ebook download Spatiotemporal Random Fields, Second Edition: Theory and Applications George Christakos - eBook PDF all chapter

Spatiotemporal Random Fields, Second
Edition: Theory and Applications

George Christakos - eBook PDF
Go to download the full and correct content document:
https://ebooksecure.com/download/spatiotemporal-random-fields-second-edition-theo
ry-and-applications-ebook-pdf/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
(eBook PDF) Turbomachinery: Basic Theory and

Applications, Second Edition
http://ebooksecure.com/product/ebook-pdf-turbomachinery-basic-
theory-and-applications-second-edition/
Kappa Distributions. Theory and Applications in Plasmas

1st Edition Edition George Livadiotis (Eds.) - eBook
PDF
https://ebooksecure.com/download/kappa-distributions-theory-and-
applications-in-plasmas-ebook-pdf/
Solutions for Biot's Poroelastic Theory in Key

Engineering Fields. Theory and Applications 1st Edition
Edition Yuanqiang Cai And Honglei Sun (Auth.) - eBook
PDF
https://ebooksecure.com/download/solutions-for-biots-poroelastic-
theory-in-key-engineering-fields-theory-and-applications-ebook-
pdf/
Geophysical potential fields: geological and

environmental applications 1st edition - eBook PDF
https://ebooksecure.com/download/geophysical-potential-fields-
geological-and-environmental-applications-ebook-pdf/
Rowan's Primer of EEG, 2e Second Edition Fields - eBook
PDF
https://ebooksecure.com/download/rowans-primer-of-eeg-2e-ebook-
pdf/
Schaum's Outline of Probability, Random Variables, and

Random Processes, Fourth Edition Hwei P. Hsu - eBook
PDF
https://ebooksecure.com/download/schaums-outline-of-probability-
random-variables-and-random-processes-fourth-edition-ebook-pdf/
(eBook PDF) Psycholinguistics: Introduction and

Applications, Second Edition
http://ebooksecure.com/product/ebook-pdf-psycholinguistics-
introduction-and-applications-second-edition-2/
(eBook PDF) Psycholinguistics Introduction and

Applications Second Edition
http://ebooksecure.com/product/ebook-pdf-psycholinguistics-
introduction-and-applications-second-edition/
(eBook PDF) Probability, Statistics, and Random Signals

by Charles Boncelet
http://ebooksecure.com/product/ebook-pdf-probability-statistics-
and-random-signals-by-charles-boncelet/
Spatiotemporal Random
Fields
This page intentionally left blank
Spatiotemporal Random
Fields
Theory and Applications
Second Edition
George Christakos
Department of Geography, San Diego State University,
San Diego, California, USA;
Institute of Islands and Coastal Ecosystems, Ocean College,
Zhejiang University, Zhoushan, Zhejiang, China
Elsevier
Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
Copyright © 2017 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including photocopying, recording, or any information storage and retrieval system, without
permission in writing from the publisher. Details on how to seek permission, further information about the
Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance
Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the Publisher (other
than as may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience broaden our
understanding, changes in research methods, professional practices, or medical treatment may become
necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using
any information, methods, compounds, or experiments described herein. In using such information or methods
they should be mindful of their own safety and the safety of others, including parties for whom they have a
professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any
liability for any injury and/or damage to persons or property as a matter of products liability, negligence or
otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the
material herein.
Library of Congress Cataloging-in-Publication Data
A catalog record for this book is available from the Library of Congress
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
ISBN: 978-0-12-803012-7
For Information on all Elsevier publications visit our

website at https://www.elsevier.com/books-and-journals
Publisher: Candice Janco

Acquisition Editor: Marisa LaFleur
Editorial Project Manager: Marisa LaFleur
Production Project Manager: Paul Prasad Chandramohan
Designer: Greg Harris
Typeset by TNQ Books and Journals
Dedicated
to
Yongmei
Contents
Preface ................................................................................................................................................. xv
CHAPTER I Space, Time, SpaceeTime, Randomness, and Probability .................. 1

