Stochastic Optimal Control in Infinite Dimension

Dynamic Programming and HJB Equations 1st

Edition Giorgio Fabbri download

https://textbookfull.com/product/stochastic-optimal-control-in-
infinite-dimension-dynamic-programming-and-hjb-equations-1st-
edition-giorgio-fabbri/

Download more ebook from https://textbookfull.com

We believe these products will be a great fit for you. Click
the link to download now, or visit textbookfull.com
to discover even more!

Adaptive Dynamic Programming with Applications in

Optimal Control 1st Edition Derong Liu

https://textbookfull.com/product/adaptive-dynamic-programming-
with-applications-in-optimal-control-1st-edition-derong-liu/

Stochastic Linear Quadratic Optimal Control Theory

Differential Games and Mean Field Problems Jingrui Sun

https://textbookfull.com/product/stochastic-linear-quadratic-
optimal-control-theory-differential-games-and-mean-field-
problems-jingrui-sun/

Theory of Simple Glasses Exact Solutions in Infinite

Dimensions 1st Edition Giorgio Parisi

https://textbookfull.com/product/theory-of-simple-glasses-exact-
solutions-in-infinite-dimensions-1st-edition-giorgio-parisi/

Optimal Control in Thermal Engineering 1st Edition

Viorel Badescu

https://textbookfull.com/product/optimal-control-in-thermal-
engineering-1st-edition-viorel-badescu/
Applied Stochastic Differential Equations 1st Edition
Simo Särkkä And Arno Solin

https://textbookfull.com/product/applied-stochastic-differential-
equations-1st-edition-simo-sarkka-and-arno-solin/

Semi Infinite Fractional Programming Ram U. Verma

https://textbookfull.com/product/semi-infinite-fractional-
programming-ram-u-verma/

Fractional Dynamic Calculus and Fractional Dynamic

Equations on Time Scales Svetlin G. Georgiev

https://textbookfull.com/product/fractional-dynamic-calculus-and-
fractional-dynamic-equations-on-time-scales-svetlin-g-georgiev/

Reinforcement learning and Optimal Control Draft

version 1st Edition Dmitri Bertsekas

https://textbookfull.com/product/reinforcement-learning-and-
optimal-control-draft-version-1st-edition-dmitri-bertsekas/

Programming Interview Problems: Dynamic Programming

(with solutions in Python) 1st Edition Leonardo Rossi

https://textbookfull.com/product/programming-interview-problems-
dynamic-programming-with-solutions-in-python-1st-edition-
leonardo-rossi/
Probability Theory and Stochastic Modelling 82

Giorgio Fabbri
Fausto Gozzi
Andrzej Święch

Stochastic
Optimal Control
in Infinite
Dimension
Dynamic Programming and
HJB Equations
With a Contribution by Marco Fuhrman
and Gianmario Tessitore
Probability Theory and Stochastic Modelling

Volume 82

Editors-in-chief
Søren Asmussen, Aarhus, Denmark
Peter W. Glynn, Stanford, CA, USA
Yves Le Jan, Orsay, France

Advisory Board
Martin Hairer, Coventry, UK
Peter Jagers, Gothenburg, Sweden
Ioannis Karatzas, New York, NY, USA
Frank P. Kelly, Cambridge, UK
Andreas Kyprianou, Bath, UK
Bernt Øksendal, Oslo, Norway
George Papanicolaou, Stanford, CA, USA
Etienne Pardoux, Marseille, France
Edwin Perkins, Vancouver, BC, Canada
Halil Mete Soner, Zürich, Switzerland
The Probability Theory and Stochastic Modelling series is a merger and
continuation of Springer’s two well established series Stochastic Modelling and
Applied Probability and Probability and Its Applications series. It publishes
research monographs that make a significant contribution to probability theory or an
applications domain in which advanced probability methods are fundamental.
Books in this series are expected to follow rigorous mathematical standards, while
also displaying the expository quality necessary to make them useful and accessible
to advanced students as well as researchers. The series covers all aspects of modern
probability theory including
• Gaussian processes
• Markov processes
• Random fields, point processes and random sets
• Random matrices
• Statistical mechanics and random media
• Stochastic analysis

as well as applications that include (but are not restricted to):

• Branching processes and other models of population growth
• Communications and processing networks
• Computational methods in probability and stochastic processes, including
simulation
• Genetics and other stochastic models in biology and the life sciences
• Information theory, signal processing, and image synthesis
• Mathematical economics and finance
• Statistical methods (e.g. empirical processes, MCMC)
• Statistics for stochastic processes
• Stochastic control
• Stochastic models in operations research and stochastic optimization
• Stochastic models in the physical sciences

More information about this series at http://www.springer.com/series/13205

Giorgio Fabbri Fausto Gozzi
•

Andrzej Święch

Stochastic Optimal Control

in Infinite Dimension
Dynamic Programming and HJB Equations

With a Contribution by
Marco Fuhrman and Gianmario Tessitore

123
Giorgio Fabbri Andrzej Święch
Aix-Marseille School of Economics School of Mathematics
CNRS, Aix-Marseille University, EHESS, Georgia Institute of Technology
Centrale Marseille Atlanta, GA
Marseille USA
France

Fausto Gozzi
Dipartimento di Economia e Finanza
Università LUISS – Guido Carli
Rome
Italy

ISSN 2199-3130 ISSN 2199-3149 (electronic)

Probability Theory and Stochastic Modelling
ISBN 978-3-319-53066-6 ISBN 978-3-319-53067-3 (eBook)
DOI 10.1007/978-3-319-53067-3

Library of Congress Control Number: 2017934613

Mathematics Subject Classification (2010): 49Lxx, 93E20, 49L20, 35R15, 35Q93, 49L25, 65H15,
37L55

© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
To my parents and to Sara
– G.F.
To my parents and to my family
– F.G.
To my parents Franciszka Święch and Jerzy
Święch
– A.Ś.
Preface

The main objective of this book is to give an overview of the theory of Hamilton–
Jacobi–Bellman (HJB) partial differential equations (PDEs) in infinite-dimensional
Hilbert spaces and its applications to stochastic optimal control of
infinite-dimensional processes and related fields. Both areas have developed very
rapidly in the last few decades. While there exist several excellent monographs on
this subject in finite-dimensional spaces (see e.g., [263, 264, 385, 453, 468, 490,
576]), much less has been written in infinite-dimensional spaces. A good account
of the infinite-dimensional case in the deterministic context can be found in [404]
(see also [562] on optimal control of deterministic PDEs). Other books that touch
on the subject are [29, 179, 468]. We attempt to fill this gap in the literature.
Infinite-dimensional diffusion processes appear naturally and are used to model
phenomena in physics, biology, chemistry, economics, mathematical finance,
engineering, and many other areas (see e.g., [124, 177, 180, 372, 569]). This book
investigates the PDE approach to their stochastic optimal control; however,
infinite-dimensional PDEs can also be used to study other properties of such pro-
cesses as large deviations, invariant measures, stochastic viability, stochastic dif-
ferential games for infinite-dimensional diffusions, etc. (see [86, 177, 179, 249, 251,
261, 465, 467, 542, 544]).
To illustrate the main theme of the book, let us begin with a model distributed
parameter stochastic optimal control problem. We want to control a process (called
the state) given by an abstract stochastic differential equation in a real, separable
Hilbert space H


dXðsÞ ¼ ðAXðsÞ þ bðs; XðsÞ; aðsÞÞÞds þ rðs; XðsÞ; aðsÞÞdW ðsÞ; s[t0
XðtÞ ¼ x 2 H;

where A is the generator of a C0 semigroup in H, b; r are some functions, and W

is a so-called Q-Wiener process1 in H. The functions aðÞ, called controls, are

1
Q is a suitable self-adjoint positive operator in H, the covariance operator for W.

vii
viii Preface

stochastic processes with values in some metric space K, which satisfy certain
measurability properties. The above abstract stochastic differential equation is very
general and includes various semilinear stochastic PDEs, as well as other equations
which can be rewritten as stochastic functional evolution equations, for instance,
stochastic differential delay equations. In a most typical optimal control problem we
want to find a control aðÞ, called optimal, which minimizes a cost functional
Z T 
Jðt; x; aðÞÞ ¼ E lðs; XðsÞ; aðsÞÞds þ gðXðTÞÞ
t

(for some T [ t) among all admissible controls for some functions

l : ½0; T  H  K ! R, g : H ! R.
The dynamic programming approach to the above problem is based on studying
the properties of the so-called value function

Vðt; xÞ ¼ inf Jðt; x; aðÞÞ

aðÞ

and characterizing it as a solution of a fully nonlinear PDE, the associated HJB

equation. Since the state XðsÞ evolves in the infinite-dimensional space H, this PDE
is defined in ½0; T  H. The link between the value function V and the HJB
equation is established by the Bellman principle of optimality known as the
dynamic programming principle (DPP),
Z g 
V ðt; xÞ ¼ inf E lðs; X ðsÞ; aðsÞÞds þ V ðg; X ðgÞÞ ; for all g 2 ½t; T:
aðÞ t

Heuristically, the DPP can be used to define a two-parameter nonlinear evolution

system and the associated HJB equation
8 h i
>
< Vt þ hAx; DV i þ inf a2K f12 Tr ðrðt; x; aÞQ2 Þðrðt; x; aÞQ2 Þ D2 V
1 1

> þ hbðt; x; aÞ; DV i þ lðt; x; aÞg ¼ 0;

:
VðT; xÞ ¼ gðxÞ
ð1Þ

is its generating equation. Such a PDE is called infinite-dimensional or a PDE in

infinitely many variables. We also call it unbounded since it has a term with an
unbounded operator A which is well defined only on the domain of A. Other terms
may also be undefined for some values of DV and D2 V, the Fréchet derivatives of
V, which we may identify with elements of H and with bounded, self-adjoint
operators in H respectively. In particular, the term Tr½ðrQ2 ÞðrQ2 Þ D2 V is well
1 1

1 1 
defined only if ðrQ2 ÞðrQ2 Þ D2 V is of trace class.
The main idea is to use the HJB equation to study the properties of the value
function, find conditions for optimality, obtain formulas for synthesis of optimal
Preface ix

feedback controls, etc. This approach turned out to be very successful for
finite-dimensional problems because of its clarity and simplicity and thanks to the
developments of the theory of fully nonlinear elliptic and parabolic PDEs, in par-
ticular the introduction of the notion of a viscosity solution and advances in reg-
ularity theory. However, even there many open questions remain, especially if the
HJB equations are degenerate. We hope the dynamic programming approach will
be equally valuable for infinite-dimensional problems even though a complete
theory is not available yet.
Equation (1) is an example of a fully nonlinear second-order PDE of (degen-
erate) parabolic type. In this book, we will deal with more general and different
versions of such equations and their degenerate elliptic counterparts. If K is a
singleton, (1) is just a terminal value problem for a linear Kolmogorov equation. If
K is not a singleton but the diffusion coefficient r is independent of the control
parameter a, (1) is semilinear. The theory of linear equations (and some special
semilinear equations) has been studied by many authors and can be found in the
books [29, 106, 179, 583]. The emphasis of this book is on semilinear and fully
nonlinear equations.
There are several notions of solution applicable to PDEs in Hilbert spaces which
are discussed in this book: classical solutions, strong solutions, mild solutions in the
space of continuous functions, solutions in L2 ðlÞ, and viscosity solutions. Classical
solutions are the most regular ones. This notion of solution requires C 1;2 regularity
in the Fréchet sense and imposes additional conditions so that all terms in the
equation make sense pointwise for ðt; xÞ 2 ½0; T  H. When classical solutions
exist, we can apply the classical dynamic programming approach to obtain verifi-
cation theorems and the synthesis of optimal feedback controls. Unfortunately, in
almost all interesting cases it is not possible to find such solutions; however, they
are very useful as a theoretical tool in the theory. The notions of strong solutions,
mild solutions in the space of continuous functions, and solutions in L2 ðlÞ are
introduced and studied only for semilinear equations and define solutions which
have at least first derivative (in some suitable sense). Verification theorems and
synthesis of optimal feedback controls can still be developed within their frame-
work. The notion of viscosity solutions is the most general and applies to fully
nonlinear equations; however, at the current stage there are no results on verifi-
cation theorems and synthesis of optimal feedback controls.
Infinite-dimensional problems present unique challenges, and among them are
the lack of local compactness and no equivalent of Lebesgue measure. This means
that standard finite-dimensional elliptic and parabolic techniques which are based
on measure theory cannot be carried over to the infinite-dimensional case.
Moreover, the equations are mostly degenerate and contain unbounded terms which
are singular. So the methods to find regular solutions to PDEs in infinite dimension
like ours tend to be global and are based on semigroup theory, smoothing properties
of transition semigroups (like the Ornstein–Uhlenbeck semigroups), fixed point
techniques, and stochastic analysis. These methods are mostly restricted to equa-
tions of semilinear type. On the other hand, the notion of a viscosity solution is
x Preface

