
λV ; because 1 is upwardsdirected, we may suppose that 1
t
= ¦V
n
: n ∈ N¦ for some nondecreasing
sequence ¸V
n
)
n∈N
in 1. Set A(1) =
1 ¸
1
t
.
??? Suppose, if possible, that there is a sequence ¸1
n
)
n∈N
in V such that λ
∗
(
n∈N
A(1
n
)) > 0.
(i) We have λ
∗
(X ¸
n∈N
A(1
n
)) < 1; let ¸C
n
)
n∈N
be a sequence in ( such that
X ¸
n∈N
A(1
n
) ⊆
n∈N
C
n
,
∞
n=0
λC
n
= γ
0
< 1
(see 254A254C). For each n ∈ N, express 1
t
n
as ¦V
nr
: r ∈ N¦ where ¸V
nr
)
r∈N
is nondecreasing, and set
W
n
=
1
t
n
=
r∈N
V
nr
. Let J ⊆ I be a countable set such that every C
n
and every V
nr
is dependent on
coordinates in J. Express J as
k∈N
J
k
where J
0
= ∅ and, for each k, J
k+1
is equal either to J
k
or to J
k
417E τadditive product measures 95
with one point added. (As in the proof of 254Fa, I am using a formulation which will apply equally to ﬁnite
and inﬁnite I, though of course the case of ﬁnite I is elementary once we have 417C.)
(ii) For each n ∈ N, set
W
t
n
=
k∈N
¦x : x ∈ X, f
W
n
(xJ
k
) = 1¦.
Then λ(W
t
n
¸ W
n
) = 0. PPP For any k ∈ N, if we think of λ as the product of λ
J
k
and λ
I\J
k
and of f
W
n
as a
measurable function on
i∈J
k
X
i
, we see that ¦x : f
W
n
(xJ
k
) = 1¦ is of the form F
k
i∈I\J
k
X
i
, where
F
k
⊆
i∈J
k
X
i
is measurable; and
λ((F
k
i∈I\J
k
X
i
) ¸ W
n
) =
_
F
k
(1 −f
W
n
(v))λ
J
k
(dv) = 0.
Summing over k, we see that W
t
n
¸ W
n
is negligible. QQQ
Observe that every W
t
n
, like W
n
, is determined by coordinates in J. So
n∈N
W
t
n
¸ W
n
is of the form
E
i∈I\J
X
i
where λ
J
E = 0 (254Ob). There is therefore a sequence ¸D
n
)
n∈N
of measurable cylinders in
i∈J
X
i
such that E ⊆
n∈N
D
n
and
∞
n=0
λ
J
D
n
< 1 −γ
0
. Set C
t
n
= ¦x : x ∈ X, xJ ∈ D
n
¦ ∈ ( for each
n. Then
n∈N
W
t
n
¸ W
n
⊆
n∈N
C
t
n
, so
(X ¸
n∈N
A(1
n
)) ∪
n∈N
W
t
n
¸ W
n
⊆
n∈N
C
n
∪
n∈N
C
t
n
,
γ =
∞
n=0
λC
n
+
∞
n=0
λC
t
n
< 1,
while each C
n
and each C
t
n
is dependent on coordinates in a ﬁnite subset of J.
(iii) For k ∈ N, let P
k
be the set of those v ∈
i∈J
k
X
i
such that
∞
n=0
f
C
n
(v) +f
C
n
(v) ≤ γ, f
V
(v) ≤ f
W
n
(v) for every n ∈ N, V ∈ 1
n
.
Our hypothesis is that
∞
n=0
f
C
n
(∅) +f
C
n
(∅) =
∞
n=0
λC
n
+λC
t
n
≤ γ,
and the 1
t
n
were chosen such that
f
V
(∅) = λV ≤ λW
n
= f
W
n
(∅)
for every n ∈ N, V ∈ 1
n
; that is, ∅ ∈ P
0
.
(iv) Now if k ∈ N, v ∈ P
k
there is a v
t
∈ P
k+1
extending v. PPP If J
k+1
= J
k
we can take v
t
= v.
Otherwise, J
k+1
= J
k
∪ ¦j¦ for some j ∈ I ¸ J
k
. Now
γ ≥
∞
n=0
f
C
n
(v) +f
C
n
(v) =
∞
n=0
_
f
C
n
(v
t) +f
C
n
(v
t) µ
j
(dt)
((bi) above)
=
_
∞
n=0
f
C
n
(v
t) +f
C
n
(v
t) µ
j
(dt),
so
H = ¦t : t ∈ X
j
,
∞
n=0
f
C
n
(v
t) +f
C
n
(v
t)µ
j
(dt) ≤ γ¦
has positive measure.
Next, for V ∈ 
s
, set g
V
(t) = f
V
(v
t) for each t ∈ X
j
. Then g
V
is lower semicontinuous, by (bii)
above. For each n ∈ N, ¦g
V
: V ∈ 1
n
¦ is an upwardsdirected family of lower semicontinuous functions, so
its supremum g
∗
n
is also lower semicontinuous, and because µ
j
is τadditive,
_
g
∗
n
dµ
j
= sup
V ∈\
n
_
g
V
dµ
j
= sup
V ∈\
n
f
V
(v) ≤ f
W
n
(v) =
_
f
W
n
(v
t)µ
j
(dt)
(using 414B and (bi) again). But also, because ¸V
nr
)
r∈N
is nondecreasing and has union W
n
,
f
W
n
(v
t) = sup
r∈N
f
V
nr
(v
t) ≤ g
∗
n
(t)
for every t ∈ X
j
. So we must have
96 Topologies and Measures 417E
f
W
n
(v
t) = g
∗
n
(t) a.e.(t).
And this is true for every n ∈ N.
There is therefore a t ∈ H such that
f
W
n
(v
t) = g
∗
n
(v
t) for every n ∈ N.
Fix on such a t and set v
t
= v
t ∈
i∈J
k+1
X
i
; then v
t
∈ P
k+1
, as required. QQQ
(v) We can therefore choose a sequence ¸v
k
)
k∈N
such that v
k
∈ P
k
and v
k+1
extends v
k
for each k.
Choose x ∈ X such that x(i) = v
k
(i) whenever k ∈ N, i ∈ J
k
and x(i) belongs to the support of µ
i
whenever
i ∈ I ¸ J. (Recall from 411Nd that µ
i
does have a support.)
We need to know that if k, n ∈ N and V ∈ 1
n
then f
V \W
n
(v
k
) = 0. PPP For any r ∈ N there is a V
t
∈ 1
n
such that V ∪ V
nr
⊆ V
t
, so
f
V ∪V
nr
(v
k
) ≤ f
V
(v
k
) ≤ f
W
n
(v
k
),
and
f
V \W
n
(v
k
) ≤ f
V \V
nr
(v
k
) = f
V ∪V
nr
(v
k
) −f
V
nr
(v
k
) ≤ f
W
n
(v
k
) −f
V
nr
(v
k
) → 0
as r → ∞. QQQ
(vi) If n ∈ N, then x / ∈ C
n
∪ C
t
n
. PPP C
n
and C
t
n
are determined by coordinates in a ﬁnite subset
of J, so must be determined by coordinates in J
k
for some k ∈ N. Now f
C
n
(v
k
) + f
C
n
(v
k
) ≤ γ < 1, so
¦y : yJ
k
= v
k
¦ cannot be included in C
n
∪ C
t
n
, and must be disjoint from it; accordingly x / ∈ C
n
∪ C
t
n
. QQQ
(vii) Because
(X ¸
n∈N
A(1
n
)) ∪
n∈N
W
t
n
¸ W
n
⊆
n∈N
C
n
∪
n∈N
C
t
n
,
there is some n ∈ N such that
x ∈ A(1
n
) ¸ (W
t
n
¸ W
n
) ⊆ (
1
n
) ¸ W
t
n
,
that is, there is some V ∈ 1
n
such that x ∈ V ¸ W
t
n
. Let U ∈  be such that x ∈ U ⊆ V . Express U as
U
t
∩ U
tt
where U
t
∈  is determined by coordinates in a ﬁnite subset of J, and U
tt
∈  is determined by
coordinates in a ﬁnite subset of I ¸ J. Let k ∈ N be such that U
t
is determined by coordinates in J
k
. Then
f
U\W
n
(v
k
) ≤ f
V \W
n
(v
k
) = 0
by (v) above. Now
¦w : w ∈
i∈I\J
k
X
i
, (v
k
, w) ∈ U ¸ W
n
¦ = ¦w : (v
k
, w) ∈ U
tt
¸ W
n
¦
(because (v
k
, w) = (xJ
k
, w) ∈ U
t
for every w), while
¦w : (v
k
, w) ∈ U
tt
¦, ¦w : (v
k
, w) ∈ W
n
¦
are stochastically independent because the former depends on coordinates in I ¸ J, while the latter depends
on coordinates in J ¸ J
k
. So we must have
0 = f
U\W
n
(v
k
) = λ
I\J
k
¦w : (v
k
, w) ∈ U ¸ W
n
¦
= λ
I\J
k
¦w : (v
k
, w) ∈ U
tt
¸ W
n
¦
= λ
I\J
k
¦w : (v
k
, w) ∈ U
tt
¦(1 −λ
I\J
k
¦w : (v
k
, w) ∈ W
n
¦).
At this point, recall that x(i) belongs to the support of µ
i
for every i ∈ I ¸ J, while x ∈ U
tt
. So if
U
tt
= ¦y : y(i) ∈ H
i
for i ∈ K¦, where K ⊆ I ¸ J is ﬁnite and H
i
⊆ X
i
is open for every i, we must have
µ
i
H
i
> 0 for every i, and
λ
I\J
k
¦w : (v
k
, w) ∈ U
tt
¦ =
i∈K
µ
i
H
i
> 0.
On the other hand, we are also supposing that x / ∈ W
t
n
, so
λ
I\J
k
¦w : (v
k
, w) ∈ W
n
¦ = f
W
n
(v
k
) < 1.
But this means that we have expressed 0 as the product of two nonzero numbers, which is absurd. XXX
417F τadditive product measures 97
(d) Thus λ
∗
(
n∈N
A(1
n
)) = 0 for every sequence ¸1
n
)
n∈N
in V. Accordingly there is an extension of λ
to a measure
˜
λ on X such that
˜
λA(1) = 0 for every 1 ∈ V, the domain of
˜
λ being
˜
Λ = ¦W´A : W ∈ Λ, A ∈ /
∗
¦,
where /
∗
is the σideal generated by ¦A(1) : V ∈ V¦ (417A).
Now
˜
λ is a topological measure. PPP If W ⊆ X is open, then 1 = ¦V : V ∈ 
s
, V ⊆ W¦ belongs to V,
and
1 = W. Now
1
t
∈ Λ (because 1
t
is countable), so
W =
1
t
∪ A(1)
is measured by
˜
λ. QQQ
Also,
˜
λ is τadditive. PPP Let J be a nonempty upwardsdirected family of open subsets of X with union
W
∗
. Set
1 = ¦V : V ∈ 
s
, ∃ W ∈ J, V ⊆ W¦.
Then 1 ∈ V and
1 = W
∗
, so
˜
λA(1) = 0 and
˜
λW
∗
=
˜
λ(
1
t
) = sup
V ∈\
˜
λV ≤ sup
W∈V
˜
λW ≤
˜
λW
∗
(using the fact that 1
t
is upwardsdirected). As J is arbitrary,
˜
λ is τadditive. QQQ
Of course it follows at once that λ is also τadditive.
(e) Now for the supplementary properties (i)(vi) listed in the theorem.
(i) If Q ∈
˜
Λ,
˜
λQ = 0 and Q
t
⊆ Q, then Q is expressible as W´A where λW = 0 and A ∈ /
∗
. But in
this case (because λ is complete) λ(Q
t
∩ W) = 0 while Q
t
¸ W ⊆ A ∈ /
∗
, so Q
t
∈
˜
Λ. Thus
˜
λ is complete.
(ii) As always, the construction ensures that every member of
˜
Λ diﬀers by a
˜
λnegligible set from some
member of Λ.
(iii) Let W ⊆ X be an open set. Set 1 = ¦V : V ∈ 
s
, V ⊆ W¦. Then
˜
λW = sup
V ∈\
˜
λV = sup
V ∈\
λV ≤ λ
∗
W ≤
˜
λW
just because
˜
λ is a τadditive extension of λ. Now if F ⊆ X is closed,
˜
λF = 1 −
˜
λ(X ¸ F) = 1 −λ
∗
(X ¸ F) = λ
∗
F.
(iv) For each i ∈ I write F
i
for the support of µ
i
, and set F =
i∈I
F
i
. This is closed because every
F
i
is. Its complement is covered by the negligible open sets ¦x : x ∈ X, x(i) ∈ X
i
¸ F
i
¦ as i runs over I; as
λ is τadditive, the union of the negligible open sets is negligible, and F is conegligible. If W ⊆ X is open
and x ∈ F ∩ W, let U ∈  be such that x ∈ U ⊆ W. Express U as
i∈I
G
i
where G
i
∈ T
i
for every i ∈ I
and J = ¦i : G
i
,= X
i
¦ is ﬁnite. Then x(i) ∈ G
i
∩ F
i
, so µ
i
G
i
> 0, for every i. Accordingly
λ(W ∩ F) = λW = λU =
i∈J
µ
i
U
i
> 0.
Thus F is selfsupporting and is the support of λ.
(v), (vi), (vi) Use the same arguments as in the corresponding parts of 417C, this time using 412U
to conﬁrm that λ is inner regular with respect to the given family of sets.
417F Corollary Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such
that µ
i
is inner regular with respect to the Borel sets for each i. Then there is a unique complete τadditive
topological measure
˜
λ on X =
i∈I
X
i
which extends the ordinary product measure and is inner regular
with respect to the Borel sets.
proof By 417E(v) we have a measure
˜
λ with the right properties. If λ
t
is any other complete τadditive
topological measure, extending λ and inner regular with respect to the family B of Borel sets, then λ
t
W =
˜
λW
for every open set W ⊆ X. PPP By the argument of (eiii) of the proof of 417E, λ
t
W = λ
∗
W =
˜
λW. QQQ By
414L, applied to the Borel measures λ
t
B and
˜
λB, λ
t
W =
˜
λW for every Borel set W. Now λ
t
and
˜
λ are
supposed to be complete topological probability measures inner regular with respect to B, so they must be
identical, by 412L or otherwise.
98 Topologies and Measures 417G
417G Notation In the context of 417D or 417F, I will call
˜
λ the τadditive product measure on
i∈I
X
i
.
Note that the uniqueness assertions in 417D and 417F mean that for the products of ﬁnitely many
probability spaces we do not need to distinguish between the two constructions. The latter also shows that
we can relate 415E to the new method: if every T
i
is separable and metrizable and every µ
i
is strictly positive,
then the ‘ordinary’ product measure λ is a complete topological measure. Since it is also inner regular with
respect to the Borel sets (412Uc), it must be exactly the τadditive product measure as described here.
417H Fubini’s theorem for τadditive product measures Let (X, T, Σ, µ) and (Y, S, T, ν) be two
complete locally determined eﬀectively locally ﬁnite τadditive topological measure spaces such that both µ
and ν are inner regular with respect to the Borel sets. Let
˜
λ be the τadditive product measure on X Y ,
and
˜
Λ its domain.
(a) Let f be a [−∞, ∞]valued function such that
_
f d
˜
λ is deﬁned in [−∞, ∞] and (X Y ) ¸ ¦(x, y) :
(x, y) ∈ domf, f(x, y) = 0¦ can be covered by a set of the form X
n∈N
Y
n
where νY
n
< ∞ for every
n ∈ N. Then the repeated integral
__
f(x, y)ν(dy)µ(dx) is deﬁned and equal to
_
fd
˜
λ.
(b) Let f : X Y → [0, ∞] be lower semicontinuous. Then
__
f(x, y)ν(dy)µ(dx) =
__
f(x, y)µ(dx)ν(dy) =
_
fd
˜
λ
in [0, ∞].
(c) Let f be a
˜
Λmeasurable realvalued function deﬁned
˜
λa.e. on XY . If either
__
[f(x, y)[ν(dy)µ(dx)
or
__
[f(x, y)[µ(dx)ν(dy) is deﬁned and ﬁnite, then f is
˜
λintegrable.
proof (a) I use 252B.
(i) Write J for the set of those W ∈
˜
Λ such that
_
νW[¦x¦]µ(dx) is deﬁned in [0, ∞] and equal to
˜
λW. Then open sets belong to J, by 417C(iv). Next, any Borel subset of an open set of ﬁnite measure
belongs to J. PPP If W
0
is an open set of ﬁnite measure, then ¦W : W ⊆ X Y, W ∩W
0
∈ J¦ is a Dynkin
class containing every open set, so contains all Borel subsets of X Y . QQQ
Now suppose that W ⊆ X Y is
˜
λnegligible and included in X
n∈N
Y
n
, where νY
n
< ∞ for every
n. Then W ∈ J. PPP Set A = ¦x : x ∈ X, ν
∗
W[¦x¦] > 0¦. For each n, let H
n
⊆ Y be an open set of ﬁnite
measure such that ν(Y
n
¸ H
n
) ≤ 2
−n
; we may arrange that H
n+1
⊇ H
n
for each n. Set H =
n∈N
H
n
, so
that W[¦x¦] ¸ H ⊆
n∈N
Y
n
¸ H is negligible for every x ∈ X.
Fix an open set G ⊆ X of ﬁnite measure and n ∈ N for the moment. Because
˜
λ is inner regular with
respect to the Borel sets, there is a Borel set V ⊆ (GH
n
) ¸ W such that
˜
λV =
˜
λ((GH
n
) ¸ W), that is,
˜
λV
t
= 0, where V
t
= (GH
n
) ¸ V ⊇ (GH
n
) ∩ W. We know that V
t
∈ J, so
_
νV
t
[¦x¦]dx =
˜
λV
t
= 0,
and νV
t
[¦x¦] = 0 for almost every x ∈ X; but this means that H
n
∩ W[¦x¦] is negligible for almost every
x ∈ G.
At this point, recall that n was arbitrary, so H ∩ W[¦x¦] and W[¦x¦] are negligible for almost every
x ∈ G, that is, A ∩ G is negligible. This is true for every open set G ⊆ X of ﬁnite measure. Because µ is
inner regular with respect to subsets of open sets of ﬁnite measure, and is complete and locally determined,
A is negligible (412Jb). But this means that
_
νW[¦x¦]µ(dx) is deﬁned and equal to zero, so that W ∈ J.
QQQ
(ii) Now suppose that
_
fd
˜
λ is deﬁned in [−∞, ∞] and that there is a sequence ¸Y
n
)
n∈N
of sets of
ﬁnite measure in Y such that f(x, y) is deﬁned and zero whenever x ∈ X and y ∈ Y ¸
n∈N
Y
n
. Set
Z =
n∈N
Y
n
. Write λ for the c.l.d. product measure on X Y and Λ for its domain. Then there is a
Λmeasurable function g : X Y → [−∞, ∞] which is equal
˜
λalmost everywhere to f. PPP For q ∈ Q set
W
q
= ¦(x, y) : (x, y) ∈ domf, f(x, y) ≥ q¦ ∈
˜
Λ, and choose V
q
∈ Λ such that
˜
λ(W
q
´V
q
) = 0 (417C(ii));
set g(x, y) = sup¦q : q ∈ Q, (x, y) ∈ V
q
¦ for x ∈ X, y ∈ Y , interpreting sup ∅ as −∞. QQQ Adjusting g if
necessary, we may suppose that it is zero on X (Y ¸ Z). Set
A = (X Y ) ¸ ¦(x, y) : f(x, y) = g(x, y)¦,
417I τadditive product measures 99
so that A is
˜
λnegligible and included in X Z. By (i), νA[¦x¦] = 0, that is, y → f(x, y) and y → g(x, y)
are equal νa.e., for µalmost every x. Write λ
X·Z
for the subspace measure induced by λ on X Z; note
that this is the c.l.d. product of µ with the subspace measure ν
Z
on Z, by 251P(iiα).
Now we have
_
fd
˜
λ =
_
g d
˜
λ =
_
g dλ
(by 235Ib, because the identity map from (X Y,
˜
λ) to (X Y, λ) is inversemeasurepreserving)
=
_
X·Z
g dλ =
_
X·Z
g dλ
X·Z
=
__
Z
g(x, y)ν
Z
(dy)µ(dx)
(by 252B, because ν
Z
is σﬁnite)
=
__
g(x, y)ν(dy)µ(dx)
(because g(x, y) = 0 if y ∈ Y ¸ Z)
=
__
f(x, y)ν(dy)µ(dx).
(b) If f is nonnegative and lower semicontinuous, set
W
ni
= ¦(x, y) : f(x, y) > 2
−n
i¦
for n, i ∈ N, and
f
n
= 2
−n
4
n
i=1
χW
ni
for n ∈ N. Applying 417C(iv) we see that
_
f
n
d
˜
λ =
__
f
n
(x, y)dydx =
__
f
n
(x, y)dxdy
in [0, ∞] for every n; taking the limit as n → ∞,
_
fd
˜
λ =
__
f(x, y)dydx =
__
f(x, y)dxdy,
because ¸f
n
)
n∈N
is a nondecreasing sequence with limit f.
(c) ??? Suppose, if possible, that γ =
__
[f(x, y)[dydx is ﬁnite, but that f is not integrable. Because
˜
λ is
semiﬁnite, there must be a nonnegative
˜
λsimple function g such that g ≤
a.e.
[f[ and
_
g d
˜
λ > γ (213B).
For each n ∈ N, there are open sets G
n
⊆ X, H
n
⊆ Y of ﬁnite measure such that
˜
λ(¦(x, y) : g(x, y) ≥
2
−n
¦ ¸ (G
n
H
n
)) ≤ 2
−n
, by 417C(iii); now g χ(G
n
H
n
) → g a.e., so there is some n such that
_
G
n
·H
n
g d
˜
λ > γ. In this case, setting g
t
(x, y) = min(g(x, y), [f(x, y)[) for (x, y) ∈ (G
n
H
n
) ∩ domf, 0
otherwise, we have g = g
t
a.e. on G
n
H
n
, so that
_
g
t
d
˜
λ > γ. But we can apply (a) to g
t
to see that
γ <
_
g
t
d
˜
λ =
__
g
t
(x, y)dydx ≤
__
[f(x, y)[dydx ≤ γ,
which is absurd. XXX
So if
__
[f(x, y)[dydx is ﬁnite, f must be
˜
λintegrable. Of course the same arguments, reversing the roles
of X and Y , show that f is
˜
λintegrable if
__
[f(x, y)[dxdy is deﬁned and ﬁnite.
417I The constructions here have most of the properties one would hope for. I give several in the
exercises (417Xd417Xf, 417Xj). One fact which is particularly useful, and also has a trap in it, is the
following.
Proposition Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological measure
spaces in which the measures are inner regular with respect to the Borel sets, and
˜
λ the τadditive product
measure on X Y . Suppose that A ⊆ X and B ⊆ Y , and write µ
A
, ν
B
for the corresponding subspace
measures; assume that both µ
A
and ν
B
are semiﬁnite. Then these are also eﬀectively locally ﬁnite, τ
additive and inner regular with respect to the Borel sets, and the subspace measure
˜
λ
A·B
induced by
˜
λ on
AB is just the τadditive product measure of µ
A
and ν
B
.
100 Topologies and Measures 417I
proof (a) To check that µ
A
and ν
B
are eﬀectively locally ﬁnite, τadditive and inner regular with respect to
the Borel sets, see 414K and 412P. Of course
˜
λ
A·B
inherits the same properties from
˜
λ, and is also complete
and locally determined, by 214Id.
(b) Now if C ∈ domµ
A
and D ∈ domν
B
, then
˜
λ
∗
(C D) = µ
A
C ν
B
D. PPP (α) There are E ∈ Σ, F ∈ T
such that C ⊆ E, D ⊆ F, µE = µ
∗
C and νF = ν
∗
D; in which case
˜
λ
∗
(C D) ≤
˜
λ(E F) = λ(E F) = µE νF
(251J)
= µ
∗
C ν
∗
D = µ
A
C ν
B
D.
(β) If γ < µ
A
Cν
B
D then, because µ
A
and ν
B
are semiﬁnite, there are C
t
⊆ C, D
t
⊆ D such that both have
ﬁnite outer measure and µ
∗
C
t
ν
∗
D
t
≥ γ. In this case, take E
t
∈ Σ, F
t
∈ T such that C
t
⊆ E
t
, D
t
⊆ F
t
and
both E
t
and F
t
have ﬁnite measure. Now if W ∈ dom
˜
λ and C D ⊆ W, we have C
t
D
t
⊆ W ∩(E F),
so that ν(W ∩ (E F))[¦x¦] ≥ ν
∗
D
t
for every x ∈ C
t
, and
˜
λW ≥
_
E
ν(W ∩ (E F))[¦x¦]µ(dx) ≥ µ
∗
C
t
ν
∗
D
t
≥ γ,
by 417Ha. As W is arbitrary,
˜
λ
∗
(C D) ≥ γ; as γ is arbitrary,
˜
λ
∗
(C D) ≥ µ
A
C ν
B
D. QQQ
(c) In particular, if U ⊆ A and V ⊆ B are relatively open,
˜
λ
A·B
(U V ) =
˜
λ
∗
(U V ) = µ
A
U ν
B
V .
But now 417D tells us that
˜
λ
A·B
must be exactly the τadditive product measure of µ
A
and ν
B
.
417J In order to use 417H eﬀectively in the theory of inﬁnite products, we need an ‘associative law’
corresponding to 254N.
Theorem Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that every
µ
i
is inner regular with respect to the Borel sets, and ¸K
j
)
j∈J
a partition of I. For each j ∈ J let
˜
λ
j
be the τadditive product measure on Z
j
=
i∈K
j
X
i
, and write
˜
λ for the τadditive product measure on
X =
i∈I
X
i
. Then the natural bijection
x → φ(x) = ¸xK
j
)
j∈J
: X →
j∈J
Z
j
identiﬁes
˜
λ with the τadditive product of the family ¸
˜
λ
j
)
j∈J
.
In particular, if K ⊆ I is any set, then
˜
λ can be identiﬁed with the τadditive product of the τadditive
product measures on
i∈K
X
i
and
i∈I\K
X
i
.
proof We have a lot of measures to keep track of; I hope that the following notation will not be too
opaque. Write λ for the ordinary product measure on X, and for j ∈ J write λ
j
,
˜
λ
j
for the ordinary and
τadditive product measures on Z
j
. Write θ for the ordinary product measure on Z =
j∈J
Z
j
of the
τadditive product measures
˜
λ
j
, and
˜
θ for the τadditive product of the
˜
λ
j
. Write
˜
λ
#
for the measure on X
corresponding to
˜
θ on Z. (If you like,
˜
λ
#
is the image measure
˜
θ(φ
−1
)
−1
deﬁned from
˜
θ and the function
φ
−1
: Z → X.) Then
˜
λ
#
, like
˜
θ, is a complete τadditive topological measure, inner regular with respect
to the Borel sets, because φ : X → Z is a homeomorphism. If C ⊆ X is a measurable cylinder, it is of the
form
i∈I
E
i
where E
i
∈ Σ
i
for each i and ¦i : i ∈ I, E
i
,= Σ
i
¦ is ﬁnite. So φ[C] is of the form
j∈J
C
j
,
where C
j
=
i∈K
j
E
i
, and
˜
λ
#
C =
˜
θ(
j∈J
C
j
) = θ(
j∈J
C
j
) =
j∈J
˜
λ
j
C
j
=
j∈J
λ
j
C
j
=
j∈J
i∈K
j
E
i
=
i∈I
E
i
= λC.
But this means, applying 254G to the identity map from (X,
˜
λ
#
) to (X, λ), that
˜
λ
#
extends λ. So it
is a complete τadditive topological measure, inner regular with respect to the Borel sets, extending the
ordinary product measure, and by the uniqueness declared in 417F, must be identical to the τadditive
product measure on X, as claimed.
417M τadditive product measures 101
417K Proposition Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such
that every µ
i
is inner regular with respect to the Borel sets, and let (X,
˜
Λ,
˜
λ) be their τadditive product.
For J ⊆ I let
˜
λ
J
be the τadditive product measure on X
J
=
i∈J
X
i
, and
˜
Λ
J
its domain; let π
J
: X → X
J
be the canonical map. Then
˜
λ
J
is the image measure
˜
λπ
−1
J
. In particular, if W ∈
˜
Λ is determined by
coordinates in J ⊆ I, then π
J
[W] ∈
˜
Λ
J
and
˜
λ
J
π
J
[W] =
˜
λW.
proof Because
˜
λ is an extension of the ordinary product measure λ on X,
˜
λπ
−1
J
is an extension of λπ
−1
J
,
which is the ordinary product measure on X
J
(254Oa). Because
˜
λ is a τadditive topological measure and
π
J
is continuous,
˜
λπ
−1
J
is a τadditive topological measure; because
˜
λ is a complete probability measure,
so is
˜
λπ
−1
J
. Finally,
˜
λπ
−1
J
is inner regular with respect to the Borel sets. PPP Recall that we may identify
˜
λ
with the τadditive product of
˜
λ
J
and
˜
λ
I\J
(417J). If V ∈ dom
˜
λπ
−1
J
, that is, π
−1
J
[V ] ∈
˜
Λ, we can think of
π
−1
J
[V ] ⊆ X as V X
I\J
⊆ X
I
X
J
. In this case, we must have
˜
λπ
−1
J
[V ] =
_
˜
λ
J
V dλ
I\J
,
by Fubini’s theorem for τadditive products (417Ha); that is,
˜
λ
J
V must be deﬁned and equal to
˜
λπ
−1
J
[V ].
Now if γ <
˜
λπ
−1
J
[V ], there must be a Borel set V
t
⊆ V such that
˜
λ
J
V
t
≥ γ. In this case, because π
J
is
continuous, π
−1
J
[V
t
] is also Borel, and
˜
λπ
−1
J
[V
t
] is deﬁned. As with V , this measure must be
˜
λ
J
V
t
≥ γ.
Since V and γ are arbitrary,
˜
λπ
−1
J
is inner regular with respect to the Borel sets, as claimed. QQQ
By the uniqueness assertion in 417F,
˜
λπ
−1
J
must be
˜
λ
J
exactly.
If now W ∈
˜
Λ is determined by coordinates in J, then
˜
λ
J
π
J
[W] =
˜
λπ
−1
J
[π
J
[W]] =
˜
λW.
417L Corollary Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that
every µ
i
is inner regular with respect to the Borel sets, and (X,
˜
Λ,
˜
λ) their τadditive product. Let ¸K
j
)
j∈J
be a disjoint family of subsets of I, and for j ∈ J write
˜
Λ
j
for the σalgebra of members of Λ determined by
coordinates in K
j
. Then ¸
˜
Λ
j
)
j∈J
is a stochastically independent family of σalgebras (deﬁnition: 272Ab).
proof It is enough to consider the case in which J is ﬁnite (272Bb), no K
j
is empty (since if K
j
= ∅
then
˜
Λ
j
= ¦∅, X¦) and
j∈J
K
j
= I (adding an extra term if necessary). In this case, if W
j
∈
˜
Λ
j
for
each j, then the identiﬁcation between X and
j∈J
i∈K
j
X
i
, as described in 417J, matches
j∈J
W
j
with
j∈J
π
K
j
[W
j
], writing π
K
j
(x) for xK
j
. Now if
˜
λ
j
is the τadditive product measure on Z
j
=
i∈K
j
X
i
, we
have
˜
λ
j
π
K
j
[W
j
] =
˜
λW
j
, by 417K. Since
˜
λ can be identiﬁed with the τadditive product of ¸
˜
λ
j
)
j∈J
(417J),
˜
λ(
j∈J
W
j
) =
j∈J
˜
λ
j
π
K
j
[W
j
] =
j∈J
˜
λW
j
.
As ¸W
j
)
j∈J
is arbitrary, ¸
˜
Λ
j
)
j∈J
is independent.
417M Proposition Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such
that every µ
i
is inner regular with respect to the Borel sets and strictly positive. For J ⊆ I let π
J
be the
canonical map from X onto X
J
=
i∈J
X
i
; write λ
J
,
˜
λ
J
for the ordinary and τadditive product measures
on X
J
, and Λ
J
,
˜
Λ
J
for their domains. Set
˜
λ =
˜
λ
I
,
˜
Λ =
˜
Λ
I
, λ = λ
I
, Λ = Λ
I
.
(a) Let F ⊆ X be a closed selfsupporting set, and J the smallest subset of I such that F is determined
by coordinates in J (4A2B(gii)). Then
(i) if W ∈
˜
Λ is such that W´F is
˜
λnegligible and determined by coordinates in K ⊆ I, then K ⊇ J;
(ii) J is countable;
(iii) there is a W ∈ Λ, determined by coordinates in J, such that W´F is
˜
λnegligible.
(b)
˜
λ is inner regular with respect to the family of sets of the form
n∈N
V
n
where each V
n
∈
˜
Λ is
determined by ﬁnitely many coordinates.
(c) If W ∈
˜
Λ, there are a countable J ⊆ I and sets W
t
, W
tt
∈
˜
Λ, determined by coordinates in J, such
that W
t
⊆ W ⊆ W
tt
and
˜
λ(W
tt
¸ W
t
) = 0. Consequently
˜
λπ
−1
J
[π
J
[W]] =
˜
λW.
102 Topologies and Measures 417M
proof (a)(i) ??? Suppose, if possible, otherwise. Then F is not determined by coordinates in K, so there
are x ∈ F, y ∈ X ¸ F such that xK = yK. Let U be an open set containing y, disjoint from F, and of the
form
i∈I
G
i
, where G
i
∈ T
i
for every i and L = ¦i : G
i
,= X
i
¦ is ﬁnite. Set
U
t
= ¦z : z ∈ X, z(i) ∈ G
i
for every i ∈ L ∩ K¦,
U
tt
= ¦z : z(i) ∈ G
i
for every i ∈ L ¸ K¦.
Then U
t
∩ W is determined by coordinates in K, while U
tt
is determined by coordinates in I ¸ K, so
0 =
˜
λ(F ∩ U) =
˜
λ(W ∩ U) =
˜
λ(W ∩ U
t
∩ U
tt
) =
˜
λ(W ∩ U
t
)
˜
λU
tt
(by 417L)
=
˜
λ(F ∩ U
t
)
˜
λU
tt
=
˜
λ(F ∩ U
t
)
i∈L\K
µ
i
G
i
.
But y ∈ U
t
, and xK = yK, so x ∈ F ∩ U
t
; as F is selfsupporting,
˜
λ(F ∩ U
t
) > 0. Because every µ
i
is
strictly positive, and no G
i
is empty,
i∈L\K
µ
i
G
i
> 0; and this is impossible. XXX
(ii) By 417E(ii), there is a W
0
∈ Λ such that
˜
λ(F´W
0
) = 0. By 254Oc there is a W
1
∈ Λ, determined
by coordinates in a countable subset K of I, such that λ(W
0
´W
1
) = 0. Now
˜
λ(F´W
1
) = 0, so (i) tells us
that J ⊆ K is countable.
(iii) By 417K, π
J
[F] ∈
˜
Λ
J
. By 417E(ii), there is a V ∈ Λ
J
such that V ´π
J
[F] is
˜
λ
J
negligible.
Set W = π
−1
J
[V ]. Then W ∈ Λ, W is determined by coordinates in J, and W´F = π
−1
J
[V ´π
J
[F]] is
˜
λnegligible.
(b)(i) Write 1 for the set of those members of
˜
Λ which are determined by ﬁnitely many coordinates,
and 1
δ
for the set of intersections of sequences in 1. Because 1 is closed under ﬁnite unions, so is 1
δ
; 1
δ
is
surely closed under countable intersections, and ∅, X belong to 1
δ
.
(ii) We need to know that every selfsupporting closed set F ⊆ X belongs to 1
δ
. PPP By (a), F is
determined by a countable set J of coordinates. Express J as the union of a nondecreasing sequence
¸J
n
)
n∈N
of ﬁnite sets. Then F
n
= π
−1
J
n
[π
J
n
[F]] ∈ 1 for each n, and F =
n∈N
F
n
∈ 1
δ
. QQQ
(iii) Let / be the family of subsets of X which are either open or closed. Then if A ∈ /, V ∈
˜
Λ and
˜
λ(A∩ V ) > 0, there is a K ∈ 1
δ
∩ / such that K ⊆ A and
˜
λ(K ∩ V ) > 0. PPP (α) If A is open, set
 = ¦U : U ∈ 1 is open, U ⊆ A¦.
Because 1 includes a base for the topology of X,
 = A; because
˜
λ is τadditive and 1 is closed under
ﬁnite unions, there is a U ∈  such that U ⊆ A and
˜
λU >
˜
λA −
˜
λ(A ∩ V ), so that
˜
λ(U ∩ V ) > 0. (β) If A
is closed, then it includes a selfsupporting closed set V of the same measure (414F), which belongs to 1
δ
,
by (ii) just above. QQQ
(iv) By 412C, the restriction
˜
λB of
˜
λ to the Borel σalgebra of X is inner regular with respect to 1
δ
.
But
˜
λ is just the completion of
˜
λB, so is also inner regular with respect to 1
δ
(412Ha).
(c) By (b), we have sequences ¸V
n
)
n∈N
, ¸V
t
n
)
n∈N
in 1
δ
such that V
n
⊆ W, V
t
n
⊆ X¸ W,
˜
λV
n
≥
˜
λW −2
−n
and
˜
λV
t
n
≥
˜
λ(X¸W) −2
−n
for every n ∈ N. Each V
n
, V
t
n
is determined by a countable set of coordinates, so
there is a single countable set J ⊆ I such that every V
n
and every V
t
n
is determined by coordinates in J. Set
W
t
=
n∈N
V
n
, W
tt
= X¸
n∈N
V
t
n
; then W
t
, W
tt
are both determined by coordinates in J, W
t
⊆ W ⊆ W
tt
and
˜
λ(W
tt
¸ W
t
) = 0, as required.
417N Theorem Let (X, T, Σ, µ) and (Y, S, T, ν) be two quasiRadon measure spaces. Then the τ
additive product measure
˜
λ on XY is a quasiRadon measure, the unique quasiRadon measure on XY
such that
˜
λ(E F) = µE νF for every E ∈ Σ, F ∈ T.
proof
˜
λ is a complete, locally determined, eﬀectively locally ﬁnite, τadditive topological measure, inner
regular with respect to the closed sets (417C(vii)). But this says just that it is a quasiRadon measure. By
417D, it is the unique quasiRadon measure with the right values on measurable rectangles.
417S τadditive product measures 103
417O Theorem Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of quasiRadon probability spaces. Then the τ
additive product measure
˜
λ on X =
i∈I
X
i
is a quasiRadon measure, the unique quasiRadon measure
on X extending the ordinary product measure.
proof By 417E(vi),
˜
λ is inner regular with respect to the closed sets, so is a quasiRadon measure, which
is unique by 417F.
417P Theorem Let (X, T, Σ, µ) and (Y, S, T, ν) be Radon measure spaces. Then the τadditive product
measure
˜
λ on XY is a Radon measure, the unique Radon measure on XY such that
˜
λ(EF) = µE νF
whenever E ∈ Σ, F ∈ T.
proof Of course XY is Hausdorﬀ, and
˜
λ is locally ﬁnite (because
˜
λ(GH) = µG νH is ﬁnite whenever
µG and νH are ﬁnite). By 417C(viii),
˜
λ is tight, so is a Radon measure. As in 417N, it is uniquely deﬁned
by its values on measurable rectangles.
417Q Theorem Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of Radon probability spaces, and
˜
λ the τadditive
product measure on X =
i∈I
X
i
. For each i ∈ I, let Z
i
⊆ X
i
be the support of µ
i
. Suppose that
J = ¦i : i ∈ I, Z
i
is not compact¦ is countable. Then
˜
λ is a Radon measure, the unique Radon measure on
X extending the ordinary product measure.
proof Of course X, being a product of Hausdorﬀ spaces, is Hausdorﬀ, and
˜
λ, being totally ﬁnite, is locally
ﬁnite. Now, given ∈ ]0, 1], let ¸
j
)
j∈J
be a family of strictly positive numbers such that
j∈J
j
≤ , and
for j ∈ J choose a compact set K
j
⊆ X
j
such that µ
j
K
j
≥ 1 −
j
; for i ∈ I ¸ J, set K
i
= Z
i
, so that K
i
is
compact and µ
i
K
i
= 1. Consider K =
i∈I
K
i
. Then, using 417E(iii) and 254Lb for the two equalities,
˜
λK = λ
∗
K =
i∈I
µ
i
K
i
≥
j∈J
1 −
j
≥ 1 −,
where λ is the ordinary product measure on X. As is arbitrary,
˜
λ satisﬁes the condition (iv) of 416C, and
is a Radon measure. As in 417F, it is the unique Radon measure on X extending λ.
417R Notation I will use the phrase quasiRadon product measure for a τadditive product measure
which is in fact a quasiRadon measure; similarly, a Radon product measure is a τadditive product
measure which is a Radon measure.
417S Later I will give an example in which a τadditive product measure is diﬀerent from the corre
sponding c.l.d. product measure (419E). In 415E415F, 415Ye and 416U I have described cases in which
c.l.d. measures are τadditive product measures. It remains very unclear when to expect this to happen.
I can however give a couple of results which show that sometimes, at least, we can be sure that the two
measures coincide.
Proposition (a) Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological measure
spaces such that both µ and ν are inner regular with respect to the Borel sets, and λ the c.l.d. product
measure on XY . If every open subset of XY is measured by λ, then λ is the τadditive product measure
on X Y .
(b) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that every µ
i
is
inner regular with respect to the Borel sets, and λ the ordinary product measure on X =
i∈I
X
i
. If every
open subset of X is measured by λ, then λ is the τadditive product measure on X.
(c) In (b), let λ
J
be the ordinary product measure on X
J
=
i∈J
X
i
for each J ⊆ I, and
˜
λ
J
the τadditive
product measure. If λ
J
=
˜
λ
J
for every ﬁnite J ⊆ I, and every µ
i
is strictly positive, then λ =
˜
λ
I
is the
τadditive product measure on X.
proof (a), (b) In both cases, λ is a complete locally determined eﬀectively locally ﬁnite τadditive measure
which is inner regular with respect to the Borel sets (assembling facts from 251I, 254F, 412S, 412U, 417C and
417E). The extra hypothesis added here is that λ is a topological measure, so itself satisﬁes the conditions
of 417D or 417F, and is the τadditive product measure.
104 Topologies and Measures 417S
(c)(i) The ﬁrst step is to note that λ
J
=
˜
λ
J
for every countable J ⊆ I. PPP Express J as
n∈N
J
n
where
¸J
n
)
n∈N
is a nondecreasing sequence of ﬁnite sets. If F ⊆ X
J
is closed, then it is
n∈N
π
−1
n
[π
n
[F]], where
π
n
: X
J
→ X
J
n
is the canonical map for each n. But every π
n
[F] is a closed subset of X
J
n
, therefore
measured by λ
J
n
; because π
n
is inversemeasurepreserving (417K), π
−1
n
[π
n
[F]] ∈ domλ
J
for each n, and
F ∈ domλ
J
. Thus every closed set, therefore every open set is measured by λ
J
, and λ
J
is a topological
measure; by (b), λ
J
=
˜
λ
J
. QQQ
(ii) Suppose that W ⊆ X is open. By 417M, there are W
t
, W
tt
measured by
˜
λ such that W
t
⊆ W ⊆ W
tt
,
both W
tt
and W
t
are determined by coordinates in a countable set, and
˜
λ
I
(W
tt
¸ W
t
) = 0. Let J ⊆ I be a
countable set such that W
t
and W
tt
depend on coordinates in J. Then λ
J
=
˜
λ
J
measures π
J
[W
t
], by 417K,
so λ measures W
t
= π
−1
J
[π
J
[W]], by 254Oa. Similarly, λ measures W
tt
. Now λ(W
tt
¸W
t
) =
˜
λ
I
(W
tt
¸W
t
) = 0,
so λ measures W. As W is arbitrary, λ is a topological measure and must be the τadditive product measure,
by (a).
417T Proposition Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological
measure spaces such that both µ and ν are inner regular with respect to the Borel sets, and λ the c.l.d.
product measure on X Y . If X has a conegligible subset with a countable network (e.g., if X is separable
and metrizable), then λ is the τadditive product measure on X Y .
proof (a) Suppose to begin with that µ and ν are totally ﬁnite, and that X has a countable network; let
¸A
n
)
n∈N
run over a network for X. Let ˆ µ be the completion of µ and
ˆ
Σ its domain. Let
˜
λ be the τadditive
product measure on X Y . (We are going to need Fubini’s theorem both for λ and for
˜
λ. I will use a
sprinkling of references to ¸¸251252 to indicate which parts of the argument below depend on the properties
of λ.)
Let W ⊆ X Y be an open set. For each n ∈ N, set
H
n
=
¦H : H ∈ S, A
n
H ⊆ W¦,
so that H
n
is open. Then W =
n∈N
A
n
H
n
. PPP Of course A
n
H
n
⊆ W for every n ∈ N. If (x, y) ∈ W,
there are open sets G ⊆ X, H ⊆ Y such that (x, y) ∈ G H ⊆ W; now there is an n ∈ N such that
x ∈ A
n
⊆ G, so that H ⊆ H
n
and (x, y) ∈ A
n
H
n
. QQQ
By 417C(iv), there is an open set W
0
in the domain Λ of λ such that W
0
⊆ W and
˜
λ(W ¸ W
0
) = 0. By
417Ha, applied to χ(W ¸ W
0
), A = ¦x : ν(W[¦x¦] ¸ W
0
[¦x¦]) > 0¦ is µnegligible. For each n ∈ N, x ∈ X
set f
n
(x) = ν(H
n
∩ W
0
[¦x¦]); then 252B tells us that
_
f
n
dµ is deﬁned and equal to λ(W
0
∩ (X H
n
)).
In particular, f
n
is
ˆ
Σmeasurable. Set E
n
= ¦x : f
n
(x) = νH
n
¦ ∈
ˆ
Σ. If x ∈ A
n
, then H
n
⊆ W[¦x¦], so
A
n
¸ E
n
⊆ A.
Now, by 252B again,
λ((E
n
H
n
) ¸ W
0
) =
_
E
n
ν(H
n
¸ W
0
[¦x¦])µ(dx)
=
_
E
n
νH
n
−ν(H
n
∩ W
0
[¦x¦])µ(dx) = 0.
So if we set W
1
=
n∈N
E
n
H
n
, W
1
¸ W ⊆ W
1
¸ W
0
is λnegligible. On the other hand,
W ¸ W
1
⊆
n∈N
(A
n
¸ E
n
) H
n
⊆ AY
is also λnegligible. Because λ is complete, W ∈ Λ. As W is arbitrary, λ is a topological measure and is
equal to
˜
λ, by 417Sa.
(b) Now consider the general case. Let Z be a conegligible subset of X with a countable network; since
any subset of a space with a countable network again has a countable network (4A2Na), we may suppose
that Z ∈ Σ. Again let W be an open set in XY . This time, take arbitrary E ∈ Σ, F ∈ T of ﬁnite measure,
and consider the subspace measures µ
E∩Z
and ν
F
. These are still eﬀectively locally ﬁnite and τadditive
(414K), and are now totally ﬁnite. Also E ∩ Z has a countable network. So (a) tells us that the relatively
open set W∩((E∩Z)F) is measured by the c.l.d. product of µ
E∩Z
and ν
F
, which is the subspace measure
on (E∩Z) F induced by λ (251P). Since λ surely measures EF, it measures W ∩(Z Y ) ∩(EF). As
417Xd τadditive product measures 105
E and F are arbitrary, λ measures W ∩(Z Y ) (251H). But λ((X ¸ Z) Y ) = µ(X Z) νY = 0 (251Ia),
so λ also measures W. As W is arbitrary, λ is the τadditive product measure.
417U Proposition Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces. Let
λ be the ordinary product probability measure on X =
i∈I
X
i
and Λ its domain. Then every continuous
function f : X →R is Λmeasurable, so Λ includes the Baire σalgebra of X.
proof (a) Let
˜
λ be a τadditive topological measure extending λ (417E), and
˜
Λ its domain; then f is
˜
Λmeasurable, just because
˜
λ is a topological measure. For α ∈ R, set
G
α
= ¦x : x ∈ X, f(x) < α¦, H
α
= ¦x : x ∈ X, f(x) > α¦,
F
α
= ¦x : x ∈ X, f(x) = α¦.
Then ¸F
α
)
α∈R
is disjoint, so A = ¦α : α ∈ R,
˜
λF
α
> 0¦ is countable, and A
t
= R ¸ A is dense in R; let
Q ⊆ A
t
be a countable dense set.
For each q ∈ Q, let V
q
⊆ G
q
, W
q
⊆ H
q
be such that
λV
q
= λ
∗
G
q
=
˜
λG
q
, λW
q
= λ
∗
H
q
=
˜
λH
q
(413Ea, 417E(iii)). Then
λ
∗
(G
q
¸ V
q
) ≤ λ(X ¸ (V
q
∪ W
q
)) = 1 −λV
q
−λW
q
=
˜
λ(X ¸ (G
q
∪ H
q
)) = 0.
Because λ is complete, G
q
¸ V
q
and G
q
belong to Λ. But now, if α ∈ R,
¦x : f(x) < α¦ =
q∈Q,q<α
G
q
∈ Λ,
so f is Λmeasurable.
(b) It follows that every zero set belongs to Λ, so that Λ must include the Baire σalgebra of X.
417V Proposition Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological
measure spaces, and (X Y, Λ, λ) their c.l.d. product. Then every continuous function f : X Y → R is
Λmeasurable, and the Baire σalgebra of X Y is included in Λ.
proof Let Z ⊆ X Y be a zero set. If E ∈ Σ, F ∈ T are sets of ﬁnite measure, then Z ∩(E F) is a zero
set for the relative topology of E F. Now the subspace measures µ
E
and ν
F
are τadditive topological
measures (414K), so Z ∩ (E F) is measured by the c.l.d. product µ
E
ν
F
of µ
E
and ν
F
. PPP If either µE
or νF is zero, this is trivial. Otherwise, they have scalar multiples µ
t
E
, ν
t
F
which are probability measures,
and of course are still τadditive topological measures. By 417U, Z∩(EF) is measured by µ
t
E
ν
t
F
. Since
µ
E
ν
F
is just a scalar multiple of µ
t
E
ν
t
F
, Z ∩ (E F) is measured by µ
E
ν
F
. QQQ But µ
E
ν
F
is the
subspace measure λ
E·F
(251P), so Z ∩ (E F) ∈ Λ. As E and F are arbitrary, Z ∈ Λ (251H).
Thus every zero set belongs to Λ; accordingly Λ must include the Baire σalgebra, and every continuous
function must be Λmeasurable.
417X Basic exercises (a) Let (X, Σ, µ) be a semiﬁnite measure space and / a family of subsets of X.
Show that the following are equiveridical: (i) there is a measure µ
t
on X, extending µ, such that µ
t
A = 0
for every A ∈ /; (ii) µ
∗
(
n∈N
A
n
) = 0 for every sequence ¸A
n
)
n∈N
in /.
(b) Let (X, Σ, µ) and (Y, T, ν) be measure spaces with topologies with respect to which µ and ν are
locally ﬁnite. Show that the c.l.d. product measure on X Y is locally ﬁnite.
>>>(c) Let (X, T, Σ, µ) and (Y, S, T, ν) be topological measure spaces such that µ and ν are both eﬀectively
locally ﬁnite τadditive Borel measures. Show that there is a unique eﬀectively locally ﬁnite τadditive Borel
measure λ
t
on X Y such that λ
t
(GH) = µG νH for all open sets G ⊆ X, H ⊆ Y .
>>>(d) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of topological probability spaces in which every µ
i
is a τ
additive Borel measure. Show that there is a unique τadditive Borel measure λ
t
on X =
i∈I
X
i
such that
λ
t
(
i∈I
F
i
) =
i∈I
µ
i
F
i
whenever F
i
⊆ X
i
is closed for every i ∈ I.
106 Topologies and Measures 417Xe
(e) Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological measure spaces in
which the measures are inner regular with respect to the Borel sets, and
˜
λ the τadditive product measure
on X Y . Let ¸X
i
)
i∈I
, ¸Y
j
)
j∈J
be decompositions for µ, ν respectively (deﬁnition: 211E). Show that
¸X
i
Y
j
)
i∈I,j∈J
is a decomposition for
˜
λ. (Cf. 251N.)
>>>(f ) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that µ
i
is inner
regular with respect to the Borel sets for every i, and
˜
λ the τadditive product measure on X =
i∈I
X
i
.
Take A
i
⊆ X
i
for each i ∈ I. (i) Show that if µ
∗
i
A
i
= 1 for every i, then the subspace measure induced by
˜
λ
on A =
i∈I
A
i
is just the τadditive product
˜
λ
#
of the subspace measures on the A
i
. (Hint: show that if
we set λ
t
W =
˜
λ
#
(W ∩ A) for Borel sets W ⊆ X, then λ
t
satisﬁes the conditions of 417Xd.) (ii) Show that
in any case
˜
λ
∗
A =
i∈I
µ
∗
i
A
i
. (Cf. 254L.)
(g) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
and ¸(Y
i
, S
i
, T
i
, ν
i
))
i∈I
be two families of τadditive topological probability
spaces in which every µ
i
and every ν
i
is inner regular with respect to the Borel sets. Let
˜
λ,
˜
λ
t
be the
τadditive product measures on X =
i∈I
X
i
and Y =
i∈I
Y
i
respectively. Suppose that for each i ∈ I we
are given a continuous inversemeasurepreserving function φ
i
: X
i
→ Y
i
. Show that the function φ : X → Y
deﬁned by setting φ(x)(i) = φ
i
(x(i)) for x ∈ X, i ∈ I is inversemeasurepreserving.
(h) Let (X, T, Σ, µ) and (Y, S, T, ν) be two complete locally determined eﬀectively locally ﬁnite τadditive
topological measure spaces such that both µ and ν are inner regular with respect to the Borel sets. Let
˜
λ
be the τadditive product measure on X Y , and
˜
Λ its domain. Suppose that ν is σﬁnite. Show that for
any W ∈
˜
Λ, W[¦x¦] ∈ T for almost every x ∈ X, and x → νW[¦x¦] is measurable.
>>>(i) Let (X, T, Σ, µ) be [0, 1] with its usual topology and Lebesgue measure, and let (Y, S, T, ν) be [0, 1]
with the discrete topology and counting measure. (i) Show that both are Radon measure spaces. (ii) Show
that the c.l.d. product measure on X Y is a Radon measure. (Hint: 252Kc, or use 417T and 417P.) (iii)
Show that 417Ha can fail if we omit the hypothesis on ¦(x, y) : f(x, y) ,= 0¦.
(j) Let (X, T, Σ, µ) and (Y, S, T, ν) be two eﬀectively locally ﬁnite τadditive topological measure spaces.
Let λ be the c.l.d. product measure and
˜
λ the τadditive product measure on XY . Show that λ
∗
(AB) =
˜
λ
∗
(AB) for all sets A ⊆ X, B ⊆ Y . (Hint: start with A, B of ﬁnite outer measure, so that 417I applies.)
(k) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces with strictly positive
measures all inner regular with respect to the Borel sets, and (X, T,
˜
Λ,
˜
λ) their τadditive product. For
J ⊆ I let
˜
λ
J
be the τadditive product measure on X
J
=
i∈J
X
i
, and
˜
Λ
J
its domain. (i) Show that if f is
a realvalued
˜
Λmeasurable function deﬁned
˜
λalmost everywhere on X, we can ﬁnd a countable set J ⊆ I
and a
˜
Λ
J
measurable function g, deﬁned
˜
λ
J
almost everywhere on X
J
, such that f extends gπ
J
. (ii) In (i),
show that
_
fd
˜
λ =
_
g d
˜
λ
J
if either is deﬁned in [−∞, ∞].
(l) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that every µ
i
is
inner regular with respect to the Borel sets, and (X, T,
˜
Λ,
˜
λ) their τadditive product. Show that for any
W ∈
˜
Λ there is a smallest set J ⊆ I for which there is a W
t
∈
˜
Λ, determined by coordinates in J, with
˜
λ(W´W
t
) = 0. (Hint: 254R.)
(m) What needs to be added to 417M and 415Xk to complete a proof of 415E?
(n) Let (X, T, Σ, µ) be an atomless τadditive topological probability space such that µ is inner regular
with respect to the Borel sets, and I a set of cardinal at most that of the support of µ. Show that the set
of injective functions from I to X has full outer measure for the τadditive product measure on X
I
.
>>>(o) Let (X, T, Σ, µ) and (Y, S, T, ν) be Radon measure spaces. Show that the Radon product measure
on XY is the unique Radon measure
˜
λ such that
˜
λ(KL) = µK νL for all compact sets K ⊆ X, L ⊆ Y .
>>>(p) Let I be an uncountable set, and for each i ∈ I let X
i
be ¦0, 1¦, T
i
= T¦0, 1¦ the usual topology,
and µ
i
the measure on Σ
i
= T¦0, 1¦ deﬁned by saying that µ
i
A = 1 if 0 ∈ A, 0 otherwise. Check that
417Yb τadditive product measures 107
(X
i
, T
i
, Σ
i
, µ
i
) is a Radon probability space. Let λ,
˜
λ be the ordinary and τadditive product measures on
X =
i∈I
X
i
= ¦0, 1¦
I
. Show that they are diﬀerent. Show that the support of
˜
λ is not determined by any
countable set of coordinates. Find a
˜
λnegligible open set W ⊆ X such that its projection onto ¦0, 1¦
J
is
conegligible for every countable J ⊆ I.
(q) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of Radon probability spaces, and
˜
λ the quasiRadon product
measure on X =
i∈I
X
i
. For each i ∈ I, let Z
i
⊆ X
i
be the support of µ
i
. Show that
˜
λ is a Radon measure
iﬀ ¦i : i ∈ I, Z
i
is not compact¦ is countable. In particular, show that the ordinary product measure on
[0, 1[
I
, where I is uncountable and each copy of [0, 1[ is given Lebesgue measure, is a quasiRadon measure,
but not a Radon measure.
(r) Let ¸(X
n
, T
n
, Σ
n
, µ
n
))
n∈N
be a sequence of Radon probability spaces. Show that the Radon product
measure on X =
n∈N
X
n
is the unique Radon measure
˜
λ on X such that
˜
λ(
n∈N
K
n
) =
∞
n=0
µ
n
K
n
whenever K
n
⊆ X
n
is compact for every n.
(s) Let (X, T, Σ, ν) and (Y, S, T, ν) be two topological measure spaces, of which (X, T) has a countable
network. (i) Show that the c.l.d. product measure λ on X Y is a topological measure. (ii) Show directly,
without relying on ideas from 417D, that λ is τadditive if ν is.
(t) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that every µ
i
is inner regular with respect to the Borel sets, and λ,
˜
λ the ordinary and τadditive product measures on
X =
i∈I
X
i
. Show that if A ⊆ X has
˜
λnegligible boundary, then A is measured by λ.
(u) Let us say that a topological space X is chargeable if there is an additive functional ν : TX → [0, ∞[
such that νG > 0 for every nonempty open set G ⊆ X. (i) Show that if there is a σﬁnite measure µ on
X such that µ
∗
G > 0 for every nonempty open set G, then X is chargeable. (Hint: 215B(vii), 391G.) (ii)
Show that any separable space is chargeable. (iii) Show that X is chargeable iﬀ its regular open algebra
is chargeable in the sense of 391X. (Hint: see the proof of 314P.) (iv) Show that any open subspace of a
chargeable space is chargeable. (v) Show that if Y ⊆ X is dense, then X is chargeable iﬀ Y is chargeable.
(vi) Show that if X is expressible as the union of countably many chargeable subspaces, then it is chargeable.
(vii) Show that any product of chargeable spaces is chargeable. (Cf. 391Xb(iii).) (viii) Show that if ¸X
i
)
i∈I
is a family of chargeable spaces with product X, then every regular open subset of X and every Baire subset
of X is determined by coordinates in a countable set. (Hint: 4A2Eb, 4A3Mb.)
(v) Let (X, T, Σ, µ) and (Y, S, T, ν) be quasiRadon measure spaces such that µX νY > 0. Show that
the quasiRadon product measure on XY is completion regular iﬀ it is equal to the c.l.d. product measure
and µ and ν are both completion regular. (Hint: 412Sc; if µE, νF are ﬁnite and Z ⊆ E F is a zero set of
positive measure, use Fubini’s theorem to show that Z has sections of positive measure.)
(w) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of quasiRadon probability spaces. Show that the quasiRadon
product measure on
i∈I
X
i
is completion regular iﬀ it is equal to the ordinary product measure and every
µ
i
is completion regular.
(x) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that every µ
i
is
inner regular with respect to the Borel sets, and λ the τadditive product measure on X =
i∈I
X
i
; write
Λ for its domain. (i) Show that if W ∈ Λ, λW > 0 and > 0 then there are a ﬁnite J ⊆ I and a W
t
∈ Λ
such that λW
t
≥ 1 − and for every x ∈ W
t
there is a y ∈ W such that xI ¸ J = yI ¸ J. (Cf. 254Sb.) (ii)
Show that if A ⊆ X is determined by coordinates in I ¸ ¦i¦ for every i ∈ I then λ
∗
A ∈ ¦0, 1¦. (Cf. 254Sa.)
417Y Further exercises (a) Give an example to show that, in 417A, λ can be strictly localizable while
λ
t
is not.
(b) Let (X, T, Σ, µ) and (Y, S, T, ν) be eﬀectively locally ﬁnite τadditive topological measure spaces such
that µ and ν are both inner regular with respect to the Borel sets.
(i) Fix open sets G ⊆ X, H ⊆ Y of ﬁnite measure. Let J
GH
be the set of those W ⊆ X Y such
that θ
GH
(W) =
_
G
ˆ ν(W[¦x¦] ∩ H)dx is deﬁned, where ˆ ν is the completion of ν. (α) Show that every
108 Topologies and Measures 417Yb
open set belongs to J
GH
. (β) Show that θ
GH
is countably additive in the sense that θ
GH
(
n∈N
W
n
) =
∞
n=0
θ
GH
(W
n
) for every disjoint sequence ¸W
n
)
n∈N
in J
GH
, and τadditive in the sense that θ
GH
(
1) =
sup
V ∈\
θ
GH
(V ) for every nonempty upwardsdirected family 1 of open sets in X Y . (γ) Show that
every Borel set belongs to J
GH
. (Hint: Monotone Class Theorem.) (δ) Writing B for the Borel σalgebra
of X Y , show that θ
GH
B is a τadditive Borel measure; let λ
GH
be its completion. () Show that
λ
GH
= θ
GH
Λ
GH
, where Λ
GH
= domλ
GH
. (ζ) Show that λ
GH
(E F) is deﬁned and equal to µE νF
whenever E ∈ Σ, F ∈ T, E ⊆ G and F ⊆ H. (Hint: start with open E and F, move to Borel E and F with
the Monotone Class Theorem.) (η) Writing λ for the c.l.d. product measure on X Y , show that λ
GH
(W)
is deﬁned and equal to λ(W ∩ (GH)) whenever W ∈ domλ.
(ii) Now take
˜
Λ to be
¦Λ
GH
: G ∈ T, H ∈ S, µG < ∞, νH < ∞¦ and
˜
λW = sup
G,H
λ
GH
(W) for
W ∈
˜
Λ. Show that
˜
λ is an extension of λ to a complete locally determined eﬀectively locally ﬁnite τadditive
topological measure on X Y which is inner regular with respect to the Borel sets, so is the τadditive
product measure as deﬁned in 417G.
(c) Let (X, Σ, µ) and (Y, T, ν) be complete measure spaces with topologies T, S. Suppose that µ and
ν are eﬀectively locally ﬁnite and τadditive and moreover that their domains include bases for the two
topologies. Show that the c.l.d. product measure on X Y has the same properties. (Hint: start by
assuming that µX and νY are both ﬁnite. If 1 is an upwardsdirected family of measurable open sets with
measurable open union W, look at g
V
(x) = νV [¦x¦] for V ∈ 1.)
(d) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of τadditive topological probability spaces such that every µ
i
is inner regular with respect to the Borel sets, and (X, T,
˜
Λ,
˜
λ) their τadditive product. (i) Show that the
following are equiveridical: (α) µ
i
is strictly positive for all but countably many i ∈ I; (β) whenever W ∈
˜
Λ
there are a countable J ⊆ I and W
1
, W
2
∈
˜
Λ, determined by coordinates in J, such that W
1
⊆ W ⊆ W
2
and
˜
λ(W
2
¸ W
1
) = 0. (ii) Show that when these are false,
˜
λ cannot be equal to the ordinary product measure
on X.
(e) Let (X, Σ, µ) and (Y, T, ν) be measure spaces with topologies T, S such that both µ and ν are inner
regular with respect to the families of sequentially compact sets in each space. Show that the c.l.d. product
measure λ on X Y is also inner regular with respect to the sequentially compact sets, so has an extension
to a topological measure which is inner regular with respect to the sequentially compact sets. (Hint: 412R,
416Yc.)
(f ) Let ¸(X
i
, Σ
i
, µ
i
))
i∈I
be a family of probability spaces with topologies T
i
such that every µ
i
is inner
regular with respect to the family of closed countably compact sets in X
i
and every X
i
is compact. Show that
the ordinary product measure λ on X =
i∈I
X
i
is also inner regular with respect to the closed countably
compact sets, so has an extension to a topological measure
˜
λ which is inner regular with respect to the
closed countably compact sets in X. Show that this can be done in such a way that for every W ∈ dom
˜
λ
there is a V ∈ domλ such that
˜
λ(W´V ) = 0. (Hint: 412T, 416Yb.)
(g) Let ¸(X
n
, Σ
n
, µ
n
))
n∈N
be a sequence of probability spaces with topologies T
n
such that every µ
n
is
inner regular with respect to the family of sequentially compact sets in X
n
. Show that the ordinary product
measure λ on X =
n∈N
X
n
is also inner regular with respect to the sequentially compact sets, so has an
extension to a topological measure
˜
λ which is inner regular with respect to the sequentially compact sets in
X. Show that this can be done in such a way that for every W ∈ dom