1. Introduction ...................................................................................................................1
2. SpaceeTime Continuum and Kolmogorov Probability Space ....................................4
2.1 SpaceeTime Arguments: Points, Lags, Separations, and Metrics....................... 4
2.2 Transformations and Invariance in SpaceeTime ............................................... 17
2.3 SpaceeTime Interpretations................................................................................ 23
2.4 Functions of SpaceeTime Arguments................................................................ 26
3. Random Variables in SpaceeTime.............................................................................31
3.1 Kolmogorov’s Probability Theory ...................................................................... 31
3.2 Useful Inequalities............................................................................................... 35
3.3 Convergence of Random Variable Sequences .................................................... 37
CHAPTER II Spatiotemporal Random Fields........................................................... 39
1. Introduction .................................................................................................................40
1.1 The SpaceeTime Component ............................................................................. 41
1.2 The Randomness Component ............................................................................. 42
2. Characterization of Scalar Spatiotemporal Random Fields .......................................42
2.1 Probabilistic Structure ......................................................................................... 43
2.2 The Characteristic Function ................................................................................ 50
2.3 Spatiotemporal Variability Functions: Complete (or Full) and Partial .............. 51
2.4 Analysis in the Spectral Domain ........................................................................ 56
2.5 Data-Independent Spatiotemporal Variability Function ..................................... 57
2.6 Some Noticeable Special Cases of the Spatiotemporal Random
Field Theory ........................................................................................................ 59
2.7 SpaceeTime Separability.................................................................................... 60
3. Physical Insight Behind the Random Field Concept .................................................61
3.1 Random Field Realizations ................................................................................. 61
3.2 Probable Versus Actual ....................................................................................... 63
3.3 Probability and the Observation Effect............................................................... 64
3.4 Self-consistency and Physical Fidelity ............................................................... 65
4. Geometry of Spatiotemporal Random Fields.............................................................69
5. Vector Spatiotemporal Random Fields .......................................................................70
6. Complex Spatiotemporal Random Fields...................................................................73
7. Classifications of the Spatiotemporal Random Field Model .....................................73
7.1 First Classification: Discrete Versus Continuous Arguments............................. 74
7.2 Second Classification: Scalar Versus Vector Random Fields and Arguments... 74
vii
viii CONTENTS
7.3 Third Classification: Probability Law Shapes .................................................... 74

7.4 Fourth Classification: SpaceeTime Variability .................................................. 75
7.5 Fifth Classification: Spatiotemporal Random Field Memory Versus
Independence ....................................................................................................... 77
8. Closing Comments......................................................................................................78
8.1 The Methodological Importance of SpaceeTime .............................................. 78
8.2 A Conceptual Meeting Point for Modelers and Experimentalists ..................... 80
8.3 There Is No Assumptionless Modeling .............................................................. 81
CHAPTER III SpaceeTime Metrics ......................................................................... 83
1. Basic Notions ..............................................................................................................83
1.1 Formal and Physical Aspects of SpaceeTime Metric Determination ............... 84
1.2 SpaceeTime Metric Forms ................................................................................. 87
1.3 Derived SpaceeTime Metrics............................................................................. 92
1.4 SpaceeTime Metric Differentials ....................................................................... 93
1.5 Specifying SpaceeTime Relationships in the Covariance Function ................. 95
2. Covariance Differential Formulas.............................................................................100
3. SpaceeTime Metric Determination From Physical Considerations........................106
4. Examples ...................................................................................................................108
5. Concerning the Zeta Coefficients .............................................................................117
6. Closing Comments....................................................................................................118
CHAPTER IV SpaceeTime Correlation Theory ..................................................... 121
1. Focusing on SpaceeTime Variability Functions......................................................121
1.1 Basics of SpaceeTime Correlation Theory...................................................... 122
1.2 Physical Investigations Based on SpaceeTime Correlation Theory................ 123
2. SpaceeTime Variability Functions in Terms of Scalar SpaceeTime Statistics......124
2.1 Locality: One-Point SpaceeTime Variability Functions.................................. 125
2.2 Nonlocality: Omnidirectional Two-Point SpaceeTime Variability
Functions ........................................................................................................... 127
2.3 Nonlocality: Direction-Specific SpaceeTime Variability Functions ............... 133
2.4 Physical Considerations and Assumptions of SpaceeTime Variability
Functions ........................................................................................................... 133
2.5 Formal and Physical Covariance Permissibility ............................................... 137
3. Basic Properties of Covariance Functions................................................................139
4. CrosseSpaceeTime Variability Functions...............................................................141
5. Correlation of Gaussian and Related Spatiotemporal Random Fields ....................144
5.1 General Considerations ..................................................................................... 144
5.2 Gaussian Properties ........................................................................................... 144
CONTENTS ix
6. Correlation Theory of Complex Spatiotemporal Random Fields ............................146