perfectly suited for fully nonlinear equations. It is local, and it does not require any
regularity of solutions except continuity. As in finite dimension, it is based on a
maximum principle through the idea of “differentiation by parts,” i.e., replacing the
nonexisting derivatives of viscosity subsolutions (respectively, supersolutions) by
the derivatives of smooth test functions at points where their graphs touch the
graphs of subsolutions (respectively, supersolutions) from above (respectively,
below). However, as the readers will see, this idea has to be carried out very
carefully in infinite dimension.
This book includes chapters on the most important topics in HJB equations and
the DPP approach to infinite-dimensional stochastic optimal control.
Chapter 1 contains the basic material on infinite-dimensional stochastic calculus
which is needed in subsequent chapters. It is, however, not intended to be an intro-
duction to stochastic calculus, which the reader is expected to have some familiarity
with. Chapter 1 is included to make the book more self-contained. Most of the results
presented there are well known; hence, we only provide references where the reader
can find proofs and more information about concepts, examples, etc. We provide proofs
only in cases where we could not find good references in the literature.
In Chap. 2, we introduce a general stochastic optimal control problem and prove a
key result in the theory, namely the dynamic programming principle. We formulate it
in an abstract and general form so that it can be used in many cases without the need
to prove it again. Solutions of stochastic PDEs must be interpreted in various ways
(strong, mild, variational, etc.), and our formulation of the DPP tries to capture this
phenomenon. Our proof of the DPP is based on standard ideas; however, we have
tried to avoid heavy probabilistic methods regarding weak uniqueness of solutions of
stochastic differential equations. Our proof is thus more analytical.
We also introduce many examples of stochastic optimal control problems which
can be studied in the framework of the approach presented in the book. They should
give the readers an idea of the range and applicability of the material.
Chapter 3 is devoted to the theory of viscosity solutions. The reader should keep
in mind the following principle when it comes to unbounded PDEs in infinite
dimension: There is no single definition of viscosity solutions that applies to all
equations. This is due to the fact that there are many different PDEs which contain
different unbounded operators and terms which are continuous in various norms.
Also the solutions have to be continuous with respect to weaker topologies.
However, the main idea of the notion of viscosity solutions is always the same as
we described before. What changes is the choice of test functions, spaces,
topologies, and the interpretation of various terms in the equation. In this book, we
focus on the notion of a so-called B-continuous viscosity solution which was
introduced by Crandall and Lions in [141, 142] for first-order equations and later
adapted to second-order equations in [539]. The key result in the theory is the
comparison principle, which is very technical. Its main component is the so-called
maximum principle for semicontinuous functions. The proof of such a result in
finite dimension was first obtained in [370] and was later simplified and generalized
in [137–139, 360]. It is heavily based on measure theory and is not applicable to
infinite dimension. Thus, the theory uses a finite-dimensional reduction technique
Preface xi

introduced by Lions in [413]. It restricts the class of equations which can be

considered; in particular, they have to be highly degenerated in the second-order
terms. We present three techniques to obtain the existence of viscosity solutions.
The first and most important for this book is the DPP and the stochastic optimal
control interpretation, showing directly that the value function is a viscosity solu-
tion. This technique applies to HJB equations. The other techniques are
finite-dimensional approximations and Perron’s method. Both can be applied to
more general equations, for instance, Isaacs equations associated to two-player,
zero-sum stochastic differential games; however, they have limitations of their own.
Moreover, we discuss other topics in the theory of viscosity solutions such as
consistency and singular perturbations. Several special equations are also studied in
this book because of their importance and because they are good examples to show
how the definition of viscosity solutions and some techniques can be adjusted to
particular cases. They are the HJB equations for the optimal control of the Duncan–
Mortensen–Zakai equation, stochastic Navier–Stokes equations, and stochastic
boundary control. In particular, the last one also contains ideas on how to handle
HJB equations which may be nondegenerate, for instance, if Q is not of trace class.
Finally, we present applications to the infinite-dimensional Black–Scholes–
Barenblatt equations of mathematical finance.
Chapter 4 is devoted to the theory of mild and strong solutions in spaces of
continuous functions through fixed point techniques based on the smoothing
properties of transition semigroups such as Ornstein–Uhlenbeck semigroups. This
theory applies only to semilinear equations, i.e., when the coefficient r does not
depend on the control parameter a, and historically it was the first approach
introduced in the literature. The theory was initiated by Barbu and Da Prato [29]
and later improved and developed in various papers, see e.g., [89, 90, 105, 107,
302, 307, 308, 311].
Chapter 4 is divided into four main parts. In the first part (Sects. 4.2 and 4.3), we
present the basic tools needed for the analysis: the theory of generalized gradients
and the smoothing of transition semigroups. In the second part (Sects. 4.4–4.7), we
develop the theory for a general type of semilinear HJB equation (parabolic and
elliptic) without connection with optimal control problems. The main idea behind
this approach is the following. Consider the HJB equation (1) in the semilinear case
when the coefficient r is time-independent:


Vt þ AV þ inf a2K fhbðt; x; aÞ; DV i þ lðt; x; aÞg ¼ 0;
ð2Þ
VðT; xÞ ¼ gðxÞ;

where A is the linear operator

1 h i
Au ¼ hAx; Dui þ Tr ðrðxÞQ2 ÞðrðxÞQ2 Þ D2 u :
1 1

2
xii Preface

If such an operator generates a semigroup etA then, by the variation of constants

formula, one can rewrite Eq. (2) in the integral form as
Z T 
Vðt; xÞ ¼ eðTtÞA gðxÞ þ eðTsÞA Fðs; Þ ðxÞ s;
t

where Fðs; xÞ :¼ inf a2K fhbðs; x; aÞ; DV i þ lðs; x; aÞg. The solution of this integral
equation is called a mild solution and is obtained by fixed point techniques. To
define it, the solution must at least have a first-order spatial Gâteaux derivative,
possibly only in some directions needed to give sense to the nonlinear term, the
so-called G-derivative. Thus, one needs suitable smoothing properties of the
semigroup etA (which is the Ornstein–Uhlenbeck semigroup in the simplest case).
Since this semigroup is not strongly continuous, except in very special cases, one
needs to use the theory of p-semigroups introduced in [493] or that of weakly
continuous (or K-continuous) semigroups [101, 108, 301]. Sects. 4.4 and 4.5
consider a general type of operator A, possibly depending on t, while Sects. 4.6 and
4.7 focus on the case when A is of Ornstein–Uhlenbeck type, where stronger results
can be proved.
In the third part (Sect. 4.8), we develop a connection with stochastic optimal
control problems. The fact that mild solutions have a first-order spatial derivative
allows us to give a meaning to formulae for optimal feedbacks. However, the proofs
of the verification theorems and optimal feedback formulae cannot be done
straightforwardly as one needs to apply Itô’s formula in infinite dimension, which
requires smooth functions. For this reason (following [307]), we introduce the
notion of a strong solution of the HJB equation (2) as a suitable limit of classical
solutions and prove that any mild solution is also a strong solution.
The fourth and last part of the chapter (Sects. 4.9 and 4.10) deals with some
special equations. In Sect. 4.9, we show how the techniques developed in the
previous sections can be adapted to HJB equations and analysis of optimal control
problems for the stochastic Burgers equation, stochastic Navier–Stokes equations
and stochastic reaction diffusion equations. In Sect. 4.10, we discuss some equa-
tions for which explicit representations of the solutions can be found. Such cases
are always of interest in applications.
Chapter 5 is devoted to a relatively new and promising theory of mild and strong
solutions in spaces of L2 functions with respect to a suitable measure l (see [3, 4,
125, 299]). The contents of this chapter are similar to the previous one as the main
ideas behind the definition of mild and strong solutions of HJB equations are the
same. The difference is in the fact that the reference space is not the space of
continuous functions but the space of square-integrable functions with respect to the
measure l. The results are similar: existence and uniqueness of solutions of HJB
equations through fixed point arguments, verification theorem through approxi-
mations, and existence of optimal feedbacks. The advantage of this approach is that
the results require weaker assumptions on the data, thus enlarging the range of
possible applications, including the control of delay equations; however, at a cost of
Preface xiii

weaker statements, for example, the first-order spatial derivative is now defined in a
Sobolev weak sense and is not in general a Gâteaux or Fréchet derivative. The main
tools used here are the theory of invariant measures for infinite-dimensional
stochastic differential equations and the properties of transition semigroups in the
space of integrable functions with respect to such measures.
Chapter 6 is devoted to a different and in many respects complementary tech-
nique of Backward Stochastic Differential Equations (BSDEs). The chapter was
written independently and autonomously by M. Fuhrman and G. Tessitore, who are
well-recognized experts in the field. We are grateful for their invaluable contribu-
tion. BSDEs are Itô type equations in which the initial condition is replaced by a
final condition and a new unknown process appears corresponding to a suitable
martingale term. In the nonlinear, finite-dimensional case BSDEs were introduced
in [476] while their direct connection with optimal stochastic control was first
investigated in [212] and [483]. Since then, the general theory of BSDEs has
developed considerably, see [78, 80, 210, 378, 421, 475]. Besides stochastic con-
trol, applications were given to many fields, for instance, to optimal stopping,
stochastic differential games, nonlinear partial differential equations and many
topics related to mathematical finance. Infinite-dimensional BSDEs have also been
considered, see for instance, [130, 285, 331, 351, 477]. The interest for us is that
BSDEs provide an alternative way to represent the value function of an optimal
control problem and consequently to study the corresponding HJB equation and to
solve the control problem. It turns out that the most suitable notion of solution for
the HJB equation is, in this context, that of a mild solution on spaces of continuous
functions but, unlike in Chap. 4, the BSDE method seems particularly adapted to
treating degenerate cases in which the transition semigroup has no smoothing
properties. The price to pay is that normally we need more regular coefficients and a
structural condition (imposing, roughly speaking, that the control acts within the
image of the noise). If these requirements are satisfied, the BSDE techniques are
revealed to be very flexible. In particular, in Chap. 6 we will show how they allow
us to treat both parabolic and elliptic HJB equations (see [77, 286, 352, 436, 478]).
The parabolic case is treated for nonconstant diffusion and Lipschitz nonlinearity,
while the elliptic case is considered for a constant diffusion operator with locally
Lipschitz (with respect to the gradient) nonlinearity and a mild dissipativity
assumption (with respect to the solution). We also report (without proofs) the
results of [286] concerning elliptic HJB equations with nonconstant diffusion, a
globally Lipschitz Hamiltonian and strong dissipativity. A detailed discussion of the
literature on BSDEs in infinite dimension is contained in the bibliographical notes
of Chap. 6.
It is impossible to cover all aspects of the theory of HJB equations in infinite
dimension and its connections to stochastic optimal control. In particular, the theory
of integro-PDEs is an emerging area which is not presented in the book. We do not
discuss first-order equations and extensions to Banach spaces. Equations in the
space of probability measures is another emerging topic. We have chosen a
selection of topics which give a broad overview of the field and enough information
so that the readers can start exploring the subject on their own. There are already
xiv Preface

enough important applications to justify the interest in the subject. The readers
should not be restricted to the boundaries drawn by the book. We hope that this
book will spur interest and research in the field among theoretical and applied
mathematicians, and that it will be useful to all kinds of scientists and researchers
working in areas related to stochastic control.
Suggestions for reading. The readers who are familiar with probability and
stochastic analysis in infinite dimension can skip Chap. 1 and go directly to Chap. 2.
Chapter 2 is needed for the understanding of the other chapters; however, some
material in Sect. 2.3 related to technical details of the proof of the dynamic pro-
gramming principle can be omitted during the first reading. Chaps. 3–6 are to a large
extent independent of each other, and hence the reader can pass from Chap. 2
directly to any of them.