˜
λ there is a V ∈ domλ such that
˜
λ(W´V ) = 0.
(h) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
be a family of quasiRadon probability spaces, and λ,
˜
λ the ordinary and
quasiRadon product measures on X =
i∈I
X
i
. Suppose that all but one of the T
i
have countable networks
and all but countably many of the µ
i
are strictly positive. Show that λ =
˜
λ.
(i) Let us say that a quasiRadon measure space (X, T, Σ, µ) has the simple product property if
the c.l.d. product measure on X Y is equal to the quasiRadon product measure for every quasiRadon
measure space (Y, S, T, ν). (i) Show that if (X, T) has a countable network then (X, T, Σ, µ) has the simple
417 Notes τadditive product measures 109
product property. (ii) Show that if a quasiRadon measure space has the simple product property so do all
its subspaces. (iii) Show that the quasiRadon product of two quasiRadon measure spaces with the simple
product property has the simple product property. (iv) Show that the quasiRadon product of any family
of quasiRadon probability spaces with the simple product property has the simple product property. (v)
Show that the real line with the rightfacing Sorgenfrey topology (415Xc) and Lebesgue measure has the
simple product property.
417 Notes and comments The general problem of determining just when a measure can be extended to
a measure with given properties is one which will recur throughout this volume. I have more than once
mentioned the BanachUlam problem; if you like, this is the question of whether there can ever be an
extension of the countablecocountable measure on a set X to a measure deﬁned on the whole algebra TX.
This particular question appears to be undecidable from the ordinary axioms of set theory; but for many
sets (for instance, if X = ω
1
) it is known that the answer is ‘no’. (See 419G, 438C.) This being so, we have
to take each manifestation of the general question on its own merits. In 417C and 417E the challenge is
to take a product measure λ deﬁned in terms of the factor measures alone, disregarding their topological
properties, and extend it to a topological measure, preferably τadditive. Of course there are important
cases in which λ is itself already a topological measure; for instance, we know that the c.l.d. product of
Lebesgue measure on R with itself is Lebesgue measure on R
2
(251M), and other examples are in 415E,
415Ye, 416U, 417S417T and 417Yk. But in general not every open set in the product belongs to the domain
of λ, even when we have the product of two Radon measures on compact Hausdorﬀ spaces (419E).
Once we have resolved to grasp the nettle, however, there is a natural strategy for the proof. It is easy
to see that if λ, in 417C or 417E, is to have an extension to a τadditive topological measure
˜
λ, then we
must have
˜
λA(1) = 0 for every 1 belonging to the class V. Now 417A descibes a suﬃcient (and obviously
necessary) condition for there to be an extension of λ with this property. So all we have to do is check.
The check is not perfectly straightforward; in 417E it uses all the resources of the original proof that there
is a product measure on an arbitrary product of probability spaces (which I suppose is to be expected),
with 414B (of course) to apply the hypothesis that the factor measures are τadditive, and a couple of extra
wrinkles (the W
t
n
and C
t
n
of part (cii) of the proof of 417E, and the use of supports in part (cv)).
It is worth noting that (both for ﬁnite and for inﬁnite products) the measure algebras of λ and
˜
λ are
identical (417C(ii), 417E(ii)), so there is no new work to do in identifying the measure algebra of
˜
λ and the
associated function spaces.
An obstacle we face in 417C417E is the fact that not every τadditive measure µ has an extension to a τ
additive topological measure, even when µ is totally ﬁnite and its domain includes a base for the topology. (I
give an example in 419H.) Consequently it is not enough, in 417C or 417E, to show that the ordinary product
measure λ is τadditive. But perhaps I should remark that if λ is inner regular with respect to the closed
sets, this obstacle evaporates (415L). Accordingly, for the principal applications (to quasiRadon and Radon
product measures, and in particular whenever the topological spaces involved are regular) we have rather
easier proofs available, based on the constructions of ¸415. For completely regular spaces, there is yet another
approach, because the product measures can be described in terms of the integrals of continuous functions
(415I), which by 417U and 417V can be calculated from the ordinary product measures. Of course the proof
that λ itself is τadditive is by no means trivial, especially in the case of inﬁnite products, corresponding
to 417E; but for ﬁnite products there are relatively direct arguments, applying indeed to slightly more
general situations (417Yc). If we have measures which are inner regular with respect to countably compact
classes of sets, then there may be other ways of approaching the extension, using theorems from ¸413 (see
417Ye417Yg), and for compact Radon measure spaces, λ becomes tight (412Sb, 412V), so its τadditivity
is elementary.
As always, it is important to recognise which constructions are in some sense canonical. The arguments
of 417C and 417E allow for the possibility that the factor measures are deﬁned on σalgebras going well
beyond the Borel sets. For all the principal applications, however, the measures will be c.l.d. versions of
Borel measures, and in particular will be inner regular with respect to the Borel sets. In such a context it
is natural to ask for product measures with the same property, and in this case we can identify a canonical
τadditive topological product measure, as in 417D and 417F. (If you prefer to restrict your measures to
Borel σalgebras, you again get canonical product Borel measures (417Xc417Xd).) Having done so, we
110 Topologies and measures I 417 Notes
can reasonably expect ‘commutative’ and ‘associative’ and ‘distributive’ laws, as in 417D, 417J and 417Xe.
Subspaces mostly behave themselves (417I, 417Xf).
Of course extending the product measure means that we get new integrable functions on the product, so
that Fubini’s theorem has to be renegotiated. Happily, it remains valid, at least in the contexts in which
it was eﬀective before (417Ha); we still need, in eﬀect, one of the measures to be σﬁnite. The theorem
still fails for arbitrary integrable functions on products of Radon measure spaces, and the same example
works as before (417Xi). In fact this means that we have an alternative route to the construction of the
τadditive product of two measures (417Yb). But note that on this route ‘commutativity’, the identiﬁcation
of the product measure on X Y with that on Y X, becomes something which can no longer be taken
for granted, because if we deﬁne
˜
λW to be
_
νW[¦x¦]dx we have to worry about when, and why, this will
be equal to
_
µW
−1
[¦y¦]dy.
A version of Tonelli’s theorem follows from Fubini’s theorem, as before (417Hc). We also have results
corresponding to most of the theorems of ¸254. But note that there are two traps. In the theorem that
a measurable set can be described in terms of a projection onto a countable subproduct (254O, 417M) we
need to suppose that the factor measures are strictly positive, and in the theorem that a product of Radon
measures is a Radon measure (417Q) we need to suppose that the factor measures have compact supports.
The basic examples to note in this context are 417Xp and 417Xq.
It is not well understood when we can expect c.l.d. product measures to be topological measures, even in
the case of compact Radon probability spaces. Example 419E remains a rather special case, but of course
much more eﬀort has gone into seeking positive results. Note that the ordinary product measures of this
section are always eﬀectively locally ﬁnite and τadditive (417C, 417E), so that they will be equal to the τ
additive products iﬀ they measure every open set (417S). Regarding inﬁnite products, the τadditive product
measure can fail to be the ordinary product measure in just two ways: if one of the ﬁnite product measures
is not a topological measure, or if uncountably many of the factor measures are not strictly positive (417Sc,
417Xp, 417Yd). So it is ﬁnite products which need to be studied.
Whenever we have a subset F of an inﬁnite product X =
i∈I
X
i
, it is important to know when F is
determined by coordinates in a proper subset of I; in measure theory, we are particularly interested in sets
determined by coordinates in countable subsets of I (254Mb). It may happen that there is a smallest set
J such that F is determined by coordinates in J; for instance, when we have a topological product and F
is closed (4A2Bg). When we have a product of probability spaces, we sometimes wish to identify sets J
such that F is ‘essentially’ determined by coordinates in J, in the sense that there is an F
t
, determined by
coordinates in J, such that F´F
t
is negligible. In this context, again, there is a smallest such set (254Rd),
which can be identiﬁed in terms of the probability algebra free product of the measure algebras (325Mb).
In 417Ma the two ideas come together: under the conditions there, we get the same smallest J by either
route.
In 417Ma, we have a product of strictly positive τadditive topological probability measures. If we keep
the ‘strictly positive’ but abandon everything else, we still have very striking results just because the product
topology is ccc, so that we can apply 4A2Eb. An abstract expression of this idea is in 417Xu.
418 Measurable functions and almost continuous functions
In this section I work through the basic properties of measurable and almost continuous functions, as
deﬁned in 411L and 411M. I give the results in the full generality allowed by the terminology so far introduced,
but most of the ideas are already required even if you are interested only in Radon measure spaces as the
domains of the functions involved. Concerning the codomains, however, there is a great diﬀerence between
metrizable spaces and others, and among metrizable spaces separability is of essential importance.
I start with the elementary properties of measurable functions (418A418C) and almost continuous func
tions (418D). Under mild conditions on the domain space, almost continuous functions are measurable
(418E); for a separable metrizable codomain, we can expect that measurable functions should be almost
continuous (418J). Before coming to this, I spend a couple of paragraphs on image measures: a locally
ﬁnite image measure under a measurable function is Radon if the measure on the domain is Radon and the
function is almost continuous (418I).
418B Measurable functions and almost continuous functions 111
418L418Q are important results on expressing given Radon measures as image measures associated with
continuous functions, ﬁrst dealing with ordinary functions f : X → Y (418L) and then coming to Prokhorov’s
theorem on projective limits of probability spaces (418M).
The machinery of the ﬁrst part of the section can also be used to investigate representations of vector
valued functions in terms of product spaces (418R418T).
418A Proposition Let X be a set, Σ a σalgebra of subsets of X, Y a topological space and f : X → Y
a measurable function.
(a) f
−1
[F] ∈ Σ for every Borel set F ⊆ Y .
(b) If A ⊆ X is any set, endowed with the subspace σalgebra, then fA : A → Y is measurable.
(c) Let (Z, T) be another topological space. Then gf : X → Z is measurable for every Borel measurable
function g : Y → Z; in particular, for every continuous function g : Y → Z.
proof (a) The set ¦F : F ⊆ Y, f
−1
[F] ∈ Σ¦ is a σalgebra of subsets of Y containing every open set, so
contains every Borel subset of Y .
(b) is obvious from the deﬁnition of ‘subspace σalgebra’ (121A).
(c) If H ⊆ Z is open, then g
−1
[H] is a Borel subset of Y so (gf)
−1
[H] = f
−1
[g
−1
[H]] belongs to Σ.
418B Proposition Let X be a set and Σ a σalgebra of subsets of X.
(a) If Y is a metrizable space and ¸f
n
)
n∈N
is a sequence of measurable functions from X to Y such that
f(x) = lim
n→∞
f
n
(x) is deﬁned in Y for every x ∈ X, then f : X → Y is measurable.
(b) If Y is a topological space, Z is a separable metrizable space and f : X → Y , g : X → Z are functions,
then x → (f(x), g(x)) : X → Y Z is measurable iﬀ f and g are measurable.
(c) If Y is a hereditarily Lindel¨of space,  a family of open sets generating its topology, and f : X → Y
a function such that f
−1
[U] ∈ Σ for every U ∈ , then f is measurable.
(d) If ¸Y
i
)
i∈I
is a countable family of separable metrizable spaces, with product Y , then a function
f : X → Y is measurable iﬀ π
i
f : X → Y
i
is measurable for every i, writing π
i
(y) = y(i) for y ∈ Y , i ∈ N.
proof (a) Let ρ be a metric deﬁning the topology of Y . Let G ⊆ Y be any open set, and for each n ∈ N set
F
n
= ¦y : y ∈ Y, ρ(y, z) ≥ 2
−n
for every z ∈ Y ¸ G¦.
Then F
n
is closed, so f
−1
i
[F
n
] ∈ Σ for every n, i ∈ N. But this means that
f
−1
[G] =
n∈N
i≥n
f
−1
i
[F
i
] ∈ Σ.
As G is arbitrary, f is measurable.
(b)(i) The functions (y, z) → y, (y, z) → z are continuous, so if x → (f(x), g(x)) is measurable, so are f
and g, by 418Ac.
(ii) Now suppose that f and g are measurable, and that W ⊆ Y Z is open. By 4A2P(ai), the
topology of Z has a countable base H; let ¸H
n
)
n∈N
be a sequence running over H∪ ¦∅¦. For each n, set
G
n
=
¦G : G ⊆ Y is open, GH
n
⊆ W¦;
then G
n
is open and G
n
H
n
⊆ W. Accordingly W ⊇
n∈N
G
n
H
n
. But in fact W =
n∈N
G
n
H
n
.
PPP If (y, z) ∈ W, there are open sets G ⊆ Y , H ⊆ Z such that (y, z) ∈ GH ⊆ W. Now there is an n ∈ N
such that z ∈ H
n
⊆ H, in which case GH
n
⊆ W and G ⊆ G
n
and (y, z) ∈ G
n
H
n
. QQQ
Accordingly
¦x : (f(x), g(x)) ∈ W¦ =
n∈N
f
−1
[G
n
] ∩ g
−1
[H
n
] ∈ Σ.
As W is arbitrary, x → (f(x), g(x)) is measurable.
(c) This is just 4A3Db.
(d) If f is measurable, so is every π
i
f, by 418Ac. If every π
i
f is measurable, set
 = ¦π
−1
i
[H] : i ∈ I, H ⊆ Y
i
is open¦.
Then  generates the topology of Y , and if U = π
−1
i
[H] then f
−1
[U] = (π
i
f)
−1
[H], so f
−1
[U] ∈ Σ for every
U. Also Y is hereditarily Lindel¨of (4A2P(aiii)), so f is measurable, by (c).
112 Topologies and measures I 418C
418C Proposition Let (X, Σ, µ) be a measure space and Y a Polish space. Let ¸f
n
)
n∈N
be a sequence
of measurable functions from X to Y . Then
¦x : x ∈ X, lim
n→∞
f
n
(x) is deﬁned in Y ¦
belongs to Σ.
proof (Compare 121H.) Let ρ be a complete metric on Y deﬁning the topology of Y .
(a) For m, n ∈ N and δ > 0, the set ¦x : ρ(f
m
(x), f
n
(x)) ≤ δ¦ belongs to Σ. PPP The function x →
(f
m
(x), f
n
(x)) : X → Y
2
is measurable, by 418Bb, and the function ρ : Y
2
→ R is continuous, so x →
ρ(f
m
(x), f
n
(x)) is measurable and ¦x : ρ(f
m
(x), f
n
(x)) ≤ δ¦ ∈ Σ. QQQ
(b) Now ¸f
n
(x))
n∈N
is convergent iﬀ it is Cauchy, because Y is complete. But
¦x : x ∈ X, ¸f
n
(x))
n∈N
is Cauchy¦ =
n∈N
_
m∈N
i≥m
¦x : ρ(f
i
(x), f
m
(x)) ≤ 2
−n
¦
belongs to Σ.
418D Proposition Let (X, Σ, µ) be a measure space and T a topology on X.
(a) Suppose that Y is a topological space. Then any continuous function from X to Y is almost continuous.
(b) Suppose that Y and Z are topological spaces, f : X → Y is almost continuous and g : Y → Z is
continuous. Then gf : X → Z is almost continuous.
(c) Suppose that (Y, S, T, ν) is a σﬁnite topological measure space, Z is a topological space, g : Y → Z is
almost continuous and f : X → Y is inversemeasurepreserving and almost continuous. Then gf : X → Z
is almost continuous.
(d) Suppose that µ is semiﬁnite, and that ¸Y
i
)
i∈I
is a countable family of topological spaces with product
Y . Then a function f : X → Y is almost continuous iﬀ f
i
= π
i
f is almost continuous for every i ∈ I, writing
π
i
(y) = y(i) for i ∈ I, y ∈ Y .
proof (a) is trivial.
(b) The set ¦A : A ⊆ X, gfA is continuous¦ includes ¦A : A ⊆ X, fA is continuous¦; so if µ is inner
regular with respect to the latter, it is inner regular with respect to the former.
(c) Take E ∈ Σ and γ < µE; take > 0. We have a cover of Y by a nondecreasing sequence ¸Y
n
)
n∈N
of measurable sets of ﬁnite measure; now ¸f
−1
[Y
n
])
n∈N
is a nondecreasing sequence covering E, so there is
an n ∈ N such that µ(E ∩ f
−1
[Y
n
]) ≥ γ. Because f is inversemeasurepreserving, E ∩ f
−1
[Y
n
] has ﬁnite
measure. Now we can ﬁnd measurable sets F ⊆ Y
n
, E
1
⊆ E ∩ f
−1
[Y
n
] such that fE
1
, gF are continuous
and νF ≥ νY
n
− , µE
1
≥ µ(E ∩ f
−1
[Y
n
] ¸ E
1
) − . In this case E
0
= E
1
∩ f
−1
[F] has measure at least
γ −2 and gfE
0
is continuous. As E, γ and are arbitrary, gf is almost continuous.
(d)(i) If f is almost continuous, every f
i
must be almost continuous, by (b).
(ii) Now suppose that every f
i
is almost continuous. Take E ∈ Σ and γ < µE. There is an E
0
⊆ E
such that E
0
∈ Σ and γ < µE
0
< ∞. Let ¸
i
)
i∈I
be a family of strictly positive real numbers such that
i∈I
i
≤ µE
0
−γ. For each i ∈ I choose a measurable set F
i
⊆ E
0
such that µF
i
≥ µE
0
−
i
and f
i
F
i
is
continuous. Then F = E
0
∩
i∈I
F
i
is a subset of E with measure at least γ, and fF is continuous because
f
i
F is continuous for every i (3A3Ib).
418E Theorem Let (X, T, Σ, µ) be a complete locally determined topological measure space, Y a topo
logical space, and f : X → Y an almost continuous function. Then f is measurable.
proof Set / = ¦K : K ∈ Σ, fK is continuous¦; then µ is inner regular with respect to /. If H ⊆ Y is
open and K ∈ /, then K ∩ f
−1
[H] is relatively open in K, that is, there is an open set G ⊆ X such that
K ∩ f
−1
[H] = K ∩ G. Because µ is a topological measure, G ∈ Σ so K ∩ f
−1
[H] ∈ Σ. As K is arbitrary,
and µ is complete and locally determined, f
−1
[H] ∈ Σ (412Ja). As H is arbitrary, f is measurable.
418H Measurable functions and almost continuous functions 113
418F Proposition Let (X, T, Σ, µ) be a semiﬁnite topological measure space, Y a metrizable space,
and f : X → Y a function. Suppose there is a sequence ¸f
n
)
n∈N
of almost continuous functions from X to
Y such that f(x) = lim
n→∞
f
n
(x) for almost every x ∈ X. Then f is almost continuous.
proof Suppose that E ∈ Σ and that γ < µE, > 0. Then there is a measurable set F ⊆ E such that
γ ≤ µF < ∞; discarding a negligible set if necessary, we may arrange that f(x) = lim
n→∞
f
n
(x) for every
x ∈ F. Let ρ be a metric on Y deﬁning its topology. For each n ∈ N, let F
n
⊆ F be a measurable set
such that f
n
F
n
is continuous and µ(F
n
¸ F) ≤ 2
−n
; set G =
n∈N
F
n
, so that µG ≥ γ − 2 and f
n
G is
continuous for every n ∈ N.
For m, n ∈ N, the functions x → (f
m
(x), f
n
(x)) : G → Y
2
and x → ρ(f
m
(x), f
n
(x)) : G → R are
continuous, therefore measurable, because µ is a topological measure. Also ¸f
n
(x))
n∈N
is a Cauchy sequence
for every x ∈ G. So if we set G
kn
= ¦x : x ∈ G, ρ(f
i
(x), f
j
(x)) ≤ 2
−k
for all i, j ≥ n¦, ¸G
kn
)
n∈N
is a non
decreasing sequence of measurable sets with union G for each k ∈ N, and we can ﬁnd a strictly increasing
sequence ¸n
k
)
k∈N
such that µ(G¸G
kn
k
) ≤ 2
−k
for every k. Setting H =
k∈N
G
kn
k
, µH ≥ µG−2 ≥ γ−4
and ρ(f
i
(x), f
n
k
(x)) ≤ 2
−k
whenever x ∈ H and i ≥ n
k
; consequently ρ(f(x), f
n
k
(x)) ≤ 2
−k
whenever x ∈ H
and k ∈ N. But this means that ¸f
n
k
)
k∈N
converges to f uniformly on H, while every f
n
k
is continuous on
H, so fH is continuous (3A3Nb). And of course H ⊆ E.
As E, γ and are arbitrary, f is almost continuous.
418G Proposition Let (X, T, Σ, µ) be a σﬁnite quasiRadon measure space, Y a metrizable space and
f : X → Y an almost continuous function. Then there is a conegligible set X
0
⊆ X such that f[X
0
] is
separable.
proof (a) Let / be the family of selfsupporting measurable sets K of ﬁnite measure such that fK is
continuous. Then µ is inner regular with respect to /. PPP If E ∈ Σ and γ < µE, there is an F ∈ Σ such
that F ⊆ E and γ < µF < ∞; there is an H ∈ Σ such that H ⊆ F, γ ≤ µH and fH is continuous; and
there is a measurable selfsupporting K ⊆ H with the same measure as H (414F), in which case K ∈ / and
K ⊆ E and µK ≥ γ. QQQ
(b) Now f[K] is ccc for every K ∈ /. PPP If ( is a disjoint family of nonempty relatively open subsets
of f[K], then ¸K∩f
−1
[G])
G∈Ç
is a disjoint family of nonempty relatively open subsets of K, because fK
is continuous, and
G∈Ç
µ(K ∩ f
−1
[G]) ≤ µK. Because K is selfsupporting, µ(K ∩ f
−1
[G]) > 0 for every
G ∈ (; because µK is ﬁnite, ( is countable. As ( is arbitrary, f[K] is ccc. QQQ
Because Y is metrizable, f[K] must be separable (4A2Pd).
(c) Because µ is σﬁnite, there is a countable family L ⊆ / such that X
0
=
L is conegligible (412Ic).
Now f[X
0
] =
L∈1
f[L] is a countable union of separable spaces, so is separable (4A2B(ei)).
418H Proposition (a) Let X and Y be topological spaces, µ an eﬀectively locally ﬁnite τadditive
measure on X, and f : X → Y an almost continuous function. Then the image measure µf
−1
is τadditive.
(b) Let (X, T, Σ, µ) be a totally ﬁnite quasiRadon measure space, (Y, S) a regular topological space, and
f : X → Y an almost continuous function. Then there is a unique quasiRadon measure ν on Y such that
f is inversemeasurepreserving for µ and ν.
proof (a) Let H be an upwardsdirected family of open subsets of Y , all measured by µf
−1
, and suppose
that H
∗
=
H is also measurable. Take any γ < (µf
−1
)(H
∗
) = µf
−1
[H
∗
]. Then there is a measurable
set E ⊆ f
−1
[H
∗
] such that µE ≥ γ and fE is continuous. Consider ¦E ∩ f
−1
[H] : H ∈ H¦. This is an
upwardsdirected family of relatively open measurable subsets of E with measurable union E. By 414K, the
subspace measure on E is τadditive, so
γ ≤ µE ≤ sup
H∈1
µ(E ∩ f
−1
[H]) ≤ sup
H∈1
µf
−1
[H].
As γ is arbitrary, µf
−1
[H
∗
] ≤ sup
H∈1
µf
−1
[H]; as H is arbitrary, µf
−1
is τadditive.
(b) By 418E, f is measurable. Let ν
0
be the restriction of µf
−1
to the Borel σalgebra of Y ; by (a),
ν
0
is τadditive, and f is inversemeasurepreserving with respect to µ and ν
0
. Because Y is regular, the
completion ν of ν
0
is a quasiRadon measure (415Cb). Because µ is complete, f is still inversemeasure
preserving with respect to µ and ν (235Hc).
114 Topologies and measures I 418H
To see that ν is unique, observe that its values on Borel sets are determined by the requirement that f
be inversemeasurepreserving, so that 415H gives the result.
418I The next theorem is one of the central properties of Radon measures. I have already presented
what amounts to a special case in 256G.
Theorem Let (X, T, Σ, µ) be a Radon measure space, Y a Hausdorﬀ space, and f : X → Y an almost
continuous function. If the image measure ν = µf
−1
is locally ﬁnite, it is a Radon measure.
proof (a) By 418E, f is measurable, that is, f
−1
[H] ∈ Σ for every open set H ⊆ Y ; but this means that
the domain T of ν contains every open set, and ν is a topological measure.
(b) ν is tight (that is, inner regular with respect to the compact sets). PPP If F ∈ T and νF > 0, then
µf
−1
[F] > 0, so there is an E ⊆ f
−1
[F] such that µE > 0 and fE is continuous. Next, there is a compact
set K ⊆ E such that µK > 0. In this case, L = f[K] is a compact subset of F, and
νL = µf
−1
[L] ≥ µK > 0.
By 412B, this is enough to prove that ν is tight. QQQ Note that because ν is locally ﬁnite, νL < ∞ for every
compact L ⊆ Y (411Ga).
(c) Because µ is complete, so is ν (212Bd). Next, ν is locally determined. PPP Suppose that H ⊆ Y is
such that H ∩ F ∈ T whenever νF < ∞. Then, in particular, H ∩ f[K] ∈ T whenever K ⊆ X is compact
and fK is continuous. But setting
/ = ¦K : K ⊆ X is compact, fK is continuous¦,
µ is inner regular with respect to / (412Ac). And if K ∈ /,
K ∩ f
−1
[H] = K ∩ f
−1
[H ∩ f[K]] ∈ Σ.
Because µ is complete and locally determined, this is enough to show that f
−1
[H] ∈ Σ (412Ja), that is,
H ∈ T. As H is arbitrary, ν is locally determined. QQQ
(d) Thus ν is a complete locally determined tight locally ﬁnite topological measure; that is, it is a Radon
measure.
418J Theorem Let (X, Σ, µ) be a semiﬁnite measure space and T a topology on X such that µ is inner
regular with respect to the closed sets. Suppose that Y is a separable metrizable space and f : X → Y is
measurable. Then f is almost continuous.
proof Let H be a countable base for the topology of Y , and ¸H
n
)
n∈N
a sequence running over H ∪ ¦∅¦.
Take E ∈ Σ and γ < µE. Choose ¸E
n
)
n∈N
inductively, as follows. There is an E
0
∈ Σ such that E
0
⊆ E
and γ < µE
0
< ∞. Given E
n
∈ Σ with γ < µE
n
< ∞, E
n
¸ f
−1
[H
n
] ∈ Σ, so there is a closed set F
n
∈ Σ
such that
F
n
⊆ E
n
¸ f
−1
[H
n
], µ((E
n
¸ f
−1
[H
n
]) ¸ F
n
) < µE
n
−γ;
set E
n+1
= (E
n
∩ f
−1
[H
n
]) ∪ F
n
, so that
E
n+1
∈ Σ, E
n+1
⊆ E
n
, µE
n+1
> γ, E
n+1
¸ f
−1
[H
n
] = F
n
.
Continue.
At the end of the induction, set F =
n∈N
E
n
. Then F ⊆ E, µF ≥ γ, and for every n ∈ N
F ∩ f
−1
[H
n
] = F ∩ E
n+1
∩ f
−1
[H
n
] = F ¸ F
n
is relatively open in F. It follows that fF is continuous (4A2B(aii)). As E, γ are arbitrary, f is almost
continuous.
Remark For variations on this idea, see 418Yg, 433E and 434Yb; also 418Yh.
418K Corollary Let (X, T, Σ, µ) be a quasiRadon measure space and Y a separable metrizable space.
Then a function f : X → Y is measurable iﬀ it is almost continuous.
proof Put 418E and 418J together.
Remark This generalizes 256F.
418L Measurable functions and almost continuous functions 115
418L In all the results above, the measure starts on the left of the diagram f : X → Y ; in 418I418K, it
is transferred to an image measure on Y . If X has enough compact sets, a measure can move in the reverse
direction, as follows.
Theorem Let (X, T) be a Hausdorﬀ space, (Y, S, T, ν) a Radon measure space and f : X → Y a continuous
function such that whenever F ∈ T and νF > 0 there is a compact set K ⊆ X such that ν(F ∩ f[K]) > 0.
Then there is a Radon measure µ on X such that ν is the image measure µf
−1
and the inversemeasure
preserving function f induces an isomorphism between the measure algebras of ν and µ.
proof (a) Note ﬁrst that ν is inner regular with respect to L = ¦f[K] : K ∈ /¦, where / is the family of
compact subsets of X. PPP If νF > 0, there is a K ∈ / such that ν(F ∩ f[K]) > 0; now there is a closed set
F
t
⊆ F ∩f[K] such that νF
t
> 0, and K
t
= K∩f
−1
[F
t
] is compact, while f[K
t
] ⊆ F has nonzero measure.
As L is closed under ﬁnite unions, this is enough to show that ν is inner regular with respect to L (412Aa).
QQQ
(b) Consequently there is a disjoint set L
0
⊆ L such that every nonnegligible F ∈ T meets some member
of L
0
in a nonnegligible set (412Ib). We can express L
0
as ¦f[K] : K ∈ /
0
¦ where /
0
⊆ / is disjoint. Set
X
0
=
/
0
.
(c) Set
Σ
0
= ¦X
0
∩ f
−1
[F] : F ∈ T¦.
Then Σ
0
is a σalgebra of subsets of X
0
. If F, F
t
∈ T and νF ,= νF
t
, then there must be some K ∈ /
0
such
that f[K]∩(F´F
t
) ,= ∅, so that X
0
∩f
−1
[F] ,= X
0
∩f
−1
[F
t
]; we therefore have a functional µ
0
: Σ
0
→ [0, ∞]
deﬁned by setting µ
0
(X
0
∩ f
−1
[F]) = νF whenever F ∈ T. It is easy to check that µ
0
is a measure on
X
0
. Now µ
0
is inner regular with respect to /. PPP If E ∈ Σ
0
and µE > 0, there is an F ∈ T such that
E = X
0
∩f
−1
[F] and νF > 0. There are a K ∈ /
0
such that ν(F ∩f[K]) > 0, and a closed set F
t
⊆ F ∩f[K]
such that νF
t
> 0; now K ∩ f
−1
[F
t
] = X
0
∩ f
−1
[F
t
] belongs to Σ
0
∩ /, is included in E and has measure
greater than 0. Because / is closed under ﬁnite unions, this is enough to show that µ
0
is inner regular with
respect to /. QQQ
(c) Set
Σ
1
= ¦E : E ⊆ X, E ∩ X
0
∈ Σ
0
¦, µ
1
E = µ
0
(E ∩ X
0
) for every E ∈ Σ
1
.
Then µ
1
is a measure on X (being the image measure µ
0
ι
−1
, where ι : X
0
→ X is the identity map), and is
inner regular with respect to /. If F ∈ T, then
µ
1
f
−1
[F] = µ
0
(X
0
∩ f
−1
[F]) = νF,
so f is inversemeasurepreserving for µ
1
and ν. Consequently µ
1
is locally ﬁnite. PPP If x ∈ X, there is an
open set H ⊆ Y such that f(x) ∈ H and νH < ∞; now f
−1
[H] is an open subset of X of ﬁnite measure
containing x. QQQ In particular, µ
∗
1
K < ∞ for every compact K ⊆ X (411Ga).
(d) By 413O, there is an extension of µ
1
to a complete locally determined measure µ on X which is inner
regular with respect to /, deﬁned on every member of /, and such that whenever E belongs to the domain
Σ of µ and µE < ∞, there is an E
1
∈ Σ
1
such that µ(E´E
1
) = 0. Now µ is locally ﬁnite because µ
1
is,
so µ is a Radon measure; and f is inversemeasurepreserving for µ and ν because it is inversemeasure
preserving for µ
1
and ν.
(e) The image measure µf
−1
extends ν, so is locally ﬁnite, and is therefore a Radon measure (418I);
since it agrees with ν on the compact subsets of Y , it must be identical with ν.
(f ) I have still to check that the corresponding measurepreserving homomorphism π from the measure
algebra B of ν to the measure algebra A of µ is actually an isomorphism, that is, is surjective. If a ∈ A
and ¯ µa < ∞, we can ﬁnd E ∈ Σ such that E
•
= a and E
1
∈ Σ
1
such that µ(E´E
1
) = 0. Now E
1
∩ X
0
=
f
−1
[F] ∩ X
0
for some F ∈ T; but in this case
µ(E
1
´f
−1
[F]) = µ
1
(E
1
´f
−1
[F]) = 0, a = E
•
1
= (f
−1
[F])
•
= πF
•
.
Accordingly π[B] includes ¦a : ¯ µa < ∞¦, and is orderdense in A. But as π is injective and B is
Dedekind complete (being the measure algebra of a Radon measure, which is strictly localizable), it follows
that π[B] = A (314Ib). Thus π is an isomorphism, as required.
116 Topologies and measures I 418L
Remarks Of course this result is most commonly applied when X and Y are both compact and f is a
surjection, in which case the condition
(*) whenever F ∈ T and νF > 0 there is a compact set K ⊆ X such that ν(F ∩ f[K]) > 0
is trivially satisﬁed.
Evidently (*) is necessary if there is to be any Radon measure on X for which f is inversemeasure
preserving, so in this sense the result is best possible. In 433D, however, there is a version of the theorem
in which f is not required to be continuous.
418M Prokhorov’s theorem Suppose that (I, ≤), ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
, ¸f
ij
)
i≤j∈I
, (X, T) and ¸g
i
)
i∈I
are such that
(I, ≤) is a nonempty upwardsdirected partially ordered set,
every (X
i
, T
i
, Σ
i
, µ
i
) is a Radon probability space,
f
ij
: X
j
→ X
i
is an inversemeasurepreserving function whenever i ≤ j in I,
(X, T) is a Hausdorﬀ space,
g
i
: X → X
i
is a continuous function for every i ∈ I,
g
i
= f
ij
g
j
whenever i ≤ j in I.
Suppose moreover that
for every > 0 there is a compact set K ⊆ X such that µ
i
g
i
[K] ≥ 1 − for every i ∈ I.
Then there is a Radon probability measure µ on X such that every g
i
is inversemeasurepreserving for µ.
proof (a) Set
T = ¦g
−1
i
[E] : i ∈ I, E ∈ Σ
i
¦ ⊆ TX.
Then T is a subalgebra of TX. PPP (i) There is an i ∈ I, so ∅ = g
−1
i
[∅] belongs to T. (ii) If H ∈ T there are
i ∈ I, E ∈ Σ
i
such that H = g
−1
i
[E]; now X ¸ H = g
−1
i
[X
i
¸ E] belongs to T. (iii) If G, H ∈ T, there are i,
j ∈ I and E ∈ Σ
i
, F ∈ Σ
j
such that G = g
−1
i
[E] and H = g
−1
j
[F]. Now I is upwardsdirected, so there is
a k ∈ I such that i ≤ k and j ≤ k. Because f
ik
and f
jk
are inversemeasurepreserving, f
−1
ik
[E] and f
−1
jk
[F]
belong to Σ
k
, so that
G∩ H = g
−1
i
[E] ∩ g
−1
j
[F] = (f
ik
g
k
)
−1
[E] ∩ (f
jk
g
k
)
−1
[F]
= g
−1
k
[f
−1
ik
[E] ∩ f
−1
jk
[F]] ∈ T. QQQ
(b) There is an additive functional ν : T → [0, 1] deﬁned by writing νg
−1
i
[E] = µ
i
E whenever i ∈ I and
E ∈ Σ
i
.
PPP (i) Suppose that i, j ∈ I and E ∈ Σ
i
, F ∈ Σ
j
are such that g
−1
i
[E] = g
−1
j
[F]. Let k ∈ I be such that
i ≤ k and j ≤ k. Then
g
−1
k
[f
−1
ik
[E]´f
−1
jk
[F]] = g
−1
i
[E]´g
−1
j
[F] = ∅,
so g
k
[X] ∩ (f
−1
ik
[E]´f
−1
jk
[F]) = ∅. But now remember that for every > 0 there is a set K ⊆ X such that
µ
k
g
k
[K] ≥ 1 −. This means that µ
k
g
k
[X] must be 1, so that f
−1
ik
[E]´f
−1
jk
[F] must be negligible, and
µ
i
E = µ
k
f
−1
ik
[E] = µ
k
f
−1
jk
[F] = µ
j
F.
Thus the proposed formula for ν deﬁnes a function on T.
(ii) Now suppose that G, H ∈ T are disjoint. Again, take i, j ∈ I and E ∈ Σ
i
, F ∈ Σ
j
such that
G = g
−1
i
[E] and H = g
−1
j
[F], and k ∈ I such that i ≤ k and j ≤ k. Then
νG+νH = µ
i
E +µ
j
F = µ
k
f
−1
ik
[E] +µ
k
f
−1
jk
[F]
= µ
k
(f
−1
ik
[E] ∪ f
−1
jk
[F]) +µ
k
(f
−1
ik
[E] ∩ f
−1
jk
[F])
= νg
−1
k
[f
−1
ik
[E] ∪ f
−1
jk
[F]] +νg
−1
k
[f
−1
ik
[E] ∩ f
−1
jk
[F]]
= ν(G∪ H) +ν(G∩ H).
418Nf Measurable functions and almost continuous functions 117
But as ν∅ is certainly 0, we get ν(G∪ H) = νG+νH. As G, H are arbitrary, ν is additive. QQQ
Note that νX = 1.
(c) νG = sup¦νH : H ∈ T, H ⊆ G, H is closed¦ for every G ∈ T. PPP If γ < νG, there are an i ∈ I
and an E ∈ Σ
i
such that G = g
−1
i
[E]. In this case µ
i
E = νG > γ; let L ⊆ E be a compact set such that
µ
i
L ≥ γ; then H = g
−1
i
[L] is a closed subset of G and νH = µ
i
L ≥ γ. QQQ
νX = sup
K⊆X is compact
inf
G∈T,G⊇K
νG. PPP If > 0, there is a compact K ⊆ X such that µ
i
g
i
[K] ≥ 1−
for every i ∈ I, by the ﬁnal hypothesis of the theorem. If G ∈ T and G ⊇ K, there are an i ∈ I and an
E ∈ Σ
i
such that G = g
−1
i
[E], in which case g
i
[K] ⊆ E, so that
νG = µ
i
E ≥ µ
i
g
i
[K] ≥ 1 −.
Thus inf
G∈T,G⊇K
νG ≥ 1 −; as is arbitrary, we have the result. QQQ
This means that the conditions of 416O are satisﬁed, and there is a Radon measure µ extending ν. Of
course this means that every g
i
is inversemeasurepreserving.
418N Remarks (a) Taking I to be a singleton, we get a version of 418L in which Y is a probability
space, and omitting the check that the function g induces an isomorphism of the measure algebras. Taking
I to be the family of ﬁnite subsets of a set T, and every X
i
to be a product
t∈i
Z
t
of Radon probability
spaces with its product Radon measure, we obtain a method of constructing products of arbitrary families
of compact probability spaces from ﬁnite products.
(b) In the hypotheses of 418M, I asked only that the f
ij
should be measurable, and omitted any check on
the compositions f
ij
f
jk
when i ≤ j ≤ k. But it is easy to see that the f
ij
must in fact be almost continuous,
and that f
ij
f
jk
must be equal almost everywhere to f
ik
(418Xu), just as in 418P below.
(c) In the theorem as written out above, the space X and the functions g
i
: X → X
i
are part of the data.
Of course in many applications we start with a structure
(¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
, ¸f
ij
)
i≤j∈I
),
and the ﬁrst step is to ﬁnd a suitable X and g
i
, as in 418O and 418P.
(d) There are important questions concerning possible relaxations of the hypotheses in 418M, especially in
the special case already mentioned, in which X
i
=
t∈i
Z
t
, f
ij
(x) = xi when i ⊆ j ∈ [T]
<ω
, X =
t∈T
Z
t
,
and g
i
(x) = xi for x ∈ X and i ∈ I, but there is no suggestion that the µ
i
are product measures. For a
case in which we can dispense with auxiliary topologies on the X
i
, see 451Yb.
(e) A typical class of applications of Prokhorov’s theorem is in the theory of stochastic processes, in which
we have large families ¸X
t
)
t∈T
of random variables; for deﬁniteness, imagine that T = [0, ∞[, so that we
are looking at a system evolving over time. Not infrequently our intuition leads us to a clear description of
the joint distributions ν
J
of ﬁnite subfamilies ¸X
t
)
t∈J
without providing any suggestion of a measure space
on which the whole family ¸X
t
)
t∈T
might be deﬁned. (As I tried to explain in the introduction to Chapter
27, probability spaces themselves are often very shadowy things in true probability theory.) Each ν
J
can be
thought of as a Radon measure on R
J
, and for I ⊆ J ∈ [T]
<ω
we have a natural map f
IJ
: R
J
→R
I
, setting
f
IJ
(y) = yI for y ∈ R
J
. If our distributions ν
J
mean anything at all, every f
IJ
will surely be inverse
measurepreserving; this is simply saying that ν
I
is the joint distribution of a subfamily of ¸X
t
)
t∈J
. If we
can ﬁnd a Hausdorﬀ space Ω and a continuous function g : Ω → R
T
such that, for every ﬁnite J ⊆ T and
> 0, there is a compact set K ⊆ Ω such that ν
J
g
J
[K] ≥ 1 − (where g
J
(x) = g(x)J), then Prokhorov’s
theorem will give us a measure µ on Ω which will then provide us with a suitable realization of ¸X
t
)
t∈T
as
a family of random variables on a genuine probability space, writing X
t
(ω) = g(ω)(t). That they become
continuous functions on a Radon measure space is a valuable shield against irrelevant complications.
Clearly, if this can be done at all it can be done with Ω = R
T
; but some of the central results of probability
theory are speciﬁcally concerned with the possibility of using other sets Ω (e.g., Ω = C(T), as in 455D).
(f ) In (e) above, we do always have the option of regarding each ν
J
as a measure on the compact space
[−∞, ∞]
J
. In this case, by 418O or otherwise, we can be sure of ﬁnding a measure on [−∞, ∞]
T
to support
functions X
t
, at the cost of either allowing the values ±∞ or (as I should myself ordinarily do) accepting
118 Topologies and measures I 418Nf
that each X
t
would be undeﬁned on a negligible set. The advantage of this is just that it gives us conﬁdence
in applying the KolmogorovLebesgue theory to the whole family ¸X
t
)
t∈T
at once, rather than to ﬁnite or
countable subfamilies. For an example of what can happen if we try to do similar things with noncompact
measures, see 419K. For an example of the problems which can arise with uncountable families, see 418Xv.
418O I mention two cases in which we can be sure that the projective limit (X, ¸g
i
)
i∈I
) required in
Prokhorov’s theorem will exist.
Proposition Suppose that (I, ≤), ¸(X
i
, T
i
))
i∈I
and ¸f
ij
)
i≤j∈I
are such that
(I, ≤) is a nonempty upwardsdirected partially ordered set,
every (X
i
, T
i
) is a compact Hausdorﬀ space,
f
ij
: X
j
→ X
i
is a continuous surjection whenever i ≤ j in I,
f
ij
f
jk
= f
ik
whenever i ≤ j ≤ k in I.
Then there are a compact Hausdorﬀ space (X, T) and a family ¸g
i
)
i∈I
such that g
i
: X → X
i
is a continuous
surjection and f
ij
g
j
= f
i
whenever i ≤ j ∈ I; so that if we endow the X
i
with Radon probability measures for
which the f
ij
are inversemeasurepreserving, (I, ¸X
i
)
i∈I
, ¸f
ij
)
i≤j∈I
, X, ¸g
i
)
i∈I
) will satisfy all the hypotheses
of 418M.
proof Set
X = ¦x : x ∈
i∈I
X
i
, f
ij
x(j) = x(i) whenever i ≤ j ∈ I¦,
g
i
(x) = x(i) for x ∈ X, i ∈ I.
Of course g
i
= f
ij
g
j
whenever i ≤ j. To see that every g
i
is surjective, observe that if y ∈ X
i
and J ⊆ I is
ﬁnite, then
F
J
= ¦x : x ∈
i∈I
X
i
, x(i) = y, f
jk
x(k) = x(j) whenever j ≤ k ∈ J¦
is a closed set. F
J
is always nonempty, because if k is an upper bound of J ∪ ¦i¦ there is a z ∈ X
k
such
that f
ik
(z) = y, in which case x ∈ F
J
whenever x(j) = f
jk
(z) for every j ∈ J ∪ ¦i¦. Now ¦F
J
: J ∈ [I]
<ω
¦
is a downwardsdirected family of nonempty closed sets in the compact space
j∈I
X
j
, so has nonempty
intersection, and if x is any point of the intersection then x ∈ X and g
i
(x) = y.
418P Proposition Let (I, ≤), ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
and ¸f
ij
)
i≤j∈I
be such that
(I, ≤) is a countable nonempty upwardsdirected partially ordered set,
every (X
i
, T
i
, Σ
i
, µ
i
) is a Radon probability space,
f
ij
: X
j
→ X
i
is an inversemeasurepreserving almost continuous function whenever i ≤ j in
I,
f
ij
f
jk
= f
ik
µ
k
a.e. whenever i ≤ j ≤ k in I.
Then there are a Radon probability space (X, T, Σ, µ) and continuous inversemeasurepreserving functions
g
i
: X → X
i
such that g
i
= f
ij
g
j
whenever i ≤ j in I.
proof (a) We can use the same formula as in 418O:
X = ¦x : x ∈
i∈I
X
i
, f
ij
x(j) = x(i) whenever i ≤ j ∈ I¦,
g
i
(x) = x(i) for x ∈ X, i ∈ I.
As before, the consistency relation g
i
= f
ij
g
j
is a trivial consequence of the deﬁnition of X. For the rest, we
have to check that the ﬁnal condition of 418M is satisﬁed. Fix ∈ ]0, 1[. Start by taking a family ¸
ij
)
i≤j∈I
of strictly positive numbers such that
i≤j∈I
ij
≤
1
2
. (This is where we need to know that I is countable.)
Set
j
=
i≤j
ij
for each j, so that
j∈I
j
≤
1
2
.
For i ≤ j ≤ k in I, set
E
ijk
= ¦x : x ∈ X
k
, f
ik
(x) = f
ij
f
jk
(x)¦,
418Q Measurable functions and almost continuous functions 119
so that E
ijk
is µ
k
conegligible; set E
k
=
i≤j≤k
E
ijk
, so that E
k
is µ
k
conegligible. For i ≤ j ∈ I, choose
compact sets K
ij
⊆ E
j
such that µ
j
K
ij
≥ 1 −
ij
and f
ij
K
ij
is continuous. Now we seem to need a
threestage construction, as follows:
for j ∈ I, set K
j
=
i≤j
K
ij
;
for j ∈ I, set K
∗
j
= K
j
∩
i≤j
f
−1
ij
[K
i
];
ﬁnally, set K = X ∩
i∈I
K
∗
i
.
Let us trace the properties of these sets stage by stage.
(b) For each j ∈ I, K
j
⊆ K
jj
⊆ E
j
is compact and
µ
j
(X
j
¸ K
j
) ≤
i≤j
µ
j
(X
j
¸ K
ij
) ≤
i≤j
ij
=
j
,
so that µ
j
K
j
≥ 1 −
j
. Note that f
ik
agrees with f
ij
f
jk
on K
k
whenever i ≤ j ≤ k, and that f
ij
K
j
is
continuous whenever i ≤ j.
(c) Every K
∗
j
is compact, and if i ≤ j ≤ k then f
ik
agrees with f
ij
f
jk
on K
∗
k
, while f
ij
K
∗
j
is always
continuous. Also
µ
j
(X
j
¸ K
∗
j
) ≤ µ
j
(X
j
¸ K
j
) +
i≤j
µ
j
(X
j
¸ f
−1
ij
[K
i
])
≤
j
+
i≤j
i
≤ ,
so µ
j
K
∗
j
≥ 1 −, for every j ∈ I.
The point of moving from K
j
to K
∗
j
is that f
jk
[K
∗
k
] ⊆ K
∗
j
whenever j ≤ k in I. PPP K
∗
k
⊆ f
−1
jk
[K
j
], so
f
jk
[K
∗
k
] ⊆ K
j
. If i ≤ j, then
K
∗
k
= K
∗
k
∩ f
−1
ik
[K
i
] = K
∗
k
∩ f
−1
jk
[f
−1
ij
[K
i
]]
because f
ij
f
jk
agrees with f
ik
on K
∗
k
. So f
jk
[K
∗
k
] ⊆ f
−1
ij
[K
i
]. As i is arbitrary, f
jk
[K
∗
k
] ⊆ K
∗
j
. QQQ
Again because f
ik
agrees with f
ij
f
jk
on K
∗
k
, we have f
ik
[K
∗
k
] = f
ij
[f
jk
[K
∗
k
]] ⊆ f
ij
[K
∗
j
] whenever i ≤ j ≤ k.
And because f
ij
K
∗
j
is always continuous, all the sets f
ij
[K
∗
j
] are compact.
(d)(i) K is compact. PPP
K = ¦x : x ∈
i∈I
K
∗
i
, f
ij
x(j) = x(i) whenever i ≤ j ∈ I¦
is closed in
i∈I
K
∗
i
because f
ij
K
∗
j
is always continuous (and every X
i
is Hausdorﬀ). Since
i∈I
K
∗
i
is
compact, so is K. QQQ
(ii) µ
i
g
i
[K] ≥ 1− for every i ∈ I. PPPBy (c), f
ik
[K
∗
k
] ⊆ f
ij
[K
∗
j
] whenever i ≤ j ≤ k. So ¦f
ij
[K
∗
j
] : j ≥ i¦
is a downwardsdirected family of compact sets; write L for their intersection. Since
µ
i
f
ij
[K
∗
j
] = µ
j
f
−1
ij
[f
ij
[K
∗
j
]] ≥ µ
j
K
∗
j
≥ 1 −
for every j ≥ i, µ
i
L ≥ 1 − (414C). If z ∈ L, then for every k ≥ i the set
F
k
= ¦x : x ∈
j∈I
K
∗
j
, x(k) = z, f
jk
x(k) = x(j) whenever j ≤ k¦
is a nonempty closed set in
j∈I
K
∗
j
, while F
k
⊆ F
j
when j ≤ k; so that ¦F
k
: k ≥ i¦ is a downwardsdirected
family of nonempty closed sets in a compact space, and has nonempty intersection. But if x ∈
k≥i
F
k
,
then x ∈ K and x(k) = z, so z ∈ g
i
[K]. Thus g
i
[K] ⊇ L and µ
i
g
i
[K] ≥ 1 −. QQQ
(e) As is arbitrary, the ﬁnal condition of 418M is satisﬁed. But now 418M tells us that (β) is true.
418Q Corollary Let ¸(X
n
, T
n
, Σ
n
, µ
n
))
n∈N
be a sequence of Radon probability spaces, and suppose we
are given an inversemeasurepreserving almost continuous function f
n
: X
n+1
→ X
n
for each n. Set
X = ¦x : x ∈
n∈N
X
n
, f
n
(x(n + 1)) = x(n) for every n ∈ N¦.
Then there is a unique Radon probability measure µ on X such that all the coordinate maps x → x(n) :
X → X
n
are inversemeasurepreserving.
120 Topologies and measures I 418Q
proof (a) For i ≤ j ∈ N, deﬁne f
ij
: X
j
→ X
i
by writing
f
ii
(x) = x for every x ∈ X
i
,
f
i,j+1
= f
ij
f
j
for every j ≥ i.
It is easy to check that f
ij
f
jk
= f
ik
whenever i ≤ j ≤ k, and that every f
ij
is inversemeasurepreserving
and almost continuous (using 418Dc). So we are exactly in the situation of 418P, and we know that there
is a Radon probability measure on X for which every g
i
is inversemeasurepreserving.
(b) To see that µ is unique, observe that if K ⊆ X is compact and x ∈ X ¸ K then there is some ﬁnite
set I ⊆ N such that ¦y : yI = xI¦ is disjoint from K. But as g
n
= f
n
g
n+1
for every n, this means that
¦y : g
n
(y) = g
n
(x)¦ is disjoint from K for all n large enough. As x is arbitrary, we see that ¸g
−1
n
[g
n
[K]])
n∈N
is a nonincreasing sequence with intersection K, so that
µK = lim
n→∞
µg
−1
n
[g
n
[K]] = lim
n→∞
µ
n
g
n
[K].
Thus the values of µ on compact sets are determined by the construction. By 416E(bii), µ is uniquely
deﬁned.
418R I turn now to a special kind of measurable function, corresponding to a new view of product
spaces.
Theorem Let X be a set, Σ a σalgebra of subsets of X, and (Y, T, ν) a σﬁnite measure space. Give L
0
(ν)
the topology of convergence in measure (¸245). Write L
0
(Σ
´
⊗T) for the space of Σ
´
⊗Tmeasurable functions
h : X Y → R, where Σ
´
⊗T is the σalgebra of subsets of X Y generated by ¦E F : E ∈ Σ, F ∈ T¦.
Then for a function f : X → L
0
(ν) the following are equiveridical:
(i) f[X] is separable and f is measurable;
(ii) there is an h ∈ L
0
(Σ
´
⊗T) such that f(x) = h
•
x
for every x ∈ X, where h
x
(y) = h(x, y) for x ∈ X,
y ∈ Y .
proof Let ¸Y
n
)
n∈N
be a nondecreasing sequence of subsets of Y of ﬁnite measure covering Y .
(a)(i)⇒(ii) For each n ∈ N, let ρ
n
be the continuous pseudometric on L
0
(ν) deﬁned by saying that
ρ
n
(g
•
1
, g
•
2
) =
_
Y
n
min(1, [g
1
− g
2
[)dν for g
1
, g
2
∈ L
0
(T), writing L
0
(T) for the space of Tmeasurable real
valued functions on Y (245A). Then ¦ρ
n
: n ∈ N¦ deﬁnes the topology of L
0
(ν) (see the proof of 245Eb).
Because f[X] is separable, there is a sequence ¸v
k
)
k∈N
in L
0
(ν) such that f[X] ⊆ ¦v
k
: k ∈ N¦. For each k,
choose g
k
∈ L
0
(T) such that g
•
k
= v
k
. For n, k ∈ N set
E
nk
= ¦x : x ∈ X, ρ
n
(f(x), v
k
) < 2
−n
¦,
H
nk
= E
nk
¸
i<k
E
ni
.
Then every E
nk
belongs to Σ (because f is measurable) and
k∈N
E
nk
= X (because ¦v
k
: k ∈ N¦ is dense);
so ¸H
nk
)
k∈N
is a disjoint sequence of measurable sets covering X. Set h
(n)
(x, y) = g
k
(y) for x ∈ H
nk
; then
h
(n)
∈ L
0
(Σ
´
⊗T).
Fix x ∈ X for the moment. Then for each n ∈ N there is a unique k
n
such that x ∈ H
nk
n
, and
ρ
n
(f(x), v
k
n
) ≤ 2
−n
. So if n ≤ m,
_
Y
n
min(1, [h
(m+1)
x
−h
(m)
x
[) =
_
Y
n
min(1, [g
k
m+1
−g
k
m
[) = ρ
n
(g
•
k
m+1
, g
•
k
m
)
≤ ρ
n
(g
•
k
m+1
, f(x)) +ρ
n
(f(x), g
•
k
m
)
≤ ρ
m+1
(g
•
k
m+1
, f(x)) +ρ
m
(f(x), g
•
k
m
) ≤ 3 2
−m−1
.
But this means that
∞
m=0
_
Y
n
min(1, [h
(m+1)
x
−h
(m)
x
[) is ﬁnite, so that ¸h
(m)
x
)
m∈N
must be convergent almost
everywhere on Y
n
. As this is true for every n, ¸h
(m)
x
)
m∈N
is convergent a.e. on Y . Moreover,
lim
m→∞
(h
(m)
x
)
•
= lim
m→∞
g
•
k
m
= f(x)
in L
0
(ν).
418S Measurable functions and almost continuous functions 121
Since this is true for every x,
W = ¦(x, y) : ¸h
(m)
(x, y))
m∈N
converges in R¦
has conegligible vertical sections, while of course W ∈ Σ
´
⊗T because every h
(m)
is Σ
´
⊗Tmeasurable (418C).
If we set h(x, y) = lim
m→∞
h
(m)
(x, y) for (x, y) ∈ W, 0 for other (x, y) ∈ X Y , then h ∈ L
0
(Σ
´
⊗T), while
(by 245Ca)
h
•
x
= lim
m→∞
(h
(m)
x
)
•
= f(x)
in L
0
(ν) for every x ∈ X. So we have a suitable h.
(ii)⇒(i) Let Φ be the set of those h ∈ L
0
(Σ
´
⊗T) such that (i) is satisﬁed; that is, x → h
•
x
is measurable,
and ¦h
•
x
: x ∈ X¦ is separable. Then
(ααα) Φ is closed under addition. PPP If h,
˜
h belong to Φ, set A = ¦h
•
x
: x ∈ X¦,
˜
A = ¦
˜
h
•
x
: x ∈ X¦. Then
both A and
˜
A are separable metrizable spaces, so A
˜
A is separable and metrizable and x → (h
•
x
,
˜
h
•
x
) :
X → A
˜
A is measurable (418Bb). But addition on L
0
(ν) is continuous (245D), so
x → h
•
x
+
˜
h
•
x
= (h +
˜
h)
•
x
is measurable (418Ac), and
¦(h +
˜
h)
•
x
: x ∈ X¦ ⊆ ¦u + ˜ u : u ∈ A, ˜ u ∈
˜
A¦
is separable (4A2Be). Thus h +
˜
h ∈ Φ. QQQ
(βββ) Φ is closed under scalar multiplication, just because u → αu : L
0
(ν) → L
0
(ν) is always continuous.
(γγγ) If ¸h
(n)
)
n∈N
is a sequence in Φ and h(x, y) = lim
n→∞
h
(n)
(x, y) for all x ∈ X, y ∈ Y , then h ∈ Φ.
PPP Setting A
n
= ¦(h
(n)
x
)
•
: x ∈ X¦ for each n, then A = ¦h
•
x
: x ∈ X¦ is included in
n∈N
A
n
, which is
separable (4A2B(ei)), so A is separable (4A2P(aiv)); moreover, h
•
x
= lim
n→∞
(h
(n)
x
)
•
for every x ∈ X, so
x → h
•
x
is measurable, by 418Ba. QQQ
(δδδ) What this means is that if we set J = ¦W : W ∈ Σ
´
⊗T, χW ∈ Φ¦, then W ¸ W
t
∈ J whenever
W, W
t
∈ J and W
t
⊆ W, and that
n∈N
W
n
∈ J whenever ¸W
n
)
n∈N
is a nondecreasing sequence in
J. Also, it is easy to see that E F ∈ J whenever E ∈ Σ and F ∈ T. By the Monotone Class Theorem
(136B), J includes the σalgebra generated by ¦EF : E ∈ Σ, F ∈ T¦, that is, is equal to Σ
´
⊗T. It follows
at once, from (α) and (β), that
n
i=0
α
i
χW
i
∈ Φ whenever W
0
, . . . , W
n
∈ Σ
´
⊗T and α
0
, . . . , α
n
∈ R, and
hence (using (γ)) that L
0
(Σ
´
⊗T) ⊆ Φ, which is what we had to prove.
418S Corollary Let (X, Σ, µ) and (Y, T, ν) be σﬁnite measure spaces with c.l.d. product (X Y, Λ, λ).
Give L
0
(ν) the topology of convergence in measure. Write L
0
(λ) for the space of Λmeasurable realvalued
functions deﬁned λa.e. on X Y , as in ¸241.
(a) If h ∈ L
0
(λ), set h
x
(y) = h(x, y) whenever this is deﬁned. Then
¦x : f(x) = h
•
x
is deﬁned in L
0
(ν)¦
is µconegligible, and includes a conegligible set X
0
such that f : X
0
→ L
0
(ν) is measurable and f[X
0
] is
separable.
(b) If f : X → L
0
(ν) is measurable and there is a conegligible set X
0
⊆ X such that f[X
0
] is separable,
then there is an h ∈ L
0
(λ) such that f(x) = h
•
x
for almost every x ∈ X.
proof (a) The point is that λ is just the completion of its restriction to Σ
´
⊗T (251K). So there is a conegligible
set W ∈ Σ
´
⊗T such that hW is Σ
´
⊗Tmeasurable (212Fa). Setting
˜
h(x, y) = h(x, y) for (x, y) ∈ W, 0
otherwise, and setting
˜
f(x) =
˜
h
•
x
for every x ∈ X, we see from 418R that
˜
f is measurable and that
˜
f[X] is
separable. But 252D tells us that
X
0
= ¦x : ((X Y ) ¸ W)[¦x¦] is negligible¦
is conegligible; and if x ∈ X
0
then h
x
=
˜
h
x
νa.e., so that f(x) is deﬁned and equal to
˜
f(x). This proves
the result.
122 Topologies and measures I 418S
(b) fX
0
satisﬁes 418R(i). So, setting f
1
(x) = f(x) for x ∈ X
0
, 0 otherwise, there is some h ∈ L
0
(Σ
´
⊗T)
such that f
1
(x) = h
•
x
for every x, so that f(x) = h
•
x
for almost every x, and (ii) is true.
418T Corollary (Mauldin & Stone 81) Let (Y, T, ν) be a σﬁnite measure space, and (B, ¯ ν) its
measure algebra, with its measurealgebra topology (¸323).
(a) Let X be a set, Σ a σalgebra of subsets of X, and f : X → B a function. Then the following are
equiveridical:
(i) f[X] is separable and f is measurable;
(ii) there is a W ∈ Σ
´
⊗T such that f(x) = W[¦x¦]
•
for every x ∈ X.
(b) Let (X, Σ, µ) be a σﬁnite measure space and Λ the domain of the c.l.d. product measure λ on XY .
(i) Suppose that ν is complete. If W ∈ Λ, then
¦x : f(x) = W[¦x¦]
•
is deﬁned in B¦
is µconegligible, and includes a conegligible set X
