100% found this document useful (21 votes)
159 views81 pages

Measure Theory and Nonlinear Evolution Equations 1st Edition Flavia Smarrazzo Instant Download

The document discusses the book 'Measure Theory and Nonlinear Evolution Equations' by Flavia Smarrazzo and Alberto Tesei, which integrates measure theory with the theory of partial differential equations (PDEs). It aims to provide a unified framework for understanding measure-valued solutions and their applications in PDEs, highlighting analytical methods and functional analysis. The book is designed for both students and advanced mathematicians, serving as a self-contained reference.

Uploaded by

eusumdkz077
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (21 votes)
159 views81 pages

Measure Theory and Nonlinear Evolution Equations 1st Edition Flavia Smarrazzo Instant Download

The document discusses the book 'Measure Theory and Nonlinear Evolution Equations' by Flavia Smarrazzo and Alberto Tesei, which integrates measure theory with the theory of partial differential equations (PDEs). It aims to provide a unified framework for understanding measure-valued solutions and their applications in PDEs, highlighting analytical methods and functional analysis. The book is designed for both students and advanced mathematicians, serving as a self-contained reference.

Uploaded by

eusumdkz077
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 81

Measure Theory and Nonlinear Evolution Equations

1st Edition Flavia Smarrazzo pdf download

https://ebookgate.com/product/measure-theory-and-nonlinear-
evolution-equations-1st-edition-flavia-smarrazzo/

Get the full ebook with Bonus Features for a Better Reading Experience on ebookgate.com
Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Nonlinear Hyperbolic Equations Spectral Theory and Wavelet


Transformations A Volume of Advances in Partial
Differential Equations 1st Edition P. R. Popivanov (Auth.)
https://ebookgate.com/product/nonlinear-hyperbolic-equations-spectral-
theory-and-wavelet-transformations-a-volume-of-advances-in-partial-
differential-equations-1st-edition-p-r-popivanov-auth/
ebookgate.com

Nonlinear Ordinary Differential Equations 4th Edition


Dominic Jordan

https://ebookgate.com/product/nonlinear-ordinary-differential-
equations-4th-edition-dominic-jordan/

ebookgate.com

Focus on Evolution Equations 1st Edition Gaston M.


N'Guerekata

https://ebookgate.com/product/focus-on-evolution-equations-1st-
edition-gaston-m-nguerekata/

ebookgate.com

Nonlinear H Infinity Control Hamiltonian Systems and


Hamilton Jacobi Equations 1st Edition M.D.S. Aliyu

https://ebookgate.com/product/nonlinear-h-infinity-control-
hamiltonian-systems-and-hamilton-jacobi-equations-1st-edition-m-d-s-
aliyu/
ebookgate.com
Nonlinear H Infinity Control Hamiltonian Systems and
Hamilton Jacobi Equations 1st Edition M.D.S. Aliyu

https://ebookgate.com/product/nonlinear-h-infinity-control-
hamiltonian-systems-and-hamilton-jacobi-equations-1st-edition-m-d-s-
aliyu-2/
ebookgate.com

Topics in Measure Theory and Real Analysis 1st Edition


Alexander B. Kharazishvili

https://ebookgate.com/product/topics-in-measure-theory-and-real-
analysis-1st-edition-alexander-b-kharazishvili/

ebookgate.com

Handbook of Nonlinear Partial Differential Equations


Second Edition Andrei D. Polyanin

https://ebookgate.com/product/handbook-of-nonlinear-partial-
differential-equations-second-edition-andrei-d-polyanin/

ebookgate.com

Inverse Problems for Kinetic and Other Evolution Equations


Yu. E. Anikonov

https://ebookgate.com/product/inverse-problems-for-kinetic-and-other-
evolution-equations-yu-e-anikonov/

ebookgate.com

Measure for Measure William Shakespeare

https://ebookgate.com/product/measure-for-measure-william-shakespeare/

ebookgate.com
Flavia Smarrazzo and Alberto Tesei
Measure Theory and Nonlinear Evolution Equations
De Gruyter Studies in
Mathematics

|
Edited by
Carsten Carstensen, Berlin, Germany
Gavril Farkas, Berlin, Germany
Nicola Fusco, Napoli, Italy
Fritz Gesztesy, Waco, Texas, USA
Niels Jacob, Swansea, United Kingdom
Zenghu Li, Beijing, China
Karl-Hermann Neeb, Erlangen, Germany

Volume 86
Flavia Smarrazzo and Alberto Tesei

Measure Theory and


Nonlinear Evolution
Equations

|
Mathematics Subject Classification 2010
Primary: 28-01, 28Axx, 28Bxx, 31A15; Secondary: 35D55, 35K55, 35L65, 35R25

Authors
Dr. Flavia Smarrazzo Prof. Dr. Alberto Tesei
Università Campus Bio-Medico di Roma Università Sapienza di Roma
Via Alvaro del Portillo 21 Dipartimento di Matematica
00128 Roma Piazzale Aldo Moro 5
Italy 00185 Roma RM
[email protected] Italy
[email protected]

ISBN 978-3-11-055600-1
e-ISBN (PDF) 978-3-11-055690-2
e-ISBN (EPUB) 978-3-11-055604-9
ISSN 0179-0986

Library of Congress Control Number: 2022930253

Bibliographic information published by the Deutsche Nationalbibliothek


The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available on the Internet at http://dnb.dnb.de.

© 2022 Walter de Gruyter GmbH, Berlin/Boston


Typesetting: VTeX UAB, Lithuania
Printing and binding: CPI books GmbH, Leck

www.degruyter.com
Contents
Preface | XI

Introduction | XIII

Part I: General theory

1 Measure theory | 5
1.1 Preliminaries | 5
1.2 Families of sets | 5
1.2.1 Measurable spaces | 5
1.2.2 Borel σ-algebras | 7
1.3 Measures | 8
1.3.1 General properties | 8
1.3.2 Borel and Radon measures | 9
1.3.3 Null sets | 12
1.4 Measures and outer measures | 15
1.4.1 Carathéodory construction | 15
1.4.2 Extension of measures | 18
1.5 Lebesgue and Lebesgue–Stieltjes measures | 22
1.5.1 Lebesgue measure in ℝN | 22
1.5.2 Lebesgue–Stieltjes measure | 26
1.6 Metric outer measures and capacities | 30
1.6.1 Capacities | 30
1.6.2 Metric outer measures | 32
1.7 Hausdorff measure and capacities | 36
1.8 Signed measures | 39
1.8.1 Hahn and Jordan decompositions | 40
1.8.2 The Banach space of finite signed measures | 43
1.8.3 Absolutely continuous and singular measures | 43
1.8.4 Concentrated and diffuse measures | 47
1.9 Vector measures | 47
1.9.1 Definitions and general results | 48

2 Scalar integration and differentiation | 53


2.1 Measurable functions | 53
2.1.1 Definition and general properties | 53
2.1.2 Convergence results | 57
2.1.3 Quasi-continuous functions | 59
VI | Contents

2.2 Integration | 61
2.2.1 Definition of integral | 61
2.3 Product measures | 65
2.3.1 Tonelli and Fubini theorems | 65
2.4 Applications | 67
2.4.1 A useful equality | 67
2.4.2 Steiner symmetrization | 68
2.5 Young measure | 70
2.6 Riesz representation theorem: positive functionals | 71
2.7 Riesz representation theorem: bounded functionals | 78
2.8 Convergence in Lebesgue spaces | 84
2.8.1 Preliminary remarks | 84
2.8.2 Uniform integrability | 85
2.8.3 Strong convergence | 91
2.8.4 Weak and weak∗ convergence | 93
2.9 Differentiation | 100
2.9.1 Radon–Nikodým derivative | 100
2.9.2 Differentiation of Radon measures on ℝN | 103

3 Function spaces and capacity | 109


3.1 Function spaces | 109
3.1.1 Distributional derivative | 109
3.1.2 Sobolev spaces | 111
3.1.3 Bessel potential spaces | 113
3.1.4 Functions of bounded variation | 114
3.1.5 Sobolev functions | 119
3.2 Capacities associated with a kernel | 121
3.2.1 Preliminaries and definitions | 121
3.2.2 The capacities Cg,p | 123
3.2.3 Dependence of Cg,p on p ∈ [1, ∞) | 128
3.3 Bessel, Riesz, and Sobolev capacities | 135
3.3.1 Bessel and Riesz capacities | 135
3.3.2 Metric properties of the Bessel capacity | 136
3.3.3 Sobolev capacity | 139
3.4 Relationship between different concepts of capacity | 142
3.4.1 Bessel versus Hausdorff | 142
3.4.2 Bessel versus Riesz | 145
3.4.3 Bessel versus Sobolev | 145
3.4.4 Sobolev versus Hausdorff | 147
3.4.5 Laplacian capacity versus Sobolev | 147
3.4.6 (m, p)-concentrated and (m, p)-diffuse measures | 148
Contents | VII

4 Vector integration | 153


4.1 Measurability of vector functions | 153
4.1.1 Measurability | 153
4.1.2 μ-measurability | 154
4.1.3 Weak and weak∗ measurability | 159
4.2 Integration of vector functions | 164
4.2.1 Bochner integrability | 164
4.2.2 Fubini theorem | 171
4.2.3 Integration with respect to vector measures | 174
4.2.4 Weaker notions of integral | 175
p p
4.3 The spaces Lp (X ; Y ), Lw (X ; Y ) and Lw ∗ (X ; Y ∗ ) | 177
4.3.1 Definition and general properties | 177
4.3.2 Spaces of continuous functions | 180
4.3.3 Convergence theorems | 181
4.3.4 Approximation results and separability | 182
4.4 Duality of vector Lebesgue spaces | 183
4.4.1 The Radon–Nikodým property | 183
4.4.2 Duality and Radon–Nikodým property | 191
4.4.3 Duality results | 197
4.4.4 Duality results: separable Y ∗ | 198
4.4.5 Duality results: separable Y | 199
4.5 Vector Lebesgue spaces of real-valued functions | 203
4.6 Vector Sobolev spaces | 207
4.6.1 Vector distributions | 207
4.6.2 Definition and general properties | 208
4.6.3 Continuous embedding | 210
4.6.4 Compact embedding | 213

5 Sequences of finite Radon measures | 219


5.1 Notions of convergence | 219
5.1.1 Strong convergence | 219
5.1.2 Weak∗ convergence | 219
5.1.3 Narrow convergence | 220
5.1.4 Prokhorov distance | 224
5.1.5 Narrow convergence and tightness | 228
5.2 Parameterized measures and disintegration | 234
5.3 Young measures revisited | 240
5.3.1 Weak∗ convergence | 244
5.3.2 Narrow convergence and tightness | 245
5.4 Sequences of Young measures associated with functions | 249
5.4.1 Weak∗ convergence of {νuj } | 250
5.4.2 Narrow convergence of {νuj } | 252
VIII | Contents

5.4.3 Uniform integrability of {uj } | 256


5.4.4 Biting lemma | 260

Part II: Applications

6 Case study 1: quasilinear parabolic equations | 279


6.1 Statement of the problem and preliminary results | 279
6.1.1 Weak solutions | 279
6.1.2 Weak entropy solutions | 282
6.2 Persistence | 283
6.3 Uniqueness | 285
6.4 Existence and regularity results | 289
6.4.1 Existence | 289
6.4.2 Regularity | 290
6.5 Proof of existence results: the approximating problems (Pn ) | 292
6.5.1 Approximation of the initial data | 292
6.5.2 A priori estimates | 296
6.6 Proof of existence results: letting n → ∞ | 302
6.7 The case of unbounded ϕ | 314
6.7.1 Definition of solution | 315
6.7.2 Persistence and uniqueness | 316
6.7.3 Existence | 318
6.7.4 Regularization | 319

7 Case study 2: hyperbolic conservation laws | 321


7.1 Statement of the problem | 321
7.1.1 Assumptions and preliminary remarks | 321
7.1.2 Definition of solution | 322
7.2 Uniqueness | 325
7.3 Existence and regularity results | 337
7.4 Proof of existence results: the approximating problems | 338
7.5 Proof of existence results | 346
7.6 Proof of regularity results | 362

8 Case study 3: forward–backward parabolic equations | 371


8.1 Statement of the problem and preliminary results | 371
8.1.1 Notions of solution | 372
8.2 The regularized problem | 374
8.2.1 Existence | 374
8.2.2 A priori estimates | 376
8.2.3 Letting ϵ → 0+ | 380
Contents | IX

8.3 Existence | 382


8.4 Asymptotic behavior | 387
8.5 Characterization of the limiting Young measure | 396

Bibliography | 403

Appendix A Topological spaces | 409

List of Symbols | 415

Index | 417
Preface
The study of measure-valued solutions of partial differential equations (PDEs in the
sequel) combines two distant mathematical areas, measure theory and theory of PDEs.
Since measure-valued solutions describe singularities of solutions of PDEs, a subject
related to the concept of capacity, fundamental results of potential theory also appear
in the discussion.
As a consequence, gathering from the literature what is needed for the study is
often difficult. In books devoted to PDEs, often the results of measure theory (e. g.,
on Young measures) that are strictly necessary for applications are presented, to the
detriment of the understanding of the theory as a whole. On the other hand, books on
measure theory and probability frequently use terminology and arguments unfamiliar
to the PDE scholar, while aspects related to the analytic functional framework remain
in the background. To some extent, similar remarks apply to treatises on potential
theory.
This book is aimed at presenting the topics mentioned above in a unified frame-
work; analytical methods of proof are mostly used, and general aspects of functional
analysis are highlighted. It is written for a wide range of possible interested parties,
including the students and advanced mathematicians. Being self-contained, it is also
intended both for self-study and as a reference book for well-known and less well-
known things. The reader is expected to have a background in real analysis, topology,
and functional analysis at the level of textbooks like [90]. Anyway, necessary prelim-
inaries on topology are recalled in Appendix A at the end of the book.
A detailed description of the contents of the chapters is given at the beginning of
each part. We do not consider it useful to suggest specific paths to the reader, who will
proceed for himself according to his own taste and interests.

https://doi.org/10.1515/9783110556902-201
Introduction
Measure-valued solutions have an important role in problems of calculus of variations
and nonlinear partial differential equations (PDEs) suggested by physics, chemistry,
biology, and engineering.
Speaking of “measure-valued solutions” is actually ambiguous, since this term
refers both to Radon measure-valued and to Young measure-valued solutions, two no-
tions that remarkably fit in the same mathematical framework. It is informative to
highlight some areas of research in which one or the other notion, or both, have played
an important role. This is the content of the following section, which, though, is not
intended to cover the whole spectrum of evolution problems where measure-valued
solutions arise (in particular, we refer the reader to [6] for the theory of gradient flows
in the Wasserstein space of probability measures and its applications to the transport
equation).

Motivations
1 Radon measure-valued solutions of elliptic and parabolic PDE’s

In the elliptic case, we are concerned with equations of the form

Lu = μ in Ω, (E0 )

where Ω ⊆ ℝN is open, u = u(x) (x ∈ Ω), L is a linear or nonlinear elliptic opera-


tor, and μ is a Radon measure on Ω. If Ω ⊂ ℝN properly, boundary conditions at the
boundary 𝜕Ω must be satisfied:

u=ν on 𝜕Ω, (BC)

the boundary data ν possibly being a Radon measure.


In the parabolic case, we think of initial value problems of the form

𝜕t u + Lu = μ in Ω × (0, T) =: Q,
{ (P0 )
u = u0 in Ω × {0},

where Ω is as above, T > 0, u = u(x, t) ((x, t) ∈ Q), and either μ or u0 , or both, are
Radon measures. Again, if Ω ⊂ ℝN properly, suitable boundary conditions at the lat-
eral boundary 𝜕Ω × (0, T), possibly involving Radon measures, must be satisfied.
A specimen of equation (E0 ) and problem (P0 ) are the Poisson equation and the
Cauchy problem for the heat equation, respectively. These cases (which correspond to
the choice Ω = ℝN , L = −Δ, μ = δx0 in (E0 ) and μ = 0, u0 = δx0 in (P0 ), δx0 being the
Dirac mass located at some point x0 ∈ Ω) played a central role in the linear theory of

https://doi.org/10.1515/9783110556902-202
XIV | Introduction

PDEs, leading to the investigation of fundamental solutions, whose superposition by


linearity allowed representing more general solutions.
On the strength of the linear theory, it seemed natural to study analogous prob-
lems in the nonlinear case. However, it soon became apparent that even in simple
semilinear cases, there were obstacles to the very existence of solutions in the usual
L1 -framework.