6.1 Basic Notions .................................................................................................... 146
6.2 Other Types of Complex Covariance Functions............................................... 148
6.3 Gaussian Complex Spatiotemporal Random Fields ......................................... 151
6.4 Complex-Valued Versus Real-Valued Covariance Functions
of SpaceeTime Homostationary Random Fields ............................................. 152
6.5 Some Methodological Considerations .............................................................. 153
CHAPTER V Transformations of Spatiotemporal Random Fields......................... 155
1. Introduction ...............................................................................................................155
2. Fourier Transformation .............................................................................................157
2.1 Characteristic Functions .................................................................................... 157
2.2 Harmonizable Random Fields and Covariance Functions ............................... 158
2.3 Transfer Function and Evolutionary Mean Power............................................ 165
2.4 Fourier Transform of Vector Spatiotemporal Random Fields.......................... 167
3. Space Transformation ...............................................................................................168
3.1 Basic Notions .................................................................................................... 168
3.2 Space Transformation of Spatiotemporal Random Fields................................ 173
3.3 Space Transformation for Spatiotemporal Variability Functions..................... 175
3.4 Space Transformation in the Simulation of Spatiotemporal
Random Fields................................................................................................... 177
3.5 Space Transformation in the Solution of Stochastic Partial Differential
Equation............................................................................................................. 180
4. The Traveling Transformation ..................................................................................180
4.1 Basic Notions .................................................................................................... 181
4.2 Determination of the Traveling Vector ............................................................. 186
4.3 Traveling Transformation in Spatiotemporal Random Field Estimation:
The SpaceeTime Projection Technique ........................................................... 195
5. Closing Comments....................................................................................................201
CHAPTER VI Geometrical Properties of Spatiotemporal Random Fields ............ 203
1. Introduction ...............................................................................................................203
2. Stochastic Convergence ............................................................................................204
3. Stochastic Continuity ................................................................................................208
3.1 Basic Types of Stochastic Continuity ............................................................... 209
3.2 Equivalence, Modification, and Separability .................................................... 212
4. Stochastic Differentiation .........................................................................................216
4.1 Basic Notation and Definitions ......................................................................... 217
4.2 Covariances of Random Field Derivatives ....................................................... 223
4.3 Mean Squarely Differentiability Conditions..................................................... 227
4.4 Almost Surely Differentiability Conditions...................................................... 231
x CONTENTS
5. The Central Limit Theorem......................................................................................233