Marseille, France Giorgio Fabbri

Rome, Italy Fausto Gozzi
Atlanta, GA, USA Andrzej Święch
Acknowledgements

The writing of this book was a daunting task which took several years to complete.
We greatly benefited from comments, remarks, and advice from many people who
read parts of the manuscript or provided useful suggestions regarding the book.
Their input improved the content of this book and the presentation of the material
and reduced the number of mistakes and errors. The list, in alphabetical order,
includes Elena Bandini, Daniel Bauer, Enrico Biffis, Sandra Cerrai, Andrea Cosso,
Giuseppe Da Prato, Cristina Di Girolami, Salvatore Federico, Ben Goldys, Carlo
Marinelli, Federica Masiero, Chenchen Mou, Mauro Rosestolato, Nizar Touzi, and
Jerzy Zabczyk. We thank all of them for their valuable help.
G. Fabbri wishes to express his gratitude to his wife Sara for her constant
understanding and encouragement.
F. Gozzi is grateful to his family (Enrica, Matteo, and Marta) who supported him
in this long work and to all his friends who encouraged him to accomplish this
book. He also expresses special thanks to G. Da Prato, for introducing him to the
theory of HJB equations in infinite dimension, for constant encouragement in this
work, and also for reading part of the manuscript.
A. Święch would also like to express his gratitude to M.G. Crandall who
introduced him to viscosity solutions and PDEs in infinite-dimensional spaces and
who greatly influenced his mathematical career.
M. Fuhrman and G. Tessitore would like to thank G. Da Prato and J. Zabczyk for
introducing them to stochastic analysis and for their constant help and support.
Finally, we are grateful to Boris Rozovski, the former editor of the series, for his
support and Marina Reizakis and the Springer production team for their patience
with us and for their very professional handling of the project.

xv
Contents

1 Preliminaries on Stochastic Calculus in Infinite Dimension . . . . . . . . 1

1.1 Basic Probability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Probability Spaces, r-Fields . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.3 The Bochner Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.1.4 Expectation, Covariance and Correlation . . . . . . . . . . . . 12
1.1.5 Conditional Expectation and Conditional Probability . . . 13
1.1.6 Gaussian Measures on Hilbert Spaces
and the Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . 17
1.2 Stochastic Processes and Brownian Motion . . . . . . . . . . . . . . . . . 22
1.2.1 Stochastic Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.2.2 Martingales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.2.3 Stopping Times . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.2.4 Q-Wiener Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.2.5 Simple and Elementary Processes . . . . . . . . . . . . . . . . . . 32
1.3 The Stochastic Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
1.3.1 Definition of the Stochastic Integral . . . . . . . . . . . . . . . . 35
1.3.2 Basic Properties and Estimates . . . . . . . . . . . . . . . . . . . . 39
1.4 Stochastic Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . 45
1.4.1 Mild and Strong Solutions . . . . . . . . . . . . . . . . . . . . . . . 45
1.4.2 Existence and Uniqueness of Solutions . . . . . . . . . . . . . . 48
1.4.3 Properties of Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . 50
1.4.4 Uniqueness in Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
1.5 Further Existence and Uniqueness Results in Special Cases . . . . 57
1.5.1 SDEs Coming from Boundary Control Problems . . . . . . 58
1.5.2 Semilinear SDEs with Additive Noise . . . . . . . . . . . . . . 63
1.5.3 Semilinear SDEs with Multiplicative Noise . . . . . . . . . . 67
1.6 Transition Semigroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
1.7 Itô’s and Dynkin’s Formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

xvii
xviii Contents

1.8 Bibliographical Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

2 Optimal Control Problems and Examples. . . . . . . . . . . . . . . . . . . . . . 91
2.1 Stochastic Optimal Control Problems: General Formulation. . . . . 92
2.1.1 Strong Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
2.1.2 Weak Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
2.2 The Dynamic Programming Principle: Setup
and Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 97
2.2.1 The Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 97
2.2.2 The General Assumptions . . . . . . . . . . . . . . . . . . . . . . .. 99
2.2.3 The Assumptions in the Case of Control Problems
for Mild Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 103
2.3 The Dynamic Programming Principle: Statement and Proof. . . .. 106
2.3.1 Pullback to the Canonical Reference
Probability Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
2.3.2 Independence of Reference Probability Spaces . . . . . . . . 108
2.3.3 The Proof of the Abstract Principle of Optimality . . . . . 109
2.4 Infinite Horizon Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
2.5 The HJB Equation and Optimal Synthesis
in the Smooth Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 119
2.5.1 The Finite Horizon Problem: Parabolic
HJB Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 120
2.5.2 The Infinite Horizon Problem: Elliptic HJB Equation . .. 127
2.6 Some Motivating Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 131
2.6.1 Stochastic Controlled Heat Equation: Distributed
Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 133
2.6.2 Stochastic Controlled Heat Equation: Boundary
Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 137
2.6.3 Stochastic Controlled Heat Equation: Boundary
Control and Boundary Noise . . . . . . . . . . . . . . . . . . . .. 142
2.6.4 Optimal Control of the Stochastic Burgers Equation . . .. 145
2.6.5 Optimal Control of the Stochastic Navier–Stokes
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 148
2.6.6 Optimal Control of the Duncan–Mortensen–Zakai
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
2.6.7 Super-Hedging of Forward Rates . . . . . . . . . . . . . . . . . . 158
2.6.8 Optimal Control of Stochastic Delay Equations . . . . . . . 161
2.7 Bibliographical Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
3 Viscosity Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
3.1 Preliminary Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
3.1.1 B-Continuity and Weak and Strong B-Conditions . . . . . . 172
Contents xix

3.1.2 Estimates for Solutions of Stochastic Differential

Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
3.1.3 Perturbed Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . 188
3.2 A Maximum Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
3.3 Viscosity Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
3.3.1 Bounded Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
3.4 Consistency of Viscosity Solutions . . . . . . . . . . . . . . . . . . . . . . . 202
3.5 Comparison Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
3.5.1 Degenerate Parabolic Equations . . . . . . . . . . . . . . . . . . . 206
3.5.2 Degenerate Elliptic Equations . . . . . . . . . . . . . . . . . . . . . 216
3.6 Existence of Solutions: Value Function . . . . . . . . . . . . . . . . . . . . 222
3.6.1 Finite Horizon Problem. . . . . . . . . . . . . . . . . . . . . . . . . . 225
3.6.2 Improved Version of the Dynamic Programming
Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
3.6.3 The Infinite Horizon Problem . . . . . . . . . . . . . . . . . . . . . 245
3.7 Existence of Solutions: Finite-Dimensional Approximations . . . . 249
3.8 Singular Perturbations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
3.9 Perron’s Method and Half-Relaxed Limits . . . . . . . . . . . . . . . . . . 272
3.10 The Infinite-Dimensional Black–Scholes–Barenblatt Equation . . . 283
3.11 The HJB Equation for Control of the Duncan–
Mortensen–Zakai Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 286
3.11.1 Variational Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . .. 287
3.11.2 Weighted Sobolev Spaces . . . . . . . . . . . . . . . . . . . . . . .. 292
3.11.3 Optimal Control of the Duncan–Mortensen–Zakai
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
3.11.4 Estimates for the DMZ Equation . . . . . . . . . . . . . . . . . . 296
3.11.5 Viscosity Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
3.11.6 The Value Function and Existence of Solutions . . . . . . . 307
3.12 HJB Equations for Boundary Control Problems . . . . . . . . . . . . . . 310
3.12.1 Definition of a Viscosity Solution . . . . . . . . . . . . . . . . . . 311
3.12.2 Comparison and Existence Theorem . . . . . . . . . . . . . . . . 312
3.12.3 A Stochastic Control Problem . . . . . . . . . . . . . . . . . . . . . 320
3.13 HJB Equations for Control of Stochastic Navier–Stokes
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
3.13.1 Estimates for Controlled SNS Equations . . . . . . . . . . . . . 335
3.13.2 The Value Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
3.13.3 Viscosity Solutions and the Comparison Theorem . . . . . 344
3.13.4 Existence of Viscosity Solutions . . . . . . . . . . . . . . . . . . . 352
3.14 Bibliographical Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
4 Mild Solutions in Spaces of Continuous Functions . . . . . . . . . . . . . . . 367
4.1 The Setting and an Introduction to the Methods . . . . . . . . . . . . . 368
4.1.1 The Method in the Parabolic Case . . . . . . . . . . . . . . . . . 369
4.1.2 The Method in the Elliptic Case . . . . . . . . . . . . . . . . . . . 372
xx Contents

4.2 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373

4.2.1 G-Derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
4.2.2 Weighted Spaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
4.3 Smoothing Properties of Transition Semigroups. . . . . . . . . . . . . . 392
4.3.1 The Case of the Ornstein–Uhlenbeck Semigroup . . . . . . 393
4.3.2 The Case of a Perturbed Ornstein–Uhlenbeck
Semigroup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
4.3.3 The Case of an Invertible Diffusion Coefficient . . . . . . . 423
4.4 Mild Solutions of HJB Equations . . . . . . . . . . . . . . . . . . . . . . . . 431
4.4.1 The Parabolic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431
4.4.2 The Elliptic Case with a Big Discount Factor . . . . . . . . . 452
4.5 Approximation of Mild Solutions: Strong Solutions. . . . . . . . . . . 466
4.5.1 The Parabolic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
4.5.2 The Elliptic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
4.6 HJB Equations of Ornstein–Uhlenbeck Type: Lipschitz
Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 482
4.6.1 The Parabolic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 487
4.6.2 The Elliptic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 497
4.7 HJB Equations of Ornstein–Uhlenbeck Type: Locally
Lipschitz Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 514
4.7.1 The Parabolic Case . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 514
4.8 Stochastic Control: Verification Theorems and Optimal
Feedbacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
4.8.1 The Finite Horizon Case . . . . . . . . . . . . . . . . . . . . . . . . . 529
4.8.2 The Infinite Horizon Case . . . . . . . . . . . . . . . . . . . . . . . . 552
4.8.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
4.9 Mild Solutions of HJB for Two Special Problems . . . . . . . . . . . . 569
4.9.1 Control of Stochastic Burgers and Navier–Stokes
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
4.9.2 Control of Reaction-Diffusion Equations . . . . . . . . . . . . 580
4.10 Regular Solutions Through “Explicit” Representations . . . . . . . . 589
4.10.1 Quadratic Hamiltonians. . . . . . . . . . . . . . . . . . . . . . . . . . 589
4.10.2 Explicit Solutions in a Homogeneous Case. . . . . . . . . . . 591
4.11 Bibliographical Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
4.11.1 The First Papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597
4.11.2 Development of the Method . . . . . . . . . . . . . . . . . . . . . . 598
4.11.3 Beyond the Ornstein–Uhlenbeck Semigroup . . . . . . . . . . 599
4.11.4 Explicit Solutions of HJB Equations . . . . . . . . . . . . . . . . 602
4.11.5 The Results and the Proofs of This Chapter
Compared with the Literature . . . . . . . . . . . . . . . . . . . .. 602
Contents xxi