1.1 The elliptic case


A first case of nonexistence is described in vivid terms in [14] concerning the elliptic
semilinear equation

− Δu + |u|p−1 u = μ in ℝN (p > 0) (0.1)

with N ≥ 3. Equation (0.1) is related to the equation

λv − Δ(|v|α−1 v) = μ in ℝN (λ, α > 0) (0.2)

(setting u = |v|α−1 v, p = α1 , and scaling out λ), which is the resolvent equation for the
porous medium equation

𝜕t v = Δ(|v|α−1 v) in ℝN × (0, T) (α, T > 0). (0.3)

Since the nonlinear semigroup theory suggests to study equation (0.3) in the function
space L1 (ℝN ), it is natural to assume that μ ∈ L1 (ℝN ) in (0.1)–(0.2). A different moti-
vation to study (0.1) came from the Thomas–Fermi model for the electron density of
large atoms, in which case μ is a finite superposition of Dirac masses [65].
Let Ω ⊆ ℝN be open and bounded with smooth boundary 𝜕Ω. It was proven in
[28] that for any μ ∈ L1 (Ω) and p > 0, there exists a unique solution u ∈ W01,1 (Ω), with
|u|p ∈ L1 (Ω), of the homogeneous Dirichlet problem

−Δu + |u|p−1 u = μ in Ω,
{
u=0 in 𝜕Ω.

Well-posedness results for equation (0.1) were subsequently proven in [15].


Concerning the case where the right-hand side of (0.1) is a Radon measure, it was
soon realized that the condition
N
p< (0.4)
N −2

plays a central role. In fact, let Rf (Ω) denote the Banach space of finite (signed) Radon
measures on Ω ⊆ ℝN , endowed with the norm ‖μ‖Rf (Ω) := |μ|(Ω) (|μ|(Ω) < ∞ being the
total variation of the measure μ; see Subsection 1.8.2). If (0.4) holds, then for every
Motivations | XV

N
μ ∈ Rf (ℝN ), there exists a unique function u in the Marcinkiewicz space M N−2 (ℝN )
that solves equation (0.1) in the following sense: we have v := Δu + μ ∈ L1 (ℝN ) and
v = |u|p−1 u a. e. in ℝN (see [14, Appendix A]). On the other hand, no function exists
that satisfies (0.1) with μ = δx0 (x0 ∈ ℝN ) if condition (0.4) is not satisfied (this follows
by a simple direct argument; see [14, Remark A.4]).
The above results are now well understood. In fact, the method used to prove the
existence of solutions of (0.1) is natural and widely used (possibly with some variant;
e. g., see [27]). Consider the family of approximating problems

− Δun + |un |p−1 un = μn in ℝN , (0.5)

where {μn } ⊆ L1 (ℝN ) ∩ C ∞ (ℝN ) is an approximating sequence of μ such that

‖μn ‖L1 (ℝN ) ≤ ‖μ‖Rf (ℝN ) , (0.6)

lim ∫ ρ μn dx = ∫ ρ dμ for any ρ ∈ C0 (ℝN ) (0.7)


n→∞
ℝN ℝN

(namely, limn→∞ μn = μ weakly∗ in Rf (ℝN ); see Section 5.1). By the results in [15],
N
for each n ∈ ℕ, there exists a unique solution un ∈ M N−2 (ℝN ) of (0.5). A priori esti-
mates of un and the uniform bound (0.6) ensure that (i) the sequence {un } is relatively
compact in L1loc (ℝN ) and (ii) the sequence {vn }, vn := Δun + μn , is bounded in L1 (ℝN ).
Moreover, if (0.4) holds, then the sequence {vn } is uniformly integrable on the bounded
subsets of ℝN and hence relatively weakly compact in L1loc (ℝN ) by the Dunford–Pettis
theorem (see [14, Lemma A.1] and Theorem 2.8.18). Then letting n → ∞ in (0.5) and
using (0.7) the stated existence result follows.
If (0.4) does not hold, then the uniform integrability of {vn } fails. In view of the
Dunford–Pettis theorem, in this case, {vn } is not relatively weakly compact in L1loc (ℝN ),
in agreement with the above nonexistence statement. However, even in this case the
sequence {vn } is bounded in L1 (ℝN ) ⊆ Rf (ℝN ) and thus relatively compact in the
weak∗ topology of Rf (ℝN ) (see Theorem 5.1.7). Hence there exists a finite Radon mea-
sure v such that

lim ∫ ρ vn dx = ∫ ρ dv for any ρ ∈ C0 (ℝN ).


n→∞
ℝN ℝN

This suggests to seek the existence of solutions of (a suitably modified version of)
equation (0.1) in a larger class of Radon measure-valued solutions (see [97]).
On the other hand, the measures μ ∈ Rf (ℝN ) for which equation (0.1) has a so-
lution u ∈ Lp (ℝN ) have been characterized in [10] (see also [52]): for any p ∈ (1, ∞),
this happens if and only if μ ∈ L1 (ℝN ) + W −2,p (ℝN ). By Theorem 3.4.15 an equivalent
statement is that (0.1) has a solution u ∈ Lp (ℝN ) if and only if μ is diffuse with respect
XVI | Introduction

p
to the Sobolev capacity C2,q , where q := p−1
is the conjugate exponent of p, that is, if
and only if

μ(E) = 0 for any Borel set E ⊆ ℝN such that C2,q (E) = 0 (0.8)

(see Subsections 1.8.4 and 3.3.3).


By Remark 3.3.2(ii) there are no Borel sets E ⊆ ℝN such that C2,q (E) = 0 if p ∈
N
(1, ∞) and 2q > N ⇔ p < N−2 , and hence condition (0.8) is void if (0.4) holds (see
N
Proposition 3.4.11). On the other hand, when p ≥ N−2 , we have C2,q ({x0 }) = 0 for any
x0 ∈ ℝN , thus the Dirac mass δx0 is concentrated with respect to C2,q , and condition
(0.8) is not satisfied (see Definition 1.8.10 and Proposition 3.3.5). In other words, if
N
p ≥ N−2 , then the Dirac mass is not a removable singularity for (0.1) (see [105] and
references therein). This explains the above existence and nonexistence results for
equation (0.1).

Remark 0.1.1. With reference to equation (0.2), it follows from the above remarks that
no function solves

λv − Δ(|v|α−1 v) = μ in ℝN (λ, α > 0)

N−2
if N ≥ 3, μ is concentrated with respect to the Sobolev capacity C2, 1 , and α ≤ N
. This
1−α
is in agreement with the parabolic results in [26] and [80, Theorems 1, 2] (see below).

Let us mention that the above existence and nonexistence results concerning
equation (0.1) stimulated subsequent work in various directions, in particular, con-
cerning different nonlinear elliptic operators in problem (E0 ) and/or the case where
𝜕Ω ≠ 0 and g ∈ Rf (𝜕Ω) in (BC) (see [14, 13, 37, 55, 69, 106] and references therein).

1.2 The parabolic case


Consider the Cauchy problem for the fast diffusion porous medium equation

𝜕t u = Δuα in ℝN × (0, T) =: S,
{ (C)
u = u0 in ℝN × {0},

where u0 belongs to the cone R+f (ℝN ) of nonnegative finite Radon measures on ℝN
(N ≥ 3), and α ∈ (0, N−2 N
]. In [80] a solution of problem (C) is, by definition, any u ∈
L1 (S) that satisfies the first equation in distributional sense and the second in the sense
of the narrow convergence, namely,

ess lim+ ∫ ρ u(⋅, t) dx = ∫ ρ du0 for any ρ ∈ Cb (ℝN )


t→0
ℝN ℝN
Motivations | XVII

(see Definition 5.1.3; here Cb (ℝN ) denotes the space of continuous bounded real
functions on ℝN ). It is proven that such a solution – more precisely, a solution
u ∈ C(0, T; L1 (ℝN )) – exists if and only if u0 is diffuse with respect to the Sobolev
capacity C2, 1 (ℝN ) (see [80, Theorems 1, 2]). This is in agreement with [26], where it
1−α

was observed that such a solution of (C) with u0 = δ0 exists if and only if α ∈ ( N−2
N
, 1);
in fact, for a point x0 ∈ ℝN , we have C2, 1 ({x0 }) = 0, and thus the Dirac mass δx0 is
1−α

concentrated with respect to C2, 1 , if and only if α ∈ (0, N−2


N
] (see Proposition 3.4.11).
1−α
The result also agrees with Theorem 6.7.1 (see equality (6.135)). The connection with
the above elliptic results is apparent (see Remark 0.1.1).
Following (C), initial value problems for quasilinear parabolic equations whose
initial datum is a Radon measure have been widely investigated (in particular, see [16,
26, 35, 36, 81, 104] and references therein), always seeking function-valued solutions.
To highlight this point, consider the initial-boundary value problem

{
{𝜕t u = Δϕ(u) in Ω × (0, T) =: Q,
{
{u=0 on 𝜕Ω × (0, T), (P)
{
{
{u = u0 in Ω × {0},

which will be discussed at length in Chapter 6; here Ω ⊆ ℝN is an open bounded set


with smooth boundary 𝜕Ω, T > 0, u0 ∈ Rf (Ω), and ϕ is assumed to be continuous and
nondecreasing in ℝ. In our parlance, a solution u of (P) is function-valued if for any
t > 0, u(⋅, t) ∈ L1 (Ω). Such a solution can be viewed as a particular Radon measure-
valued solution, namely, a solution u of (P) such that u(⋅, t) ∈ Rf (Ω) for positive times
(see [85, 86, 87] and references therein). Whether or not such a Radon measure-valued
solution becomes function-valued for some t > 0 is a regularity problem, which can be
also regarded as a problem of removal of singularities. Not surprisingly, as we will see
below, this issue is related to some characteristic capacity pertaining to the evolution
equation under consideration.
Defining Radon measure-valued solutions of (P) raises the question of how to de-
fine nonlinear functions of such measures. In [11] a natural definition was proposed for
a similar problem by a heuristic argument, in agreement with the notion of a nonlin-
ear function of measures given in [38, 39]. Accordingly, we think of a Radon measure-
valued solution of (P) as a map from (0, T) to Rf (Ω) that satisfies (P) in the following
weak sense:
T

∫⟨u(⋅, t), 𝜕t ζ (⋅, t)⟩dt = ∬ ∇[ϕ(ur )] ⋅ ∇ζ dxdt − ⟨u0 , ζ (⋅, 0)⟩


0 Q

for a suitable class of test functions ζ (see Chapter 6 for technical details). Specifically,
the measure u ∈ Rf (Q) is required to belong to the space L∞ w∗ (0, T; Rf (Ω)), thus the
Radon measure u(⋅, t) on Ω is defined for a. e. t ∈ (0, T), and the function ϕ(ur ) is
XVIII | Introduction

required to belong to the space L1 (0, T; W01,1 (Ω)). Here ur ∈ L1 (Q) is the density of the
absolutely continuous part uac of u with respect to the Lebesgue measure, and ⟨⋅, ⋅⟩
denotes the duality map between C0 (Ω) and Rf (Ω) (see Chapters 2 and 4; by abuse of
notation, we sometimes further identify ur with uac ).
Radon measure-valued solutions of (P) can be constructed as outlined before in
the elliptic case, approximating the initial data u0 by a suitable sequence {u0n } ⊆ L1 (Ω)
and studying the convergence in L∞ w∗ (0, T; Rf (Ω)) of the corresponding sequence of
solutions of the approximating problems (see Chapter 6). As for regularity, if the func-
N−2
tion |ϕ(z)| diverges more than |z| N as |z| → ∞, then every constructed solution
of (P) belongs to L∞ (0, T; L1 (Ω)), that is, an instantaneous Rf -L1 regularization oc-
curs (see Proposition 6.7.9). On the other hand, if for some M > 0 and α ∈ (0, N−2 N
],
α
|ϕ(z)| ≤ M(1 + |z|) (z ∈ ℝ), then the C2, 1 -concentrated part of any weak solution of
1−α
(P) is constant in time, in agreement with [80] (see Theorem 6.7.1). Therefore, in this
case, every such solution is Radon measure-valued for any positive time.

2 Young and Radon measure-valued solutions of hyperbolic conservation laws

Consider the Cauchy problem

𝜕t u + div[ϕ(u)] = 0 in S,
{ (CL)
u = u0 in ℝN × {0},

with ϕ ≡ (ϕ1 , . . . , ϕN ) ∈ (C 1 (ℝ))N , ϕj : ℝ 󳨃→ ℝ nonlinear (j = 1, . . . , N).

2.1 Young measure-valued solutions


Let u0 ∈ L∞ (ℝN ). A classical way to prove the existence of weak solutions of (CL) is
the vanishing viscosity method, which relies on the companion parabolic problem

𝜕t uϵ + div[ϕ(uϵ )] = ϵ Δuϵ in S
{ (ϵ > 0), (CLϵ )
uϵ = u0ϵ in ℝN × {0}

where u0ϵ ∈ Cb (ℝN ) and ‖u0ϵ ‖L∞ (ℝN ) ≤ ‖u0 ‖L∞ (ℝN ) . For every ϵ > 0, there exists
a unique classical solution uϵ of (CLϵ ), which by the maximum principle satisfies the
inequality

‖uϵ ‖L∞ (S) ≤ ‖u0 ‖L∞ (ℝN ) . (0.9)

By inequality (0.9) and the Banach–Alaoglu theorem there exist a sequence {uk } ≡
{uϵk } ⊆ L∞ (S) with ϵk → 0+ as k → ∞ and u ∈ L∞ (S) such that limk→∞ uk = u weakly∗
Motivations | XIX

in L∞ (S). Then u is an obvious candidate to be a weak solution of (CL). However, letting


k → ∞ in the right-hand side of the weak formulation of (CLϵk ),

∬{uk 𝜕t ζ + ϵk uk Δζ } dxdt + ∫ u0 ζ (x, 0) dx = − ∬[ϕ(uk )] ⋅ ∇ζ dxdt, (0.10)


S ℝ S

is cumbersome, since the weak∗ convergence in L∞ (S) of uk to u does not imply that
of ϕj (uk ) to ϕj (u), j = 1, . . . , N (see Example 2.8.1(i); here ζ is a suitable test function).
To overcome this difficulty, we need further information, for instance, the conver-
gence uk → u a. e. in S (of some subsequence of {uk }, not relabeled for simplicity). In
fact, by inequality (0.9), the continuity of ϕ, and the dominated convergence theorem,
this convergence would imply that

lim ∬[ϕ(uk )] ⋅ ∇ζ dxdt = ∬[ϕ(u)] ⋅ ∇ζ dxdt.


k→∞
S S

Then letting k → ∞ in (0.10), we would obtain, as anticipated,

∬{u 𝜕t ζ + [ϕ(u)] ⋅ ∇ζ } dxdt + ∫ u0 ζ (x, 0) dx = 0. (0.11)


S ℝ

To establish the a. e. convergence of {uk } in S, a typical approach is setting ℝN =


⋃∞n=1 Kn with each Kn compact and then proving by the Fréchet–Kolmogorov theorem
that for any fixed n ∈ ℕ, the sequence of restrictions {uk |Kn ×(0,T) } are relatively compact
in L1 (Kn × (0, T)). Then by classical results and a diagonal argument the result follows
(see the proof of [68, Theorem 4.62 of Chapter 2]).
Let us outline a different method, which makes use of Young measures (e. g., see
[68] and references therein). Let λN denote the Lebesgue measure on ℝN , let U ⊆ S ⊆
ℝN+1 be open and bounded, and let R+f (U × ℝ) be the cone of finite (positive) Radon
measures on U × ℝ. By definition a Young measure on U × ℝ is any ν ∈ R+f (U × ℝ) such
that

ν(E × ℝ) = λN+1 (E) for any Borel set E ⊆ U.