0
such that f : X
0
→ B is measurable and f[X
0
] is
separable.
(ii) If f : X → B is measurable and there is a conegligible set X
0
⊆ X such that f[X
0
] is separable,
then there is a W ∈ Σ
´
⊗T such that f(x) = W[¦x¦]
•
for almost every x ∈ X.
proof Everything follows directly from 418R and 418S if we observe that B is homeomorphically embedded
in L
0
(ν) by the function F
•
→ (χF)
•
for F ∈ T (323Xf, 367R). We do need to check, for (i)⇒(ii) of part
(a), that if h ∈ L
0
(Σ
´
⊗T) and h
•
x
is always of the form (χF)
•
, then there is some W ∈ Σ
´
⊗T such that
h
•
x
= (χW[¦x¦])
•
for every x; but of course this is true if we just take W = ¦(x, y) : h(x, y) = 1¦. Now (bii)
follows from (a) just as 418Sb followed from 418R.
418X Basic exercises >>>(a) Let (X, Σ, µ) be a measure space, Y a set and h : X → Y a function; give
Y the image measure µh
−1
. Show that for any function g from Y to a topological space Z, g is measurable
iﬀ gh : X → Z is measurable.
>>>(b) Let X be a set, Σ a σalgebra of subsets of X, ¸Y
n
)
n∈N
a sequence of topological spaces with product
Y , and f : X → Y a function. Show that f is measurable iﬀ ψ
n
f : X →
i≤n
Y
i
is measurable for every
n ∈ N, where ψ
n
(y) = (y(0), . . . , y(n)) for y ∈ Y , n ∈ N.
(c) Let (X, Σ, µ) be a semiﬁnite measure space, (Y, S) a metrizable space, and ¸f
n
)
n∈N
a sequence of
measurable functions from X to Y such that ¸f
n
(x))
n∈N
is convergent for almost every x ∈ X. Show that
µ is inner regular with respect to ¦E : ¸f
n
E)
n∈N
is uniformly convergent¦. (Cf. 412Xr.)
>>>(d) Set Y = [0, 1]
[0,1]
, with the product topology. For x ∈ [0, 1], n ∈ N deﬁne f
n
(x) ∈ Y by saying that
f
n
(x)(t) = max(0, 1−2
n
[x−t[) for t ∈ [0, 1]. Check that (i) each f
n
is continuous, therefore measurable; (ii)
f(x) = lim
n→∞
f
n
(x) is deﬁned in Y for every x ∈ [0, 1]; (iii) for each t ∈ [0, 1], the coordinate functional
x → f(x)(t) is continuous except at t, and in particular is almost continuous and measurable; (iv) fF is
not continuous for any inﬁnite closed set F ⊆ [0, 1], and in particular f is not almost continuous; (v) every
subset of [0, 1] is of the form f
−1
[H] for some open set H ⊆ Y ; (vi) f is not measurable; (vii) the image
measure µf
−1
, where µ is Lebesgue measure on [0, 1], is neither a topological measure nor tight.
(e) Let (X, T, Σ, µ) be a quasiRadon measure space, Y a topological space, and f : X → Y a function.
Suppose that for every x ∈ X there is an open set G containing x such that fG is almost continuous with
respect to the subspace measure on G. Show that f is almost continuous.
(f ) For i = 1, 2 let (X
i
, T
i
, Σ
i
, µ
i
) and (Y
i
, S
i
, T
i
, ν
i
) be quasiRadon measure spaces, and f
i
: X
i
→ Y
i
an almost continuous inversemeasurepreserving function. Show that (x
1
, x
2
) → (f
1
(x
1
), f
2
(x
2
)) is inverse
measurepreserving for the quasiRadon product measures.
(g) Let ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
and ¸(Y
i
, T
i
, Σ
i
, ν
i
))
i∈I
be two families of topological spaces with τadditive
Borel probability measures, and let µ, ν be the τadditive product measures on X =
i∈I
X
i
, Y =
i∈I
Y
i
.
Suppose that every ν
i
is strictly positive. Show that if f
i
: X
i
→ Y
i
is almost continuous and inverse
measurepreserving for each i, then x → ¸f
i
(x(i)))
i∈I
: X → Y is inversemeasurepreserving, but need not
be almost continuous.
418Xt Measurable functions and almost continuous functions 123
(h) Let (X, T, Σ, µ) and (Y, S, T, ν) be quasiRadon measure spaces, (Z, U) a topological space and
f : X Y → Z a function which is almost continuous with respect to the quasiRadon product measure on
X Y . Suppose that ν is σﬁnite. Show that y → f(x, y) is almost continuous for almost every x ∈ X.
(i) Let (X, T, Σ, µ) be an eﬀectively locally ﬁnite τadditive topological measure space, Y a topological
space and f : X → Y an almost continuous function. (i) Show that the image measure µf
−1
is τadditive.
(ii) Show that if µ is a totally ﬁnite quasiRadon measure and the topology on Y is regular, then µf
−1
is
quasiRadon.
(j) Let (X, T, Σ, µ) be a topological measure space and U a linear topological space. Show that if
f : X → U and g : X → U are almost continuous, then f +g : X → U is almost continuous.
(k) Let (X, T, Σ, µ) and (Y, S, T, ν) be topological measure spaces, and (Z, U) a topological space; let
f : X → Y be almost continuous and inversemeasurepreserving, and g : Y → Z almost continuous. Show
that if either µ is a Radon measure and ν is locally ﬁnite or µ is τadditive and eﬀectively locally ﬁnite and
ν is eﬀectively locally ﬁnite, then gf : X → Z is almost continuous. (Hint: show that if µE > 0 there is a
set F such that νF < ∞ and µ(E ∩ f
−1
[F]) > 0.)
(l) Let (X, Σ, µ) be a complete strictly localizable measure space, φ : Σ → Σ a lower density such that
φX = X, and T the associated density topology on X (414P). Let f : X →R be a function. Show that the
following are equiveridical: (i) f is measurable; (ii) f is almost continuous; (iii) f is continuous at almost
every point; (iv) there is a conegligible set H ⊆ X such that fH is continuous. (Cf. 414Xk.)
(m) Let (X, Σ, µ) be a complete strictly localizable measure space, φ : Σ → Σ a lifting, and S the lifting
topology on X (414Q). Let f : X → R be a function. Show that the following are equiveridical: (i) f is
measurable; (ii) f is almost continuous; (iii) there is a conegligible set H ⊆ X such that fH is continuous.
(Cf. 414Xr.)
(n) Let (X, T, Σ, µ) be a quasiRadon measure space, (Y, S) a regular topological space and f : X → Y
an almost continuous function. Show that there is a quasiRadon measure ν on Y such that f is inverse
measurepreserving for µ and ν iﬀ
¦f
−1
[H] : H ⊆ Y is open, µf
−1
[H] < ∞¦ is conegligible in X.
(o) Let X and Y be Hausdorﬀ spaces and f : X → Y a continuous injective function. Show that if µ
1
and µ
2
are distinct totally ﬁnite Radon measures on X then µ
1
f
−1
,= µ
2
f
−1
.
(p) Let (X, T, Σ, µ) be a Radon measure space, (Y, S) and (Z, U) Hausdorﬀ spaces, f : X → Y an almost
continuous function such that ν = µf
−1
is locally ﬁnite, and g : Y → Z a function. Show that g is almost
continuous with respect to ν iﬀ gf is almost continuous with respect to µ.
(q) Let (X, T, Σ, µ) and (Y, S, T, ν) be topological probability spaces, and f : X → Y a measurable
function such that µf
−1
[H] ≥ νH for every H ∈ S. Show that (i)
_
gfdµ =
_
g dν for every g ∈ C
b
(Y )
(ii) µf
−1
[F] = νF for every Baire set F ⊆ Y (iii) if µ is a Radon measure and f is almost continuous, then
µf
−1
[F] = νF for every Borel set F ⊆ Y , so that if in addition ν is complete and inner regular with respect
to the Borel sets then it is a Radon measure.
(r) Let (X, T, Σ, µ) be a totally ﬁnite topological measure space in which the topology T is normal and µ
is inner regular with respect to the closed sets. Show that if f : X →R is a measurable function and > 0
there is a continuous g : X →R such that µ¦x : g(x) ,= f(x)¦ ≤ .
(s) Let X and Y be Hausdorﬀ spaces, ν a totally ﬁnite Radon measure on Y , and f : X → Y an injective
continuous function. Show that the following are equiveridical: (i) there is a Radon measure µ on X such
that f is inversemeasurepreserving; (ii) f[X] is conegligible and f
−1
: f[X] → X is almost continuous.
(t) Let (X, T, Σ, µ) and (Y, S, T, ν) be Radon measure spaces and f : X → Y an almost continuous
inversemeasurepreserving function. Show that (i) µ
∗
A ≤ ν
∗
f[A] for every A ⊆ X (ii) ν is precisely the
image measure µf
−1
.
124 Topologies and measures I 418Xu
(u) In 418M, show that all the f
ij
must be almost continuous. Show that if i ≤ j ≤ k then f
ij
f
jk
= f
ik
almost everywhere on X
k
.
>>>(v) Let 1 be the family of ﬁnite subsets of [0, 1], and let (X
I
, T
I
, Σ
I
, µ
I
) be [0, 1] ¸ I with its subspace
topology and measure for each I ∈ 1. For I ⊆ J ∈ 1, y ∈ X
J
set f
IJ
(y) = y. Show that these X
I
, f
IJ
satisfy nearly all the hypotheses of 418O, but that there are no X, g
I
which satisfy the hypotheses of 418M.
(w) Let T be any set, and X the set of total orders on T. (i) Regarding each member of X as a subset
of T T, show that X is a closed subset of T(T T). (ii) Show that there is a unique Radon measure µ on
X such that Pr(t
1
≤ t
2
≤ . . . ≤ t
n
) =
1
n!
for all distinct t
1
, . . . , t
n
∈ T. (Hint: for I ∈ [T]
<ω
, let X
I
be the
set of total orderings on I with the uniform probability measure giving the same measure to each singleton;
show that the natural map from X
I
to X
J
is inversemeasurepreserving whenever J ⊆ I.)
(x) In 418S, suppose that f
1
: X → L
0
(ν) and f
2
: X → L
0
(ν) correspond to h
1
, h
2
∈ L
0
(λ). Show that
f
1
(x) ≤ f
2
(x) µa.e.(x) iﬀ h
1
≤ h
2
λa.e. Hence show that (if we assign appropriate algebraic operations to
the space of functions from X to L
0
(ν)) we have an falgebra isomorphism between L
0
(λ) and the space of
equivalence classes of measurable functions from X to L
0
(ν) with separable ranges.
418Y Further exercises (a) Let X be a set, Σ a σalgebra of subsets of X, Y a topological space and
f : X → Y a function. Set T = ¦F : F ⊆ Y, f
−1
[F] ∈ Σ¦. Suppose that Y is hereditarily Lindel¨of and its
topology is generated by some subset of T. Show that f is measurable.
(b) Let (X, Σ, µ) be a measure space, Y and Z topological spaces and f : X → Y , g : X → Z measurable
functions. Show that if Z has a countable network consisting of Borel sets (e.g., Z is secondcountable, or
Z is regular and has a countable network), then x → (f(x), g(x)) : X → Y Z is measurable.
(c) Let X be a set, Σ a σalgebra of subsets of X, and ¸Y
i
)
i∈I
a countable family of topological spaces
with product Y . Suppose that every Y
i
has a countable network, and that f : X → Y is a function such
that π
i
f is measurable for every i ∈ I, writing π
i
(y) = y(i). Show that f is measurable.
(d) Find strictly localizable Hausdorﬀ topological measure spaces (X, T, Σ, µ), (Y, S, T, ν) and (Z, U, Λ, λ)
and almost continuous inversemeasurepreserving functions f : X → Y , g : Y → Z such that gf is not
almost continuous.
(e) Let (X, Σ, µ) be a σﬁnite measure space and T a topology on X such that µ is eﬀectively locally
ﬁnite and τadditive. Let Y be a topological space and f : X → Y an almost continuous function. Show
that there is a conegligible subset X
0
of X such that f[X
0
] is ccc.
(f ) Show that if µ is Lebesgue measure on R, T is the usual topology on R and S is the rightfacing
Sorgenfrey topology, then the identity map from (R, T, µ) to (R, S) is measurable, but not almost continuous,
and the image measure is not a Radon measure.
(g) Let (X, Σ, µ) be a semiﬁnite measure space and T a topology on X such that µ is inner regular with
respect to the closed sets. Suppose that Y is a topological space with a countable network consisting of
Borel sets, and that f : X → Y is measurable. Show that f is almost continuous.
(h) Find a topological probability space (X, T, Σ, µ) in which µ is inner regular with respect to the closed
sets, a topological space Y with a countable network and a measurable function f : X → Y which is not
almost continuous.
(i) Let (X, Σ, µ) be a semiﬁnite measure space and T a topology on X such that µ is inner regular with
respect to the closed sets. Let A ⊆ L
∞
(µ) be a normcompact set. Show that there is a set B of bounded
realvalued measurable functions on X such that (i) A = ¦f
•
: f ∈ B¦ (ii) B is normcompact in
∞
(X)
(iii) µ is inner regular with respect to ¦E : fE is continuous for every f ∈ B¦.
418 Notes Measurable functions and almost continuous functions 125
(j) Let µ be Lebesgue measure on [0, 1]. For t ∈ [0, 1] set u
t
= χ[0, t]
•
∈ L
0
(µ). Show that A = ¦u
t
: t ∈
[0, 1]¦ is normcompact in L
p
(µ) for every p ∈ [1, ∞[ and also compact for the topology of convergence in
measure on L
0
(µ). Show that if B is a set of measurable functions such that A = ¦f
•
: f ∈ B¦ then µ is
not inner regular with respect to ¦E : fE is continuous for every f ∈ B¦.
(k) Suppose that (I, ≤), ¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
and ¸f
ij
)
i≤j∈I
are such that (α) (I, ≤) is a nonempty
upwardsdirected partially ordered set (β) every (X
i
, T
i
, Σ
i
, µ
i
) is a completely regular Hausdorﬀ quasi
Radon probability space (γ) f
ij
: X
j
→ X
i
is a continuous inversemeasurepreserving function whenever
i ≤ j in I (δ) f
ij
f
jk
= f
ik
whenever i ≤ j ≤ k in I. Let X
t
i
be the support of µ
i
for each i; show that
f
ij
[X
t
j
] is a dense subset of X
t
i
whenever i ≤ j. Let Z
i
be the Stone
ˇ
Cech compactiﬁcation of X
t
i
and let
˜
f
ij
: Z
j
→ Z
i
be the continuous extension of f
ij
X
t
j
for i ≤ j; let ˜ µ
i
be the Radon probability measure on
Z
i
corresponding to µ
i
TX
t
i
(416V). Show that Z
i
,
˜
f
ij
satisfy the conditions of 418O, so that we have a
projective limit Z, ¸g
i
)
i∈I
, µ as in 418M.
(l) Suppose that (I, ≤), ¸(X
i
, Σ
i
, µ
i
))
i∈I
, X, ¸f
i
)
i∈I
and ¸f
ij
)
i≤j∈I
are such that (α) (I, ≤) is a non
empty upwardsdirected partially ordered set (β) every (X
i
, Σ
i
, µ
i
) is a probability space (γ) f
ij
: X
j
→ X
i
is inversemeasurepreserving whenever i ≤ j in I (δ) f
ij
f
jk
= f
ik
whenever i ≤ j ≤ k () f
i
: X → X
i
is a
function for every i ∈ I (ζ) f
i
= f
ij
f
j
whenever i ≤ j (η) whenever ¸i
n
)
n∈N
, ¸x
n
)
n∈N
are such that ¸i
n
)
n∈N
is a nondecreasing sequence in I, x
n
∈ X
i
n
for every n ∈ N and f
i
n
i
n+1
(x
n+1
) = x
n
for every n ∈ N, then
there is an x ∈ X such that f
i
n
(x) = x
n
for every n. Show that there is a probability measure on X such
that every f
i
is inversemeasurepreserving.
418 Notes and comments The message of this section is that measurable functions are dangerous, but
that almost continuous functions behave themselves. There are two fundamental problems with measurable
functions: a function x → (f(x), g(x)) may not be measurable when the components f and g are measurable
(419Xg), and an image measure under a measurable function can lose tightness, even when both domain
and codomain are Radon measure spaces and the function is inversemeasurepreserving (419Xh). (This is
the ‘image measure catastrophe’ mentioned in 235J and the notes to ¸343.) Consequently, as long as we
are dealing with measurable functions, we often have to impose strong conditions on the range spaces –
commonly, we have to restrict ourselves to separable metrizable spaces (418B, 418C), or something similar,
which indeed often means that a measurable function is actually almost continuous (418J, 433E). Indeed, for
functions taking values in metrizable spaces, ‘almost continuity’ is very close to ‘measurable with essentially
separable range’ (418G, 418J). The condition ‘separable and metrizable’ is a little stronger than is strictly
necessary (418Yb, 418Yc, 418Yg), but covers the principal applications other than 433E. If we keep the
‘metrizable’ we can very substantially relax the ‘separable’ (438E, 438F), and it is in fact the case that
a measurable function from a Radon measure space to any metrizable space is almost continuous (451S).
These extensions apply equally to the results in 418R418T (438Xg438Xh, 451Xn). But both take us deeper
into set theory than seems appropriate at the moment.
For almost continuous functions, the two problems mentioned above do not arise (418Dd, 418I, 418Xt).
Indeed we rather expect almost continuous functions to behave as if they were continuous. But we still
have to be careful. The limit of a sequence of almost continuous functions need not be almost continuous
(418Xd), unless the codomain is metrizable (418F); and if we have a function f from a topological measure
space to an uncountable product of topological spaces, it can happen that every coordinate of f is an almost
continuous function while f is not (418Xd again). But for many purposes, intuitions gained from the study
of measurable functions between Euclidean spaces can be transferred to general almost continuous functions.
Theorems 418L and 418M are of a quite diﬀerent kind, but seem to belong here as well as anywhere.
Even in the simplest application of 418L (when Y = [0, 1] with Lebesgue measure, and X ⊆ [0, 1]
2
is a closed
set meeting every vertical line) it is not immediately obvious that there will be a measure with the right
projection onto the horizontal axis, though there are at least two proofs which are easier than the general
case treated in 413N413O418L.
As I explain in 418N, the really interesting question concerning 418M is when, given the projective system
¸(X
i
, T
i
, Σ
i
, µ
i
))
i∈I
, ¸f
ij
)
i≤j∈I
, we can expect to ﬁnd X and ¸g
i
)
i∈I
satisfying the rest of the hypotheses,
and once past the elementary results 418O418Q this can be hard to determine. I describe a method in
126 Topologies and measures 418 Notes
418Yk which can sometimes be used, but (like the trick in 418Nf) it is too easy and too abstract to be often
illuminating. See 454G below for something rather deeper.
The results of 418R418T stand somewhat aside from anything else considered in this chapter, but they
form part of an important technique. A special case has already been mentioned in 253Yg. I do not discuss
vectorvalued measurable functions in this book, except incidentally, but 418R is one of the fundamental
results on their representation; it means, for instance, that if V is any of the Banach function spaces of
Chapter 36 we can expect to represent Bochner integrable V valued functions (253Yf) in terms of functions
on product spaces, because V will be continuously embedded in an L
0
space (367P). The measurealgebra
version in 418T will be very useful in Volume 5 when establishing relationships between properties of measure
spaces and corresponding properties of measure algebras.
419 Examples
In ¸216, I went much of the way to describing examples of spaces with all the possible combinations
of the properties considered in Chapter 21. When we come to topological measure spaces, the number of
properties involved makes it unreasonable to seek any such comprehensive list. I therefore content myself
with seven examples to indicate some of the boundaries of the theory developed here.
The ﬁrst example (419A) is supposed to show that the hypothesis ‘eﬀectively locally ﬁnite’ which appears
in so many of the theorems of this chapter cannot as a rule be replaced by ‘locally ﬁnite’. The next two
(419C419D) address technical questions concerning the deﬁnition of ‘Radon measure’, and show how small
variations in the deﬁnition can lead to very diﬀerent kinds of measure space. The fourth example (419E)
shows that the τadditive product measures of ¸417 are indeed new constructions. 419H is there to show that
extension theorems of the types proved in ¸415 and ¸417 cannot be taken for granted. The classical example
419K exhibits one of the obstacles to generalizations of Prokhorov’s theorem (418M, 418Q). Finally, I return
to the split interval (419L) to describe its standard topology and its relation to the measure introduced in
343J.
419A Example There is a locally compact Hausdorﬀ space X with a complete, σﬁnite, locally ﬁnite,
τadditive topological measure µ, inner regular with respect to the closed sets, which has a closed subset Y ,
of measure 1, such that the subspace measure µ
Y
on Y is not τadditive. In particular, µ is not eﬀectively
locally ﬁnite.
proof (a) Let Q be a countably inﬁnite set, not containing any ordinal. Fix an enumeration ¸q
n
)
n∈N
of Q,
and for A ⊆ Q set νA =
¦
1
n+1
: q
n
∈ A¦. Let 1 be the ideal ¦A : A ⊆ Q, νA < ∞¦. For any sets I, J,
say that I ⊆
∗
J if I ¸ J is ﬁnite; then ⊆
∗
is a reﬂexive transitive relation.
Let κ be the smallest cardinal of any family / ⊆ 1 for which there is no I ∈ 1 such that K ⊆
∗
I for
every K ∈ /. Then κ is uncountable. PPP (i) Of course κ is inﬁnite. (ii) If ¸I
n
)
n∈N
is any sequence in 1, then
for each n ∈ N we can ﬁnd a ﬁnite I
t
n
⊆ I
n
such that ν(I
n
¸ I
t
n
) ≤ 2
−n
; setting I =
n∈N
I
n
¸ I
t
n
, we have
νI ≤ 2 < ∞, while I
n
⊆
∗
I for every n. Thus κ > ω. QQQ
(b) There is a family ¸I
ξ
)
ξ<κ
in 1 such that (i) I
η
⊆
∗
I
ξ
whenever η ≤ ξ < κ (ii) there is no I ∈ 1 such
that I
ξ
⊆
∗
I for every ξ < κ. PPP Take a family ¸K
ξ
)
ξ<κ
in 1 such that there is no I ∈ 1 such that K
ξ
⊆
∗
I
for every ξ < κ. Choose ¸I
ξ
)
ξ<κ
in 1 inductively in such a way that
K
ξ
⊆
∗
I
ξ
, I
η
⊆
∗
I
ξ
for every η < ξ.
(This can be done because ¦K
ξ
¦ ∪ ¦I
η
: η < ξ¦ will always be a subset of 1 of cardinal less than κ.) If now
I
ξ
⊆
∗
I for every ξ < κ, then K
ξ
⊆
∗
I for every ξ < κ, so I / ∈ 1. QQQ
Hence, or otherwise, we see that κ is regular. PPP If A ⊆ κ and #(A) < κ, then there is an I ∈ 1 such
that I
ζ
⊆
∗
I for every ζ ∈ A; now there must be a ξ < κ such that I
ξ
,⊆
∗
I, in which case ζ < ξ for every
ζ ∈ A, and A is not coﬁnal with κ. QQQ
(c) Set X = Q∪ κ. (This is where it is helpful to have arranged at the start that no ordinal belongs to
Q, so that Q∩ κ = ∅.) Let T be the family of sets G ⊆ X such that
G∩ κ is open for the order topology of κ,
for every ξ ∈ G∩ κ ¸ ¦0¦ there is an η < κ such that I
ξ
¸ I
η
⊆
∗
G,
419A Examples 127
if 0 ∈ G then I
0
⊆
∗
G.
(i) This is a Hausdorﬀ topology on X. PPP (α) It is easy to check that X ∈ T, ∅ ∈ T and
( ∈ T for
every ( ⊆ T. (β) Suppose that G, H ∈ T. Then (G ∩ H) ∩ κ = (G ∩ κ) ∩ (H ∩ κ) is open for the order
topology of κ. If ξ ∈ G∩ H ∩ κ ¸ ¦0¦ there are η, ζ < ξ such that I
ξ
¸ I
η
⊆
∗
G and I
ξ
¸ I
ζ
⊆
∗
H, and now
α = max(η, ζ) < ξ, I
η
∪ I
ζ
⊆
∗
I
α
,
so
I
ξ
¸ I
α
⊆
∗
(I
ξ
¸ I
η
) ∩ (I
ξ
¸ I
ζ
) ⊆
∗
G∩ H.
Finally, if 0 ∈ G ∩ H then I
0
⊆
∗
G ∩ H. So G ∩ H ∈ T. Thus T is a topology on X. (γ) For any ξ < κ,
the set E
ξ
= (ξ + 1) ∪ I
ξ
is openandclosed for T; for any q ∈ Q, ¦q¦ is openandclosed. Since these sets
separate the points of X, T is Hausdorﬀ. QQQ
(ii) The sets E
ξ
of the last paragraph are all compact for T. PPP Let T be an ultraﬁlter on X containing
E
ξ
. (α) If a ﬁnite set K belongs to T, then T must contain ¦x¦ for some x ∈ K, and converges to x. So
suppose henceforth that T contains no ﬁnite set. (β) If E
0
∈ T, then for any open set G containing 0, E
0
¸G
is ﬁnite, so does not belong to T, and G ∈ T; as G is arbitrary, T → 0. (γ) If E
0
/ ∈ T, let η ≤ ξ be the least
ordinal such that E
η
∈ T. If G is an open set containing η, there are ζ
t
, ζ
tt
< η such that I
η
¸ I
ζ
⊆
∗
G,
]ζ
tt
, η] ⊆ G; so that E
η
¸ E
ζ
⊆
∗
G, where ζ = max(ζ
t
, ζ
tt
) < η. Now E
η
∈ T, E
ζ
/ ∈ T and (E
η
¸ E
ζ
) ¸ G / ∈ T,
so that G ∈ T. As G is arbitrary, T → η. (δ) As T is arbitrary, E
ξ
is compact. QQQ
(iii) It follows that T is locally compact. PPP For q ∈ Q, ¦q¦ is a compact open set containing q; for
ξ < κ, E
ξ
is a compact open set containing ξ. QQQ
(iv) The deﬁnition of T makes it clear that Q ∈ T, that is, that κ is a closed subset of X. We need
also to check that the subspace topology T
κ
on κ induced by T is just the order topology of κ. PPP (α) By
the deﬁnition of T, G ∩ κ is open for the order topology of κ for every G ∈ T. (β) For any ξ < κ, E
ξ
is
openandclosed for T so ξ +1 = E
ξ
∩κ is openandclosed for T
κ
. But this means that all sets of the forms
[0, ξ[ =
η<ξ
η + 1 and ]ξ, κ[ = κ ¸ (ξ + 1) belong to T
κ
; as these generate the order topology, every open
set for the order topology belongs to T
κ
, and the two topologies are equal. QQQ
(d) Now let T be the ﬁlter on X generated by the coﬁnal closed sets in κ. Because the intersection of
any sequence of closed coﬁnal sets in κ is another (4A1Bc), the intersection of any sequence in T belongs
to T. So
Σ = T ∪ ¦X ¸ F : F ∈ T¦
is a σalgebra of subsets of X, and we have a measure µ
1
: Σ → ¦0, 1¦ deﬁned by saying that µ
1
F = 1,
µ
1
(X ¸ F) = 0 if F ∈ T.
(e) Set µE = ν(E∩Q) +µ
1
E for E ∈ Σ. Then µ is a measure. Let us work through the properties called
for.
(i) If µE = 0 and A ⊆ E, then X ¸ A ⊇ X ¸ E ∈ T, so A ∈ Σ. Thus µ is complete.
(ii) µ(κ) = 1 and µ¦q¦ is ﬁnite for every q ∈ Q, so µ is σﬁnite.
(iii) If G ⊆ X is open, then κ ¸ G is closed, in the order topology of κ; if it is coﬁnal with κ, it belongs
to T; otherwise, κ ∩ G ∈ T. Thus in either case G ∈ Σ, and µ is a topological measure.
(iv) The next thing to note is that µG = ν(G ∩ Q) for every open set G ⊆ X. PPP If G / ∈ T this is
trivial. If G ∈ T, then κ¸G cannot be coﬁnal with κ, so there is a ξ < κ such that κ¸ξ ⊆ G. ??? If G∩Q ∈ 1,
then (G∩Q) ∪I
ξ
∈ 1. There must be a least η < κ such that I
η
,⊆
∗
(G∩Q) ∪I
ξ
; of course η > ξ, so η ∈ G.
There is some ζ < η such that I
η
¸ I
ζ
⊆
∗
G; but as I
ζ
⊆
∗
G ∪ I
ξ
, by the choice of η, we must also have
I
η
⊆
∗
G∪ I
ξ
, which is impossible. XXX Thus G∩ Q / ∈ 1 and µG = ν(G∩ Q) = ∞. QQQ
(v) It follows that µ is τadditive. PPP Suppose that ( ⊆ T is a nonempty upwardsdirected set with
union H. Then
µH = ν(H ∩ Q) = sup
G∈Ç
ν(G∩ Q) = sup
G∈Ç
µG
128 Topologies and measures 419A
because ν is τadditive (indeed, is a Radon measure) with respect to the discrete topology on Q. QQQ
(vi) µ is inner regular with respect to the closed sets. PPP Take E ∈ Σ, γ < µE. Then γ −µ
1
(E ∩ κ) <
ν(E∩Q), so there is a ﬁnite I ⊆ E∩Q such that νI > γ −µ(E∩κ). If µ
1
(E∩κ) = 0, then I ⊆ E is already
a closed set with µI > γ. Otherwise, E∩κ ∈ T, so there is a coﬁnal closed set F ⊆ κ such that F ⊆ E; now
F is closed in X (because κ is closed in X and the subspace topology on κ is the order topology), so I ∪ F
is closed, and µ(I ∪ F) > γ. As E and γ are arbitrary, µ is inner regular with respect to the closed sets. QQQ
(vii) µ is locally ﬁnite. PPP For any ξ < κ, E
ξ
is an open set containing ξ, and µE
ξ
= νI
ξ
is ﬁnite. For
any q ∈ Q, ¦q¦ is an open set containing q, and µ¦q¦ = ν¦q¦ is ﬁnite. QQQ
(viii) Now consider Y = κ. This is surely a closed set, and µκ = 1. I noted in (biv) above that the
subspace topology T
κ
is just the order topology of κ. But this means that ¦ξ : ξ < κ¦ is an upwardsdirected
family of negligible relatively open sets with union κ, so that the subspace measure µ
κ
= µ
1
is not τadditive.
(ix) It follows from 414K that µ cannot be eﬀectively locally ﬁnite; but it is also obvious from the work
above that κ is a measurable set, of nonzero measure, such that µ(κ ∩ G) = 0 whenever G is an open set
of ﬁnite measure.
419B Lemma For any nonempty set I, there is a dense G
δ
set in [0, 1]
I
which is negligible for the usual
measure on [0, 1]
I
.
proof Fix on some i
0
∈ I, and set π(x) = x(i
0
) for each x ∈ [0, 1]
I
, so that π is continuous and inverse
measurepreserving for the usual topologies and measures on [0, 1]
I
and [0, 1]. For each n ∈ N let G
n
⊇
[0, 1] ∩ Q be an open subset of [0, 1] with measure at most 2
−n
, so that π
−1
[G
n
] is an open set of measure
at most 2
−n
, and E =
n∈N
π
−1
[G
n
] is a G
δ
set of measure 0. If H ⊆ [0, 1]
I
is any nonempty open set,
its image π[H] is open in [0, 1], so contains some rational number, and meets
n∈N
G
n
; but this means that
H ∩ E ,= ∅, so E is dense.
419C Example (Fremlin 75b) There is a completion regular Radon measure space (X, T, Σ, µ) such
that
(i) there is an E ∈ Σ such that µ(F´E) > 0 for every Borel set F ⊆ X, that is, not every element of the
measure algebra of µ can be represented by a Borel set;
(ii) µ is not outer regular with respect to the Borel sets;
(iii) writing ν for the restriction of µ to the Borel σalgebra of X, ν is a locally ﬁnite, eﬀectively locally
ﬁnite, tight (that is, inner regular with respect to the compact sets) τadditive topological measure, and
there is a set Y ⊆ X such that the subspace measure ν
Y
is not semiﬁnite.
proof (a) For each ξ < ω
1
set X
ξ
= [0, 1]
ω
1
\ξ
, and take µ
ξ
to be the usual measure on X
ξ
; write Σ
ξ
for its
domain. Note that µ
ξ
is a completion regular Radon measure for the usual topology T
ξ
of X
ξ
(416U). Set
X =
ξ<ω
1
X
ξ
, and let µ be the direct sum measure on X (214K), that is, write
Σ = ¦E : E ⊆ X, E ∩ X
ξ
∈ Σ
ξ
for every ξ < ω
1
¦,
µE =
ξ<ω
1
µ
ξ
(E ∩ X
ξ
) for every E ∈ Σ.
Then µ is a complete locally determined (in fact, strictly localizable) measure on X. Write Σ for its domain.
(b) For each η < ω
1
let ¸β
ξη
)
ξ≤η
be a summable family of strictly positive real numbers with β
ηη
= 1
(4A1P). Deﬁne g
η
: X →R by setting
g
η
(x) =
1
β
ξη
x(η) if x ∈ X
ξ
where ξ ≤ η,
= 0 if x ∈ X
ξ
where ξ > η.
Now deﬁne f : X → ω
1
R
ω
1
by setting
f(x) = (ξ, ¸g
η
(x))
η<ω
1
)
if x ∈ X
ξ
. Note that f is injective. Let T be the topology on X deﬁned by f, that is, the family ¦f
−1
[W] :
W ⊆ ω
1
R
ω
1
is open¦, where ω
1
and R
ω
1
are given their usual topologies (4A2S, 3A3K), and their product
419C Examples 129
is given its product topology. Because f is injective, T can be identiﬁed with the subspace topology on f[X];
it is Hausdorﬀ and completely regular.
(c) For ξ, η < ω
1
, g
η
X
ξ
is continuous for the compact topology T
ξ
. Consequently fX
ξ
is continuous,
and the subspace topology on X
ξ
induced by T must be T
ξ
exactly. It follows that µ is a Radon measure for
T. PPP (i) We know already that µ is complete and locally determined. (ii) If G ∈ T then G∩ X
ξ
∈ T
ξ
⊆ Σ
ξ
for every ξ < ω
1
, so G ∈ Σ; thus µ is a topological measure. (iii) If E ∈ Σ and µE > 0, there is a ξ < ω
1
such that µ
ξ
(E ∩X
ξ
) > 0. Because µ
ξ
is a Radon measure, there is a T
ξ
compact set F ⊆ E ∩X
ξ
such that
µ
ξ
F > 0. Now F is Tcompact and µF > 0. As E is arbitrary, µ is tight (using 412B). (iv) If x ∈ X, take
that ξ < ω
1
such that x ∈ X
ξ
, and consider
G = f
−1
[(ξ + 1) ¦w : w ∈ R
ω
1
, w(ξ) < 2¦].
Because ξ + 1 is open in ω
1
, G ∈ T. Because g
ξ
(x) = x(ξ) ≤ 1, x ∈ G. Now for ζ ≤ ξ,
µ
ζ
(G∩ X
ζ
) = µ
ζ
¦x : x ∈ X
ζ
, g
ξ
(x) < 2¦
= µ
ζ
¦x : x ∈ X
ζ
, β
−1
ζξ
x(ξ) < 2¦
= µ
ζ
¦x : x ∈ X
ζ
, x(ξ) < 2β
ζξ
¦ ≤ 2β
ζξ
,
so
µG =
ζ≤ξ
µ
ζ
(G∩ X
ζ
) ≤ 2
ζ≤ξ
β
ζξ
< ∞.
As x is arbitrary, µ is locally ﬁnite, therefore a Radon measure. QQQ
We also ﬁnd that µ is completion regular. PPP If E ⊆ X and µE > 0, then there is a ξ < ω
1
such that
µ(E ∩ X
ξ
) > 0. Because µ
ξ
is completion regular, there is a set F ⊆ E ∩ X
ξ
, a zero set for T
ξ
, such that
µF > 0. Now X
ξ
is a G
δ
set in X (being the intersection of the open sets
η<ζ<ξ+1
X
ζ
for η < ξ, unless
ξ = 0, in which case X
ξ
itself is open), so F is a G
δ
set in X (4A2C(aiv)); being a compact G
δ
set in a
completely regular space, it is a zero set (4A2F(hv)).
Thus every set of positive measure includes a zero set of positive measure. So µ is inner regular with
respect to the zero sets (412B). QQQ
(d) The key to the example is the following fact: if G ⊆ X is open, then either there is a coﬁnal closed
set V ⊆ ω
1
such that G ∩ X
ξ
= ∅ for every ξ ∈ V or ¦ξ : µ(G ∩ X
ξ
) ,= 1¦ is countable. PPP Suppose that
A = ¦ξ : G ∩ X
ξ
,= ∅¦ meets every coﬁnal closed set, that is, is stationary (4A1C). Then B = A ∩ Ω is
stationary, where Ω is the set of nonzero countable limit ordinals (4A1Bb, 4A1Cb). Let H ⊆ ω
1
R
ω
1
be
an open set such that G = f
−1
[H].
For each ξ ∈ B choose x
ξ
∈ G∩X
ξ
. Then f(x
ξ
) ∈ H, so there must be a ζ
ξ
< ξ, a ﬁnite set I
ξ
⊆ ω
1
, and
a δ
ξ
> 0 such that z ∈ H whenever z = (γ, ¸t
η
)
η∈ω
1
) ∈ ω
1
R
ω
1
, ζ
ξ
< γ ≤ ξ and [t
η
−g
η
(x)[ < δ
ξ
for every
η ∈ I
ξ
. Because ξ is a nonzero limit ordinal, ζ
t
ξ
= sup(¦ζ
ξ
¦ ∪ (I
ξ
∩ ξ)) < ξ.
By the PressingDown Lemma (4A1Cc), there is a ζ < ω
1
such that C = ¦ξ : ξ ∈ B, ζ
t
ξ
= ζ¦ is
uncountable. ??? Suppose, if possible, that ζ < η < ω
1
and µ(G ∩ X
η
) < 1. Then there is a measurable
subset F of X
η
¸ G, determined by coordinates in a countable set J ⊆ ω
1
¸ η, such that µF = µ
η
F > 0
(254Ob). Let ξ ∈ C be such that η < ξ and J ⊆ ξ, and take any y ∈ F. If we deﬁne y
t
∈ X
η
by setting
y
t
(γ) = y(γ) for γ ∈ ξ ¸ η,
= x
ξ
(γ) for γ ∈ ω
1
¸ ξ,
then y
t
∈ F. But also ζ
ξ
< η < ξ and ξ ¸ η ⊆ ξ ¸ ζ
t
ξ
is disjoint from I
ξ
, so g
γ
(y
t
) = g
γ
(x
ξ
) for every γ ∈ I
ξ
,
since both are zero if γ ≤ η and otherwise y
t
(γ) = x
ξ
(γ). By the choice of ζ
ξ
, I
ξ
we must have f(y
t
) ∈ H
and y
t
∈ F ∩ G; which is impossible. XXX
Thus µ(G∩ X
η
) = 1 for every η > ζ, as required by the second alternative. QQQ
(e) For each ξ < ω
1
, let 1
ξ
be the family of negligible meager subsets of X
ξ
. Then 1
ξ
is a σideal; note
that it contains every closed negligible set, because µ
ξ
is strictly positive. Set
T
ξ
= 1
ξ
∪ ¦X
ξ
¸ F : F ∈ 1
ξ
¦,
so that T
ξ
is a σalgebra of subsets of X
ξ
, containing every conegligible open set, and µ
ξ
F ∈ ¦0, 1¦ for every
F ∈ T
ξ
. Set
130 Topologies and measures 419C
T = ¦E : E ∈ Σ, ¦ξ : E ∩ X
ξ
/ ∈ T
ξ
¦ is nonstationary¦.
Then T is a σsubalgebra of Σ (because the nonstationary sets form a σideal of subsets of ω
1
, 4A1Cb),
and contains every open set, by (d); so includes the Borel σalgebra B of X.
If we set
E
ξ
= ¦x : x ∈ X
ξ
, x(ξ) ≤
1
2
¦ for each ξ < ω
1
, E =
ξ<ω
1
E
ξ
,
then E ∈ Σ. But if F ⊆ X is a Borel set, F ∈ T so µ(E´F) = ∞. This proves the property (i) claimed for
the example.
(f ) Next, for each ξ < ω
1
, take a negligible dense G
δ
set E
t
ξ
⊆ X
ξ
(419B). Set Y =
ξ<ω
1
E
t
ξ
, so
that µY = 0. If F ⊇ Y is a Borel set, then F ∩ X
ξ
⊇ E
ξ
/ ∈ 1
ξ
for every ξ < ω
1
, while F ∈ T, so
¦ξ : µ
ξ
(F ∩ X
ξ
) = 0¦ is nonstationary and µF = ∞. Thus µ is not outer regular with respect to the Borel
sets. Taking ν = µB, the subspace measure ν
Y
is not semiﬁnite. PPP We have just seen that ν
Y
Y = ν
∗
Y
is inﬁnite. If F ∈ B and νF < ∞, then A = ¦ξ : µ
ξ
(F ∩ X
ξ
) > 0¦ is countable, so F
0
=
ξ∈A
E
t
ξ
and
F
1
= F ¸
ξ∈A
X
ξ
are negligible Borel sets; since F ∩Y ⊆ F
0
∪F
1
, ν
Y
(F ∩Y ) = 0. But this means that ν
Y
takes no values in ]0, ∞[ and is not semiﬁnite. QQQ
Remark X here is not locally compact. But as it is Hausdorﬀ and completely regular, it can be embedded as
a subspace of a locally compact Radon measure space (X
t
, T
t
, Σ
t
, µ
t
) (416T). Now µ
t
still has the properties
(i)(iii).
419D Example (Fremlin 75b) There is a complete locally determined τadditive completion regular
topological measure space (X, T, Σ, µ) in which µ is tight and compact sets have ﬁnite measure, but µ is not
localizable.
proof (a) Let I be a set of cardinal greater than c. Set X = [0, 1]
I
. For i ∈ I, t ∈ [0, 1] set X
it
= ¦x :
x ∈ X, x(i) = t¦. Give X
it
its natural topology T
it
and measure µ
it
, with domain Σ
it
, deﬁned from the
expression of X
it
as [0, 1]
I\¦i]
¦t¦, each factor [0, 1] being given its usual topology and Lebesgue measure,
and the singleton factor ¦t¦ being given its unique (discrete) topology and (atomic) probability measure.
By 416U, µ
it
is a completion regular Radon measure. Set
T = ¦G : G ⊆ X, G∩ X
it
∈ T
it
for all i ∈ I, t ∈ [0, 1]¦,
Σ = ¦E : E ⊆ X, E ∩ X
it
∈ Σ
it
for all i ∈ I, t ∈ [0, 1]¦,
µE =
i∈I,t∈[0,1]
µ
it
(E ∩ X
it
) for every E ∈ Σ.
(Compare 216D.) Then it is easy to check that T is a topology. T is Hausdorﬀ because it is ﬁner(= larger)
than the usual topology S on X; because each T
it
is the subspace topology induced by S, it is also the
subspace topology induced by T. Next, the deﬁnition of µ makes it a locally determined measure; it is a
tight complete topological measure because every µ
it
is.
(b) If K ⊆ X is compact, µK < ∞. PPP??? Otherwise, M = ¦(i, t) : i ∈ I, t ∈ [0, 1], µ
it
(K ∩ X
it
) > 0¦
must be inﬁnite. Take any sequence ¸(i
n
, t
n
))
n∈N
of distinct elements of M. Choose a sequence ¸x
n
)
n∈N
in
K inductively, as follows. Given ¸x
m
)
m<n
, then set C
ni
= ¦x
m
(i) : m < n¦ for each i ∈ I, and
A
n
= ¦x : x ∈ X
i
n
t
n
, x(i) / ∈ C
ni
for i ∈ I ¸ ¦i
n
¦¦;
then µ
∗
i
n
t
n
A
n
= 1 (254Lb). Since µ
i
n
t
n
(X
it
∩ X
ju
) = 0 whenever (i, t) ,= (j, u), there must be some
x
n
∈ K ∩ A
n
¸
m,=n
X
i
m
,t
m
. Continue.
This construction ensures that that if i ∈ I and m < n, either i ,= i
n
so x
n
(i) / ∈ C
ni
and x
n
(i) ,= x
m
(i),
or i = i
n
,= i
m
and x
m
/ ∈ X
i
n
t
n
so x
n
(i) = t
n
,= x
m
(i), or i = i
m
= i
n
and x
n
(i) = t
n
,= t
m
= x
m
(i). But
this means that ¦x
n
: n ∈ N¦ is an inﬁnite set meeting each X
it
in at most one point, and is closed for T;
so ¸x
n
)
n∈N
has no cluster point for T, which is impossible. XXXQQQ
(c) µ is not localizable. PPP Fix on any k ∈ I and consider c = ¦X
kt
: t ∈ [0, 1]¦. ??? If E ∈ Σ is an
essential supremum for c, then E ∩ X
kt
must be µ
kt
conegligible for every t ∈ [0, 1]. We can therefore ﬁnd
a countable set J
t
⊆ I and a µ
kt
conegligible set F
t
⊆ E ∩ X
kt
, determined by coordinates in J
t
. At this
419F Examples 131
point recall that #(I) > c, so there is some j ∈ I ¸ (¦k¦ ∪
t∈[0,1]
J
t
). Since X
j0
∩X
kt
is negligible for every
t ∈ [0, 1], X
j0
∩ E must be negligible, and
_
1
0
νH
t
dt = 0, where
H
t
= ¦y : y ∈ [0, 1]
I\¦j,k]
, (y, 0, t) ∈ E¦
and ν is the usual measure on [0, 1]
I\¦j,k]
, identifying X with [0, 1]
I\¦j,k]
[0, 1] [0, 1]. But because F
t
is
determined by coordinates in I ¸ ¦j¦, we can identify it with F
t
t
[0, 1] ¦t¦ where F
t
t
is a νconegligible
subset of [0, 1]
I\¦j,k]
, and F
t
t
⊆ H
t
, so νH
t
= 1 for every t, which is absurd. XXX
Thus c has no essential supremum in Σ, and µ cannot be localizable. QQQ
(d) I have still to check that µ is completion regular. PPP If E ∈ Σ and µE > 0, there are i ∈ I, t ∈ [0, 1]
such that µ
it
(E ∩ X
it
) > 0, and an F ⊆ E ∩ X
it
, a zero set for the subspace topology of X
it
, such that
µ
it
F > 0. But now observe that X
it
is a zero set in X for the usual topology S, so that F is a zero set for
S (4A2Gc) and therefore for the ﬁner topology T. By 412B, this is enough to show that µ is inner regular
with respect to the zero sets. QQQ
Remark It may be worth noting that the topology T here is not regular. See Fremlin 75b, p. 106.
419E Example (Fremlin 76) Let (Z, S, T, ν) be the Stone space of the measure algebra of Lebesgue
measure on [0, 1], so that ν is a strictly positive completion regular Radon probability measure (411P). Then
the c.l.d. product measure λ on Z Z is not a topological measure, so is not equal to the τadditive product
measure
˜
λ, and
˜
λ is not completion regular.
proof Consider the sets W,
˜
W described in 346K. We have W ∈ Λ = domλ and
˜
W =
1, where
1 = ¦GH : G, H ⊆ Z are openandclosed, (GH) ¸ W is negligible¦.
˜
W is a union of open sets, therefore must be open in Z
2
. And λ
∗
˜
W ≤ λW. PPP??? Otherwise, there is a
V ∈ Λ such that V ⊆
˜
W and λV > λW. Now λ is tight, by 412Sb, so there is a compact set K ⊆ V such
that K ∈ Λ and λK > λW. There must be U
0
, . . . , U
n
∈ 1 such that K ⊆
i≤n
U
i
. But λ(U
i
¸ W) = 0 for
every i, so λ(K ¸ W) = 0 and λK ≤ λW. XXXQQQ
However, the construction of 346K arranged that λ
∗
˜
W should be 1 and λW strictly less than 1. So
λ
∗
˜
W < λ
∗
˜
W and
˜
W / ∈ Λ. Accordingly λ is not a topological measure and cannot be equal to the Radon
measure
˜
λ of 417P.
We know that λ is inner regular with respect to the zero sets (412Sc) and is deﬁned on every zero set
(417V), while
˜
λ properly extends λ. But this means that
˜
λ cannot be inner regular with respect to the zero
sets, by 412L, that is, cannot be completion regular.
419F Theorem (Rao 69) T(ω
1
ω
1
) = Tω
1
´
⊗Tω
1
, the σalgebra of subsets of ω
1
generated by ¦EF :
E, F ⊆ ω
1
¦.
proof (a) Because ω
1
≤ c, there is an injection h : ω
1
→ ¦0, 1¦
N
; set E
i
= ¦ξ : h(ξ)(i) = 1¦ for each i ∈ N.
(b) Suppose that A ⊆ ω
1
has countable vertical sections. Then A ∈ Tω
1
´
⊗Tω
1
. PPP Set B = A
−1
[ω
1
]
and for ξ ∈ B let f
ξ
: N → A[¦ξ¦] be a surjection. Set g
n
(ξ) = f
ξ
(n) for ξ ∈ B and n ∈ N, and
A
n
= ¦(ξ, f
ξ
(n)) : ξ ∈ B¦ for n ∈ N. Then
A
n
= ¦(ξ, η) : ξ ∈ B, η = g
n
(ξ)¦
= ¦(ξ, η) : ξ ∈ B, η < ω
1
, h(g
n
(ξ)) = h(η)¦
=
i∈N
_
¦(ξ, η) : ξ ∈ g
−1
n
[E
i
], η ∈ E
i
¦ ∪ ¦(ξ, η) : ξ ∈ B ¸ g
−1
n
[E
i
], η ∈ ω
1
¸ E
i
¦
_
∈ Tω
1
´
⊗Tω
1
.
So
A =
n∈N
A
n
∈ Tω
1
´
⊗Tω
1
. QQQ
(c) Similarly, if a subset of ω
1
ω
1
has countable horizontal sections, it belongs to Tω
1
´
⊗Tω
1
. But for
any A ⊆ ω
1
ω
1
, A = A
t
∪ A
tt
where
132 Topologies and measures 419F
A
t
= ¦(ξ, η) : (ξ, η) ∈ A, η ≤ ξ¦ has countable vertical sections,
A
tt
= ¦(ξ, η) : (ξ, η) ∈ A, ξ ≤ η¦ has countable horizontal sections,
so both A
t
and A
tt
belong to Tω
1
´
⊗Tω
1
and A also does.
419G Corollary (Ulam 30) Let Y be a set of cardinal at most ω
1
and µ a σﬁnite measure with domain
TY . Then µ is pointsupported: in particular, there is a countable conegligible set A ⊆ Y .
proof (a) Let µ
0
be the pointsupported part of µ, that is, µ
0
A =
y∈A
µ¦y¦ for every A ⊆ Y ; then µ
0
is a measure (112Bd) and so is ν = µ − µ
0
. If A ⊆ Y is countable, then of course µ
0
A = µA, so νA = 0.
Because µ is σﬁnite, so is ν.
(b) If Y is countable, we can stop. For the case in which #(Y ) = ω
1
, it is enough to consider the case in
which Y is actually equal to ω
1
. Let λ = ν ν be the product measure on ω
1
ω
1
. By 419F, the domain of
λ is the whole of T(ω
1
ω
1
); in particular, it contains the set V = ¦(ξ, η) : ξ ≤ η < ω
1
¦. Now by Fubini’s
theorem
λV =
_
νV [¦ξ¦]ν(dξ) =
_
ν(ω
1
¸ ξ)ν(dξ) = (νω
1
)
2
,
and also
λV =
_
νV
−1
[¦η¦]ν(dη) =
_
ν(η + 1)ν(dη) = 0.
So νω
1
= 0 and µ = µ
0
. Now ¦y : µ¦y¦ > 0¦ is a countable set (because µ is σﬁnite) and is conegligible.
Remark I ought to remark that this result, though not 419F, is valid for many other cardinals besides ω
1
;
see, in particular, 438C below. There will be more on this topic in Volume 5.
419H Example There is a complete probability space (X, Σ, µ) with a topology T such that µ is τ
additive and inner regular with respect to the Borel sets, T is generated by T ∩ Σ, but µ has no extension
to a topological measure.
proof (a) Set Y = ω
1
+ 1 = ω
1
∪ ¦ω
1
¦. Let T be the σalgebra of subsets of Y generated by the σideal of
countable subsets of ω
1
. Let ν be the probability measure with domain T deﬁned by the formula
νF =
1
2
#(F ∩ ¦0, ω
1
¦) for every F ∈ T.
Set
S = ¦∅, Y ¦ ∪ ¦H : 0 ∈ H ⊆ ω
1
¦.
This is a topology on Y , and every subset of Y is a Borel set for S; so ν is surely inner regular with respect
to the Borel sets.
Note that
¦¦0, α¦ : α < ω
1
¦ ∪ ¦Y ¦
is a base for S included in T.
(b) Let λ be the product probability measure on Y
N
, and Λ its domain; let Λ
0
⊆ Λ be the σalgebra of
subsets of Y
N
generated by sets of the form ¦y : y(i) ∈ F¦, where i ∈ N and F ∈ T. Let S
∗
be the product
topology on Y
N
, so that S
∗
∩ Λ
0
is a base for S
∗
. Then λ is inner regular with respect to the Borel sets
(412Uc, or otherwise). Deﬁne φ : ¦0, 1¦
N
→ Y by setting
φ(u)(n) = 0 if u(n) = 0,
= ω
1
if u(n) = 1.
Let ν
ω
be the usual measure on ¦0, 1¦
N
; then φ is inversemeasurepreserving for ν
ω
and λ, by 254H.
If V ∈ Λ
0
there is an α < ω
1
such that
whenever x ∈ V , y ∈ Y
N
and y(i) = x(i) whenever min(x(i), y(i)) < α, then y ∈ V .
419I Examples 133
PPP Let J be the family of sets V ∈ Λ with this property. Then J is a σalgebra of subsets of Y
N
including
the family ( of measurable cylinders, so includes Λ
0
. QQQ
(c) Let X be Y
N
¸ ¦0, ω
1
¦
N
. Then λ
∗
X = 1. PPP Take V ∈ Λ such that V ⊇ X and λV = λ
∗
X.
Then there is a V
t
∈ Λ
0
such that V ⊆ V
t
and λV
t
= λV (254Ff). As remarked in (b), there is an
α ∈ ]0, ω
1
[ such that y ∈ V
t
whenever x ∈ V
t
, y ∈ Y
N
and y(i) = x(i) whenever min(x(i), y(i)) < α. But as
¦0, α¦
N
¸ ¦0¦ ⊆ X ⊆ V
t
, ¦0, ω
1
¦
N
¸ ¦0¦ ⊆ V
t
. Accordingly
λ
∗
X = λV = λV
t
= ν
ω
φ
−1
[V
t
] = ν
ω
(¦0, 1¦
N
¸ ¦0¦) = 1. QQQ
Give X the subspace measure µ induced by λ, with domain Σ = ¦X ∩ W : W ∈ Λ¦, and the subspace
topology T induced by S
∗
. Then µX = λ
∗
X = 1, and µ(X ∩ V ) = λV for every V ∈ Λ. Σ∩ T is a base for
T, just because Λ ∩ S
∗
is a base for S
∗
; and µ is inner regular with respect to the Borel sets, by 412Pb.
(d) ??? Suppose, if possible, that µ is not τadditive. Let ( ⊆ Σ ∩ T be a nonempty upwardsdirected
set with union W ∈ Σ ∩ T such that µW > sup
G∈Ç
µG. Let (
0
be a countable subset of ( such that
µ(G¸
(
0
) = 0 for every G ∈ (
0
, and set W
1
= W ¸
(
0
; then µ(
(
0
) ≤ sup
G∈Ç
µG, so µW
1
> 0, while
µ(G∩ W
1
) = 0 for every G ∈ (.
Express W
1
as V
1
∩ X where V
1
∈ Λ. Then there is a V
2
∈ Λ
0
such that V
2
⊆ V
1
and λ(V
1
¸ V
2
) = 0, so
that
λV
2
= λV
1
= µW
1
> 0.
Let α < ω
1
be such that y ∈ V
2
whenever x ∈ V
2
, y ∈ Y
N
and y(i) = x(i) whenever min(x(i), y(i)) < α.
Let F ⊆ φ
−1
[V
2
] be a nonnegligible measurable selfsupporting set for the Radon measure ν
ω
(416U,
414F). By Lemma 345E, there are u, u
t
∈ F which diﬀer at exactly one coordinate; let i ∈ N be that
coordinate, and suppose that u(i) = 1, u
t
(i) = 0.
We know that φ(u) ∈ V
2
, and φ(u)(i) = ω
1
. Deﬁne x ∈ X by saying that x(j) = φ(u)(j) for j ,= i,
x(i) = α. Then x ∈ V
2
, by the choice of α. Now V
2
∩ X ⊆ W
1
⊆
(, so there is a G ∈ ( containing x. Let
V ⊆ Y
N
be a basic open set such that x ∈ X ∩ V ⊆ G; express V as
j∈N
H
j
where H
j
∈ S for every j
and J = ¦j : H
j
,= Y ¦ is ﬁnite. Observe that 0 ∈ H
j
and x(j) < ω
1
for every j ∈ J, just because 0 belongs
to every nonempty open subset of Y and the only open set containing ω
1
is Y itself. But this means that
u(j) = 0 for every j ∈ J ¸ ¦i¦, so u
t
(j) = 0 for every j ∈ J, and φ(u
t
) ∈ V ; thus the open set U = φ
−1
[V ]
meets F. Because F is selfsupporting,
0 < ν
ω
(F ∩ U) ≤ λ(V
2
∩ V ) = µ(V
2
∩ V ∩ X) ≤ µ(W
1
∩ G),
which is impossible. XXX
Thus µ is τadditive.
(e) ??? But suppose, if possible, that there were a topological measure ˜ µ on X agreeing with µ on every
open set in the domain of µ. For each i ∈ N, set π
i
(x) = x(i) for x ∈ X. Every subset of Y is a Borel set
for S; because π
i
is continuous, the image measure ˜ µπ
−1
i
is deﬁned on TY . Now #(Y ) = ω
1
, so there must
be a countable conegligible set (419G), and there must be some α
i
< ω
1
such that ˜ µπ
−1
i
(ω
1
¸ α
i
) = 0. On
the other hand,
˜ µπ
−1
i
(α
i
¸ ¦0¦) = µπ
−1
i
(α
i
¸ ¦0¦) = λ¦y : 0 < y(i) < α
i
¦ = ν(α
i
¸ ¦0¦) = 0,
so ˜ µπ
−1
i
(ω
1
¸ ¦0¦) = 0.
But the deﬁnition of X was exactly devised so that
X =
i∈N
π
−1
i
(ω
1
¸ ¦0¦),
so this is impossible. XXX
So we have the required example.
Remark I note that the topology of X is not regular. Of course the phenomenon here cannot arise with
regular spaces, by 415M.
419I For the next example it will be helpful to know some basic facts about Lebesgue measure which
seemed a little advanced for Volume 1 and for which I have not found a suitable place since.
134 Topologies and measures 419I
Lemma (a) If (X, T, Σ, µ) is an atomless Radon measure space and E ∈ Σ has nonzero measure, then
#(E) ≥ c.
(b) The number of closed subsets of R is c.
proof (a) There must be a compact set K ⊆ E such that µK > 0, and a selfsupporting closed K
t
⊆ K
such that µK
t
= µK. Because µ¦x¦ = 0 for every x ∈ X, K
t
can have no isolated points. So #(K
t
) ≥ c
(4A2G(iii)) and #(E) ≥ c.
(b) Write c for the family of closed subsets of R. Let ¸U
n
)
n∈N
enumerate a base for the topology of R
(4A2Ua). For each I ⊆ N, set
F
I
= R ¸
n∈I
U
n
.
Because every open set is expressible as a union of some U
i
, the map I → F
I
: TN → c is surjective. So
#(c) ≤ #(TN) = c. On the other hand, the map x → [0, x] : [0, 1] → c is injective, so #(c) ≥ #([0, 1]) = c.
Remark In fact, of course, the number of Borel subsets of R is c; see 4A1O.
419J The next result is a strengthening of 134D.
Lemma Let µ be Lebesgue measure on R, and H any measurable subset of R. Then there is a disjoint
family ¸A
α
)
α<c
of subsets of H such that H is a measurable envelope of every A
α
; in particular, µ
∗
A
α
= 0
and µ
∗
A
α
= µH for every α < c.
proof If µH = 0, we can take every A
α
to be empty; so suppose that µH > 0. Let c be the family of
closed subsets of H of nonzero measure. By 419Ib, #(c) ≤ c; enumerate c c as ¸(F
ξ
, α
ξ
))
ξ<c
(3A1Ca).
Choose ¸x
ξ
)
ξ<c
inductively, as follows. Given ¸x
η
)
η<ξ
, where ξ < c, F
ξ
has cardinal (at least) c, by 419Ia,
so cannot be included in ¦x
η
: η < ξ¦; take any x
ξ
∈ F
ξ
¸ ¦x
η
: η < ξ¦, and continue.
At the end of the induction, set
A
α
= ¦x
ξ
: ξ < c, α
ξ
= α¦.
Then the A
α
are disjoint just because the x
ξ
are distinct.
??? Suppose, if possible, that H is not a measurable envelope of A
α
for some α. Then µ
∗
(H ¸ A
α
) > 0
(413Ei), so there is a nonnegligible measurable set E ⊆ H ¸ A
α
. Now there is an F ∈ c such that F ⊆ E.
Let ξ < c be such that F = F
ξ
and α = α
ξ
; then x
ξ
∈ A
α
∩ F, which is impossible. XXX
Thus H is always a measurable envelope of A
α
. It follows from the deﬁnition of ‘measurable envelope’
that µ
∗
A
α
= µH. But also, if α < c, µ
∗
A
α
≤ µ
∗
(H ¸ A
α+1
), which is 0, as we have just seen. So we have a
suitable family.
419K Example (Blackwell 56) There are sequences ¸X
n
)
n∈N
, ¸T
n
)
n∈N
and ¸ν
n
)
n∈N
such that (i)
for each n, (X
n
, T
n
) is a separable metrizable space and ν
n
is a quasiRadon probability measure on Z
n
=
i≤n
X
i
(ii) for m ≤ n the canonical map π
mn
: Z
n
→ Z
m
is inversemeasurepreserving (iii) there is no
probability measure on Z =
i∈N
X
i
such that all the canonical maps from Z to Z
n
are inversemeasure
preserving.
proof Let ¸A
n
)
n∈N
be a disjoint sequence of subsets of [0, 1] such that µ
∗
([0, 1] ¸A
n
) = 0, that is, µ
∗
A
n
= 1
for every n, where µ is Lebesgue measure (using 419J). Set X
n
=
i≥n
A
i
, so that ¸X
n
)
n∈N
is a non
increasing sequence of sets of outer measure 1 with empty intersection. For each n ≥ 1, we have a map
f
n
: X
n
→ Z
n
deﬁned by setting f
n
(x)(i) = x for every i ≤ n, x ∈ X
n
. Let ν
n
be the image measure µ
X
n
f
−1
n
,
where µ
X
n
is the subspace measure on X
n
induced by µ. Note that f
n
is a homeomorphism between X
n
and the diagonal ∆
n
= ¦z : z ∈ Z
n
, z(i) = z(j) for all i, j ≤ n¦, which is a closed subset of Z
n
; so that ν
n
,
like µ
X
n
, is a quasiRadon probability measure.
If m ≤ n, then π
mn
is inversemeasurepreserving, where π
mn
(z)(i) = z(i) for z ∈ Z
n
, i ≤ m. PPP If
W ⊆ Z
m
is measured by ν
m
, then f
−1
m
[W] is measured by µ
X
m
, so is of the form X
m
∩ E where E is
Lebesgue measurable. But in this case f
−1
m
[π
−1
mn
[W]] = X
n
∩ E, so that
ν
n
(π
−1
mn
[W]) = µ
X
n
(f
−1
n
[π
−1
mn
[W]]) = µ
∗
(X
n
∩ E) = µE = µ
∗
(X
m
∩ E) = ν
m
W. QQQ
??? But suppose, if possible, that there is a probability measure ν on Z =
i∈N
X
i
such that π
n
: Z → Z
n
is inversemeasurepreserving for every n, where π
n
(z)(i) = z(i) for z ∈ Z, i ≤ n. Then
419Xc Examples 135
νπ
−1
n
[∆
n
] = ν
n
∆
n
= µ
X
n
f
−1
n
[∆
n
] = 1
for each n, so
1 = ν(
n∈N
π
−1
n
[∆
n
]) = ν¦z : z ∈ Z, z(i) = z(j) for all i, j ∈ N¦ = ν∅,
because
n∈N
X
n
= ∅; which is impossible. XXX
419L The split interval again (a) For the sake of an example in ¸343, I have already introduced the
‘split interval’ or ‘double arrow space’. As this construction gives us a topological measure space of great
interest, I repeat it here. Let I