5 Mild Solutions in L2 Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605

5.1 Introduction to the Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
5.2 Preliminaries and the Linear Problem . . . . . . . . . . . . . . . . . . . . . 609
5.2.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
5.2.2 The Reference Measure m and the Main
Assumptions on the Linear Part . . . . . . . . . . . . . . . . . .. 609
5.2.3 The Operator A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 616
5.2.4 The Gradient Operator DQ and the Space
WQ1;2 ðH; mÞ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
5.2.5 The Operator R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
5.2.6 Two Key Lemmas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
5.3 The HJB Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
5.4 Approximation of Mild Solutions. . . . . . . . . . . . . . . . . . . . . . . . . 640
5.5 Application to Stochastic Optimal Control . . . . . . . . . . . . . . . . . . 646
5.5.1 The State Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
5.5.2 The Optimal Control Problem and the HJB Equation . . . 648
5.5.3 The Verification Theorem . . . . . . . . . . . . . . . . . . . . . . . . 649
5.5.4 Optimal Feedbacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
5.5.5 Continuity of the Value Function
and Non-degeneracy of the Invariant Measure . . . . . . . . 662
5.6 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665
5.6.1 Optimal Control of Delay Equations . . . . . . . . . . . . . . . . 665
5.6.2 Control of Stochastic PDEs of First Order . . . . . . . . . . . 666
5.6.3 Second-Order SPDEs in the Whole Space . . . . . . . . . . . 668
5.7 Results in Special Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 669
5.7.1 Parabolic HJB Equations . . . . . . . . . . . . . . . . . . . . . . . . 669
5.7.2 Applications to Finite Horizon Optimal Control
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 673
5.7.3 Elliptic HJB Equations . . . . . . . . . . . . . . . . . . . . . . . . .. 676
5.8 Bibliographical Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 680
6 HJB Equations Through Backward Stochastic Differential
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 685
6.1 Complements on Forward Equations with Multiplicative
Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 686
6.1.1 Notation on Vector Spaces and Stochastic Processes . .. 686
6.1.2 The Class G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 688
6.1.3 The Forward Equation: Existence, Uniqueness
and Regularity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 692
6.2 Regular Dependence on Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
6.2.1 Differentiability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695
6.2.2 Differentiability in the Sense of Malliavin . . . . . . . . . . . 698
xxii Contents

6.3 Backward Stochastic Differential Equations (BSDEs) . . . . . . . . . 711

6.3.1 Well-Posedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711
6.3.2 Regular Dependence on Data . . . . . . . . . . . . . . . . . . . . . 719
6.3.3 Forward–Backward Systems . . . . . . . . . . . . . . . . . . . . . . 727
6.4 BSDEs and Mild Solutions to HJB . . . . . . . . . . . . . . . . . . . . . . . 731
6.5 Applications to Optimal Control Problems . . . . . . . . . . . . . . . . . . 736
6.6 Application: Controlled Stochastic Equation with Delay . . . . . . . 742
6.7 Elliptic HJB Equation with Arbitrarily Growing Hamiltonian . . . 746
6.8 The Associated Forward–Backward System . . . . . . . . . . . . . . . . . 747
6.8.1 Differentiability of the BSDE and a Priori Estimate
on the Gradient . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 753
6.9 Mild Solution of the Elliptic PDE . . . . . . . . . . . . . . . . . . . . . . . . 757
6.10 Application to Optimal Control in an Infinite Horizon . . . . . . . . . 762
6.11 Application: The Heat Equation with Additive Noise . . . . . . . . . 767
6.12 Elliptic HJB Equations with Non-constant Diffusion . . . . . . . . . . 770
6.12.1 The Heat Equation with Multiplicative Noise . . . . . . . . . 778
6.13 Bibliographical Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 779
Appendix A: Notation and Function Spaces . . . . . . . . . . . . . . . . . . . . . . . 783
Appendix B: Linear Operators and C0 -Semigroups. . . . . . . . . . . . . . . . . 793
Appendix C: Parabolic Equations with Non-homogeneous
Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 843
Appendix D: Functions, Derivatives and Approximations . . . . . . . . . . . . 855
Appendix E: Viscosity Solutions in RN . . . . . . . . . . . . . . . . . . . . . . . . . . . 867
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 875
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 901
Notation Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 911
About the Authors

Giorgio Fabbri is a CNRS Researcher at the Aix-Marseille School of Economics,

Marseille, France. He works on optimal control of deterministic and stochastic
systems, notably in infinite dimensions, with applications to economics. He has also
published various papers in several economic areas, in particular in growth theory
and development economics.
Fausto Gozzi is a Full Professor of Mathematics for Economics and Finance at
Luiss University, Roma, Italy. His main research field is the optimal control of finite
and infinite-dimensional systems and its economic and financial applications. He is
the author of many papers in various areas, from Mathematics, to Economics and
Finance.
Andrzej Święch is a Full Professor at the School of Mathematics, Georgia Institute
of Technology, Atlanta, USA. His main research interests are in nonlinear PDEs
and integro-PDEs, PDEs in infinite-dimensional spaces, viscosity solutions,
stochastic and deterministic optimal control, stochastic PDEs, differential games,
mean-field games, and calculus of variations.

About the Contributors

Marco Fuhrman is a Full Professor of Probability and Mathematical Statistics at
the University of Milano, Italy. His main research topics are stochastic differential
equations in infinite dimensions and backward stochastic differential equations for
optimal control of stochastic processes.

Gianmario Tessitore is a Full Professor of Probability and Mathematical Statistics

at Milano-Bicocca University. He is the author of several scientific papers on
control of stochastic differential equations in finite and infinite dimensions. He is, in
particular, interested in the applications of backward stochastic differential equa-
tions in stochastic control.

xxiii
Chapter 1
Preliminaries on Stochastic Calculus
in Infinite Dimension

1.1 Basic Probability

We recall some basic notions of measure theory and give a short introduction to
random variables and the theory of the Bochner integral.

1.1.1 Probability Spaces, σ-Fields

Definition 1.1 (π-system, σ-field) Consider a set  and denote by P() the power
set of .

(i) A non-empty class of subsets of , F ⊂ P(), is called a π-system if it is

closed under finite intersections.
(ii) A class of subsets of , F ⊂ P(), is called a σ-field in  if  ∈ F and F
is closed under complements and countable unions.
(iii) A class of subsets of , F ⊂ P(), is called a λ-system if:
•  ∈ F;
• if A, B ∈ F , A ⊂ B, then B \ A ∈ F ;
• if Ai ∈ F , i = 1, 2, ..., Ai ↑ A, then A ∈ F .

If G and F are two σ-fields in  and G ⊂ F , we say that G is a sub-σ-field of

F . Given a class C ⊂ P(), the smallest σ-field containing C is called the σ-field
generated by C . It is denoted by σ(C ). A σ-field F in  is said to be countably
generated if there exists a countable class of subsets C ⊂ P() such that σ(C ) = F .
If C ⊂ P() and A ⊂  we define C ∩ A := {B ∩ A : B ∈ C }. We denote by
σ A (C ∩ A) the σ-field of subsets of A generated by C ∩ A. It is easy to see that
σ A (C ∩ A) = σ(C ) ∩ A (see, for instance, [18], p. 5).

© Springer International Publishing AG 2017 1

G. Fabbri et al., Stochastic Optimal Control in Infinite Dimension,
Probability Theory and Stochastic Modelling 82,
DOI 10.1007/978-3-319-53067-3_1
2 1 Preliminaries on Stochastic Calculus in Infinite Dimension

For A ⊂  we denote its complement by Ac :=  \ A, and for A, B ⊂  we

denote their symmetric difference by AB := (A \ B) ∪ (B \ A). We will write
+
R+ = [0, +∞), R = [0, +∞) ∪ {+∞}, R = R ∪ {±∞}.
Theorem 1.2 Let G be a π-system and F be a λ-system in some set , such that
G ⊂ F . Then σ(G ) ⊂ F .
Proof See [370], Theorem 1.1, p. 2. 
Corollary 1.3 Let G be a π-system and F be the smallest family of subsets of 
such that:
• G ⊂ F;
• if A ∈ F then Ac ∈ F ;
∞
• if Ai ∈ F , Ai ∩ A j = ∅ for i, j = 1, 2, ..., i = j, then ∪i=1 Ai ∈ F .
Then σ(G ) = F .
Proof Since σ(G ) satisfies the three conditions for F , we obviously have F ⊂
σ(G ). For the opposite inclusion it remains to observe that F is a λ-system. (For a
self-contained proof, see also [180], Proposition 1.4, p. 17.) 
Definition 1.4 (Measurable space) If  is a set and F is a σ-field in , the pair
(, F ) is called a measurable space.
Definition 1.5 (Probability measure, probability space) Consider a measurable
space (, F ). A function μ : F → [0, +∞) ∪ {+∞} is called a measure on (, F )
if μ(∅) = 0, and whenever Ai ∈ F , Ai ∩ A j = ∅ for i, j = 1, 2, ..., i = j, then
∞  ∞
 
μ Ai = μ(Ai ).
i=1 i=1

The triplet (, F , μ) is called

 a measure space. If μ() < +∞ we say that μ is a
bounded measure. If  = ∞ n=1 An , where An ∈ F , μ(An ) < +∞, n = 1, 2, ..., we
say that μ is a σ-finite measure. If μ() = 1 we say that μ is a probability measure.
We will use the symbol P to denote probability measures. The triplet (, F , P) is
called a probability space.
Thus a probability measure is a σ-additive function P : F → [0, 1] such that
P() = 1.
Given a measure space (, F , μ), we define N := {F ⊂  : ∃G ∈ F , F ⊂ G,
μ(G) = 0}. The elements of N are called μ-null sets. If N ⊂ F , the measure space
(, F , μ) is said to be complete. The σ-field F := σ(F , N ) is called the completion
of F (with respect to μ). It is easy to see that σ(F , N ) = {A ∪ B : A ∈ F , B ∈ N }.
If G ⊂ F is another σ-field then σ(G , N ) is called the augmentation of G by the
null sets of F . The augmentation of G may be different from its completion, as the
latter is just the augmentation of G by the subsets of the sets of measure zero in G .
We also have σ(G , N ) = {A ⊂  : AB ∈ N for some B ∈ G }.
1.1 Basic Probability 3

Let μ, ν be two measures on a measurable space (, F ). We say that μ is

absolutely continuous with respect to ν (we write μ << ν) if for every A ∈ F such
that ν(A) = 0 we have μ(A) = 0. If μ << ν and ν << μ, we say that the measures
μ and ν are equivalent (we write μ ∼ ν). If there exists a set A ∈ F such that for
every B ∈ F we have μ(B) = μ(A ∩ B), we say that μ is concentrated on the set
A. If μ and ν are concentrated on disjoint sets we say that μ and ν are (mutually)
singular and we write μ ⊥ ν.

Lemma 1.6 Let μ1 , μ2 be two bounded measures on a measurable space (, F ),

and let G be a π-system in  such that  ∈ G and σ(G ) = F . Then μ1 = μ2 if and
only if μ1 (A) = μ2 (A) for every A ∈ G .

Proof See [370], Lemma 1.17, p. 9. 