If f : U 󳨃→ ℝ is measurable, then the Young measure νf such that

νf (E × F) = λN+1 (E ∩ f −1 (F)) for all Borel sets E ⊆ U and F ⊆ ℝ

is called the Young measure associated with the function f (see Definition 2.5.3).
Let us denote by Y(U; ℝ) the set of Young measures on U × ℝ (U ⊆ S) and by
Cb (U × ℝ) that of bounded Carathéodory integrands on U × ℝ (see Definition 5.3.1). The
following results are particular consequences of Proposition 5.3.1, Theorem 5.4.5, and
Proposition 5.4.10.
XX | Introduction

Proposition 0.2.1. Let ν ∈ Y(U; ℝ). Then for a. e. (x, t) ∈ U, there exists a probability
measure ν(x,t) on ℝ such that for any g ∈ Cb (U; ℝ), the map (x, t) → ∫ℝ g(x, t, y) dν(x,t) (y)
is integrable on U, and

∫ g dν = ∫ dxdt ∫ g(x, t, y) dν(x,t) (y).


U×ℝ U ℝ

The family {ν(x,t) } ≡ {ν(x,t) }(x,t)∈U is a parameterized measure on ℝ, called disinte-


gration of ν (see Definitions 5.2.1 and 5.2.2). The function u∗ defined as

u∗ (x, t) := ∫ y dν(x,t) (y) ((x, t) ∈ U) (0.12)


is called the barycenter of the disintegration {ν(x,t) }.

Theorem 0.2.2. Let {un } ⊆ L1 (U) be bounded, and let {νn } ≡ {νun } be the sequence of
associated Young measures. Then:
(i) there exist a subsequence {unk } ⊆ {un } and a Young measure ν on U × ℝ such that
νnk → ν narrowly;
(ii) for any f ∈ C(ℝ) such that the sequence {f (unk )} ⊆ L1 (U) is bounded and uniformly
integrable, we have

∫ dxdt ∫󵄨󵄨󵄨f (y)󵄨󵄨󵄨 dν(x,t) (y) < ∞, f (unk ) ⇀ f ∗ in L1 (U),


󵄨 󵄨
U ℝ

where

f ∗ (x, t) := ∫ f (y) dν(x,t) (y) for a. e. (x, t) ∈ U; (0.13)


(iii) for any Carathéodory function g : U × ℝ → ℝ such that the sequence {g(⋅, unk )} ⊆
L1 (U) is bounded and uniformly integrable, we have

lim ∫ g(x, t, unk (x, t)) dxdt = ∫ g dν.


k→∞
U U×ℝ

(iv) [Biting lemma] there exist a subsequence {unk } ≡ {unj } ⊆ {unk } and a sequence
j
of measurable sets {Uj } such that Uj+1 ⊆ Uj ⊆ U for any j ∈ ℕ, λN+1 (Uj ) → 0 as
j → ∞, and the sequence {unj χU\Uj } is uniformly integrable. Moreover, u∗ ∈ L1 (U),
and unj χU\Uj ⇀ u∗ in L1 (U).

Let us go back to the problem of letting k → ∞ in (0.10). Set again ℝN =


⋃∞
n=1 Kn with each Kn compact. For any fixed n ∈ ℕ, consider the sequence of re-
strictions {uk |Kn ×(0,T) }, which is bounded in L1 (Kn × (0, T)) and uniformly integrable
Motivations | XXI

by Lemma 2.8.12. Clearly, the same holds for the sequence {ϕ(uk )|Kn ×(0,T) }. Then by
Theorem 0.2.2(ii) (see also Remark 5.4.2(ii)) and a diagonal argument there exist a
subsequence of {uk } (not relabeled for simplicity) and a Young measure ν ∈ Y(S; ℝ)
such that

uk ⇀ u∗ , ϕj (uk ) ⇀ ϕ∗j in L1 (U)

for any bounded open U ⊆ S, where

ϕ∗j (x, t) := ∫ ϕj (y) dν(x,t) (y) for a. e. (x, t) ∈ U (j = 1, . . . , N). (0.14)


Since limk→∞ uk = u weakly∗ in L∞ (S), it follows that u = u∗ . We say that the couple
(u, ν) is a weak Young measure-valued solution of (CL), meaning that

u(x, t) = ∫ y dν(x,t) (y) for a. e. (x, t) ∈ S, (0.15a)


and for any ζ ∈ C 1 ([0, T]; Cc1 (ℝ)) with ζ (⋅, T) = 0 in ℝ, we have

∬{u 𝜕t ζ + ϕ∗ ⋅ ∇ζ } dxdt + ∫ u0 ζ (x, 0) dx = 0 (0.15b)


S ℝ

with ϕ∗ ≡ (ϕ∗1 , . . . , ϕ∗N ) given by (0.14).


So far only inequality (0.9) has been used. More can be said if each component
ϕj of the flux is genuinely nonlinear (see assumption (A2 ) in Chapter 7). In this case,
it can be proven that ν is in fact the Young measure associated with u, that is, ν = νu .
Then by Proposition 5.3.2 ν(x,t) = δu(x,t) for a. e. (x, t) ∈ S (see (5.53)), whence by (0.14)
ϕ∗ = ϕ(u) a. e. in S, and equality (0.15b) reduces to (0.11). Equivalently, knowing that
ν = νu implies that, up to a subsequence, uk → u a. e. in S (see Proposition 5.4.1(ii)),
and thus we can argue as before to get (0.11).
The proof of the key equality ν = νu relies on the so-called Murat–Tartar equal-
ity, which in turn makes use of the div–curl lemma (see [46, Corollary 1.3.1 and Theo-
rem 5.2.1], [68, Section 3.3]; we will use a similar argument to prove Proposition 7.5.6).

2.2 Radon measure-valued solutions


Problem (CL) in one space dimension,

𝜕t u + 𝜕x [ϕ(u)] = 0 in ℝ × (0, ∞),


{ (CL1 )
u = u0 in ℝ × {0},
XXII | Introduction

was studied in [67] with u0 ∈ R+f (ℝ) for a class of fluxes ϕ superlinear at infinity (the
model case being ϕ(u) = um , m > 1). Entropy solutions of (CL1 ) were meant in the
following sense:
– for positive times, u is a function: u ∈ L∞ (0, T; L1 (ℝ)) ∩ L∞ ((τ, T) × ℝ) for every τ ∈
(0, T);
– for all k ∈ ℝ and ζ ∈ Cc1 (S), ζ ≥ 0, we have the entropy inequality

∬{|u − k| 𝜕t ζ + sgn(u − k)[ϕ(u) − ϕ(k)]𝜕x ζ } dxdt ≥ 0;


S

– the initial condition is satisfied in the sense of the narrow convergence:

ess lim+ ∫ u(⋅, t)ρ dx = ∫ ρ du0 for any ρ ∈ Cb (ℝ).


t→0
ℝ ℝ

The following result was proven.

Proposition 0.2.3. Let ϕ : [0, ∞) 󳨃→ [0, ∞), ϕ(0) = 0 be increasing, and let z 󳨃→
[ϕ(z)]1−α be convex on (0, ∞) for some α ∈ (0, 1). Then there exists a unique entropy
solution u ≥ 0 of (CL1 ).

The existence part of Proposition 0.2.3 follows by a constructive procedure, in


which the initial Radon measure u0 ≥ 0 is approximated by a sequence {u0n } ⊆
L∞ (ℝ) ∩ L1 (ℝ), u0n → u0 narrowly in ℝ. If ϕ(u) = um with m > 1 and u0 = δ0 , then this
procedure gives the fundamental solution of the Burgers equation, called N-wave:
1/(m−1)
x
u(x, t) = ( ) χA (x, t), (0.16a)
mt
(m−1)/m
m
A := {(x, t) | 0 ≤ x ≤ ( ) t 1/m , 0 < t ≤ T}. (0.16b)
m−1

Hence the solution of (CL1 ) in this case is function valued (plainly, the same holds for
m ∈ (0, 1)). Instead, in the linear case m = 1 the initial singularity δ0 obviously persists,
and thus the solution is Radon measure valued for all times.
A novel situation prevails when studying (CL1 ) with bounded ϕ, a problem (moti-
vated by a mathematical model of ion etching [88]; see also [50]) that is addressed in
Chapter 7. To illustrate its main features, consider the model

1
𝜕t u + 𝜕x [1 − (1+u)m
] =0 in ℝ × (0, ∞), m > 0,
{ (CL2 )
u = δ0 in ℝ × {0},

1
where ϕ(z) = 1 − (1+z) m is increasing and concave and belongs to a class for which the

constructed entropy solution of problem (CL2 ) is unique (see Definition 7.3.1 and The-
orems 7.3.3–7.3.4). As in [67], we construct entropy solutions of (CL2 ) as limiting points
Motivations | XXIII

(in a suitable topology) of the sequence of entropy solutions of the approximating Rie-
mann problems:

{𝜕t un + 𝜕x [1 − (1+u1 )m ] = 0 in S,
n
(Rn )
un = n χ(− 1 , 1 ) in ℝ × {0}.
{
{ 2n 2n

1
An elementary analysis of (Rn ) shows that a shock starts from the point (− 2n , 0), a rar-
1
efaction fan from ( 2n , 0). They meet at a point (xn , tn ),

tn ϕ(n) 1
xn := [ + ϕ′ (n)], tn := (n ∈ ℕ),
2 n ϕ(n) − nϕ′ (n)

where a new shock starts. It is easily seen that limn→∞ tn = 1 and limn→∞ xn = 0. This
explains the following result (see [18]).

Proposition 0.2.4. The unique (constructed) entropy solution of problem (CL2 ) is u =


ur + us , where
1
mt 1+m
ur (x, t) := [( ) − 1] χA (x, t), us (⋅, t) := [1 − t]+ δ0 ,
x (0.17)
A := {(x, t) | 0 < x ≤ mt, 0 ≤ t ≤ 1} ∪ {(x, t) | ξ (t) ≤ x ≤ mt, 1 < t ≤ T},

and ξ solves the problem


m
(mtξ −1 )− 1+m −1
{ξ ′ = − 1 in (1, T),
(mtξ −1 ) 1+m −1
{
{ξ (1, 1) = 0.

In view of (0.17), in this case the singular part of the solution of (CL1 ) is nonzero
until the waiting time t = 1 and vanishes for t ≥ 1, and hence we have deferred regu-
larization. Another important remark is that the singular part us (⋅, t) is concentrated
at x0 = 0 for 0 < t < τ1 = 1 (see (0.17)). When approaching supp us (⋅, t) = {0}, for any
t ∈ (0, τ1 ), the density ur (⋅, t) satisfies the compatibility conditions

lim ϕ(ur (x, t)) ≡ ϕ(ur (0− , t)) = 0, lim ϕ(ur (x, t)) ≡ ϕ(ur (0+ , t)) = 1. (0.18)
x→0− x→0+

We refer the reader to Chapter 7 to highlight the essential role played by the compati-
bility conditions in the uniqueness proof for problem (CL1 ) with bounded ϕ (see Defi-
nition 7.2.1 and Theorem 7.2.4).
XXIV | Introduction

3 Young and Radon measure-valued solutions of ill-posed problems

Let us now consider the forward-backward parabolic equation

𝜕t u = Δϕ(u) in Ω × (0, T) =: Q, (FB)

where Ω ⊆ ℝN (N ≥ 1) is open, and ϕ ∈ C 1 (ℝ) is nonmonotonic. Equation (FB) can be


seen as the gradient flow (in the H −1 topology) of the functional

ℰ [u] := ∫ Φ(u) dx, (0.19)


Ω

where Φ′ = ϕ. Since ϕ is nonmonotonic, the functional ℰ is nonconvex, and initial-


boundary value problems for (FB) are ill-posed. The dynamics associated with (FB)
is relevant whenever nonconvex functionals arise, e. g., in phase transitions, nonlin-
ear elasticity, and image processing (see [29, 73, 79]). Ill-posedness is revealed by the
lack of uniqueness of solutions (see [59], [108] for ϕ satisfying (A1 ), respectively, (A2 )
below).
In the Landau theory of phase transitions the function Φ(u) = (1 − u2 )2 is the
double-well potential, and thus ϕ is a cubic:

ϕ′ (u) > 0 if u ∈ (−∞, b) ∪ (a, ∞), ϕ′ (u) < 0 if u ∈ (b, a);


{ (A1 )
ϕ(u) → ±∞ as u → ±∞.

The three monotone branches of the graph of v = ϕ(u) are denoted by

u := s1 (v), v ∈ (−∞, B) ⇔ v = ϕ(u), u ∈ (−∞, b),


u := s0 (v), v ∈ (A, B) ⇔ v = ϕ(u), u ∈ (b, a), (0.20)
u := s2 (v), v ∈ (A, ∞) ⇔ v = ϕ(u), u ∈ (a, ∞)

(here A := ϕ(a), B := ϕ(b); see Fig. 1). Heuristically, the branches s1 and s2 correspond
to stable phases, s0 corresponds to an unstable phase, and equation (FB) describes
the dynamics of transition between stable phases.
Ill-posedness calls for several refinements of the Landau theory, including in par-
ticular nonlocal spatial effects and/or time delay effects. In the first case, starting from
the so-called Allen–Cahn functional

κ
ℰ [u] := ∫{Φ(u) + |∇u|2 } dx,
2
Ω

we obtain the Cahn–Hilliard equation

𝜕t u = Δ[ϕ(u) − κΔu] (κ > 0), (0.21)


Motivations | XXV

Figure 1: Cubic ϕ.

namely a fourth-order regularization of (FB). Instead, taking time delay effects into
account leads to the Sobolev regularization of (FB):

𝜕t u = Δ[ϕ(u) + ϵ 𝜕t u] (ϵ > 0).

If both effects are included, then we obtain the viscous Cahn–Hilliard equation

𝜕t u = Δ[ϕ(u) − κΔu + ϵ 𝜕t u].

Other regularizations of (FB), based on time or space discretization, have been used
by several authors [12, 43, 57, 61].
Beside phase transitions, a motivation for studying equation (FB) comes from
population dynamics, where it arises as a quasi-continuous approximation of a dis-
crete model for aggregating populations (see [76, 77] and references therein). In this
framework, u ≥ 0 has the meaning of population density, and ϕ ∈ C 2 ([0, ∞)) is a
nonmonotonic function with the following properties:

{
{ ϕ(u) > 0 if u > 0, ϕ(0) = 0;
{
{
{ϕ (u) > 0
{ ′
if 0 ≤ u < u,̄ ϕ′ (u) < 0 if u > u;̄
{ (A2 )
{ϕ′ (u)̄ = 0, ϕ′′ (u)̄ ≠ 0, ϕ(u) → 0 as u → ∞;
{
{
{
{ p
{ϕ ∈ L (0, ∞) for some p ∈ [1, ∞)

(see Fig. 2). Functions satisfying (A2 ) are often called “of Perona–Malik type”, since
for N = 1, the Perona–Malik equation reads

u
𝜕t z = 𝜕x [ϕ(𝜕x z)] with ϕ(u) = (u ≥ 0), (0.22)
1 + u2
XXVI | Introduction

Figure 2: ϕ of Perona–Malik type.

and ϕ above satisfies (A2 ) (see [79]). Equation (0.22) also models the formation of layers
of constant temperature (or salinity) in the ocean [11]. Setting u := 𝜕x z in (0.22) gives

u
𝜕t u = 𝜕xx ( ) (u ≥ 0), (0.23)
1 + u2

a particular case of (FB) (let us mention that equation (FB) with N = 1 and cubic ϕ was
studied in [77]). Observe that the first equation in (0.22) is the gradient flow of (0.19)
with the Perona–Malik potential Φ(u) = log(1 + u2 ).
Sound modeling arguments were produced in [11] to motivate the pseudoparabolic
regularization of (0.22):

𝜕t z = 𝜕x [ϕ(𝜕x z)] + ϵ 𝜕tx [ψ(𝜕x z)],

where the regularization term with


u
ϕ(z)
ψ(u) := −ϕ(u) + ∫ dz (u ≥ 0) (0.24)
z
0

is obtained introducing in (FB) time delay effects. For the function ϕ in (0.22), we ob-
u
tain ψ(u) = − 1+u 2 + arctg u, and hence

π
ψ′ (u) > 0 for all u ≥ 0, ψ(u) → as u → ∞.
2

In general, if ϕ satisfies (A2 ), then the function ψ : [0, ∞) → [0, ∞) satisfies the fol-
lowing requirements:

ψ′ (u) > 0 for all u ≥ 0, ψ(u) → ψ∞ ∈ (0, ∞) as u → ∞. (A3 )


Motivations | XXVII

Clearly, the pseudoparabolic regularization is weaker than the Sobolev regularization,


which formally corresponds to the case ψ(u) = u. In agreement with the correspon-
dence between (0.22) and (0.23), the pseudoparabolic regularization of (FB) reads

𝜕t u = Δϕ(u) + ϵ 𝜕t Δ[ψ(u)] (u ≥ 0).