6. Stochastic Integration................................................................................................234
CHAPTER VII Auxiliary Hypotheses of Spatiotemporal Variation ....................... 239
1. Introduction ...............................................................................................................239
1.1 Hypothesis 1: Homostationarity........................................................................ 241
1.2 Hypothesis 2: Isostationarity............................................................................. 243
1.3 Hypothesis 3: Heterogeneity ............................................................................. 245
1.4 Hypothesis 4: Ergodicity................................................................................... 246
1.5 Hypothesis 5: Separability ................................................................................ 248
1.6 Hypothesis 6: Symmetry ................................................................................... 249
1.7 Hypothesis 7: Locational Divergence ............................................................... 249
2. SpaceeTime Homostationarity.................................................................................250
2.1 Omnidirectional Spatiotemporal Variability Functions .................................... 251
2.2 Direction-Specific Spatiotemporal Covariance Function ................................. 255
2.3 Anisotropic Features.......................................................................................... 255
2.4 Spatiotemporal Variogram and Structure Functions: Omnidirectional
and Direction Specific ....................................................................................... 257
3. Spectral Representations of SpaceeTime Homostationarity ...................................260
3.1 Spectral Functions of SpaceeTime Homostationary Random
Fields ................................................................................................................. 261
3.2 Properties of the Spectral Density Function ..................................................... 266
3.3 Partial Spectral Representations........................................................................ 268
3.4 More on Dispersion Relations........................................................................... 272
3.5 Separability Aspects .......................................................................................... 273
4. The Geometry of SpaceeTime Homostationarity ...................................................276
4.1 Differentiation Formulas: Physical and Spectral Domains .............................. 276
4.2 Stochastic Continuity and Differentiability ...................................................... 285
4.3 Spatiotemporal Random Field Integrability ..................................................... 296
5. Spectral Moments and Linear Random Field Transformations ...............................297
CHAPTER VIII Isostationary Scalar Spatiotemporal Random Fields ................... 303
1. Introduction ...............................................................................................................303
1.1 Basic Considerations ......................................................................................... 303
1.2 Power-Law Correlations.................................................................................... 309
1.3 Physical Considerations of Variogram Functions............................................. 313
2. Relationships Between Covariance Derivatives and SpaceeTime
Isostationarity ............................................................................................................314
3. Higher-Order Spatiotemporal Variogram and Structure Functions .........................319
4. Separable Classes of SpaceeTime Isostationary Covariance Models.....................320
5. A Survey of SpaceeTime Covariance Models ........................................................324
CONTENTS xi
6. Scales of Spatiotemporal Dependence and the Uncertainty Principle ....................329

6.1 Scales for Spatiotemporal Random Fields With Restricted SpaceeTime
Variability .......................................................................................................... 330
6.2 Relationships Between Physical and Spectral Domains: The Uncertainty
Principle............................................................................................................. 334
7. On the Ergodicity Hypotheses of Spatiotemporal Random Fields ..........................336
CHAPTER IX Vector and Multivariate Random Fields.......................................... 347
1. Introduction ...............................................................................................................347
2. Homostationary and Homostationarily Connected CrosseSpatiotemporal
Variability Functions and CrosseSpectral Density Functions.................................349
2.1 Basic Notions and Interpretations ..................................................................... 350
2.2 Geometry of Vector Spatiotemporal Random Fields ....................................... 355
3. Some Special Cases of Covariance Functions .........................................................356
4. Solenoidal and Potential Vector Spatiotemporal Random Fields ............................362
5. Partial Cross-Covariance and Cross-Spectral Functions ..........................................365
6. Higher-Order CrosseSpatiotemporal Variability Functions ....................................366
7. Isostationary Vector Spatiotemporal Random Fields ...............................................369
7.1 Direct (Lag-Based) SpaceeTime Isostationarity.............................................. 369
7.2 Composite Lag-FieldeBased SpaceeTime Isostationarity.............................. 372
7.3 Links With Solenoidal and Potential Spatiotemporal
Random Fields................................................................................................... 378
8. Effective Distances and Periods ...............................................................................381
CHAPTER X Special Classes of Spatiotemporal Random Fields......................... 383
1. Introduction ...............................................................................................................383
2. Frozen Spatiotemporal Random Fields and Taylor’s Hypothesis............................384
2.1 Basic Notions .................................................................................................... 385
2.2 Spectral Domain Analysis................................................................................. 389
2.3 Differential Equation Representations .............................................................. 391
2.4 Extensions of the Frozen Random Field Model............................................... 395
2.5 Integrals of Frozen Spatiotemporal Random Fields......................................... 399
2.6 Vector Frozen Spatiotemporal Random Fields ................................................. 399
3. Plane-Wave Spatiotemporal Random Fields ............................................................400
4. Lognormal Spatiotemporal Random Fields..............................................................402
5. Spherical Spatiotemporal Random Fields ................................................................402
6. Lagrangian Spatiotemporal Random Fields .............................................................407
CHAPTER XI Construction of Spatiotemporal Probability Laws .......................... 409
1. Introduction ...............................................................................................................409
2. Direct Probability Density Model Construction Techniques ...................................411
2.1 The Independency Techniques .......................................................................... 412
xii CONTENTS
2.2 The Spherical Symmetry Technique................................................................. 412