Let t , t ∈ T be a family of sets. We will denote the Cartesian product of the

family t by ×t∈T t . If T is finite (T = {1, ..., n}) or countable (T = N), we will
also write 1 × ... × n , respectively 1 × 2 × .... If each t is a topological
space, we endow ×t∈T t with the product topology. If each t has a σ-field Ft , we
define the product σ-field ⊗t∈T Ft in ×t∈T t as the σ-field generated by the one-
dimensional cylinder sets At × ×s =t s . If T = {1, ..., n} (respectively, T = N) we
will just write ⊗t∈T Ft = F1 ⊗ ... ⊗ Fn (respectively, ⊗t∈T Ft = F1 ⊗ F2 ⊗ ...).
If S is a topological space, the σ-field generated by the open sets of S is called
the Borel σ-field. It will be denoted by B(S). If S is a metric space, unless stated
otherwise, its default σ-field will always be B(S). It is not difficult to see that if
S1 , S2 , ... are separable metric spaces, then

B(S1 × S2 × ...) = B(S1 ) ⊗ B(S2 ) ⊗ ....

If (S, ρ) is a metric space, A ⊂ S, and we consider (A, ρ) as a metric space, then

B(A) = A ∩ B(S). A complete separable metric space is called a Polish space. Also
+
B(R ) = σ(B(R+ ), {+∞}), B(R) = σ(B(R), {−∞}, {+∞}).
A measurable space (, F ) is called countably determined (or F is called count-
ably determined) if there is a countable set F0 ⊂ F such that any two probability
measures on (, F ) that agree on F0 must be the same. It follows from Lemma 1.6
that if F is countably generated then F is countably determined. If S is a Polish
space then B(S) is countably generated.
If (i , Fi , μi ), i = 1, ..., n, are measure spaces, their product measure on (1 ×
... × n , F1 ⊗ ... ⊗ Fn ) is denoted by μ1 ⊗ ... ⊗ μn .
If S is a metric space, a bounded measure μ on (S, B(S)) is called regular if

μ(A) = sup{μ(C) : C ⊂ A, C closed} = inf{μ(U ) : A ⊂ U, U open} ∀A ∈ B(S).

Every bounded measure on (S, B(S)) is regular (see [478], Chap. II, Theorem 1.2).
A bounded measure μ on (S, B(S)) is called tight if for every ε > 0 there exists
a compact set K ε ⊂ S such that μ(S \ K ε ) < ε. If S is a Polish space then every
bounded measure on (S, B(S)) is tight (see [478], Chap. II, Theorem 3.2).
4 1 Preliminaries on Stochastic Calculus in Infinite Dimension

We refer to [58, 61, 267, 370, 478] for more on the general theory of measure
and probability.

1.1.2 Random Variables

Definition 1.7 (Random variable) A measurable map X between two measurable

spaces (, F ) and (, ˜ G ) is a called a random variable. This means that X is a
random variable if X −1 (A) ∈ F for every A ∈ G . We write it shortly as X −1 (G ) ⊂
F . Sometimes we will just say that X is F /G -measurable.
If ˜ = R (resp. R+ ) and G is the Borel σ-field B(R) (resp. B(R+ )) then X is said
to be a real random variable (resp. positive random variable).
If , ˜ are topological spaces and F , G are the Borel σ-fields then X is said to
be Borel measurable.
If (, F , μ) is a measure space and X, X 1 :  → ,˜ we say that X 1 is a version
of X if X = X 1 μ-a.e.
Given a random variable X : (, F ) → (, ˜ G ) we denote by σ(X ) the smallest
sub-σ-field of F that makes X measurable, i.e. σ(X ) := X −1 (G ). It is called the
σ-field generated by X . Given a set of indices I and a family of random variables
˜ G ), i ∈ I , the σ-field σ (X i : i ∈ I ) generated by {X i }i∈I is the
X i : (, F ) → (,
smallest sub-σ-field of F that makes all the functions X i : (, σ (X i : i ∈ I )) →
˜ G ) measurable, i.e. σ (X i : i ∈ I ) = σ X i−1 (G ) : i ∈ I .
(,
Lemma 1.8 Let (, F ) be a measurable space. Then:
˜ G ) is a measurable space, X :  → ,
(i) If (, ˜ and C ⊂ G is such that σ(C ) =
G , then X is F /G -measurable if and only if X −1 (C ) ⊂ F . Moreover, σ(X ) =
σ(X −1 (C )).
(ii) If X n :  → R, n = 1, 2, ..., are random variables, then supn X n , inf n X n ,
lim supn X n , lim inf n X n are random variables.
(iii) Let X n :  → S, n = 1, 2, ..., be random variables, where S is a metric
space. Then:
• if S is complete then {ω : X n (ω) converges} ∈ F ;
• if X n → X on , then X is a random variable.
(iv) Let (i , Fi ), i = 1, 2, be measurable spaces, and X : 1 × 2 →  be
(F1 ⊗ F2 )/F -measurable. Then, for every ω1 ∈ 1 , X ω1 (·) = X (ω1 , ·) is F2 /F -
measurable, and, for every ω2 ∈ 2 , X ω2 (·) = X (·, ω2 ) is F1 /F -measurable. 
Proof See, for instance, [370], Lemmas 1.4, 1.9, 1.10, and [520], Theorem 7.5,
p. 138. 
Theorem 1.9 Let (, F ) and (, ˜ G ) be two measurable spaces and (S, d) a Pol-
˜
ish space. Let X : (, F ) → (, G ) and φ : (, F ) → (S, B(S)) be two random
variables. Then φ is measurable as a map from (, σ(X )) to (S, B(S)) if and only
˜ G ) → (S, B(S)) such that φ = η ◦ X .
if there exists a measurable map η : (,
1.1 Basic Probability 5

Proof See [370], Lemma 1.13, p. 7, or [575] Theorem 1.7, p. 5. 

We refer to [58, 267, 370, 520] for more on measurability and for the general
theory of integration.
Definition 1.10 (Borel isomorphism) Let (, F ) and (, ˜ G ) be two measurable
˜ is called a Borel isomorphism if f is F /G -
spaces. A bijection f from  onto 
˜ G ) are
measurable and f −1 is G /F -measurable. We then say that (, F ) and (,
Borel isomorphic.
Definition 1.11 (Standard measurable space) A measurable space (, F ) is called
standard if it is Borel isomorphic to one of the following spaces:
(i) ({1, .., n}, B({1, .., n})),
 B(N)),
(ii) (N, 
(iii) {0, 1}N , B({0, 1}N ) ,
where we have the discrete topologies in {1, .., n} and N, and the product topology
in {0, 1}N .
The following theorem collects results that can be found in [478] (Chap. I, Theorems
2.8 and 2.12).
Theorem 1.12 If S is a Polish space, then (S, B(S)) is standard. If a Borel subset of
S is uncountable, then it is Borel isomorphic to {0, 1}N . Two Borel subsets of S are
Borel isomorphic if and only if they have the same cardinality. If (, F ) is standard
and A ∈ F , then (A, F ∩ A) is standard.
In particular, we have the following result.
Theorem 1.13 If (, F ) is standard, then it is Borel isomorphic to a closed subset
of [0, 1] (with its induced Borel sigma field).
Definition 1.14 (Simple random variable) Let (, F ) be a measurable space, and
(S, d) be a metric space (endowed with the Borel σ-field induced by the distance).
A random variable X : (, F ) → (S, B(S)) is called simple (or a simple function)
if it has a finite number of values.
Lemma 1.15 Let f : (, F ) → S be a measurable function between a measur-
able space (, F ) and a separable metric space (S, d) (endowed with the Borel
σ-field induced by the distance). Then there exists a sequence f n :  → S of simple,
F /B(S)-measurable functions, such that d ( f (ω), f n (ω)) is monotonically decreas-
ing to 0 for every ω ∈ .
Proof See [180], Lemma 1.3, p. 16. 
Lemma 1.16 Let S be a Polish space with metric d. Let (, F , P) be a complete
probability space and let G1 , G2 ⊂ F be two σ-fields with the following property:
for every A ∈ G2 there exists a B ∈ G1 such that P(AB) = 0. Let f : (, G2 ) →
(S, B(S)) be a measurable function. Then there exists a function g : (, G1 ) →
(S, B(S)) such that f = g, P-a.e., and simple functions gn : (, G1 ) → (S, B(S))
such that d( f (ω), gn (ω)) monotonically decreases to 0, P-a.e.
6 1 Preliminaries on Stochastic Calculus in Infinite Dimension

Proof The proof follows the lines of the proof of Lemma 1.25, p. 13, in [370].
Step 1: Let us assume first that f = x1 A (1 A denotes the characteristic function
of the set A) for some A ∈ G2 and x ∈ S. By hypothesis, we can find B ∈ G1 s.t.
P(AB) = 0 and then the claim is proved if we choose gn ≡ g = x1 B . The same
argument holds for a simple function f .
Step 2: For the case of a general f , thanks to Lemma 1.15 we can find a sequence
of simple, G2 -measurable functions f n such that d( f (ω), f n (ω)) monotonically
decreases to 0. By Step 1, we can find simple, G1 -measurable functions gn such
that f n = gn , P-a.e. Thus the claim follows by taking g(ω) := lim gn (ω) if the limit
exists and g(ω) = s (for some s ∈ S) otherwise. 
Lemma 1.17 Let (, F ) be a measurable space, and V ⊂ E be two real separable
Banach spaces such that the embedding of V into E is continuous. Then:
(i) B(E) ∩ V ⊂ B(V ) and B(V ) ⊂ B(E).
(ii) If X :  → V is F /B(V )-measurable, then it is F /B(E)-measurable.
(iii) If X :  → E is F /B(E)-measurable, then X · 1{X ∈V } is F /B(V )-
measurable.
(iv) X :  → E is F /B(E)-measurable if and only if for every f ∈ E ∗ , f ◦ X is
F /B(R)-measurable.
Proof The embedding of V into E is continuous, so B(E) ∩ V ⊂ B(V ). Since the
embedding is also one-to-one, it follows from [478], Theorem 3.9, p. 21, that B(V ) ⊂
B(E), which completes the proof of (i). Parts (ii) and (iii) are direct consequences
of (i). f () is separable because E is separable, so Part (iv) is a particular case of
the Pettis theorem, see [488] Theorem 1.1. 
Lemma 1.18 Let (, F ) be a measurable space and (S1 , ρ1 ), (S2 , ρ2 ) be two metric
spaces with S1 separable. Let f :  × S1 → S2 be such that
(i) for each x ∈ S1 , the function f (·, x) :  → S2 is F /B(S2 )-measurable;
(ii) for each ω ∈  the function f (ω, ·) : S1 → S2 is continuous.
Then f :  × S1 → S2 is F ⊗ B(S1 )/B(S2 )-measurable.
Proof See Lemma 4.51, p. 153 of [8]. 
Notation 1.19 If E is a Banach space we denote by | · | E its norm. Given two Banach
spaces E and F, we denote by L(E, F) the Banach space of all continuous linear
operators from E to F. If E = F we will usually write L(E) instead of L(E, F). If
H is a Hilbert space we denote by ·, · its inner product. We will always identify H
with its dual via Riesz representation theorem. If V, H are two real separable Hilbert
spaces, we denote by L2 (V, H ) the space of Hilbert–Schmidt operators from V to
H (see Appendix B.3). The space L2 (V, H ) is a real separable Hilbert space with
the inner product ·, ·2 , see Proposition B.25. 
Lemma 1.20 Let (, F ) be a measurable space and V, H be real separable Hilbert
spaces. Suppose that F :  → L2 (V, H ) is a map such that for every v ∈ V , F(·)v
is F /B(H )-measurable. Then F is F /B(L2 (V, H ))-measurable.
1.1 Basic Probability 7

Proof Since L2 (V, H ) is separable, by Lemma 1.17-(iv) it is enough to show that

for every T ∈ L2 (V, H )