3.1 Young measure-valued solutions


Consider the initial-boundary value problem for (FB) with Sobolev regularization:

{
{ 𝜕t u = Δv in Q,
{
𝜕 v=0 in 𝜕Ω × (0, T), (Sϵ )
{ ν
{
{
{u = u 0 in Ω × {0},

where v := ϕ(u) + ϵ 𝜕t u, Ω ⊆ ℝN (N ≥ 1) is open and bounded with smooth boundary


𝜕Ω, and 𝜕ν denotes the outer normal derivative. The following results were proven in
[74] (see also [76]).

Lemma 0.3.1. Let ϕ ∈ C 1 (ℝ), and let u0 ∈ C(Ω). Then for every ϵ > 0, there exists
Tϵ > 0 such that problem (Sϵ ) has a unique strong solution (uϵ , vϵ ) in QTϵ := Ω × (0, Tϵ ).
Moreover:
u
(i) let g ∈ C 1 (ℝ), g ′ ≥ 0, and G(u) := ∫0 g(ϕ(s))ds + c (c ∈ ℝ). Then

∫ G(uϵ (x, t)) dx ≤ ∫ G(u0 (x)) dx for any t ∈ (0, Tϵ ); (0.25)


Ω Ω

(ii) let u1 , u2 ∈ ℝ, u1 < u2 , and

ϕ(u1 ) ≤ ϕ(u) ≤ ϕ(u2 ) for any u ∈ [u1 , u2 ]. (0.26)

Let u0 (x) ∈ [u1 , u2 ] for every x ∈ Ω. Then uϵ (x, t) ∈ [u1 , u2 ] for every (x, t) ∈ QTϵ .

Proof. Claim (ii) above follows from (i) by a proper choice of the function g. To prove
(i), observe that

𝜕t [G(uϵ )] = g(ϕ(uϵ ))𝜕t uϵ = g(vϵ )Δvϵ + [g(ϕ(uϵ )) − g(vϵ )]Δvϵ


vϵ − ϕ(uϵ )
= div[g(vϵ )∇vϵ ] − g ′ (vϵ )|∇vϵ |2 + [g(ϕ(uϵ )) − g(vϵ )] .
ϵ
⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟
≤0

Integrating on Ω the above inequality gives (0.25).

If (A1 ) holds, then the function ϕ satisfies (0.26) for a suitable choice of u1 < 0 < u2
with |u1 |, |u2 | sufficiently large. Hence we get a uniform L∞ -estimate of {uϵ }. The other
XXVIII | Introduction

estimates in (0.27) below follow similarly from the proof of inequality (0.25), and thus
we have the following result.

Proposition 0.3.2. Let (A1 ) be satisfied, and let u0 ∈ C(Ω). Then:


(i) for every ϵ > 0, there exists a unique strong solution (uϵ , vϵ ), with vϵ = ϕ(uϵ )+ϵ 𝜕t uϵ ,
of the regularized problem (Sϵ );
(ii) there exists M > 0 such that for any ϵ > 0,

max{󵄩󵄩󵄩uϵ 󵄩󵄩󵄩L∞ (Q) , √ϵ 󵄩󵄩󵄩𝜕t uϵ 󵄩󵄩󵄩L2 (Q) , 󵄩󵄩󵄩vϵ 󵄩󵄩󵄩L∞ (Q) , 󵄩󵄩󵄩vϵ 󵄩󵄩󵄩L2 (0,T;H 1 (Ω)) } ≤ M. (0.27)
󵄩 󵄩 󵄩 󵄩 󵄩 󵄩 󵄩 󵄩

By inequality (0.27) there exist sequences {ϵk } with ϵk → 0+ as k → ∞, {uk } ≡ {uϵk },


and {vk } ≡ {vϵk }, and u ∈ L∞ (Q) and v ∈ L∞ (Q) ∩ L2 (0, T; H 1 (Ω)) such that
∗ ∗
uk ⇀ u in L∞ (Q), ϕ(uk ) ⇀ v in L∞ (Q),
{ ∗ (0.28)
vk ⇀ v in L∞ (Q), vk ⇀ v in L2 (0, T; H 1 (Ω))

(observe that the sequences {vk } and {ϕ(uk )} admit the same weak* limit in L∞ (Q)). By
(0.28), letting k → ∞ in the weak formulation of (Sϵk ), we easily obtain

∬{u 𝜕t ζ − ∇v ⋅ ∇ζ } dxdt + ∫ u0 ζ (x, 0) dx = 0 (0.29)


Q Ω

for any ζ ∈ C 1 ([0, T]; C 1 (Ω)), ζ (⋅, T) = 0 in Ω.


Since the convergences in (0.28) do not imply the equality v = ϕ(u), u need not be
a weak solution of the problem

{
{𝜕t u = Δ[ϕ(u)] in Q,
{
{𝜕ν [ϕ(u)] = 0 in 𝜕Ω × (0, T), (S)
{
{
{u = u0 in Ω × {0}.

However, a weak Young measure-valued solution of (S) is easily obtained. Since


‖uk ‖L∞ (Q) ≤ M (see (0.27)) and Q is bounded, the sequence {uk } is bounded in L1 (Q)
and uniformly integrable (see Lemma 2.8.12), and the same holds for the sequence
{ϕ(uk )}. Then there exist a subsequence of {uk } (not relabeled) and a Young measure
ν ∈ Y(Q; ℝ) such that

uk ⇀ u∗ , ϕ(uk ) ⇀ ϕ∗ in L1 (Q),

where u∗ and ϕ∗ are given by (0.12) and (0.13), respectively, with U = Q and f = ϕ.
From (0.28) and the above convergences we get u = u∗ , v = ϕ∗ , whence by (0.29)

∬{u 𝜕t ζ − ∇ϕ∗ ⋅ ∇ζ } dxdt + ∫ u0 ζ (x, 0) dx = 0. (0.30)


Q ℝ

Therefore the couple (u, ν) is a weak Young measure-valued solution of (S).


Motivations | XXIX

If ϕ satisfies assumption (A1 ), then we have the following characterization of the


Young measure ν ([83]; see also [82, 84]):

2
ν(x,t) = ∑ λi (x, t)δsi (v(x,t)) for a. e. (x, t) ∈ Q. (0.31)
i=0

Here the coefficients λi take values in the interval [0, 1], ∑2i=0 λi = 1, and the functions
si (v) are defined by (0.20). By equalities (0.12) and (0.31) we have

2
u(x, t) = ∑ λi (x, t)si (v(x, t)) for a. e. (x, t) ∈ Q. (0.32)
i=0

Remark 0.3.1. In view of (0.32), u is a superposition of different phases. By definition


a two-phase solution of (S) is any solution as above such that λ0 = 0 a. e. in Q and λi = 1
a. e. in Vi (i = 1, 2), where Q = V 1 ∪ V 2 , V1 ∩ V2 = 0 with smooth common boundary
V 1 ∩ V 2.
When N = 1, a major qualitative feature of two-phase solutions is that they display
hysteresis effects: the only admissible phase changes are those that take place from
the extremum points (b, B) and (c, A) of the cubic (see Fig. 1). The proof is analogous
to that of the Oleinik entropy condition for piecewise smooth solutions of hyperbolic
conservation laws (which to some extent are the counterpart of two-phase solutions
of (S); see [48, 54, 70, 71, 82, 83, 84, 98]).
Among others, this feature points out that solutions of (S) obtained by Sobolev
regularization are definitely different from those obtained by Cahn–Hilliard regular-
ization. In fact, it was shown in [12] that the limiting dynamics as κ → 0+ of solutions
of (0.21) on the one-dimensional torus is governed by the Maxwell equal area law:

1 1

∫ ϕ∗∗ (u) du = 0 = ∫ ϕ(u) du,


−1 −1

which is incompatible with hysteresis phenomena (here ϕ∗∗ := (Φ∗∗ )′ , Φ∗∗ being
the convex envelope of the double-well potential Φ). Similar results were obtained by
an implicit variational scheme in [43], concerning an ill-posed problem for a gradient
equation. Problems of the same kind were also investigated by Sobolev and Cahn–
Hilliard regularizations in [30, 95] (see Chapter 8).
XXX | Introduction

3.2 Radon measure-valued solutions


Let ϕ and ψ satisfy (A2 ) and (A3 ), respectively. Since ψ′ (u) → 0 as u → ∞, the initial-
boundary value problem

{
{ 𝜕t u = 𝜕xx [ϕ(u)] + ϵ 𝜕txx [ψ(u)] in Q,
{
{ϕ(u) + ϵ 𝜕t [ψ(u)] = 0 on 𝜕 Ω × (0, T), u ≥ 0, (PMϵ )
{
{
{u = u0 in Ω × {0},

can be guessed to have for ϵ > 0 the same qualitative features displayed as ϵ → 0+ by
the problem

{
{ 𝜕t u = 𝜕xx [ϕ(u)] + ϵ 𝜕txx u in Q,
{
{ ϕ(u) + ϵ 𝜕t u = 0 on 𝜕 Ω × (0, T), u ≥ 0, (Σϵ )
{
{
{u = u 0 in Ω × {0},

with Sobolev regularization, as far as the behavior of singularities is concerned.


Problem (PMϵ ) with u0 ∈ R+f (Ω) was studied in [99, 100]. Its most striking fea-
ture, clearly connected to the weakness of the pseudoparabolic regularization, is the
spontaneous appearance of singularities for positive times when u0 is smooth (see [22,
Theorem 4.2], Definition 1.8.9 and Theorem 3.1.8):

Proposition 0.3.3. Let (A2 )–(A3 ) be satisfied. There exist solutions of (PMϵ ) with u0 ∈
C(Ω), u0 ≥ 0, such that for some t ∈ (0, T), either us (⋅, t) is purely atomic and contains
countably many Dirac masses, or us (⋅, t) is singular continuous.

This result also depends on the negativity of ϕ′ (u) for large values of u, which
gives rise to concentration phenomena. It is in formal agreement with the result in [11]
that solutions of the problem

{
{ 𝜕t z = 𝜕x [ϕ(𝜕x z)] + ϵ 𝜕tx [ψ(𝜕x z)] in Q,
{
{ ϕ(𝜕x z) + ϵ 𝜕t [ψ(𝜕x z)] = 0 in 𝜕Ω × (0, T),
{
{ 1
{z = z0 ∈ C (Ω) in Ω × {0},

may become discontinuous with respect to x for positive times.


Finally, it is interesting to compare the effect of (A1 ) versus (A2 ) when letting
ϵ → 0+ in problem (Σϵ ). If ϕ is a cubic, then every interval [u1 , u2 ] sufficiently large
satisfies condition (0.26), and thus Lemma 0.3.1(ii) gives a uniform L∞ -estimate of
{uϵ }. Therefore, as ϵ → 0+ , we obtain equality (0.30) with u ∈ L∞ (Q), and hence
no singularities arise in this case. On the other hand, if ϕ is of Perona–Malik type,
then only those intervals [u1 , u2 ] ⊆ [0, ∞) where ϕ′ is positive satisfy (0.26). Hence a
uniform L∞ -estimate of {uϵ } only holds if {uϵ } takes values in the stable phase, which
is trivial.
Motivations | XXXI

However, in the latter case, it can be proven that the half-line [0, ∞) is positively
invariant for solutions of (PMϵ ) (see [96]). As a consequence, if u0 ∈ L1 (Ω), u0 ≥ 0,
then the conservation of mass,

∫ uϵ (x, t) dx = ∫ u0 (x) dx (t ∈ (0, T)),


Ω Ω

gives a uniform estimate of {uϵ } in L∞ (0, T; L1 (Ω)):

󵄩󵄩 ϵ ϵ
󵄩󵄩u (⋅, t)󵄩󵄩󵄩L1 (Ω) = ∫ u (x, t) dx = ∫ u0 (x) dx = ‖u0 ‖L1 (Ω) .
󵄩
Ω Ω

It follows that

󵄩󵄩 ϵ 󵄩󵄩
󵄩󵄩u 󵄩󵄩L1 (Q) ≤ T ‖u0 ‖L1 (Ω) for any ϵ > 0.

Then by the biting lemma (see Theorem 0.2.2(iv) and Theorem 5.4.12) there exist a se-
quence {uk } ≡ {uϵk } ⊆ {uϵ }, a Young measure ν ∈ Y(Q; ℝ), a Radon measure σ ∈ R+f (Q),
and a sequence Qk+1 ⊆ Qk ⊆ Q, with Lebesgue measure λN+1 (Qk ) vanishing as ϵk → 0+ ,
such that (possibly up to a subsequence, not relabeled)

uk χQ\Qk ⇀ u∗ := ∫ y dν(x,t) (y) in L1 (Q)


[0,∞)

and

uk χQk ⇀ σ in Rf (Q). (0.33)

As in Proposition 0.3.2, it is also easily checked that there exists v ∈ L∞ (Q) ∩


L2 (0, T; H01 (Ω)) such that


vk ⇀ v in L∞ (Q), vk ⇀ v in L2 (0, T; H01 (Ω)),

where vk := ϕ(uk ) + ϵk 𝜕t uk . Then letting k → ∞ in the weak formulation of (Σϵk ), we


easily obtain

∬(u∗ 𝜕t ζ − ∇v ⋅ ∇ζ ) dxdt + ∫ u0 ζ (x, 0) dx = −⟨σ, 𝜕t ζ ⟩


Q Ω

for any ζ ∈ C 1 ([0, T]; Cc1 (Ω)), ζ (⋅, T) = 0 in Ω.


The right-hand side of the above equality, which arises because of the convergence
in (0.33), accounts for possible concentration phenomena in (Σϵ ) as ϵ → 0+ (already
observed in (PMϵ ) for ϵ > 0 because of the weaker regularization). On the other hand,
as in the case of cubic ϕ, it can be checked that v = ϕ∗ with ϕ∗ (x, t) := ∫ℝ ϕ(y) dν(x,t) (y)
XXXII | Introduction

for a. e. (x, t) ∈ Q. Therefore both Radon and Young measures are needed to describe
the vanishing viscosity limit of (Σϵ ) in the Perona–Malik case.
The above discussion suggests the following picture:
(i) the behavior of ϕ at infinity determines the behavior of the singular part of the
solution with respect to the Lebesgue measure, whereas the changes of mono-
tonicity affect that of the absolutely continuous part;
(ii) the behavior of the singular part is described by Radon measures and is related to
disappearance, persistence, or appearance of singularities (concentration or reg-
ularization phenomena);
(iii) the behavior of the absolutely continuous part is described by Young measures
and is related to oscillation phenomena.

