Elements of concave analysis and

applications Prem K. Kythe

Visit to download the full and correct content document:
https://textbookfull.com/product/elements-of-concave-analysis-and-applications-prem-
k-kythe/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Analysis and Implementation of Isogeometric Boundary

Elements for Electromagnetism Felix Wolf

https://textbookfull.com/product/analysis-and-implementation-of-
isogeometric-boundary-elements-for-electromagnetism-felix-wolf/

Advanced Flight Dynamics with Elements of Flight

Control 1st Edition Nandan K. Sinha

https://textbookfull.com/product/advanced-flight-dynamics-with-
elements-of-flight-control-1st-edition-nandan-k-sinha/

Seismic Analysis of Structures and Equipment Praveen K.

Malhotra

https://textbookfull.com/product/seismic-analysis-of-structures-
and-equipment-praveen-k-malhotra/

Newborn Surgery, Fourth Edition Prem Puri

https://textbookfull.com/product/newborn-surgery-fourth-edition-
prem-puri/
Climate Change and the White World Prem Shankar Goel

https://textbookfull.com/product/climate-change-and-the-white-
world-prem-shankar-goel/

Advance Elements of Optoisolation Circuits Nonlinearity

Applications in Engineering 1st Edition Ofer Aluf
(Auth.)

https://textbookfull.com/product/advance-elements-of-
optoisolation-circuits-nonlinearity-applications-in-
engineering-1st-edition-ofer-aluf-auth/

Vibration theory and applications with finite elements

and active vibration control 1st Edition Palazzolo

https://textbookfull.com/product/vibration-theory-and-
applications-with-finite-elements-and-active-vibration-
control-1st-edition-palazzolo/

Freshwater Ecology Concepts and Environmental

Applications of Limnology Walter K Dodds

https://textbookfull.com/product/freshwater-ecology-concepts-and-
environmental-applications-of-limnology-walter-k-dodds/

Comprehensive inorganic chemistry II from elements to

applications 2nd Edition Jan Reedijk

https://textbookfull.com/product/comprehensive-inorganic-
chemistry-ii-from-elements-to-applications-2nd-edition-jan-
reedijk/
 Elements of
Concave Analysis
and Applications
 Elements of
Concave Analysis
and Applications

Prem K. Kythe
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2018 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed on acid-free paper

Version Date: 20180414

International Standard Book Number-13: 978-1-138-70528-9 (Hardback)

This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information
storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access
www.copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization
that provides licenses and registration for a variety of users. For organizations that have been granted
a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging-in-Publication Data

Names: Kythe, Prem K., author.

Title: Elements of concave analysis and applications / Prem K. Kythe.
Description: Boca Raton, Florida : CRC Press, [2018] | Includes
bibliographical references and index.
Identifiers: LCCN 2018006905| ISBN 9781138705289 (hardback : alk. paper) |
ISBN 9781315202259 (ebook : alk. paper)
Subjects: LCSH: Concave functions--Textbooks. | Convex functions--Textbooks.
| Functions of real variables--Textbooks. | Matrices--Textbooks.
Classification: LCC QA353.C64 K96 2018 | DDC 515/.88--dc23
LC record available at https://lccn.loc.gov/2018006905

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
 To the memory of
Dr. James H. & Mrs. Mickey Abbott
with reverence and love
Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Notations, Definitions, and Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
1 Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Matrix Inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3.1 Cofactor Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Systems of Linear Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4.1 Solution with the Inverse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.2 Cramer’s Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.3 Gaussian Elimination Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Definite and Semidefinite Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.6 Special Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6.1 Jacobian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6.2 Hessian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.6.3 Bordered Hessian: Two Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.6.4 Bordered Hessian: Single Function . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2 Differential Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.1 Limit of a Function at a Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2 Theorems on Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.2.1 Limit at Infinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.2 Infinite Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3 Global and Local Extrema of Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.4 First and Second Derivative Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.4.1 Definition of Concavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.5 Vector-Valued Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.5.1 Geometric Meaning of the Inflection Point . . . . . . . . . . . . . . . . . . . 40
2.6 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.7 Multivariate Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.7.1 Geometric Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
viii CONTENTS

2.7.2 Gradient at a Point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

2.8 Mathematical Economics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.8.1 Isocost Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.8.2 Supply and Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.8.3 IS-LM Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.8.4 Marginal of an Economic Function . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.8.5 Marginal Rate of Technical Substitution . . . . . . . . . . . . . . . . . . . . . 50
2.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3 Concave and Convex Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.1 Convex Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.2 Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2.1 Properties of Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.3 Jensen’s Inequality for Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . 69
3.4 Convex Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.4.1 Properties of Convex Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.4.2 Jensen’s Inequality for Convex Functions . . . . . . . . . . . . . . . . . . . . 73
3.5 Differentiable Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.6 Unconstrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4 Concave Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.1 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.1.1 Unconstrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.2 Method of Lagrange Multipliers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.2.1 Constrained Optimization with Equality Constraint . . . . . . . . . . 89
4.3 Karush-Kuhn-Tucker Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.3.1 Equality and Inequality Constraints . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.3.2 Necessary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.3.3 Regularity Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.3.4 Sufficient Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.4 Inequality Constrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.5 Application to Mathematical Economics . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.5.1 Peak Load Pricing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.6 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5 Convex Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.1 Minimization Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.1.1 Unconstrained Minimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.1.2 Equality Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.1.3 Equality Constraints: General Case . . . . . . . . . . . . . . . . . . . . . . . . 123
5.1.4 Inequality Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5.1.5 General Linear Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.2 Nonlinear Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.2.1 Two Inequality Constraints and One Equality Constraint . . . 131
5.2.2 Two Inequality Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
 CONTENTS ix

5.3 Fritz John Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

5.3.1 Feasibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.3.2 Slater’s Condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
5.4 Lagrangian Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
5.4.1 Geometrical Interpretation of Duality . . . . . . . . . . . . . . . . . . . . . . 141
5.4.2 Saddle Point Sufficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.4.3 Strong Duality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6 Quasi-Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.1 Quasi-Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.2 Differentiable Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
6.3 Theorems on Quasi-Concavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.4 Three-Dimensional Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.5 Multivariate Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.6 Sums of Quasi-Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
6.7 Strictly Quasi-Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.7.1 Sums of Strictly Quasi-Concave Functions . . . . . . . . . . . . . . . . . . 167
6.8 Quasi-Concave Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
6.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
6.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
7 Quasi-Convex Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
7.1 Quasi-Convex Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
7.2 Properties of Quasi-Convex Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.3 Bordered Hessian Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
7.3.1 Properties of the Bordered Hessian . . . . . . . . . . . . . . . . . . . . . . . . . 186
7.4 Quasi-Convex Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.4.1 No Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.4.2 Equality Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
7.4.3 Inequality Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7.4.4 Convex Feasibility Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7.4.5 Equality and Inequality Constraints . . . . . . . . . . . . . . . . . . . . . . . . 190
7.4.6 Minmax Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
7.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
8 Log-Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
8.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
8.1.1 Log-Concavity Preserving Operations . . . . . . . . . . . . . . . . . . . . . . 198
8.2 Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.2.1 General Results on Log-Concavity . . . . . . . . . . . . . . . . . . . . . . . . . . 200
8.2.2 Log-Concavity of Density and Left-Side Integral . . . . . . . . . . . . . 202
8.2.3 Reliability Theory and Right-Side Integral . . . . . . . . . . . . . . . . . . 203
8.2.4 Mean Residual Lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
8.3 Asplund Sum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
8.3.1 Derivatives of Integrals of Log-Concave Functions . . . . . . . . . . 204
x CONTENTS

8.3.2 Adding Log-Concave Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

8.3.3 Asplund Sum and Conjugate Functions . . . . . . . . . . . . . . . . . . . . . 205
8.3.4 Integral Functional . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
8.3.5 Area Measure of Log-Concave Functions . . . . . . . . . . . . . . . . . . . . 207
8.4 Log-Concavity of Nonnegative Sequences . . . . . . . . . . . . . . . . . . . . . . . . 207
8.5 Log-Concave Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
7.10 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
9 Quadratic Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
9.1 Quadratic Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
9.2 Hildreth-D’Esopo Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
9.3 Beale’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
9.4 Wolfe’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
9.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
10 Optimal Control Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
10.1 Hamiltonian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
10.2 Optimal Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
10.2.1 Sufficient Conditions for Optimization . . . . . . . . . . . . . . . . . . . . . 235
10.3 Free Endpoint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
10.4 Inequality Constraints at the Endpoints . . . . . . . . . . . . . . . . . . . . . . . 237
10.5 Discounted Optimal Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
10.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
11 Demands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
11.1 Shephard’s Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
11.2 Marshallian Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
11.3 Hicksian Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
11.4 Slutsky Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
11.4.1 Giffen Goods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
11.4.2 Veblen Goods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
11.5 Walrasian Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
11.6 Cost Minimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
11.7 Expenditure Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
11.8 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
11.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
12 Black-Scholes Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
12.1 Black-Scholes Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
12.1.2 Itô’s Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
12.1.2 Derivation of Black-Scholes Equation . . . . . . . . . . . . . . . . . . . . . . 272
12.2 Solution of Black-Scholes Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
12.2.1 Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
12.2.2 Solution of the Heat Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
12.2.3 Black-Scholes Call Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
12.2.4 Some Finance Terms and Arbitrage . . . . . . . . . . . . . . . . . . . . . . . 277
12.2.5 Self-Financing Portfolio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
12.2.6 Implied Volatility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
 CONTENTS xi