2.3 The Transformation Technique ......................................................................... 413
3. Factora-Based Probability Density Model Construction Techniques ......................414
4. Copula-Based Probability Density Model Construction Techniques.......................418
5. Stochastic Differential EquationeBased Probability Density Model
Construction Techniques...........................................................................................421
5.1 The Transformation of Variables Approach ..................................................... 422
5.2 The Characteristic Function Approach ............................................................. 425
5.3 The Functional Approach.................................................................................. 426
6. Bayesian Maximum EntropyeBased Multivariate Probability Density
Model Construction Techniques ...............................................................................428
7. Methodological and Technical Comments ...............................................................431
CHAPTER XII Spatiotemporal Random Functionals ............................................. 433
1. Continuous Linear Random Functionals in the SpaceeTime Domain ...................433
1.1 Basic Notions .................................................................................................... 433
1.2 Generalized Fourier Transform ......................................................................... 436
1.3 SpaceeTime Characteristic Functionals........................................................... 439
1.4 Functional Derivatives....................................................................................... 441
2. Gaussian Functionals ................................................................................................447
CHAPTER XIII Generalized Spatiotemporal Random Fields ................................ 455
1. Basic Notions ............................................................................................................455
1.1 The Notion of Generalized Spatiotemporal Random Field.............................. 456
1.2 Generalized Spatiotemporal Random Field Properties and Physical
Significance ....................................................................................................... 461
1.3 Homostationary Generalized Spatiotemporal Random Fields ......................... 464
2. Spatiotemporal Random Fields of Orders n/m..........................................................468
2.1 Departure From SpaceeTime Homostationarity .............................................. 468
2.2 SpaceeTime Detrending ................................................................................... 470
2.3 Ordinary Spatiotemporal Random Field-n/m Representations of the
Generalized Spatiotemporal Random Field-n/m ............................................... 474
2.4 Determination of the Operator Qn/m and Its Physical Significance .................. 475
3. The Correlation Structure of Spatiotemporal Random Field-n/m ............................477
3.1 SpaceeTime Functional Statistics .................................................................... 477
3.2 Generalized Spatiotemporal Covariance Functions.......................................... 479
3.3 Generalized Spectral Representations and Permissibility of Generalized
Covariances........................................................................................................ 481
3.4 Generalized Covariance Function Models........................................................ 484
CONTENTS xiii
4. Discrete Linear Representations of Spatiotemporal Random Fields .......................490