Mathematical tools
Let us draw some conclusions from the above discussion.
Addressing nonlinear differential problems that involve measures always makes
use of some approximation procedure. The solutions of the approximating problems
are a family in some function space, which is naturally embedded in the Banach space
of finite Radon measures. The limiting points of the family in some suitable topology,
as some regularization parameter goes to zero, are natural candidates as solutions of
the original problem.
Therefore the main tool to prove the existence of solutions is provided by results
concerning convergence and compactness of sequences of finite Radon measures in
various topologies. These results fit in (and their proofs make use of) the general
framework of Lebesgue spaces of measure-valued functions defined on some given
Banach space. Indeed, the clarity and elegance of the proofs greatly benefit from a
clear understanding of the functional analytic context.
Applying the general theory to the sequence of solutions of the approximating
problems gives convergence results, which allow us to take the limit in the weak for-
mulation of the approximating problems. Depending on the nature of the problem,
such results account both for concentration and for oscillation phenomena, if any (see
Section 5.4).
In this general framework a major issue is that of regularity of Radon measure-
valued solutions and thus of the possible presence, location, and evolution of singu-
larities. As outlined before, these features are related to suitable capacities which are
characteristic of the problem. Hence the theory of capacities is also an important tool
to study problems of this kind.
Contents and brief user’s guide | XXXIII

Contents and brief user’s guide


Part I of the book is devoted to the general theory. Chapters 1 and 2 have a prelimi-
nary character, presenting general results of abstract measure theory, respectively, of
measurability and integration of scalar functions. Chapters 3, 4, and 5 are the core of
the book, dealing respectively with capacity theory, vector integration, and Lebesgue
spaces of vector functions, and convergence of sequences of finite Radon measures.
In Part II (Chapters 6, 7, and 8), we describe three applications of the general
theory. Each application fits in one of the three research areas (quasilinear parabolic
problems, hyperbolic conservation laws, ill-posed evolution problems) mentioned in
this Introduction.
A description of the content of the chapters is given at the beginning of each part.
Necessary preliminaries on topology are recalled in Appendix A at the end of the book.
Although far from being exhaustive, the bibliography contains ample references to the
different topics treated in the book.
Let us add some practical remarks:
1. By a “scalar function” we mean any real-valued function, as opposite to “vector
function”, a map taking values in any Banach space other than ℝ.
2. All vector spaces considered are real.
3. Within each section theorems, propositions, lemmata and corollaries are labeled
sequentially (e. g., Theorem 1.2.1 is followed by Proposition 1.2.2, etc.), whereas
definitions, remarks, and examples are numbered separately. Equations are num-
bered sequentially within each chapter.
|
Part I: General theory
Outline of Part I
The first chapter provides general results of abstract measure theory, including top-
ics like outer measures and abstract capacity, Carathéodory construction, Hausdorff
measure and capacity, general results of decomposition and vector measures.
In the second chapter, after recalling classical results on measurability and inte-
gration of scalar functions, we introduce two subjects which are fundamental in the
development of the book, Young measures and Riesz representation theorems (both
for positive and for bounded functionals). Various notions of convergence in Lebesgue
spaces are then discussed, pointing out the important role of uniform integrability (Vi-
tali Theorem, Dunford–Pettis Theorem). The final section deals with a major result of
differentiation, the Radon–Nikodým Theorem, and with differentiation of Radon mea-
sures on the Euclidean space ℝN .
In the third chapter first we present some material concerning Sobolev and Bessel
potential spaces, then we recall results concerning functions of bounded variations
and Sobolev functions. Building on these notions we discuss in detail various con-
cepts of capacity on subsets of ℝN (Bessel, Riesz, Sobolev and Hausdorff capacities)
and their mutual relationships. The results concerning Bessel and Riesz capacities
are derived from the Meyer’s theory of capacities associated with a kernel. Measures
concentrated or diffuse with respect to the Sobolev capacity are also discussed and
characterized.
Chapter 4 deals with vector integration. After introducing several concepts of mea-
surability of vector functions and discussing their mutual relationships (in particu-
lar, the Pettis Theorem), we present results concerning both the Bochner integral and
weaker notions of integral. The subsequent step is introducing the vector Lebesgue
spaces Lp (X; Y), Lpw (X; Y) and Lpw∗ (X; Y ∗ ) (where p ∈ [1, ∞], and X, Y Banach spaces),
proving completeness and separability results. Then we discuss the duality theory of
such spaces (in particular, of Lpw∗ (X; Y ∗ ) when Y ∗ or Y is separable), pointing out the
central role of the so-called Radon–Nikodým property, or equivalently of the Riesz rep-
resentability of linear continuous operators from L1 (X) to Y. The application we have
in mind is the case where Y = C0 (Z) and Y ∗ = Rf (Z), the Banach space of finite Radon
N
measures on some metric space Z, since the space L∞ w∗ (0, T; Rf (Ω)) (T > 0, Ω ⊆ ℝ ) is
the natural framework for the evolution problems dealt with in Part II of the book. The
last two sections of the chapter deal with vector Lebesgue spaces where Y is a space
of real functions, and with vector Sobolev spaces, respectively.
Chapter 5 is devoted to convergence of sequences of finite Radon measures.
Whereas boundedness in the Banach space Rf (X) ensures compactness in the weak∗
topology (by the Riesz and Banach–Alaoglu Theorems), compactness in the stronger
narrow topology requires both boundedness and tightness, a concept which plays
a central role in this context. Criteria both for narrow convergence (given by the
Portmanteau Theorem, or using the Prokhorov distance) and for tightness are pro-
vided.

https://doi.org/10.1515/9783110556902-001
4 | General theory

The subsequent step is to prove the Disintegration Theorem for finite Radon mea-
sures on the product of Hausdorff spaces. Applying the Disintegration Theorem to
some sequence {νuj } of Young measures associated with measurable functions uj :
X 󳨃→ ℝ (j ∈ ℕ), it is possible to provide a relation between the boundedness of the
sequence {uj } in Lebesgue spaces and the (relative) compactness of {νuj } with respect
to the narrow topology. In particular, if {uj } is bounded in L1 (X), the sequence {νuj } of
the associated Young measures converges in the narrow sense (possibly up to a sub-
sequence) to some Young measure ν. On the other hand, by the Banach–Alaoglu The-

orem there exist {ujk } ⊆ {uj } and σ ∈ Rf (X) such that ujk ⇀ σ in Rf (X). Then the Biting
Lemma (which is also proven) provides an accurate description of the concentration
phenomena connected with such convergence, as well as a relationship between the
limiting measure σ and the aforementioned Young measure ν.
1 Measure theory
1.1 Preliminaries
We will use the usual notations of set theory. As usual, ℕ, ℤ, ℚ, ℝ, and ℝN (N ∈ ℕ;
ℝ1 ≡ ℝ) denote the natural numbers, integers, rational numbers, real numbers, and
N-tuples of real numbers, respectively. The familiar notations of intervals, (a, b), [a, b],
(a, b], [a, b] will be used also for subintervals of the extended real line ℝ := ℝ ∪ {±∞}.
Occasionally, we use the symbol [−∞, +∞] for ℝ; we also write ∞ instead of +∞.
Let X be a set, and let 𝒫 (X) := {E | E ⊆ X}. For any E ⊆ X, we denote by E c :=
{x ∈ X | x ∈ ̸ E} the complementary set of E in X, by 0 = X c the empty set, and by
χE : X 󳨃→ [0, ∞) the characteristic function of E,

1 if x ∈ E,
χE (x) := {
0 otherwise.

For any E, F ⊆ X, we denote by E\F := E∩F c the difference and by E△F := (E\F)∪(F \E)
the symmetric difference of E and F. If ℱ1 ⊆ ℱ2 ⊆ 𝒫 (X) and f : ℱ2 󳨃→ [−∞, ∞] is any
set function, then the restriction of f to ℱ1 is denoted by f |ℱ1 . For any x ∈ X, the set
{x} ∈ 𝒫 (X) is called a singleton. A cover of E ⊆ X is any nonempty family {Fi }i∈I ⊆ 𝒫 (X)
such that E ⊆ ⋃i∈I Fi . A family {Fi }i∈I ⊆ 𝒫 (X) is disjoint if Fi ∩ Fj = 0 for all i, j ∈ I, i ≠ j.
A finite disjoint family {F1 , . . . , Fn } ⊆ 𝒫 (X) such that X = ⋃ni=1 Fi is called a partition of X.
A sequence {En } ⊆ 𝒫 (X) is nondecreasing (respectively, nonincreasing) if En ⊆ En+1
(En ⊇ En+1 , respectively) for all n ∈ ℕ. Nondecreasing or nonincreasing sequences are
called monotone.

1.2 Families of sets


1.2.1 Measurable spaces

Let us recall the following definitions.

Definition 1.2.1. A nonempty family 𝒮 ⊆ 𝒫 (X) is called a semialgebra if:


(i) E, F ∈ 𝒮 ⇒ E ∩ F ∈ 𝒮 ;
(ii) for any E ∈ 𝒮 , there exists a partition of E c .

A family 𝒜 ⊆ 𝒫 (X) is called an algebra if:


(i) 0 ∈ 𝒜;
(ii) E ∈ 𝒜 ⇒ E c ∈ 𝒜;
(iii) E, F ∈ 𝒜 ⇒ E ∪ F ∈ 𝒜.

https://doi.org/10.1515/9783110556902-002
6 | 1 Measure theory

For any family ℱ ⊆ 𝒫 (X), the intersection 𝒜0 (ℱ ) := ⋂{𝒜 | 𝒜 algebra, 𝒜 ⊇ ℱ } is


an algebra, which contains ℱ and is contained in all algebras 𝒜 ⊇ ℱ . It is called the
algebra generated by ℱ or the minimal algebra containing ℱ .

Example 1.2.1. Let a ≡ (a1 , . . . , an ) and b ≡ (b1 , . . . , bn ) with ai , bi ∈ [−∞, ∞] and ai < bi
for every i = 1, . . . , n (n ∈ ℕ). Consider the N-cell

(a, b] := {x ≡ (x1 , . . . , xn ) | xi ∈ [−∞, ∞], ai < xi ≤ bi ∀ i = 1, . . . , n}

with volume vol((a, b]) := ∏ni=1 (bi − ai ) ≤ ∞. It is easily seen that the family ℐn
of the N-cells is a semialgebra and the algebra 𝒜0 (ℐn ) generated by ℐn consists of
finite disjoint unions of N-cells, that is, for every E ∈ 𝒜0 (ℐn ), there exist disjoint
(a1 , b1 ], . . . , (ap , bp ] such that E = ⋃pk=1 (ak , bk ].

Definition 1.2.2. A family 𝒜 ⊆ 𝒫 (X) is called a σ-algebra if:


(i) 0 ∈ 𝒜;
(ii) E ∈ 𝒜 ⇒ E c ∈ 𝒜;
(iii) for any sequence {Ek } ⊆ 𝒜, we have ⋃∞k=1 Ek ∈ 𝒜.

Definition 1.2.3. Let 𝒜 ⊆ 𝒫 (X) be a σ-algebra. The couple (X, 𝒜) is called a measurable
space, and the elements of 𝒜 are called measurable sets.

Remark 1.2.1. Let F ⊆ X. It is easily seen that if 𝒜 is a σ-algebra, then the family 𝒜∩F :=
{E ∩ F | E ∈ 𝒜} ⊆ 𝒫 (F) is also a σ-algebra, called the trace of 𝒜 on F. The measurable
space (F, 𝒜 ∩ F) is called a measurable subspace of (X, 𝒜).

Definition 1.2.4. A nonempty family ℳ ⊆ 𝒫 (X) is called a monotone class if for any
nondecreasing (respectively, nonincreasing) sequence {Ek } ⊆ ℳ, we have ⋃∞ k=1 Ek ∈
ℳ (respectively, ⋂k=1 Ek ∈ ℳ).

Clearly, every σ-algebra is a monotone class. As in the case of algebras, for any
family ℱ ⊆ 𝒫 (X), we can consider the minimal σ-algebra

σ0 (ℱ ) := ⋂{𝒜 | 𝒜 σ-algebra, 𝒜 ⊇ ℱ }

and the minimal monotone class

ℳ0 (ℱ ) := ⋂{ℳ | ℳ monotone class, ℳ ⊇ ℱ }

generated by ℱ .

Definition 1.2.5. A measurable space (X, 𝒜) is called separable if there exists a count-
able family 𝒮 ⊆ 𝒫 (X) such that 𝒜 = σ0 (𝒮 ).

Theorem 1.2.1. Let 𝒜 ⊆ 𝒫 (X) be an algebra. Then the minimal monotone class ℳ0 (𝒜)
and the minimal σ-algebra σ0 (𝒜) coincide.
1.2 Families of sets | 7

Proof. Clearly, 0 ∈ ℳ0 (𝒜). Let us first show that the monotone class ℳ0 (𝒜) is an al-
gebra. Set 𝒜̃ := {E ∈ ℳ0 (𝒜) | E c ∈ ℳ0 (𝒜)}. Since 𝒜 is an algebra and 𝒜 ⊆ ℳ0 (𝒜), we
have 𝒜 ⊆ 𝒜.̃ On the other hand, it is easily seen that 𝒜̃ is a monotone class, and hence
ℳ0 (𝒜) ⊆ ℳ0 (𝒜)̃ ⊆ 𝒜.̃ Since by definition 𝒜̃ ⊆ ℳ0 (𝒜), it follows that 𝒜̃ = ℳ0 (𝒜).
Therefore, for any E ∈ ℳ0 (𝒜), we have E c ∈ ℳ0 (𝒜). It is similarly seen that for any
E, F ∈ ℳ0 (𝒜), we have E ∪ F ∈ ℳ0 (𝒜). Hence the claim follows.
j
Let {Ek } ⊆ ℳ0 (𝒜) and set Fj := ⋃k=1 Ek (j, k ∈ ℕ). Then the sequence {Fj } is non-
decreasing, and {Fj } ⊆ ℳ0 (𝒜) since ℳ0 (𝒜) is an algebra. Since ℳ0 (𝒜) is a monotone
class, we obtain that ⋃∞k=1 Ek = ⋃j=1 Fj ∈ ℳ0 (𝒜). Therefore ℳ0 (𝒜) is a σ-algebra.

Since ℳ0 (𝒜) is a σ-algebra containing 𝒜, we have ℳ0 (𝒜) ⊇ σ0 (𝒜). On the


other hand, the σ-algebra σ0 (𝒜) is a monotone class containing 𝒜, and thus σ0 (𝒜) ⊇
ℳ0 (𝒜). Hence the conclusion follows.

Let us define the product of two measurable spaces (X1 , 𝒜1 ) and (X2 , 𝒜2 ). Consider
the family of measurable rectangles ℛ ⊆ 𝒫 (X1 × X2 ),

ℛ := {E1 × E2 | E1 ∈ 𝒜1 , E2 ∈ 𝒜2 }.

Definition 1.2.6. Let (X1 , 𝒜1 ) and (X2 , 𝒜2 ) be measurable spaces. The minimal σ-alge-
bra σ0 (ℛ), denoted 𝒜1 × 𝒜2 , is called a product σ-algebra. The measurable space (X1 ×
X2 , 𝒜1 × 𝒜2 ) is called a product measurable space.

Let E ⊆ X1 × X2 . The set

Ex1 := {x2 ∈ X2 | (x1 , x2 ) ∈ E} (x1 ∈ X1 ) (1.1)

is called the x1 -section of E. The x2 -section of E with x2 ∈ X2 is similarly defined. Ob-


serve that
∞ ∞
(E c )x = (Exi )c , ( ⋃ En ) = ⋃ (En )xi (i = 1, 2).
i
n=1 xi n=1

Proposition 1.2.2. Let (X1 , 𝒜1 ) and (X2 , 𝒜2 ) be measurable spaces, and let E ∈ 𝒜1 × 𝒜2 .
Then Ex1 ∈ 𝒜2 for any x1 ∈ X1 , and Ex2 ∈ 𝒜1 for any x2 ∈ X2 .

Proof. Set Ω := {E ∈ 𝒜1 × 𝒜2 | Ex1 ∈ 𝒜2 ∀ x1 ∈ X1 }. It is easily checked that Ω is a


σ-algebra containing the family ℛ of measurable rectangles, and hence Ω = 𝒜1 × 𝒜2 .
It is similarly seen that Ex2 ∈ 𝒜1 for any x2 ∈ X2 , and thus the result follows.