+∞

ω→F(ω), T 2 = F(ω)ek , T ek 
k=1

is F /B(R)-measurable, where {ek } is any orthonormal basis of V . But this is clear

since for every ω
F(ω), T 2 = lim FnT (ω),
n→+∞

where

n
FnT (ω) = F(ω)ek , T ek 
k=1

and FnT (ω) is F /B(R)-measurable because it is a finite sum of functions that are
F /B(R)-measurable. 
Let I be an interval in R, E, F be two real Banach spaces, and let E be separable.
If f : I × E → F is Borel measurable then for every t ∈ I the function f (t, ·) :
E → F is Borel measurable (by Lemma 1.8-(iv)).
Assume now that, for all t ∈ I and for some m ≥ 0, f (t, ·) ∈ Bm (E, F) (the space
of Borel measurable functions with polynomial growth m, see Appendix A.2 for the
precise definition). It is not true in general that the function

I → Bm (E, F), t→ f (t, ·)

is Borel measurable. As a counterexample1 one can take the function

[0, 1] × L 2 (R) → L 2 (R), (t, x)→St x,

where (St )t≥0 is the semigroup of left translations. Indeed, the map

[0, 1] → L(L 2 (R)), t→St

is not measurable (see e.g. [180], Sect. 1.2). Since L(L 2 (R))⊂B1 (L 2 (R), L 2 (R))
and the norm in L(L 2 (R)) is equivalent to the one induced by B1 (L 2 (R), L 2 (R)),
the claim follows in a straightforward way.
On the other hand, we have the following useful result.
Lemma 1.21 Let I and  be two Polish spaces. Let μ be a measure defined on the
Borel σ-field B(I ) and denote by B(I ) the completion of B(I ) with respect to μ.
Let f : I ×  → R be Borel measurable and such that for every t ∈ I , f (t, ·) is
bounded from below (respectively, above). Then the function