12.3 Black-Scholes Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282

12.4 Use of Greek Letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
12.5 Log-normal Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
12.5.1 Log-normal c.d.f and p.d.f . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
12.5.2 Log-normal Conditional Expected Value . . . . . . . . . . . . . . . . . . . 286
12.6 Black-Scholes Call Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
12.6.1 Black-Scholes Economy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
12.6.2 Black-Scholes under a Different Numéraire . . . . . . . . . . . . . . . . 288
12.6.3 Black-Scholes by Direct Integration . . . . . . . . . . . . . . . . . . . . . . . 290
12.6.4 Feynman-Kac Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
12.6.5 CAPM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
12.6.6 CAPM for Assets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
12.7 Dividends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
12.7.1 Continuous Dividends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294
12.7.2 Lumpy Dividends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
12.8 Solutions of SDEs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
12.8.1 Stock Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
12.8.2 Bond Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
12.8.3 Discounted Stock Price . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
12.8.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
12.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
A Probability Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
B Differentiation of Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
C Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
D Laplace Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
E Implicit Function Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
F Locally Nonsatiated Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
Preface

This textbook on concave analysis aims at two goals. Firstly, it provides sim-
ple yet comprehensive subject matter to the readers who are undergraduate
seniors and beginning graduate students in mathematical economics and busi-
ness mathematics. For most readers the only prerequisites are courses in ma-
trix algebra and differential calculus including partial differentiation; however,
for the last chapter a thorough working knowledge of linear partial differen-
tial equations and the Laplace transforms is required. The readers can omit
this chapter if not required. The subject of the book centers mostly around
concave and convex optimization; other related topics are also included. The
details are provided below in the overview section.
Although there are many excellent books on the market, almost all of
them are at times difficult to understand. They are very heavy on theoreti-
cal aspects, and generally fail to provide ample worked-out examples to give
readers easy understanding and workability.
The second goal is elucidated below in the section ‘To Readers’.

Motivation
The subject of convexity and quasi-convexity has been a model for economic
theorists to make decisions about cost minimization and revenue maximiza-
tion. This has resulted in a lot of publications in convex optimization. So
why is there keen interest in concave and quasi-concave functions? Firstly,
economic theory dictates that all utility functions are quasi-convex and that
all cost functions are concave in input prices. Therefore, a cost function that
is not concave in input prices is not a cost function. Secondly, the standard
model in economic theory consists in a set of alternatives and an ordering of
these alternatives, according to different priorities and interests. The process
that a decision maker follows is to choose a favorite alternative with the prop-
erty that no other alternative exceeds the ordering. In such a situation the
decision maker often uses a function that ‘represents’ this ordering. Thus, for
example, suppose there are four alternatives, say, a, b, c and d, and suppose
that the decision maker prefers a to b and treats both c and d as equally
desirable. Any function, like f , with f (a) > f (b) > f (c) = f (d) may rep-
xiv PREFACE

resent the ordering, irrespective of whatever numerical values (level curves)

such ordering has. However, the situation changes when the decision maker is
a consumer who is choosing between different goods and prices. In this case
the consumer’s ordering, based on the level curves (or indifference curves) of
chosen alternatives, can look quite different from that of the businessperson.
It so happens that the consumer’s ordering becomes concave (less expense,
more goods). But this situation implies that any function that represents the
consumer’s interest is, at best, quasi-concave. This natural phenomenon is, of
course, based on a deep but simple result that concave and convex functions
are diametrically opposite to each other in behavior and intent.
However, the subject of concave analysis, with emphasis on concave, quasi-
concave and log-concave functions, has appeal to both the consumer and busi-
ness organizations.

Overview
A general description of the topics covered in the book is as follows: Chap-
ter 1 introduces a review of matrix algebra that includes definitions, matrix
inversion, solutions of systems of linear algebraic equations, definite and semi-
definite matrices, Jacobian, two types of Hessian matrices, and the Hessian
test. Chapter 2 is a review of calculus, with topics dealing with limits, deriva-
tive, global and local extrema, first and second derivative tests, vector-valued
functions, optimization, multivariate functions, and basic concepts of mathe-
matical economics.
Concave and convex functions are introduced in Chapter 3, starting with
the notion of convex sets, Jensen’s inequalities for both concave and convex
functions, and unconstrained optimization. Chapter 4 deals with concave
programming; it is devoted to optimization problems on maximization mostly
with inequality constraints, and using the Lagrange method of multipliers and
the KKT necessary and sufficient conditions. Applications to mathematical
economics include the topic of peak price loading, and comparative statics is
discussed. Optimization problems focusing on minimization are introduced
in Chapter 5 on convex programming, in order to compare it with concave
optimization. Nonlinear programming is discussed; the Fritz John and Slater
conditions are presented, and the topic of Lagrangian duality is discussed.
Chapters 6 and 7 deal with quasi-concave and quasi-convex functions.
Both topics are important in their own applications. The single-function
bordered Hessian test on quasi-concavity and quasi-convexity is presented,
and optimization problems with types of functions and the minmax theorem
are provided. Chapter 8 deals with log-concave functions; general results
on log-concavity are presented, with application on mean residual life; and
the Asplund sum is introduced, with its algebra, derivatives, and area mea-
sure. Log-concavity of nonnegative sequences is discussed, and all log-concave
 PREFACE xv

probability distributions with their density functions and cumulative density

functions are presented in detail.
Chapter 9 deals with the quadratic programming to optimization prob-
lems, and presents the following numerical methods: (i) Hildreth-D’Esopo
method; (ii) Beale’s method; and (iii) Wolfe’s method. The optimal con-
trol theory is discussed in Chapter 10, using the Hamiltonian and different
types of optimization problems. Three types of demands, namely Marshal-
lian, Hicksian, and Walrasian, are introduced in Chapter 11, using the Shep-
hard’s lemma and the Slusky equation with applications to cost minimiza-
tion, and the Giffen and Veblen goods. Chapter 12 is exclusively devoted to
the Black-Scholes differential equation, its solution, log-normal distribution,
Black-Scholes call price, Feynman-Kac theorem, capital asset pricing model,
dividends, stocks, bonds, and discounted stock prices.
There are six appendices: (A) Some useful topics on probability; (B) differ-
entiation of operators involving the Gateaux differential, Fréchet derivative,
the concept of the gradient, and Taylor’s series first- and second-order approx-
imations; (C) a list of probability distributions; (D) a self-contained detailed
discussion of the Laplace transforms; (E) implicit function theorem; and (F)
locally nonsatiated function. The bibliography toward the end of the book
contains the references cited in the book, followed by the Index.
The book contains over 330 examples and exercises; most exercises are
provided with solutions, simpler ones with hints and answers. Since the book
uses and discusses vectors and matrices, care is taken, in order to avoid confu-
sion, to set all (row/column) vectors and matrices in bold lowercase and bold
uppercase, respectively.
This is an introductory textbook that provides a good combination of
methodology, applications, and hands-on projects for students with diverse
interests from mathematical economics, business mathematics, engineering,
and other related applied mathematics courses.

To Readers
The second goal concerns specifically the abuse and misuse of a couple of
standard mathematical notations in this field of scientific study. They are
the gradient ∇f and the Laplacian ∇2 f of a function f (x) in Rn . Somehow,
and somewhere, a tradition started to replace the first-order partials of the
function f by its gradient ∇f . It seems that this tradition started without
any rigorous mathematical argument in its support. This book has provided
a result (Theorem 2.18) that establishes that only under a specific necessary
condition the column vector [∂f /∂x1 · · · ∂f /∂xn ] can replace the gradient
vector ∇f , and these two quantities, although isomorphic to each other, are
not equal. Moreover, it is shown that any indiscriminate replacement between
these two quantities leads to certain incorrect results (§3.5).
xvi PREFACE

The other misuse deals with the Laplacian ∇2 f , which has been used to
represent the Hessian matrix (§1.6.2), without realizing that ∇2 f is the trace
(i.e., sum of the diagonal elements) of the Hessian matrix itself. This abuse
makes a part equal to the whole. Moreover, ∇2 is the well-known linear partial
differential operator of the elliptic type known as the Laplacian.
It appears that this misuse perhaps happened because of the term ‘vector’,
which is used (i) as a scalar quantity, having only magnitude, as in the row
or column vectors (in the sense of a matrix), and (ii) as a physical quantity,
such as force, velocity, acceleration, and momentum, having both magnitude
and direction. The other factor for the abuse in the case of the gradient is
the above-mentioned linear isomorphic mapping between the gradient vector
∇f and the (scalar) column vector [∂f /∂x1 · · · ∂f /∂xn ]T . This isomorphism
has been then literally used as ‘equality’ between these two quantities. Once
the case for ∇f became the tradition, the next choice ∇2 f for the Hessian
matrix became another obvious, but incorrect, tradition.
As readers, you will find an attention symbol, !!! , at different parts of the
book. It is used to point out the significance of the statements found there.
The other less important notations are the ≺, the ⊕ and the ⊙ symbols.
Although borrowed from physics and astronomy, these symbols are acceptable
with a different but almost similar meaning provided that they are properly
defined as given in the section on Notations. Moreover, the ⊕ and the ⊙
symbols have now become so common due to the advancement in cell phones
and related electronic technology that they are probably losing their rigorous
mathematical significance.