1.2.2 Borel σ-algebras

Definition 1.2.7. Let (X, 𝒯 ) be a topological space. The σ-algebra σ0 (𝒯 ) generated by


the topology 𝒯 (denoted ℬ ≡ ℬ(X) ≡ ℬ(X, 𝒯 )) is called the Borel σ-algebra. Every set
E ∈ ℬ is called a Borel set.
8 | 1 Measure theory

Example 1.2.2. The family 𝒢δ of countable intersections of open sets and the family
ℱσ of countable unions of closed sets are contained in ℬ.

Definition 1.2.8. Let (X, 𝒯 ) be a compact topological space, and let Gδ ⊆ 𝒢δ be the
family of compact sets E ∈ 𝒢δ . The σ-algebra generated by the family Gδ (denoted
ℬa (X)) is called the Baire σ-algebra. Every set E ∈ ℬ is called a Baire set.

Remark 1.2.2. Let X = ℝ, and let 𝒯 = 𝒯 (ℝ) be the real topology. It is easily seen
that the Borel σ-algebra ℬ(ℝ) coincides with the σ-algebra generated by the family
of open bounded intervals. It is generated also by other families of intervals, e. g., by
the family of half-lines {(a, ∞) | a ∈ ℝ}. Similar remarks hold for the Borel σ-algebra
ℬ(ℝN ) (N ≥ 2).

Let us mention the following result, concerning the product of Borel σ-algebras.

Theorem 1.2.3. Let (X1 , 𝒯1 ) and (X2 , 𝒯2 ) be topological spaces with countable bases, and
let (X1 , ×X2 , 𝒯1 × 𝒯2 ) be the product topological space. Then

ℬ(X1 , 𝒯1 ) × ℬ(X2 , 𝒯2 ) = ℬ(X1 × X2 , 𝒯1 × 𝒯2 ).

In particular, ℬ(ℝM ) × ℬ(ℝN ) = ℬ(ℝM+N ) (M, N ∈ ℕ).

1.3 Measures

1.3.1 General properties

Definition 1.3.1. Let 0 ∈ ℱ ⊆ 𝒫 (X). A set function φ : ℱ → [0, ∞] is called:


(i) monotone if φ(E) ≤ φ(F) for any E, F ∈ ℱ such that E ⊆ F;
(ii) additive if φ(⋃nk=1 Ek ) = ∑nk=1 φ(Ek ) for any finite disjoint family {E1 , . . . , En } ⊆ ℱ
such that ⋃nk=1 Ek ∈ ℱ ;
(iii) σ-subadditive if φ(⋃∞ k=1 Ek ) ≤ ∑k=1 φ(Ek ) for any sequence {Ek } ⊆ ℱ such that

⋃k=1 Ek ∈ ℱ ;

(iv) σ-additive if φ(⋃∞ k=1 Ek ) = ∑k=1 φ(Ek ) for any disjoint sequence {Ek } ⊆ ℱ such that

⋃k=1 Ek ∈ ℱ .

Definition 1.3.2. Let (X, 𝒜) be a measurable space. A set function μ : 𝒜 → [0, ∞] is


called a (positive) measure on 𝒜 (or on X) if μ(0) = 0 and μ is σ-additive. The triple
(X, 𝒜, μ) is called a measure space.
A measure μ is called finite if μ(X) < ∞; it is called σ-finite if there exists a se-
quence {Ek } ⊆ 𝒜 such that X = ⋃∞ k=1 Ek and μ(Ek ) < ∞ for all k ∈ ℕ. A measure space
(X, 𝒜, μ) is called finite (respectively, σ-finite) if μ is finite (σ-finite, respectively).
1.3 Measures | 9

If μ(X) = 1, then the measure μ is called a probability measure, and the space
(X, 𝒜, μ) is called a probability space. The set of probability measures on X will be
denoted by P(X).

If ℱ ⊆ 𝒫 (X) is any family of subsets, then the above definition can be generalized
as follows. A map μ : ℱ → [0, ∞] is a measure on ℱ if μ(0) = 0 when 0 ∈ ℱ and for any
disjoint sequence {Ek } ⊆ ℱ such that ⋃∞ k=1 Ek ∈ ℱ , we have μ(⋃k=1 Ek ) = ∑k=1 μ(Ek ).
∞ ∞

A measure μ on ℱ is σ-finite if there exists a sequence {Ek } ⊆ ℱ such that X = ⋃∞ k=1 Ek


and μ(Ek ) < ∞ for every k ∈ ℕ.

Remark 1.3.1. Let (X, 𝒜, μ) be a measure space, and let F ∈ 𝒜. The restriction μ|𝒜∩F of
μ to the trace σ-algebra 𝒜 ∩ F is a measure induced by μ on 𝒜 ∩ F. The measure space
(F, 𝒜 ∩ F, μ|𝒜∩F ) is called a measure subspace of (X, 𝒜, μ).

The following properties of measures are easily proven.

Proposition 1.3.1. Let (X, 𝒜, μ) be a measure space. Then:


(i) μ(⋃nk=1 Ek ) = ∑nk=1 μ(Ek ) for any finite disjoint family {E1 , . . . , En } ⊆ 𝒜;
(ii) E ⊆ F ⇒ μ(E) ≤ μ(F) for any E, F ∈ 𝒜;
(iii) μ(⋃∞k=1 Ek ) ≤ ∑k=1 μ(Ek ) for any sequence {Ek } ⊆ 𝒜;

(iv) μ(⋃∞ k=1 Ek ) = limk→∞ μ(Ek ) for any nondecreasing sequence {Ek } ⊆ 𝒜;
(v) μ(⋂∞ k=1 Ek ) = limk→∞ μ(Ek ) for any nonincreasing sequence {Ek } ⊆ 𝒜 such that
μ(E1 ) < ∞.

Example 1.3.1. (i) Let X be a set. The map μ# : 𝒫 (X) → [0, ∞] defined as

number of elements of E if E is finite,


μ# (E) := {
∞ otherwise

is a measure called the counting measure. Clearly, μ# is σ-finite if X is countable.


(ii) Let X ≠ 0, and let x ∈ X. The map δx : 𝒫 (X) → [0, ∞) defined as

1 if x ∈ E,
δx (E) := {
0 otherwise

is a measure called the Dirac measure concentrated in {x}.

1.3.2 Borel and Radon measures

Let (X, 𝒯 ) be a Hausdorff space, and let ℬ = ℬ(X) be the Borel σ-algebra generated by
the topology 𝒯 . Let 𝒦 ⊆ 𝒫 (X) denote the family of compact subsets.

Definition 1.3.3. Let 𝒜 ⊇ ℬ be a σ-algebra, and let μ : 𝒜 → [0, ∞] be a measure.


10 | 1 Measure theory

(a) μ is called locally finite if for any x ∈ X, there exists a neighborhood U of x such
that μ(U) < ∞. A locally finite measure μ : 𝒜 → [0, ∞] is called a Borel measure
on X.
(b) A set E ∈ 𝒜 is called μ-outer regular (or outer regular) if

μ(E) = inf{μ(A) | A ⊇ E, A ∈ 𝒯 }.

The measure μ is called outer regular if every E ∈ 𝒜 is μ-outer regular.


(c) A set E ∈ 𝒜 is called μ-inner regular (or inner regular) if

μ(E) = sup{μ(K) | K ⊆ E, K ∈ 𝒦}.

The measure μ is called inner regular if every E ∈ 𝒜 is μ-inner regular.


(d) A set E ∈ 𝒜 is called μ-regular (or regular) if it is both μ-outer regular and μ-inner
regular. The measure μ is called regular if every E ∈ 𝒜 is μ-regular.
(e) An inner regular Borel measure is called a Radon measure.
(f) A Borel measure is called moderate if there exists {An } ⊆ 𝒯 such that X = ⋃∞ n=1 An
and μ(An ) < ∞ for all n ∈ ℕ.
(g) The support of a Borel measure μ (denoted by supp μ) is the closed set of points
x ∈ X such that μ(U) > 0 for any neighborhood U of x.

The collections of Borel and Radon measures on X will be denoted by B+ (X) and
R (X), respectively.
+

Remark 1.3.2. Let X, 𝒜, ℬ, and μ be as in Definition 1.3.3.


(i) A set E ∈ 𝒜 is μ-outer regular if and only if for any ϵ > 0, there exists A ∈ 𝒯 such
that A ⊇ E and μ(A \ E) < ϵ. A set E ∈ 𝒜 is μ-inner regular if and only if for any
ϵ > 0, there exists K ∈ 𝒦 such that K ⊆ E and μ(E \ K) < ϵ. If E ∈ 𝒜 is μ-outer
regular, then there exists G ∈ ℬ such that G ⊇ E and μ(G \ E) = 0.
(ii) A σ-compact set E ⊆ X is μ-inner regular. In fact, by definition there exists a non-
decreasing sequence {Kn } ⊆ 𝒦 such that E = ⋃∞ n=1 Kn , and thus E ∈ ℬ . Then by
Proposition 1.3.1(iv) we have

μ(E) = lim μ(Kn ) = sup{μ(K) | K ⊆ E, K ∈ 𝒦}.


n→∞

(iii) If μ is locally finite (in particular, if μ is a Radon measure), then μ(K) < ∞ for any
K ∈ 𝒦. In fact, for any K ∈ 𝒦, there exists A ∈ 𝒯 such that A ⊇ K and μ(A) < ∞,
and thus μ(K) ≤ μ(A) < ∞.
(iv) If the Hausdorff space X is locally compact, then μ is locally finite if and only if
μ(K) < ∞ for any K ∈ 𝒦. In fact, the “if” part of the claim follows from the very
definition of locally compact space, whereas the “only if” part follows from (iii).
In particular, if X is a locally compact Hausdorff space, then μ ∈ B+ (X) if and only
if μ(K) < ∞ for any K ∈ 𝒦.
Another Random Scribd Document
with Unrelated Content
"Well, why have you come back again?"
She looked at him intently: "Guess!"
"I—well, no—perhaps—tell me, Kate, yourself."
"You are different, now, you look different, David."
"Am I changed! Better or worse?"
She did not reply and he continued:
"You, too, are changed, I can't tell how it is, or where, but you are."
"O, I am changed, much changed," murmured Kate.
"Have you been well?"
"Yes."
"And happy?"
"Yes."
"Then how unwise of you to come back."
"I have come back," said Kate, "to be happier. But somehow you are
different."
"You are different, too. Shall we ever be happy again?"
"Why—why not?" said Kate.
"Come on!" he cried hilariously, "let us make a day of it, come
along!"
Out in the streets they wandered until rain began to fall. "Come in
here for a while." They were passing a roomy dull building, the
museum, and they went in together. It was a vast hollow-sounding
flagstone place that had a central brightness fading into dim
recesses and galleries of gloom. They examined a monster skeleton
of something like an elephant, three stuffed apes and a picture of
the dodo. Kate stood before them without interest or amusement,
she just contemplated them. What did she want with an elephant,
an ape, or a dodo! The glass exhibit cases were leaned upon by
them, the pieces of coal neatly arranged and labelled were stared at,
besides the pieces of granite or coloured rock with long names
ending in orite, dorite and sorite and so on to the precious gems
including an imitation, as big as a bun, of a noted diamond. They
leaned over them, repeating the names on the labels with the
quintessence of vacuity. They hated it. There were beetles and
worms of horror, butterflies of beauty, and birds that had been
stuffed for so long that they seemed to be intoxicated; their beaks
fitted them as loosely as a drunkard's hat, their glassy eyes were
pathetically vague. After ascending a flight of stone steps David and
Kate stooped for a long time over a case of sea-anemones that had
been reproduced in gelatine by a German with a fancy for such
things. From the railed balcony they could peer down into the well of
the fusty-smelling museum. No one else was visiting it, they were
alone with all things dead, things that had died millions of years ago
and were yet simulating life. A footfall sounded so harsh in the
corridors, boomed with such clangour, that they took slow diffident
steps, almost tiptoeing, while Kate scarcely spoke at all and he
conversed in murmurs. Whenever he coughed the whole place
seemed to shudder. In the recess, hidden from prying eyes, David
clasped her willing body in his arms. For once she was unshrinking
and returned his fervour. The vastness, the emptiness, the deadness,
worked upon their feelings with intense magic.
"Love me, David," she murmured, and when they moved away from
the gelatinous sea-urchins she kept both her arms clasped around
him as they walked the length of the empty corridor. He could not
perceive her intimations, their meaning was dark to him. She was so
altered, this was another Kate.
"I have come home to make it all up to you," she repeated, and he
scarcely dared to understand her.
They approached a lecture room; the door was open, the room was
empty, they went in and stood near the platform. The place was
arranged like a tiny theatre, tiers of desks rising in half circles on
three sides high up towards the ceiling. A small platform with a
lecturer's desk confronted the rising tiers; on the wall behind it a
large white sheet; a magic lantern on a pedestal was near and a
blackboard on an easel. A pencil of white chalk lay broken on the
floor. Behind the easel was a piano, with a duster on its lid. The
room smelled of spilled acids. The lover's steps upon the wooden
floor echoed louder than ever after their peregrination upon the
flagstones; they were timid of the sound and stood still, close
together, silent. He touched her bosom and pressed her to his heart,
but all her surrender seemed strange and nerveless. She was almost
violently different; he had liked her old rejections, they were fiery
and passionate. He scarce knew what to do, he understood her less
than ever now. Dressed as she was in thick winter clothes it was like
embracing a tree, it tired him. She lay in his arms waiting, waiting,
until he felt almost stifled. Something like the smell of the acids
came from her fur necklet. He was glad when she stood up, but she
was looking at him intently. To cover his uneasiness he went to the
blackboard and picking up a piece of chalk he wrote the first
inconsequent words that came into his mind. Kate stood where he
had left her, staring at the board as he traced the words upon it: We
are but little children weak.
Laughing softly she strolled towards him.
"What do you write that for? I know what it is."
"What it is!—well, what is it?"
She took the chalk from his fingers.
"It's a hymn," she went on "it goes...."
"A hymn!" he cried, "I did not know that."
Underneath the one he had written she was now writing another line
on the board: Nor born to any high estate.
"Of course," he whispered, "I remember it now, I sang it as a child—
at school—go on, go on."
But she had thereupon suddenly turned away, silent, dropping her
hands to her side. One of her old black moods had seized her. He let
her go and picking up another fragment of chalk completed the
verse:

What can we do for Jesu's sake


Who is so high and good and great?