4.1 SpaceeTime Random Increments .................................................................... 490
4.2 SpaceeTime Variogram Analysis ..................................................................... 495
CHAPTER XIV Physical Considerations................................................................ 501
1. Spatiotemporal Variation and Laws of Change........................................................501
2. Empirical Algebraic Equations.................................................................................504
3. Physical Differential Equations ................................................................................506
4. Links Between Stochastic Partial Differential Equation and Generalized
Random Fields ..........................................................................................................512
4.1 Links in Terms of the Random Functional....................................................... 513
4.2 Links in Terms of the Detrending Operator ..................................................... 515
5. Physical Constraints in the Form of Integral Relationships, Domain
Restrictions, and Dispersion Equations ....................................................................518
CHAPTER XV Permissibility in SpaceeTime........................................................ 521
1. Concerning Permissibility.........................................................................................521
2. Bochnerian Analysis .................................................................................................522
2.1 Main Results...................................................................................................... 523
2.2 Limitations of Bochnerian Analysis ................................................................. 525
3. Metric Dependence ...................................................................................................528
4. Formal and Physical Permissibility Conditions for Covariance Functions .............529
4.1 Permissibility Conditions for SpaceeTime Homostationary Covariance
Functions ........................................................................................................... 530
4.2 Permissibility Conditions for SpaceeTime Isostationary Covariance
Functions ........................................................................................................... 532
4.3 Permissibility Conditions for Generalized Spatiotemporal Covariance
Functions ........................................................................................................... 535
4.4 Permissibility Conditions for Spatiotemporal Covariance Matrices ................ 537
5. More Consequences of Permissibility ......................................................................540
CHAPTER XVI Construction of Spatiotemporal Covariance Models ................... 543
1. Introduction.............................................................................................................543
2. Probability Density FunctioneBased and Related Techniques .............................545
2.1 Linking Directly Covariance Models and Probability Density
Functions ......................................................................................................... 545
2.2 Using Polynomial-Exponential Functions ...................................................... 548
2.3 Using Spectral Functions ................................................................................ 550
3. Delta and Related Techniques ................................................................................552
3.1 Basic Decomposition....................................................................................... 552
xiv CONTENTS
3.2 Normalized Angular Spectrum Decomposition.............................................. 554