Acknowledgments
I take this opportunity to thank Mr. Sarfraz Khan, Executive Editor, Taylor
& Francis, for his support, and Mr. Callum Fraser for coordinating the book
project. I also thank the Project Editor Michele A. Dimont for doing a great
job of editing the text. Thanks are due to the reviewers and to some of my
colleagues who made some very valuable suggestions to improve the book.
Lastly, I thank my friend Michael R. Schäferkotter for help and advice freely
given whenever needed.

Prem K. Kythe
Notations, Definitions, and Acronyms

A list of the notations, definitions, abbreviations, and acronyms used in this

book is given below.

a.s., almost surely

A, matrix
AT , transpose of the matrix A
|A|, determinant of a matrix A
adj A, adjoint of a matrix A
det(A) or |A|, determinant of a matrix A
A−1 , inverse of a matrix A
aij , element of a matrix A in the ith row and jth column
B(c, r), ball with center c and radius r
B(X, Y ), class of all bounded linear operators from X into Y
B, bordered Hessian matrix: one function
C 0 (D), class of functions continuous on a region D
C k (D), class of continuous functions with kth continuous derivative
on a region D, 0 ≤ k < ∞
C ∞ (D), class of continuous functions infinitely differentiable on a region D
C-function, same as a C 0 -function; continuous function
C, cofactor matrix
Cij , cofactor of the element in the ith row and jth column
C T , transpose of C = adj A
Ci , eigenvectors
C, cost; consumption
C(Q), production cost
CAPM, capital asset pricing model
CES, constant elasticity of substitution
CLPD, constant positive linear dependence constraint qualification
CLT, central limit theorem
CQ, constraint qualification
CRCQ, constant rank constraint qualification
CU, concave upward
CD, concave downward
xviii NOTATIONS, DEFINITIONS, AND ACRONYMS

Cov, covariance
c.d.f., cumulative distribution function
c(w, y), cost function
Dt , derivative with respect to t
Df (x), derivative of f (x) in Rn
D, aggregated demand
D, domain, usually in the z-plane
dist(A, B), distance between points (or sets) A and B
dom(f ), domain of a function f
DRS, decreasing return to scale
e, expenditure function
E, amount allocated for expenditure
E[X], expected value of a random vector X
E(f ), entropy of f
Eq(s)., Equation(s) (when followed by an equation number)
ei , ith unit vector, i = 1, . . . , n
[e], set of the unit vectors ei in Rn
epi(f ), epigraph of f
e(p, u), expenditure function
F , field
f : X 7→ Y , function f maps the set X into (onto) the set Y
f ◦ g, composite function of f and g: (f ◦ g)(·) = f (g(·))
f ′ , first derivative of f
f ′′ , second derivative of f
f (n) , nth derivative of f
∂f (x)
, first-order partials of f in Rn , also written fi , for i = 1, . . . , n; also
∂xi
∂f ∂f ∂f
written as fx , fy , fx for , , in R3
∂x ∂y ∂z
∂ 2 f (x)
, second-order partials of f in Rn , also written as fij for i, j = 1, . . . , n;
∂xi ∂xj
∂2f ∂2f ∂2f
also written as fxx , fyy , fzx for , , in R3
∂x2 ∂y 2 ∂z 2
(f ◦ g)(x),= f (g(x)), composition of functions f and g
Rt Rt
f ⋆ g, convolution of f (t) and g(t) (= 0 f (t − u)g(u) du = 0 f (ug (t − u) du =
L−1 {G(s)F (s)})
FJ, Fritz John conditions R∞
F (s), Laplace transform of f (t) (= 0 est f (t) dt)
G, government expenditure; constrained set
Gmin , positive minimal accepted level of profit
G(·; ·), Green’s function
 NOTATIONS, DEFINITIONS, AND ACRONYMS xix

Geom(p), geometric distribution with probability distribution

H, Hamiltonian
hyp, hypograph of f
hj (p, u), Hicksian demand for good j
h, hours of work, h = T − τ
H, Hessian matrix
H̄, bordered Hessian matrix: two functions, f and the constraint g
iff, if and only if
i, income
I, income
I(p), e-beam intensity at position p
IK (x) = 0 if x ∈ K; ∞ if x ∈ /K
IS, commodity equilibrium
Int (D), interior of a domain D
IRS, increasing return to scale
I, identity matrix
j , goods number; input factor
|J|, or J, or simply J, Jacobian determinant
K, capital
KKT, Karush-Kuhn-Tucker conditions
L, Lagrangian function; also, labor
∂L ∂L
Lx , Ly , = , first-order partials of L
∂x ∂y
∂ 2 L ∂ 2 L ∂ 2L ∂ 2 L
Lxx , Lyy , Lxy , Lyx , = , , , , second-order partials of L
∂x2 ∂y 2 ∂x∂y ∂y∂x
L{f (t)}, Laplace transform of f (t), also denoted by F (s)
L−1 {f (t)}, inverse Laplace transform, also denoted by f (t)
L2 {f (t); s}, Laplace2 transform (also known as L2 -transform) of f (t; s)
LICQ, linear independence constraint qualification
LM, monetary equilibrium
LM, monetary equilibrium
LQC, linearity constrained qualifications
mrl(x), mean residual lifetime function
Mij , minor of the element aij of a matrix A
Md , demand for money
Ms , supply of money
Mt , transition-precautionary demand for money
Mz , speculative demand for money
MC, marginal cost
MFCQ, Mangasarian-Fromovitz constraint qualification
ME, marginal expenditure
xx NOTATIONS, DEFINITIONS, AND ACRONYMS

MPC, marginal propensity to consume

MR, marginal revenue
MRTS, marginal rate of technical substitution
N, set of natural numbers (positive integers)
N(A), null space of matrix A
N , numéraire
ND, negative definite
NSD, negative semidefinite
p.d.f., probability density function, or density function
p, prices (a vector with component pj )
P , profit; supply; nonlabor income
PV , probability function for a random variable V
PK price for K
PL , price for L
P, probability
PD, positive definite
PSD, positive semidefinite
q, Cobb-Douglas production function; also, amount of consumed good
Q, quantity of output produced; risk neutral measure
Qs , supply
Qd , demand
Q, level of output
QNCQ, quasi-normality constraint qualification
QP, quadratic programming
r, eigenvalue of a matrix A
r(t), vector-valued function (= f (t)i + g(t)j + h(t)k)
R, revenue
R(Q), sales revenue
R(f ), range of function f
R, the real line; real plane
R3 , three-dimensional real space
Rn , n-dimensional real space
R+ , nonnegative real numbers
R++ , positive real numbers
s, variable of Laplace transform
S, slackness variable; savings
S(p, w), Slutsky matrix
SC, Slater condition
SDE, stochastic differential equation
SOSC, second-order sufficient condition
tr(A), trace of a matrix A (sum of the diagonal elements)
 NOTATIONS, DEFINITIONS, AND ACRONYMS xxi

T , total time available

TC, total cost
TE, total expenditure
TP, total product
TPk , totally positive of order k
TR, total revenue
u, utility
u̇, time-derivative of a function u
U (y | f ), upper contour set of f at y
Uα , upper-level set
v(p, m), indirect utility function
V , value function
w, wage (a vector with components wj );
wh, labor income(wage/hour)
x, quantity (vector)
xj (w, y), conditional factor demand for each input factor or good j
⌊x⌋, greatest integer function, floor of x
x, eigenvector of a matrix A; a point in R
x∗ , critical number, critical point (maximizer or minimizer)
x  y, componentwise inequality between vectors x and y
x ≻ y, strict componentwise inequality between vectors x and y
X, real normed space
X, exports
y, output;
Y , income
Yc , convex hull of Y ⊂ X
Z, imports
Z, set of integers
Z+ , set of nonnegative integers
δf (x, h), Gateaux differential (or G-differential) of f at x and h real
λ, λ Lagrange multiplier(s)
µ, µ Lagrange multiplier(s); also measure, Lebesgue measure
π, profit
ρ, rank of a matrix; ρ(A), rank of the matrix A; correlation coefficient
Σ, σ-algebra
τ , labor; also, time spent for leisure
χK (x) = 1 if x ∈ K; 0 if x ∈ / K, characteristic function defined on a convex
set K
k · k, norm
0, null (or zero) vector, = [0 0 . . . 0] in Rn
xxii NOTATIONS, DEFINITIONS, AND ACRONYMS