be the set ¦a
+
: a ∈ [0, 1]¦ ∪ ¦a
−
: a ∈ [0, 1]¦. Order it by saying that
a
+
≤ b
+
⇐⇒ a
−
≤ b
+
⇐⇒ a
−
≤ b
−
⇐⇒ a ≤ b, a
+
≤ b
−
⇐⇒ a < b.
Then it is easy to check that I

is a totally ordered space, and that it is Dedekind complete. (If A ⊆ [0, 1]
is a nonempty set, then sup
a∈A
a
−
= (sup A)
−
, while sup
a∈A
a
+
is either (sup A)
+
or (sup A)
−
, depending
on whether sup A belongs to A or not.) Its greatest element is 1
+
and its least element is 0
−
. Consequently
the order topology on I

is a compact Hausdorﬀ topology (4A2Rc, 4A2Ri). Note that Q = ¦q
+
: q ∈
[0, 1] ∩ Q¦ ∪ ¦q
−
: q ∈ [0, 1] ∩ Q¦ is dense, because it meets every nontrivial interval in I

. By 4A2E(aii)
and 4A2Rn, I

is ccc and hereditarily Lindel¨of.
(b) If we deﬁne h : I

→ [0, 1] by writing h(a
+
) = h(a
−
) = a for every a ∈ [0, 1], then h is continuous,
because ¦x : h(x) < a¦ = ¦x : x < a
−
¦, ¦x : h(x) > a¦ = ¦x : x > a
+
¦ for every a ∈ [0, 1]. Now
we can describe the Borel sets of I

, as follows: a set E ⊆ I

is Borel iﬀ there is a Borel set F ⊆ [0, 1]
such that E´h
−1
[F] is countable. PPP Write Σ
0
for the family of subsets E of I

such that E´h
−1
[F]
is countable for some Borel set F ⊆ [0, 1]. It is easy to check that Σ
0
is a σalgebra of subsets of I

.
(If E´h
−1
[F] is countable, so is (I

¸ E)´h
−1
[[0, 1] ¸ F]; if E
n
´h
−1
[F
n
] is countable for every n, so is
(
n∈N
E
n
)´h
−1
[
n∈N
F
n
].) Because the topology of I

is Hausdorﬀ, every singleton set is closed, so every
countable set is Borel. Also h
−1
[F] is Borel for every Borel set F ⊆ [0, 1], because h is continuous (4A3Cd).
So if E´h
−1
[F] is countable for some Borel set F ⊆ [0, 1], E = h
−1
[F]´(E´h
−1
[F]) is a Borel set in I

.
Thus Σ
0
is included in the Borel σalgebra B of I

. On the other hand, if J ⊆ I

is an interval, h[J] is also
an interval, therefore a Borel set, and h
−1
[h[J]] ¸ J can contain at most two points, so J ∈ Σ
0
. If G ⊆ I

is
open, it is expressible as
i∈I
J
i
, where ¸J
i
)
i∈I
is a disjoint family of nonempty open intervals (4A2Rj). As
X is ccc, I must be countable. Thus G is expressed as a countable union of members of Σ
0
and belongs to
Σ
0
. But this means that the Borel σalgebra B must be included in Σ
0
, by the deﬁnition of ‘Borel algebra’.
So B = Σ
0
, as claimed. QQQ
(c) In 343J I described the standard measure µ on I

; its domain is the set Σ = ¦h
−1
[F]´M : F ∈
Σ
L
, M ⊆ I

, µ
L
h[M] = 0¦, where Σ
L
is the set of Lebesgue measurable subsets of [0, 1] and µ
L
is Lebesgue
measure, and µE = µ
L
h[E] for E ∈ Σ. h is inversemeasurepreserving for µ and µ
L
.
The new fact I wish to mention is: µ is a completion regular Radon measure. PPP I noted in 343Ja that it
is a complete probability measure; a fortiori, it is locally determined and locally ﬁnite. If G ⊆ I

is open,
then we can express it as h
−1
[F]´C for some Borel set F ⊆ [0, 1], countable set C ⊆ I

((b) above), so it
belongs to Σ; thus µ is a topological measure. If E ∈ Σ and µE > γ, then F = [0, 1] ¸ h[I

¸ E] is Lebesgue
measurable, and µE = µ
L
F. So there is a compact set L ⊆ F such that µ
L
L ≥ γ. But now K = h
−1
[L] ⊆ E
is closed, therefore compact, and µK ≥ γ. Moreover, L is a zero set, being a closed set in a metrizable space
(4A2Lc), so K is a zero set (4A2C(biv)). As E and γ are arbitrary, µ is inner regular with respect to the
compact zero sets, and is a completion regular Radon measure. QQQ
419X Basic exercises (a) Show that the topological space X of 419A is zerodimensional.
(b) Give an example of a compact Radon probability space in which every dense G
δ
set is conegligible.
(Hint: 411P.)
(c) In 419E, show that we can start from any atomless probability measure in place of Lebesgue measure
on [0, 1].
136 Topologies and measures 419Xd
>>>(d) (i) Show that if E ⊆ R
2
is Lebesgue measurable, with nonzero measure, then it cannot be covered
by fewer than c lines. (Hint: if H = ¦t : µ
1
E[¦t¦] > 0¦, where µ
1
is Lebesgue measure on R, then µ
1
H > 0,
so #(H) = c. So if we have a family L of lines, with #(L) < c, there must be a t ∈ H such that L
t
= ¦t¦R
does not belong to L. Now #(L
t
∩ E) = c and each member of L meets L
t
∩ E in at most one point.)
(ii) Show that there is a subset A of R
2
, of full outer measure, which meets every vertical line and every
horizontal line in exactly one point. (Hint: enumerate R as ¸t
ξ
)
ξ<c
and the closed sets of nonzero measure
as ¸F
ξ
)
ξ<c
. Choose ¸I
ξ
)
ξ<c
such that every I
ξ
is ﬁnite, no two points of I
ξ
∪
η<ξ
I
η
lie on any horizontal
or vertical line, the lines ¦t
ξ
¦ R and R ¦t
ξ
¦ both meet I
ξ
∪
η<ξ
I
η
, and I
ξ
meets F
ξ
.) (iii) Show that
there is a subset B of R
2
, of full outer measure, such that every straight line meets A in exactly two points.
(Hint: enumerate the straight lines in R
2
as ¸L
ξ
)
ξ<c
. Choose ¸J
ξ
)
ξ<c
such that every J
ξ
is ﬁnite, no three
points of J
ξ
∪
η<ξ
J
η
lie on any line, L
ξ
∩ (J
ξ
∪
η<ξ
J
η
) has just two points and J
ξ
∩ F
ξ
,= ∅.)
(e) Let µ be Lebesgue measure on [0, 1], and A ⊆ [0, 1] a set with inner measure 0 and outer measure
1; let T be the usual topology on [0, 1]. Let 1 be the family of sets I ⊆ A such that every point of A has
a neighbourhood containing at most one point of I. Show that S = ¦G ¸ I : G ∈ T, I ∈ 1¦ is a topology
on [0, 1] with a countable network. Show that the identity map from [0, 1] to itself, regarded as a map from
([0, 1], T, µ) to ([0, 1], S), is measurable but not almost continuous.
(f ) Let R

be the split line, that is, the set ¦a
+
: a ∈ R¦ ∪ ¦a
−
: a ∈ R¦, ordered by the rules in
419L. Show that R

is a Dedekind complete totally ordered set, so that its order topology T is locally
compact. Write µ
L
for Lebesgue measure on R, Σ
L
for its domain. Set h(a
+
) = h(a
−
) = a for a ∈ R,
Σ = ¦E : E ⊆ X, h[E] ∈ Σ
L
, µ
L
(h[E] ∩ h[X ¸ E]) = 0¦, µE = µ
L
h[E] for E ∈ Σ. Show that µ is a
completion regular Radon measure on R

and that h is continuous and inversemeasurepreserving for µ and
µ
L
. Show that the set ¦a
+
: a ∈ R¦, with the induced topology and measure, is isomorphic, as quasiRadon
measure space, to the rightfacing Sorgenfrey line (415Xc) with Lebesgue measure. Show that R

and the
Sorgenfrey line are hereditarily Lindel¨of.
(g) Let µ be Lebesgue measure on [0, 1] and Σ its domain. Let I

be the split interval. (i) Show
that the functions x → x
+
: [0, 1] → I

and x → x
−
: [0, 1] → I

are measurable. (Hint: 419Lb.) (ii)
Show that the function x → (x
+
, x
−
) : [0, 1] → (I

)
2
is not measurable. (Hint: the subspace topology on
¦(x
+
, x
−
) : x ∈ [0, 1]¦ is discrete.)
>>>(h) (i) Again writing I

for the split interval, show that the function which exchanges x
+
and x
−
for
every x ∈ [0, 1] is a Borel automorphism and an automorphism for the usual Radon measure ν on I