1 This example has been suggested to us by Mauro Rosestolato.

Exploring the Variety of Random
Documents with Different Content
Alice preferred the mountains, and chose a delightful little salon
upstairs as her own sitting-room, whilst her bedroom was close at
hand. There was a balcony, and she liked to sit there in the mild air
during the warmest and brightest hours.
Mr. Stedman's powerful and active nature suffered from their
monotonous life at the villa, and he needed exercise both for the
body and the mind. Alice perceived this, and, well knowing that it
was impossible for her father to do any thing except in her service,
plotted a little scheme by which she hoped to make him take the
exercise and the interest in outward things which in these sad days
were more than ever necessary to him.
"Papa," she said one day, "I think if I'd a little regular work to do, it
would do me good. I wish you would go geologizing for me, and
bring me specimens. You might botanize a little, too,
notwithstanding the time of the year; it would be amusing to puzzle
out some of the rarer plants. It's a very curious country, isn't it,
papa? I'm sure, if I were well, we should find a great deal of work to
do together here." Then she began to question him about the
geology and botany of the district, and made him buy some books
which have been written upon these subjects by scientific
inhabitants of Avignon. Her little trick succeeded. Mr. Stedman,
under the illusion that he was working to please his poor Alice,
trudged miles and miles in the country, and extended his
explorations to the very slopes of Mount Ventoux itself. In this way
he improved the tone of his physical constitution, and Alice saw with
satisfaction that it would be better able to endure the impending
sorrow.
He had long ceased to treat Philip Stanburne with coldness or
distrust. His manner with his young friend was now quite gentle, and
even affectionate, tenderly and sadly genial. The one point on which
they disagreed was no longer a sore point for either of them. One
day, when they were together, they met a religious procession, with
splendid sacerdotal costumes and banners, and Philip kneeled as the
host was carried by. Their conversation, thus briefly interrupted, was
resumed without embarrassment, and Mr. Stedman asked some
questions about the especial purpose of the procession, without the
slightest perceptible expression of contempt for it. He began to take
an interest in the charities of the place, and having visited the
hospital, said he thought he should like to give something, and
actually left a bank-note for five hundred francs, though the
managers of the institution, and the nurses, and the patients, were
Romanists without exception. Meanwhile, he read his Bible very
diligently every day, and the prayers of the little household, in which
Philip willingly joined.
During one of Mr. Stedman's frequent absences on the little scientific
missions ordered by his daughter Alice, she and Philip had a
conversation which he ever afterwards remembered.
"Philip," she said, "do you ever think much about what might have
been, if just one circumstance had been otherwise? I have been
thinking a great deal lately, almost constantly, about what might
have been, for us two, if my health had been strong and good.
People say that love such as ours is only an illusion—only a short
dream—but I cannot believe that. It might have changed, as our
features change, with time, but it would have remained with us all
our lives. Do you ever fancy us a quiet respectable old couple, living
at the Tower, and coming sometimes to Sootythorn together? I do. I
fancy that, and all sorts of things that might have been—and some
of them would have been, too—if I had lived. There's one thing
vexes me, and that is, that I never saw the Tower. I wish I had just
seen it once, so that I might fancy our life there more truly. How
glad dear papa would have been to come and stay with us, and
botanize and geologize amongst your rocks there! You would have
let him come, wouldn't you, dear?—I am sure you would have been
very kind to him. You will be kind to him, won't you, my love, when
he has no longer his poor little Lissy to take care of him? Don't leave
him altogether by himself. I am afraid his old age will be very sad
and lonely. It grieves me to think of that, for he will be old in a few
years now, and his poor little daughter will not be near him to keep
him cheerful. Fancy him coming home every evening from the mill,
and nobody but servants in the house! Go and stay with him
sometimes, dear, at Chesnut Hill, and get him to go to the Tower,
and you will sometimes talk together about Alice, and it will do you
both good."
Philip had kept up manfully as long as he was able, but the vivid
picture that these words suggested of a world without Alice was too
much for him to bear, and he burst into passionate tears. As for
Alice, she remained perfectly calm, but when she spoke again it was
with an ineffable tenderness. She took his hand in hers, and drew
him towards her, and kissed him. Again and again she kissed him,
smoothing his hair caressingly with her fingers—gentle touches that
thrilled through his whole being. "You don't know, my darling," she
said, "how much I love you, and how miserable it made me when I
thought we must be separated in this world. It isn't so hard to be
separated by death; but to live both of us in the same world, seeing
the same sun, and moon, and stars, even the same hills, and not to
be together, but always living out of sight and hearing of each other,
and yet so near—it would have been a trial beyond my strength! And
isn't it something, my love, to be together as we are now for the last
few weeks and days? You don't know how happy it makes me to see
you and papa getting on so nicely as you do. Isn't he nice, now? I
don't believe he thinks a bit the worse of you for being a Catholic.
We shall all meet again, darling—shall we not?—in the same heaven,
and then we shall have the same perfect knowledge, and our errors
and differences will be at an end for ever."
She was a good deal exhausted with saying this, and leaned back in
her chair, closing her eyes for a while. Philip gradually recovered his
usual melancholy tranquillity, and they sat thus without speaking, he
holding both her hands in his, and gently chafing and caressing
them. He had not courage to speak to Alice—indeed, in all their
saddest and most serious conversations, the courage was mainly on
her side.
Whilst they were sitting thus, the sky became suddenly overcast,
and there came a few pattering drops of rain. Alice started suddenly,
and seemed to be agitated by an unknown terror. She grasped
Philip's hand in a nervous way, and complained of a strange
suffering and foreboding. "I felt so calm and peaceful all the
morning," she said; "I wish I could feel so now."
The agitation increased, and it was evident to Philip that a great
change had taken place. Alice threw her arms round him, and
clasped him to her. "O Philip!" she cried, wildly, "don't leave me now
—don't leave me even for a minute! Stay, darling, stay; it is coming,
coming!"
The pattering of the rain had ceased. It had been nothing but a few
drops—scarcely even a shower—and it had ceased.
But the air was not clearer after the rain. On the contrary, it had
been clearer before it than it was now. The snowy summit of Mount
Ventoux was hidden in an opaque, thick atmosphere; mist it was
not, as we northerns understand mist, but a substantial thickening of
the air.
Soon there was the same thickening, the same opacity in the
atmosphere of the remote plain that stretched to the mountain's
foot. It was invisible now, the Mount Ventoux, the Mountain of the
Winds.
And as the plain grew dark the Rhone as suddenly whitened. It
whitened and whitened, nearer and nearer Avignon; then a dull
distant roar became audible, steadily increasing. A violent brief
squall shook the villa. What! so frightened already? Poor children, it
is nothing yet!
Over the terrified plain, over the foaming river, comes the Mistral,
careering in his strength! Well for you, walls of Avignon, that you
were built for the shocks of battle! well for thee, most especially, O
palace of the transplanted Papacy, that thy fortress-heights were
erected less for pleasure than for resistance!
Louder and louder, nearer and nearer! How the trees bend like
fishing-rods! Crash, crash—they break before the tempest. What a
clatter against the windows! It is a volley of pebbles that the Mistral
carries with it as a torrent does. Bang, bang—the shutters are torn
off their iron hinges and pitched nobody knows where—into the
court, on the roof-top, it may be, or into the neighbor's garden!
The intensity of the noise made all human voices inaudible. The
Mistral likes to make an uproar—it is his amusement, when he
comes to Avignon from his mountain. And he whistles at once in a
thousand chimneys, as a boy whistles in two steel keys; and he
makes such a clatter with destroying things, that the most insured
house-property leaves no peace to its possessor. But straight in the
midst of his path rise the towers of the fortress-palace, and Peter
Obreri, its architect, knows in the world of spirits that they resist the
Mistral yet.
But alas for our poor little Alice! This wind does not suit her at all;
this unceasing, this wearisome wind—this agitating, terrible wind!
She did not fear death before, in the calm serene weather, when it
seemed that her soul might rise in the blue ether, and be borne by
floating angels. But to go out into the bleak, stern tempest—to leave
his encircling arms, and be dashed no one knows whither along the
desolate, unfamiliar Provence, with twigs, and dead leaves, and
pebbles, and that choking cloud of sand!
"Forgive me these foolish fancies," she prayed, from the depths of
this horror. "My soul knows her way to the haven of thy rest, O Lord,
my Guide and my Redeemer!"
 CHAPTER XXXIV.
JACOB OGDEN FREE AGAIN.
Early in the month of February there came a black-edged letter to
Arkwright Lodge, with a French stamp upon it. The letter was from
Philip Stanburne, and it announced Alice Stedman's death.
Two days after the arrival of that letter another letter arrived at
Milend for Jacob Ogden. It bore the Whittlecup post-mark, and had
an exact outward resemblance to several other letters which had
come from the same place, but its contents were of a new character.
Miss Anison expressed her regret that in consequence of Mr. Jacob
Ogden's neglect, of his readiness to postpone his visits on the
slightest pretexts, of the rarity and coldness of his letters, she felt
compelled, from a due regard to her own happiness, to put an end
to the engagement which had existed between them.
The accusations in this letter were perfectly well founded, though it
is quite certain that they would never have been made if Philip
Stanburne's communication had been edged with silver instead of
black. Margaret Anison had remarked with secret satisfaction that
Jacob Ogden's behavior as a lover gave her good reasons for
retreating from her engagement, whenever she might determine on
that decisive step; but in the mean while she had never reproached
him with it, had never appeared aware of it when he did come, but
always received him in the same uniformly gracious way, as if he had
been the most assiduous of adorers. She had kept this accusation of
negligence to be used against him whenever it might be convenient
to throw the blame of a rupture upon him; but if she had finally
decided to marry him, this and all other faults would have been
affectionately overlooked. It had been highly convenient to let him
sink deeper and deeper in that sin of negligence, till at last, from
mere carelessness and an aversion to all letter-writing that was not
upon business, he had actually reached that depth in crime that he
no longer observed the common forms of society, and did not even
write a line of apology or excuse. Margaret never expected him to be
attentive to her as a husband: she intended to spend his money,
and, so long as that was forthcoming, cared little about Jacob
Ogden's manners. But it was charming to be able to back out of her
engagement, now that Alice was dead, and do it in a dignified and
honorable manner. For of all sins that a lover can commit, the chief
is the sin of neglect; and in this case any competent and just jury
would have pronounced the verdict "guilty."
To this letter Jacob Ogden made no reply. His feelings on receiving it
were, first, the most unfeigned astonishment (for he thought he had
been very attentive, and that "courtin'" had absorbed far too much
of his time); next, a paroxysm of indignation, with a sense of injury;
and then, when this subsided, a sense of relief so exquisite, so
delicious, and so complete, that nobody can have any idea of it
unless at some period of his existence a wearing and persistent
anxiety has been suddenly removed for ever. The love of Margaret
Anison had been one of those masterful passions which sometimes
force the most prudent men to folly. He had made his offer in the
height of his temporary insanity, but after the engagement had been
entered upon, his old self had gradually returned; and though he
was fully determined to "go through with it," as a business which
had to be done, he by no means looked forward to the conjugal
state as an improvement upon his accustomed life. It was like
embarking on an unknown and perilous sea, in utter ignorance of
the art of navigation, and that sea might be a sea of troubles. The
complex details of married life, its endless little duties, were
perplexing to a man whose time and thoughts were already taken up
by the government of a heavy business, and the care of an
increasing estate. And now to escape from these new and unfamiliar
troubles—to remain in the old quiet life at Milend—to have full
control over his own expenditure, with no female criticism or
interference—to see his fortune growing and growing without sons
to establish or daughters to dower, or an expensive houseful of
servants to eat the bank-notes in his pocket-book like so many
nattering mice,—ah! it was sweet to him to think of this in his
innermost and sincerest self! He had loved his bachelor life well
enough before, but he had never felt the full luxury of its
independence as he did now!
Jacob Ogden enjoyed a privilege highly favorable to happiness, but
not so favorable to moral or intellectual growth. He lived at peace
with himself, and looking back on his life, he approved of its whole
course, with the single exception of that hour of folly at Whittlecup.
He felt and believed that no man could be wiser or more perfect
than he was. When he humbly called his faculties "common-sense,"
he by no means understood the word as meaning a sense which he
had in common with others, but rather a special faculty, to himself
vouchsafed by the bounteous gift of nature. He lived in absolute
independence of the good opinion of others, because his mind was
at peace with itself—because he always manfully did to-day what he
was sure to approve to-morrow, or ten years after to-morrow. Am I
painting the portrait of a man of pre-eminent virtues? Not exactly,
but of a man who would have been pre-eminently virtuous, or pre-
eminently learned, if virtue or knowledge had been his ideal. For he
had a manly resolution, a steady unflinching determination, to live
up to the standard which he fixed for himself. And the inward peace
which he enjoyed was due to his obedience to the laws of his own
nature, which thus ever remained in harmony with itself in serene
strength and efficiency.
This peace had for a while been lost to him, and he had felt a
strange change and diminution in the inward satisfactions. His
communings with himself had lost their old sweetness, and he no
longer masticated the cud of contentment in the fair pastures of
reflection and imagination. To go back to those happy pastures once
more—to chew that sweet cud again, after months of privation—
what a deep, strengthening, cheering, encouraging, replenishing
delight it was!
Yet there was one drawback to the plenitude of Ogden's happiness,
even though he had escaped the misery of the wedding-day. That
new mansion had been begun, he had spent £400 upon it already,
and spoilt a pretty meadow, and he had spent some money on
presents for Margaret—not very much, for his ideas on the subject of
gift-making were not very large ideas, yet still enough to plague and
torment him, for the loss of a sovereign would do that. To be jilted
did not trouble him much, but to have been cheated into wasting his
money! that thought would not let him rest. It followed and
harassed him wherever he went, and it was the cause of the
following letter, which was received by Mr. Joseph Anison:—
"Sir,—I am instructed by my client, Mr. Jacob Ogden, to
lay before you the following statement of facts. Your
daughter, Miss Margaret Anison, by a letter bearing date
——, and which is in our possession, accepted his
proposal of marriage, and promised marriage; which
promise she now, by a letter bearing date ——, refuses to
execute. In consequence of her promise, and in
conformity with her desires, our client has been led into
considerable expense, especially in the erection of a
mansion, of which Miss Anison herself selected the site.
The works were immediately stopped when it became
known to our client that Miss Anison had determined upon
a breach of promise, but a heavy sum had been already
expended, which, so far as our client is concerned, is
money utterly thrown away. We beg to call your attention
to the fact that our client and his mother offered another
most commodious and suitable residence to Miss Anison,
situated at Milend, and that she declined this, and induced
our client to commence the erection of a new and costly
mansion on a site which he would never have selected for
himself. We therefore claim for our client damages to the
amount of one thousand pounds (£1,000), and beg to
inform you, that unless this sum is paid before the
expiration of one calendar month from this date, we shall
institute a suit for breach of promise of marriage, and
claim damages on that score to a far heavier amount. The
present claim, we desire it to be understood, is not made
on the ground of breach of promise, but is merely a claim
for compensation on account of outlay which our client
has been induced to incur. Our client has no desire to
push matters to the extremity of a public exposure, but
will not shrink from doing so if his present just claim is
refused.
"I am, sir, your obedient servant,
 "Jonas Hanby."
"You may decide for yourself, Margaret," said Mr. Anison, "whether
you prefer that I should pay this out of your fortune, or stand an
action for breach of promise. It is not usual to bring actions of this
sort against women, but Ogden is a most determined fellow, and he
doesn't care much for what people may say. He will bring his action
if we don't send him a cheque, and I don't think such an action
would be very pleasant to you. Considering circumstances, too,
especially the building of that new house, I am inclined to think that
he would get rather heavy damages, certainly at least as much as he
is asking for. Such an action would make a tremendous noise, and
we should be in all the newspapers. We must consider your sisters,
too, who wouldn't be much benefited by publicity of this kind. In
short, my advice is to send the cheque."
The cheque was accordingly sent to Mr. Hanby, and duly
acknowledged. The presents had been returned a few days before.
These last had been purchased of a jeweller in St. Ann's Square,
Manchester, who took them back in exchange for an excellent
gentleman's watch and a big cameo brooch. The watch went into
Jacob Ogden's own fob, and the brooch adorned his already
sufficiently ornamented mother. All things considered, Jacob Ogden