∂ ∂ ∂
∇, ‘del’ operator, ∇ = i + j + k , ((x, y, z) ∈ R3 ); an operator defined
∂x ∂y ∂z
∂ ∂
in Rn as ∇ = e1 + · · · + en
∂x1 ∂xn
∂f ∂f
∇f , gradient of a function f , a vector in R3 , defined by ∇ = i +j +
∂x ∂y
∂f ∂f ∂f
k ; a vector in Rn defined by ∇ = e1 + · · · + en for x =
∂z ∂x1 ∂xn
∂f ∂f
(x1 , . . . , xn ) ∈ Rn (dimension 1 × n), or by ∇ = e1 + · · · + en for
∂x1 ∂xn
x = (x1 , . . . , xn ) ∈ Rn (dimension n × 1)
∂2 ∂2
∇2 , Laplacian operator defined on Rn as + · · ·+ ; it is a linear elliptic
∂x21 ∂x2n
partial differential operator
∂2f ∂2f
∇2 f (x) = 2 + ··· + , x = (x1 , . . . , xn ) ∈ Rn , Laplacian of f (x); also
∂x1 ∂x2n
the trace of the Hessian matrix H
kxk1 , l1 -norm of a vector x
kxk2 , l2 -norm, or Euclidean norm, of a vector x
kxk∞ , l∞ -norm of a vector x
≻, , (subordination (predecessor): A  B, matrix inequality between ma-
trices A and B; A ≻ B, strict matrix inequality between matrices A and
B
≺, , subordination (successor) , e.g., f ≺ g is equivalent to f (0) = g(0) and
f (E) ⊂ g(E), where E is the open disks ; but here x ≺ y is used for
componentwise strict inequality, and x  y for componentwise inequality
between vectors x and y
(f ⊕ g)(z), = sup{f (x)g(y)}, where f and g are log-concave functions
x+y
x

(s ⊙ f )(x), = sf , where f is a log-concave function, and s > 0
s
n

n! n


k , binomial coefficient = k! (n − k)! = n−k
iso iso
= , isomorphic to; for example, A = B means A is isomorphic to B, and
conversely
 end of a proof, or an example
!!! attention symbol
1
Matrix Algebra

Some basic concepts and results from linear and matrix algebra, and from
finite-dimensional vector spaces are presented. Proofs for most of the results
can be found in many books, for example, Bellman [1970], Halmos [1958],
Hoffman and Kunze [1961] Lipschutz [1968], and Michel and Herget [2007].

1.1 Definitions
A matrix A is a rectangular array of elements (numbers, parameters, or vari-
ables), where the elements in a horizontal line are called rows, and those in
a vertical line columns. The dimension of a matrix is defined by the number
of rows m and the number of columns n, and we say that such a matrix has
dimension m × n, or simply that the matrix is m × n. If m = n, then we have
a square matrix. If the matrix is 1 × n, we call it a row vector, and if the
matrix is m × 1, then it is called a column vector. A matrix that converts the
rows of a matrix A to columns and the columns of A to rows is called the
transpose of A and is denoted by AT .
Let two 3 × 3 matrices A and B be defined as
   
a11 a12 a13 b11 b12 b13
A =  a21 a22 a23  , B =  b21 b22 b23  . (1.1.1)
a31 a32 a33 b31 b32 b33

Then, for example,

 
a11 a21 a31
AT =  a12 a22 a32  ,
a13 a23 a33

and similarly for BT . For addition (or subtraction) of two matrices (A + B,

or A − B) the two matrices A and B must be of equal dimension. Each
element of one matrix (B) is added to (or subtracted from) the corresponding
element of the other matrix (A). Thus, the element b11 in B is added to (or
2 1 MATRIX ALGEBRA

subtracted from) a11 in A; b12 to (or from) a12 , and so on. Multiplication of
a matrix by a number or scalar involves multiplication of each element of the
matrix by the scalar, and it is called scalar multiplication, since it scales the
matrix up or down by the size of the scalar.
A row vector A and a column vector B are written, respectively, as
 
b11
A = [ a11 a12 a13 ]1×3 , B =  b21  .
b31 3×1

However, to save space, B is often written as its transpose, i.e.,

T
B = [ b11 b21 b31 ] .

Multiplication of a row vector A by a column vector B requires that each

vector has the same number of elements. This multiplication is then carried
out by multiplying each element of the row vector by its corresponding element
in the column vector and summing the product. Thus,
T
AB = [ a11 a12 a13 ]1×3 × [ b11 b21 b31 ]3×1
= [a11 b11 + a12 b21 + a13 b31 ]1×1 .

The above technique of multiplication of a row and a column vector is used

to obtain the multiplication of any two vectors with a precondition that the
number of rows and columns of one matrix must be the same as the number
of columns and rows of the other matrix. Then the two matrices are said to
be conformable for multiplication. Thus, an m × n matrix can be multiplied
by another n × m matrix, and the resulting matrix will be m × m.
Example 1.1. Given
 
  5 13  
3 6 11 1 4 7
A= , B = 7 8  , E= ,
12 8 5 2 4 9
2×3 9 10 3×2 2×3

the matrices A and B, and B and E are conformable for multiplication, but
A and C are not conformable. Thus,
   
3 · 5 + 6 · 7 + 11 · 9 3 · 13 + 6 · 8 + 11 · 10 156 97
AB = = ,
12 · 5 + 8 · 7 + 5 · 9 12 · 13 + 8 · 8 + 5 · 10 161 270 2×2
   
5 · 1 + 13 · 2 5 · 4 + 13 · 4 5 · 7 + 13 × 9 31 72 152
BE =  7 · 1 + 8 · 2 7·4+8·4 7 · 7 + 8 × 9  =  23 60 121  .
9 · 1 + 10 · 2 9 · 4 + 10 · 4 9 · 7 + 10 × 9 29 76 153 3×3
 1.2 PROPERTIES 3

Example 1.2. In business, one method of keeping track of sales of different

types of products at different outlets is to keep the inventory in the form of a
matrix. Thus, suppose that a construction company has four different outlets
selling (a) bricks, (b) lumber, (c) cement, and (d) roof shingles. The inventory
and the price of each item are expressed as matrices A and P:

(a) (b) (c) (d)

   
100 110 80 115 220
 210 230 150 400   65 
A =  , P=  .
165 95 68 145 114
150 190 130 300 4×4 168 4×1

Since the two matrices are conformable, their product AP will give the values
V (in dollars) of the stock at each outlet:
   
100 · 220 + 110 · 65 + 80 · 114 + 115 · 168 57, 590
 210 · 220 + 230 · 65 + 150 · 114 + 400 · 168   145, 450 
V = AP =  = . 
165 · 220 + 95 · 65 + 68 · 114 + 145 · 168 74, 587
150 · 220 + 190 · 65 + 130 · 114 + 300 · 168 110, 520

1.2 Properties
The following properties of matrices are useful.
1. Matrix addition is commutative and associative i.e., A + B = B + A, and
(A+B)+C = A+(B+C). These properties also hold for matrix subtraction,
since A − B = A + (−B).
2. Matrix multiplication, with a few exceptions, is not commutative, i.e.,
AB 6= BA. Scalar multiplication is commutative, i.e., cA = Ac. If three or
more matrices are conformable, i.e., if Aj×k , Bm×n , Cp×q , where k = m and
n = p, the associative law applies as long as matrices are multiplied in their
order of conformability. Thus, (AB)C = A(BC). Under the same conditions
the matrix multiplication is also distributive, i.e., A(B + C) = AB + BC.
Example 1.3. Given
  
7 4   7
4 3 10
A= 1 5
  , B= , C = 8 ,
3 5 6 2×3
8 9 3×2 9 3×1

we get
   
7·4+4·3 7·3+4·5 7 · 10 + 4 · 6 40 41 94
AB =  1 · 4 + 5 · 3 1 · 3 + 5 · 5 1 · 10 + 5 · 6  =  19 28 40  ;
8·4+9·3 8·3+9·5 8 · 10 + 9 · 6 59 69 134 3×3
4 1 MATRIX ALGEBRA
 
     
40 41 94 7 40 · 7 + 41 · 8 + 94 · 9 1454
(AB)C =  19 28 40   8  =  19 · 7 + 28 · 8 + 40 · 9 = 717  ;
59 69 134 3×3 9 3×1 59 · 7 + 69 · 8 + 134 · 9 2171 3×1
 
  7    
4 3 10 4 · 7 + 3 · 8 + 10 · 9 142
BC =  8  = = ;
3 5 6 2×3 3·7+5·8+6·9 115 2×1
9 3×1
     
7 4   7 · 142 + 4 · 115 1454
142
A(BC) =  1 5  =  1 · 142 + 5 · 115  =  717  .
115
8 9 3×2 8 · 142 + 9 · 115 3×1 2171 3×1
3. An identity matrix I is a square matrix whose diagonal elements are all
1 and all remaining elements are 0. An n × n identity matrix is sometimes
denoted by In . The identity matrix I is the unity in matrix algebra just as
the numeral 1 is the unity in algebra. Thus, the multiplication of a matrix
by an identity matrix leaves the original matrix unchanged; so also the multi-
plication of an identity matrix by itself leaves the identity matrix unchanged.
Hence, AI = IA = A, and I × I = I2 = I.
4. A matrix A for which A = AT is called a symmetric matrix. A symmetric
matrix A for which A × A = A is an idempotent matrix. The identity matrix
I is both symmetric and idempotent.
5. A null matrix 0 is composed of all 0s and can have any dimension; it is not
necessarily square. Obviously, addition or subtraction of null matrices leaves
the original matrix unchanged; multiplication by a null matrix yields a null
matrix. A scalar zero 0 has dimension 1 × 1.
6. A matrix with zero elements everywhere below (or above) the principal
diagonal is called upper (or lower) triangular matrix, also known as upper or
lower echelon form. Thus,
   
a11 a12 ··· a1,n−1 a1,n a11 0 ··· 0 0
 0 a22 ··· a2,n−1 a2n   a21 a22 ··· 0 0 
 , or  
··· ··· ··· ··· ··· ··· ··· ··· ··· ···
0 0 ··· 0 ann an1 an2 ··· an,n−1 ann

represent the upper or lower triangular matrix.