, but is
not almost continuous. (ii) Show that if we set f(x) = x
+
for x ∈ [0, 1], then f is inversemeasurepreserving
for Lebesgue measure µ
L
on [0, 1], but the image measure µ
L
f
−1
is not ν (nor, indeed, a Radon measure).
419Y Further exercises (a) In the example of 419E, show that there is a Borel set V ⊆ Z
2
such that
˜
λV = 0 and λ
∗
V = 1.
(b) Show that if / ⊆ Tω
1
is any family with #(/) ≤ ω
1
, there is a countably generated σalgebra Σ of
subsets of ω
1
such that / ⊆ Σ.
(c) Show that the split interval with its usual topology and measure has the simple product property
(417Yi).
419 Notes and comments The construction of the locally compact space X in 419A from the family
¸I
ξ
)
ξ<κ
is a standard device which has been used many times. The relation ⊆
∗
also appears in many
contexts. In eﬀect, part of the argument is taking place in the quotient algebra A = TQ/[Q]
<ω
, since
I ⊆
∗
J iﬀ I
•
⊆ J
•
in A; setting 1
#
= ¦I
•
: I ∈ 1¦, the cardinal κ is min¦#(A) : A ⊆ 1
#
has no upper
bound in 1
#
¦, the ‘additivity’ of the partially ordered set 1
#
. Additivities of partially ordered sets will be
one of the important concerns of Volume 5. I remark that we do not need to know whether (for instance)
κ = ω
1
or κ = c. This is an early taste of the kind of manoeuvre which has become a staple of settheoretic
analysis. It happens that the cardinal κ here is one of the most important cardinals of settheoretic measure
419 Notes Examples 137
theory; it is ‘the additivity of Lebesgue measure’ (529Xa), and under that name will appear repeatedly in
Chapter 52.
Observe that the measure µ of 419A only just fails to be a quasiRadon measure; it is locally ﬁnite instead
of being eﬀectively locally ﬁnite. And it would be a Radon measure if it were inner regular with respect to
the compact sets, rather than just with respect to the closed sets.
419C and 419D are relevant to the question: have I given the ‘right’ deﬁnition of Radon measure space?
419C is perhaps more important. Here we have a Radon measure space (on my deﬁnition) for which the
associated Borel measure is not localizable. (If A is the measure algebra of the measure µ, and Bthe measure
algebra of µB where B is the Borel σalgebra of X, then the embedding B ⊆ Σ induces an embedding of B
in A which represents B as an orderdense subalgebra of A, just because µ is inner regular with respect to
B. Property (i) of 419C shows that B ,= A, so B cannot be Dedekind complete in itself, by 314Ia.) Since
(I believe) localizable versions of measure spaces should almost always be preferred, I take this as strong
support for my prejudice in favour of insisting that ‘Radon’ measure spaces should be locally determined
as well as complete. Property (ii) of 419C is not I think of real signiﬁcance, but is further evidence, to be
added to 415Xh, that outer regularity is like an exoskeleton: it may inhibit growth above a certain size.
In 419D I explore the consequences of omitting the condition ‘locally ﬁnite’ from the deﬁnition of Radon
measure. Even if we insist instead that compact sets should have ﬁnite measure, we are in danger of getting
a nonlocalizable measure. Of course this particular space is pathological in terms of most of the criteria of
this chapter – for instance, every nonempty open set has inﬁnite measure, and the topology is not regular.
Perhaps the most important example in the section is 419E. The analysis of τadditive product measures
in ¸417 was long and diﬃcult, and if these were actually equal to the familiar product measures in all
important cases the structure of the theory would be very diﬀerent. But we ﬁnd that for one of the standard
compact Radon probability spaces of the theory, the c.l.d. product measure on its square is not a Radon
measure, and something has to be done about it.
I present 419H here to indicate one of the obstacles to any simpliﬁcation of the arguments in 417C and
417E. It is not signiﬁcant in itself, but it oﬀers a welcome excuse to describe some fundamental facts about
ω
1
(419F419G). Similarly, 419K asks for some elementary facts about Lebesgue measure (419I419J) which
seem to have got left out. This example really is important in itself, as it touches on the general problem of
representing stochastic processes, to which I will return in Chapter 45.
138 Descriptive set theory
Chapter 42
Descriptive set theory
At this point, I interpolate an auxiliary chapter, in the same spirit as Chapters 31 and 35 in the last
volume. As with Boolean algebras and Riesz spaces, it is not just that descriptive set theory provides
essential tools for modern measure theory; it also oﬀers deep intuitions, and for this reason demands study
well beyond an occasional glance at an appendix. Several excellent accounts have been published; the closest
to what we need here is probably Rogers 80; at a deeper level we have Moschovakis 80, and an admirable
recent treatment is Kechris 95. Once again, however, I indulge myself by extracting those parts of the
theory which I shall use directly, giving proofs and exercises adapted to the ideas I am trying to emphasize
in this volume and the next.
The ﬁrst section describes Souslin’s operation and its basic settheoretic properties up to the theory
of ‘constituents’ (421N421Q), mostly steering away from topological ideas, but with some remarks on σ
algebras and SouslinF sets. ¸422 deals with uscocompact relations and Kanalytic spaces, working through
the topological properties which will be useful later, and giving a version of the First Separation Theorem
(422I422J). ¸423 looks at ‘analytic’ or ‘Souslin’ spaces, treating them primarily as a special kind of K
analytic space, with the von NeumannJankow selection theorem (423N). ¸424 is devoted to ‘standard Borel
spaces’; it is largely a series of easy applications of results in ¸423, but there is one substantial theorem on
Borel measurable actions of Polish groups (424H).
421 Souslin’s operation
I introduce Souslin’s operation o (421B) and show that it is idempotent (421D). I describe alternative
characterizations of members of o(c), where c ⊆ TX, as projections of sets in N
N
X (421G421J). I brieﬂy
mention SouslinF sets (421J421L) and a special property of ‘inner Souslin kernels’ (421M). At the end of
the section I set up an abstract theory of ‘constituents’ for kernels of Souslin schemes and their complements
(421N421Q).
421A Notation Throughout this chapter, and frequently in the next, I shall regard a member of N as
the set of its predecessors, so that N
k
can be identiﬁed with the set of functions from k to N, and if φ ∈ N
N
and k ∈ N, we can speak of the restriction φk ∈ N
k
. In the same spirit, identifying functions with their
graphs, I can write ‘σ ⊆ φ’ when σ ∈ N
k
, φ ∈ N
N
and φ extends σ. On occasion I may write #(σ) for the
‘length’ of a ﬁnite function σ – again identifying σ with its graph – so that #(σ) = k if σ ∈ X
k
. And if
k = 0, identiﬁed with ∅, then the only function from k to X is the empty function, so X
0
becomes ¦∅¦.
I shall sometimes refer to the ‘usual topology of N
N
’; this is the product topology if each copy of N is
given its discrete topology. Writing I
σ
= ¦φ : φ ∈ N
N
, φ ⊇ σ¦ for σ ∈ S
∗
=
k∈N
N
k
, ¦I
σ
: σ ∈ S
∗
¦ is a
base for the topology of N
N
consisting of openandclosed sets. If σ ∈ N
k
, i ∈ N I write σ
i for the member
τ of N
k+1
such that τ(k) = i and τ(j) = σ(j) for j < k.
421B Deﬁnition Let S be the set
k≥1
N
k
. If c is a family of sets, I write o(c) for the family of sets
expressible in the form
_
φ∈N
N
k≥1
E
φk
for some family ¸E
σ
)
σ∈S
in c.
A family ¸E
σ
)
σ∈S
is called a Souslin scheme; the corresponding set
φ∈N
N
k≥1
E
φk
is its kernel; the
operation
¸E
σ
)
σ∈S
→
_
φ∈N
N
k≥1
E
φk
is Souslin’s operation or operation /. Thus o(c) is the family of sets obtainable from sets in c by
Souslin’s operation. If c = o(c), we say that c is closed under Souslin’s operation.
421Ce Souslin’s operation 139
Remark I should perhaps warn you that some authors use S
∗
=
k∈N
N
k
here in place of S; so that their
Souslin kernels are of the form
φ∈N
N
k≥0
E
φk
⊆ E
∅
. Consequently, for such authors, any member of
o(c) is included in some member of c. If c has a greatest member (or, fractionally more generally, if any
sequence in c is bounded above in c) this makes no diﬀerence; but if, for instance, c is the family of compact
subsets of a topological space, the two deﬁnitions of o may not quite coincide. I believe that on this point,
for once, I am following the majority.
421C Elementary facts (a) It is worth noting straight away that if c is any family of sets, then
n∈N
E
n
and
n∈N
E
n
belong to o(c) for any sequence ¸E
n
)
n∈N
in c. PPP Set
F
σ
= E
σ(0)
for every σ ∈ S,
G
σ
= E
k
whenever k ∈ N, σ ∈ N
k+1
;
then
n∈N
E
n
=
φ∈N
N
k≥1
F
φk
∈ o(c),
n∈N
E
n
=
φ∈N
N
k≥1
G
φk
∈ o(c). QQQ
In particular, c ⊆ o(c). But note that there is no reason why E ¸ F should belong to o(c) for E, F ∈ c.
(b) Let X and Y be sets, and f : X → Y a function. Let ¸F
σ
)
σ∈S
be a Souslin scheme in TY , with
kernel B. Then f
−1
[B] is the kernel of the Souslin scheme ¸f
−1
[F
σ
])
σ∈S
. PPP
f
−1
[B] = f
−1
[
φ∈N
N
n≥1
F
φn
] =
φ∈N
N
n≥1
f
−1
[F
φn
]. QQQ
(c) Let X and Y be sets, and f : X → Y a function. Let T be a family of subsets of Y . Then
¦f
−1
[B] : B ∈ o(T)¦ = o(¦f
−1
[F] : F ∈ T¦).
PPP For a set A ⊆ X, A ∈ o(¦f
−1
[F] : F ∈ T¦) iﬀ there is some Souslin scheme ¸E
σ
)
σ∈S
in ¦f
−1
[F] : F ∈ T¦
such that A is the kernel of ¸E
σ
)
σ∈S
, that is, iﬀ there is some Souslin scheme ¸F
σ
)
σ∈S
in T such that A is
the kernel of ¸f
−1
[F
σ
])
σ∈S
, that is, iﬀ A = f
−1
[B] where B is the kernel of some Souslin scheme in T. QQQ
(d) Let X and Y be sets, and f : X → Y a surjective function. Let T be a family of subsets of Y . Then
o(T) = ¦B : B ⊆ Y, f
−1
[B] ∈ o(¦f
−1
[F] : F ∈ T¦)¦.
PPP If B ∈ o(T), then f
−1
[B] ∈ o(¦f
−1
[F] : F ∈ T¦), by (c) above. If B ⊆ Y and f
−1
[B] ∈ o(¦f
−1
[F] :
F ∈ T¦), then there is a Souslin scheme ¸F
σ
)
σ∈S
in T such that f
−1
[B] is the kernel of ¸f
−1
[F
σ
])
σ∈S
, that
is, f
−1
[B] = f
−1
[C] where C is the kernel of ¸F
σ
)
σ∈S
. Because f is surjective, B = C ∈ o(T). QQQ
(e) Souslin’s operation can be thought of as a projection operator, as follows. Let ¸E
σ
)
σ∈S
be a Souslin
scheme with kernel A. Set
R =
n≥1
σ∈N
n
I
σ
E
σ
,
writing I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦ as usual. Then R[N
N
] = A. PPP For any x, and any φ ∈ N
N
,
(φ, x) ∈ R ⇐⇒ for every n ≥ 1 there is a σ ∈ N
n
such that x ∈ E
σ
, φ ∈ I
σ
⇐⇒ x ∈ E
φn
for every n ≥ 1.
But this means that
x ∈ R[N
N
] ⇐⇒ there is a φ ∈ N
N
such that (φ, x) ∈ R
⇐⇒ there is a φ ∈ N
N
such that x ∈
n≥1
E
φn
⇐⇒ x ∈ A. QQQ
140 Descriptive set theory 421D
421D The ﬁrst fundamental theorem is that the operation o is idempotent.
Theorem (Souslin 17) For any family c of sets, o(c) is closed under Souslin’s operation.
proof (a) Let ¸A
σ
)
σ∈S
be a family in o(c), and set A =
φ∈N
N
k≥1
A
φk
; I have to show that A ∈ o(c).
For each σ ∈ S, let ¸E
στ
)
τ∈S
be a family in c such that A
σ
=
ψ∈N
N
m≥1
E
σ,ψm
. Then
A =
_
φ∈N
N
k≥1
_
ψ∈N
N
m≥1
E
φk,ψm
=
_
φ∈N
N
Ψ∈(N
N
)
N\{0}
k,m≥1
E
φk,ψ
k
m
,
writing Ψ = ¸ψ
k
)
k≥1
for Ψ ∈ (N
N
)
N\¦0]
. The idea of the proof is simply that N
N
(N
N
)
N\¦0]
is essentially
identical to N
N
, so that all we have to do is to organize new names for the E
στ
. But as it is by no means a
trivial matter to devise a coding scheme which really works, I give the details at length.
(b) The ﬁrst step is to note that S and S
2
are countable, so there is a sequence ¸H
n
)
n∈N
running over
¦E
στ
: σ, τ ∈ S¦. Next, choose any injective function q : N N → N ¸ ¦0¦ such that q(0, 0) = 1 and
q(0, 1) = 2. For k, m ≥ 1 set J
km
= ¦(i, 0) : i < k¦ ∪ ¦(i, k) : i < m¦, so that J
11
= ¦(0, 0), (0, 1)¦, and
choose a family ¸(k
n
, m
n
))
n≥3
running over (N ¸ ¦0¦)
2
such that q[J
k
n
,m
n
] ⊆ n for every n ≥ 3. (The pairs
(k
n
, m
n
) need not all be distinct, so this is easy to achieve.)
Now, for υ ∈ N
n
, where n ≥ 3, set F
υ
= E
στ
where
σ ∈ N
k
n
, σ(i) = υ(q(i, 0)) for i < k
n
,
τ ∈ N
m
n
, τ(i) = υ(q(i, k
n
)) for i < m
n
;
these are welldeﬁned because q[J
k
n
,m
n
] ⊆ n. For υ ∈ N
1
∪ N
2
, set F
υ
= H
υ(0)
.
(c) This deﬁnes a Souslin scheme ¸F
υ
)
υ∈S
in c. Let A
t
be its kernel, so that A
t
∈ o(c). The point is
that A
t
= A.
PPP (i) If x ∈ A, there must be φ ∈ N
N
, Ψ ∈ (N
N
)
N\¦0]
such that x ∈
k,m≥1
E
φk,ψ
k
m
. Choose θ ∈ N
N
such that
H
θ(0)
= E
φ1,ψ
1
1
,
θ(q(i, 0)) = φ(i) for every i ∈ N,
θ(q(i, k)) = ψ
k
(i) for every k ≥ 1, i ∈ N.
(This is possible because q : N
2
→N ¸ ¦0¦ is injective.) Now
F
θ1
= F
θ2
= H
θ(0)
= E
φ1,ψ
1
1
certainly contains x. And for n ≥ 3, F
θn
= E
στ
where σ(i) = θ(q(i, 0)) for i < k
n
, τ(i) = θ(q(i, k
n
)) for
i < m
n
, that is, σ = φk
n
and τ = ψ
k
n
m
n
, so again x ∈ F
θn
. Thus
x ∈
n≥1
F
θn
⊆ A
t
.
As x is arbitrary, A ⊆ A
t
.
(ii) Now take any x ∈ A
t
. Let θ ∈ N
N
be such that x ∈
n≥1
F
θn
. Deﬁne φ ∈ N
N
, Ψ ∈ (N
N
)
N\¦0]
by
setting
φ(i) = θ(q(i, 0)) for i ∈ N,
ψ
k
(i) = θ(q(i, k)) for k ≥ 1, i ∈ N.
If k, m ≥ 1, let n ≥ 3 be such that k = k
n
, m = m
n
. Then x ∈ F
θn
= E
στ
, where
σ(i) = θ(q(i, 0)) for i < k
n
, τ(i) = θ(q(i, k
n
)) for i < m
n
,
that is, σ = φk
n
= φk and τ = ψ
k
n
m
n
= ψ
k
m. As m and n are arbitrary,
x ∈
m,n≥1
E
φk,ψ
k
m
⊆ A.
As x is arbitrary, A
t
⊆ A. QQQ
Accordingly we must have A ∈ o(c), and the proof is complete.
421G Souslin’s operation 141
421E Corollary For any family c of sets, o(c) is closed under countable unions and intersections.
proof For 421Ca tells us that the union and intersection of any sequence in o(c) will belong to oo(c) =
o(c).
421F Corollary Let X be a set and c a family of subsets of X. Suppose that X and ∅ belong to o(c)
and that X ¸ E ∈ o(c) for every E ∈ c. Then o(c) includes the σalgebra of subsets of X generated by c.
proof The set
Σ = ¦F : F ∈ o(c), X ¸ F ∈ o(c)¦
is closed under complements (necessarily), contains ∅ (because ∅ and X belong to o(c)), and is also closed
under countable unions, by 421E. So it is a σalgebra; but the hypotheses also ensure that c ⊆ Σ, so that
the σalgebra generated by c is included in Σ and in o(c).
421G Proposition Let c be a family of sets such that ∅ ∈ c. For σ ∈ S =
k≥1
N
k
set I
σ
= ¦φ : φ ∈
N
N
, φ ⊇ σ¦. Then
o(c) = ¦R[N
N
] : R ∈ o(¦I
σ
E : σ ∈ S, E ∈ c¦)¦
= ¦R[N
N
] : R ∈ o(¦I
σ
E : σ ∈ S, E ∈ c¦), R
−1
[¦x¦] is closed for every x¦.
proof Set T = ¦I
σ
E : σ ∈ S, E ∈ c¦.
(a) Suppose ﬁrst that A ∈ o(c). Let ¸E
σ
)
σ∈S
be a Souslin scheme in c with kernel A. Set
R =
k≥1
σ∈N
k
I
σ
E
σ
.
Then R ∈ o(T), by 421E, and R[N
N
] = A, by 421Ce. Also
R
−1
[¦x¦] =
k≥1
¦I
σ
: σ ∈ N
k
, x ∈ E
σ
¦
is closed, for every x.
(b) Now suppose that A = R[N
N
] for some R ∈ o(T). Let ¸I
τ(σ)
E
σ
)
σ∈S
be a Souslin scheme in T
with kernel R. For k ≥ 1, σ ∈ N
k
set
F
σ
= E
σ
if
1≤n≤k
I
τ(σn)
,= ∅,
= ∅ otherwise.
Then ¸F
σ
)
σ∈S
is a Souslin scheme in c, so its kernel A
t
belongs to o(c).
The point is that A
t
= A. PPP (i) If x ∈ A, there are a φ ∈ N
N
such that (φ, x) ∈ R and a ψ ∈ N
N
such
that (φ, x) ∈
n≥1
I
τ(ψn)
E
ψn
. Now, for any k ≥ 1, we have
φ ∈
1≤n≤k
I
τ(ψn)
=
1≤n≤k
I
τ((ψk)n)
,
so that F
ψk
= E
ψk
contains x; thus x ∈
k≥1
F
ψk
⊆ A
t
. As x is arbitrary, A ⊆ A
t
. (ii) If x ∈ A
t
, take
ψ ∈ N
N
such that x ∈
n≥1
F
ψn
. In this case we must have F
ψk
,= ∅, so
1≤n≤k
I
τ(ψn)
,= ∅, for every
k ≥ 1. But what this means is that, setting τ
n
= τ(ψn) for each n ≥ 1, τ
n
(i) = τ
m
(i) whenever i ∈ N is
such that both are deﬁned. So ¦τ
n
: n ≥ 1¦ must have a common extension φ ∈ N
N
, and φ ∈
n≥1
I
τ(ψn)
.
Now
(φ, x) ∈
n≥1
I
τ(ψn)
E
ψn
⊆ R,
so x ∈ A. Thus A
t
⊆ A and the two are equal. QQQ
This shows that
¦R[N
N
] : R ∈ o(T)¦ ⊆ o(c),
and the proof is complete.
142 Descriptive set theory 421H
421H When the class c is a σalgebra, the last proposition can be extended.
Proposition Let X be a set, and Σ a σalgebra of subsets of X. Let B be the algebra of Borel subsets of
N
N
. For σ ∈ S =
k≥1
N
k
, set I
σ
= ¦φ : φ ∈ N
N
, φ ⊇ σ¦. Then
o(Σ) = ¦R[N
N
] : R ∈ B
´
⊗Σ¦
= ¦R[N
N
] : R ∈ o(¦I
σ
E : σ ∈ S, E ∈ Σ¦)¦
= ¦R[N
N
] : R ∈ o(B
´
⊗Σ)¦.
Notation Recall that B
´
⊗Σ is the σalgebra of subsets of N
N
X generated by ¦H E : H ∈ B, E ∈ Σ¦.
proof (a) Suppose ﬁrst that A ∈ o(Σ). As in 421G, let ¸E
σ
)
σ∈S
be a Souslin scheme in Σ with kernel A,
and set
R =
k≥1
σ∈N
k
I
σ
E
σ
,
so that A = R[N
N
]. Because every I
σ
is an openandclosed set in N
N
, R ∈ B
´
⊗Σ. Thus
o(Σ) ⊆ ¦R[N
N
] : R ∈ B
´
⊗Σ¦.
(b) Set T = ¦I
σ
E : σ ∈ S, E ∈ Σ¦. Then o(B
´
⊗Σ) = o(T). PPP If E ∈ Σ and σ ∈ N
k
then
(N
N
X) ¸ (I
σ
E) = (I
σ
(X ¸ E)) ∪
τ∈N
k
,τ,=σ
I
τ
X ∈ o(T).
Also
N
N
X =
σ∈N
1
I
σ
X, ∅ = I
τ
∅
(where τ is any member of S) belong to o(T). By 421F, o(T) includes the σalgebra Λ of sets generated
by T. Now if E ∈ Σ and H ⊆ N
N
is open, H =
σ∈T
I
σ
for some T ⊆ S; as T is necessarily countable,
H E =
σ∈T
I
σ
E ∈ Λ.
Since ¦F : F ⊆ N
N
, F E ∈ Λ¦ is a σalgebra of subsets of N
N
, and we have just seen that it contains all
the open sets, it must include B; thus F E ∈ Λ for every F ∈ B, E ∈ Σ. So B
´
⊗Σ ⊆ Λ ⊆ o(T), and
o(T) ⊆ o(B
´
⊗Σ) ⊆ oo(T) = o(T)
(421D). QQQ
(c) Now we have
o(Σ) ⊆ ¦R[N
N
] : R ∈ B
´
⊗Σ¦
(by (a))
⊆ ¦R[N
N
] : R ∈ o(B
´
⊗Σ)¦ = ¦R[N
N
] : R ∈ o(T)¦
(by (b))
= o(Σ)
by 421G.
421I There is a particularly simple description of sets obtainable by Souslin’s operation from closed sets
in a topological space.
Lemma Let X be a topological space and R ⊆ N
N
X a closed set. Then
R[A] =
φ∈A
n≥1
R[I
φn
].
for any A ⊆ N
N
, writing I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦ as usual. In particular, R[N
N
] is the kernel of the Souslin
scheme ¸R[I
σ
])
σ∈S
.
421M Souslin’s operation 143
proof Set
B =
φ∈A
n≥1
R[I
φn
].
(i) If x ∈ R[A], there is a φ ∈ A such that (φ, x) ∈ R. In this case, φ ∈ I
φn
so
x ∈ R[I
φn
] ⊆ R[I
φn
]
for every n, and x ∈ B. Thus R[A] ⊆ B. (ii) If x ∈ B, let φ ∈ A be such that x ∈ R[I
φn
] for every n ∈ N.
??? If (φ, x) / ∈ R, then (because R is closed) there are a σ ∈ S and an open G ⊆ X such that φ ∈ I
σ
, x ∈ G
and (I
σ
G) ∩ R = ∅. But this means that G∩ R[I
σ
] = ∅ so G∩ R[I
σ
] = ∅ and x / ∈ R[I
σ
]; which is absurd,
because σ = φn for some n ≥ 1. XXX Thus (φ, x) ∈ R and x ∈ R[A]. As x is arbitrary, B ⊆ R[A] and
B = R[A], as required.
421J Proposition Let X be a topological space, and T the family of closed subsets of X. Then a set
A ⊆ X belongs to o(T) iﬀ there is a closed set R ⊆ N
N
X such that A is the projection of R on X.
proof (a) Suppose that A ∈ o(T). Let ¸F
σ
)
σ∈S
be a Souslin scheme in T with kernel A. Set
R =
n≥1
σ∈N
n
I
σ
F
σ
.
For each n ≥ 1,
σ∈N
n
I
σ
F
σ
= (N
N
X) ¸
σ∈N
n
I
σ
(X ¸ F
σ
)
is closed in N
N
X, so R is closed; and the projection R[N
N
] is A, by 421Ce.
(b) Suppose that R ⊆ N
N
is a closed set with projection A. Then A is the kernel of the Souslin scheme
¸R[I
σ
])
σ∈S
, by 421I, so belongs to o(T).
421K Deﬁnition Let X be a topological space. A subset of X is a SouslinF set in X if it is obtainable
from closed subsets of X by Souslin’s operation; that is, is the projection of a closed subset of N
N
X.
For a subset of R
r
, or, more generally, of any Polish space, it is common to say ‘Souslin set’ for ‘SouslinF
set’; see 421Xm.
421L Proposition Let X be any topological space. Then every Baire subset of X is SouslinF.
proof Let Z be the family of zero sets in X. If F ∈ Z then X ¸ F is a countable union of zero sets
(4A2C(bvi)), so belongs to o(Z). By 421F, the σalgebra generated by Z is included in o(Z) ⊆ o(T),
where T is the family of closed subsets of X; that is, every Baire set is SouslinF.
421M Proposition Let c be any family of sets such that ∅ ∈ c and E ∪ E
t
,
n∈N
E
n
belong to c for
every E, E
t
∈ c and all sequences ¸E
n
)
n∈N
in c. (For instance, c could be the family of closed subsets of a
topological space, or a σalgebra of sets.) Let ¸E
σ
)
σ∈S
be a Souslin scheme in c, and K ⊆ N
N
a set which
is compact for the usual topology on N
N
. Then
φ∈K
n≥1
E
φn
∈ c.
proof Set A =
φ∈K
n≥1
E
φn
. For k ∈ N, set K
k
= ¦φk : φ ∈ K¦; note that K
k
⊆ N
k
is compact,
because φ → φk is continuous, therefore ﬁnite, because the topology of N
k
is discrete. Set
H =
k≥1
σ∈K
k
1≤n≤k
E
σn
.
Because c is closed under ﬁnite unions and countable intersections, H ∈ c. Now A = H. PPP (i) If x ∈ A,
take φ ∈ K such that x ∈ E
φn
for every n ≥ 1; then φk ∈ K
k
and x ∈
1≤n≤k
E
(φk)n
for every k ≥ 1,
so x ∈ H. Thus A ⊆ H. (ii) If x ∈ H, then for each k ∈ N we have a σ
k
∈ K
k
such that x ∈
1≤n≤k
E
σ
k
n
.
Choose φ
k
∈ K such that φ
k
k = σ
k
for each k. Now K is supposed to be compact, so the sequence ¸φ
k
)
k∈N
has a cluster point φ in K.
If n ≥ 1, then I
φn
is a neighbourhood of φ in N
N
, so must contain φ
k
for inﬁnitely many k; let k ≥ n be
such that φ
k
n = φn. In this case
x ∈ E
σ
k
n
= E
φ
k
n
= E
φn
.
As n is arbitrary,
144 Descriptive set theory 421M
x ∈
n≥1
E
φn
⊆ A.
As x is arbitrary, H ⊆ A and H = A, as claimed. QQQ
So A ∈ c.
*421N I now embark on preparations for the theory of ‘constituents’ of analytic and coanalytic sets. It
turns out that much of the work can be done in the abstract context of this section.
Trees and derived trees (a) Let T be the family of subsets T of S =
n≥1
N
n
such that σk ∈ T whenever
σ ∈ T and 1 ≤ k ≤ #(σ). Note that the intersection and union of any nonempty family of members of T
again belong to T . Members of T are often called trees.
(b) For T ∈ T , set
∂T = ¦σ : σ ∈ S, ∃ i ∈ N, σ
i ∈ T¦,
so that ∂T ∈ T and ∂T ⊆ T. Of course ∂T
0
⊆ ∂T
1
whenever T
0
, T
1
∈ T and T
0
⊆ T
1
.
(c) For T ∈ T , deﬁne ¸∂
ξ
T)
ξ<ω
1
inductively by setting ∂
0
T = T and, for ξ > 0, ∂
ξ
T = ∂(
η<ξ
∂
η
T). An
easy induction shows that ∂
ξ
T ∈ T , ∂
ξ
T ⊆ ∂
η
T and ∂
ξ+1
T = ∂(∂
ξ
T) whenever η ≤ ξ < ω
1
.
(d) For any T ∈ T , there is a ξ < ω
1
such that ∂
ξ
T = ∂
η
T whenever ξ ≤ η < ω
1
. PPP Set T
1
=
ξ<ω
1
∂
ξ
T.
For each σ ∈ S ¸ T
1
, there is a ξ
σ
< ω
1
such that σ / ∈ ∂
ξ
σ
T. Set ξ = sup¦ξ
σ
: σ ∈ S ¸ T
1
¦; because S is
countable, ξ < ω
1
, and we must now have ∂
ξ
T = T
1
, so that ∂
ξ
T = ∂
η
T whenever ξ ≤ η < ω
1
. QQQ
(e) For T ∈ T , its rank is the ﬁrst ordinal r(T) < ω
1
such that ∂
r(T)
T = ∂
r(T)+1
T; of course ∂
r(T)
T =
∂
η
T whenever r(T) ≤ η < ω
1
, and ∂(∂
r(T)
T) = ∂
r(T)
T.
(f ) For T ∈ T , the following are equiveridical: (α) ∂
r(T)
T ,= ∅; (β) there is a φ ∈ N
N
such that φn ∈ T
for every n ≥ 1. PPP (i) If σ ∈ ∂
r(T)
T then σ ∈ ∂(∂
r(T)
T) so there is an i ∈ N such that σ
i ∈ ∂
r(T)
T. We
can therefore choose ¸σ
n
)
n∈N
inductively so that σ
n
∈ ∂
r(T)
T and σ
n+1
properly extends σ
n
for every n. At
the end of the induction, φ =
n∈N
σ
n
belongs to N
N
and
φn = σ
n
n ∈ ∂
r(T)
T ⊆ T
for every n ≥ 1. (ii) If φ ∈ N
N
is such that φn ∈ T for every n ≥ 1, then an easy induction shows that
φn ∈ ∂
ξ
T for every ξ < ω
1
and every n ≥ 1, so that ∂
r(T)
T is nonempty. QQQ
(g) Now suppose that ¸A
σ
)
σ∈S
is a Souslin scheme. For any x we have a tree T
x
∈ T deﬁned by saying
that
T
x
= ¦σ : σ ∈ S, , x ∈
1≤i≤#(σ)
A
σi
¦.
Now the kernel of ¸A
σ
)
σ∈S
is just
A = ¦x : ∃ φ ∈ N
N
, x ∈
n≥1
A
φn
¦
= ¦x : ∃ φ ∈ N
N
, φn ∈ T
x
∀ n ≥ 1¦ = ¦x : ∂
r(T)
T ,= ∅¦
by (f).
The sets
¦x : x ∈ X ¸ A, r(T
x
) = ξ¦ = ¦x : x ∈ X, r(T
x
) = ξ, ∂
ξ
T
x
= ∅¦,
for ξ < ω
1
, are called constituents of X ¸ A. (Of course they should properly be called ‘the constituents of
the Souslin scheme ¸A
σ
)
σ∈S
’.)
*421Q Souslin’s operation 145
*421O Theorem Let X be a set and Σ a σalgebra of subsets of X. Let ¸A
σ
)
σ∈S
be a Souslin scheme
in Σ with kernel A, and for x ∈ X set
T
x
= ¦σ : σ ∈ S, x ∈
1≤i≤#(σ)
A
σi
¦ ∈ T
as in 421Ng.
(a) For every ξ < ω
1
and σ ∈ S, ¦x : x ∈ X, σ ∈ ∂
ξ
T
x
¦ ∈ Σ.
(b) For every ξ < ω
1
, ¦x : x ∈ A, r(T
x
) ≤ ξ¦ and ¦x : x ∈ X ¸ A, r(T
x
) ≤ ξ¦ belong to Σ. In particular,
all the constituents of X ¸ A belong to Σ.
proof (a) Induce on ξ. For ξ = 0, we have
¦x : x ∈ X, σ ∈ ∂
0
T
x
¦ = ¦x : x ∈ X, σ ∈ T
x
¦ =
1≤i≤#(σ)
A
σi
∈ Σ.
For the inductive step to ξ > 0, we have
¦x : σ ∈ ∂
ξ
T
x
¦ = ¦x : σ ∈ ∂(
η<ξ
∂
η
T
x
)¦ =
_
i∈N
¦x : σ
i ∈
η<ξ
∂
η
T
x
¦
=
_
i∈N
η<ξ
¦x : σ
i ∈ ∂
η
T
x
¦ ∈ Σ
because ξ is countable and all the sets ¦x : σ
i ∈ ∂
η
T
x
¦ belong to Σ by the inductive hypothesis.
(b) Now, given ξ < ω
1
, we see that r(T
x
) ≤ ξ iﬀ ∂
ξ+1
T
x
⊆ ∂
ξ
T
x
, so that if we set E
ξ
= ¦x : x ∈
X, r(T
x
) ≤ ξ¦ then
E
ξ
=
σ∈S
¦x : x ∈ X, σ ∈ ∂
ξ+1
T
x
or σ / ∈ ∂
ξ
T
x
¦
belongs to Σ. Now, given that x ∈ E
ξ
, so that ∂
r(T
x
)
T
x
= ∂
ξ
T
x
, 421Ng tells us that x ∈ A iﬀ ∅ ∈ ∂
ξ
T
x
; so
that E
ξ
∩ A and E
ξ
¸ A both belong to Σ.
Now the constituents of X ¸ A are the sets (E
ξ
¸ A) ¸
η<ξ
E
η
for ξ < ω
1
, so all belong to Σ.
*421P Corollary Let X be a set and Σ a σalgebra of subsets of X. If A ∈ o(Σ) then both A and X¸A
can be expressed as the union of at most ω
1
members of Σ.
proof In the language of 421O, we have
A =
ξ<ω
1
E
ξ
∩ A, X ¸ A =
ξ<ω
1
E
ξ
¸ A.
*421Q Lemma Let X be a set and ¸A
σ
)
σ∈S
and ¸B
σ
)
σ∈S
two Souslin schemes of subsets of X. Suppose
that whenever φ, ψ ∈ N
N
there is an n ≥ 1 such that
1≤i≤n
A
φi
∩ B
ψi
= ∅. For x ∈ X set
T
x
=
n≥1
¦σ : σ ∈ N
n
, x ∈
1≤i≤n
A
σi
¦
as in 421Ng, and let B be the kernel of ¸B
σ
)
σ∈S
. Then sup
x∈B
r(T
x
) < ω
1
.
proof For σ ∈ S set A
t
σ
=
1≤i≤#(σ)
A
σi
, B
t
σ
=
1≤i≤#(σ)
B
σi
. Then T
x
= ¦σ : σ ∈ S, x ∈ A
t
σ
¦ for each
x ∈ X, B is the kernel of ¸B
t
σ
)
σ∈S
, and for every φ, ψ ∈ N
N
there is an n ∈ N such that A
t
φn
∩ B
t
ψn
= ∅.
Deﬁne ¸Q
ξ
)
ξ<ω
1
inductively by setting
Q
0
= ¦(σ, τ) : σ, τ ∈ S, A
t
σ
∩ B
t
τ
,= ∅¦,
and, for 0 < ξ < ω
1
,
Q
ξ
= ¦(σ, τ) : σ, τ ∈ S, ∃ i, j ∈ N, (σ
i, τ
j) ∈
η<ξ
Q
η
¦.
Then the same arguments as in 421Na421Nd show that there is a ζ < ω
1
such that Q
ζ+1
= Q
ζ
. ??? If
Q
ζ
,= ∅, then, just as in 421Nf, there must be φ, ψ ∈ N
N
such that (φm, ψn) ∈ Q
ζ
⊆ Q
0
for every m,
n ≥ 1; but this means that A
t
φn
∩ B
t
ψn
,= ∅ for every n ≥ 1, which is supposed to be impossible. XXX
Now suppose that x ∈ B. Then there is a ψ ∈ N
N
such that x ∈ B
t
ψn
for every n ≥ 1. But this
means that (σ, ψn) ∈ Q
0
for every σ ∈ T
x
and every n ≥ 1. An easy induction shows that (σ, ψn) ∈ Q
ξ
whenever ξ < ω
1
, σ ∈ ∂
ξ
T
x
and n ≥ 1. But as Q
ζ
= ∅ we must have ∂
ζ
T
x
= ∅ and r(T
x
) ≤ ζ. Thus
sup
x∈B
r(T
x
) ≤ ζ < ω
1
, and the proof is complete.
146 Descriptive set theory 421X
421X Basic exercises (a) Let X be a set and c a family of subsets of X. (i) Show that ∅ ∈ o(c) iﬀ
there is a sequence in c with empty intersection. (ii) Show that X ∈ o(c) iﬀ there is a sequence in c with
union X.
(b) Let c be a family of sets and F any set. Show that
o(¦E ∩ F : E ∈ c¦) = ¦A∩ F : A ∈ o(c)¦,
o(¦E ∪ F : E ∈ c¦) = ¦A∪ F : A ∈ o(c)¦.
(c) Let c be a family of sets and F any set. Show that
o(c ∪ ¦F¦) = ¦F¦ ∪ ¦A∩ F : A ∈ o(c)¦ ∪ ¦B ∪ F : B ∈ o(c)¦
∪ ¦(A∩ F) ∪ B : A, B ∈ o(c)¦.
(d) Suppose that c is a family of sets with #(c) ≤ c. Show that #(o(c)) ≤ c. (Hint: #(c
S
) ≤
#((TN)
S
) = #(T(N S)).)
(e) Let c be the family of halfopen intervals [2
−n
k, 2
−n
(k + 1)[, where n ∈ N, k ∈ Z; let ( be the set of
open subsets of R; let T be the set of closed subsets of R; let / be the set of compact subsets of R; let B be
the Borel σalgebra of R. Show that o(c) = o(T) = o(() = o(/) = o(B). (Hint: 421F.)
(f ) Let 1 be the family ¦I
σ
: σ ∈
k∈N
N
k
¦ (421A); let ( be the set of open subsets of N
N
; let T be the
set of closed subsets of N
N
; let / be the set of compact subsets of N
N
; let B be the Borel σalgebra of N
N
.
Show that o(1) = o(T) = o(() = o(B), but that o(/) is strictly smaller than these. (Hint: if ¸K
n
)
n∈N
is any sequence in /, set φ(i) = 1 + sup
ψ∈K
i
ψ(i) for each i ∈ N, so that φ / ∈
n∈N
K
n
; hence show that
N
N
/ ∈ o(/).)
(g) Let X be a separable metrizable space with at least two points; let  be any base for its topology,
and B its Borel σalgebra. Show that o() = o(B). What can happen if #(X) ≤ 1?
(h) Let X be a topological space; let Z be the set of zero sets in X, ( the set of cozero sets, and Ba the
Baire σalgebra. Show that o(Z) = o(() = o(Ba).
(i) Let X be a set, c a family of subsets of X, and Σ the σalgebra of subsets of X generated by c. Show
that if #(c) ≤ c then #(Σ) ≤ c. (Hint: #(c
S
) = #(T(N N)) = c and Σ ⊆ o(c ∪ ¦X ¸ E : E ∈ c¦).)
(j) Let X be a topological space such that every open set is SouslinF. Show that every Borel set is
SouslinF.
(k) Let X be a topological space and B(X) its Borel σalgebra. Show that o(B(X)) is just the set of
projections on X of Borel subsets of N
N
X. (Hint: 4A3G.)
(l) Let X and Y be topological spaces, f : X → Y a continuous function and F ⊆ Y a SouslinF set.
Show that f
−1
[F] is a SouslinF set in X.
(m) Let X be any metrizable space; let ( be the set of open subsets of X, T the set of closed subsets,
and B the Borel σalgebra. Show that o(() = o(T) = o(B).
(n) Let us say that a Souslin scheme ¸E
σ
)
σ∈S
is regular if E
σ
⊆ E
τ
whenever σ, τ ∈ S, #(τ) ≤ #(σ)
and σ(i) ≤ τ(i) for every i < #(σ). Let c be a family of sets such that E ∪ F, E ∩ F ∈ c for all E, F ∈ c.
Show that every member of o(c) can be expressed as the kernel of a regular Souslin scheme in c. (Hint:
if ¸E
σ
)
σ∈S
is any Souslin scheme in c with kernel A, set F
σ
=
τ⊆σ
E
τ
, G
σ
=
τ≤σ
F
τ
, where τ ≤ σ if
τ(i) ≤ σ(i) for i < #(τ) = #(σ); show that A is the kernel of ¸F
σ
)
σ∈S
and of ¸G
σ
)
σ∈S
, using an idea from
421M for the latter.)
421 Notes Souslin’s operation 147
(o) Let X be a Hausdorﬀ topological space and ¸K
σ
)
σ∈S
a Souslin scheme in which every K
σ
is a compact
subset in X. Show that
φ∈K
n≥1
K
φn
is compact for any compact K ⊆ N
N
.
421Y Further exercises (a) Let X be a topological space, Y a Hausdorﬀ space and f : X → Y a
continuous function. Let / be the family of closed countably compact subsets of X. Show that for any
c ⊆ /,
¦f[A] : A ∈ o(c)¦ = o(¦f[E] : E ∈ c¦).
(b) Let X be a topological space, and Ba its Baire σalgebra. Show that o(Ba) is just the family of sets
expressible as f
−1
[B] where f is a continuous function from X to some metrizable space Y and B ⊆ Y is
SouslinF.
(c) Let X be a set, c a family of subsets of X, and Σ the smallest σalgebra of subsets of X including
c and closed under Souslin’s operation. Show that if #(c) ≤ c then #(Σ) ≤ c. (Hint: deﬁne ¸c
ξ
)
ξ<ω
1
by setting c
ξ
= o(¦X ¸ E : E ∈ c ∪
η<ξ
c
ξ
¦) for each ξ. Show that #(c
ξ
) ≤ c for every ξ and that
Σ =
ξ<ω
1
c
ξ
.)
(d) Let X be a compact space and A a SouslinF set in X. Show that there is a family ¸F
ξ
)
ξ<ω
1
of Borel
sets such that X ¸ A =
ξ<ω
F
ξ
and whenever B ⊆ X ¸ A is a SouslinF set there is a ξ < ω
1
such that
B ⊆ F
ξ
. (Hint: take F
ξ
= ¦x : r(T
x
) ≤ ξ¦ ¸ A as in 421Ob, and apply 421Q.)
421 Notes and comments In 111G, I deﬁned the Borel sets of R to be the members of the smallest
σalgebra containing every open set. In 114E, I deﬁned a set to be Lebesgue measurable if it behaves in
the right way with respect to Lebesgue outer measure. The latter formulation, at least, provides some sort
of testing principle to determine whether a set is Lebesgue measurable. But the deﬁnition of ‘Borel set’
does not. The only tool so far available for proving that a set E ⊆ R is not Borel is to ﬁnd a σalgebra
containing all open sets and not containing E; conversely, the only method we have for proving properties
of Borel sets is to show that a property is possessed by every member of some σalgebra containing every
open set. The revolutionary insight of Souslin 17 was a construction which could build every Borel set
from rational intervals. (See 421Xe.) For fundamental reasons, no construction of this kind can provide
all Borel sets without also producing other sets, and to actually characterize the Borel σalgebra a further
idea is needed (423Fa); but the class of analytic sets, being those constructible by Souslin’s operation from
rational intervals (or open sets, or closed sets, or Borel sets – the operation is robust under such variations),
turns out to have remarkable properties which make it as important in modern real analysis as the Borel
algebra itself.
The guiding principle of ‘descriptive set theory’ is that the properties of a set may be analysed in the
light of a construction for that set. Thus we can think of a closed set F ⊆ R as
R ¸
(q,q
)∈I
]q, q
t
[
where I ⊆ QQ. The principle can be eﬀective because we often have such descriptions in terms of objects
fundamentally simpler than the set being described. In the formula above, for instance, Q Q is simpler
than the set F, being a countable set with a straightforward description from N. The set T(Q Q) is
relatively complex; but a single subset I of QQ can easily be coded as a single subset of N (taking some
more or less natural enumeration of Q
2
as a sequence ¸(q
n
, q
t
n
))
n∈N
, and matching I with ¦n : (q
n
, q
t
n
) ∈ I¦).
So, subject to an appropriate coding, we have a description of closed subsets of R in terms of subsets of N.
At the most elementary level, this shows that there are at most c closed subsets of R. But we can also set
out to analyse such operations as intersection, union, closure in terms of these descriptions. The details are
complex, and I shall go no farther along this path here; but investigations of this kind are at the heart of
some of the most exciting developments of twentiethcentury real analysis.
The particular descriptive method which concerns us in the present section is Souslin’s operation. Starting
from a relatively simple class c, we proceed to the larger class o(c). The most fundamental property of o
is 421D: oo(c) = o(c). This means, for instance, that if c ⊆ o(T) and T ⊆ o(c), then o(c) will be equal
to o(T); consequently, diﬀerent classes of sets will often have the same Souslin closures, as in 421Xe421Xh.
148 Descriptive set theory 421 Notes
After a little practice you will ﬁnd that it is often easy to see when two classes c and T are at the same
level in this sense; but watch out for traps like the class of compact subsets of N
N
(421Xf) and odd technical
questions (421Xg).
Souslin’s operation, and variations on it, will be the basis of much of the next chapter; it has dramatic
applications in general topology and functional analysis as well as in real analysis and measure theory. An
important way of looking at the kernel of a Souslin scheme ¸E
σ
)
σ∈S
is to regard it as the projection on the
second coordinate of the corresponding set R =
k≥1
σ∈N
k
I
σ
E
σ
(421Ce). We ﬁnd that many other
sets R ⊆ N
N
X will also have projections in o(c) (421G, 421H). Let me remark that it is essential here
that the ﬁrst coordinate should be of the right type. In one sense, indeed, N
N
is the only thing that will do;
but its virtue transfers to analytic spaces, as we shall see in 423M423O below. We shall often want to deal
with members of o(c) which are most naturally deﬁned in terms of some such auxiliary space.
I have moved into slightly higher gear for 421N421Q because these are not essential for most of the work
of the next chapter. From the point of view of this section 421P is very striking but the signiﬁcance of 421Q
is unlikely to be apparent. It becomes important in contexts in which the condition
∀ φ, ψ ∈ N
N
∃ n ≥ 1,
1≤i≤n
A
φi
∩ B
ψi
= ∅
is satisﬁed for natural reasons. I will expand on these in the next two sections. In the meantime, I oﬀer
421Yd as an example of what 421O and 421Q together can tell us.
422 Kanalytic spaces
I introduce Kanalytic spaces, deﬁned in terms of uscocompact relations. The ﬁrst step is to deﬁne the
latter (422A) and give their fundamental properties (422B422E). I reach Kanalytic spaces themselves in
422F, with an outline of the most important facts about them in 422G422K.
422A Deﬁnition Let X and Y be Hausdorﬀ spaces. A relation R ⊆ X Y is uscocompact if
(α) R[¦x¦] is a compact subset of Y for every x ∈ X,
(β) R
−1
[F] is a closed subset of X for every closed set F ⊆ Y .
(Relations satisfying condition (β) are sometimes called ‘upper semicontinuous’.)
422B The following elementary remark will be useful.
Lemma Let X and Y be Hausdorﬀ spaces and R ⊆ XY an uscocompact relation. If H is an open subset
of Y including R[¦x¦] where x ∈ X, there is an open set G ⊆ X, containing x, such that R[G] ⊆ H.
proof Set G = X ¸ R
−1
[Y ¸ H]. Because Y ¸ H is closed, so is R
−1
[Y ¸ H], and G is open. Of course
R[G] ⊆ H, and x ∈ G because R[¦x¦] ⊆ H.
422C Proposition Let X and Y be Hausdorﬀ spaces. Then a subset R of X Y is an uscocompact
relation iﬀ whenever T is an ultraﬁlter on XY , containing R, such that the ﬁrstcoordinate image π
1
[[T]]
of T has a limit in X, then T has a limit in R.
proof Recall that, writing π
1
(x, y) = x and π
2
(x, y) = y for (x, y) ∈ X Y ,
π
1
[[T]] = ¦A : A ⊆ X, π
−1
1
[A] ∈ T¦ = ¦A : A ⊆ X, AY ∈ T¦
(2A1Ib), and that T → (x, y) iﬀ π
1
[[T]] → x and π
2
[[T]] → y (3A3Ic).
(a) Suppose that R is uscocompact and that T is an ultraﬁlter on XY , containing R, such that π
1
[[T]]
has a limit x ∈ X. ??? If T has no limit in R, then, in particular, it does not converge to (x, y) for any
y ∈ R[¦x¦]; that is, π
2
[[T]] does not converge to any point of R[¦x¦], that is, every point of R[¦x¦] belongs
to an open set not belonging to π
2
[[T]]. Because R[¦x¦] is compact, it is covered by a ﬁnite union of open
sets not belonging to π
2
[[T]]; but as π
2
[[T]] is an ultraﬁlter (2A1N), there is an open set H ⊇ R[¦x¦] such
that Y ¸ H ∈ π
2
[[T]].
Now 422B tells us that there is an open set G containing x such that R[G] ⊆ H. In this case, G ∈ π
1
[[T]]
so GY ∈ T; at the same time, X (Y ¸ H) ∈ T. So
422D Kanalytic spaces 149
R ∩ (GY ) ∩ (X (Y ¸ H)) ∈ T.
But this is empty, by the choice of G; which is intolerable. XXX
Thus T has a limit in R, as required.
(b) Now suppose that R has the property described.
(i) Let x ∈ X, and suppose that ( is an ultraﬁlter on Y containing R[¦x¦]. Set h(y) = (x, y) for
y ∈ Y ; then T = h[[(]] is an ultraﬁlter on XY containing R. The image π
1
[[T]] is just the principal ﬁlter
generated by ¦x¦, so certainly converges to x; accordingly T must converge to some point (x, y) ∈ R, and
( = π
2
[[T]] converges to y ∈ R[¦x¦]. As ( is arbitrary, R[¦x¦] is compact (2A3R).
(ii) Let F ⊆ Y be closed, and x ∈ R
−1
[F] ⊆ X. Consider
c = ¦R, X F¦ ∪ ¦GY : G ⊆ X is open, x ∈ G¦.
Then c has the ﬁnite intersection property. PPP If G
0
, . . . , G
n
are open sets containing x, then R
−1
[F] meets
G
0
∩ . . . ∩ G
n
in z say, and now (z, y) ∈ R ∩ (X F) ∩
i≤n
(G
i
Y ) for some y ∈ F. QQQ Let T be an
ultraﬁlter on X Y including c (4A1Ia). Because G Y ∈ c ⊆ T for every open set G containing x,
π
1
[[T]] → x, so T converges to some point (x, y) of R. Because X F is a closed set belonging to c ⊆ T,
y ∈ F and x ∈ R
−1
[F]. As x is arbitrary, R
−1
[F] is closed; as F is arbitrary, R satisﬁes condition (β) of
422A, and is uscocompact.
422D Lemma (a) Let X and Y be Hausdorﬀ spaces. If R ⊆ X Y is an uscocompact relation, then
R is closed in X Y .
(b) Let X and Y be Hausdorﬀ spaces. If R ⊆ X Y is an uscocompact relation and R
t
⊆ R is a closed
set, then R
t
is uscocompact.
(c) Let X and Y be Hausdorﬀ spaces. If f : X → Y is a continuous function, then its graph is an
uscocompact relation.
(d) Let ¸X
i
)
i∈I
and ¸Y
i
)
i∈I
be families of Hausdorﬀ spaces, and R
i
⊆ X
i
Y
i
an uscocompact relation
for each i. Set X =
i∈I
X
i
, Y =
i∈I
Y
i
and
R = ¦(x, y) : x ∈ X, y ∈ Y, (x(i), y(i)) ∈ R
i
for every i ∈ I¦.
Then R is uscocompact in X Y .
(e) Let X and Y be Hausdorﬀ spaces, and R ⊆ X Y an uscocompact relation. Then (i) R[K] is a
compact subset of Y for any compact subset K of X (ii) R[L] is a Lindel¨of subset of Y for any Lindel¨of
subset L of X.
(f) Let X, Y and Z be Hausdorﬀ spaces, and R ⊆ X Y , S ⊆ Y Z uscocompact relations. Then the
composition
S
◦
R = ¦(x, z) : there is some y ∈ Y such that (x, y) ∈ R and (y, z) ∈ S¦
is uscocompact in X Z.
(g) Let X and Y be Hausdorﬀ spaces and Y
0
any subset of Y . Then a relation R ⊆ XY
0
is uscocompact
when regarded as a relation between X and Y
0
iﬀ it is uscocompact when regarded as a relation between
X and Y .
proof (a) If (x, y) ∈ R, there is an ultraﬁlter T containing R and converging to (x, y) (4A2Bc). By 422C,
T must have a limit in R; but as X Y is Hausdorﬀ, this limit must be (x, y), and (x, y) ∈ R. As (x, y) is
arbitrary, R is closed.
(b) It is obvious that R
t
will satisfy the condition of 422C if R does.
(c) f[¦x¦] = ¦f(x)¦ is surely compact for every x ∈ X, and f
−1
[F] is closed for every closed set F ⊆ Y
because f is continuous.
(d) For i ∈ I, x ∈ X, y ∈ Y set φ
i
(x, y) = (x(i), y(i)). If T is an ultraﬁlter on X Y containing R such
that π
1
[[T]] has a limit in X, then
π
1
φ
i
[[T]] = ψ
i
π
1
[[T]]
150 Descriptive set theory 422D
has a limit in X
i
for every i ∈ I, writing ψ
i
(x) = x(i) for i ∈ I, x ∈ X. But φ
i
[[T]] contains R
i
, so has a
limit (x
0
(i), y
0
(i)) in X
i
Y
i
, for each i. Accordingly (x
0
, y
0
) is a limit of T in X Y (3A3Ic). As T is
arbitrary, R is uscocompact.
(e) For the moment, let L be any subset of X. Let H be a family of open sets in Y covering R[L]. Let
( be the family of those open sets G ⊆ X such that R[G] can be covered by ﬁnitely many members of H.
Then ( covers L. PPP If x ∈ L, then R[¦x¦] is a compact subset of R[L] ⊆
H, so there is a ﬁnite set H
t
⊆ H
covering R[¦x¦]. Now there is an open set G containing x such that R[G] ⊆
H
t
, by 422B. QQQ
(i) If L is compact, then there must be a ﬁnite subfamily (
t
of ( covering L; now R[L] ⊆ R[
(
t
] is
covered by ﬁnitely many members of H. As H is arbitrary, R[L] is compact.
(ii) If L is Lindel¨of, then there must be a countable subfamily (
t
of ( covering L; now R[L] ⊆ R[
(
t
]
is covered by countably many members of H. As H is arbitrary, R[L] is Lindel¨of.
(f ) If x ∈ X then R[¦x¦] ⊆ Y is compact, so (SR)[¦x¦] = S[R[¦x¦] is compact, by (ei). If F ⊆ Z is
closed then S
−1
[F] ⊆ Y is closed so (SR)
−1
[F] = R
−1
[S
−1
[F]] is closed.
(g)(i) Suppose that R is uscocompact when regarded as a subset of X Y
0
. Set S = ¦(y, y) : y ∈ Y
0
¦;
by (c), S is uscocompact when regarded as a subset of Y
0
Y , so by (f) R = SR is uscocompact when
regarded as a subset of X Y .
(ii) If R is uscocompact when regarded as a subset of X Y , and x ∈ X, then R[¦x¦] is a subset of
Y
0
which is compact for the topology of Y , therefore for the subspace topology of Y
0
. If F ⊆ Y
0
is closed
for the subspace topology, it is of the form F
t
∩ Y
0
for some closed F
t
⊆ Y , so R
−1
[F] = R
−1
[F
t
] is closed
in X. As x and F are arbitrary, R is uscocompact in X Y
0
.
422E The following lemma is actually very important in the structure theory of Kanalytic spaces (see
422Yb). It will be useful to us in 423C below.
Lemma Let X and Y be Hausdorﬀ spaces, and R ⊆ X Y an uscocompact relation. If X is regular, so
is R (in its subspace topology).
proof ??? Suppose, if possible, otherwise; that there are a closed set F ⊆ R and an (x, y) ∈ R ¸ F which
cannot be separated from F by open sets (in R). If G, H are open sets containing x, y respectively, then
R ∩ (G H), R ¸ (G H) are disjoint relatively open sets in R, so the latter cannot include F; that is,
F ∩ (G H) ,= ∅ whenever G, H are open and x ∈ G, y ∈ H. Accordingly there is an ultraﬁlter T on
X Y such that F ∩ (GH) ∈ T whenever G ⊆ X, H ⊆ Y are open sets containing x, y respectively. In
this case R ∈ T, and G ∈ π
1
[[T]] for every open set G containing x. Because the topology of X is regular,
every open set containing x includes G for some smaller open set G containing x, and belongs to π
1
[[T]];
thus π
1
[[T]] → x in X. Because R is uscocompact, T has a limit in R, which must be of the form (x, y
t
).
Because F ∈ T is closed (in R), (x, y
t
) ∈ F. But also y
t
∈ H for every open set H containing y, since XH
is a closed set belonging to T; because the topology of Y is Hausdorﬀ, y
t
must be equal to y, and (x, y) ∈ F,
which is absurd. XXX
422F Deﬁnition (Frol´ık 61) Let X be a Hausdorﬀ space. Then X is Kanalytic if there is an
uscocompact relation R ⊆ N
N
X such that R[N
N
] = X.
If X is a Hausdorﬀ space, we call a subset of X Kanalytic if it is a Kanalytic space in its subspace
topology.
422G Theorem (a) Let X be a Hausdorﬀ space. Then a subset A of X is Kanalytic iﬀ there is an
uscocompact relation R ⊆ N
N
X such that R[N
N
] = A.
(b) N
N
is Kanalytic.
(c) Compact Hausdorﬀ spaces are Kanalytic.
(d) If X and Y are Hausdorﬀ spaces and R ⊆ XY is an uscocompact relation, then R[A] is Kanalytic
whenever A ⊆ X is Kanalytic. In particular, a Hausdorﬀ continuous image of a Kanalytic Hausdorﬀ space
is Kanalytic.
(e) A product of countably many Kanalytic Hausdorﬀ spaces is Kanalytic.
422H Kanalytic spaces 151
(f) A closed subset of a Kanalytic Hausdorﬀ space is Kanalytic.
(g) A Kanalytic Hausdorﬀ space is Lindel¨of, so a regular Kanalytic Hausdorﬀ space is completely regular.
proof (a) A is Kanalytic iﬀ there is an uscocompact relation R ⊆ N
N
A with projection A. But a subset
of N
N
X with projection A is uscocompact in N
N
A iﬀ it is uscocompact in N
N
X, by 422Dg.
(b) The identity function from N
N
to itself is an uscocompact relation, by 422Dc.
(c) If X is compact, then R = N
N
X is an uscocompact relation (because R[¦φ¦] = X is compact for
every φ ∈ N
N
, while R
−1
[F] is either N
N
or ∅ for every closed F ⊆ X), so X = R[N
N
] is Kanalytic.
(d) By (a), there is an uscocompact relation S ⊆ N
N
X such that S[N
N
] = A. Now R
◦
S ⊆ N
N
Y is
uscocompact, by 422Df, and RS[N
N
] = R[A].
In particular, if X itself is Kanalytic and f : X → Y is a continuous surjection, f is an uscocompact
relation (422Dc), so Y = f[X] is Kanalytic.
(e) Let ¸X
i
)
i∈I
be a countable family of Kanalytic Hausdorﬀ spaces with product X. If I = ∅ then
X = ¦∅¦ is compact, therefore Kanalytic. Otherwise, choose for each i ∈ I an uscocompact relation
R
i
⊆ N
N
X
i
such that R
i
[N
N
] = X
i
. Set
R = ¦(φ, x) : φ ∈ (N
N
)
I
, x ∈ X, (φ(i), x(i)) ∈ R
i
for every i ∈ I¦.
By 422Dd, R is an uscocompact relation in (N
N
)
I
X, and it is easy to see that R[(N
N
)
I
] = X. But
(N
N
)
I ∼
= N
N·I
is homeomorphic to N
N
, because I is countable, so we can identify R with a relation in
N
N
X which is still uscocompact, and X is Kanalytic.
(f ) Let X be a Kanalytic Hausdorﬀ space and F a closed subset. Let R ⊆ N
N
X be an uscocompact
relation such that R[N
N
] = X. Set R
t
= R∩ (N
N
F). Then R
t
is a closed subset of R, so is uscocompact
(422Da). By (a), F = R
t
[N
N
] is Kanalytic.
(g) Let X be a Kanalytic Hausdorﬀ space. N
N
is Lindel¨of (4A2Ub), and there is an uscocompact
relation R such that R[N
N
] = X, so that X is Lindel¨of, by 422D(eii). 4A2H(bi) now tells us that if X is
regular it is completely regular.
422H Theorem (a) If X is a Hausdorﬀ space, then any Kanalytic subset of X is SouslinF in X.
(b) If X is a Kanalytic Hausdorﬀ space, then a subset of X is Kanalytic iﬀ it is SouslinF in X.
(c) For any Hausdorﬀ space X, the family of Kanalytic subsets of X is closed under Souslin’s operation.
proof (a) If A ⊆ X is Kanalytic, there is an uscocompact relation R ⊆ N
N
X such that R[N
N
] = A, by
422Ga. By 422Da, R is a closed set; so A is SouslinF by 421J.
(b) Now suppose that X itself is Kanalytic, and that A ⊆ X is SouslinF in X. Then there is a closed
set R ⊆ N
N
X such that R[N
N
] = A (421J, in the other direction). N
N
X is Kanalytic (422Gb, 422Ge),
and R is closed, therefore itself Kanalytic (422Gf); so its continuous image A is Kanalytic, by 422Gd.
(c)(i) The ﬁrst step is to show that the union of a sequence of Kanalytic subsets of X is Kanalytic.
PPP Let ¸A
n
)
n∈N
be a sequence of Kanalytic sets, with union A. For each n ∈ N, let R
n
⊆ N
N
X be an
uscocompact relation such that R
n
[N
N
] = A
n
. In (N N
N
) X let R be the set
¦((n, φ), x) : n ∈ N, (φ, x) ∈ R
n
¦.
If (n, φ) ∈ N N
N
, then R[¦(n, φ)¦] = R
n
[¦φ¦] is compact; if F ⊆ X is closed, then
R
−1
[F] = ¦(n, φ) : n ∈ N, φ ∈ R
−1
n
[F]¦
is closed in N N
N
. So R is uscocompact, and of course
R[N N
N
] =
n∈N
R
n
[N
N
] = A.
As N N
N
is Kanalytic (in fact, homeomorphic to N
N
), A is Kanalytic. QQQ
(ii) Now suppose that ¸A
σ
)
σ∈S
is a Souslin scheme consisting of Kanalytic sets with kernel A. Then
X
t
=
σ∈S
A
σ
is Kanalytic, by (i). By (a), every A
σ
is SouslinF when regarded as a subset of X
t
. But
since the family of SouslinF subsets of X
t
is closed under Souslin’s operation, by 421D, A is also SouslinF
in X
t
. By (b) of this theorem, A is Kanalytic. As ¸A
σ
)
σ∈S
is arbitrary, we have the result.
152 Descriptive set theory 422I
422I It seems that for the measuretheoretic results of ¸432, at least, the following result (the ‘First
Separation Theorem’) is not essential. However I do not think it possible to get a ﬁrm grasp on Kanalytic
and analytic spaces without knowing some version of it, so I present it here. It is most often used through
the forms in 422J and 422Xd below.
Lemma Let X be a Hausdorﬀ space. Let c be a family of subsets of X such that (i)
n∈N
E
n
,
n∈N
E
n
∈ c
whenever ¸E
n
)
n∈N
is a sequence in ∈ c (ii) whenever x, y are distinct points of X, there are disjoint E,
F ∈ c such that x ∈ int E and y ∈ int F. Then whenever A, B are disjoint nonempty Kanalytic subsets of
X, there are disjoint E, F ∈ c such that A ⊆ E and B ⊆ F.
proof (a) We need to know that if K, L are disjoint nonempty compact subsets of X, there are disjoint
E, F ∈ c such that K ⊆ int E and L ⊆ int F. PPP For any point (x, y) ∈ K L, we can ﬁnd disjoint E
xy
,
F
xy
∈ c such that x ∈ int E and y ∈ int F. Because L is compact and nonempty, there is for each x ∈ K
a nonempty ﬁnite set I
x
⊆ L such that L ⊆
y∈I
x
int F
xy
. Set E
x
=
y∈I
x
E
xy
, F
x
=
y∈I
x
F
xy
; then E
x
,
F
x
are disjoint members of c, x ∈ int E
x
and L ⊆ int F
x
. Because K is compact and not empty, there is a
nonempty ﬁnite set J ⊆ K such that K ⊆
x∈J
int E
x
. Set E =
x∈J
E
x
, F =
x∈J
F
x
; then E, F ∈ c,
E ∩ F = ∅, K ⊆ int E and L ⊆ int F, as required. QQQ
(b) Let Q, R ⊆ N
N
X be uscocompact relations such that Q[N
N
] = A and R[N
N
] = B. For each
σ ∈ S
∗
=
n∈N
N
n
, set
I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦, A
σ
= Q[I
σ
], B
σ
= R[I
σ
],
so that A = A
∅
and A
σ
=
i∈N
A
σ
i
for every σ.
(c) Write T for the set of pairs
¦(σ, τ) : σ, τ ∈ S
∗
and there are disjoint E, F ∈ c such that A
σ
⊆ E, B
τ
⊆ F¦.
If σ, τ ∈ S
∗
are such that (σ
i, τ
j) ∈ T for every i, j ∈ N, then (σ, τ) ∈ T. PPP For each i, j ∈ N take
disjoint E
ij
, F
ij
∈ c such that
A
σ
i
⊆ E
ij
, B
τ
j
⊆ F
ij
.
Then E =
i∈N
j∈N
E
ij
, F =
i∈N
j∈N
F
ij
are disjoint and belong to c, and A
σ
⊆ E, B
τ
⊆ F. So
(σ, τ) ∈ T. QQQ
(d) ??? Now suppose, if possible, that there are no disjoint E, F ∈ c such that A ⊆ E and B ⊆ F; that is,
that (∅, ∅) / ∈ T. By (c), used repeatedly, we can ﬁnd sequences ¸φ(i))
i∈N
, ¸ψ(i))
i∈N
such that (φn, ψn) / ∈ T
for every n ∈ N. Set K = Q[¦φ¦], L = R[¦ψ¦]. These are compact (because R is uscocompact) and disjoint
(because K ⊆ A and L ⊆ B). By (a), there are disjoint E, F ∈ c such that K ⊆ int E and L ⊆ int F.
By 422B, there are open sets U, V ⊆ N
N
such that
φ ∈ U, Q[U] ⊆ int E, ψ ∈ V , R[V ] ⊆ int F.
But now there is some n ∈ N such that I
φn
⊆ U and I
ψn
⊆ V , in which case
A
φn
⊆ E, B
ψn
⊆ F,
and (φn, ψn) ∈ T, contrary to the choice of φ and ψ. XXX
This contradiction shows that the lemma is true.
422J Corollary Let X be a Hausdorﬀ space and A, B disjoint Kanalytic subsets of X. Then there is
a Borel set which includes A and is disjoint from B.
proof Apply 422I with c the Borel σalgebra of X.
*422K I give the next step in the theory of ‘constituents’ begun in 412O412R.
Theorem Let X be a Hausdorﬀ space.
(i) Suppose that X is regular. Let A ⊆ X be a Kanalytic set. Then there is a nondecreasing family
¸E
ξ
)
ξ<ω
1
of Borel sets in X, with union X ¸ A, such that every SouslinF subset of X disjoint from A is
included in some E
ξ
.
422Xd Kanalytic spaces 153
(ii) Suppose that X is regular. Let A ⊆ X be a SouslinF set. Then there is a nondecreasing family
¸E
ξ
)
ξ<ω
1
of Borel sets in X, with union X ¸ A, such that every Kanalytic subset of X ¸ A is included in
some E
ξ
.
(iii) Let A ⊆ X be a Kanalytic set. Then there is a nondecreasing family ¸E
ξ
)
ξ<ω
1
of Borel sets in X,
with union X ¸ A, such that every Kanalytic subset of X ¸ A is included in some E
ξ
.
proof (a) The ﬁrst two parts depend on the following fact: if X is regular, R ⊆ N
N
X is uscocompact,
¸F
σ
)
σ∈S
is a Souslin scheme consisting of closed sets with kernel B, and R[N
N
] ∩ B = ∅, then for any φ,
ψ ∈ N
N
there is an n ≥ 1 such that R[I
φn
] ∩
1≤i≤n
F
ψi
is empty, where I write I
σ
= ¦θ : σ ⊆ θ ∈ N
N
¦ for
σ ∈ S =
n≥1
N
n
. PPP We know that K = R[¦φ¦] is a compact set disjoint from the closed set
n≥1
F
ψn
.
So there is some m ≥ 1 such that K ∩ F = ∅ where F =
1≤i≤m
F
ψi
. Because X is regular, there are
disjoint open sets G, H ⊆ X such that K ⊆ G and F ⊆ H (4A2F(hii)). Now R
−1
[X ¸ G] is a closed set
not containing φ, so there is some n such that R[I
φn
] ⊆ G. Of course we can take n ≥ m, and in this case
R[I
φn
] ∩
1≤i≤n
F
ψi
⊆ G∩ F = ∅,
as required. QQQ
(b)(i) Suppose that A ⊆ X is Kanalytic. Then there is an uscocompact set R ⊆ N
N
X such that
R[N
N
] = A, and R is closed (422Da), so that A is the kernel of the Souslin scheme ¸R[I
σ
])
σ∈S
(421I). For
x ∈ X set T
x
= ¦σ : σ ∈ S, x ∈ R[I
σ
]¦, as in 421Ng, and let r(T
x
) < ω
1
be the rank of the tree T
x
. Then
E
ξ
= ¦x : x ∈ X ¸ A, r(T
x
) ≤ ξ¦ is a Borel set for every ξ < ω
1
, by 421Ob. Now suppose that B ⊆ X ¸ A
is a SouslinF set. Then it is the kernel of a Souslin scheme ¸F
σ
)
σ∈S
consisting of closed sets. If φ, ψ ∈ N
N
then by (a) above there is an n ≥ 1 such that R[I
φn
] ∩
1≤i≤n
F
ψi
is empty. By 421Q, there must be some
ξ < ω
1
such that B ⊆ E
ξ
. So ¸E
ξ
)
ξ<ω
1
is a suitable family.
(ii) The other part is almost the same. Suppose that A ⊆ X is SouslinF. Then it is the kernel of a
Souslin scheme ¸F
σ
)
σ∈S
consisting of closed sets. For x ∈ X set
T
x
=
n≥1
¦σ : σ ∈ N
n
, x ∈
1≤i≤n
F
σi
¦,
and let r(T
x
) < ω
1
be the rank of the tree T
x
. Then E
ξ
= ¦x : x ∈ X ¸ A, r(T
x
) ≤ ξ¦ is a Borel set for
every ξ < ω
1
. Now let B ⊆ X ¸ A be a Kanalytic set. There is an uscocompact set R ⊆ N
N
X such that
R[N
N
] = B, and B is the kernel of the Souslin scheme ¸R[I
σ
])
σ∈S
. If φ, ψ ∈ N
N
then by (a) above there is
an n ≥ 1 such that
1≤i≤n
F
ψi
∩R[I
φn
] is empty. So there must be some ξ < ω
1
such that B ⊆ E
ξ
. Thus
here again ¸E
ξ
)
ξ<ω
1
is a suitable family.
(c) If X is not regular, we still have a version of the result in (a), as follows: if R, S ⊆ N
N
X and
R[N
N
] ∩S[N
N
] = ∅, then for any φ, ψ ∈ N
N
there is an n ≥ 1 such that R[Iφn] ∩S[I
ψn
] is empty. PPP This
time, R[¦φ¦] and S[¦ψ¦] are disjoint compact sets, so there are disjoint open sets G, H with R[¦φ¦] ⊆ G
and S[¦ψ¦] ⊆ H (4A2F(hi)). Now R[I
φn
] ⊆ G and S[I
ψn
] ⊆ H for all n large enough. QQQ
Now the argument of (bi), with F
σ
= S[I
σ
], gives part (iii).
422X Basic exercises (a) Let X and Y be Hausdorﬀ spaces and X
0
a closed subset of X. Show that a
relation R ⊆ X
0
Y is uscocompact when regarded as a relation between X
0
and Y iﬀ it is uscocompact
when regarded as a relation between X and Y .
(b) Show that a locally compact Hausdorﬀ space is Kanalytic iﬀ it is Lindel¨of iﬀ it is σcompact.
>>>(c) Prove 422Hc from ﬁrst principles, without using 421D. (Hint: if ¸R
σ
)
σ∈S
is a Souslin scheme of
uscocompact relations in N
N
X,
¦((φ, ¸ψ
σ
)
σ∈S
), x) : (ψ
φn
, x) ∈ R
φn
for every n¦
is uscocompact in (N
N
(N
N
)
S
) X.)
(d) Let X be a completely regular Hausdorﬀ space and A, B disjoint Kanalytic subsets of X. Show that
there is a Baire set including A and disjoint from B.
154 Descriptive set theory 422Xe
(e) Let X be a Hausdorﬀ Kanalytic space. (i) Show that every Baire subset of X is Kanalytic. (Hint:
136G.) (ii) Show that if X is regular, it is perfectly normal iﬀ it is hereditarily Lindel¨of iﬀ every open subset
of X is Kanalytic.
(f ) Let X be a Hausdorﬀ space in which every open set is Kanalytic. Show that every Borel set is
Kanalytic. (Hint: apply 136Xi to the family of Kanalytic subsets of X.)
422Y Further exercises (a) Let X be a completely regular Hausdorﬀ space, and βX its Stone
ˇ
Cech
compactiﬁcation. Show that X is Kanalytic iﬀ it is a SouslinF set in βX.
(b) Show that a Hausdorﬀ space is Kanalytic iﬀ it is a continuous image of a K
σδ
set in a compact
Hausdorﬀ space, that is, a set expressible as
m∈N
n∈N
K
mn
where every K
mn
is compact. (Hint: Write
/
∗
for the class of Hausdorﬀ continuous images of K
σδ
subsets of compact Hausdorﬀ spaces. (i) Show that
N
N
is a K
σδ
set in Y
N
, where Y is the onepoint compactiﬁcation of N. (ii) Show that if X is a compact
Hausdorﬀ space and R ⊆ N
N
X is closed, then R ∈ /
∗
. (iii) Show that if X is a compact Hausdorﬀ space,
then every SouslinF subset of X belongs to /
∗
. (iv) Show that if X is a regular Kanalytic Hausdorﬀ space,
then X ∈ /
∗
. (v) Show that if X is any Hausdorﬀ space and R ⊆ N
N
X is an uscocompact relation, then
R ∈ /
∗
. See Jayne 76.)
(c) Let X be a normal space and ( the family of countably compact closed subsets of X. Let A, B be
disjoint sets obtainable from ( by Souslin’s operation. (For instance, if X itself is countably compact, A
and B could be disjoint SouslinF sets.) Show that there is a Borel set including A and disjoint from B.
(Hint: adapt the proof of 422I.)
(d) Let X be a Hausdorﬀ space and ¸A
n
)
n∈N
a sequence of Kanalytic subsets of X such that
n∈N
A
n
= ∅.
Show that there is a sequence ¸E
n
)
n∈N
of Borel sets such that A
n
⊆ E
n
for every n ∈ N and
n∈N
E
n
= ∅.
(Hint: for each n ∈ N choose an uscocompact R
n
⊆ N
N
X with projection A
n
. Consider the set
T = ¦¸σ
n
)
n∈N
: ∃ Borel E
n
, R
n
[I
σ
n
] ⊆ E
n
∀ n,
n∈N
E
n
= ∅¦.)
(e) Explain how to prove 422J from 421Q, without using 422I.
(f ) Let X be a set and S, T two Hausdorﬀ topologies on X such that S ⊆ T and (X, T) is Kanalytic.
Show that S, T yield the same Kanalytic subspaces of X.
422 Notes and comments In a sense, this section starts at the deep end of its topic. ‘Descriptive set
theory’ originally developed in the context of the real line and associated spaces, and this remains the centre
of the subject. But it turns out that some of the same arguments can be used in much more general contexts,
and in particular greatly illuminate the theory of Radon measures on Hausdorﬀ spaces. I ﬁnd that a helpful
way to look at Kanalytic spaces is to regard them as a common generalization of compact Hausdorﬀ spaces
and SouslinF subsets of R; if you like, any theorem which is true of both these classes has a fair chance of
being true of all Kanalytic spaces. In the next section we shall come to the special properties of the more
restricted class of ‘analytic’ spaces, which are much closer to the separable metric spaces of the original
theory.
The phrase ‘uscocompact’ is neither elegant nor transparent, but is adequately established and (in view
of the frequency with which it is needed) seems preferable to less concise alternatives. If we think of a
relation R ⊆ X Y as a function x → R[¦x¦] from x to TY , then an uscocompact relation is one which
takes compact values and is ‘upper semicontinuous’ in the sense that ¦x : R[¦x¦] ⊆ H¦ is open for every
open set H ⊆ Y ; just as a realvalued function is upper semicontinuous if ¦x : f(x) < α¦ is open for every
α.
This is not supposed to be a book on general topology, and in my account of the topological properties
of Kanalytic spaces I have concentrated on facts which are useful when proving that spaces are Kanalytic,
on the assumption that these will be valuable when we seek to apply the results of ¸432 below. Other
properties are mentioned only when they are relevant to the measuretheoretic results which are my real