She turned when he had finished and without a word walked loudly
to the piano, fetched the duster and rubbed out the words they had
written on the blackboard. She was glaring at him.
"How absurd you are"—he was annoyed—"let us go out and get
some tea." He wandered off to the door, but she did not follow. He
stood just outside gazing vacantly at a stuffed jay that had an indigo
eye. He looked into the room again. She was there still just as he
had left her; her head bent, her hands hanging clasped before her,
the dimness covering and caressing her—a figure full of sad
thoughts. He ran to her and crushed her in his arms again:
"Kate, my lovely."
She was saying brokenly: "You know what I said. I've come to make
it all up to you. I promised, didn't I?"
Something shuddered in his very soul—too late, too late, this was no
love for him. The magic lantern looked a stupid childish toy, the
smell of the acid was repulsive. Of all they had written upon the
blackboard one word dimly remained: Jesu.
She stirred in his arms. "You are changed, David."
"Changed, yes, everything is changed."
"This is just like a theatre, like a play, as if we were acting."
"Yes, as if we were acting. But we are not acting. Let us go up and
sit in the gallery."
They ascended the steps to the top ring of desks and looked down
to the tiny platform and the white curtain. She sat fondling his
hands, leaning against him.
"Have you ever acted—you would do it so well?"
"Why do you say that? Am I at all histrionic?"
"Does that mean insincere? O no. But you are the person one
expects to be able to do anything."
"Nonsense! I've never acted. I suppose I could. It isn't difficult, you
haven't to be clever, only courageous. I should think it very easy to
be only an ordinary actor, but I'm wrong, no doubt. I thought it was
easy to write—to write a play—until I tried. I once engaged myself
to write a little play for some students to act. I had never done such
a thing before and like other idiots I thought I hadn't ever done it
simply because I hadn't ever wanted to. Heavens, how harassed I
was and how ashamed! I could not do it. I got no further than the
author's speech."
"Well that was something. Tell me it."
"It's nothing to do with the play. It's what the author says to the
audience when the play is finished."
She insisted on hearing it whatever it was. "O well," he said at last.
"Let's do that properly, at least. I'll go down there and deliver it from
the stage. You must pretend that you are the enthusiastic audience.
Come and sit in the stalls."
They went down together.
"Now imagine that this curtain goes up and I suddenly appear."
Kate faintly clapped her hands. He stood upon the platform facing
her and taking off his hat, began:
"Ladies and Gentlemen.
"I am so deeply touched by the warmth of this reception, this utterly
undeserved appreciation, that—forgive me—I have forgotten the
speech I had carefully prepared in anticipation of it. Let me meet my
obligation by telling you a story; I think it is true, I made it up
myself. Once upon a time there was a poor playwright—something
like me—who wrote a play—something like this—and at the end of
the performance the audience, a remarkably handsome well-fed
intellectual audience—something like this—called him before the
curtain and demanded a speech. He protested that he was
unprepared and asked them to allow him to tell them a story—
something like this. Well, that, too, was a remarkably handsome
well-fed intellectual audience, so they didn't mind and he began
again, Once upon a time a poor playwright,—and was just about to
repeat the story I have already twice told you when suddenly,
without a word of warning, without a sound, without a compunction,
the curtain swooped down and chopped him clean in half."
Masterman made an elaborate obeisance and stepped off the
platform.
"Is that all?" asked Kate.
"That's all."
At that moment a loud bell clanged throughout the building
signifying that the museum was about to close.
"Come along!" he cried, but Kate did not move, she still sat in the
stalls.
"Don't leave me, David, I want to hear the play," she said archly.
"There was no play and there is no play. Come, or we shall be locked
in for the night."
Still she sat on. He went to her and seized her hands.
"What does it matter!" she whispered, embracing him. "I want to
make it all up to you."
He was astoundingly moved. She was marvellously changed. If she
hadn't the beauty of perfection she had some of the perfection of
beauty. He adored her.
"But no," he said, "it won't do, it really won't. Come, I have got to
buy you something at once, a ring with a diamond in it, as big as a
bun, an engagement ring, quickly, or the shops will be shut."
He dragged the stammering bewildered girl away, down the stairs
and into the street. The rain had ceased, the sunset sky was bright
and Masterman was intensely happy.
A BROADSHEET BALLAD
At noon the tiler and the mason stepped down from the roof of the
village church which they were repairing and crossed over the road
to the tavern to eat their dinner. It had been a nice little morning but
there were clouds massing in the south; Sam, the tiler, remarked
that it looked like thunder. The two men sat in the dim little taproom
eating, Bob, the mason, at the same time reading from a newspaper
an account of a trial for murder.
"I dunno what thunder looks like," Bob said, "but I reckon this chap
is going to be hung, though I can't rightly say for why. To my
thinking he didn't do it at all: but murder's a bloody thing and
someone ought to suffer for it."
"I don't think," spluttered Sam, as he impaled a flat piece of
beetroot on the point of a pocket-knife and prepared to contemplate
it with patience until his stuffed mouth was ready to receive it, "he
ought to be hung."
"There can be no other end for him though, with a mob of lawyers
like that, and a jury too ... why the rope's half round his neck this
minute; he'll be in glory within a month; they only have three
Sundays, you know, between the sentence and the execution. Well,
hark at that rain then!"
A shower that began as a playful sprinkle grew to a powerful steady
summer downpour. It splashed in the open window and the dim
room grew more dim, and cool.
"Hanging's a dreadful thing," continued Sam, "and 'tis often unjust
I've no doubt, I've no doubt at all."
"Unjust! I tell you ... at majority of trials those who give their
evidence mostly knows nothing at all about the matter; them as
knows a lot—they stays at home and don't budge, not likely!"
"No? But why?"
"Why? They has their reasons. I know that, I knows it for truth ...
hark at that rain, it's made the room feel cold."
They watched the downfall in complete silence for some moments.
"Hanging's a dreadful thing," Sam at length repeated, with almost a
sigh.
"I can tell you a tale about that, Sam, in a minute," said the other.
He began to fill his pipe from Sam's brass box, which was labelled
cough lozenges and smelled of paregoric.
"Just about ten years ago I was working over in Cotswold country. I
remember I'd been into Gloucester one Saturday afternoon and it
rained. I was jogging along home in a carrier's van; I never seen it
rain like that afore, no, nor never afterwards, not like that. Br..r..r..r!
it came down ... bashing! And we came to a cross-roads where
there's a public house called The Wheel of Fortune, very lonely and
onsheltered it is just there. I seed a young woman standing in the
porch awaiting us, but the carrier was wet and tired and angry or
something and wouldn't stop. 'No room'—he bawled out to her—'full
up, can't take you!' and he drove on. 'For the love o' God, mate,'—I
says—pull up and take that young creature! She's ... she's ... can't
you see!' 'But I'm all behind as 'tis', he shouts to me, 'you knows
your gospel, don't you: time and tide wait for no man?' 'Ah, but
dammit all, they always call for a feller,' I says. With that he turned
round and we drove back for the girl. She clumb in and sat on my
knees; I squat on a tub of vinegar, there was nowhere else and I
was right and all, she was going on for a birth. Well, the old van
rattled away for six or seven miles; whenever it stopped you could
hear the rain clattering on the tarpolin, or sounding outside on the
grass as if it was breathing hard, and the old horse steamed and
shivered with it. I had knowed the girl once in a friendly way, a
pretty young creature, but now she was white and sorrowful and
wouldn't say much. By and bye we came to another cross-roads
near a village, and she got out there. 'Good day, my gal,' I says,
affable like, and; 'Thank you, sir,' says she, and off she popped in
the rain with her umbrella up. A rare pretty girl, quite young, I'd met
her before, a girl you could get uncommon fond of, you know, but I
didn't meet her afterwards: she was mixed up in a bad business. It
all happened in the next six months while I was working round those
parts. Everybody knew of it. This girl's name was Edith and she had
a younger sister Agnes. Their father was old Harry Mallerton, kept
The British Oak at North Quainy; he stuttered. Well this Edith had a
love affair with a young chap William, and having a very loving
nature she behaved foolish. Then she couldn't bring the chap up to
the scratch nohow by herself, and of course she was afraid to tell
her mother or father: you know how girls are after being so pesky
natural, they fear, O they do fear! But soon it couldn't be hidden any
longer as she was living at home with them all, so she wrote a letter
to her mother. 'Dear Mother,' she wrote, and told her all about her
trouble.
"By all accounts the mother was angry as an old lion, but Harry took
it calm like and sent for young William, who'd not come at first. He
lived close by in the village, so they went down at last and fetched
him.
"'All right, yes,' he said, 'I'll do what's lawful to be done. There you
are, I can't say no fairer, that I can't.'
"'No,' they said, 'you can't.'
"So he kissed the girl and off he went, promising to call in and settle
affairs in a day or two. The next day Agnes, which was the younger
girl, she also wrote a note to her mother telling her some more
strange news.
"'God above!' the mother cried out, 'can it be true, both of you girls,
my own daughters, and by the same man! O, whatever were you
thinking on, both of ye! Whatever can be done now!'"
"What!" ejaculated Sam, "both on 'em, both on 'em!"
"As true as God's my mercy—both on 'em—same chap. Ah! Mrs.
Mallerton was afraid to tell her husband at first, for old Harry was
the devil born again when he were roused up, so she sent for young
William herself, who'd not come again, of course, not likely. But they
made him come, O yes, when they told the girls' father.
"'Well, may I go to my d..d..d..damnation at once!' roared old Harry
—he stuttered you know—'at once, if that ain't a good one!' So he
took off his coat, he took up a stick, he walked down street to
William and cut him off his legs. Then he beat him till he howled for
his mercy, but you couldn't stop old Harry once he were roused up—
he was the devil born again. They do say as he beat him for a solid
hour; I can't say as to that, but then old Harry picked him up and
carried him off to The British Oak on his own back, and threw him
down in his own kitchen between his own two girls like a dead dog.
They do say that the little one Agnes flew at her father like a raging
cat until he knocked her senseless with a clout over head; rough
man he was."
"Well, a' called for it sure," commented Sam.
"Her did;" agreed Bob, "but she was the quietest known girl for
miles round those parts, very shy and quiet."
"A shady lane breeds mud," said Sam.
"What do you say?—O ah!—mud, yes. But pretty girls both, girls you
could get very fond of, skin like apple bloom, and as like as two
pinks they were. They had to decide which of them William was to
marry."
"Of course, ah!"
"'I'll marry Agnes,' says he."
"'You'll not,' says the old man, 'you'll marry Edie.'"
"'No I won't,' William says, 'it's Agnes I love and I'll be married to
her or I won't be married to e'er of 'em.' All the time Edith sat quiet,
dumb as a shovel, never a word, crying a bit; but they do say the
young one went on like a ... a young ... Jew."
"The Jezebel!" commented Sam.
"You may say it; but wait, my man, just wait. Another cup of beer?
We can't go back to church until this humbugging rain have
stopped."
"No, that we can't."
"It's my belief the 'bugging rain won't stop this side of four o'clock."
"And if the roof don't hold it off it 'ull spoil they Lord's
Commandments that's just done up on the chancel front."
"O, they be dry by now." Bob spoke re-assuringly and then
continued his tale. "'I'll marry Agnes or I won't marry nobody,'
William says, and they couldn't budge him. No, old Harry cracked on
but he wouldn't have it, and at last Harry says: 'It's like this.' He
pulls a half-crown out of his pocket and, 'heads it's Agnes,' he says,
'or tails it's Edith,' he says."
"Never! Ha! ha!" cried Sam.
"Heads it's Agnes, tails it's Edie, so help me God. And it come down
Agnes, yes, heads it was—Agnes—and so there they were."
"And they lived happy ever after?"
"Happy! You don't know your human nature, Sam; wherever was
you brought up? 'Heads it's Agnes,' said old Harry, and at that Agnes
flung her arms round William's neck and was for going off with him
then and there; ha! But this is how it happened about that. William
hadn't any kindred, he was a lodger in the village, and his landlady
wouldn't have him in her house one mortal hour when she heard all
of it; give him the rightabout there and then. He couldn't get
lodgings anywhere else, nobody would have anything to do with
him. So of course, for safety's sake, old Harry had to take him, and
there they all lived together at The British Oak—all in one happy
family. But they girls couldn't bide the sight of each other, so their
father cleaned up an old outhouse in his yard that was used for carts
and hens and put William and his Agnes out in it. And there they
had to bide. They had a couple of chairs, a sofa, and a bed and that
kind of thing, and the young one made it quite snug."
"'Twas a hard thing for that other, that Edie, Bob."
"It was hard, Sam, in a way, and all this was happening just afore I
met her in the carrier's van. She was very sad and solemn then; a
pretty girl, one you could like. Ah, you may choke me, but there they
lived together. Edie never opened her lips to either of them again,
and her father sided with her, too. What was worse, it came out
after the marriage that Agnes was quite free of trouble—it was only
a trumped-up game between her and this William because he
fancied her better than the other one. And they never had no child,
them two, though when poor Edie's mischance come along I be
damned if Agnes weren't fonder of it than its own mother, a jolly
sight more fonder, and William—he fair worshipped it."
"You don't say!"
"I do. 'Twas a rum go, that, and Agnes worshipped it, a fact, can
prove it by scores of people to this day, scores, in them parts.
William and Agnes worshipped it, and Edie—she just looked on, long
of it all, in the same house with them, though she never opened her
lips again to her young sister to the day of her death."
"Ah, she died? Well, it's the only way out of such a tangle, poor
woman."
"You're sympathising with the wrong party." Bob filled his pipe again
from the brass box; he ignited it with deliberation; going to the open
window he spat into a puddle in the road. "The wrong party, Sam;
'twas Agnes that died. She was found on the sofa one morning stone
dead, dead as a adder."
"God bless me," murmured Sam.
"Poisoned," added Bob, puffing serenely.
"Poisoned!"
Bob repeated the word 'poisoned.' "This was the way of it," he
continued; "one morning the mother went out in the yard to collect
her eggs, and she began calling out; 'Edie, Edie, here a minute,
come and look where that hen have laid her egg; I would never
have believed it,' she says. And when Edie went out her mother led
her round the back of the outhouse, and there on the top of a wall
this hen had laid an egg. 'I would never have believed it, Edie,' she
says, 'scooped out a nest there beautiful, ain't she; I wondered
where her was laying. T'other morning the dog brought an egg
round in his mouth and laid it on the doormat. There now, Aggie,
Aggie, here a minute, come and look where the hen have laid that
egg.' And as Aggie didn't answer the mother went in and found her
on the sofa in the outhouse, stone dead."
"How'd they account for it?" asked Sam, after a brief interval.
"That's what brings me to the point about this young feller that's
going to be hung," said Bob, tapping the newspaper that lay upon
the bench. "I don't know what would lie between two young women
in a wrangle of that sort; some would get over it quick, but some
would never sleep soundly any more not for a minute of their mortal
lives. Edie must have been one of that sort. There's people living
there now as could tell a lot if they'd a mind to it. Some knowed all
about it; could tell you the very shop where Edith managed to get
hold of the poison, and could describe to me or to you just how she
administrated it in a glass of barley water. Old Harry knew all about
it, he knew all about everything, but he favoured Edith and he never
budged a word. Clever old chap was Harry, and nothing came out
against Edie at the inquest—nor the trial neither."
"Was there a trial then?"
"There was a kind of a trial. Naturally. A beautiful trial. The police
came and fetched poor William, they took him away and in due
course he was hanged."
"William! But what had he got to do with it?"
"Nothing. It was rough on him, but he hadn't played straight, and so
nobody struck up for him. They made out a case against him—there
was some unlucky bit of evidence which I'll take my oath old Harry
knew something about—and William was done for. Ah, when things
take a turn against you it's as certain as twelve o'clock, when they
take a turn; you get no more chance than a rabbit from a weasel.
It's like dropping your matches into a stream, you needn't waste the
bending of your back to pick them out—they're no good on, they'll
never strike again. And Edith, she sat in court through it all, very
white and trembling and sorrowful, but when the judge put his black
cap on they do say she blushed and looked across at William and
gave a bit of a smile. Well, she had to suffer for his doings, so why
shouldn't he suffer for hers. That's how I look at it...."
"But God-a-mighty...!"
"Yes, God-a-mighty knows. Pretty girls they were, both, and as like
as two pinks."
There was quiet for some moments while the tiler and the mason
emptied their cups of beer. "I think," said Sam then, "the rain's give
over now."
"Ah, that it has," cried Bob. "Let's go and do a bit more on this
'bugging church or she won't be done afore Christmas."
COTTON
At the place where the road from Carnaby Down ends in the main
western highway that goes towards Bath there stands, or once
stood, a strongly built stone cottage confronting, on the opposite
side of the high road, a large barn and some cattle stalls. A man
named Cotton lived with his wife lonely in this place, their whole
horizon bounded by the hedges and fences of their farm. His
Christian name, for some unchristian reason, was Janifex; people
called him Jan, possibly because it rhymed with his wife's name,
which was Ann. And Ann was a robust managing woman of five and
thirty, childless, full of desolating cleanliness and kindly tyrannies,
with no perceptions that were not determined by her domestic
ambition, and no sympathies that could interfere with her diurnal