7. The sum of all the elements on the principal diagonal of an n × n matrix
A is called the trace of the matrix A and is denoted by tr(A).
See Exercise 1.6 for the dimensions of some most used expressions in Rn .
Remember that dimensions are always listed row by column, and a vector a
is a column vector, and scalars are of dimension 1 × 1.
 1.3 MATRIX INVERSION 5

1.3 Matrix Inversion

The determinant of a 2 × 2 matrix A is called a second-order determinant,
and it is obtained by taking the product of the two elements on the principal
diagonal and subtracting it from the product of the two elements off the
principal diagonal. Thus, for the 2 × 2 matrix
 
a11 a12
A= ,
a21 a22

the determinant is

a11 a12
|A| = = a11 a22 − a12 a21 . (1.3.1)
a21 a22

The determinant |A| is sometimes denoted by det(A). It is a number or a

scalar and is obtained only for square matrices. If |A| = 0, then the deter-
minant is said to vanish and the matrix A is said to be singular. A singular
matrix is one in which there exists a linear dependence between at least two
rows or columns. If |A| 6= 0, then the matrix A is nonsingular and all its rows
and columns are linearly independent.
Generally, in the case of systems of linear equations, if there is linear de-
pendence, the system may have no solution or an infinite number of possible
solutions, and a unique solution cannot be determined. Thus, the determi-
nant is an important test that points to specific problems before the system
should be solved. Given a system of equations with the coefficient matrix A,
(i) if |A| = 0, the matrix is singular, indicating that there is a linear
dependence among the equations, and so no unique solution exists.
(ii) if |A| 6= 0, the matrix is nonsingular and there is linear independence
among the equations, and so a unique solution exists and can be determined.
The rank ρ of a matrix is defined as the maximum number of linearly
independent rows or columns in the matrix. The rank of a matrix is used
for a simple test of linear dependence, as follows: Assume that A is a square
matrix of order n. Then
(i) if ρ(A) = n, the A is nonsingular and there is linear independence;
(ii) if ρ(A) < n, then A is singular and there is linear dependence.
   
7 4 6 9
Example 1.4. Consider A = and B = . Then |A| =
8 9 8 12
7(9) − 4(8) = 63 − 32 = 31 6= 0, and so the matrix A is nonsingular, i.e., there
is linear independence between any rows or columns, and ρ(A) = 2. On the
other hand, |B| = 6(12) − 9(8) = 72 − 72 = 0, and so the matrix B is singular
and linear dependence exists between its rows and columns (casual inspection
reveals that row 2 is 4/3 times row 1, and column 2 is 3/2 times column 1),
and thus ρ(B) = 1. 
6 1 MATRIX ALGEBRA

The determinant of a 3 × 3 matrix A

 
a11 a12 a13
A =  a21 a22 a23  (1.3.2)
a31 a32 a32

is called a third-order determinant, and it is a sum of three products, which

are derived as follows:
1. Take the first element a11 of the first row and (mentally) delete the row
and the column in which it appears. Then multiply a11 by the determinant
of the remaining second-order matrix.
2. Take the second element a12 of the first row and (mentally) delete the
row and the column in which it appears. Then multiply a12 by (−1) and the
determinant of the remaining second-order matrix.
3. Take the first element a13 of the first row and (mentally) delete the row
and the column in which it appears. Then multiply a13 by the determinant
of the remaining second-order matrix.
This process yields

a22 a23 a a23 a a22

|A| = a11 + a12 (−1) 21 + a13 21
a32 a33 a31 a33 a31 a32
= a11 (a22 a33 − a23 a32 ) − a12 (a21 a32 − a23 a31 ) + a13 (a21 a32 − a22 a31 ),
(1.3.3)

which is a scalar quantity.

The determinant of a 4 × 4 matrix can be determined similarly as the sum
of four products, and so on.
The first line of Eq (1.3.3) can also be written as

|A| = a11 |M11 | + a12 (−1)|M12 | + a13 |M13 |, (1.3.4)

where
a22 a23 a21 a23 a21 a22
|M11 | = , |M12 | = , |M13 | = ,
a32 a33 a31 a33 a31 a32

where |M11 | is the minor of a11 , |M12 | the minor of a12 , and |M13 | the minor
of a13 . A cofactor |Cij | is a minor with a prescribed sign, which follows the
rule
|Cij | = (−1)i+j |Mij |. (1.3.5)
Thus, depending on an even or odd power of (−1) we have

if i + j is an even number, then |Cij | = |Mij |,
if i + j is an odd number, then |Cij | = −|Mij |.
Another random document with
no related content on Scribd:
with her handkerchief before starting. She had not adjusted the
music-stool. She had not pushed back the music-rest as far as it
would go. She had not played the chord and arpeggio inversions of
A flat major and paused dramatically before beginning the
composition of Liszt. All these things she had forgotten. People
would think she was an inexperienced player. Anyhow, she made up
as well as she could for her initial deficiencies during the progress of
the piece. She “swanked,” according to the popular expression. She
was very conscious of the effect her hair was or ought to be
producing....
As a matter of fact, nobody was either looking at her or listening
to her with any particular interest or eagerness.
She was awakened from her egoistic dreams by the half-hearted
applause of those people who by divine instinct know when a piece
is coming to an end several bars ahead, and start their applause at
the last bar but one.... She bowed graciously in front of the piano,
and tripped lightly behind the scenes. The applause did not justify an
encore.... She had made up her mind as she played the concluding
chords of the Concert Study: If I am given an encore, I will do all
those things I omitted to do before: I will polish my hands, adjust the
stool, push back the music-rest, have the sound-board lifted, run up
with arpeggios on the tonic....
But she was not given an encore.
In the artists’ room behind the scenes nobody took much notice
of her. Fred Hitchcock, a local tenor with baritone leanings, was
giving final frenzied directions to his accompanist, a large-featured
female with an excessively low and powdered neck.
“Go slow over that twiddly bit,” he whispered, catching hold of her
to lead her on to the platform. “And don’t forget to give me the
leading note in the adagio.” His hoarse voice merged into the buzz of
sound that came down the corridor leading to the platform.
She overheard a conversation.
“What was that thing that girl played?”
“What girl?”
 “The girl with the red hair.”
“Oh, I don’t know—some Liszt thing, I think.”
“Classical?”
“S’pose so ... of course, nobody listens to pianoforte solos
nowadays....”
“They’re too common, that’s what it is. Everybody strums on the
piano, more or less.”
“I suppose you went to hear Razounov?”
“No, I couldn’t get a seat. The Hippodrome was full of people who
went to see him do something eccentric.”
“Did he?”
“No, as it happened. A friend told me he just came on the
platform, played like an angel for two hours, and went off again. Of
course everybody was greatly disappointed.”
“Naturally....”
“Bockley isn’t a musical suburb. It doesn’t even think it is.
Whereas Upton Rising thinks it is and isn’t.... I wish that pianoforte
player of ours wouldn’t show so much of her red hair and try to look
like a female Beethoven....”
“Oh, shut up—she’s probably somewhere about, she’ll hear
you....”
Catherine put on her hat and cloak and went out by the side door.
She was not angry, but she was suffering from one of those
periodical fits of disillusionment which were the aftermath of her
dreaming. She walked out into the Ridgeway, where the gas lamps
glowed amongst the sprouting trees. Far away she could hear the
clang of trams along the High Road. She passed the corner house
where, it seemed now an age ago, she had discovered her soul in
the murmur of a grand piano. Swiftly she walked along the tarred
asphalt, thinking to reach Gifford Road and have supper. She felt
disappointed. The evening had been lacking in that species of
adventure it had seemed to promise. She had not seen George
Trant. That, she told herself, had nothing to do with it.
Down the Ridgeway a newsboy came running bearing a placard-
sheet in front of him.
“Suicide of a Bockley Schoolmaster,” it said. An awful excitement
seized her. Eagerly she bought a paper and searched the front page.
It took some moments to discover the announcement. It was only
a small paragraph on an inside page: the placard had evidently been
printed to stimulate local circulation.
“Mr. Weston,” she read, “of 24, Kitchener Road, Bockley, an
elementary school teacher at the Downsland Road Council School ...
throat cut....”
She leaned up against the iron railing round a tree. Then,
discovering that she was attracting the attention of passers-by, she
walked on more swiftly than before. In her excitement she took the
opposite direction, towards the Bockley High Street....