concern, and readers already acquainted with this area may be startled by some of my omissions. For a
proper treatment of the subject, I refer you to Rogers 80. As usual, however, I take technical details
423B Analytic spaces 155
seriously in the material I do cover. I hope you will not ﬁnd that such results as 422Dg and 422Ga try
your patience too far. I think a moment’s thought will persuade you that it is of the highest importance
that Kanalyticity (like compactness) is an intrinsic property. In contrast, the property of being ‘SouslinF’,
like the property of being closed, depends on the surrounding space. A completely regular Hausdorﬀ space
is compact iﬀ it must be a closed set in any surrounding Hausdorﬀ space iﬀ it is closed in its Stone
ˇ
Cech
compactiﬁcation; and it is Kanalytic iﬀ it must be a SouslinF set in any surrounding Hausdorﬀ space iﬀ it
is a SouslinF set in its Stone
ˇ
Cech compactiﬁcation (422Ya).
For regular spaces, 422K gives us another version of the First Separation Theorem. But this one is
simultaneously more restricted in its scope (it does not seem to have applications to Baire σalgebras, for
instance) and very much more powerful in its application. When all Borel sets are SouslinF, as in the
next section, it tells us something very important about the coﬁnal structure of the SouslinF subsets of the
complement of a Kanalytic set.
423 Analytic spaces
We come now to the original class of Kanalytic spaces, the ‘analytic’ spaces. I deﬁne these as continuous
images of N
N
(423A), but move as quickly as possible to their characterization as Kanalytic spaces with
countable networks (423C), so that many other fundamental facts (423E423G) can be regarded as simple
corollaries of results in ¸422. I give two versions of Lusin’s theorem on injective images of Borel sets
(423I), and a form of the von NeumannJankow measurable selection theorem (423N). I end with notes on
constituents of coanalytic sets (423P423Q).
423A Deﬁnition A Hausdorﬀ space is analytic or Souslin if it is either empty or a continuous image
of N
N
.
423B Proposition (a) A Polish space (deﬁnition: 4A2A) is analytic.
(b) A Hausdorﬀ continuous image of an analytic Hausdorﬀ space is analytic.
(c) A product of countably many analytic Hausdorﬀ spaces is analytic.
(d) A closed subset of an analytic Hausdorﬀ space is analytic.
(e) An analytic Hausdorﬀ space has a countable network consisting of analytic sets.
proof (a) Let X be a Polish space. If X = ∅, we can stop. Otherwise, let ρ be a metric on X, inducing its
topology, under which X is complete. For σ ∈ S
∗
=
n∈N
N
n
choose X
σ
⊆ X as follows. X
∅
= X. Given
that X
σ
is a closed nonempty subset of X, where σ ∈ S
∗
, then X
σ
is separable, because X is separable
and metrizable (4A2P(aiv)), and we can choose a sequence ¸x
σi
)
i∈N
in X
σ
such that ¦x
σi
: i ∈ N¦ is dense
in X
σ
. Set X
σ
i
= X
σ
∩ B(x
σi
, 2
−n
) for each i ∈ N, where B(x, δ) = ¦y : ρ(y, x) ≤ δ¦, and continue. Note
that because ¦x
σi
: i ∈ N¦ is dense in X
σ
, X
σ
=
i∈N
X
σ
i
, for every σ ∈ S
∗
.
For each φ ∈ N
N
, ¸X
φn
)
n∈N
is a nonincreasing sequence of nonempty closed sets, and diam(X
φn+1
)
≤ 2
−n+1
for every n. Because X is complete under ρ,
n∈N
X
φn
is a singleton ¦f(φ)¦ say. (f(φ) is the
limit of the Cauchy sequence ¸x
φn,φ(n)
)
n∈N
.) Thus we have a function f : N
N
→ X. f is continuous
because ρ(f(ψ), f(φ)) ≤ 2
−n+1
whenever φn + 1 = ψn + 1 (since in this case both f(ψ) and f(φ) belong
to X
φn+1
). f is surjective because, given x ∈ X, we can choose ¸φ(i))
i∈N
inductively so that x ∈ X
φn
for every n; at the inductive step, we have x ∈ X
φn
=
i∈N
X
(φn)
i
, so we can take φ(n) such that
x ∈ X
(φn)
φ(n)
= X
φn+1
.
Thus X is a continuous image of N
N
, as claimed.
(b) If X is an analytic Hausdorﬀ space and Y is a Hausdorﬀ continuous image of X, then either X is a
continuous image of N
N
and Y is a continuous image of N
N
, or X = ∅ and Y = ∅.
(c) Let ¸X
i
)
i∈I
be a countable family of analytic Hausdorﬀ spaces, with product X. Then X is Hausdorﬀ
(3A3Id). If I = ∅ then X = ¦∅¦ is a continuous image of N
N
, therefore analytic. If there is some i ∈ I such
that X
i
= ∅, then X = ∅ is analytic. Otherwise, we have for each i ∈ I a continuous surjection f
i
: N
N
→ X
i
.
Setting f(φ) = ¸f
i
(φ(i)))
i∈I
for φ ∈ (N
N
)
I
, f : (N
N
)
I
→ X is a continuous surjection. But (N
N
)
I ∼
= N
N·I
is
homeomorphic to N
N
, so X is analytic.
156 Descriptive set theory 423B
(d) Let X be an analytic Hausdorﬀ space and F a closed subset of X. Then F is Hausdorﬀ in its
subspace topology (4A2F(ai)). If X = ∅ then F = ∅ is analytic. Otherwise, there is a continuous surjection
f : N
N
→ X. Now H = f
−1
[F] is a closed subset of the Polish space N
N
, therefore Polish in its induced
topology (4A2Qd). By (a), H is analytic, so its continuous image F = f[H] is also analytic, by (b).
(e) Let X be an analytic Hausdorﬀ space. If it is empty then of course it has a countable network
consisting of analytic sets. Otherwise, there is a continuous surjection f : N
N
→ X. For σ ∈ S
∗
set
I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦; then ¦I
σ
: σ ∈ S
∗
¦ is a base for the topology of N
N
, so ¦f[I
σ
] : σ ∈ S
∗
¦ is a network
for the topology of X (see the proof of 4A2Nd). But I
σ
is homeomorphic to N
N
, so f[I
σ
] is analytic, for
every σ ∈ S
∗
, and ¦f[I
σ
] : σ ∈ S
∗
¦ is a countable network consisting of analytic sets.
423C Theorem A Hausdorﬀ space is analytic iﬀ it is Kanalytic and has a countable network.
proof (a) Let X be an analytic Hausdorﬀ space. By 423Be, it has a countable network. If X = ∅ then
surely it is Kanalytic. Otherwise, X is a continuous image of N
N
. But N
N
is Kanalytic (422Gb), so X also
is Kanalytic, by 422Gd.
(b) Now suppose that X is a Kanalytic Hausdorﬀ space and has a countable network.
(i) If X ⊆ N
N
then X is analytic. PPP Let R ⊆ N
N
X be an uscocompact relation such that R[N
N
] = X.
Then R is still uscocompact when regarded as a subset of N
N
N
N
(422Dg), so is closed in N
N
N
N
(422Da).
But N
N
N
N ∼
= N
N
is analytic, so R is in itself an analytic space (423Bd), and its continuous image X is
analytic, by 423Bb. QQQ
(ii) Now suppose that X is regular. By 4A2Ng, X has a countable network c consisting of closed sets.
Adding ∅ to c if need be, we may suppose that c , = ∅. Let ¸E
n
)
n∈N
be a sequence running over c. For each
n ∈ N, let ¸F
ni
)
i∈N
be a sequence running over ¦E
n
¦ ∪ ¦E : E ∈ c, E ∩ E
n
= ∅¦. Now consider the relation
R = ¦(φ, x) : φ ∈ N
N
, x ∈
n∈N
F
n,φ(n)
¦ ⊆ N
N
X.
(ααα) R is closed in N
N
X. PPP Because every F
ni
is closed,
(N
N
X) ¸ R =
i,n∈N
¦(φ, x): φ(n) = i and x / ∈ F
ni
¦
is open. QQQ
(βββ) R[N
N
] = X. PPP For every n ∈ N,
X ¸ E
n
=
¦E : E ∈ c, E ⊆ X ¸ E
n
¦
because c is a network and E
n
is closed, so
i∈N
F
ni
= X. So, given x ∈ X, we can ﬁnd for each n a φ(n)
such that x ∈ F
n,φ(n)
, and (φ, x) ∈ R. QQQ
(γγγ) R is the graph of a function. PPP??? Suppose that we have (φ, x) and (φ, y) in R where x ,= y.
Because the topology of X is Hausdorﬀ, there is an n ∈ N such that x ∈ E
n
and y / ∈ E
n
. But in this case
x ∈ E
n
∩ F
n,φ(n)
, so F
n,φ(n)
= E
n
, while y ∈ F
n,φ(n)
¸ E
n
, so F
n,φ(n)
,= E
n
; which is absurd. XXXQQQ
(δδδ) Set A = R
−1
[X], so that R is the graph of a function from A to X; in recognition of its new
status, give it a new name f. Then f is continuous. PPP Suppose that φ ∈ A and that x = f(φ) ∈ G, where
G ⊆ X is open. Then there is an n ∈ N such that x ∈ E
n
⊆ G. In this case x ∈ F
n,φ(n)
, because (φ, x) ∈ R,
so F
n,φ(n)
= E
n
. Now if ψ ∈ A and ψ(n) = φ(n), we must have
f(ψ) ∈ F
n,ψ(n)
= E
n
⊆ G.
Thus f
−1
[G] includes a neighbourhood of φ in A. As φ and G are arbitrary, f is continuous. QQQ
() At this point recall that X is Kanalytic. It follows that N
N
X is Kanalytic (422Ge), so that
its closed subset R is Kanalytic (422Gf) and A, which is a continuous image of R, is Kanalytic (422Gd).
But now A is a Kanalytic subset of N
N
, so is analytic, by (i) just above. And, ﬁnally, X is a continuous
image of A, so is analytic.
(iii) Thus any regular Kanalytic space with a countable network is analytic. Now suppose that X is
an arbitrary Kanalytic Hausdorﬀ space with a countable network. Let R ⊆ N
N
X be an uscocompact
423G Analytic spaces 157
relation such that R[N
N
] = X. Then R is a closed subset of N
N
X, so is itself a Kanalytic space with a
countable network. But it is also regular, by 422E. So R is analytic and its continuous image X is analytic.
This completes the proof.
423D Corollary (a) Any analytic Hausdorﬀ space is hereditarily Lindel¨of.
(b) In a regular analytic Hausdorﬀ space, closed subsets are zero sets and the Baire and Borel σalgebras
coincide.
(c) A compact subset of an analytic Hausdorﬀ space is metrizable.
(d) A metrizable space is analytic iﬀ it is Kanalytic.
proof (a)(c) These are true just because there is a countable network (4A2Nb, 4A2Na, 4A2Nh, 4A3Kb).
(d) Let X be a metrizable space. If X is analytic, of course it is Kanalytic. If X is Kanalytic, it is
Lindel¨of (422Gg) therefore separable (4A2Pd) and has a countable network (4A2P(aiii)), so is analytic.
423E Theorem (a) For any Hausdorﬀ space X, the family of subsets of X which are analytic in their
subspace topologies is closed under Souslin’s operation.
(b) Let (X, T) be an analytic Hausdorﬀ space. For a subset A of X, the following are equiveridical:
(i) A is analytic;
(ii) A is Kanalytic;
(iii) A is SouslinF;
(iv) A can be obtained by Souslin’s operation from the family of Borel subsets of X.
In particular, all Borel sets in X are analytic.
proof (a) Let X be a Hausdorﬀ space and / the family of analytic subsets of X. Let ¸A
σ
)
σ∈S
be a Souslin
scheme in / with kernel A. Then every A
σ
is Kanalytic, so A is Kanalytic, by 422Hc. Also every A
σ
has a
countable network, so A
t
=
σ∈S
A
σ
has a countable network (4A2Nc); as A ⊆ A
t
, A also has a countable
network (4A2Na) and is analytic.
(b) Because X has a countable network, so does A. So 423C tells us at once that (i) ⇐⇒ (ii). In
particular, X is Kanalytic, so 422Hb tells us that (ii) ⇐⇒ (iii). Of course (iii)⇒(iv).
Now suppose that G ⊆ X is open. Then G ∈ /. PPP If X = ∅ then G = ∅ is open. Otherwise, there is a
continuous surjection f : N
N
→ X. Set H = f
−1
[G], so that H ⊆ N
N
is open and G = f[H]. Being an open
set in a metric space, H is F
σ
(4A2Lc), so, in particular, is SouslinF; but N
N
is analytic, so H is analytic
and its continuous image G is analytic. QQQ
We have already seen that closed subsets of X belong to / (423Bd). Because / is closed under Souslin’s
operation, it contains every Borel set, by 421F. It therefore contains every set obtainable by Souslin’s
operation from Borel sets, and (i)⇒(i).
Remark See also 423Yb below.
423F Proposition Let (X, T) be an analytic Hausdorﬀ space.
(a) A set E ⊆ X is Borel iﬀ both E and X ¸ E are analytic.
(b) If S is a coarser(= smaller) Hausdorﬀ topology on X, then S and T have the same Borel sets.
proof (a) If E is Borel, then E and X ¸ E are analytic, by 423Eb. If E and X ¸ E are analytic, they are
Kanalytic (423Eb) and disjoint, so there is a Borel set F ⊇ E which is disjoint from X¸ E (422J); but now
of course F = E, so E must be Borel.
(b) Because the identity map from (X, T) to (X, S) is continuous, S is an analytic topology (423Bb)
and every SBorel set is TBorel. If E ⊆ X is TBorel, then it and its complement are Tanalytic, therefore
Sanalytic (423Bb), and E is SBorel by (a).
423G Lemma Let X and Y be analytic Hausdorﬀ spaces and f : X → Y a Borel measurable function.
(a) (The graph of) f is an analytic set.
(b) f[A] is an analytic set in Y for any analytic set (in particular, any Borel set) A ⊆ X.
(c) f
−1
[B] is an analytic set in X for any analytic set (in particular, any Borel set) B ⊆ Y .
158 Descriptive set theory 423G
proof (a) Let c be a countable network for the topology of Y . Set
R =
E∈L
(X E) ∪ ((X ¸ f
−1
[E]) Y ).
Then R is a Borel set in X Y . But also R is the graph of f. PPP If f(x) = y, then surely y ∈ E whenever
x ∈ f
−1
[E], so (x, y) ∈ R. On the other hand, if x ∈ X, y ∈ Y and f(x) ,= y, there are disjoint open sets G,
H ⊆ Y such that f(x) ∈ G and y ∈ H; now there is an E ∈ c such that f(x) ∈ E ⊆ G, so that f(x) ∈ E
but y / ∈ E, and (x, y) / ∈ R. QQQ
Because X Y is analytic (423Bc), R is analytic (423Eb).
(b) If A ⊆ X is analytic, then A Y and R ∩ (A Y ) are analytic (423Ea), so f[A] = R[A], which is a
continuous image of R ∩ (AY ), is analytic.
(c) Similarly, if B ⊆ Y is analytic, then f
−1
[B] is a continuous image of R ∩ (X B), so is analytic.
423H Lemma Let (X, T) be an analytic Hausdorﬀ space, and ¸E
n
)
n∈N
any sequence of Borel sets in X.
Then the topology T
t
generated by T ∪ ¦E
n
: n ∈ N¦ is analytic.
proof If X = ∅ this is trivial. Otherwise, there is a continuous surjection f : N
N
→ X. Set F
n
= f
−1
[E
n
]
for each n; then F
n
is a Borel subset of N
N
, so there is a Polish topology S
t
on N
N
, ﬁner than the usual
topology, for which every F
n
is open, by 4A3I. But now f is continuous for S
t
and T
t
, so T
t
is analytic, by
423Ba and 423Bb. (Of course T
t
is Hausdorﬀ, because it is ﬁner than T.)
423I Theorem Let X be a Polish space, E ⊆ X a Borel set, Y a Hausdorﬀ space and f : E → Y an
injective function.
(a) If f is continuous, then f[E] is Borel.
(b) If Y has a countable network (e.g., is an analytic space or a separable metrizable space), and f is
Borel measurable, then f[E] is Borel.
proof (a)(i) Since there is a ﬁner Polish topology on X for which E is closed (4A3I), therefore Polish in the
subspace topology (4A2Qd), and f will still be continuous for this topology, we may suppose that E = X.
(ii) Let ¸U
n
)
n∈N
run over a base for the topology of X (4A2P(ai)). For each pair m, n ∈ N such that
U
m
∩ U
n
is empty, f[U
m
] and f[U
n
] are analytic sets in Y (423Eb, 423Bb) and are disjoint (because f is
injective), so there is a Borel set H
mn
including f[U
m
] and disjoint from f[U
n
] (422J). Set
E
n
= f[U
n
] ∩
¦H
nm
¸ H
mn
: m ∈ N, U
m
∩ U
n
= ∅¦
for each n ∈ N; then E
n
is a Borel set in Y including f[U
n
]. Note that if U
m
∩ U
n
is empty, then
E
m
∩ E
n
⊆ (H
mn
¸ H
nm
) ∩ (H
nm
¸ H
mn
) is also empty.
Fix a metric ρ on X, inducing its topology, for which X is complete, and for k ∈ N set
F
k
=
¦E
n
: n ∈ N, diamU
n
≤ 2
−k
¦,
so that F
k
is Borel. Let F =
k∈N
F
k
; then F is also a Borel subset of Y .
The point is that F = f[X]. PPP (i) If x ∈ X, then for every k ∈ N there is an n ∈ N such that
x ∈ U
n
⊆ ¦y : ρ(y, x) ≤ 2
−k−1
¦;
now diamU
n
≤ 2
−k
, so
f(x) ∈ f[U
n
] ⊆ E
n
⊆ F
k
.
As k is arbitrary, f(x) ∈ F
k
; as x is arbitrary, f[X] ⊆ F. (ii) If y ∈ F, then for each k ∈ N we can ﬁnd an
n(k) such that y ∈ E
n(k)
and diamU
n(k)
≤ 2
−k
. Since f[U
n(k)
] ⊇ E
n(k)
is not empty, nor is U
n(k)
, and we
can choose x
k
∈ U
n(k)
. Indeed, for any j, k ∈ N, E
n(j)
∩E
n(k)
contains y, so is not empty, and U
n(j)
∩U
n(k)
cannot be empty; but this means that there is some x in the intersection, and
ρ(x
j
, x
k
) ≤ ρ(x
j
, x) +ρ(x, x
k
) ≤ diamU
n(j)
+ diamU
n(k)
≤ 2
−j
+ 2
−k
.
This means that ¸x
k
)
k∈N
is a Cauchy sequence. But X is supposed to be complete, so ¸x
k
)
k∈N
has a limit
x say.
423K Analytic spaces 159
??? If f(x) ,= y, then (because Y is Hausdorﬀ) there is an open set H containing f(x) such that y / ∈ H.
Now f is continuous, so there is a δ > 0 such that f(x
t
) ∈ H whenever ρ(x
t
, x) ≤ δ. There is a k ∈ N such
that 2
−k
+ρ(x
k
, x) ≤ δ. If x
t
∈ U
n(k)
, then
ρ(x
t
, x) ≤ ρ(x
t
, x
k
) +ρ(x
k
, x) ≤ δ;
thus f[U
n(k)
] ⊆ H, and
E
n(k)
⊆ f[U
n(k)
] ⊆ H.
But y ∈ E
n(k)
¸ H. XXX
Thus y = f(x) belongs to f[X]; as y is arbitrary, F ⊆ f[X]. QQQ Accordingly f[X] = F is a Borel subset
of Y , as claimed.
(b) By 4A2Nf, there is a countable family 1 of open sets in Y such that whenever y, y
t
are distinct points
of Y there are disjoint V , V
t
∈ 1 such that y ∈ V and y
t
∈ V
t
. Let S
t
be the topology generated by 1;
then S
t
is Hausdorﬀ. For each V ∈ 1, f
−1
[V ] is a Borel set in X, so there is a Polish topology T
t
on X,
ﬁner than the original topology, for which every f
−1
[V ] is open (4A3I). Now f is continuous for T
t
and S
t
(4A2B(aii)), and E is T
t
Borel, so f[E] is a S
t
Borel set in Y , by (a). Since S
t
is coarser than the original
topology S on Y , f[X] is also SBorel.
423J Lemma If X is an uncountable analytic Hausdorﬀ space, it has subsets homeomorphic to ¦0, 1¦
N
and N
N
.
proof (a) Let f : N
N
→ X be a continuous surjection. Write S
∗
=
n∈N
N
n
, I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦ for
σ ∈ S
∗
,
T = ¦σ : σ ∈ S
∗
, f[I
σ
] is uncountable¦.
Then if σ ∈ T there are τ, τ
t
∈ T, both extending σ, such that f[I
τ
] ∩ f[I
τ
] = ∅. PPP Set
A =
¦f[I
τ
] : τ ∈ S
∗
¸ T¦.
Then A is a countable union of countable sets, so is countable. There must therefore be distinct points x,
y of f[I
σ
] ¸ A; express x as f(φ) and y as f(ψ) where φ and ψ belong to I
σ
. Because X is Hausdorﬀ, there
are disjoint open sets G, H such that x ∈ G and y ∈ H. Because f is continuous, there are m, n ∈ N such
that I
φm
⊆ f
−1
[G] and I
ψn
⊆ f
−1
[H]. Of course both τ = φm and τ
t
= ψn must extend σ, and they
belong to T because x ∈ f[I
τ
] ¸ A, y ∈ f[I
τ
] ¸ A. QQQ
(b) We can therefore choose inductively a family ¸τ(υ))
υ∈S
∗
2
in T, where S
∗
2
=
n∈N
¦0, 1¦
n
, such that
τ(∅) = ∅,
τ(υ
i) ⊇ τ(υ) whenever υ ∈ S
∗
2
, i ∈ ¦0, 1¦,
f[I
τ(υ
0
)] ∩ f[I
τ(υ
1
] = ∅ for every υ ∈ S
∗
2
.
Note that #(τ(υ)) ≥ #(υ) for every υ ∈ S
2
. For each z ∈ ¦0, 1¦
N
, ¸τ(zn))
n∈N
is a sequence in S
∗
in which
each term strictly extends its predecessor, so there is a unique g(z) ∈ N
N
such that τ(zn) ⊆ g(z) for every
n. Now g(z
t
)n = g(z)n whenever zn = z
t
n, so g and fg : ¦0, 1¦
N
→ X are continuous. If w, z are
distinct points of ¦0, 1¦
N
, there is a ﬁrst n such that w(n) ,= z(n), in which case fg(w) ∈ f[I
τ(wn)
w(n)
]
and fg(z) ∈ f[I
τ(wn)
z(n)
] are distinct. So fg : ¦0, 1¦
N
→ X is a continuous injection, therefore a
homeomorphism between ¦0, 1¦
N
and its image, because ¦0, 1¦
N
is compact (3A3Dd).
(c) Thus X has a subspace homeomorphic to ¦0, 1¦
N
. Now ¦0, 1¦
N
has a subspace homeomorphic to N.
PPP For instance, setting d
n
(n) = 1, d
n
(i) = 0 for i ,= n, D = ¦d
n
: n ∈ N¦ is homeomorphic to N. QQQ Now
D
N
is homeomorphic to N
N
and is a subspace of (¦0, 1¦
N
)
N ∼
= ¦0, 1¦
N·N ∼
= ¦0, 1¦
N
, so ¦0, 1¦
N
has a subspace
homeomorphic to N
N
. Accordingly X also has a subspace homeomorphic to N
N
.
423K Corollary Any uncountable Borel set in any analytic space has cardinal c.
160 Descriptive set theory 423K
proof If X is an analytic space and E ⊆ X is an uncountable Borel set, then E is analytic (423E), so
includes a copy of ¦0, 1¦
N
and must have cardinal at least #(¦0, 1¦
N
) = c. On the other hand, E is also a
continuous image of N
N
, so has cardinal at most #(N
N
) = c.
423L Proposition Let X be an uncountable analytic Hausdorﬀ space. Then it has a nonBorel analytic
subset.
proof (a) I show ﬁrst that there is an analytic set A ⊆ N
N
N
N
such that every analytic subset of N
N
is a
vertical section of A. PPP Let  be a countable base for the topology of N
N
N
N
, containing ∅, and ¸U
n
)
n∈N
an enumeration of . Write
M = (N
N
N
N
N
N
) ¸
m,n∈N
(¦x : x(m) = n¦ U
n
).
Then M is a closed subset of (N
N
N
N
) N
N
, therefore analytic (423Ba, 423Bd), so its continuous image
A = ¦(x, z) : there is some y such that (x, y, z) ∈ M¦
is analytic (423Bb).
Now let E be any analytic subset of N
N
. By 423E, E is SouslinF; by 421J, there is a closed set
F ⊆ N
N
N
N
such that E = ¦z : ∃ y, (y, z) ∈ F¦. Let ¸x(m))
m∈N
be a sequence running over ¦n : n ∈
N, U
n
∩ F = ∅¦, so that
F = (N
N
N
N
) ¸
m∈N
U
x(m)
= ¦(y, z) : (x, y, z) ∈ M¦.
Now
¦z : (x, z) ∈ A¦ = ¦z : there is some y such that (x, y, z) ∈ M¦
= ¦z : there is some y such that (y, z) ∈ F¦ = E,
and E is a vertical section of A, as required. QQQ
(b) It follows that there is a nonBorel analytic set B ⊆ N
N
. PPP Take A from (a) above, and try
B = ¦x : (x, x) ∈ A¦.
Because B is the inverse image of A under the continuous map x → (x, x), it is analytic (423Gc). ??? If B
were a Borel set, then B
t
= N
N
¸ B would also be Borel, therefore analytic (423E), and there would be an
x ∈ N
N
such that B
t
= ¦y : (x, y) ∈ A¦. But in this case
x ∈ B ⇐⇒ (x, x) ∈ A ⇐⇒ x ∈ B
t
,
which is a diﬃculty you may have met before. XXXQQQ
(c) Now return to our arbitrary uncountable analytic Hausdorﬀ space X. By 423J, X has a subset Z
homeomorphic to N
N
. By (b), Z has an analytic subset A which is not Borel in Z, therefore cannot be a
Borel subset of X.
423M I devote a few paragraphs to an important method of constructing selectors.
Theorem Let X be an analytic Hausdorﬀ space, Y a set, and ( ⊆ TY . Write T for the σalgebra of subsets
of Y generated by o((), where o is Souslin’s operation, and 1 for o(¦F C : F ⊆ X is closed, C ∈ (¦). If
W ∈ 1, then W[X] ∈ T and there is a Tmeasurable function f : W[X] → X such that (f(y), y) ∈ W for
every y ∈ W[X].
proof Write T for ¦F C : F ⊆ X is closed, C ∈ (¦.
(a) Consider ﬁrst the case in which X = N
N
and all the horizontal sections W
−1
[¦y¦] of W are closed.
Let c be the family of closed subsets of Y . For σ ∈ S
∗
=
n∈N
N
n
set I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦.
Then W ∩ (I
σ
Y ) ∈ 1. PPP Because Souslin’s operation is idempotent (421D), o(1) = 1. The set
¦V : V ∩(I
σ
Y ) ∈ 1¦ is therefore closed under Souslin’s operation (apply 421Cc to the identity map from
I
σ
Y to X Y , or otherwise); since it includes T, it is the whole of 1, and contains W. QQQ
By 421G, W[I
σ
] = (W∩(I
σ
Y ))[N
N
] belongs to o(() ⊆ T for every σ. In particular, W[N
N
] = W[I
∅
] ∈ T.
Choose ¸Y
σ
)
σ∈S
∗ in T inductively, as follows. Y
∅
= W[N
N
]. Given that Y
σ
∈ T and that Y
σ
⊆ W[I
σ
], set
423N Analytic spaces 161
Y
σ
j
= Y
σ
∩ W[I
σ
j
] ¸
i<j
W[I
σ
i
]
for every j ∈ N. Continue.
At the end of the induction, we have
j∈N
Y
σ
j
= Y
σ
∩
j∈N
W[I
σ
j
] = Y
σ
∩ W[I
σ
] = Y
σ
for every σ ∈ S
∗
, while ¸Y
σ
j
)
j∈N
is always disjoint. So for each y ∈ Y
∅
= W[N
N
] we have a unique
f(y) ∈ N
N
such that y ∈ Y
f(y)n
for every n. Since f
−1
[I
σ
] = Y
σ
∈ T for every σ ∈ S
∗
, f is Tmeasurable
(4A2H(cii)). Also (f(y), y) ∈ W for every y ∈ W[N
N
]. PPP For each n ∈ N, y ∈ Y
f(y)n
= W[I
f(y)n
], so
there is an x
n
∈ N
N
such that x
n
n = f(y)n and (x
n
, y) ∈ W. But this means that f(y) = lim
n→∞
x
n
;
since we are supposing that the horizontal sections of W are closed, (f(y), y) ∈ W. QQQ
Thus the theorem is true if X = N
N
and W has closed horizontal sections.
(b) Now suppose that X = N
N
and that W ⊆ N
N
Y is any set in 1. Then there is a Souslin scheme
¸F
σ
C
σ
)
σ∈S
in T with kernel W; of course I mean you to suppose that F
σ
⊆ N
N
is closed and C
σ
∈ ( for
every σ. Set
˜
W =
k≥1
σ∈N
k
I
σ
F
σ
C
σ
⊆ N
N
N
N
Y .
Then W is the projection of
˜
W onto the last two coordinates, by 421Ce. If y ∈ Y , then
¦(φ, ψ) : (φ, ψ, y) ∈
˜
W¦ =
k≥1
¦I
σ
F
σ
: σ ∈ N
k
, y ∈ C
σ
¦
is closed in N
N
N
N
. (If J is any subset of N
k
, then
(N
N
N
N
) ¸
σ∈J
I
σ
F
σ
=
σ∈J
I
σ
(N
N
¸ F
σ
) ∪
σ∈N
k
\J
I
σ
N
N
is open.) Also I
σ
F
σ
is a closed subset of N
N
N
N
for every σ, and N
N
N
N
is homeomorphic to N
N
. We
can therefore apply (a) to
˜
W, regarded as a subset of (N
N
N
N
) Y , to see that W[N
N
] =
˜
W[N
N
N
N
] ∈ T
and that there is a Tmeasurable function h = (g, f) : W[N
N
] →N
N
N
N
such that (g(y), f(y), y) ∈
˜
W for
every y ∈ W[N
N
]. Now, of course, f : W[N
N
] →N
N
is Tmeasurable and (f(y), y) ∈ W for every y ∈ W[N
N
].
(c) Finally, suppose only that X is an analytic Hausdorﬀ space and that W ∈ 1. If X is empty, so is Y ,
and the result is trivial. Otherwise, there is a continuous surjection h : N
N
→ X. Set
˜
h(φ, y) = (h(φ), y) for
φ ∈ N
N
and y ∈ Y ; then
˜
h : N
N
Y → X Y is a continuous surjection, and
˜
W =
˜
h
−1
[W] is the kernel of
a Souslin scheme in
¦
˜
h
−1
[F C] : F ⊆ N
N
is closed, C ∈ (¦ = ¦h
−1
[F] C : F ⊆ N
N
is closed, C ∈ (¦
by 421Cb. So we can apply (b) to see that W[X] =
˜
W[N
N
] ∈ T and there is a Tmeasurable g : W[X] →N
N
such that (g(y), y) ∈
˜
W for every y ∈ Y . Finally f = hg : W[X] → X is Tmeasurable and (f(y), y) ∈ W
for every y ∈ Y . This completes the proof.
423N The expression
1 = o(¦F C : F ⊆ X is closed, C ∈ (¦)
in 423M is a new formulation, and I had better describe one of the basic cases in which we can use the
result.
Corollary Let X be an analytic Hausdorﬀ space and Y any topological space. Let T be the σalgebra of
subsets of Y generated by o(B(Y )), where B(Y ) is the Borel σalgebra of Y . If W ∈ o(B(X Y )), then
W[X] ∈ T and there is a Tmeasurable function f : W[X] → X such that (f(y), y) ∈ W for every y ∈ W[X].
proof (a) Suppose to begin with that X = N
N
. In 423M, set ( = B(Y ). Then every open subset and every
closed subset of XY belongs to 1 as deﬁned in 423M. PPP For σ ∈ S
∗
=
k∈N
N
k
, set I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦.
If V ⊆ X Y is open, set
H
σ
=
¦H : H ⊆ Y is open, I
σ
H
σ
⊆ V ¦
for each σ ∈ S
∗
. Because ¦I
σ
: σ ∈ S
∗
¦ is a base for the topology of N
N
, V =
σ∈S
∗
I
σ
H
σ
∈ 1.
As for the complement of V , we have
162 Descriptive set theory 423N
(N
N
Y ) ¸ V =
k∈N
σ∈N
k
(N
N
Y ) ¸ (I
σ
H
σ
)
=
k∈N
σ∈N
k
((N
N
¸ I
σ
) Y ) ∪ (N
N
(Y ¸ H
σ
))
=
k∈N
σ∈N
k
_
τ∈N
k
τ=σ
(I
τ
Y ) ∪ (N
N
(Y ¸ H
σ
))
again belongs to 1, because 1 is closed under countable unions and intersections and contains I
τ
Y and
N
N
(Y ¸ H
σ
) for all σ, τ ∈ S
∗
. QQQ
By 421F, 1 contains every Borel subset of N
N
Y , so includes o(B(N
N
Y )). So in this case we can
apply 423M directly to get the result.
(b) Now suppose that X is any analytic space. If X is empty, the result is trivial. Otherwise, let
h : N
N
→ X be a continuous surjection. Set
˜
h(φ, y) = (h(φ), y) for φ ∈ N
N
, y ∈ Y , so that
˜
h : N
N
Y → XY
is continuous. Set
˜
W =
˜
h
−1
[W]. If V ∈ B(XY ) then
˜
h
−1
[V ] ∈ B(N
N
Y ) (4A3Cd), so
˜
W ∈ o(B(N
N
Y ))
(421Cc). By (a),
˜
W[N
N
] ∈ T and there is a Tmeasurable function g :
˜
W[N
N
] →N
N
such that (g(y), y) ∈
˜
W
for every y ∈
˜
W[N
N
]. It is now easy to check that W[X] =
˜
W[N
N
] ∈ T (this is where we need to know that
h is surjective), that f = hg : W[X] → X is Tmeasurable, and that (f(y), y) ∈ W for every y ∈ W[X], as
required.
Remark This is a version of the von NeumannJankow selection theorem.
423O Corollary Let X and Y be analytic Hausdorﬀ spaces, and f : X → Y a Borel measurable
surjection. Let T be the σalgebra of subsets of Y generated by the SouslinF sets in Y . Then there is a
Tmeasurable function g : Y → X such that fg is the identity on Y .
proof The point is that the graph Γ = ¦(x, f(x)) : x ∈ X¦ of f is analytic (423Ga), therefore a SouslinF
set in X Y (423Eb). Also Γ[X] = f[X] = Y . By 423N, there is a Tmeasurable function g : Y → X such
that (g(y), y) ∈ Γ, that is, f(g(y)) = y, for every y ∈ Y .
*423P Constituents of coanalytic sets: Theorem Let X be a Hausdorﬀ space, and A ⊆ X an
analytic subset of X. Then there is a nondecreasing family ¸E
ξ
)
ξ<ω
1
of Borel subsets of X, with union
X ¸ A, such that every analytic subset of X ¸ A is included in some E
ξ
.
proof Put 422K(iii) and 423C together.
*423Q Remarks (a) Let A be an analytic set in an analytic space X and ¸E
ξ
)
ξ<ω
1
a family of Borel
sets as in 423P. There is nothing unique about the E
ξ
. But if ¸E
t
ξ
)
ξ<ω
1
is another such family, then every
E
t
ξ
is an analytic subset of X¸ A, by 423E, so is included in some E
η
; and, similarly, every E
ξ
is included in
some E
t
η
. We therefore have a function f : ω
1
→ ω
1
such that E
t
ξ
⊆ E
f(ξ)
and E
ξ
⊆ E
t
f(ξ)
for every ξ < ω.
If we set C = ¦ξ : ξ < ω
1
, f(η) < ξ for every η < ξ¦, then C is a closed coﬁnal set in ω
1
(4A1Bd), and
η<ξ
E
η
=
η<ξ
E
t
η
for every ξ ∈ C. If X ¸ A is itself analytic, that is, if A is a Borel set, then we shall
have to have X ¸ A = E
ξ
= E
t
ξ
for some ξ < ω
1
.
Another way of expressing the result in 423P is to say that if we write 1 = ¦B : B ⊆ X ¸ A is analytic¦,
then ¦E : E ∈ 1, E is Borel¦ is coﬁnal with 1 (this is the First Separation Theorem) and cf 1 ≤ ω
1
.
(b) It is a remarkable fact that, in some models of set theory, we can have nonBorel coanalytic sets
in Polish spaces such that all their constituents are countable (Jech 78, p. 529, Cor. 2). (Note that, by
(a), this is the same thing as saying that X ¸ A is uncountable but all its Borel subsets are countable.)
But in ‘ordinary’ cases we shall have, for every Borel subset E of X ¸ A, an uncountable Borel subset of
(X¸A) ¸E; so that for any family ¸G
ξ
)
ξ<ω
1
of Borel constituents of X¸A, there must be uncountably many
uncountable G
ξ
. To see that this happens at least sometimes, take any nonBorel analytic subset A
0
of N
N
(423L), and consider A = A
0
N
N
⊆ (N
N
)
2
. Then A is analytic (423B). If E ⊆ (N
N
)
2
¸ A is Borel, then
423Yc Analytic spaces 163
π
1
[E] = ¦x : (x, y) ∈ E¦ is an analytic subset of N
N
¸ A, so is not the whole of N
N
¸ A
0
(by 423Fa). Taking
any x ∈ (N
N
¸ A
0
) ¸ π
1
[E], ¦x¦ N
N
is an uncountable Borel subset of ((N
N
)
2
¸ A) ¸ E. For an alternative
construction, see 423Ye.
423X Basic exercises >>>(a) For a Hausdorﬀ space X, show that the following are equiveridical: (i) X
is analytic; (ii) X is a continuous image of a Polish space; (iii) X is a continuous image of a closed subset
of N
N
.
(b) Write out a direct proof of 423Ea, not quoting 423C or 421D.
>>>(c) Let X be a set and S a Hausdorﬀ topology on X, T an analytic topology on X such that S ⊆ T.
Show that S and T have the same analytic sets. (Hint: 423F.)
>>>(d) Let X be an analytic Hausdorﬀ space. (i) Show that its Borel σalgebra B(X) is countably generated
as σalgebra. (Hint: use 4A2Nf and 423Fb.) (ii) Show that there is an analytic subset Y of R such that
(X, /
X
) is isomorphic to (Y, /
Y
), where /
X
, /
Y
are the families of SouslinF subsets of X, Y respectively.
(Hint: show that there is an injective Borel measurable function from X to R (cf. 343E), and use 423G.) (iii)
Show that (X, B(X)) is isomorphic to (Y, B(Y )), where B(Y ) is the Borel σalgebra of Y . (Hint: 423Fa.)
(iv) Let T
X
, T
Y
be the σalgebras generated by /
X
, /
Y
respectively. Show that (X, T
X
) and (Y, T
Y
) are
isomorphic.
(e) Let S be the rightfacing Sorgenfrey topology on R (415Xc). Show that S has the same Borel sets as
the usual topology T on R. (Hint: S is hereditarily Lindel¨of (419Xf) and has a base consisting of TBorel
sets.) Show that S is not analytic.
(f ) Let X be an analytic Hausdorﬀ space and Y any topological space. Let T be the σalgebra of subsets
of Y generated by the SouslinF sets. Show that if W ⊆ X Y is SouslinF, then W[X] ∈ T and there is
a Tmeasurable function f : W[X] → Y such that (f(y), y) ∈ W for every y ∈ W[X]. (Hint: start with
X = N
N
, as in 423N.)
(g) Let X and Y be Hausdorﬀ spaces, and R ⊆ XY an analytic set such that R
−1
[Y ] = X. Show that
there is a function g : X → Y , measurable with respect to the σalgebra generated by the SouslinF subsets
of X, such that (x, g(x)) ∈ R for every x ∈ X.
>>>(h) (i) Show that the family of analytic subsets of [0, 1] has cardinal c. (Hint: 421Xd.) (ii) Show that
the σalgebra T of subsets of [0, 1] generated by the analytic sets has cardinal c. (Hint: 421Xi.) (iii) Show
that there is a set A ⊆ [0, 1] which does not belong to T.
(i) Let X = Y = [0, 1]. Give Y the usual topology, and give X the topology corresponding to the
onepoint compactiﬁcation of the discrete topology on [0, 1[, that is, a set G ⊆ X is open if either 1 / ∈ G or
G is coﬁnite. Show that the identity map f : X → Y is a Borel measurable bijection, but that f
−1
is not
measurable for the σalgebra of subsets of Y generated by the SouslinF sets.
423Y Further exercises (a) Show that a space with a countable network is hereditarily separable
(that is, every subset is separable), therefore countably tight.
(b) Show that if X is a Hausdorﬀ space with a countable network, then every analytic subset of X is
obtainable by Souslin’s operation from the open subsets of X.
(c) Let (X, T) and (Y, S) be analytic Hausdorﬀ spaces and f : X → Y a Borel measurable function. (i)
Show that there is a zerodimensional separable metrizable topology S
t
on Y with the same Borel sets and
the same analytic sets as S. (Hint: 423Xd.) (ii) Show that there is a zerodimensional separable metrizable
topology T
t
on X, with the same Borel sets and the same analytic sets as T, such that f is continuous for the
topologies T
t
, S
t
. (iii) Explain how to elaborate these ideas to deal with any countable family of analytic
spaces and Borel measurable functions between them.
164 Descriptive set theory 423Yd
(d) Let X be an analytic Hausdorﬀ space, Y a Hausdorﬀ space with a countable network, and f : X → Y
a Borel measurable surjection. Let T be the σalgebra of subsets of Y generated by the SouslinF sets in Y .
Show that there is a Tmeasurable function g : Y → X such that fg is the identity on Y .
(e) Set S =
n≥1
N
n
and consider TS with its usual topology. Let T ⊆ TS be the set of trees (412O);
show that T is closed, therefore a compact metrizable space. Set F
σ
= ¦T : σ ∈ T ∈ T ¦ for σ ∈ S, and let
A be the kernel of the Souslin scheme ¸F
σ
)
σ∈s
. Show that the constituents of T ¸ A for this scheme are just
the sets G
ξ
= ¦T : r(T) = ξ¦, where r is the rank function of 421Ne. Show by induction on ξ that all the
G
ξ
is nonempty, so that A is not a Borel set. Show that #(G
ξ
) = c for 1 ≤ ξ < ω
1
. Show that if X is any
topological space and B ⊆ X is a SouslinF set, there is a Borel measurable function f : X → T such that
B = f
−1
[A].
423 Notes and comments We have been dealing, in this section and the last, with three classes of
topological space: the class of analytic spaces, the class of Kanalytic spaces and the class of spaces with
countable networks. The ﬁrst is more important than the other two put together, and I am sure many people
would ﬁnd it more comfortable, if more timeconsuming, to learn the theory of analytic spaces thoroughly
ﬁrst, before proceeding to the others. This was indeed my own route into the subject. But I think that the
theory of Kanalytic spaces has now matured to the point that it can stand on its own, without constant
reference to its origin as an extension of descriptive set theory on the real line; and that our understanding
of analytic spaces is usefully advanced by seeing how easily their properties can be deduced from the fact
that they are Kanalytic spaces with countable networks.
‘Selection theorems’ appear everywhere in mathematics. The axiom of choice is a selection theorem: if
¸A
i
)
i∈I
is a family of nonempty sets, there is a selector for the relation ¦(i, x) : i ∈ I, x ∈ A
i
¦. The
Lifting Theorem (¸341) is a special kind of selection theorem, where we seek a selector which is a Boolean
homomorphism. In general topology we look for continuous selectors; in measure theory, for measurable
selectors. Any selection theorem will have expressions either as a theorem on right inverses of functions, as
in 423O, or as a theorem on selectors for relations, as in 423M423N. In the language of this section, however,
the strongest results are most easily set out in terms of relations, because the essence of the method is that
we can ﬁnd selectors taking values in analytic spaces, and the relations of 423M423N can be very far from
being analytic spaces in themselves, even when we have a natural topology on the space Y . The value of
these results in measure theory will become clearer in ¸431, where we shall see that there are important
σalgebras which are closed under Souslin’s operation, so that they can include the algebras T of 423M423O.
Typical applications are in
As in ¸422, I have made no attempt to cover the general theory of analytic spaces, nor even to give a
balanced introduction. I have tried instead to give a condensed account of the principal methods for showing
that spaces are analytic, with some of the ideas which can be applied to make them more accessible to the
imagination (423J, 423Xc423Xd, 423Yb423Yc). Lusin’s theorem 423I does not mention ‘analytic’ sets
in its statement, but it depends essentially on the separation theorem 422J, so cannot really be put with
the other results on Polish spaces in 4A2Q. You must of course know that not all analytic sets are Borel
(423L) and that not all sets are analytic (423Xh). For further information about this fascinating subject,
see Rogers 80, Kechris 95 and Moschovakis 80.
‘Selection theorems’ appear everywhere in mathematics. The axiom of choice is a selection theorem; it says
that whenever R ⊆ XY is a relation and R[X] = Y , there is a function f : Y → X such that (f(y), y) ∈ R
for every y ∈ Y . The Lifting Theorem (¸341) asks for a selector which is a Borel homomorphism. In general
topology we look for continuous selectors. In measure theory, naturally, we are interested in measurable
selectors, as in 423M423O. Any selection theorem will have expressions either as a theorem on right inverses
of functions, as in 423O, or as a theorem on selectors for relations, as in 423M423N. In the language here,
however, we get better theorems by examining relations, because the essence of the method is that we can
ﬁnd measurable functions into analytic spaces, and the relations of 423M can be very far from being analytic,
even when there is a natural topology on the space Y . The value of these results will become clearer in ¸431,
when we shall see that the σalgebras T of 423M423O are often included in familiar σalgebras. Typical
applications are in 433F433G below.
424C Standard Borel spaces 165
424 Standard Borel spaces
This volume is concerned with topological measure spaces, and it will come as no surprise that the
topological properties of Polish spaces are central to the theory. But even from the point of view of unadorned
measure theory, not looking for topological structures on the underlying spaces, it turns out that the Borel
algebras of Polish spaces have a very special position. It will be useful later on to be able to refer to some
fundamental facts concerning them.
424A Deﬁnition Let X be a set and Σ a σalgebra of subsets of X. We say that (X, Σ) is a standard
Borel space if there is a Polish topology on X for which Σ is the algebra of Borel sets.
Warning! Many authors reserve the phrase ‘standard Borel space’ for the case in which X is uncountable.
I have seen the phrase ‘Borel space’ used for what I call a ‘standard Borel space’.
424B Proposition (a) If (X, Σ) is a standard Borel space, then Σ is countably generated as σalgebra
of sets.
(b) If ¸(X
i
, Σ
i
))
i∈I
is a countable family of standard Borel spaces, then (
i∈I
X
i
,
´
i∈I
Σ
i
) (deﬁnition:
254E) is a standard Borel space.
(c) Let (X, Σ) and (Y, T) be standard Borel spaces and f : X → Y a (Σ, T)measurable surjection. Then
(i) if E ∈ Σ is such that f[E] ∩ f[X ¸ E] = ∅, then f[E] ∈ T;
(ii) T = ¦F : F ⊆ Y, f
−1
[F] ∈ Σ¦;
(iii) if f is a bijection it is an isomorphism.
(d) Let (X, Σ) and (Y, T) be standard Borel spaces and f : X → Y a (Σ, T)measurable injection. Then
Z = f[X] ∈ T and f is an isomorphism between (X, Σ) and (Z, T
Z
), where T
Z
is the subspace σalgebra.
proof (a) Let T be a Polish topology on X such that Σ is the algebra of Borel sets. Then T has a countable
base , which generates Σ (4A3Da/4A3E).
(b) For each i ∈ I let T
i
be a Polish topology on X
i
such that Σ
i
is the algebra of T
i
Borel sets. Then
X =
i∈I
T
i
, with the product topology T, is Polish (4A2Qc). By 4A3Dc/4A3E, Σ =
´
i∈I
Σ
i
is just the
Borel σalgebra of X, so (X, Σ) is a standard Borel space.
(c) Let T, S be Polish topologies on X, Y respectively for which Σ and T are the Borel σalgebras. Then
f is Borel measurable.
(i) By 423Gb and 423Eb, f[E] and f[X ¸ E] are analytic subsets of Y . But they are complementary,
so they are Borel sets, by 423Fa.
(ii) f
−1
[F] ∈ Σ for every F ∈ T, just because f is measurable. On the other hand, if F ⊆ Y and
E = f
−1
[F] ∈ Σ, then F = f[E] ∈ T by (i).
(iii) follows at once.
(d) Give X and Y Polish topologies for which Σ, T are the Borel σalgebras. By 423Ib, f[E] ∈ T for every
E ∈ Σ; in particular, Z = f[X] belongs to T. Also f
−1
[F] ∈ Σ for every F ∈ T
Z
, so f is an isomorphism
between (X, Σ) and (Z, T
Z
).
424C Theorem Let (X, Σ) be a standard Borel space.
(a) If X is countable then Σ = TX.
(b) If X is uncountable then (X, Σ) is isomorphic to (N
N
, B(N
N
)), where B(N
N
) is the algebra of Borel
subsets of N
N
.
proof Let T be a Polish topology on X such that Σ is its Borel σalgebra.
(a) Every singleton subset of X is closed, so must belong to Σ. If X is countable, every subset of X is a
countable union of singletons, so belongs to Σ.
(b) (Rao & Srivastava 94) The strategy of the proof is to ﬁnd Borel sets Z ⊆ X, W ⊆ N
N
such that
(Z, Σ
Z
)
∼
= (N
N
, B(N
N
)) and (W, B(W))
∼
= (X, Σ) (writing Σ
Z
, B(W) for the subspace σalgebras), and use
a form of the Schr¨oderBernstein theorem.
166 Descriptive set theory 424C
(i) By 423J, X has a subset Z homeomorphic to N
N
; let h : N
N
→ Z be a homeomorphism. By 424Bd,
h is an isomorphism between (N
N
, B(N
N
)) and (Z, Σ
Z
).
(ii) Let ¸U
n
)
n∈N
run over a base for the topology of X. Deﬁne g : X → ¦0, 1¦
N
⊆ N
N
by setting
g(x) = ¸χU
n
(x))
n∈N
for every x ∈ N. Then g is injective, because X is Hausdorﬀ. Also g is Borel measurable,
by 4A3D(cii). By 424Bd, g is an isomorphism between (X, Σ) and (W, B(W)), where W = g[X] belongs to
B(N
N
).
(iii) We have Z ∈ Σ, W ∈ B(N
N
) such that (Z, Σ
Z
)
∼
= (N
N
, B(N
N
)) and (W, B(W))
∼
= (X, Σ). By
344D, (X, Σ)
∼
= (N
N
, B(N
N
)), as claimed.
424D Corollary (a) If (X, Σ) and (Y, T) are standard Borel spaces and #(X) = #(Y ), then (X, Σ) and
(Y, T) are isomorphic.
(b) If (X, Σ) is an uncountable standard Borel space then #(X) = #(Σ) = c.
proof These follow immediately from 424C, if we recall that if B(N
N
) is the Borel σalgebra of N
N
then
#(B(N
N
)) = c (4A3Fb).
424E Proposition Let X be a set and Σ a σalgebra of subsets of X; suppose that (X, Σ) is countably
separated in the sense that there is a countable set c ⊆ Σ separating the points of X. If A ⊆ X is such that
(A, Σ
A
) is a standard Borel space, where Σ
A
is the subspace σalgebra, then A ∈ Σ.
proof Give A a Polish topology T such that Σ
A
is the Borel σalgebra of A, and let S be the topology on
X generated by c ∪ ¦X ¸ E : E ∈ c¦. Then S is secondcountable (4A2Oa), so has a countable network
(4A2Oc), and is Hausdorﬀ because c separates the points of X. The identity map from A to X is Borel
measurable for T and S, so 423Ib tells us that A is SBorel; but of course the SBorel σalgebra is just the
σalgebra generated by c (4A3Da), so is included in Σ.
424F Corollary Let X be a Polish space and A ⊆ X any set which is not Borel. Let B(A) be the Borel
σalgebra of A. Then (A, B(A)) is not a standard Borel space.
424G Proposition Let (X, Σ) be a standard Borel space. Then (E, Σ
E
) is a standard Borel space for
every E ∈ Σ, writing Σ
E
for the subspace σalgebra.
proof Let T be a Polish topology on X for which Σ is the Borel σalgebra. Then there is a Polish topology
T
t
⊇ T for which E is closed (4A3I), therefore itself Polish in the subspace topology T
t
E
(4A2Qd). But T
t
and T have the same Borel sets (423Fb), so Σ
E
is just the Borel σalgebra of E for T
t
E
, and (E, Σ
E
) is a
standard Borel space.
*424H For the full strength of a theorem in ¸448 we need a remarkable result concerning group actions
on Polish spaces.
Theorem (Becker & Kechris 96) Let G be a Polish group, (X, T) a Polish space and
•
a Borel measurable
action of G on X. Then there is a Polish topology T
t
on X, yielding the same Borel sets as T, such that the
action is continuous for T
t
and the given topology of G.
proof (a) Fix on a righttranslationinvariant metric ρ on G deﬁning the topology of G (4A5Q), and let D
be a countable dense subset of G; write e for the identity of G. Let Z be the set of 1Lipschitz functions
from G to [0, 1], that is, functions f : G → [0, 1] such that [f(g) − f(h)[ ≤ ρ(g, h) for all g, h ∈ G. Then
Z, with the topology of pointwise convergence inherited from the product topology of [0, 1]
G
, is a compact
metrizable space. PPP It is a closed subset of [0, 1]
G
, so is a compact Hausdorﬀ space. Writing q(f) = fD
for f ∈ Z, q : Z → [0, 1]
D
is injective, because D is dense and every member of Z is continuous; but this
means that Z is homeomorphic to q[Z], which is metrizable, by 4A2Pc. QQQ
We see also that the Borel σalgebra of Z is the σalgebra generated by sets of the form W
gα
= ¦f : f(g) <
α¦ where g ∈ G and α ∈ [0, 1]. PPP This σalgebra contains every set of the form ¦f : f ∈ Z, α < f(g) < β¦,
where g ∈ G and α, β ∈ R; since these sets generate the topology of Z, the σalgebra they generate is the
Borel σalgebra of Z, by 4A3Da. QQQ
*424H Standard Borel spaces 167
(b) There is a continuous action of G on Z deﬁned by setting
(g
•
f)(h) = f(hg)
for f ∈ Z and g, h ∈ G. PPP (i) If f ∈ Z and g, h
1
, h
2
∈ G, then
[(g
•
f)(h
1
) −(g
•
f)(h
2
)[ = [f(h
1
g) −f(h
2
g)[ ≤ ρ(h
1
g, h
2
g) = ρ(h
1
, h
2
)
because ρ is righttranslationinvariant. So g
•
f ∈ Z for every f ∈ Z, g ∈ G. (ii)
(e
•
f)(h) = f(he) = f(h)
for every h ∈ G, so e
•
f = f for every f ∈ Z. (iii) If f ∈ Z and g
1
, g
2
, h ∈ G, then
(g
1
•
(g
2
•
f))(h) = (g
2
•
f)(hg
1
) = f(hg
1
g
2
) = ((g
1
g
2
)
•
f)(h);
as h is arbitrary, (g
1
g
2
)
•
f = g
1
•
(g
2
•
f). Thus
•
is an action of G on Z. (iv) Suppose that g
0
∈ G, f
0
∈ Z,
h ∈ G and > 0. Set
V = ¦g : g ∈ G, ρ(hg, hg
0
) <
1
2
¦,
W = ¦f : f ∈ Z, [f(hg
0
) −f
0
(hg
0
)[ <
1
2
¦.
Then V is an open set in G containing g
0
(because g → ρ(hg, hg
0
) is continuous) and W is an open set in
Z containing f
0
. If g ∈ V and f ∈ W,
[(g
•
f)(h) −(g
0
•
f
0
)(h)[ = [f(hg) −f
0
(hg
0
)[
≤ [f(hg) −f(hg
0
)[ +[f(hg
0
) −f
0
(hg
0
)[
≤ ρ(hg, hg
0
) +
1
2
≤ .
As f
0
, g
0
, are arbitrary, the map (g, f) → (g
•
f)(h) is continuous; as h is arbitrary, the map (g, f) → g
•
f
is continuous. QQQ
(c) Let B(X) be the Borel σalgebra of X. For x ∈ X and B ∈ B(X), set
P
B
(x) = ¦g : g ∈ G, g
•
x ∈ B¦,
Q
B
(x) =
¦V : V ⊆ G is open, V ¸ P
B
(x) is meager¦,
f
B
(x)(g) = inf(¦1¦ ∪ ¦ρ(g, h) : h ∈ G¸ Q
B
(x)¦)
for g ∈ G. It is easy to check that
f
B
(x)(g
t
) ≤ ρ(g, g
t
) +f
B
(x)(g
t
)
for all g, g
t
∈ G, so that every f
B
(x) belongs to Z.
Every P
B
(x) is a Borel set, because
•
is Borel measurable. It follows that Q
B
(x)´P
B
(x) is meager. PPP
By 4A3Ra, there is an open set V
0
⊆ G such that V
0
´(X ¸ P
B
(x)) is meager and V
0
∩V is empty whenever
V ⊆ G is open and V ¸ P
B
(x) is meager. Now Q
B
(x) ∩V
0
is empty, so Q
B
(x) ¸ P
B
(x) ⊆ (X ¸ P
B
(x)) ¸ V
0
is
meager. Also
(X ¸ V
0
) ¸ P
B
(x) ⊆ (X ¸ P
B
(x)) ¸ V
0
is meager, so X ¸ V
0
⊆ Q
B
(x) and
P
B
(x) ¸ Q
B
(x) ⊆ (V
0
¸ V
0
) ∪ (V
0
¸ (X ¸ P
B
(x))
is meager. QQQ Consequently Q
B
(x) ¸ P
B
(x) is empty, because the only meager open set in G is empty, by
Baire’s theorem for complete metric spaces (4A2Ma).
Let 1 be a countable base for the topology of G containing G.
(d) For each B ∈ B(X), the map f
B
: X → Z is Borel measurable. PPP Because the Borel σalgebra of Z
is generated by the sets W
gα
of (a) above, it is enough to show that
¦x : f
B
(x) ∈ W
gα
¦ = ¦x : f
B
(x)(g) < α¦
168 Descriptive set theory *424H
always belongs to B(X), because ¦W : f
−1
B
[W] ∈ B¦ is surely a σalgebra of subsets of Z. But if α > 1 this
set is X, while if α ≤ 1 it is
¦x : there is some h ∈ X ¸ Q
B
(x) such that ρ(g, h) < α¦
= ¦x : there is some V ∈ 1 such that V ¸ Q
B
(x) ,= ∅
and ρ(g, h) < α for every h ∈ V ¦
=
_
V ∈\
¦x : V ,⊆ Q
B
(x)¦
(where 1
t
= ¦V : V ∈ 1, ρ(g, h) < α for every h ∈ V ¦)
=
_
V ∈\
¦x : V ¸ P
B
(x) is not meager¦.
But for any ﬁxed V ∈ 1,
¦x : V ¸ P
B
(x) is not meager¦ = X ¸ ¦x : W[¦x¦] is meager¦
where
W = ¦(y, g) : y ∈ X, g ∈ V, g
•
y ∈ X ¸ B¦
is a Borel subset of X G, because
•
is supposed to be Borel measurable; and therefore W ∈ B(X)
´
⊗B(G),
writing B(G) for the Borel σalgebra of G (4A3G). Now B(G) ⊆
´
B(G), the Baire property algebra of G
(4A3Rb), so W ∈ B(X)
´
⊗
´
B(G). By 4A3Rc, the quotient algebra
´
B// has a countable orderdense set,
where / is the σideal of meager sets, so 4A3Sa tells us that ¦x : W[¦x¦] is meager¦ ∈ B(X). Accordingly
¦x : V ¸ P
B
(x) is not meager¦ is Borel for every V , and ¦x : f
B
(x)(g) < α¦ is Borel. QQQ
(e) If g ∈ G, B ∈ B(X) and x ∈ X, then g
•
f
B
(x) = f
B
(g
•
x). PPP
P
B
(g
•
x) = ¦h : h
•
(g
•
x) ∈ B¦ = ¦h : hg ∈ P
B
(x)¦ = P
B
(x)g
−1
.
Because the map h → hg
−1
: G → G is a homeomorphism, Q
B
(g
•
x) = Q
B
(x)g
−1
; because it is an isometry,
f
B
(g
•
x)(h) = min(1, ρ(h, X ¸ Q
B
(g
•
x))) = min(1, ρ(h, X ¸ Q
B
(x)g
−1
))
= min(1, ρ(hg, X ¸ Q
B
(x))) = f
B
(x)(hg) = (g
•
f
B
(x))(h)
for every h ∈ G. QQQ
(f ) Let ¸B
0m
)
m∈N
run over a base for T containing X. We can now ﬁnd a countable set c ⊆ B(X)
such that (i) the topology T
∗
generated by c is a Polish topology ﬁner than T (ii) f
B
is T
∗
continuous for
every B ∈ c (iii) X ∈ c. PPP Let J be a countable base for the topology of Z. Enumerate N N J as
¸(k
n
, m
n
, W
n
))
n∈N
in such a way that k
n
≤ n for every n. Having chosen Borel sets B
ij
⊆ X for i ≤ n,
j ∈ N in such a way that the topology S
n
generated by ¦B
ij
: i ≤ n, j ∈ N¦ is a Polish topology ﬁner than
T, consider the set
C
n
= ¦x : f
B
k
n
,m
n
(x) ∈ W
n
¦.
This is TBorel, therefore S
n
Borel, so by 4A3H there is a Polish topology S
n+1
⊇ S
n
such that C
n
∈ S
n+1
.
Let ¸B
n+1,m
)
m∈N
run over a base for S
n+1
; by 423Fb, every B
n+1,m
belongs to B(X). Continue.
Let T
∗
be the topology generated by c = ¦B
ij
: i, j ∈ N¦. By 4A2Qf, T
∗
is Polish, because it is the
topology generated by
n∈N
S
n
. If W ∈ J, B ∈ c there are i, j ∈ N such that B = B
ij
and an n ∈ N such
that (i, j, W) = (k
n
, m
n
, W
n
); now
f
−1
B
[W] = C
n
∈ S
n+1
⊆ T
∗
.
As W is arbitrary, f
B
is T
∗
continuous. Also
X ∈ ¦B
0m
: m ∈ N¦ ⊆ c,
as required. QQQ
*424H Standard Borel spaces 169
(g) Deﬁne θ : X → Z
L
by setting θ(x)(E) = f
E
(x) for x ∈ X, E ∈ c. Then θ is injective. PPP Suppose
that x, y ∈ X and that x ,= y. For every g ∈ G,
g
−1
•
(g
•
x) = x ,= y = g
−1
•
(g
•
y),
so g
•
x ,= g
•
y and there is some m ∈ N such that g
•
x ∈ B
0m
while g
•
y / ∈ B
0m
, that is, g ∈ P
B
0m
(x) ¸P
B
0m
(y).
Thus G =
m∈N
P
B
0m
(x) ¸ P
B
0m
(y); by Baire’s theorem, there is some m ∈ N such that P
B
0m
(x) ¸ P
B
0m
(y)
is nonmeager. Because Q
B
0m
(x)´P
B
0m
(x) and Q
B
0m
(y)´P
B
0m
(y) are both meager,
¦g : f
B
0m
(x)(g) > 0¦ = Q
B
0m
(x) ,= Q
B
0m
(y) = ¦g : f
B
0m
(y)(g) > 0¦,
and
θ(x)(B
0m
) = f
B
0m
(x) ,= f
B
0m
(y) = θ(y)(B
0m
).
So θ(x) ,= θ(y). QQQ Because f
E
is T
∗
continuous for every E ∈ c, θ is T
∗
continuous.
(h) Let T
t
be the topology on X induced by θ; that is, the topology which renders θ a homeomorphism
between X and θ[X]. Because θ[X] ⊆ Z
L
is separable and metrizable, T
t
is separable and metrizable.
Because θ is T
∗
continuous, T
t
⊆ T
∗
.
(i) The action of G on X is continuous for the given topology S on G and T
t
on X. PPP For any E ∈ c,
(g, x) → θ(g
•
x)(E) = f
E
(g
•
x) = g
•
f
E
(x)
((e) above) is ST
t
continuous because the action of G on Z is continuous ((b) above) and f
E
: X → Z is
T
t
continuous (by the deﬁnition of T
t
). But this means that (g, x) → θ(g
•
x) is S T
t
continuous, so that
(g, x) → g
•
x is (ST
t
, T
t
)continuous. QQQ
(j) Let σ be a complete metric on G deﬁning the topology S, and τ a complete metric on X deﬁning
the topology T
∗
. (We do not need to relate σ to ρ in any way beyond the fact that they both give rise to
the same topology S.) For E ∈ c, V ∈ 1 and n ∈ N let S
EV n
be the set of those φ ∈ Z
L
such that either
φ(E)(g) = 0 for every g ∈ V or there is an F ∈ c such that F ⊆ E, diam
τ
(F) ≤ 2
−n
and φ(F)(g) > 0 for
some g ∈ V . Then S
EV n
is the union of a closed set and an open set, so is a G
δ
set in Z
L
(4A2C(ai)).
Consequently
Y = ¦φ : φ ∈ Z
L
, φ(X) = χG¦ ∩ θ[X] ∩
E∈L,V ∈\,n∈N
S
EV n
is a G
δ
subset of Z
L
, being the intersection of countably many G
δ
sets.
(k) θ[X] ⊆ Y . PPP Let x ∈ X. (i) P
X
(x) = G so Q
X
(x) = G and θ(x)(X) = f
X
(x) = χG. (ii) Of course
θ(x) ∈ θ[X]. (iii) Suppose that E ∈ c, V ∈ 1 and n ∈ N. If Q
E
(x) ∩V = ∅ then θ(x)(E)(g) = f
E
(x)(g) = 0
for every g ∈ V , and θ(x)(E) ∈ S
EV n
. Otherwise, V ∩ P
E
(x) is nonmeager. But
E =
¦F : F ∈ c, F ⊆ E, diam
τ
(F) ≤ 2
−n
¦,
so
P
E
(x) =
¦P
F
(x) : F ∈ c, F ⊆ E, diam
τ
(F) ≤ 2
−n
¦;
because c is countable, this is a countable union and there is an F ∈ c such that F ⊆ E, diam
τ
(F) ≤ 2
−n
and P
F
(x) ∩ V is nonmeager. In this case Q
F
(x) ∩ V is nonempty and θ(x)(F) = f
F
(x) is nonzero at
some point of V ; thus again θ(x) ∈ S
EV n
. As E, V and n are arbitrary, we have the result. QQQ
(l) (The magic bit.) Y ⊆ θ[X]. PPP Take any φ ∈ Y . Choose ¸E
n
)
n∈N
in c, ¸V
n
)
n∈N
in 1 and ¸˜ g
n
)
n∈N
in G as follows. E
0
= X and V
0
= G. Given that φ(E
n
) is nonzero at some point of V
n
, then, because
φ ∈ S
E
n
V
n
n
, there is an E
n+1
∈ c such that E
n+1
⊆ E
n
, diam
τ
(E
n+1
) ≤ 2
−n
and φ(E
n+1
) is nonzero
at some point of V
n
; say ˜ g
n
∈ V
n
is such that φ(E
n+1
)(˜ g
n
) > 0. Now we can ﬁnd a V
n+1
∈ 1 such that
˜ g
n
∈ V
n+1
⊆ V
n
and diam
σ
(V
n+1
) ≤ 2
−n
. Continue.
We are supposing also that φ ∈ θ[X], so we have a sequence ¸x
i
)
i∈N
in X such that ¸θ(x
i
))
i∈N
→ φ, that
is, ¸f
E
(x
i
))
i∈N
→ φ(E) for every E ∈ c. In particular,
lim
i→∞
f
E
n+1
(x
i
)(˜ g
n
) = φ(E
n+1
)(˜ g
n
) > 0
for every n. Let ¸i
n
)
n∈N
be a strictly increasing sequence in N such that f
E
n+1
(x
i
n
)(˜ g
n
) > 0 for every n.
Then
170 Descriptive set theory *424H
˜ g
n
∈ V
n+1
∩ Q
E
n+1
(x
i
n
) ⊆ V
n+1
∩ P
E
n+1
(x
i
n
);
there is therefore some g
n
∈ V
n+1
∩ P
E
n+1
(x
i
n
), so that g
n
•
x
i
n
∈ E
n+1
.
¸V
n
)
n∈N
is a nonincreasing sequence of sets with σdiameters converging to 0, so ¸g
n
)
n∈N
is a Cauchy
sequence for the complete metric σ. Similarly, ¸g
n
•
x
i
n
)
n∈N
is a Cauchy sequence for the complete metric
τ, because diam
τ
(E
n+1
) ≤ 2
−n
. We therefore have g ∈ G, y ∈ X such that ¸g
n
)
n∈N
→ g for S and
¸g
n
•
x
i
n
)
n∈N
→ y for T
∗
. In this case, ¸g
n
•
x
i
n
)
n∈N
→ y for the coarser topology T
t
, while ¸g
−1
n
)
n∈N
→ g
−1
for S. Because the action is (ST
t
, T
t
)continuous,
¸x
i
n
)
n∈N
= ¸g
−1
n
•
(g
n
•
x
i
n
))
n∈N
→ g
−1
•
y
for T
t
. But θ is continuous for T
t
, so
θ(g
−1
•
y) = lim
n→∞
θ(x
i
n
) = φ,
and φ ∈ θ[X]. As φ is arbitrary, Y ⊆ θ[X]. QQQ
(m) Thus θ[X] = Y is a G
δ
set in the compact metric space Z
L
, and is a Polish space in its induced
topology (4A2Qd). But this means that (X, T
t
), which is homeomorphic to θ[X], is also Polish.
(n) I have still to check that T
t
has the same Borel sets as T. But T, T
∗
and T
t
are all Polish topologies
and T
∗
is ﬁner than both the other two. By 423Fb, T
∗
has the same Borel sets as either of the others.
This completes the proof.
424X Basic exercises >>>(a) Let ¸(X
i
, Σ
i
))
i∈I
be a countable family of standard Borel spaces, and (X, Σ)
their direct sum, that is, X = ¦(x, i) : i ∈ I, x ∈ X
i
¦, Σ = ¦E : E ⊆ X, ¦x : (x, i) ∈ E¦ ∈ Σ
i
for every i¦.
Show that (X, Σ) is a standard Borel space.
>>>(b) Let (X, Σ) be a standard Borel space and T a countably generated σsubalgebra of Σ. Show that
there is an analytic Hausdorﬀ space Z such that T is isomorphic to the Borel σalgebra of Z. (Hint: by
4A3I, we can suppose that X is a Polish space and T is generated by a sequence of openandclosed sets,
corresponding to a continuous function from X to ¦0, 1¦
N
.)
>>>(c) Let (X, Σ) be a standard Borel space and T
1
, T
2
two countably generated σsubalgebras of Σ which
separate the same points, in the sense that if x, y ∈ X then there is an E ∈ T
1
such that x ∈ E, y ∈ X ¸ E
iﬀ there is an E
t
∈ T
2
such that x ∈ E
t
, y ∈ X ¸ E
t
. Show that T
1
= T
2
. (Hint: 424Xb, 423Fb.) In
particular, if T
1
separates the points of X then T
1
= Σ.
(d) Let A be a Dedekind σcomplete Boolean algebra. Show that A is isomorphic to the Borel σalgebra of
an analytic Hausdorﬀ space iﬀ it is isomorphic to a countably generated σsubalgebra of the Borel σalgebra
of [0, 1].
>>>(e) Let U be a separable Banach space. Show that its Borel σalgebra is generated, as σalgebra, by
the sets of the form ¦u : h(u) ≤ α¦ as h runs over the dual U
∗
and α runs over R.
>>>(f ) Let X be a compact metrizable space. Show that the Borel σalgebra of C(X) (the Banach space
of continuous realvalued functions on X) is generated, as σalgebra, by the sets ¦u : u ∈ C(X), u(x) ≥ α¦
as x runs over X and α runs over R.
(g) Let (X, Σ) be a standard Borel space, Y any set, and T a σalgebra of subsets of Y . Write T
∗
for
the σalgebra of subsets of Y generated by o(T), the family of sets obtainable from sets in T by Souslin’s
operation. Let W ∈ o(Σ
´
⊗T). Show that W[X] ∈ T
∗
and that there is a (T
∗
, Σ)measurable function
f : W[X] → X such that (f(y), y) ∈ W for every y ∈ W[X]. (Hint: 423M.)
(h) Let (X, Σ) be a standard Borel space, Y any set, and T a countably generated σalgebra of subsets of
Y . Let f : X → Y be a (Σ, T)measurable function, and write T
∗
for the σalgebra of subsets of Y generated
by o(T), the family of sets obtainable from sets in T by Souslin’s operation. Show that (i) f[X] ∈ T
∗
(ii)
there is a (T
∗
, Σ)measurable function g : f[X] → X such that gf is the identity on X. (Hint: start with
the case in which T separates the points of Y , so that the graph of f belongs to Σ
´
⊗T.)
424 Notes Standard Borel spaces 171
>>>(i) Show that 424Xc and 424Xh are both false if we omit the hypothesis ‘countably generated’. (Hint:
consider (i) the countablecocountable algebra of R (ii) the split interval (iii) 423Xi.)
(j) Let (X, Σ) be a standard Borel space. Show that if X is uncountable, Σ has a countably generated
σsubalgebra not isomorphic either to Σ or to TI for any set I. (Hint: 423L.)
(k) Let (X, Σ) be a standard Borel space. (i) Show that the Boolean algebra A = Σ/[X]
≤ω
is homoge
neous. (ii) Show that for every sequentially ordercontinuous Boolean homomorphism π : A → A there is a
(Σ, Σ)measurable f : X → X representing π in the sense that π(E
•
) = (f
−1
[E])
•
for every E ∈ Σ. (Hint:
reduce to the case X = ¦0, 1¦
N
.) (iii) Show that for every π ∈ Aut A there is an automorphism f of (X, Σ)
representing π. (Hint: 344Ya.)
424Y Further exercises (a) Let (X, T) be a Polish space and T the family of closed subsets of X. Let
Σ be the σalgebra of subsets of T generated by the sets c
H
= ¦F : F ∈ T, F ∩ H ,= ∅¦ as H runs over the
open subsets of X. (i) Show that (T, Σ) is a standard Borel space. (Hint: take a complete metric ρ deﬁning
the topology of X. Set S
∗
=
n∈N
N
n
and choose a family ¸U
σ
)
σ∈S
∗ of open sets in X such that U
∅
= X,
diamU
σ
≤ 2
−n
whenever #(σ) = n + 1, U
σ
=
i∈N
U
σ
i
for every σ, U
σ
i
⊆ U
σ
for every σ, i. Deﬁne
f : T → ¦0, 1¦
S
∗
by setting f(F)(σ) = 1 if F ∩ U
σ
,= ∅, 0 otherwise. Show that Z = f[T] is a Borel set and
that f is an isomorphism between Σ and the Borel σalgebra of Z.) (This is the Eﬀros Borel structure
on T.) (ii) Show that [X]
n
∈ Σ for every n ∈ N.
(b) Let (X, Σ) be a standard Borel space. Let T be the family of Polish topologies on X for which Σ is
the Borel σalgebra. Show that any sequence in T has an upper bound in T, and that any sequence with a
lower bound has a least upper bound.
(c) Let (X, Σ) be a standard Borel space. Say that C ⊆ X is coanalytic if its complement belongs to
o(Σ). Show that for any such C the partially ordered set Σ ∩ TC has coﬁnality 1 if C ∈ Σ and coﬁnality
ω
1
otherwise. (Hint: 423P.)
(d) Let I