now felt that he could look back upon the whole business with a
mind at ease. He had done his duty by himself. After deducting the
outlay on the house, and the outlay necessary for restoring the field
to its pristine verdure, he found that there remained to him a clear
surplus of four hundred and fifteen pounds seven shillings and
twopence, which he entered in the column of profits. "It's been
rather a good business for once, has this courtin'," said Jacob to
himself; "but it's devilish risky, and there's nobody'll catch me at it
again. If she'd nobbut stuck to me, she'd 'ave wenly ruined me."
So, when the walls of the mansion that was to have been were
levelled with the ground, and the foundations buried under the earth
that they might be no more seen, Jacob Ogden buried with them the
thought and idea of marriage; and the grass grew on the field that
had been so torn, and cut, and burdened, and disturbed by the
masons and laborers who had been there.
As the field grew level and green again just as it used to be, so
flourished the mind of Jacob Ogden in serene and productive life.
But as beneath the field—beneath the waving of the rich grass—
there still lay the plan of the house that was to have been, traced
out in stony foundations, so in the mind of its owner there lay
hidden a stony memory of the plans of this strange year; and though
the surface was perfectly restored, there were hard places under his
happiness that had not been there before.
 CHAPTER XXXV.
LITTLE JACOB'S EDUCATION.
The rupture between Jacob Ogden and Miss Anison had an
immediate effect upon the fortunes of a young friend of ours, who
has for a long time been very much in the background. Little Jacob
began to occupy a larger and larger place in his uncle's thoughts.
For, though Uncle Jacob had formerly always intended, in a general
way, to remain a bachelor, this had been nothing more than a sort of
intellectual preference for bachelorhood, deduced from his general
views of life, and especially from his dominant anxiety to make a
fortune. But his objections to matrimony were no longer of this mild
kind. Like a wild animal that has once felt the noose of the trapper
round its neck, and yet succeeded in freeing itself, he had conceived
a horror of the snare which was incomparably more active and
intense than the vague alarms of the inexperienced. His former ideas
about marriage had been purely negative. He had no intention to
marry, and there was the end of his reflections on the matter. But
now his preference for celibacy had taken the shape of a passionate
and unalterable resolution.
The increase of his fortune, which might henceforth be surely relied
on, led him to think a good deal about the little boy at Twistle Farm,
who was most probably destined to inherit it; and he determined to
use a legitimate influence over his brother Isaac, so that little Jacob
might be educated in a manner suitable to his future position.
We have said that Jacob Ogden was perfectly satisfied with himself,
and that knowledge was not his ideal. But although this is true, his
views were really larger than the reader may have hitherto
suspected. He considered himself perfect in his place; but as little
Jacob would probably have a very different place in the world, he
would need different perfections. The qualities needed for making a
large fortune were, in Jacob Ogden's view, the finest qualities that a
human being can possess, and he knew that he possessed them;
but then there were certain ornaments and accomplishments which
were necessary to a rich gentleman, and which the manufacturer
had not had time to acquire. He was not foolish enough to torment
himself with regrets that he did not know Latin and Greek; he had
none of the silly humilities of weak minds that are perpetually
regretting their "deficiencies." Whatever it was necessary for his
main purpose that he should know, he always resolutely set himself
to learn, and, by strenuous application, mastered; what was
unnecessary for his purpose, he remained contentedly ignorant
about. The customary pedantries of the world, its shallow pretension
to scholarship, never humiliated him. He suspected, perhaps, that
genuine classical acquirement was much rarer than the varnish of
pseudo-scholarship, and he had not that deferential faith in
gentlemen's Latin and Greek which is sometimes found in the
uneducated. But, on the other hand, as he had learned every thing
that was necessary to a plodding Shayton cotton-spinner, so he was
determined that little Jacob should learn every thing necessary to a
perfect English gentleman. He had not read the sentence of
Emerson, "We like to see every thing do its office after its kind,
whether it be a milk-cow or a rattlesnake;" but the sentiment in it
was his own. His strong sense perceived that so long as men hold
different situations in the world, their preparatory training must be
different; and that, as a young pigeon must learn to fly, and a young
terrier to catch rats, so the youthful heir of a splendid fortune, and
the boy who has his fortune to make, ought to receive respectively a
celestial and a terrestrial training.
For Jacob Ogden, himself a terrestrial, knew that there was a heaven
above him—the heaven of aristocracy! There dwelt superior beings,
in golden houses, like gods together, far above the ill-used race of
men that cleave the soil and store their yearly dues. There is
something ludicrous, if it were not pathetic and painful, in the self-
abasement of a man so strong and resolute as Ogden before a
heaven whose saints and angels were only titled ladies and
gentlemen, mainly occupied in amusing themselves; but to him it
was the World of the Ideal. And this religion had one great
advantage—it kept him a little humbler than he ever would have
been without it. Great was the successful cotton-spinner in his eyes,
but there were beings cast by nature in a nobler mould. For Jacob
Ogden actually believed, in all sincerity and simplicity, that there was
the same natural difference between a lord and a plebeian that there
is between a thorough-bred and a cart-horse. This superstition,
though founded on a dim sense of the natural differences which do
exist, erred in making them the obedient servants of the artificial
differences. There are, no doubt, thorough-breds and cart-horses
amongst mankind, and the popular phraseology would imply that
there are also asses; but these natural differences seem to be
independent of title altogether, and dependent even upon fortune
only so far as it may help or hinder their development. The
superstition that lords, quâ lords, are wiser, and better, and braver,
and more respectable than other people, was more prevalent in
Shayton than it is in places where lords are more frequently seen.
Now, with this deeply rooted Anglican superstition about the heaven
of aristocracy and the angels that dwell therein, Uncle Jacob
naturally desired that his nephew should be qualified for admission
there. And he had a devout belief that the states of probation for a
young soul aspiring to celestial bliss were terms of residence at Eton
and at Oxford.
Little Jacob had continued his custom of staying at Milend every
Sunday, that he might benefit by the services of our friend Mr.
Prigley in the pew at Shayton Church. Isaac Ogden, though he had
come to church three Sundays in succession after the recovery of
little Jacob, and had attended divine service regularly as an officer of
militia (being in that character compulsible thereunto by martial
law), had, I regret to say, relapsed into his old habits of negligence
at Twistle Farm, and spent the Sunday there in following his own
devices. It must be admitted, however, that he did little harm, on
that day or any other, to himself or anybody else. He remained
religiously faithful to his vow of total abstinence, and spent several
hours every day in giving a sound elementary education to his son.
"I'll tell you what it is, Isaac," said Uncle Jacob one day when his
elder brother had come on one of his rare visits to Milend—"I'll tell
you what it is; if you'll just let me have my own way about th'
eddication o' th' young un, I'll leave him all my brass, and, what's
more, I don't mind payin' for his schoolin' beside. I want nowt
nobbut what's reet, but I'll make sich a gentleman on him as there
isn't i' o Shayton nor i' o Manchester nother. And to start wi', I
reckon nowt of his stoppin' up at Twistle Farm same as he is doin'
an' idlin' away auve[19] his time. Let him live at Milend regular for a
twelvemonth, and go to Prigley six hour every day, and then send
him to Eton—that's where gentlefolk sends their lads to. And afther
that, we'll send him to Hoxford College."
 CHAPTER XXXVI.
A SHORT CORRESPONDENCE.
No sooner had Mr. Prigley got into the full swing of work with his
young pupil, than he received a letter from our friend Colonel
Stanburne of Wenderholme:—
"My dear Mr. Prigley,—It would give me great pleasure,
and be of great use to me besides, if you could come over
here and stay with me for a fortnight or three weeks. We
got the house covered in just before the winter, and the
works have been going forward since in some parts of the
interior, but there are some points about internal fittings,
especially in the principal rooms, that I and my architect
don't agree about. Now, what I most want is, the advice
of a competent unprofessional friend; and as I know that
you have studied architecture much more deeply than I
have ever done myself, I look to you to help me. It will
probably be a long time before the house is finished, but
now is the time to decide about the interior arrangements.
Helena is at Lord Adisham's, and so I am left alone with
the architect. I wish you would come. He seems to want
me to adopt a different style for the finishing of the
interior to that which was generally prevalent when
Wenderholme was built. Now my notion is (puisque
l'occasion se présente) to make the place as
homogeneous as possible.
"Do come. You will stay here at the Cottage. I am living
with my mother.
"Very faithfully yours,
John Stanburne."
To this letter, which offered to Mr. Prigley's mind the most tempting
of all possible baits, for he dearly loved to dabble in architecture and
restorations, the reverend gentleman, being bound by his
engagement with the Ogdens, could only regretfully answer:—
"My dear Colonel Stanburne,—I should have accepted your
kind invitation with the greatest pleasure, and the more so
that I take a deep interest in the restoration of your noble
old mansion, but unfortunately I have a private pupil
whom I cannot leave. It is young Jacob Ogden, whose
father is one of your militia officers.
"Yours most truly,
E. Prigley."
But by return of post Mr. Prigley got the following short reply:—
"My dear Mr. Prigley,—The best solution of the difficulty
will be, to bring little Jacob with you. I know little Jacob
very well, and he knows me. Give my compliments to his
father if you have to ask his permission, and tell him we
will take good care of his little boy.
"Yours very faithfully,
J. Stanburne."
So the end of it was, that little Jacob found himself suddenly
removed to Wenderholme Cottage, where old Mrs. Stanburne lived.
The change was highly agreeable to him—not the less agreeable
that the companion of his leisure hours was the beautiful little Edith.
 CHAPTER XXXVII.
AT WENDERHOLME COTTAGE.
Wenderholme Cottage was in fact a very comfortable and
commodious house. Its claims to the humble title which it bore,
were, first, that its front was all gables, with projecting roofs, and
carved or traceried barge-boards; and, secondly, that its rooms were
small. But if they were small they were numerous; and when it
pleased Mrs. Stanburne to receive visitors—and it often pleased that
hospitable lady so to do—it was astonishing how many people the
Cottage could be made to hold.
A little kindness soon wins the affections of a child, and little Jacob
had not been more than three or four days at Wenderholme before
he began to be very fond of Mrs. Stanburne. Hers was just the sort
of influence which is necessary to a young gentleman at that age—
the influence of a woman of experience, who is at the same time a
high-bred gentlewoman. No doubt his old grandmother loved little
Jacob more than any thing else in the world; but she was narrow-
minded, and despotic, and vulgar in all her ways. Mrs. Ogden, too,
had moments of caprice and violence, in which, she was dangerous
to oppose, and difficult to pacify; in short, she was one of those
persons, too common in her class, of whom Matthew Arnold says
that they are deficient in sweetness and light. The steady unfailing
goodness of Mrs. Stanburne, her uniformly gentle manners, her
open intelligent sympathy, produced on her young guest an effect
made ten times more powerful by all his early associations. It was
like coming out of a chamber where every thing was rough and
uncouth, into a pleasant drawing-room, full of light and elegance,
where there are flowers, and music, and books. Such a change
would not be agreeable to every one: whether it would be agreeable
or not depends upon the instinctive preferences. Ladies like Mrs.
Stanburne do not put everybody at his ease, and it proves much in
little Jacob's favor that he felt happy in her presence. As Jacob
Ogden, the elder, had been formed by nature for the rude contest
with reluctant fortune, so his nephew had been created for the
refinements of an attained civilization. Therefore, henceforth, though
he still loved his grandmother, both from gratitude and habit, his
young mind saw clearly that neither her precepts nor her example
were to be accepted as authoritative, and he looked up to Mrs.
Stanburne as his preceptress.
Little Jacob's healthy honest face and simple manners recommended
him to the good lady from the first, and he had not been a week
under her roof before she took a kind interest in every thing
concerning him. The mere facts that he had no mother, no sister, no
brother, and that he had lived alone with his father in such a place
as Twistle Farm, were of themselves enough to attract attention and
awaken curiosity; but the story of his arrival at Wenderholme in the
preceding winter was also known to her, and she knew how
unendurably miserable his lonely home had been. Mrs. Stanburne
talked a good deal with Mr. Prigley about the boy, and learned with
pleasure his father's wonderful and (as now might be hoped)
permanent reformation.
"He does not seem to have neglected the little boy," she said; "he
reads very well. I asked him to read aloud to me yesterday, and was
surprised to hear how well he read—I mean, quite as if he
understood it, and not in the sing-song way children often acquire."
"He's ten years old now, and he ought to read well," replied Mr.
Prigley; "but he knows a great deal for a boy of his age. It's high
time to send him to school, though; it's too lonely for him at the
farm. I am preparing him for Eton."
Mrs. Stanburne expressed some surprise at this. "Boys in his rank in
life don't often go to Eton, do they, Mr. Prigley?"
The clergyman smiled as he answered that little Jacob's rank in life
was not yet definitively settled. Mrs. Stanburne replied that she
thought it was, since his father was a retired tradesman.
"Yes, but his uncle, Mr. Jacob Ogden of Milend, has not left business;
indeed he is greatly extending his business just now, for he has built
an immense new factory. And this little boy is to be his heir—his
uncle told me so himself three weeks since. This child will be a rich
man—nobody can tell how rich. His uncle wishes him to be educated
as a gentleman."
It is a great recommendation to a little boy to be heir to a large
fortune, and Mrs. Stanburne's natural liking for little Jacob was by no
means diminished by a knowledge of that fact. As he was going to
Eton, too, she began to look upon him as already in her own rank of
life, where boys were sent to Eton, and inherited extensive estates.
During Mr. Prigley's frequent absences with Colonel Stanburne at the
Hall, Mrs. Stanburne undertook to hear little Jacob his lessons, and
then the idea struck her that Jacob and Edith might both write
together from her dictation. In this way the boy and the girl became
class-fellows. Edith had a governess usually, but the governess had
gone to visit her relations, and Miss Edith's education was for the
present under the superintendence of her grandmamma.
So between these two children an intimacy rapidly established itself
—an intimacy which affected the course of their whole lives.
One day when they had been left alone together in the drawing-
room, little Jacob asked the young lady some question, and he
began by calling her "Miss Edith."
"Miss Edith!" said she, pouting; "why do you call me Miss? The
servants may call me Miss, but you mayn't. We're school-fellows
now, and you must call me Edith. And I shall call you Jacob. Why
haven't you got a prettier name for me to call you by? Jacob isn't
pretty at all. Haven't you another name?"
Poor little Jacob was obliged to confess his poverty in names. He had
but one, and that one uncouth and unacceptable!
"Only one name. Why, you funny little boy, only to have one name!
I've got four. I'm called Edith Maud Charlotte Elizabeth. But I'll tell
you what I'll do. As I've got four names and you've only one, I'll give
you one of mine. I can't call you Charlotte, you know, because
you're not a girl; but I can call you Charley, and I always will do. So
now I begin. Charley, come here!"
Little Jacob approached obediently.
"Ha, ha! he answers to his new name already!" she cried in delight,
clapping her hands. "What a clever little boy he is! He's a deal
cleverer than the pony was when we changed its name! But then, to
be sure, the pony never properly knew its first name either."
Suddenly she became grave, and put her fingers on the young
gentleman's arm. "Charley," she said, "this must be a secret between
us two, because if grandmamma found out, she might be angry with
me, you know. But you like to be called Charley, don't you? isn't it
nice?"
 CHAPTER XXXVIII.
ARTISTIC INTOXICATION.
The London architect who was charged with the restoration of
Wenderholme gave advice which could not be followed without a
heavy outlay; but in this respect he was surpassed by Colonel
Stanburne's amateur adviser, Mr. Prigley, whose imagination revelled
in the splendors of an ideal Elizabethan interior, full of carving and
tapestry, and all manner of barbaric magnificence. Where the
architect would have been content with paper, Mr. Prigley insisted
upon wainscot; and where the architect admitted plain panelling, the
clergyman would have it carved in fanciful little arches, or imitations
of folded napkins, or shields of arms, or large medallion portraits of
the kings of England, or bas-reliefs of history or the chase.
Only consider what Mr. Prigley's tastes and circumstances had been,
and what a painful contradiction had ever subsisted between them!
He had an intense passion for art—not for painting or sculpture in
their independent form, for of these he knew little—but Mr. Prigley
loved architecture mainly, and then all the other arts as they could
help the effect of architecture. With these tastes he lived in a degree
of poverty which utterly forbade any practical realization of them,
and surrounded by buildings of which it is enough to say that they
represented the taste of the inhabitants of Shayton. The ugliest
towns in the world are English towns—the ugliest towns in England
are in the manufacturing district—the ugliest town in the
manufacturing district was the one consigned to Mr. Prigley's
spiritual care. Here his artistic tastes dwelt in a state of suppression,
like Jack-in-the-box. Colonel Stanburne had imprudently unfastened
the lid; it flew open, and Jack sprang up with a suddenness and an
energy that was positively startling and alarming.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com