§3
Half-way down the Ridgeway she met George Trant. They were
both walking excessively fast and in opposite directions: they almost
cannoned into each other.
“Just looking for you,” he said, stopping her. He wore evening
dress beneath an overcoat. It was peculiar that her eyes should glue
themselves upon an ivory solitaire that he wore. She was half dazed.
“Looking for me?” she echoed, vaguely.
“Yes. Thought you’d gone back to your digs. I was coming to
fetch you. What I want to say is——” (That was one of his
mannerisms of speech. In his letters he had constantly written,
“What I want to say is——”) “we’re having a little supper at the
Forest Hotel after the concert’s over. Just ourselves—the
performers, I mean. Of course you’ll join us.... I didn’t think you’d be
running off so early, or I should have mentioned it before....”
 She was still staring monotonously at that ivory solitaire of his.
“Well—er—you see ... er....”
“Of course if you’re engaged for somewhere else——”
“No, I’m not engaged for anywhere else.” She paused, as if
weighing things in the balance. Then a change came over her. It was
as if she were suddenly electrified. Her eyes lifted and were found
shining with peculiar brilliance. Her body, too, which had been tiredly
swaying, jerked all at once into challenging rigidity. “All right,” she
said, and even in her voice there was a new note, “I’ll come.”
“Good.” He looked a little queerly at this transformation of her.
“Then we’ll go now.”
“But it’s not half-past nine yet. The concert won’t be over till after
ten.”
“That doesn’t matter. I’ve got to go to the hotel to fix up
arrangements. You’d better come with me.”
“Right.” The promptitude of her reply had something in it of
riotous abandon.
“We’ll go by bus to High Wood and walk the rest. It’s sooner....”
Again she acquiesced, this time by a nod that seemed to indicate
an eagerness too great to be put into words.
At the corner of the Bockley High Street they took a bus. They
occupied the front seat on the top. The night was moonless, but
stars were shining over the whole sky. In front and behind stretched
the high road with arc lights gleaming like a chain of pearls. She
thought of that other evening when she had ridden with Helen along
this very road on the top of a crowded tram-car. She remembered
how in the passing glare of the arc-lamps she had read the note
which George Trant had enclosed for her. She remembered it all as
clearly as if it had happened yesterday, though in point of time it
seemed to belong to another age. She remembered the purr of the
quickly-moving car, the hiss of the trolley-wheel along the overhead
wires, the buzz of talk all round her, and the sharp, sickly sensation
of reading a few words in spasms and fitting them into their context
when the pale light merged into the darkness.
But even while she thought of these things she became greatly
joyous. She took off her hat and stuffed it into her pocket (it was of
the kind that yields to such treatment). Her hair blew in soft spray
about her head and shoulders, and her eyes were wet with the tears
that the cool wind brought. She remembered that once he had said
“My God! ... your hair! ...” He might not say it again, but perhaps he
would think it.
“I liked your playing,” he said.
“You did?”
“Rather.... I’m not much of a judge, but I can always tell a real
musician from a false one. The real musician throws his whole soul
into his music....”
“Did I?”
“Yes. I know you did. You played almost unconsciously. I believe
you forgot all about your audience. You were just playing for the
sheer love of playing....”
“Are you sure?”
“Quite sure.”
“Well, you’re wrong, anyway,” She laughed defiantly. “I didn’t
forget about my audience a bit: I kept on remembering them the
whole time. I kept on thinking: !Did they notice that little bit? ... I
polished off that arpeggio rather nicely; I wonder if anybody noticed
it....’ And as for throwing my whole soul into my music, I’m not so
sure—whether—even—whether——”
“Yes?”
She tossed back her head so that her hair danced like flame. The
bus jerked to a standstill.
“Whether I’ve got a soul,” she said very quickly. “Come on, we’re
at High Wood.”
 They clambered down the steps.
“I’m sure you have,” he said, as he helped her off the conductor’s
platform.
“Oh, you don’t know anything about me,” she snapped, as they
entered the footpath through the Forest.
“I believe I know a very great deal about you,” he said quietly.
“Of course you believe so. Well, I don’t mind you telling me.”
He stroked his chin reflectively.
“Well, to begin with,” he said, “you’re passionate.”
She burst into sudden, uncontrollable, crackling laughter. In the
empty spaces of the Forest it sounded like musketry.
“I knew you’d say that.... I knew you would. And for the life of me
I don’t know whether you’re right or wrong. Every woman likes to
think she’s passionate. And nobody knows whether she’s any more
passionate than anybody else.... Pass on to the next point. You may
be right or you may be wrong about the last.”
“You’re impulsive—but good-natured.”
“Oh, rather. A kind heart beneath a rough exterior, eh?”
“I’m quite serious.”
“Are you? I’m not.”
She frisked along in front of him, revelling in the rustle of last
autumn’s leaves.
“Do you know what I should do if I were serious?” she asked
suddenly, when he had caught up to her.
“No.”
She walked a little way in silence, kicking up the dried leaves with
her toes.
“What would you do?” he said.
 Her voice became fierce. “I should——” she began, and stopped.
She walked a few steps as if pondering, then she laughed airily and
tossed her head. “I really don’t know what I should do. Only I’m
certain of one thing: I shouldn’t be with you here.”
She could almost feel the extent to which her conversation was
mystifying him.
Then she became quiet and submissive, nestling like a stray
kitten at his side. She took his arm.
“I’m going to lean on you,” she said; “I nearly fell over a tree root
just there.”
He looked gratified. For three or four minutes they walked on in
silence. He had plenty he wished to say, but as a matter of fact he
thought this particular silence, coming when it did, rather impressive,
and he was unwilling to curtail it by a remark unworthy of its
profundity. He was engaged in thinking of that remark, a remark that
should not so much break the silence as guide it into still more
profound depths. He had almost decided on what he should say
when quickly and without any warning she snatched her arm from
his and scampered a few paces ahead.
“Oh, George,” she cried, with an extraordinary mingling of
passion and irritation, “do say something! For God’s sake keep up
the conversation! We’ve been a quarter of an hour without a word.
Say something, anything you like—only I can’t stand this mooning
about under the trees saying nothing!”
“You’re in a very extraordinary mood to-night,” he said
deliberately. He was genuinely disappointed.
“I am, or I shouldn’t have come with you,” she replied bluntly.
“Do you dislike me, then?” he asked, with a kind of injured dignity.
“Oh no—oh, don’t let’s talk seriously. I tell you I don’t feel serious
to-night.”
“Well, you won’t need to be. We’re going to have a very jolly
evening.”
 “I hope so. That’s why I came. I feel like having a jolly evening.”
The Forest Hotel occupied a fine position on the crest of a thickly-
wooded hill overlooking one of the prettiest spots in Epping Forest. A
large balcony opened on to the dining-room, which was on the first
floor, and Chinese lanterns swung loosely from the ornamental
pilasters. As Catherine caught sight of the table, a vista of white and
silver and gleaming glass, she clapped her hands ecstatically. She
was as a little child in her enthusiasm.
“Oh, fine—fine!” she cried, clutching George once more by the
arm.
The table was on the balcony, and inside the dining-room the
floor had been cleared, presumably for dancing. A sleek grand piano
sprawled across one corner. Catherine rushed up to it and
immediately plunged into some rapid, noisy piece. It was a splendid
instrument, and the dim light (only the swaying lanterns on the
balcony were lit) threw her into rapture. George came to her side,
watching in admiration. Watching rather than listening, because, as
he had himself admitted, he was no judge of music. And also
because the red glow from the swinging lanterns kindled her hair like
a puff of wind on smouldering charcoal.
“There!” she cried, triumphant, as she executed something
difficult with her left hand. She swung into a dirge-like melody, tired
of it seemingly, and broke into energetic ragtime. George felt it was
in some way inappropriate to play ragtime at such a moment.
“Let’s come out on to the balcony,” he suggested, “we’ve only got
a quarter of an hour or so before the others come.”
“Well, we’ve nothing particular to do, have we?”
“It’s cooler.... Come on....”
They walked through the French windows and sat on the parapet
overlooking the gravel courtyard and the blurred panorama of the
Forest.
“It ought to be moonlight,” he exclaimed rapturously.
 “No, it oughtn’t,” she contradicted. “I’m glad it isn’t. Starlight is
much better.”
It was not an encouraging beginning for him.
“Do you mind if I talk to you seriously?” he asked.
She laughed a little unsympathetically.
“Not at all, only I don’t suppose I shall talk to you seriously.”
“Then it’s not much good, is it?” he remarked, crest-fallen.
“No. Much better to talk nonsense. Let’s talk nonsense. Does one
eat oysters with a spoon or a fork?”
“I can’t——”
“But I want to know. I noticed we begin with oysters, and I’m not
sure what tools to use. Surely you don’t want me to make a fool of
myself. Come, tell me, how does one masticate oysters?”
“A fork is customary, I believe.”
“Thank you. That is what I wanted to know.”
There was a pause, during which the distant sound of voices
reached them from the dim Forest background.
“They’re just coming,” she said. “They must have come by bus,
like we did.”
He ground his heel into the carpet-matting.
“What I want to say——” he started suddenly. “It’s like this. I
believe there was a—a sort of—er—misunderstanding between us at
one time. Now I’m not prepared to say that I was altogether right. In
fact——”
“I don’t remember any misunderstanding. I think I at any rate
understood you perfectly. I really don’t know what you’re talking
about.”
“Well, to put it bluntly, what happened was——”
“Excuse me. I must let them hear the piano as they come out of
the Forest. Sorry to cut short our argument, but I don’t feel
metaphysical.... What shall I play? Something appropriate....
Suggest something!”
He sat rather gracelessly on the parapet watching her as she
skipped over to the piano. The expression on his face was one of
bafflement.
“I really don’t——” he called ineffectually.
For answer she began the pianoforte accompaniment of Landon
Ronald’s “Down in the Forest.”
A moment later over the fringe of Forest still untraversed came
the voice of the soprano singer, clear and tremulous, but not
particularly musical. “Down in the Forest something stirred,” she
sang, and Catherine laughed as she caught the sound....