be the split interval. Show that there is a σalgebra Σ of subsets of I

such that (I

, Σ) is a
standard Borel space and ¦(x, y) : x, y ∈ I

, x ≤ y¦ ∈ Σ
´
⊗Σ.
424 Notes and comments In this treatise I have generally indulged my prejudice in favour of ‘complete’
measures. Consequently Borel σalgebras, as such, have taken subordinate roles. But important parts of the
theory of Lebesgue measure, and Radon measures on Polish spaces in general, are associated with the fact
that these are completions of measures deﬁned on standard Borel spaces. Moreover, such spaces provide a
suitable framework for a large part of probability theory. Of course they become deﬁcient in contexts where
we need to look at uncountable independent families of random variables, and there are also diﬃculties with
σsubalgebras, even countably generated ones, since these can correspond to the Borel algebras of general
analytic spaces, which will not always be standard Borel structures (424F, 423L). 424Xf and 424Ya suggest
the ubiquity of standard Borel structures; the former shows that they are not always presented as countably
generated algebras, while the latter is an example in which we have to make a special construction in order
to associate a topology with the algebra. The theory is of course dominated by the results of ¸423, especially
423Fb and 423I.
I include 424H in this section because there is no other convenient place for it, but I have an excuse: the
idea of ‘Borel measurable action’ can, in this context, be described entirely in terms of σalgebras, since the
Borel algebra of GX is just the σalgebra product of the Borel algebras of the factors (as in 424Bb). Of
course for the theorem as expressed here we do need to know that G has a Polish group structure; but X
could be presented just as a standard Borel space. The result is a dramatic expression of the fact that, given
a standard Borel space (X, Σ), we have a great deal of freedom in deﬁning a corresponding Polish topology
on X.
172 Topological measure spaces II
Chapter 43
Topologies and measures II
The ﬁrst chapter of this volume was ‘general’ theory of topological measure spaces; I attempted to
distinguish the most important properties a topological measure can have – inner regularity, τadditivity –
and describe their interactions at an abstract level. I now turn to rather more specialized investigations,
looking for features which oﬀer explanations of the behaviour of the most important spaces, radiating
outwards from Lebesgue measure.
In eﬀect, this chapter consists of three distinguishable parts and two appendices. The ﬁrst three sections
are based on ideas from descriptive set theory, in particular Souslin’s operation (¸431); the properties of
this operation are the foundation for the theory of two classes of topological space of particular importance
in measure theory, the Kanalytic spaces (¸432) and the analytic spaces (¸433). The second part of the
chapter, ¸¸434435, collects miscellaneous results on Borel and Baire measures, looking at the ways in which
topological properties of a space determine properties of the measures it carries. In ¸436 I present the
most important theorems on the representation of linear functionals by integrals; if you like, this is the
inverse operation to the construction of integrals from measures in ¸122. The ideas continue into ¸437,
where I discuss spaces of signed measures representing the duals of spaces of continuous functions. The
ﬁrst appendix, ¸438, looks at a special topic: the way in which the patterns in ¸¸434435 are aﬀected if we
assume that our spaces are not unreasonably complex in a rather special sense deﬁned in terms of measures
on discrete spaces. Finally, I end the chapter with a further collection of examples, mostly to exhibit
boundaries to the theorems of the chapter, but also to show some of the variety of the structures we are
dealing with.
431 Souslin’s operation
I begin the chapter with a short section on Souslin’s operation (¸421). The basic facts we need to know
are that (in a complete locally determined measure space) the family of measurable sets is closed under
Souslin’s operation (431A), and that the kernel of a Souslin scheme can be approximated from within in
measure (431D).
431A Theorem Let (X, Σ, µ) be a complete locally determined measure space. Then Σ is closed under
Souslin’s operation.
proof Let ¸E
σ
)
σ∈S
be a Souslin scheme in Σ with kernel A. Write
S
∗
=
k∈N
N
k
= S ∪ ¦∅¦.
If F ∈ Σ and µF < ∞, then A∩ F ∈ Σ. PPP For each σ ∈ S
∗
, set
A
σ
=
φ∈N
N
,φ⊇σ
n≥1
E
φn
,
and let G
σ
be a measurable envelope of A
σ
∩F. Because A
σ
⊆ E
σ
(writing E
∅
= X), we may suppose that
G
σ
⊆ E
σ
∩ F. Now, for any σ ∈ S
∗
,
A
σ
∩ F =
i∈N
A
σ
i
∩ F ⊆
i∈N
G
σ
i
,
so
H
σ
= G
σ
¸
i∈N
G
σ
i
is negligible.
Set H =
σ∈S
∗
H
σ
, so that H is negligible. Take any x ∈ G
∅
¸ H. Choose ¸φ(i))
i∈N
inductively, as
follows. Given that σ = ¸φ(i))
i<k
has been chosen and x ∈ G
σ
, then x / ∈ H
σ
, so there must be some j ∈ N
such that x ∈ G
σ
j
; set φ(k) = j, and continue. Now
x ∈
k≥1
G
φk
⊆
k≥1
E
φk
⊆ A.
Thus we see that G
∅
¸ H ⊆ A; as G
∅
⊆ F, G
∅
¸ H ⊆ A∩F. On the other hand, A∩F ⊆ G
∅
. Because H
is negligible and µ is complete, A∩ F ∈ Σ. QQQ
Because µ is locally determined, it follows that A ∈ Σ. As ¸E
σ
)
σ∈S
is arbitrary, Σ is closed under Souslin’s
operation.
431D Souslin’s operation 173
431B Corollary If (X, T, Σ, µ) is a complete locally determined topological measure space, every Souslin
F set in X (deﬁnition: 421K) is measurable.
431C Corollary Let X be a set and θ an outer measure on X. Let µ be the measure deﬁned by
Carath´eodory’s method, and Σ its domain. Then Σ is closed under Souslin’s operation.
proof Let ¸E
σ
)
σ∈S
be a Souslin scheme in Σ with kernel A. Take any C ⊆ X such that θC < ∞. Then
θ
C
= θTC is an outer measure on C; let µ
C
be the measure on C deﬁned from θ
C
by Carath´eodory’s
method, and Σ
C
its domain. If σ ∈ S, D ⊆ C then
θ
C
(D ∩ C ∩ E
σ
) +θ
C
(D ¸ (C ∩ E
σ
)) = θ(D ∩ E
σ
) +θ(D ¸ E
σ
)
= θD = θ
C
D;
as D is arbitrary, C ∩ E
σ
∈ Σ
C
. µ
C
is a complete totally ﬁnite measure, so 431A tells us that the kernel of
the Souslin scheme ¸C ∩E
σ
)
σ∈S
belongs to Σ
C
. But this is just C ∩A (applying 421Cb to the identity map
from C to X.) So
θ(C ∩ A) +θ(C ¸ A) = θ
C
(C ∩ A) +θ
C
(C ¸ A) = θ
C
C = θC.
As C is arbitrary, A ∈ Σ (113D). As ¸E
σ
)
σ∈S
is arbitrary, we have the result.
431D Theorem Let (X, Σ, µ) be a complete locally determined measure space, and ¸E
σ
)
σ∈S
a Souslin
scheme in Σ with kernel A. Then
µA = sup¦µ(
_
φ∈K
n≥1
E
φn
) : K ⊆ N
N
is compact¦
= sup¦µ(
_
φ≤ψ
n≥1
E
φn
) : ψ ∈ N
N
¦,
writing φ ≤ ψ if φ(i) ≤ ψ(i) for every i ∈ N.
proof (a) By 431A, A is measurable. For K ⊆ N
N
, set H
K
=
φ∈K
n≥1
E
φn
. Of course H
K
⊆ A, and
we know from 421M (or otherwise) that H
K
∈ Σ if K is compact. So surely µA ≥ µH
K
for every compact
K ⊆ N
N
. If ψ ∈ N
N
, then ¦φ : φ ≤ ψ¦ =
i∈N
(ψ(i) + 1) is compact. We therefore have
µA ≥ sup¦µ(
_
φ∈K
n≥1
E
φn
) : K ⊆ N
N
is compact¦
≥ sup¦µ(
_
φ≤ψ
n≥1
E
φn
) : ψ ∈ N
N
¦.
So what I need to prove is that
µA ≤ sup¦µ(
φ≤ψ
n≥1
E
φn
) : ψ ∈ N
N
¦.
(b) Fix on a set F ∈ Σ of ﬁnite measure. For σ ∈ S
∗
=
k∈N
N
k
set
A
σ
=
φ∈N
N
,φ⊇σ
n≥1
E
φn
.
We need to know that A
σ
belongs to Σ; this follows from 431A, because writing E
t
τ
= E
τ
if τ ⊆ σ or σ ⊆ τ,
∅ otherwise,
A
σ
=
φ∈N
N
n≥1
E
t
φn
∈ o(Σ) = Σ,
writing o for Souslin’s operation, as in ¸421.
Let > 0, and take a family ¸
σ
)
σ∈S
∗ of strictly positive real numbers such that
σ∈S
∗
σ
≤ . For each
σ ∈ S
∗
we have A
σ
=
i∈N
A
σ
i
, so there is an m
σ
∈ N such that
µ(F ∩ A
σ
¸
i≤m
σ
A
σ
i
) ≤
σ
.
174 Topological measure spaces II 431D
Deﬁne ψ ∈ N
N
by saying that
ψ(k) = max¦m
σ
: σ ∈ N
k
, σ(i) ≤ ψ(i) for every i < k¦
for each k ∈ N. Set
H =
φ∈N
N
,φ≤ψ
n≥1
E
φn
.
(c) Set
G =
σ∈S
∗
F ∩ A
σ
¸
i≤m
σ
A
σ
i
,
so that µG ≤ , by the choice of the
σ
and the m
σ
. Then F ∩ A ¸ G ⊆ H. PPP If x ∈ F ∩ A ¸ G, choose
¸φ(i))
i∈N
inductively, as follows. Given that φ(i) ≤ ψ(i) for i < k and x ∈ A
σ
, where σ = ¸φ(i))
i<k
,
then x / ∈ A
σ
¸
j≤m
σ
A
σ
j
, so there must be some j ≤ m
σ
such that x ∈ A
σ
j
; set φ(k) = j; because
σ ∈
i<k
(ψ(i) + 1), j ≤ m
σ
≤ ψ(k), and the induction continues. At the end of the induction, φ ≤ ψ and
x ∈
n≥1
A
φn
⊆
n≥1
E
φn
⊆ H. QQQ
(d) It follows that
µ(A∩ F) ≤ µG+µH ≤ +µH.
As F and are arbitrary, and µ is semiﬁnite,
µA ≤ sup¦µ(
φ≤ψ
n≥1
E
φn
) : ψ ∈ N
N
¦,
and the proof is complete.
431E Corollary If (X, T, Σ, µ) is a topological measure space and E ⊆ X is a SouslinF set with ﬁnite
outer measure, then µ
∗
E = sup¦µF : F ⊆ E is closed¦.
proof Let ˜ µ be the c.l.d. version of µ (213E). Let ¸E
σ
)
σ∈S
be a Souslin scheme of closed sets with kernel
E. Then 213Fb and 431D tell us that
µ
∗
E = ˜ µE = sup
K⊆N
N
is compact
µF
K
,
where F
K
=
φ∈K
n≥1
E
φn
for K ⊆ N
N
. But every F
K
is closed, by 421M. So µ
∗
E ≤ sup
F⊆E is closed
µF;
as the reverse inequality is trivial, we have the result.
*431F There is a topological version of 431A, as follows.
Theorem Let X be any topological space, and
´
B its Baire property algebra.
(a) For any A ⊆ X, there is a Baire property envelope of A, that is, a set E ∈
´
B such that A ⊆ E and
E ¸ F is meager whenever A ⊆ F ∈
´
B.
(b)
´
B is closed under Souslin’s operation.
proof (a) By 4A3Ra, there is an open set H ⊆ X such that A¸ H is meager and H ∩G is empty whenever
G ⊆ X is open and A∩ G is meager. Set E = A∪ H; then E ⊇ A and E´H = A¸ H is meager, so E ∈
´
B.
If A ⊆ F ∈
´
B, let G be an open set such that G´(X ¸ F) is meager. Then G ∩ A ⊆ G ∩ F is meager, so
G∩ H is empty and E ¸ F ⊆ (E´H) ∪ (G´(X ¸ F)) is meager. Thus E is a Baire property envelope of A.
(b) Let ¸E
σ
)
σ∈S
be a Souslin scheme in
´
B with kernel A. Write
S
∗
=
k∈N
N
k
= S ∪ ¦∅¦.
For each σ ∈ S
∗
, set
A
σ
=
φ∈N
N
,φ⊇σ
n≥1
E
φn
,
and let G
σ
be a Baire property envelope of A
σ
as described in (a). Because A
σ
⊆ E
σ
(writing E
∅
= X), we
may suppose that G
σ
⊆ E
σ
. Now, for any σ ∈ S
∗
,
A
σ
=
i∈N
A
σ
i
⊆
i∈N
G
σ
i
,
so
431 Notes Souslin’s operation 175
H
σ
= G
σ
¸
i∈N
G
σ
i
is meager.
Set H =
σ∈S
∗
H
σ
, so that H is negligible. Take any x ∈ G
∅
¸ H. Choose ¸φ(i))
i∈N
inductively, as
follows. Given that σ = ¸φ(i))
i<k
has been chosen and x ∈ G
σ
, then x / ∈ H
σ
, so there must be some j ∈ N
such that x ∈ G
σ
j
; set φ(k) = j, and continue. Now
x ∈
k≥1
G
φk
⊆
k≥1
E
φk
⊆ A.
Thus we see that G
∅
¸ H ⊆ A. On the other hand, A ⊆ G
∅
, so G
∅
´A is meager and A ∈
´
B. As ¸E
σ
)
σ∈S
is arbitrary,
´
B is closed under Souslin’s operation.
431X Basic exercises (a) Let (X, T, Σ, µ) be a complete locally determined topological measure space,
(Y, S) a topological space, and f : X → Y a measurable function. Let B(Y ) be the Borel algebra of Y .
Show that f
−1
[B] ∈ Σ for every B ∈ o(B(Y )).
(b) Let X be a topological space and µ a semiﬁnite topological measure on X which is inner regular
with respect to the SouslinF sets. Show that µ is inner regular with respect to the closed sets.
(c) Let (X, Σ, µ) be a complete locally determined measure space, and ¸E
σ
)
σ∈S
a regular Souslin scheme
in Σ (deﬁnition: 421Xn) with kernel A. Show that
µA = sup
φ∈N
N µ(
n≥1
E
φn
).
>>>(d) Let (X, Σ, µ) be a measure space with locally determined negligible sets (deﬁnition: 213I), and
¸E
σ
)
σ∈S
a Souslin scheme in Σ with kernel A. Show that
µ
∗
A = sup
ψ∈N
N µ(
φ≤ψ
n≥1
E
φn
).
>>>(e) Let (X, Σ, µ) be a semiﬁnite measure space, and ¸E
σ
)
σ∈S
a Souslin scheme in Σ with kernel A.
Show that
µ
∗
A = sup
ψ∈N
N µ(
φ≤ψ
n≥1
E
φn
).
431Y Further exercises (a) Let (X, Σ, µ) be a complete measure space with the measurable envelope
property (213Xl). Show that Σ is closed under Souslin’s operation.
(b) Let X be a set, Σ a σalgebra of subsets of X, and 1 a σideal of subsets of X such that 1 ⊆ Σ.
Suppose that for every A ⊆ X there is an F ∈ Σ such that A ⊆ F and F ¸ E ∈ 1 whenever A ⊆ E ∈ Σ.
Show that Σ is closed under Souslin’s operation.
(c) Let X be a set, Σ a σalgebra of subsets of X, and 1 an ω
1
saturated σideal of Σ; suppose that
A ∈ Σ whenever A ⊆ B ∈ 1. Show that Σ is closed under Souslin’s operation.
431 Notes and comments From the point of view of measure theory, the most important property
of Souslin’s operation, after its idempotence, is the fact that (for many measure spaces) the family of
measurable sets is closed under the operation (431A). The proof I give here is based on the concept of
measurable envelope, which can be used in other cases of great interest (431Yb, 431Yc). But for some
applications it is also very important to know that if A is the kernel of a Souslin scheme ¸E
σ
)
σ∈S
, then A
can be approximated from inside by sets of the form H =
φ≤ψ
n≥1
E
φn
(431D, 431Xd), which belong
to the σalgebra generated by the E
σ
(421M). A typical application of this idea is when every E
σ
is a Borel
subset of R; then we ﬁnd not only that A is Lebesgue measurable (indeed, measurable for every Radon
measure on R) but that (for any given Radon measure µ) the Souslin scheme itself provides a Borel subset
H of A of measure approximating the measure of A.
Let me repeat that the essence of descriptive set theory is that we are not satisﬁed merely to know that
a set of a certain type exists. We want also to know how to build it, because we expect that an explicit
176 Topologies and measures II 431 Notes
construction will be valuable later on. For instance, the construction given in 431D shows that if the Souslin
scheme consists of closed compact sets, the set H will be compact (421Xo).
I mention 431B as a typical application of 431A, even though it is both obvious and obviously less than
what can be said. The algebras Σ of this section are algebras closed under Souslin’s operation. In a complete
locally determined topological measure space, the algebra Σ of measurable sets includes the open sets (by
deﬁnition), therefore the Borel algebra B, therefore o(B); but now we can take the algebra /
1
generated
by o(B), and /
1
and o(/
1
) will also be included in Σ, so that Σ will included the algebra /
2
generated
by o(/
1
), and so on. (Note that o(/
1
) includes the σalgebra generated by /
1
, by 421F, so I do not need
to mention that separately.) We have to run through all the countable ordinals before we can be sure of
getting to the smallest algebra /
ω
1
=
ξ<ω
1
/
ξ
which contains every open set and is closed under Souslin’s
operation, and we shall then have /
ω
1
⊆ Σ.
The result in 431D is one of the special features of measures. (A similar result, based on rather diﬀerent
hypotheses, is in 432J.) But the argument of 431A can be applied in many other cases; see 431Yb431Yc.
A striking one is 431F, which will be useful in Volume 5.
432 Kanalytic spaces
I describe the basic measuretheoretic properties of Kanalytic spaces (¸422). I start with ‘elementary’
results (432A432C), assembling ideas from ¸¸421, 422 and 431. The main theorem of the section is 432D,
one of the leading cases of the general extension theorem 416P. An important corollary (432G) gives a
suﬃcient condition for the existence of pullback measures. I brieﬂy mention ‘capacities’ (432I432J).
432A Proposition Let (X, T, Σ, µ) be a complete locally determined Hausdorﬀ topological measure
space. Then every Kanalytic subset of X is measurable.
proof If A ⊆ X is Kanalytic, it is SouslinF (422Ha), therefore measurable (431B).
432B Theorem Let X be a Kanalytic Hausdorﬀ space, and µ a semiﬁnite topological measure on X.
Then
µX = sup¦µK : K ⊆ X is compact¦.
proof If γ < µX, there is an E ∈ domµ such that γ < µE < ∞; set νF = µ(E ∩ F) for every Borel set
F ⊆ X, so that ν is a totally ﬁnite Borel measure on X, and νX > γ. Let ˆ ν be the completion of ν. Let
R ⊆ N
N
X be an uscocompact relation such that R[N
N
] = X. Set F
σ
= R[I
σ
] for σ ∈ S =
k≥1
N
k
,
where I
σ
= ¦φ : σ ⊆ φ ∈ N
N
¦. Because R is closed in N
N
X (422Da), X is the kernel of the Souslin scheme
¸F
σ
)
σ∈S
(421I). By 431D, there is a compact L ⊆ N
N
such that ˆ ν(
φ∈L
n∈N
F
φn
) ≥ γ. But, by 421I, this
is just ˆ ν(R[L]); and R[L] is compact, by 422D(ei). So µR[L] is deﬁned, with µR[L] ≥ νR[L] = ˆ νR[L], and
we have a compact subset of X of measure at least γ. As γ is arbitrary, the proposition is proved.
432C Proposition Let X be a Hausdorﬀ space such that all its open sets are Kanalytic, and µ a Borel
measure on X.
(a) If µ is semiﬁnite, it is tight (that is, inner regular with respect to the compact sets).
(b) If µ is locally ﬁnite, its completion is a Radon measure on X.
proof (a) By 422Hb, every open subset of X is SouslinF. Applying 421F to the family c of closed subsets
of X, we see that every Borel subset of X is SouslinF, therefore Kanalytic (by 422Hb in the opposite
direction). Now suppose that E ⊆ X is a Borel set. Then the subspace measure µ
E
is a semiﬁnite Borel
measure on the Kanalytic space E, so by 432B µE = sup
K⊆E is compact
µK. As E is arbitrary, µ is tight.
(b) Because X is Lindel¨of (422Gg), µ is σﬁnite (411Ge), therefore semiﬁnite. So (a) tells us that µ is
tight. By 416F, its c.l.d. version is a Radon measure. But (because µ is σﬁnite) this is just its completion
(213Ha).
432G Kanalytic spaces 177
432D Theorem (Aldaz & Render 00) Let X be a Kanalytic Hausdorﬀ space and µ a locally ﬁnite
measure on X which is inner regular with respect to the closed sets. Then µ has an extension to a Radon
measure on X. In particular, µ is τadditive.
proof The point is that if µE > 0 then there is a compact K ⊆ E such that µ
∗
K > 0. PPP Write Σ for the
domain of µ. Take γ < µE. Because X is Lindel¨of (422Gg), µ is σﬁnite (411Ge), therefore semiﬁnite. Let
E
t
⊆ E be such that γ < µE
t
< ∞. Because µ is inner regular with respect to the closed sets, there is a
closed set F ⊆ E such that µF > γ. F is Kanalytic (422Gf); let R ⊆ N
N
F be an uscocompact relation
such that R[N
N
] = F. For σ ∈ S
∗
=
n∈N
N
n
set
F
σ
= ¦x : (φ, x) ∈ R for some φ ∈ N
N
such that φ(i) ≤ σ(i) for every i < #(σ)¦.
Then ¸F
σ
i
)
i∈N
is a nondecreasing sequence with union F
σ
, so
µ
∗
F
σ
= sup
i∈N
µ
∗
F
σ
i
for every σ ∈ S
∗
. We can therefore ﬁnd a sequence ψ ∈ N
N
such that
µ
∗
F
ψn
> γ
for every n ∈ N. Set
K = ¦φ : φ ∈ N
N
, φ(i) ≤ ψ(i) for every i ∈ N¦;
then K =
n∈N
(ψ(n) + 1) is compact, so R[K] is compact (422D(ei)).
??? Suppose, if possible, that µ
∗
R[K] < γ. Then there is an H ∈ Σ such that R[K] ⊆ H ⊆ F and
µ(F ¸ H) > µF −γ. Because µ is inner regular with respect to the closed sets, there is a closed set F
t
∈ Σ
such that F
t
⊆ F ¸ H and µF
t
> µF −γ. Since R[K] ∩F
t
= ∅, K∩R
−1
[F
t
] = ∅. R
−1
[F
t
] is closed, because
R is uscocompact, so there is some n such that
L = ¦φ : φ ∈ N
N
, φn = φ
t
n for some φ
t
∈ K¦
does not meet R
−1
[F
t
] (4A2F(hvi)), and R[L] ∩ F
t
= ∅. But L is just ¦φ : φ(i) ≤ ψ(i) for every i < n¦, so
R[L] = F
ψn
, and
γ < µ
∗
F
ψn
≤ µ(F ¸ F
t
) < γ,
which is absurd. XXX
Thus µ
∗
R[K] ≥ γ. As γ > 0, we have the result. QQQ
Now the theorem follows at once from 416P(ii)⇒(i).
432E Corollary Let X be a Kanalytic Hausdorﬀ space, and µ a locally ﬁnite quasiRadon measure on
X. Then µ is a Radon measure.
proof By 432D, µ has an extension to a Radon measure µ
t
. But of course µ and µ
t
must coincide, by 415H
or otherwise.
432F Corollary Let X be a Kanalytic Hausdorﬀ space, and ν a locally ﬁnite Baire measure on X.
Then ν has an extension to a Radon measure on X; in particular, it is τadditive. If the topology of X is
regular, the extension is unique.
proof Because X is Lindel¨of (422Gg), ν is σﬁnite, therefore semiﬁnite; by 412D, it is inner regular with
respect to the closed sets. So 432D tells us that it has an extension to a Radon measure on X. Since the
extension is τadditive, so is ν.
If X is regular, then it must be completely regular (4A2H(bi)), and the family ( of cozero sets is a base
for the topology closed under ﬁnite unions. If µ, µ
t
are Radon measures extending ν, they agree on (, and
must be equal, by 415H(iv).
432G Corollary Let X be a Kanalytic Hausdorﬀ space, Y a Hausdorﬀ space and ν a locally ﬁnite
measure on Y which is inner regular with respect to the closed sets. Let f : X → Y be a continuous
function such that f[X] has full outer measure in Y . Then there is a Radon measure µ on X such that f is
inversemeasurepreserving for µ and ν. If ν is Radon, it is precisely the image measure µf
−1
.
178 Topologies and measures II 432G
proof (a) Write T for the domain of ν, and set Σ
0
= ¦f
−1
[F] : f ∈ T¦, so that Σ
0
is a σalgebra of subsets
of X, and we have a measure µ
0
on X deﬁned by setting µ
0
f
−1
[F] = νF whenever F ∈ T (132Yd).
(b) If E ∈ Σ
0
and γ < µ
0
E, there is an F ∈ T such that E = f
−1
[F]. Now there is a closed set F
t
⊆ F
such that νF
t
≥ γ. Because f is continuous, f
−1
[F
t
] is closed, and we have f
−1
[F
t
] ⊆ E and µ
0
f
−1
[F
t
] ≥ γ.
As E and γ are arbitrary, µ
0
is inner regular with respect to the closed sets.
If x ∈ X, then (because ν is locally ﬁnite) there are an open set H ⊆ Y and an F ∈ T such that
f(x) ∈ H ⊆ F, νF < ∞.
In this case f
−1
[H] is an open set containing x and
µ
∗
0
f
−1
[H] ≤ µ
0
f
−1
[F] = νF < ∞.
As x is arbitrary, µ
0
is locally ﬁnite.
(c) By 432D, there is a Radon measure µ on X extending µ
0
. Because f is inversemeasurepreserving
for µ
0
and ν, it is surely inversemeasurepreserving for µ and ν.
The image measure µf
−1
extends ν, so must be locally ﬁnite; it is therefore a Radon measure (418I). So
if ν itself is a Radon measure, it must be identical with µf
−1
, by 416Eb.
432H Corollary Suppose that X is a set and that S, T are Hausdorﬀ topologies on X such that (X, T)
is Kanalytic and S ⊆ T. Then the totally ﬁnite Radon measures on X are the same for S and T.
proof Write f for the identity function on X regarded as a continuous function from (X, T) to (X, S).
If µ is a totally ﬁnite TRadon measure on X, then µ = µf
−1
is SRadon, by 418I. If ν is a totally ﬁnite
SRadon measure on X, then 432G tells us that it is of the form µ = µf
−1
for some TRadon measure µ,
that is, is itself TRadon.
432I Capacitability The next result is not exactly measure theory as studied in most of this treatise;
but it is clearly very close to the other ideas of this section, and it has important applications to measure
theory in the narrow sense.
Deﬁnition Let (X, T) be a topological space. A (Choquet) capacity on X is a function c : TX → [0, ∞]
such that
(i) c(A) ≤ c(B) whenever A ⊆ B ⊆ X;
(ii) lim
n→∞
c(A
n
) = c(A) whenever ¸A
n
)
n∈N
is a nondecreasing sequence of subsets of X with union A;
(iii) c(K) = inf¦c(G) : G ⊇ K is open¦ for every compact set K ⊆ X.
432J Theorem (Choquet 55) Let X be a Hausdorﬀ space and c a capacity on X. If A ⊆ X is
Kanalytic, then c(A) = sup¦c(K) : K ⊆ A is compact¦.
proof Take γ < c(A). Let R ⊆ N
N
X be an uscocompact relation such that R[N
N
] = A; for σ ∈ S
∗
=
n∈N
N
n
set
A
σ
= ¦x : (φ, x) ∈ R for some φ ∈ N
N
such that φ(i) ≤ σ(i) for every i < #(σ)¦.
Then ¸A
σ
i
)
i∈N
is a nondecreasing sequence with union A
σ
, so
c(A
σ
) = sup
i∈N
c(A
σ
i
)
for every σ ∈ S
∗
. We can therefore ﬁnd a sequence ψ ∈ N
N
such that c(A
ψn
) > γ for every n ∈ N. Set
K = ¦φ : φ ∈ N
N
, φ(i) ≤ ψ(i) for every i ∈ N¦;
then K =
n∈N
(ψ(n) + 1) is compact, so R[K] is compact (422D(ei)).
??? Suppose, if possible, that c(R[K]) < γ. Then, by (iii) of 432I, there is an open set G ⊇ R[K] such
that c(G) < γ. Set F = X ¸ G, so that F is closed and K ∩ R
−1
[F] = ∅. R
−1
[F] is closed, because R is
uscocompact, so there is some n such that
L = ¦φ : φ ∈ N
N
, φn = φ
t
n for some φ
t
∈ K¦
does not meet R
−1
[F] (4A2F(hvi)), and R[L] ∩F = ∅, that is, R[L] ⊆ G. But L is just ¦φ : φ(i) ≤ ψ(i) for
every i < n¦, so R[L] = A
ψn
, and
432 Notes Kanalytic spaces 179
γ < c(A
ψn
) ≤ c(G) < γ,
which is absurd. XXX
Thus c(R[K]) ≥ γ. As γ is arbitrary and R[K] is compact, we have the result.
432X Basic exercises (a) Put 422Xf, 431Xb and 432D together to prove 432C.
(b) Let X be a Kanalytic Hausdorﬀ space, and µ a measure on X which is outer regular with respect
to the open sets. Show that µX = sup
K⊆X is compact
µ
∗
K. (Hint: use the argument of part (a) of the proof
of 432D.)
>>>(c) Let X be a Kanalytic Hausdorﬀ space, and µ a semiﬁnite topological measure on X. Show that if
either µ is inner regular with respect to the closed sets or X is regular and µ is a τadditive Borel measure,
then µ is tight.
(d) Use 422Gf, 432B and 416C to prove 432E.
>>>(e) Suppose that X is a set and that S, T are Hausdorﬀ topologies on X such that (X, T) is Kanalytic
and S ⊆ T. Let (Z, U, T, ν) be a Radon measure space and f : Z → X a function which is almost continuous
for U and S. Show that f is almost continuous for U and T. (Hint: it is enough to consider totally ﬁnite ν;
show that νf
−1
is TRadon, so is inner regular for ¦K : T
K
= S
K
¦, writing T
K
for the subspace topology
induced by T on K.)
(f ) Let X be a topological space and µ a locally ﬁnite measure on X which is inner regular with respect
to the closed sets. Show that µ
∗
is a capacity.
(g) Let X be a topological space and F a closed subset of X. Deﬁne c : TX → ¦0, 1¦ by setting c(A) = 1
if A meets F, 0 otherwise. Show that c is a capacity on X.
(h) Let X and Y be Hausdorﬀ spaces, and R ⊆ X Y an uscocompact relation. Show that if c is a
capacity on Y , then A → c(R[A]) is a capacity on X.
(i) Use 432J and 432Xf to shorten the proof of 432D.
432Y Further exercises (a) Show that there are a Kanalytic Hausdorﬀ space X and a probability
measure µ on X such that (i) µ is inner regular with respect to the Borel sets (ii) the domain of µ includes
a base for the topology of X (iii) every compact subset of X is negligible. Show that there is no extension
of µ to a topological measure on X.
432 Notes and comments The measuretheoretic properties of Kanalytic spaces can largely be sum
marised in the slogan ‘Kanalytic spaces have lots of compact sets’. I said above that it is sometimes helpful
to think of Kanalytic spaces as an amalgam of compact Hausdorﬀ spaces and SouslinF subsets of R. For
the former, it is obvious that they have many compact subsets; for the latter, it is not obvious, but is of
course one of their fundamental properties, deducible from 422De. 432B and part (a) of the proof of 432D
are typical manifestations of the phenomenon. The real point of these theorems is that we can extend a
Borel or Baire measure to a Radon measure with no assumption of τadditivity (432F). A Radon measure
must be τadditive just because it is tight. A (locally ﬁnite) Borel or Baire measure must be τadditive
whenever the measurable open sets are Kanalytic.
The condition ‘every open set is Kanalytic’ in 432C is of course a very strong one in the context of
compact Hausdorﬀ spaces (422Xe). But for analytic spaces it is automatically satisﬁed (423E), and that is
the side on which the principal applications of 432C appear.
The results which I call corollaries of 432D can mostly be proved by more direct methods (see 432Xd),
but the line I choose here seems to be the most powerful technique. Indeed it can be used to deal with 432C
as well (432Xa).
In ¸434 I will discuss ‘universally measurable’ sets in topological spaces. In fact Kanalytic sets are
universally measurable in a particularly strong sense (432A). The point here is that Kanalyticity is intrinsic;
180 Topologies and measures II 432 Notes
a Kanalytic space is measurable whenever embedded as a subspace of a (complete locally determined)
topological measure space.
The theorems here touch on two phenomena of particular importance. First, in 432G we have an example
of ‘pulling back’ a measure, that is, we have a measure ν on a set Y and a function f : X → Y and seek
a Radon measure µ on X such that f is inversemeasurepreserving, or, even better, such that ν = µf
−1
.
There was a similar result in 418L. In both cases we have to suppose that f is continuous and (in eﬀect) that
ν is a Radon measure. (This is not part of the hypotheses of 432G, but of course it is an easy consequence of
them, using 432B.) In 418L, we need a special hypothesis to ensure that there are enough compact subsets
of X to carry an appropriate Radon measure; in 432G, this is an automatic result of assuming that X is
Kanalytic. Both 418L and 432G can be regarded as consequences of Henry’s theorem (416M). The diﬃculty
arises from the requirement that µ should be a Radon measure; if we do not insist on this there is a much
simpler solution, since we need suppose only that f[X] is of full outer measure (132Yd).
The next theme I wish to mention is a related one, the investigation of comparable topologies. If S and
T are (Hausdorﬀ) topologies on a set X, and S is coarser than T (so that (X, S) is a continuous image
of (X, T)), then 418I tells us that any totally ﬁnite TRadon measure is SRadon. We very much want to
know when the reverse is true, so that the (totally ﬁnite) Radon measures for the two topologies are the
same. 432H provides one of the important cases in which this occurs. The hypothesis ‘(X, T) is Kanalytic’
generalizes the alternative ‘(X, T) is compact’; in the latter case, S = T, so that the result is, from our
point of view here, trivial. (But from the point of view of elementary general topology, of course, it is one
of the pivots of the theory of compact Hausdorﬀ spaces.) In a similar vein we have a variety of important
topological consequences of the same hypotheses (422Ye, 423Fb).
The paragraphs 432I432J may appear to be no more that a minor extension of ideas already set out. I
ought therefore to say plainly that the topological and measure theory of Kanalytic spaces have coevolved
with the notion of capacity, and that 432J (‘Kanalytic spaces are capacitable’) is one of the cornerstones
of a theory of which I am giving only a minuscule part. For a idea of the vitality and scope of this theory,
see Dellacherie 80.
433 Analytic spaces
We come now to the special properties of measures on ‘analytic’ spaces, that is, continuous images of N
N
,
as described in ¸423. I start with a couple of facts about spaces with countable networks.
433A Proposition Let (X, T) be a topological space with a countable network, and µ a localizable
topological measure on X which is inner regular with respect to the Borel sets. Then µ has countable
Maharam type.
proof Let ˜ µ be the c.l.d. version of µ (213E). Then the measure algebra A of ˜ µ can be identiﬁed with the
measure algebra of µ (322D(biii)). Also ˜ µ is complete, locally determined and localizable, so every subset
of X has a measurable envelope with respect to ˜ µ (213L). Let
˜
Σ be the domain of ˜ µ, and c a countable
network for T. For each E ∈ c, let F
E
∈
˜
Σ be a measurable envelope of E.
Let B be the orderclosed subalgebra of A generated by ¦F
•
E
: E ∈ c¦, and set T = ¦F : F ∈
˜
Σ, F
•
∈ B¦.
Because B is an orderclosed subalgebra of A, T is a σsubalgebra of
˜
Σ. Now T ⊆ T. PPP If G ∈ T, set
c
0
= ¦E : E ∈ c, E ⊆ G¦. Set F =
E∈L
0
F
E
, so that F ∈ T and G ⊆ F. For each E ∈ c
0
, F
E
¸ G is
negligible, so F ¸ G is negligible, and G
•
= F
•
∈ B, so G ∈ T. QQQ
It follows that T includes the Borel σalgebra of X. Because µ is inner regular with respect to the Borel
sets, B is orderdense in A, and B = A. Thus the countable set ¦F
•
E
: E ∈ c¦ τgenerates A, and the
Maharam type of A, which is the Maharam type of µ, is countable.
433B Lemma If (X, T) is a Hausdorﬀ space with a countable network, then any topological measure on
X is countably separated in the sense of 343D.
proof By 4A2Nf, there is a countable family of open sets separating the points of X.
433G Analytic spaces 181
433C Theorem Let X be an analytic Hausdorﬀ space, and µ a Borel measure on X.
(a) If µ is semiﬁnite, it is tight.
(b) If µ is locally ﬁnite, its completion is a Radon measure on X.
proof X is Kanalytic (423C); moreover, every open subset of X is again analytic (423Eb). So 432C gives
the result at once.
Remark Compare 256C.
433D Theorem Let X and Y be analytic Hausdorﬀ spaces, ν a totally ﬁnite Radon measure on Y and
f : X → Y a Borel measurable function such that f[X] is of full outer measure in Y . Then there is a Radon
measure µ on X such that ν = µf
−1
.
proof By 423Ga, the graph R of f is an analytic set in X Y , therefore Kanalytic. Set π
1
(x, y) = x,
π
2
(x, y) = y for (x, y) ∈ R, so that π
1
and π
2
are continuous. Now π
2
[R] = f[X] is of full outer measure, so
by 432G there is a Radon measure λ on R such that ν = λπ
−1
2
. Next, because π
1
is continuous, the image
µ = λπ
−1
1
is a Radon measure on X, by 418I. But π
2
= fπ
1
, so
µf
−1
= (λπ
−1
1
)f
−1
= λ(fπ
1
)
−1
= λπ
−1
2
= ν,
as required.
433E Proposition Let (X, Σ, µ) be a semiﬁnite measure space and T a topology on X such that µ is
inner regular with respect to the closed sets. Let (Y, S) be an analytic space and f : X → Y a measurable
function. Then f is almost continuous.
proof Take E ∈ Σ and γ < µE. Then there is an F ∈ Σ such that F ⊆ E and γ < µF < ∞. For Borel
sets H ⊆ Y , set νH = µ(F ∩ f
−1
[H]). Then ν is a totally ﬁnite Borel measure on Y , so is tight (that is,
inner regular with respect to the compact sets) (433C); let K ⊆ Y be a compact set such that νK > γ, so
that µ(F ∩f
−1
[K]) > γ. The subspace measure on L = F ∩f
−1
[K] is still inner regular with respect to the
(relatively) closed sets (412Pc), and fL is still measurable; but fL is a function from L to K, and K is
metrizable, by 423Dc. So fL is almost continuous, by 418J, and there is a set F
t
⊆ L, of measure at least
γ, such that fF
t
is continuous.
As E and γ are arbitrary, f is almost continuous.
Remark Compare 418Yg.
433F I give two simple corollaries of the von NeumannJankow selection theorem (423N423O).
Proposition Let (X, T) and (Y, S) be analytic Hausdorﬀ spaces, and f : X → Y a Borel measurable
surjection. Let ν be a complete locally determined topological measure on Y , and T its domain. Then there
is a Tmeasurable function g : Y → X such that gf is the identity on X.
proof By 423O we know that there is a function g : Y → X such that gf is the identity and g is T
1

measurable, where T
1
is the σalgebra generated by the SouslinF subsets of Y . But T contains every
SouslinF subset of Y , by 431B, therefore includes T
1
, and g is actually Tmeasurable.
433G Proposition Let (X, T) be an analytic Hausdorﬀ space, (Y, T, ν) a complete locally determined
measure space, and f : X → Y a surjection. Suppose that there is some countable family T ⊆ T such
that T separates the points of Y (that is, whenever y, y
t
are distinct points of Y there is a member of T
containing one and not the other) and f
−1
[F] is a Borel subset of X for every F ∈ T. Then there is a
Tmeasurable function g : Y → X such that fg is the identity on Y .
proof Set / = T ∪ ¦Y ¸ F : F ∈ T¦. The topology T
1
on X generated by T ∪ ¦f
−1
[A] : A ∈ /¦ is
still analytic (423H). If we take S to be the topology on Y generated by /, then S is Hausdorﬀ and f is
(T
1
, S)continuous, so S is analytic (423Bb).
Because S is generated by a countable subset / of T, it is secondcountable, and S ⊆ T (4A3Da/4A3E).
So ν is a topological measure with respect to S. By 433F, there is a function g : Y → X, measurable for
T and the topology T
1
, such that gf is the identity on X; and of course g is still measurable for T and the
coarser original topology T on X.
182 Topologies and measures II 433H
433H Because analytic spaces have countable networks (423C), and their compact subsets are therefore
metrizable (4A2Nh), their measure theory is very close to that of R or [0, 1] or ¦0, 1¦
N
. I give some simple
manifestations of this principle.
Proposition Let ¸X
i
)
i∈I
be a family of analytic spaces, and for each i ∈ I let µ
i
be a Radon probability
measure on X
i
. Let λ be the ordinary product measure on X =
i∈I
X
i
, as deﬁned in ¸254.
(a) If I is countable then λ is a Radon measure.
(b) If every µ
i
is strictly positive, then λ is a quasiRadon measure.
proof (a) In this case, X is analytic (423Bc), therefore hereditarily Lindel¨of (423Da). Let Λ be the domain
of λ and T the topology of X. Then Λ∩T is a base for T; by 4A3Da, T ⊆ Λ. By 417Sb, λ is the τadditive
product measure on X; by 417Q, this is a Radon measure.
(b) By (a), the ordinary product measure on
i∈J
X
i
is a Radon measure for every ﬁnite set J ⊆ I. So
417Sc tells us that λ is the τadditive product measure on X; by 417O, this is a quasiRadon measure.
433I Proposition Let X be an analytic Hausdorﬀ space, and T a countably generated σsubalgebra of
the Borel σalgebra B(X) of X. Then any locally ﬁnite measure with domain T has an extension to a Radon
measure on X.
proof Let µ
0
be a locally ﬁnite measure with domain T.
(a) Consider ﬁrst the case in which µ
0
is totally ﬁnite. Let ¸F
n
)
n∈N
be a sequence in T generating T as
σalgebra. Deﬁne f : X → ¦0, 1¦
N
by setting
f(x)(n) = χF
n
(x) for n ∈ N, x ∈ X.
Then f is Tmeasurable (use 418Bd), so we have a Borel measure ν
0
on ¦0, 1¦
N
deﬁned by setting ν
0
E =
µ
0
f
−1
[E] for every Borel set E ⊆ ¦0, 1¦
N
. Now the completion ν of ν
0
is a Radon measure (433C). Also
f[X] must be analytic, by 423Gb, because f is B(X)measurable. So ν measures f[X] (432A), and
νf[X] = ν
∗
0
f[X] = ν
0
¦0, 1¦
N
,
that is, f[X] is νconegligible. By 433D, there is a Radon measure µ on X such that ν = µf
−1
.
Because every F
n
is expressible as f
−1
[E] for some E ∈ B(¦0, 1¦
N
), so is every member of T. If F ∈ T,
take H ∈ B(¦0, 1¦
N
) such that F = f
−1
[H]; then
µF = νH = ν
0
H = µ
0
F.
Thus µ extends µ
0
and µΣ will serve.
(b) In general, because X is Lindel¨of and µ
0
is locally ﬁnite, µ
0
is σﬁnite. Let ¸X
n
)
n∈N
be a partition
of X into members of T such that µ
0
X
n
is ﬁnite for every n, and set µ
(n)
0
F = µ
0
(F ∩ X
n
) for every n and
every F ∈ T; then every µ
(n)
0
has an extension to a Radon measure µ
(n)
, and we can set µF =
∞
n=0
µ
(n)
F
for every F ∈ Σ. Because µ
0
is locally ﬁnite, so is µ, and it is now easy to check that µ is complete, locally
determined and tight, just because every µ
(n)
is.
433J I turn now to a brief mention of ‘standard Borel spaces’. From the point of view of this chapter,
it is natural to regard the following results as simple corollaries of theorems about Polish spaces. But, as
remarked in ¸424, there are cases in which a standard Borel space is presented without any speciﬁc topology
being attached; and in any case it is interesting to look at the ways in which we can express these ideas as
theorems about σalgebras rather than about topological spaces.
Proposition Let (X, Σ) be a standard Borel space and T a countably generated σsubalgebra of Σ. Then
any σﬁnite measure with domain T has an extension to Σ.
proof Let µ
0
be a σﬁnite measure with domain T.
(a) If µ
0
is totally ﬁnite, give X a Polish topology for which Σ is the Borel σalgebra of X, and use 433I.
(b) In general, let ¸X
n
)
n∈N
be a partition of X into members of T such that µ
0
X
n
< ∞ for every n, and
set µ
(n)
0
F = µ
0
(F ∩ X
n
) for every n and every F ∈ T; then every µ
(n)
0
has an extension to a measure µ
(n)
with domain Σ, and we can set µF =
∞
n=0
µ
(n)
F for every F ∈ Σ.
433Yb Analytic spaces 183
433K Proposition Let ¸(X
n
, Σ
n
, µ
n
))
n∈N
be a sequence of probability spaces such that (X
n
, Σ
n
) is a
standard Borel space for every n. Suppose that for each n ∈ N we are given an inversemeasurepreserving
function f
n
: X
n+1
→ X
n
. Then we can ﬁnd a standard Borel space (X, Σ), a probability measure µ with
domain Σ, and inversemeasurepreserving functions g
n
: X → X
n
such that f
n
g
n+1
= g
n
for every n.
proof For each n, choose a Polish topology T
n
on X
n
such that Σ
n
is the algebra of T
n
Borel sets. Let
ˆ µ
n
be the completion of µ
n
; then ˆ µ
n
is a Radon measure (433C). Every f
n
is inversemeasurepreserving for
ˆ µ
n+1
and ˆ µ
n
, by 235Hc, and almost continuous, by 418J.
By 418Q, we have a Radon measure ˆ µ on
X = ¦x : x ∈
n∈N
X
n
, f
n
(x(n + 1)) = x(n) for every n ∈ N¦
such that the continuous maps x → x(n) = g
n
(x) : X → X
n
are inversemeasurepreserving for every n.
Now X is a Borel subset of Z =
n∈N
X
n
. PPP For each n ∈ N, let 
n
be a countable base for T
n
. Then
Z ¸ X =
n∈N
U,V ∈/
n
,U∩V =∅
¦z : z(n) ∈ U, f
n
(z(n + 1)) ∈ V ¦
is a countable union of Borel sets because ¦z : z(n) ∈ U¦ is open and ¦z : z(n + 1) ∈ f
−1
n
[V ]¦ is Borel
whenever n ∈ N and U, V ∈ 
n
. So Z ¸ X and X are Borel sets. QQQ
Accordingly (X, Σ) is a standard Borel space, where Σ is the Borel σalgebra of X (424G). So if we take
µ = ˆ µΣ, we shall have a suitable measure on X.
433X Basic exercises (a) Let (X, T) be a topological space with a countable network, and µ a topo
logical measure on X which is inner regular with respect to the Borel sets and has the measurable envelope
property (213Xl). Show that µ has countable Maharam type.
(b) Show that an eﬀectively locally ﬁnite measure on a hereditarily Lindel¨of space (in particular, on any
analytic space) is σﬁnite.
(c) Let X ⊆ [0, 1] be a set of Lebesgue outer measure 1 and inner measure 0. Show that the subspace
measure on X is a totally ﬁnite Borel measure which is not tight.
(d) Let X be a Hausdorﬀ space and µ a locally ﬁnite measure on X, inner regular with respect to the
Borel sets, such that domµ includes a base for the topology of X. Suppose that Y ⊆ X is an analytic set
of full outer measure. Show that µ has a unique extension to a Radon measure ˜ µ on X, and that Y is
˜ µconegligible.
(e) Let (X, Σ) be a standard Borel space. (i) Show that any semiﬁnite measure with domain Σ is a
compact measure (deﬁnition: 342Ac), therefore perfect. (Hint: if X is given a suitable topology, the measure
is tight.) (ii) Show that any measure with domain Σ is countably separated.
(f ) Let (X, Σ, µ) and (Y, T, ν) be atomless probability spaces such that (X, Σ) and (Y, T) are standard
Borel spaces. Show that (X, Σ, µ) and (Y, T, ν) are isomorphic. (Hint: by 344I, their completions are
isomorphic; by 344H, they have negligible sets of cardinal c; show that any isomorphism between the
completions is (Σ, T)measurable on a conegligible set; use 424Da to match residual negligible sets.)
(g) Let X be [0, 1] ¦0, 1¦, with its usual topology, and I