energies whatever they might be. Jan was a mild husbandman,
prematurely aged, with large teeth and, since 'forty winters had
besieged his brow,' but little hair. Sometimes one of the large teeth
would drop out, leaving terrible gaps when he opened his mouth,
and turning his patient smile to a hideous leer. These evacuations,
which were never restored, began with the death of Queen Victoria;
throughout the reign of her successor great events were punctuated
by similar losses, until at last Jan could masticate, in his staid old
manner, only in one overworked corner of his mouth.
He would rise of a morning throughout the moving year at five of
the clock; having eaten his bread and drunk a mug of cocoa he
would don a long white jacket and cross the road diagonally to the
gate at the eastern corner of the sheds. These were capped by the
bright figure of a golden cockerel, voiceless but useful, flaunting
always to meet the challenge of the wind. Sometimes in his
deliberate way Jan would lift his forlorn eyes in the direction of the
road coming from the east, but he never turned to the other
direction, as that would have cost him a physical effort, and bodily
flexion had ceased years and years ago. Do roads ever run backward
—leaps not forward the eye? As he unloosed the gate of the yard his
great dog would lift its chained head from some sacks under a cart,
and a peacock would stalk out from the belt of pines that partly
encircled the buildings. The man would greet them, saying, "O, ah!"
In the rickyard he would pause to release the fowls from their hut
and watch them run to the stubbles or spurn the chaff with their
claws as they ranged between the stacks. If the day were windy the
chaff would fall back in clouds upon their bustling feathers, and that
delighted his simple mind. It is difficult to account for his joy in this
thing, for though his heart was empty of cruelty it seemed to be
empty of everything else. Then he would pass into the stalls and
with a rattle of can and churn the labour of the day was begun.
Thus he lived, with no temptations, and few desires except perhaps
for milk puddings, which for some reason concealed in Ann's thrifty
bosom he was only occasionally permitted to enjoy. Whenever his
wife thought kindly of him she would give him a piece of silver and
he would traipse a mile in the evening, a mile along to the
Huntsman's Cup, and take a tankard of beer. On his return he would
tell Ann of the things he had seen, the people he had met, and other
events of his journey.
Once, in the time of spring, when buds were bursting along the
hedge coverts and birds of harmony and swiftness had begun to
roost in the wood, a blue-chinned Spaniard came to lodge at the
farm for a few weeks. He was a labourer working at some particular
contract upon the estate adjoining the Cottons' holding, and he was
accommodated with a bed and an abundance of room in a clean loft
behind the house. With curious shoes upon his feet, blazing check
trousers tightly fitting upon his thighs, a wrapper of pink silk around
his neck, he was an astonishing figure in that withdrawn corner of
the world. When the season chilled him a long black cloak with a
hood for his head added a further strangeness. Juan da Costa was
his name. He was slightly round-shouldered with an uncongenial
squint in his eyes; though he used but few words of English his ways
were beguiling. He sang very blithely shrill Spanish songs, and had a
pleasant courtesy of manner that presented a deal of attraction to
the couple, particularly Ann, whose casual heart he reduced in a few
hours to kindness, and in a few days, inexplicably perhaps, to a still
warmer emotion—yes, even in the dull blankness of that mind some
ghostly star could glimmer. From the hour of his arrival she was an
altered woman, although, with primitive subtlety, the transition from
passivity to passion was revealed only by one curious sign, and that
was the spirit of her kindness evoked for the amiable Jan, who now
fared mightily upon his favorite dishes.
Sometimes the Spaniard would follow Jan about the farm. "Grande!"
he would say, gesturing with his arm to indicate the wide-rolling
hills.
"O ah!" Jan would reply, "there's a heap o' land in the open air."
The Spaniard does not understand! He asks: "What?"
"O ah!" Jan would echo.
But it was the cleanly buxom Ann to whom Da Costa devoted
himself. He brought home daily, though not ostensibly to her, a
bunch of the primroses, a stick of snowbudded sallow, or a sprig of
hazel hung with catkins, soft caressible things. He would hold the
hazel up before Ann's uncomprehending gaze and strike the lemon-
coloured powder from the catkins on to the expectant adjacent
buds, minute things with stiff female prongs, red like the eyes of the
white rabbit which Ann kept in the orchard hutch.
One day Juan came home unexpectedly in mid-afternoon. It was a
cold dry day and he wore his black cloak and hood.
"See," he cried, walking up to Ann, who greeted him with a smile;
he held out to her a posy of white violets tied up with some blades
of thick grass. She smelt them but said nothing. He pressed the
violets to his lips and again held them out, this time to her lips. She
took them from him and tucked them into the front of her bodice
while he watched her with delighted eyes.
"You ... give ... me ... somethin ... for ... los flores?"
"Piece a cake!" said Ann, moving towards the pantry door.
"Ah ... cake...!"
As she pulled open the door, still keeping a demure eye upon him,
the violets fell out and down upon the floor, unseen by her. He
rushed towards them with a cry of pain and a torrent of his strange
language; picking them up he followed her into the pantry, a narrow
place almost surrounded by shelves with pots of pickles and jam,
plates, cups and jugs, a scrap of meat upon a trencher, a white bowl
with cob nuts and a pair of iron crackers.
"See..., lost!" he cried shrilly as she turned to him. She was about to
take them again when he stayed her with a whimsical gesture.
"Me ... me," he said, and brushing her eyes with their soft perfume
he unfastened the top button of her bodice while the woman stood
motionless; then the second button, then the third. He turned the
corners inwards and tucked the flowers between her flesh and
underlinen. They stood eyeing one another, breathing uneasily, but
with a pretence at nonchalance. "Ah!" he said suddenly; before she
could stop him he had seized a few nuts from the white bowl and
holding open her bodice where the flowers rested he dropped the
nuts into her warm bosom. "One ... two ... three!"
"Oh...!" screamed Ann mirthfully, shrinking from their tickling, but
immediately she checked her laughter—she heard footsteps. Beating
down the grasping arms of the Spaniard she darted out of the
doorway and shut him in the pantry, just in time to meet Jan coming
into the kitchen bawling for a chain he required.
"What d'ye want?" said Ann.
"That chain for the well-head, gal, it's hanging in the pantry." He
moved to the door.
"'Taint," said Ann barring his way. "It's in the barn, I took it there
yesterday, on the oats it is, you'll find it, I took it over yesterday,
clear off with your dirty boots." She 'hooshed' him off much as she
'hooshed' the hens out of the garden. Immediately he was gone she
pulled open the pantry door and was confronted by the Spaniard
holding a long clasp knife in his raised hand. On seeing her he just
smiled, threw down the knife and took the bewildered woman into
his arms.
"Wait, wait," she whispered, and breaking from him she seized a
chain from a hook and ran out after her husband with it, holding up
a finger of warning to the Spaniard as she brushed past him. She
came back panting, having made some sort of explanation to Jan;
entering the kitchen quietly she found the Spaniard's cloak lying
upon the table; the door of the pantry was shut and he had
apparently gone back there to await her. Ann moved on tiptoe round
the table; picking up the cloak she enveloped herself in it and pulled
the hood over her head. Having glanced with caution through the
front window to the farmyard, she coughed and shuffled her feet on
the flags. The door of the pantry moved slowly open; the piercing
ardour of his glance did not abash her, but her curious appearance in
his cloak moved his shrill laughter. As he approached her she seized
his wrists and drew him to the door that led into the orchard at the
back of the house; she opened it and pushed him out, saying "Go
on, go on." She then locked the door against him. He walked up and
down outside the window making lewd signs to her. He dared not
call out for fear of attracting attention from the farmyard in front of
the house. He stood still, shivered, pretended in dumb show that he
was frozen. She stood at the window in front of him and nestled
provocatively in his cloak. But when he put his lips against the pane
he drew the gleam of her languishing eyes closer and closer to meet
his kiss through the glass. Then she stood up, took off the black
cloak, and putting her hand into her bosom brought out the three
nuts, which she held up to him. She stood there fronting the
Spaniard enticingly, dropped the nuts back into her bosom ... one ...
two ... three ... and then went and opened the door.
In a few weeks the contract was finished, and one bright morning
the Spaniard bade them each farewell. Neither of them knew, so
much was their intercourse restricted, that he was about to depart,
and Ann watched him with perplexity and unhappiness in her eyes.
"Ah, you Cotton, goodbye I say, and you señora, I say goodbye."
With a deep bow he kissed the rough hand of the blushing country-
woman. "Bueno." He turned with his kit bag on his shoulder, waved
them an airy hand and was gone.
On the following Sunday Jan returned from a visit in the evening and
found the house empty; Ann was out, an unusual thing, for their
habits were fixed and deliberate as the stars in the sky. The
sunsetting light was lying in meek patches on the kitchen wall,
turning the polished iron pans to the brightness of silver, reddening
the string of onions, and filling glass jars with solid crystal. He had
just sat down to remove his heavy boots when Ann came in, not at
all the workaday Ann but dressed in her best clothes smelling of
scent and swishing her stiff linen.
"Hullo," said Jan, surprised at his wife's pink face and sparkling eyes,
"bin church?"
"Yes, church," she replied, and sat down in her finery. Her husband
ambled about the room for various purposes and did not notice her
furtive dabbing of her eyes with her handkerchief. Tears from Ann
were inconceivable.
The year moved through its seasons, the lattermath hay was duly
mown, the corn stooked in rows; Ann was with child and the ridge of
her stays was no longer visible behind her plump shoulders. Fruit
dropped from the orchard boughs, the quince was gathered from the
wall, the hunt swept over the field. Christmas came and went, and
then a child was born to the Cottons, a dusky boy, who was shortly
christened Juan.
"He was a kind chap, that man," said Ann, "and we've no relations to
please, and it's like your name—and your name is outlandish!"
Jan's delight now was to sit and muse upon the child as he had ever
mused upon chicken, lambs and calves. "O ah!" he would say,
popping a great finger into the babe's mouth, "O ah!" But when, as
occasionally happened, the babe squinted at him, a singular fancy
would stir in his mind, only to slide away before it could congeal into
the likeness of suspicion.
Snow, when it falls near spring on those Cotswold hills, falls deeply,
and the lot of the husbandmen is hard. Sickness, when it comes,
comes with a flail and in its hobnailed boots. Contagious and
baffling, disease had stricken the district; in mid-March great
numbers of the country folk were sick abed, hospitals were full, and
doctors were harried from one dawn to another. Jan would come in
of an evening and recite the calendar of the day's dooms gathered
from men of the adjacent fields.
"Amos Green 'ave gone then, pore o' chap."
"Pore Amos," the pitying Ann would say, wrapping her babe more
warmly.
"And Buttifant's coachman."
"Dear, dear, what 'ull us all come to?"
"Mrs. Jocelyn was worse 'en bad this morning."
"Never Jan! Us'll miss 'er."
"Ah, and they do say Parson Rudwent won't last out the night."
"And whom's to bury us then?" asked Ann.
The invincible sickness came to the farm. Ann one morning was
weary, sickly, and could not rise from her bed. Jan attended her in
his clumsy way and kept coming in from the snow to give her
comforts and food, but at eve she was in fever and lay helpless in
the bed with the child at her breast. Jan went off for the doctor, not
to the nearest village for he knew that quest to be hopeless, but to a
tiny town high on the wolds two miles away. The moon, large, sharp
and round, blazed in the sky, and its light sparkled upon the rolling
fields of snow; his boots were covered at every muffled step; the
wind sighed in the hedges and he shook himself for warmth. He
came to the hill at last; half-way up was a church, its windows
glowing with warm-looking light and its bells pealing cheerfully. He
passed on and higher up met a priest trotting downwards in black
cassock and saintly hat, his hands tucked into his wide sleeves,
trotting to keep himself warm and humming as he went. Jan asked a
direction of the priest, who gave it with many circumstances of
detail, and after they had parted he could hear the priest's voice call
still further instructions after him as long as he was in sight. "O ah!"
said Jan each time, turning and waving his hand. But after all his
mission was a vain one; the doctor was out and away, it was
improbable that he would be able to come, and the simple man
turned home with a dull heart. When he reached the farm Ann was
delirious but still clung to her dusky child, sleeping snugly at her
bosom. The man sat up all night before the fire waiting vainly for the
doctor, and the next day he himself became ill. And strangely
enough as he worked among his beasts the crude suspicion in his
mind about the child took shape and worked without resistance until
he came to suspect and by easy degrees to apprehend fully the time
and occasion of Ann's duplicity.
"Nasty filthy dirty thing!" he murmured from his sick mind. He was
brushing the dried mud from the hocks of an old bay horse, but it
was not of his horse he was thinking. Later he stood in the rickyard
and stared across the road at the light in their bedroom. Throwing
down the fork with which he had been tossing beds of straw he
shook his fist at the window and cried out: "I hate 'er, I does, nasty
filthy dirty thing!"
When he went into the house he replenished the fire, but found he
could take no further care for himself or the sick woman; he just
stupidly doffed his clothes and in utter misery and recklessness
stretched himself in the bed with Ann. He lay for a long while with
aching brows, a snake-strangled feeling in every limb, an
unquenchable drouth in his throat, and his wife's body burning
beside him. Outside the night was bright, beautiful and still sparkling
with frost; quiet, as if the wind had been wedged tightly in some far
corner of the sky, except for a cracked insulator on the telegraph
pole just near the window, that rattled and hummed with monstrous
uncare. That, and the ticking of the clock! The lighted candle fell
from its sconce on to the mantelpiece; he let it remain and it
flickered out. The glow from the coals was thick upon the ceiling and
whitened the brown ware of the teapot on the untidy hearth. Falling
asleep at last he began dreaming at once, so it seemed, of the shrill
cry of lambs hailing him out of wild snow-covered valleys, so wild
and prolonged were the cries that they woke him, and he knew
himself to be very ill, very ill indeed. The child was wailing piteously,
the room was in darkness, the fire out, but the man did not stir, he
could not care, what could he do with that flame behind his eyes
and the misery of death consuming him? But the child's cries were
unceasing and moved even his numbed mind to some effort. "Ann!"
he gasped. The poor wife did not reply. "Ann!" He put his hand out
to nudge her; in one instant the blood froze in his veins and then
boiled again. Ann was cold, her body hard as a wall, dead ... dead.
Stupor returned upon him; the child, unhelped, cried on, clasped to
that frozen breast until the man again roused himself to effort.
Putting his great hands across the dead wife he dragged the child
from her arms into the warmth beside him, gasping as he did so
"Nasty ... dirty ... thing." It exhausted him, but the child was still
unpacified and again he roused himself and felt for a biscuit on the
table beside the bed. He crushed a piece in his mouth and putting
the soft pap upon his finger fed thus the hungry child until it was
stilled. By now the white counterpane spread vast like a sea, heaving
and rocking with a million waves, the framework of the bedstead
moving like the tackle of tossed ships. He knew there was only one
way to stem that sickening movement. "I hate 'er, I does," rose
again upon his lips, and drawing up his legs that were at once chilly
and streaming with sweat, full of his new hatred he urged with all
his might his wife's cold body to the edge of the bed and withdrew
the bedclothes. Dead Ann toppled and slid from him and her body
clumped upon the floor with a fall that shook the room; the candle
fell from the mantelpiece, bounced upon the teapot and rolled
stupidly along the bare boards under the bed. "Hate 'er!" groaned
the man; he hung swaying above the woman and tried to spit upon
her. He sank back again to the pillow and the child, murmuring "O
ah!" and gathering it clumsily to his breast. He became tranquil then,
and the hollow sounding clock beat a dull rhythm into his mind, until
that sound faded out with all light and sound, and Jan fell into sleep
and died, with the dusky child clasped in his hard dead arms.
POMONA'S BABE
Johnny Flynn was then seventeen years old. At that age you could
not call him 'boy' without vexing him, or 'man' without causing him
to blush—his teasing ruddy and uproarious mother delighted to
produce either or both of these manifestations, for her offspring was
a pale mild creature—but he had given a deal of thought to many
manly questions. Marriage, for instance, was one of these. That was
an institution he admired, but whose joys, whatever they were, he
was not anxious to experience; its difficulties and disasters as
ironically outlined by the widow Flynn were the subject of his
grossest scepticism—scepticism in general being not the least
prominent characteristic of Johnny Flynn.
Certainly his sister Pomona was not married; she was only sixteen,
an age too early for such bliss; but all the same she was going to
have a baby. He had quarrelled with his mother about most things;
she delighted in quarrels, they amused her very much; but on this
occasion she was really very angry, or she pretended to be so—
which was worse, much worse than the real thing.
The Flynns were poor people, quite poor, living in two top-floor
rooms at the house of a shoemaker, also moderately poor, whose
pelting and hammering of soles at evening were a durable grievance
to Johnny. He was fond of the shoemaker, a kind, bulky, tall man of
fifty, though he did not like the shoemaker's wife, as bulky as her
husband, and as tall, but not kind to him or to anything except
Johnny himself; nor did he like any of the other lodgers, of whom
Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!

ebookgate.com

You might also like