§4
About twenty minutes to midnight the tenor singer (with baritone
leanings) whispered to George Trant: “I say, ol’ chap. You’d better
l’kafter tha’ l’l gaerl of yours.”
“What d’you mean?”
“Wha’ I say. She’s had too much.”
“But really, I don’t think——”
“Two glasses sherry, one hock, three champagne, two port ... I’ve
took notice.”
“She’s a bit noisy, I’ll admit.... But she was quite lively enough as
we came along. It’s her mood, I think, mostly.”
The party had left the table and split up into groups of twos and
threes. Some lingered sentimentally on the balcony; the violinist,
who was just a shade fuddled, lay sprawled across a couch with his
eyes closed. Catherine was at the piano, making the most
extraordinary din imaginable. Surrounding her were a group of young
men in evening dress, singers and comedians and monologuists and
what not. George Trant and the tenor singer stood at the French
windows, smoking cigars and listening to the sounds that proceeded
from the piano.
“We shall have the manager up,” said George, nervously.
“He’ll say we’re damaging the instrument.... I wish she’d quieten
down a bit. The whole place must be being kept awake....”
Catherine’s voice, shrill and challenging, pierced the din.
“Impressions of Bockley High Street—nine p.m. Saturday night,”
she yelled, and pandemonium raged over the keyboard. It was really
quite a creditable piece of musical post-impressionism. But the noise
was terrific. Glissandos in the treble, octave chromatics in the bass,
terrible futurist chords and bewildering rhythms, all combined to
make the performance somewhat painful. Her select audience
applauded enthusiastically.
George Trant moved rather nervously towards the piano.
“I shouldn’t make quite so much noise,” he began, but nobody
heard him. Catherine was crying out “Marbl’arch, Benk, L’pol Street,”
in the approved jargon of the omnibus conductor, and was
simultaneously making motor-bus noises on the piano. Everybody
was laughing, because the mimicry of her voice was really excellent.
George felt himself unable to raise his voice above the din. He
paused a moment immediately behind her back and then touched
her lightly on the shoulder. She did not heed. He touched her again
somewhat more violently than before. She stopped abruptly both her
instrumental and vocal effects, and swung round suddenly on the
revolving music-stool so as to face him. Her eyes were
preternaturally bright.
“Excuse me,” he began, and something in her eyes as she
looked up at him made him doubly nervous, “but perhaps it would be
better if you didn’t make quite such a noise.... You see, the other
people ...” he added vaguely.
There was absolute silence now. The last echo of the piano had
died away, and the select audience waited rather breathlessly for
what might happen.
 Catherine rose. There was that greenish-brown glint in her eyes
that made fierce harmony with her hair. For a moment she looked at
him unflinchingly. There was certainly defiance, perhaps contempt in
her eyes.
“Who are you?” she said, with quiet insolence.
Somebody tittered.
George Trant looked and felt uncomfortable. For answer he
turned slowly on his heel and walked away. It seemed on the whole
the most dignified thing to do. Catherine flushed fiercely. Like a
tigress she bounded to his side and made him stop.
“For God’s sake, don’t sulk!” she cried wildly. “Wake up and say
something! Don’t stand there like a stone sphinx! Wake up!”
With a quick leap she sprang upwards and ran her two hands
backwards and forwards through his hair. His hair was long and lank
and well plastered. After she had finished with it it stood bolt upright
on his head like a donkey fringe. Everyone roared with laughter.
During the progress of this operation the interior door had been
opened and a man had entered. In the noise and excitement of the
mêlée he was not noticed. He was tall, severe-looking and in
evening dress. When the excitement subsided they found him
standing a little awkwardly on the edge of the scuffle.
Catherine thought he was at least an underwaiter, come to
complain of the noise they were making.
He bowed very slightly, and immediately everybody felt sure he
was a waiter. Only a professional could have bowed so chillingly.
Catherine, with flushed face and dishevelled hair, leaned against
a chair, panting from her exertion.
“I do not wish to interrupt,” began the stranger, and there might
have been sarcasm in his voice, “but I have been commissioned to
deliver a message to Miss Weston. Which is Miss Weston?”
“I am Miss Weston,” gasped Catherine. Then, to everyone’s
amazement, she proceeded furiously: “I know it—I know it. You
needn’t tell me! I saw it in the papers ... I suppose they’ll say it’s all
my fault.... Do they want me? ... if so, I’ll come. I’ll come with you
now if you like....”
The stranger raised his eyebrows slightly.
“I have no desire for you to come anywhere with me.... I don’t
know what you are talking about, either. My message is contained in
this note, and there is no immediate necessity to reply to it.”
Somebody said, rather in the spirit of a heckler at a political
meeting: “Who sent it?” The stranger turned and said: “I should think
Miss Weston and not I should be asked that.” The questioner
subsided ignominiously.
Catherine took the envelope that the stranger offered her. She
put it unread into her pocket. The stranger bowed and walked out.
Silence.... Then a chatter of conversation.
“Admirer of yours,” said the violinist, thickly, from his couch.
Everybody thought he had been asleep.
“Didn’t exactly get you at a good moment,” remarked the tenor
singer, flicking away his cigar-ash.
“Looked like an undertaker,” said the soprano.
“Or the ‘salary-doubled-in-a-fortnight’ man in the efficiency
advertisements,” put in the monologuist.
Catherine started to arrange her hair.
“I’m going,” she said, and walked towards the balcony (there was
no exit that way). Near the French windows she staggered and fell,
fortunately upon the cushions of a couch. They all crowded round
her. She did not attempt to rise.
“She’s drunk,” muttered the violinist.
“Possibly ...” said George Trant, bending down to her. “Fetch
some water. I think she’s fainted....”
 CHAPTER VII
TRAGEDY
§1
ON the first of May the weather was very sultry. Downsland Road,
running past the front of the Council School, was both blazingly hot
and distressingly conscious that it was Friday afternoon. The road
was bursting out in little gouts of soft tar: costermongers were
arranging their wares for the evening’s marketing, spitting
contemplatively on the apples and polishing them afterwards on their
coat-sleeves. Children with clanking iron hoops converged from all
directions upon the four entrance gates of the Downsland Road
Council School, respectively those of the boys’, girls’, infants’ and
junior mixed departments. There they either carried or dragged them
surreptitiously along, for the trundling of hoops was forbidden in the
schoolyard.
At five minutes to two, threading his way past the groups of boys
and girls that littered the pavements and roadways, came the short,
stumpy form of Mr. Weston. He was shabbily dressed as usual, yet it
might have been said that he carried his umbrella somewhat more
jauntily than was his wont. In fact, people had lately been saying that
he was beginning to get over the loss of his wife.... At any rate he
passed the costermongers and their stalls in a slouch that was not
quite so much a slouch as usual, smiled pleasantly as he caught
sight of the announcement of a Conservative Club soirée, and had
just reached the edifice known as the Duke Street Methodist Chapel
when his attention was arrested by an awful spectacle.
The Duke Street Methodist Chapel, it may here be remarked,
was a structure of appalling ugliness situate in the very midst of
some of the worst slums in Bockley. Its architecture was that of a
continental railway station, and its offertories between a pound and
thirty shillings a Sunday. Inside the hideous building, with her back to
the blue-distempered wall of the choir, the late Mrs. Weston had for
many years yelled the hymns at the top of her voice.... And along the
brown matting of the left-hand aisle Mr. Weston, suave and supple,
collection-plate in hand, had in his time paced many miles.... Once,
when the church steward was ill, his voice had been heard aloft in
the reading of the notices. And at the left-hand door, while the
organist played the “War March of the Priests,” he had stood with
outstretched hand, saying:
“Good evening, Mrs. Lawson.... Good evening, Ethel.... ’Night,
Miss Picksley ... see you at the Band of Hope on Tuesday, I
suppose? ...”
He did not do that sort of thing now. In the chapel he was little
seen, and the Temperance Society knew him not. Only the Guild and
Mutual Improvement Society still counted him as a member, and that
was solely because they had not worried him into resigning.... At the
Guild and Mutual Improvement Society Mr. Weston’s carefully read
papers, once a session, on “Milton,” “John Wycliff, Scholar and
Saint,” “The Lake Poets,” etc., had been a well-known, but
unfortunately not always well-attended feature.
For over a year the fixture-card had lacked the name of Mr.
Weston.
And then, a fortnight ago—to be precise, on April 14th—Mr.
Weston had been stopped in the street by Miss Picksley, the
secretary of the Guild and Mutual Improvement Society. She had
said:
“Oh, Mr. Weston, do give us one of your literary evenings, will
you?”
Perhaps it was the subtle compliment contained in the phrase
“literary evenings” that caused Mr. Weston not to say “I am sorry, but,
etc., etc....” as quickly as he had intended.
Miss Picksley exploited the delay brilliantly.
 “Good!” she cried, whipping out a pencil and notebook, “I’ll get
your name down for May 1st.... What’ll be your subject?”
“But, er ... I don’t ... er——”
“Something about literature, eh? ... Oh, do, please!” purred Miss
Picksley, making eyes at him. (She was really anxious for him to
accept, because she had canvassed in vain seven other speakers.)
“Tell me your subject, then it can go down on the fixture-cards.”
Mr. Weston, to his astonishment, lost his head and struck blindly
at the first literary name that came into his disordered mind.
“Shakespeare,” he gasped.
Miss Picksley departed, calling blessing upon his head.