the split interval (419L); deﬁne f : X → I

by
setting f(t, 0) = t
−
, f(t, 1) = t
+
for t ∈ [0, 1]. (i) Give I

its usual Radon measure ν (343J, 419Lc). Show
that there is no Radon measure λ on X such that ν = λf
−1
. (ii) Let µ be the product Radon probability
measure on X, starting from Lebesgue measure on [0, 1] and the usual faircoin measure on ¦0, 1¦. Show
that f is inversemeasurepreserving for µ and ν. Show that f is not almost continuous.
433Y Further exercises (a) Find a Hausdorﬀ topological space X with a countable network and a
semiﬁnite Borel measure on X which does not have countable Maharam type.
(b) Let X be an analytic Hausdorﬀ space and µ an atomless Radon measure on X. Show that (X, µ) is
isomorphic to Lebesgue measure on some measurable subset of R. (Hint: 344I.)
184 Topologies and measures II 433Yc
(c) Let (X, T) be a Polish space without isolated points, and µ a σﬁnite topological measure on X. Show
that there is a conegligible meager set. (Hint: Show that every nonempty open set is uncountable. Find a
countable dense negligible set and a negligible G
δ
set including it.)
(d) Let ¸X
n
)
n∈N
be a sequence of analytic spaces and for each n ∈ N let µ
n
be a Borel probability measure
on X
n
. Suppose that for each n ∈ N we are given an inversemeasurepreserving function f
n
: X
n+1
→ X
n
.
Show that we can ﬁnd a standard Borel space (X, Σ), a probability measure µ with domain Σ, and inverse
measurepreserving functions g
n
: X → X
n
such that f
n
g
n+1
= g
n
for every n.
433 Notes and comments The measuretheoretic results of 433C433E are of much the same type as
those in ¸432. A characteristic diﬀerence is that Borel measurable functions between analytic spaces behave
in many ways like continuous functions. (Compare 433D and 432G.) You may feel that 423Yc oﬀers some
explanation of this. For any question which refers to the Borel algebra of an analytic space X, or to the class
of its analytic subsets, we can expect to be able to suppose that X is separable and metrizable (see 423Xd),
and that any single Borel measurable function appearing is continuous. (424H is a particularly remarkable
instance of this principle.)
433H here amounts to spelling out a special case of ideas already treated in 417S. As this territory
is relatively unfamiliar, I give detailed examples (423Xi, 433Xc, 433Xg, 439A, 439K) to show that the
theorems of this section are not generally valid for compact Hausdorﬀ spaces (the archetype of Kanalytic
spaces which need not be analytic), nor for separable metric spaces (the archetypical spaces with countable
network). They really do depend on the particular combination of properties possessed by analytic spaces.
For large parts of probability theory, standard Borel spaces provide an adequate framework, and have a
number of advantages; some of the technical problems concerning measurability which loom rather large in
this treatise disappear in such contexts. Many authors accordingly give them great prominence. I myself
believe that the simpliﬁcations are an entrapment rather than a liberation, that sooner or later everyone
has to leave the comfortable environment of Borel algebras on Polish spaces, and that it is better to be
properly equipped with a suitable general theory when one does. But it is surely important to know what
the simpliﬁcations are, and the results in 433J433K will I hope show at least that there are wonderful ideas
here, even if my own presentation tends to leave them on one side.
434 Borel measures
What one might call the fundamental question of topological measure theory is the following.
What kinds of measures can arise on what kinds of topological space?
Of course this question has inexhaustible ramiﬁcations, corresponding to all imaginable properties of mea
sures and topologies and connexions between them. The challenge I face here is that of identifying particular
ideas as being more important than others, and the chief diﬃculty lies in the bewildering variety of topo
logical properties which have been studied, any of which may have implications for the measure theory of
the spaces involved. In this section and the next I give a sample of what is known, necessarily biased and
incomplete. I try however to include the results which are most often applied and enough others for the
proofs to contain, between them, most of the nontrivial arguments which have been found eﬀective in this
area.
In 434A I set out a crude classiﬁcation of Borel measures on topological spaces. For compact Hausdorﬀ
spaces, at least, the ﬁrst question is whether they carry Borel measures which are not, in eﬀect, Radon
measures; this leads us to the deﬁnition of ‘Radon’ space (434C) which is also of interest in the context of
general Hausdorﬀ spaces. I give a brief account of the properties of Radon spaces (434F, 434N). I look also
at two special topics: ‘quasidyadic’ spaces (434O434Q) and a construction of Borel product measures by
integration of sections (434R).
In the study of Radon spaces we ﬁnd ourselves looking at ‘universally measurable’ subsets of topological
spaces (434D). These are interesting in themselves, and also interact with constructions from earlier parts
of this treatise (434S434T). Three further classes of topological space, deﬁned in terms of the types of
topological measure which they carry, are the ‘Borelmeasurecompact’, ‘Borelmeasurecomplete’ and ‘pre
Radon’ spaces; I discuss them brieﬂy in 434G434J. They provide useful methods for deciding whether
Hausdorﬀ spaces are Radon (434K).
434Da Borel measures 185
434A Types of Borel measures In ¸411 I introduced the following properties which a Borel measure
may or may not have:
(i) inner regularity with respect to closed sets;
(ii) inner regularity with respect to zero sets;
(iii) tightness (that is, inner regularity with respect to closed compact sets);
(iv) τadditivity.
These are of course interrelated. (ii)⇒(i) just because zero sets are closed, and (iii)⇒(iv) by 411E; in
a Hausdorﬀ space, (iii)⇒(i); and for an eﬀectively locally ﬁnite measure on a regular topological space,
(iv)⇒(i) (414Mb).
On a regular Hausdorﬀ space, therefore, we can divide totally ﬁnite Borel measures into four classes:
(A) measures which are not inner regular with respect to the closed sets;
(B) measures which are inner regular with respect to the closed sets, but not τadditive nor
tight;
(C) measures which are τadditive and inner regular with respect to the closed sets, but not
inner regular with respect to the compact sets;
(D) tight measures;
and each of the classes (B)(D) can be further subdivided into those which are completion regular (B
1
, C
1
,
D
1
) and those which are not (B
0
, C
0
, D
0
). Examples may be found in 434Xf (type A), 411Q and 439K
(type B
0
), 439J (type B
1
), 415Xc and 434Xa (type C
1
) and 434Xb (type D
0
), while Lebesgue measure itself
is of type D
1
, and any direct sum of spaces of types D
0
and C
1
will have type C
0
. (The space in 439J
depends for its construction on supposing that there is a cardinal which is not measurefree. It seems that
no convincing example of a space of class B
1
, that is, a completion regular, nonτadditive Borel probability
measure on a completely regular Hausdorﬀ space, is known which does not depend on some special axiom
beyond ordinary ZFC. For one of the obstacles to ﬁnding such a space, see 434Q below.)
Note that a totally ﬁnite Borel measure µ on a regular Hausdorﬀ space can be extended to a quasiRadon
measure iﬀ µ is of class C or D (415M), and that in this case the quasiRadon measure must be just the
completion ˆ µ of µ. ˆ µ will be of the same type, on the classiﬁcation here, as µ; in particular, ˆ µ will be a
Radon measure iﬀ µ is of class D (416F).
434B Compact, analytic and Kanalytic spaces For any class of topological spaces, we can enquire
which of the seven types of measure described above can be realized by measures on spaces of that class.
The enquiry is limited only by our enterprise and diligence in seeking out new classes of topological space.
For the spaces studied in ¸¸432433, however, we have something worth repeating here. On a Kanalytic
Hausdorﬀ space, a semiﬁnite Borel measure which is inner regular with respect to the closed sets is tight
(432B, 432D); consequently classes B and C of 434A cannot appear, and we are left with only the types A,
D
0
and D
1
, all of which appear on compact Hausdorﬀ spaces (434Xb, 434Xf). On an analytic Hausdorﬀ
space we have further simpliﬁcations: every semiﬁnite Borel measure is tight (433C), and (if X is regular)
every closed set is a zero set (423Db). Thus on an analytic regular Hausdorﬀ space only type D
1
, of the
seven types in 434A, can appear. (If the topology is not regular, we may also get measures of type D
0
; see
434Ya.)
434C Radon spaces: Deﬁnition For Kanalytic Hausdorﬀ spaces, therefore, we have a large gap
between the ‘bad’ measures of class A and the ‘good’ measures of class D; furthermore, we have an important
class of spaces in which type A cannot appear. It is natural to enquire further into the spaces in which every
(totally ﬁnite) Borel measure is of class D, and (given that no exact description can be found) we are led,
as usual, to a deﬁnition. A Hausdorﬀ space X is Radon if every totally ﬁnite Borel measure on X is tight.
434D Universally measurable sets Before going farther with the study of Radon spaces it will be
useful to spend a couple of paragraphs on the following concept. Let X be a topological space.
(a) I will say that a subset E of X is universally measurable (in X) if it is measured by the completion
of every totally ﬁnite Borel measure on X; that is, for every totally ﬁnite Borel measure µ on X there is a
Borel set F ⊆ X such that E´F is µnegligible.
186 Topologies and measures II 434Db
(b) A subset of X is universally measurable iﬀ it is measured by every complete locally determined
topological measure on X. PPP (i) Suppose that A ⊆ X is universally measurable and that µ is a complete
locally determined topological measure on X. Let F ⊆ X be such that µF is deﬁned and ﬁnite. Then we
have a totally ﬁnite Borel measure ν on X deﬁned by setting νE = µ(F ∩ E) for every Borel set E ⊆ X.
Now there are Borel sets E, B ⊆ X such that A´E ⊆ B and νB = 0. In this case, (A∩F)´(E∩F) ⊆ B∩F
and µ(B ∩ F) = 0, so that (because µ is complete) A ∩ F is measured by µ. Because F is arbitrary and µ
is locally determined, A is measured by µ. (ii) Suppose that A ⊆ X is measured by every complete locally
determined topological measure on X. Then, in particular, it is measured by the completion of any totally
ﬁnite Borel measure, so is universally measurable. QQQ
(c) If X is a topological space then the family Σ
um
of universally measurable subsets of X is a σalgebra
closed under Souslin’s operation and including the Borel σalgebra. (For it is the intersection of the domains
of a family of complete totally ﬁnite measures, and all these are σalgebras including the Borel σalgebra
and closed under Souslin’s operation, by 431A.) In particular, SouslinF sets are universally measurable, so
(if X is Hausdorﬀ) Kanalytic and analytic sets are (422Ha, 423C).
(d) Note that a function f : X → R is Σ
um
measurable iﬀ it is µvirtually measurable for every totally
ﬁnite Borel measure µ on X (122Q, 212F). Generally, if Y is another topological space, I will say that
f : X → Y is universally measurable if f
−1
[H] ∈ Σ
um
for every open set H ⊆ Y ; that is, if f is
(Σ
um
, B(Y ))measurable, where B(Y ) is the Borel σalgebra of Y .
(e) In fact, if f : X → Y is universally measurable, then it is (Σ
um
, Σ
(Y )
um
)measurable, where Σ
(Y )
um
is the
algebra of universally measurable subsets of Y . PPP Take F ∈ Σ
(Y )
um
and a totally ﬁnite Borel measure µ on
X. If ˆ µ is the completion of µ, then the image measure ν = ˆ µf
−1
is a complete totally ﬁnite topological
measure on Y , so measures F, and f
−1
[F] ∈ dom ˆ µ. As µ is arbitrary, f
−1
[F] ∈ Σ
um
; as F is arbitrary, f
is (Σ
um
, Σ
(Y )
um
)measurable. QQQ
(f ) It follows that if Z is a third topological space and f : X → Y , g : Y → Z are universally measurable,
then gf : X → Z is universally measurable.
434E Universally Radonmeasurable sets A companion idea is the following.
(a) Let X be a Hausdorﬀ space. I will say that a subset E of X is universally Radonmeasurable if
it is measured by every Radon measure on X.
(b) If X is a Hausdorﬀ space, the family Σ
uRm
of universally Radonmeasurable subsets of X is a σ
algebra closed under Souslin’s operation and including the algebra of universally measurable subsets of X
(and, a fortiori, including the Borel σalgebra). (Use 434Db and the idea of 434Dc.)
(c) If X is a Hausdorﬀ space, I will say that a function f from X to a topological space Y is universally
Radonmeasurable if f
−1
[H] ∈ Σ
uRm
for every open set H ⊆ Y . A function f : X → R is Σ
uRm

measurable iﬀ it is µvirtually measurable for every totally ﬁnite tight Borel measure µ on X. (Compare
434Dd.)
434F Elementary properties of Radon spaces: Proposition Let X be a Hausdorﬀ space.
(a) The following are equiveridical:
(i) X is a Radon space;
(ii) every semiﬁnite Borel measure on X is tight;
(iii) if µ is a locally ﬁnite Borel measure on X, its c.l.d. version ˜ µ is a Radon measure;
(iv) whenever µ is a totally ﬁnite Borel measure on X, and G ⊆ X is an open set with µG > 0, then
there is a compact set K ⊆ G such that µK > 0;
(v) whenever µ is a nonzero totally ﬁnite Borel measure on X, there is a Radon subspace Y of X such
that µ
∗
Y > 0.
(b) If Y ⊆ X is a subspace which is a Radon space in its induced topology, then Y is universally
measurable in X.
434F Borel measures 187
(c) If X is a Radon space and Y ⊆ X, then Y is Radon iﬀ it is universally measurable in X iﬀ it is
universally Radonmeasurable in X. In particular, all Borel subsets and all SouslinF subsets of X are
Radon spaces.
(d) The family of Radon subspaces of X is closed under Souslin’s operation and set diﬀerence.
proof (a)(i)⇒(ii) Let µ be a semiﬁnite Borel measure on X, E ⊆ X a Borel set and γ < µE. Because µ
is semiﬁnite, there is a Borel set H of ﬁnite measure such that µ(E∩H) > γ. Set νF = µ(F ∩H) for every
Borel set F ⊆ X; then ν is a totally ﬁnite Borel measure on X, and νE > γ. Because X is a Radon space,
there is a compact set K ⊆ E such that νK ≥ γ, and now µK ≥ γ. As µ, E and γ are arbitrary, (ii) is true.
(ii)⇒(i) and (i)⇒(v) are trivial.
(v)⇒(iv) Assume (v), and let µ be a totally ﬁnite Radon measure on X and G a nonnegligible open
set. Set νE = µ(E ∩G) for every Borel set E ⊆ X. Then ν is a nonzero totally ﬁnite Borel measure on X,
so there is a Radon subspace Y of X such that ν
∗
Y > 0. The subspace measure ν
Y
on Y is a Borel measure
on Y , so is tight. Since ν
Y
(Y ¸ G) = ν(X ¸ G) = 0, ν
Y
(Y ∩ G) > 0 and there is a compact set K ⊆ Y ∩ G
such that ν
Y
K > 0. Now µK > 0. As µ and G are arbitrary, (iv) is true.
not(i)⇒not(iv) If X is not Radon, there is a totally ﬁnite Borel measure µ on X which is not tight.
By 416F(iii), there is an open set G ⊆ X such that
µG > sup
K⊆G is compact
µK = γ
say. Let / be the family of compact subsets of G. By 215B(v), there is a nondecreasing sequence ¸K
n
)
n∈N
in / such that µ(K ¸ F) = 0 for every K ∈ /, where F =
n∈N
K
n
. Observe that
µF = lim
n→∞
µK
n
≤ γ < µG.
Now set νE = µ(E ∩ G¸ F) for every Borel set E ⊆ X. Then ν is a Borel measure on X, and νG > 0. If
K ⊆ G is compact, then νK = µ(K ¸ F) = 0. So ν and G witness that (iii) is false.
(i)⇒(iii) The point is that ˜ µ is tight. PPP If ˜ µE > γ, then, because ˜ µ is semiﬁnite, there is a set
E
t
⊆ E such that γ < ˜ µE
t
< ∞; now there is a Borel set H ⊆ E
t
such that µH = ˜ µE
t
(213Fc). Setting
νF = µ(H ∩ F) for every Borel set F, ν is a totally ﬁnite Borel measure on X and νH > γ, so there is a
compact set K ⊆ H such that νK ≥ γ. Now γ ≤ µK < ∞, so ˜ µK = µK ≥ γ (213Fa), while K ⊆ E. As E
and γ are arbitrary, ˜ µ is tight. QQQ
On the other hand, every point of X belongs to an open set of ﬁnite measure for µ, which is still of ﬁnite
measure for ˜ µ (213Fa again). So ˜ µ is locally ﬁnite; since it is surely complete and locally determined, it is a
Radon measure.
(iii)⇒(i) Assume (iii), and let µ be a totally ﬁnite Borel measure on X. Then its c.l.d. version ˜ µ is tight.
But ˜ µ extends µ (213Hc), so µ also is tight. As µ is arbitrary, X is a Radon space.
(b) Let µ be a totally ﬁnite Borel measure on X, and ˆ µ its completion; let > 0. Let µ
Y
be the subspace
measure on Y , so that µ
Y
is a totally ﬁnite Borel measure on Y , and is tight. There is a compact set K ⊆ Y
such that νK ≥ µ
Y
Y −. But this means that
µ
∗
Y = µ
Y
Y ≤ µ
Y
K + = µ
∗
(K ∩ Y ) + = µK + ≤ µ
∗
Y +.
As is arbitrary, µ
∗
Y = µ
∗
Y , and Y is measured by ˆ µ (413Ef); as µ is arbitrary, Y is universally measurable.
(c) (i) If Y is Radon, it is universally measurable, by (b). (ii) If Y is universally measurable, it is
universally Radonmeasurable, by 434Eb. (iii) Suppose that Y is universally Radonmeasurable, and that ν
is a totally ﬁnite Borel measure on Y . For Borel sets E ⊆ X, set µE = ν(E ∩ Y ). Then µ is a totally ﬁnite
Borel measure on X, so its c.l.d. version ˜ µ is a Radon measure on X, by (aiii). We are supposing that Y
is universally Radonmeasurable, so, in particular, it must be measured by ˜ µ. We have
˜ µ(X ¸ Y ) = sup
K⊆X\Y is compact
˜ µK = sup
K⊆X\Y is compact
µK
(213Ha, because µ is totally ﬁnite)
188 Topologies and measures II 434F
= sup
K⊆X\Y is compact
ν(K ∩ Y ) = 0,
and Y is ˜ µconegligible.
Now suppose that E ⊆ Y is a (relatively) Borel subset of Y . Then E is of the form F ∩ Y where F is a
Borel subset of X, so that
νE = µF = ˜ µF = ˜ µ(Y ∩ F) = ˜ µE
= sup
K⊆E is compact
µK = sup
K⊆E is compact
νK.
As E is arbitrary, ν is tight; as ν is arbitrary, Y is a Radon space.
By 434Dc, it follows that all Borel subsets and all SouslinF subsets of X are Radon spaces.
(d) The ﬁrst step is to note that if ¸E
n
)
n∈N
is a sequence of Radon subspaces of X with union E, then
E is Radon; this is immediate from (av) above.
Now let ¸E
σ
)
σ∈S
be a Souslin scheme, consisting of Radon subsets of X, with kernel A. We know that
E =
σ∈S
E
σ
is a Radon space. Every E
σ
is universally measurable in E, by (b), so A also is (434Dc), and
must be Radon, by (c). Thus the family of Radon subspaces of X is closed under Souslin’s operation.
If E and F are Radon subsets of X, then E∪F is Radon, and, just as above, F is universally measurable
in E ∪ F. But this means that E ¸ F = (E ∪ F) ¸ F is universally measurable in E ∪ F, so that E ¸ F is
Radon.
434G Just as we can address the question ‘when can we be sure that every Borel measure is of class
D’ in terms of the deﬁnition of ‘Radon’ space (434C), we can form other classes of topological space by
declaring that the Borel measures they support must be of certain kinds. Three deﬁnitions which lead to
interesting patterns of ideas are the following.
Deﬁnitions (a) A topological space X is Borelmeasurecompact (Gardner & Pfeffer 84) if every
totally ﬁnite Borel measure on X which is inner regular with respect to the closed sets is τadditive, that
is, X carries no measure of class B in the classiﬁcation of 434A.
(b) A topological space X is Borelmeasurecomplete (Gardner & Pfeffer 84) if every totally
ﬁnite Borel measure on X is τadditive. (If X is regular and Hausdorﬀ, this amounts to saying that X
carries no measures of classes A or B in the classiﬁcation of 434A.)
(c) A Hausdorﬀ space X is preRadon (also called ‘hyporadonian’, ‘semiradonian’) if every τ
additive totally ﬁnite Borel measure on X is tight. (If X is regular, this amounts to saying that X carries
no measure of class C in the classiﬁcation of 434A.)
434H Proposition Let X be a topological space and B its Borel σalgebra.
(a) The following are equiveridical:
(i) X is Borelmeasurecompact;
(ii) every semiﬁnite Borel measure on X which is inner regular with respect to the closed sets is
τadditive;
(iii) every eﬀectively locally ﬁnite Borel measure on X which is inner regular with respect to the closed
sets has an extension to a quasiRadon measure;
(iv) every totally ﬁnite Borel measure on X which is inner regular with respect to the closed sets has
a support;
(v) if µ is a nonzero totally ﬁnite Borel measure on X, inner regular with respect to the closed sets,
and ( is an open cover of X, then there is some G ∈ ( such that µG > 0.
(b) If X is Lindel¨of (in particular, if X is a Kanalytic Hausdorﬀ space), it is Borelmeasurecompact.
(c) If X is Borelmeasurecompact and A ⊆ X is a SouslinF set, then A is Borelmeasurecompact in its
subspace topology. In particular, any Baire subset of X is Borelmeasurecompact.
proof (a)(i)⇒(ii) Assume (i), and let µ be a semiﬁnite Borel measure on X which is inner regular with
respect to the closed sets. Let ( be an upwardsdirected family of open sets with union G
0
, and γ < µG
0
.
434I Borel measures 189
Because µ is semiﬁnite, there is an H ∈ B such that µH < ∞ and µ(H ∩ G
0
) ≥ γ. Set νE = µ(E ∩ H) for
every E ∈ B; then ν is a totally ﬁnite Borel measure on X. For any E ∈ B,
νE = µ(E ∩ H) = sup¦µF : F ⊆ E ∩ H is closed¦ ≤ sup¦νF : F ⊆ E is closed¦,
so ν is inner regular with respect to the closed sets, and must be τadditive. Now
γ ≤ νG
0
= sup
G∈Ç
νG ≤ sup
G∈Ç
µG.
As γ and ( is arbitrary, µ is τadditive.
(ii)⇒(iii) Assume (ii), and let µ be an eﬀectively locally ﬁnite Borel measure on X which is inner regular
with respect to the closed sets. Then it is semiﬁnite (411Gd), therefore τadditive. By 415L, it has an
extension to a quasiRadon measure on X.
(iii)⇒(i) If (i) is true and µ is a totally ﬁnite Borel measure on X which is inner regular with respect
to the closed sets, then µ has an extension to a quasiRadon measure, which is τadditive, so µ is also
τadditive (411C).
(i)⇒(iv) Use 411Nd.
(iv)⇒(v) Suppose that (iv) is true, that µ is a nonzero totally ﬁnite Borel measure on X which is
inner regular with respect to the closed sets, and that ( is an open cover of X. If F is the support of µ,
then µF > 0 so F ,= ∅; there must be some G ∈ ( meeting F, and now µG > 0.
not(i)⇒not(v) Suppose that there is a totally ﬁnite Borel measure µ on X, inner regular with
respect to the closed sets, which is not τadditive. Let ( be an upwardsdirected family of open sets such
that µG
∗
> γ, where G
∗
=
( and γ = sup
G∈Ç
µG. Let ¸G
n
)
n∈N
be a nondecreasing sequence in ( such
that µ(G ¸ G
∗
0
) for every G ∈ (, where G
∗
0
=
n∈N
G
n
(215B(v)). Then µG
∗
0
≤ γ, so there is a closed set
F ⊆ G
∗
¸ G
∗
0
such that µF > 0.
Let ν be the Borel measure on X deﬁned by setting µE = µ(E∩F) for every E ∈ B. As in the argument
for (i)⇒(ii), ν is inner regular with respect to the closed sets. Consider H = ( ∪ ¦X ¸ F¦; this is an open
cover of X. If G ∈ ( then νG ≤ µ(G¸ G
∗
0
) = 0, so νH = 0 for every H ∈ H; thus ν and H witness that (iv)
is false.
(b) Use (av) and 422Gg.
(c) Let µ be a Borel measure on A which is inner regular with respect to the closed sets, that is to
say, the relatively closed sets in A. Let ν be the corresponding Borel measure on X, deﬁned by setting
νE = µ(A ∩ E) for every E ∈ B. Let ˆ ν be the completion of ν. Putting 431D and 421M together, we see
that ˆ νA = sup¦ˆ νF : F ⊆ A is closed in X¦, that is, νX = sup¦µF : F ⊆ A is closed in X¦. But this means
that if E ∈ B and γ < νE, there is a closed set F in X such that F ⊆ A such that µ(E ∩F) > γ; now there
is a relatively closed set F
t
⊆ A such that F
t
⊆ E ∩ F and µF
t
≥ γ, and as F
t
must be relatively closed in
F it is closed in X, while νF
t
≥ γ. Since E and γ are arbitrary, ν is inner regular with respect to the closed
sets, and will be τadditive.
Now suppose that ( is an upwardsdirected family of relatively open subsets of A. Set H = ¦H : H ⊆ X
is open, H ∩ A ∈ (¦. Then H is upwardsdirected, so
µ(
() = ν(
H) = sup
H∈1
νH = sup
G∈Ç
µG.
As µ and ( are arbitrary, A is Borelmeasurecompact.
By 421L, it follows that any Baire subset of X is Borelmeasurecompact.
434I Proposition Let X be a topological space.
(a) The following are equiveridical:
(i) X is Borelmeasurecomplete;
(ii) every semiﬁnite Borel measure on X is τadditive;
(iii) every totally ﬁnite Borel measure on X has a support;
(iv) whenever µ is a totally ﬁnite Borel measure on X there is a base  for the topology of X such
that µ(
¦U : U ∈ , µU = 0¦) = 0.
190 Topologies and measures II 434I
(b) If X is regular, it is Borelmeasurecomplete iﬀ every eﬀectively locally ﬁnite Borel measure on X has
an extension to a quasiRadon measure.
(c) If X is Borelmeasurecomplete, it is Borelmeasurecompact.
(d) If X is Borelmeasurecomplete, so is every subspace of X.
(e) If X is hereditarily Lindel¨of (for instance, if X is separable and metrizable, see 4A2P(aiii)), it is
Borelmeasurecomplete, therefore Borelmeasurecompact.
proof (a)(i)⇒(ii) Use the argument of (i)⇒(ii) of 434Ha; this case is simpler, because we do not need to
check that the auxiliary measure ν is inner regular.
(ii)⇒(i) is trivial.
(i)⇒(iv) If X is Borelmeasurecomplete and µ is a totally ﬁnite Borel measure on X, take  to be the
family of all open subsets of X. This is surely a base for the topology, and setting 
0
= ¦U : U ∈ , µU = 0¦,

0
is upwardsdirected so µ(

0
) = sup
U∈/
0
µU = 0, as required.
(iv)⇒(iii) Assume (iv), and let µ be a totally ﬁnite Borel measure on X. Take a base  as in (iv),
so that µ(

0
) = 0, where 
0
is the family of negligible members of . Set F = X ¸

0
, so that F is a
conegligible closed set. If G ⊆ X is an open set meeting F, there is a member U of  such that U ⊆ G and
U ∩ F ,= ∅; now U / ∈  so
µ(G∩ F) = µG ≥ µU > 0.
As G is arbitrary, F is selfsupporting and is the support of µ.
(iii)⇒(i) Assume (iii), and let µ be a totally ﬁnite Borel measure on X. Let ( be an upwardsdirected
family of open sets with union G
∗
. Set γ = sup
G∈Ç
µG. Let ¸G
n
)
n∈N
be a nondecreasing sequence in (
such that µ(G¸ G
∗
0
) for every G ∈ (, where G
∗
0
=
n∈N
G
n
(215B(v)). Then µG
∗
0
≤ γ. Let ν be the Borel
measure on X deﬁned by setting µE = µ(E ∩ G
∗
¸ G
∗
0
) for every E ∈ B. Then ν has a support F say. Now
νG = 0 for every G ∈ (, so F ∩ G = ∅ for every G ∈ (, and F ∩ G
∗
= ∅; but this means that
µ(G
∗
¸ G
∗
0
) = νX = νF = 0.
Accordingly µG
∗
= γ. As µ and ( are arbitrary, X is Borelmeasurecomplete.
(b) If X is Borelmeasurecomplete and µ is an eﬀectively locally ﬁnite Borel measure on X, then µ is
τadditive, by (aii), so extends to a quasiRadon measure on X, by 415Cb. If eﬀectively locally ﬁnite Borel
measures on X extend to quasiRadon measures, then any totally ﬁnite Borel measure is τadditive, by
411C, and X is Borelmeasurecomplete.
(c) Immediate from the deﬁnitions.
(d) If Y ⊆ X and µ is a totally ﬁnite Borel measure on Y , let ν be the Borel measure on X deﬁned by
setting νE = µ(E ∩ Y ) for every Borel set E ⊆ X. It is easy to check that ν is τadditive. So if ( is an
upwardsdirected family of relatively open subsets of Y , and we set H = ¦H : H ⊆ X is open, H ∩ Y ∈ (¦,
we shall get
µ(
() = ν(
H) = sup
H∈1
νH = sup
G∈Ç
µG.
As µ and ( are arbitrary, Y is Borelmeasurecomplete.
(e) If µ is a totally ﬁnite Borel measure on X and ( is a nonempty upwardsdirected family of open
subsets of X with union G
∗
, then there is a sequence ¸G
n
)
n∈N
in ( with union G
∗
, by 4A2H(ci). Because
( is upwardsdirected, there is a nondecreasing sequence ¸G
t
n
)
n∈N
in ( such that G
t
n
⊇ G
n
for every n ∈ N,
so that
µG
∗
= lim
n→∞
µG
t
n
≤ sup
G∈Ç
µG.
As µ and ( are arbitrary, X is Borelmeasurecomplete.
434J Borel measures 191
434J Proposition Let X be a Hausdorﬀ space.
(a) The following are equiveridical:
(i) X is preRadon;
(ii) every eﬀectively locally ﬁnite τadditive Borel measure on X is tight;
(iii) whenever µ is a nonzero totally ﬁnite τadditive Borel measure on X, there is a compact set
K ⊆ X such that µK > 0;
(iv) whenever µ is a totally ﬁnite τadditive Borel measure on X, µX = sup
K⊆X is compact
µK;
(v) whenever µ is a locally ﬁnite eﬀectively locally ﬁnite τadditive Borel measure on X, the c.l.d.
version of µ is a Radon measure on X.
(b) If X is preRadon, then every locally ﬁnite quasiRadon measure on X is a Radon measure.
(c) If X is regular and every totally ﬁnite quasiRadon measure on X is a Radon measure, then X is
preRadon.
(d) If X is preRadon, then any universally Radonmeasurable subspace (in particular, any Borel subset
or SouslinF subset) A of X is preRadon.
(e) If A ⊆ X is preRadon in its subspace topology, it is universally Radonmeasurable in X.
(f) If X is Kanalytic (for instance, if it is compact), it is preRadon.
(g) If X is completely regular and
ˇ
Cechcomplete (for instance, if it is locally compact (4A2Gj), or
metrizable and complete under a metric inducing its topology (4A2Md)), it is preRadon.
(h) If X =
i∈I
X
i
where ¸X
i
)
i∈I
is a countable family of preRadon Hausdorﬀ spaces, then X is
preRadon.
(i) If every point of X belongs to a preRadon open subset of X, then X is preRadon.
proof (a)(i)⇒(ii) Suppose that X is preRadon, that µ is an eﬀectively locally ﬁnite τadditive Borel
measure on X, that E ⊆ X is Borel, and that γ < µE. Because µ is semiﬁnite, there is a Borel set H ⊆ X
of ﬁnite measure such that µ(H∩E) > γ. Set νF = µ(F ∩H) for every Borel set F ⊆ X; then ν is a totally
ﬁnite Borel measure on X, and is τadditive, by 414Ea. Now νE > γ, so there is a compact set K ⊆ E such
that γ ≤ νK ≤ µK. As E is arbitrary, µ is tight.
(ii)⇒(iii) is trivial.
(iii)⇒(iv) Assume (iii), and let µ be a totally ﬁnite τadditive Borel measure on X. Let / be the
family of compact subsets of X and set α = sup
K∈K
µK. ??? Suppose, if possible, that µX > α. Let ¸K
n
)
n∈N
be a sequence in / such that sup
n∈N
µK
n
= α, and set L =
n∈N
K
n
; then
µL = lim
n→∞
µ(
i≤n
K
i
) = α.
Set νE = µ(E ¸ L) for every Borel set E ⊆ X. Then ν is a nonzero totally ﬁnite Borel measure on X, and
is τadditive, by 414E again. So there is a K ∈ / such that νK > 0. But now there is an n ∈ N such that
νK +µK
n
> α, and in this case K ∪ K
n
∈ / and
µ(K ∪ K
n
) = µK +µK
n
−µ(K ∩ K
n
) ≥ νK +µK
n
> α,
which is impossible. XXX So µX = α, as required.
(iv)⇒(i) Assume (iv), and let µ be a totally ﬁnite τadditive Borel measure on X. Suppose that
E ⊆ X is Borel and that γ < µE. By (iii), there is a compact set K ⊆ X such that µK > µX −µE +γ, so
that µ(E ∩ K) > γ. Consider the subspace measure µ
K
on K. By 414K, this is τadditive, so inner regular
with respect to the closed subsets of K (414Mb). There is therefore a relatively closed subset F of K such
that F ⊆ K∩E and µ
K
F ≥ γ; but now F is a compact subset of E and µF ≥ γ. As E and γ are arbitrary,
µ is tight. As µ is arbitrary, X is preRadon.
(ii)⇒(v) Assume (ii), and let µ be a locally ﬁnite eﬀectively locally ﬁnite τadditive Borel measure on
X. Then µ is tight, so by 416F(ii) its c.l.d. version is a Radon measure.
(v)⇒(iv) Assume (v), and let µ be a totally ﬁnite τadditive Borel measure on X. Then the c.l.d.
version ˜ µ of µ is a Radon measure; but ˜ µ extends µ (213Hc), so
sup
K⊆X is compact
µK sup
K⊆X is compact
˜ µK = ˜ µX = µX.
(b) Let µ be a locally ﬁnite quasiRadon measure on X. By (aii), µ is tight; by 416C, µ is a Radon
measure.
192 Topologies and measures II 434J
(c) Let µ be a totally ﬁnite τadditive Borel measure on X. Because X is regular, its c.l.d. version ˜ µ is
a quasiRadon measure (415C), therefore a Radon measure; but ˜ µ extends µ (213Hc again), so µ, like ˜ µ,
must be tight. As µ is arbitrary, X is preRadon.
(d) A be a universally Radonmeasurable subset of X, and µ a totally ﬁnite τadditive Borel measure
on A. Set νE = µ(E ∩ A) for every Borel set E ⊆ X; then ν is a totally ﬁnite τadditive Borel measure on
X. So its c.l.d. version (that is, its completion ˆ ν, by 213Ha) is a Radon measure on X, by (av). Now ˆ ν
measures A, so
µA = ν
∗
A = ˆ νA = sup¦ˆ νK : K ⊆ A is compact¦ = sup¦µK : K ⊆ A is compact¦.
By (biv), A is preRadon.
(e) Let µ be a totally ﬁnite Radon measure on X. Then the subspace measure µ
A
is τadditive (414K),
so its restriction ν to the Borel σalgebra of A is still τadditive. Because A is preRadon,
µ
∗
A = µ
A
A = νA = sup¦νK : K ⊆ A is compact¦
= sup¦µK : K ⊆ A is compact¦ = µ
∗
A,
and µ measures A (413Ef). As µ is arbitrary, A is universally Radonmeasurable.
(f ) Put 432B and (aiv) together.
(g) If we identify X with a G
δ
set in a compact Hausdorﬀ space Z, then Z is preRadon, by (f), so X is
preRadon, by (d).
(h) Let µ be a totally ﬁnite τadditive Borel measure on X, and > 0. Let ¸
i
)
i∈I
be a family of strictly
positive real numbers such that
i∈I
i
≤ (4A1P). For each i ∈ I, Borel set F ⊆ X
i
set µ
i
F = µπ
−1
i
[F],
where π
i
(x) = x(i) for x ∈ X; because π
i
: X → X
i
is continuous, µ
i
is a totally ﬁnite τadditive Borel
measure on X
i
. Because X
i
is preRadon, we can ﬁnd a compact set K
i
⊆ X
i
such that µ
i
(X
i
¸ K
i
) ≤
i
,
by (aiv). Now K =
i∈I
K
i
is compact (3A3J), and X ¸ K ⊆
i∈I
π
−1
i
[X
i
¸ K
i
], so
µ(X ¸ K) ≤
i∈I
µ
i
(X
i
¸ K
i
) ≤
i∈I
i
≤ .
As and µ are arbitrary, X satisﬁes the condition of (biv), and is preRadon.
(i) Let ( be a cover of X by preRadon open sets. Let µ be a nonzero totally ﬁnite τadditive Borel
measure on X. Then µX = sup¦µ(
(
0
) : (
0
⊆ ( is ﬁnite¦, so there is some G ∈ ( such that µG > 0. Now
the subspace measure µ
G
is a nonzero totally ﬁnite τadditive Borel measure on G, so there is a compact
set K ⊆ G such that µ
G
K > 0, in which case µK > 0. As µ is arbitrary, X is preRadon, by (aiii).
434K I return to criteria for deciding whether Hausdorﬀ spaces are Radon.
Proposition (a) A Hausdorﬀ space is Radon iﬀ it is Borelmeasurecomplete and preRadon.
(b) An analytic Hausdorﬀ space is Radon. In particular, any compact metrizable space is Radon and any
Polish space is Radon.
(c) ω
1
and ω
1
+ 1, with their order topologies, are not Radon.
(d) For a set I, [0, 1]
I
is Radon iﬀ ¦0, 1¦
I
is Radon iﬀ I is countable.
(e) A hereditarily Lindel¨of Kanalytic Hausdorﬀ space is Radon; in particular, the split interval (343J,
419L) is Radon.
proof (a) Put the deﬁnitions 434C, 434Gb and 434Gc together, recalling that a tight measure is necessarily
τadditive (411E).
(b) 433Cb.
(c) Dieudonn´e’s measure (411Q) is a Borel measure on ω
1
which is not tight, so ω
1
is certainly not a
Radon space; as it is an open set in ω
1
+ 1, and the subspace topology inherited from ω
1
+ 1 is the order
topology of ω
1
(4A2S(aiii)), ω
1
+ 1 cannot be Radon (434Fc).
(d) If I is countable, then ¦0, 1¦
I
and [0, 1]
I
are compact metrizable spaces, so are Radon. If I is
uncountable, then ω
1
+ 1, with its order topology, is homeomorphic to a closed subset of ¦0, 1¦
I
. PPP Set
434M Borel measures 193
κ = #(I). For ξ ≤ ω
1
, η < κ set x
ξ
(η) = 1 if η < ξ, 0 if ξ ≤ η. The map ξ → x
ξ
: ω
1
+ 1 → ¦0, 1¦
κ
is
injective because κ ≥ ω
1
, and is continuous because all the sets ¦ξ : x
ξ
(η) = 0¦ = (ω
1
+ 1) ∩ (η + 1) are
openandclosed in ω
1
+1. Since ω
1
+1 is compact in its order topology (4A2S(ai)), it is homeomorphic to
its image in ¦0, 1¦
κ ∼
= ¦0, 1¦
I
. QQQ
By 434Fc, ¦0, 1¦
I
cannot be a Radon space. Since ¦0, 1¦
I
is a closed subset of [0, 1]
I
, [0, 1]
I
is also not a
Radon space.
(e) Suppose that X is a hereditarily Lindel¨of Kanalytic Hausdorﬀ space. Then it is Borelmeasure
complete by 434Ie and preRadon by 434Jf, so by (a) here it is Radon.
Since the split interval is compact and Hausdorﬀ and hereditarily Lindel¨of (419La), it is a Radon space.
434L It is worth noting an elementary special property of metric spaces.
Proposition If (X, ρ) is a metric space, then any quasiRadon measure on X is inner regular with respect
to the totally bounded subsets of X.
proof Let µ be a quasiRadon measure on X and Σ its domain. Suppose that E ∈ Σ and γ < µE. Then
there is an open set G of ﬁnite measure such that µ(E ∩ G) > γ; set δ = µ(E ∩ G) −γ. For n ∈ N, I ⊆ X
set H(n, I) =
x∈I
¦y : ρ(y, x) < 2
−n
¦. Then ¦H(n, I) : I ∈ [X]
<ω
¦ is an upwardsdirected family of open
sets covering X. Because µ is τadditive, there is a ﬁnite set I
n
⊆ X such that µ(G¸ H(n, I
n
)) ≤ 2
−n−1
δ.
Consider F =
n∈N
H(n, I
n
). This is totally bounded and µ(G ¸ F) ≤ δ, so E ∩ F is totally bounded and
µ(E ∩ F) ≥ γ. As E and γ are arbitrary, µ is inner regular with respect to the totally bounded sets.
434M I turn next to a couple of ideas depending on countable compactness.
Lemma Let X be a countably compact topological space and c a nonempty family of closed subsets of X
with the ﬁnite intersection property. Then there is a Borel probability measure µ on X, inner regular with
respect to the closed sets, such that µF = 1 for every F ∈ c.
proof (a) By Zorn’s lemma, c is included in a maximal family c
∗
of closed subsets of X with the ﬁnite
intersection property.
(i) If F ⊆ X is closed and F ∩ F
0
∩ . . . ∩ F
n
,= ∅ for every F
0
, . . . , F
n
∈ c
∗
, then c
∗
∪ ¦F¦ has the ﬁnite
intersection property, so F ∈ c
∗
.
(ii) If F, F
t
∈ c
∗
, then F ∩ F
t
∩ F
0
∩ . . . ∩ F
n
,= ∅ for all F
0
, . . . , F
n
∈ c
∗
, so F ∩ F
t
∈ c
∗
.
(iii) If F ⊆ X is closed and F ∩ F
t
∈ c
∗
for every F
t
∈ c
∗
, then F ∩ F
0
∩ . . . ∩ F
n
∈ c
∗
for every
F
0
, . . . , F
n
∈ c
∗
(because F
0
∩ . . . ∩ F
n
∈ c
∗
, by (ii)), so F ∈ c
∗
.
(iv) If ¸F
n
)
n∈N
is a sequence in c
∗
, with intersection F, and F
t
∈ c
∗
, then F
t
∩
i≤n
F
i
,= ∅ for every
n ∈ N. Because X is countably compact, F
t
∩F ,= ∅ (4A2G(fii)). As F
t
is arbitrary, F ∈ c
∗
, by (iii). Thus
c
∗
is closed under countable intersections.
(b) Set
Σ = ¦E : E ⊆ X, there is an F ∈ c
∗
such that either F ⊆ E or F ∩ E = ∅¦,
and deﬁne ˆ µ : Σ → ¦0, 1¦ by saying that ˆ µE = 1 if there is some F ∈ c
∗
such that F ⊆ E, 0 otherwise.
Then ˆ µ is a probability measure on X. PPP
(i) ∅ ∈ Σ because c
∗
⊇ c is not empty.
(ii) X ¸ E ∈ Σ whenever E ∈ Σ because the deﬁnition of Σ is symmetric between E and X ¸ E.
(iii) If ¸E
n
)
n∈N
is any sequence in Σ with union E, then either there are n ∈ N, F ∈ c
∗
such that
F ⊆ E
n
⊆ E and E ∈ Σ, or for every n ∈ N there is an F
n
∈ c
∗
such that F
n
∩ E
n
= ∅. In this case
F =
n∈N
F
n
∈ c
∗
, by (aiv), and E ∩ F = ∅, so again E ∈ Σ. Thus Σ is a σalgebra of subsets of X.
(iv) ˆ µ∅ = 0 because ∅ cannot belong to c
∗
.
(v) If ¸E
n
)
n∈N
is any disjoint sequence in Σ with union E, then either there is some n such that ˆ µE
n
= 1,
in which case ˆ µE
i
= 0 for every i ,= n (because any two members of c
∗
must meet) and ˆ µE = 1 =
∞
i=0
ˆ µE
i
,
or ˆ µE
i
= 0 for every i, in which case, just as in (iii), ˆ µE = 0 =
∞
i=0
ˆ µE
i
. Thus ˆ µ is a measure.
(vi) Because c
∗
,= ∅, ˆ µX = 1. Thus ˆ µ is a probability measure. QQQ
(c) If F ⊆ X is a closed set, then either F itself belongs to c
∗
, so F ∈ Σ, or there is some F
t
∈ c
∗
such
that F ∩F
t
= ∅, in which case again F ∈ Σ. So Σ contains every closed set, therefore every Borel set, and ˆ µ
194 Topologies and measures II 434M
is a topological measure. By construction, ˆ µ is inner regular with respect to c
∗
and therefore with respect
to the closed sets. Finally, if F ∈ c then F ∈ c
∗
, so ˆ µF = 1. We may therefore take µ to be the restriction
of ˆ µ to the Borel σalgebra of X, and µ will be a Borel measure on X, inner regular with respect to the
closed sets, such that µE = 1 for every E ∈ c.
434N Proposition (a) Let X be a Borelmeasurecompact topological space. Then closed countably
compact subsets of X are compact.
(b) Let X be a Borelmeasurecomplete topological space. Then closed countably compact subsets of X
are compact.
(c) Let X be a Hausdorﬀ Borelmeasurecomplete topological space. Then compact subsets of X are
countably tight.
(d) In particular, any Radon compact Hausdorﬀ space is countably tight.
proof (a) Let C be a closed countably compact subset of X. Let T be an ultraﬁlter on C. Let c be the
family of closed subsets of C belonging to T. Then c has the ﬁnite intersection property, so by 434M there
is a Borel probability measure µ on C, inner regular with respect to the closed sets, such that µE = 1
for every E ∈ c. Let ν be the Borel measure on X deﬁned by setting νH = µ(C ∩ H) for every Borel
set H ⊆ X. Then ν is also inner regular with respect to the closed sets (of either C or X); because X is
Borelmeasurecompact, ν has a support F (434H(aiv)). Since νF = νX = 1, F ∩ C ,= ∅; take x ∈ F ∩ C.
If G is any open set (in X) containing x, then ν(X¸G) < 1, so C¸G / ∈ T and C∩G ∈ T. As G is arbitrary,
T → x; as T is arbitrary, C is compact.
(b) Repeat the argument of (a). Let C be a countably compact subset of X and T be an ultraﬁlter on
C. Let c be the family of relatively closed subsets of C belonging to T. Then there is a Borel probability
measure µ on C such that µE = 1 for every E ∈ c. Let ν be the Borel measure on X deﬁned by setting
νH = µ(C ∩ H) for every Borel set H ⊆ X. Because X is Borelmeasurecomplete, ν has a support F
(434I(aiii)). Since νF = νX = 1, F ∩ C ,= ∅; take x ∈ F ∩ C. If G is any open set containing x, then
ν(X ¸ G) < 1, so C ¸ G / ∈ T and C ∩ G ∈ T. As G is arbitrary, T → x; as T is arbitrary, C is compact.
(c) Again let C be a (countably) compact subset of X. Take A ⊆ C, and set C
0
=
¦B : B ∈ [A]
≤ω
¦.
Then C
0
is countably compact. PPP If ¸y
n
)
n∈N
is any sequence in C
0
, it has a cluster point y ∈ C. For each
n ∈ N there is a countable set B
n
⊆ A such that y
n
∈ B
n
. Now B =
n∈N
B
n
is a countable subset of A,
and y ∈ B ⊆ C
0
, so y is a cluster point of ¸y
n
)
n∈N
in C
0
. As ¸y
n
)
n∈N
is arbitrary, C
0
is countably compact.
QQQ
By (b), C
0
is compact, therefore closed, and must include A. Thus every point of A is in the closure of
some countable subset of A. As A is arbitrary, C is countably tight.
(d) Finally, a compact Radon Hausdorﬀ space is Borelmeasurecomplete (434Ka) and countably compact,
therefore countably tight.
434O Quasidyadic spaces I wish now to present a result in an entirely diﬀerent direction. Measures
of type B
1
in the classiﬁcation of 434A (completion regular, but not τadditive) seem to be hard to come
by. The next theorem shows that on a substantial class of spaces they cannot appear. First, we need a
deﬁnition.
Deﬁnition A topological space X is quasidyadic if it is expressible as a continuous image of a product
of separable metrizable spaces.
I give some elementary results to indicate what kind of spaces we have here.
434P Proposition (a) A continuous image of a quasidyadic space is quasidyadic.
(b) Any product of quasidyadic spaces is quasidyadic.
(c) A space with a countable network is quasidyadic.
(d) A Baire subset of a quasidyadic space is quasidyadic.
(e) If X is any topological space, a countable union of quasidyadic subspaces of X is quasidyadic.
proof (a) Immediate from the deﬁnition.
434Q Borel measures 195
(b) Again immediate; if X
i
is a continuous image of
j∈J
i
Y
ij
, where Y
ij
is a separable metric space for
every i ∈ I, j ∈ J
i
, then
i∈I
X
i
is a continuous image of
i∈I,j∈J
i
Y
ij
.
(c) Let c be a countable network for the topology of X. On X let ∼ be the equivalence relation in which
x ∼ y if they belong to just the same members of c; let Y be the space X/ ∼ of equivalence classes, and
φ : X → Y the canonical map. Y has a separable metrizable topology with base ¦φ[E] : E ∈ c¦∪¦φ[X¸E] :
E ∈ c¦. Let I be any set such that #(¦0, 1¦
I
) ≥ #(X), and for each y ∈ Y let f
y
: ¦0, 1¦
I
→ y be a
surjection. Then we have a continuous surjection f : Y ¦0, 1¦
I
→ X given by saying that f(y, z) = f
y
(z)
for y ∈ Y , z ∈ ¦0, 1¦
I
.
(d) Let ¸Y
i
)
i∈I
be a family of separable metric spaces with product Y and f : Y → X a continuous
surjection. If W ⊆ Y is a Baire set, it is determined by coordinates in a countable subset of I (4A3Nb),
so can be regarded as W
t
i∈I\J
Y
i
, where J ⊆ I is countable and W
t
⊆
i∈J
Y
i
; as
i∈J
Y
i
and W
t
are separable metrizable spaces (4A2Pa), W can be thought of as a product of separable metrizable spaces.
Now the set ¦E : E ⊆ X, f
−1
[E] is a Baire set in Y ¦ is a σalgebra containing every zero set in X, so
contains every Baire set. Thus every Baire subset of X is a continuous image of a Baire subset of Y , and is
therefore quasidyadic.
(e) If E
n
⊆ X is quasidyadic for each n ∈ N, then Z = N
n∈N
E
n
is quasidyadic, and f : Z →
n∈N
E
n
is a continuous surjection, where f(n, ¸x
i
)
i∈N
) = x
n
. So
n∈N
E
n
is quasidyadic.
434Q Theorem(Fremlin & Grekas 95) A semiﬁnite completion regular topological measure on a
quasidyadic space is τadditive.
proof ??? Suppose, if possible, otherwise.
(a) The ﬁrst step is the standard reduction to the case in which µX = 1 and X is covered by open sets of
zero measure. In detail: suppose that X is a quasidyadic space and µ
0
is a semiﬁnite completion regular
topological measure on X which is not τadditive. Let ( be an upwardsdirected family of open sets in X
such that µ
0
(
() is strictly greater than sup
G∈Ç
µ
0
G = γ say. Let ¸G
n
)
n∈N
be a nondecreasing sequence
in ( such that lim
n→∞
µ
0
G
n
= γ, and set H
0
=
n∈N
G
n
, so that µ
0
H
0
= γ; take a closed set Z ⊆
(
such that γ < µ
0
Z < ∞. Set µ
1
E = µ
0
(E ∩ Z ¸ H
0
) for every Borel set E ⊆ X. Then µ
1
is a nonzero
totally ﬁnite Borel measure on X, and is completion regular. PPP If E ⊆ X is a Borel set and > 0, there is
a zero set F ⊆ E ∩ Z ¸ H
0
such that µ
0
F ≥ µ
0
(E ∩ X ¸ H
0
) − , and now µ
1
F ≥ µ
1
E − . QQQ Note that
µ
1
(X ¸ Z) = µ
1
G = 0 for every G ∈ (.
For Borel sets E ⊆ X, set µE = µ
1
E/µ
1
X; then µ is a completion regular Borel probability measure on
X, and ( ∪ ¦X ¸ Z¦ is a cover of X by open negligible sets.
(b) Now let ¸Y
i
)
i∈I
be a family of separable metric spaces such that there is a continuous surjection
f : Y → X, where Y =
i∈I
Y
i
. For each i ∈ I let B
i
be a countable base for the topology of Y
i
. For J ⊆ I
let ((J) be the family of all open cylinders in Y expressible in the form
¦s : s(i) ∈ B
i
∀ i ∈ K¦,
where K is a ﬁnite subset of J and B
i
∈ B
i
for each i ∈ K; thus ((I) is a base for the topology of Y . Set
(
0
(J) = ¦U : U ∈ ((J), µ
∗
f[U] = 0¦ for each J ⊆ I. Note that (because every B
i
is countable) ((J) and
(
0
(J) are countable for every countable subset J of I. It is easy to see that ((J) ∩ ((K) = ((J ∩ K) for all
J, K ⊆ I, because if U ∈ ((I) is not empty it belongs to ((J) iﬀ its projection onto X
i
is the whole of X
i
for every i / ∈ J.
For each negligible set E ⊆ X, let ¸F
n
(E))
n∈N
be a family of zero sets, subsets of X ¸ E, such that
sup
n∈N
µF
n
= 1. Then each f
−1
[F
n
(E)] is a zero set in Y , so there is a countable set M(E) ⊆ I such that
all the sets f
−1
[F
n
(E)] are determined by coordinates in M(E) (4A3Nc). Let ¸ be the family of countable
subsets J of I such that M(f[U]) ⊆ J for every U ∈ (
0
(J); then ¸ is coﬁnal with [I]
≤ω
, that is, every
countable subset of I is included in some member of ¸. PPP If we start from any countable subset J
0
of I
and set
J
n+1
= J
n
∪
¦M(f[U]) : U ∈ (
0
(J
n
)¦
for each n ∈ N, then every J
n
is countable, and
n∈N
J
n
∈ ¸, because ¸J
n
)
n∈N
is nondecreasing, so
(
0
(
n∈N
J
n
) =
n∈N
(
0
(J
n
). QQQ
196 Topologies and measures II 434Q
(c) For each J ∈ ¸, set
Q
J
=
¦
n∈N
F
n
(f[U]) : U ∈ (
0
(J)¦.
Then µQ
J
= 1 and f
−1
[Q
J
] is determined by coordinates in J, while f
−1
[Q
J
] ∩U = ∅ whenever U ∈ (
0
(J).
If G ⊆ X is an open set, then G ∩ Q
J
= ∅ whenever J ∈ ¸ and there is a negligible Baire set Q ⊇ G
such that f
−1
[Q] is determined by coordinates in J. PPP Set H = π
−1
J
[π
J
[f
−1
[G]]], where π
J
: Y →
i∈J
Y
i
is the canonical map; then H is a union of members of ((J), because f
−1
[G] is open in Y and π
J
[f
−1
[G]]
is open in
i∈J
Y
i
. Also, because f
−1
[Q] is determined by coordinates in J, H ⊆ f
−1
[Q], so f[H] ⊆ Q and
µ
∗
f[H] = 0; thus all the members of ((J) included in H actually belong to (
0
(J), and H ∩ f
−1
[Q
J
] = ∅.
But this means that f
−1
[G] ∩ f
−1
[Q
J
] = ∅ and (because f is a surjection) G ∩ Q
J
= ∅, as claimed. QQQ In
particular, if G is a negligible open set in X, then G∩ Q
J
= ∅ whenever J ∈ ¸ and J ⊇ M(G).
(d) If J ∈ ¸, there are s, s
t
∈ f
−1
[Q
J
] such that sJ = s
t
J and f(s), f(s
t
) can be separated by open
sets in X. PPP Start from any x ∈ Q
J
and take a negligible open set G including x (recall that our hypothesis
is that X is covered by negligible open sets). For each n ∈ N let h
n
: X → R be a continuous function
such that F
n
(G) = h
−1
n
[¦0¦]. We know that G ∩ Q
J
,= ∅, while G ⊆ X ¸ (
n∈N
F
n
(G) ∩ Q
J
), which is a
negligible Baire set; by (c), f
−1
[X ¸ (
n∈N
F
n
(G) ∩ Q
J
)] is not determined by coordinates in J, and there
must be some n such that f
−1
[F
n
(G) ∩ Q
J
] is not determined by coordinates in J. Accordingly there must
be s, s
t
∈ Y such that sJ = s
t
J, s ∈ f
−1
[F
n
(G) ∩ Q
J
] and s
t
/ ∈ f
−1
[F
n
(G) ∩ Q
J
]. Now s ∈ f
−1
[Q
J
],
which is determined by coordinates in J; since sJ = s
t
J, s
t
∈ f
−1
[Q
J
] and s
t
/ ∈ f
−1
[F
n
(G)]. Accordingly
h
n
(f(s)) = 0 ,= h
n
(f(s
t
)) and f(s), f(s
t
) can be separated by open sets. QQQ
(e) We are now ready to embark on the central construction of the argument. We may choose inductively,
for ordinals ξ < ω
1
, sets J
ξ
∈ ¸, negligible open sets G
ξ
, G
t
ξ
⊆ X, points s
ξ
, s
t
ξ
∈ Y and sets U
ξ
, V
ξ
, V
t
ξ
∈ ((I)
such that
J
η
⊆ J
ξ
, U
η
, V
η
, V
t
η
all belong to ((J
ξ
) and G
η
∩ Q
J
ξ
= ∅ whenever η < ξ < ω
1
(using the
results of (b) and (c) to choose J
ξ
);
s
ξ
J
ξ
= s
t
ξ
J
ξ
, s
ξ
∈ f
−1
[Q
J
ξ
] and f(s
ξ
) and f(s
t
ξ
) can be separated by open sets in X (using
(d) to choose s
ξ
, s
t
ξ
);
G
ξ
, G
t
ξ
are disjoint negligible open sets containing f(s
ξ
), f(s
t
ξ
) respectively (choosing G
ξ
, G
t
ξ
);
U
ξ
∈ ((J
ξ
), V
ξ
, V
t
ξ
∈ ((I ¸ J
ξ
), s
ξ
∈ U
ξ
∩ V
ξ
⊆ f
−1
[G
ξ
], s
t
ξ
∈ U
ξ
∩ V
t
ξ
⊆ f
−1
[G
t
ξ
] (choosing U
ξ
,
V
ξ
, V
t
ξ
, using the fact that s
ξ
J
ξ
= s
t
ξ
J
ξ
).
On completing this construction, take for each ξ < ω
1
a ﬁnite set K
ξ
⊆ J
ξ+1
such that U
ξ
, V
ξ
and V
t
ξ
all
belong to ((K
ξ
). By the ∆system Lemma (4A1Db), there is an uncountable A ⊆ ω
1
such that ¸K
ξ
)
ξ∈A
is a
∆system with root K say. For ξ ∈ A, express U
ξ
as
˜
U
ξ
∩U
t
ξ
where
˜
U
ξ
∈ ((K), U
t
ξ
∈ ((K
ξ
¸ K). Then there
are only countably many possibilities for
˜
U
ξ
, so there is an uncountable B ⊆ A such that
˜
U
ξ
is constant for
ξ ∈ B; write
˜
U for the constant value. Let C ⊆ B be an uncountable set, not containing min A, such that
K
ξ
¸ K does not meet J
η
whenever ξ, η ∈ C and η < ξ (4A1Eb). Let D