§2
Now, as Mr. Weston passed the scene of so many of his former
labours, he felt not altogether sorry that to-night, in the schoolroom
adjoining the chapel, he would address a small but certainly select
gathering on the subject of “Shakespeare.” ... He would have liked to
have expanded the title of his paper into “Shakespeare, Man or
Superman?” after the fashion of a certain Methodist preacher who
occasionally visited Bockley. However....
Mr. Weston, it may be remarked, was feeling in quite a tolerably
good humour. He was beaming genially at the world in general when
a horrible sight met his eyes. Then his brow darkened into a frown.
The smile left his face; his lips tightened ominously. He stopped,
swung down his umbrella from its jaunty attitude, and stared. His
eyes flamed. The slope of his nose became full of menace.
For there, before his eyes, chalked up in scrawly writing on the
foundation-stone of the Duke Street Methodist Chapel, was an
inscription that excited his horrified attention. “This stone was laid ...
to the glory of God ... the Rev. Samuel Smalljohn ...” he read, and
“Let your light so shine....” And underneath that, in a space that
made it most conspicuous, the brutal legend: “Daddy Weston is a
Soppy Fool....”
Entering the Downsland Road Council School in a white heat of
indignation, Mr. Weston was just able to hear the sound of
suppressed laughter and scurrying feet as he entered the classroom.
The conviction forced itself upon him that somebody had been
watching at the keyhole....

§3
Mr. Weston was not normally a hot-tempered man. He was by
nature placid, servile, lymphatic. It was solely as a measure of self-
protection that he had trained himself to lose his temper on
appropriate occasions. It was part of his disciplinary outfit.
He stood glowering fiercely behind his desk.
“I want all boys who were concerned in the chalking up of those
offensive remarks outside the school to stand up.”
Pause. No result.
“I may say that I have already a very fair idea of who they are,
and I shall be most severe with those who do not acknowledge
themselves.”
(A lie, but Mr. Weston’s disciplinary system condoned it.)
“I may also say that for every half-minute I am kept waiting I shall
keep the class in half an hour after school hours. I have already
decided to keep the class in till five for keeping me waiting so long.”
Here Mr. Weston pulled out his watch and placed it prominently
on the desk before him. (This was mere theatricalism, as the watch
did not go.)
Pause. Then a warning shuffle and seven small boys raised
themselves.
Mr. Weston dived into his desk and produced seven coloured
dusters for cleaning blackboards.
 “Come here,” he said to the seven.
The seven came.
“You will each take one of these dusters and go out into the street
and obliterate every one of the marks you have made. Then you will
return.”
It was Mr. Weston’s own invention, this disciplinary method.

§4
Slowly, ever so slowly, the afternoon crept by, and Mr. Weston
was just beginning to congratulate himself upon having proved equal
to the occasion, when an awkward but all-important fact occurred to
him. If you keep your class in you have to stay in with it. Mr. Weston,
of all people, ought to have learnt this lesson, yet somehow amidst
the heat and sultriness of the afternoon it had escaped him. For he
was tired, dead tired. And also hot. The sweat was rolling down his
forehead. Oh, how he wished he had said half-past four, and not five!
Confound it, why had he said five? Half-past four would really have
done just as well. Only, having said five, he was bound (by that
disciplinary code of his) to keep his word.
He took a sheaf of notes from his inside pockets and perused
them diffidently. “William Shakespeare.” It was to last about half an
hour, and as yet Mr. Weston had thought about William Shakespeare
only sufficiently for it to last twenty or twenty-five minutes. It would
have to be padded out. Something about the “immortal bard of
Avon....” On such a fine evening, thought Mr. Weston, the audience
would be small. Possibly about fifteen or twenty. There would be
Miss Picksley, the secretary, to receive subscriptions for the coming
session. Mrs. Hollockshaw would be there to play the hymn on the
American harmonium, and Mr. Sly would open with a word of prayer.
The Gunter girls would sit on the back row and flirt with the Merridge
boys. Possibly old Mrs. Cowburn would turn up. (Or was she dead
by this time?) ... After he had read his paper there would be a few
minutes for discussion. That would merely mean votes of thanks,
because he would take care not to say anything controversial.
Nothing about the Shakespeare-Bacon business. Then the
benediction given out by Mr. Sly. With luck the whole business would
be over by nine, and there would be time for a stroll through the
Forest at dusk. Or perhaps, though, it would be quite dark. Heavens
I Only twenty-five to five. Old Clotters was locking up in his room....
A ray of tawdry sunlight penetrated the dust and murkiness of the
atmosphere, bringing into prominence the rather obvious fact that
Mr. Weston was combining reverie with the observation of his class
through the interstices of his fingers. (This was an integral part of Mr.
Weston’s disciplinary system.) Ever and anon his eyes would focus
themselves upon a particular boy in the hope that if he were watched
long enough he would do something amiss. This happy
consummation was not long in coming. There was that boy Jones!
Jones was doing something. Surely, surely! ... Well, well, perhaps he
would do it again. There, he had done it.... His jaws had moved
perpendicularly twice within ten seconds. There could be no further
doubt about it. Jones was eating!
“Jones.”
“Yes, sir.”
“Are you eating?”
“Yes, sir.”
“Then I will report you to Mr. Clotters on Monday morning. I will
not have this sort of thing going on in my class. Your manners are
those of the lower animals. Come up here and put what you are
eating into the waste-paper basket!”

§5
Punctually at five Mr. Weston locked his desk and prepared to
observe the solemn ritual of dismissal. It was in three movements.
“Attention!” called Mr. Weston, and the class looked at him
eagerly.
 “Stand!” called Mr. Weston, and the class stood. But there must
have been some flaw in the standing, for Mr. Weston immediately
said, “Sit down again!” They sat down again.
“Now stand!” cried Mr. Weston, after a suitable pause, and this
time the manœuvre met with his approval.
“Three!” continued Mr. Weston, and at this mysterious direction
the class took a side-step into its respective gangways.
“First row—forward!”
“Second.”
“Third.”
When the last row had dissolved into the disintegrating chaos of
the corridor, Mr. Weston took up his hat and umbrella and walked
through the masters’ gateway into the frowsiness of Downsland
Road....

§6
In the hot kitchen of No. 24, Kitchener Road, Mr. Weston made
himself some tea and cut some bread and butter. He had not much
time to spare. He must add a few pages to his paper. Then he must
wash and shave and make himself respectable. During his meal he
thought once or twice of those old days when Laura, his wife, had
been there to get his tea ready for him, to fuss round the books and
papers he brought home, and to say: “Going out to-night, are you?
Because if not, there’s your slippers. And let’s ’ave your dirty
boots....” He thought, too, of Catherine: a little child, asking him
absurd questions, messing about with his exercise books, begging
him for half-used sheets to scribble on. But there was nothing
regretful in his thoughts of those past days. On the contrary, he
rather inclined to moralize: “I don’t know whether I’m not actually
better off than I was then. At any rate I’m free, and I can do what I
like. It’s not so bad, really.”
He wrote down a few sentences about Shakespeare.