3.3 Normalized Frequency Spectrum (or Coherency Function)
Decomposition................................................................................................. 555
4. Space Transformation Technique ...........................................................................557
5. Physical Equation Techniques ................................................................................560
5.1 Covariance Construction From Stochastic Partial Differential
Equation Representations ................................................................................ 560
5.2 Covariance Construction From Algebraic Empirical Relationships .............. 570
6. Closed-Form Techniques ........................................................................................572
7. Integral Representation Techniques........................................................................580
8. SpaceeTime Separation Techniques ......................................................................582
9. Dynamic Formation Technique ..............................................................................586
10. Entropic Technique .................................................................................................587
11. Attribute and Argument Transformation Techniques ............................................588
11.1 Attribute Transformation............................................................................... 588
11.2 Argument Transformation ............................................................................. 589
12. Cross-Covariance Model Construction Techniques ...............................................590
13. Revisiting the Role of Physical Constraints...........................................................593
14. Closing Comments..................................................................................................594
Exercises ............................................................................................................................................ 597
References.......................................................................................................................................... 643
Appendix............................................................................................................................................ 653
Index ................................................................................................................................................. 665
Preface
The study of real-world phenomena relies on theories of natural (physical, biological, etc.) sciences
that generally involve mathematical models. These models are usually defined by a set of equations
and supplemented by a set of logical reasoning rules for rigorously translating the quantitative analysis
results into meaningful statements about the phenomenon of interest. Additionally, and very impor-
tantly, the real-world study of a phenomenon is faced with various sources of uncertainty, ontic,
and epistemic (including phenomenal, technical, conceptual, and computational sources related to
quantitative modeling, data selection, and processing). As a result, exact deterministic model solutions
in terms of well-known analytic functions often turn out to be unrealistic and lack any visible means of
meaningful interpretation.
In light of the above considerations, in most real-world applications the mathematical models that
we currently use to describe the attributes of a natural phenomenon are stochastic in nature, i.e., these
attributes and the associated boundary/initial conditions are represented by random fields with argu-
ments in a composite spaceetime domain. In this domain, space represents the order of coexistence,
and time represents the order of successive existence of the attribute. Randomness manifests itself as
an ensemble of possible realizations regarding the attribute distribution, where the likelihood that each
one of these possible realizations occurs is expressed by the corresponding probability law. Thus,
spatiotemporal random model solutions are considerably more flexible and realistic than the determin-
istic single-valued solutions. Attribute distributions are well represented by theoretical probability
laws, and this permits us to calculate various spaceetime properties of these distributions with reason-
able accuracy.
The above considerations are the primary reasons for devoting this book to the spatiotemporal
random field theory and its potential applications in natural sciences. In this context, for any such the-
ory there is first the mathematical problem of analyzing, as far as possible, the stochastic model gov-
erning the relevant attributes together with the available data sets (hard or exact and soft or uncertain,
in general), and of finding as realistic and complete a solution as possible to the problem of interest that
maintains good contact with the real-world phenomenon in conditions of in situ uncertainty. Next
comes the interpretation (mathematical and physical) of the conclusions thus obtained, and their uti-
lization to make informative predictions. It should be pointed out that certain exact models and equa-
tions have played very important roles in the study of natural phenomena. It should also be noticed that
because many models and equations describing real-world phenomena are necessarily complicated
(multiparametered, highly nonlinear, and heterogeneous, whereas potentially critical features of the
phenomena remain unspecified), it is very useful to understand what qualitative features these models
and equations might possess, since they have been proven to offer an invaluable guide about the phe-
nomenon of interest.
Certainly, there are several important issues related to the distinction between theory and imple-
mentation. Concerning the in situ implementation of random field theory and techniques, one should
be aware that, although the fact is not always appreciated, the real-world complexities of the phenom-
ena mentioned above also mean that approximate techniques used as part of implementation could run
into hidden complications that have a tendency to distract attention from more useful issues. The sit-
uation may also be partially the fault of those of us working in this discipline, when we occasionally
xv
xvi PREFACE
propose abstract theories for the sake of greater “generality.” Yet this is not an excuse for the fact that,
as real-world experience shows, in the vast majority of cases the “ineffectiveness” is not a feature of
the theory or the modeling technique used but rather of the practitioners often attempting to use them
in a “black-box” manner. In this framework, exact models and equations that can be compared with
approximate or numerical results are very useful in checking the validity of approximation techniques
used in an application.
In addition to the above reasons for devoting this book to the theory of spatiotemporal random
fields, it should be noted that although much work has been done concerning the subject, it is often
not generally known because of the plethora of disciplines, journals, and mathematical terminologies
and notations in which it has appeared. It is hoped that one beneficial effect of the present effort will be
to save the interested readers from spending their time rediscovering already known results. And I
hope the present attempt to characterize the known results invariantly will help readers to identify
any new findings that may emerge. Throughout the book, for the basic issue regarding fundamental
concepts of probability, statistics, and random variables, I refer the reader to other texts, when
necessary.
Naturally, I begin by introducing the basic notions of the spaceetime continuum (points, lags, met-
rics, and geometrical transformations), mathematical field and related functions, probability, uncer-
tainty, and randomness (Chapter 1). Two chapters (Chapters 2 and 4) are devoted to the
presentation of standard results of the ordinary spatiotemporal random field theory, including much
of the terminology used later in the book. Among these two chapters I have interpolated one chapter
(Chapter 3) on spaceetime metrics. Its position is due to the fact that the spaceetime metric properties
can be used to elucidate the validity of certain random field issues introduced in the preceding two
chapters. This chapter also discusses the classification of spaceetime metrics for scalar and vector
random fields, and a physical lawebased metric determination technique is outlined and applied in
several cases. Intuitively, a natural attribute represented as a spatiotemporal random field is projected
on the physical (real) domain. Yet, there may be constructed other domains on which an attribute could
be projected. Such domains are the spectral domain, the reduced dimensionality, and the traveling do-
mains, which provide equivalent representations of the attribute defined on it. And while one’s intu-
ition may be better adapted to the physical domain, in certain cases it may be more convenient to
work in the alternative domains. So in Chapter 5 we discuss important concepts and methods associ-
ated with these alternative domains. Chapter 6 focuses on spatiotemporal random field geometry (con-
tinuity, differentiability, and integrability). This is one of the subjects that would warrant a book of its
own and, thus, I had to be very selective in the choice and manner of the material presented. Because of
its special physical and mathematical interest, the topic of homostationarity (space homogeneous/time
stationary) was given a chapter of its own (Chapter 7). Similar reasons are valid for devoting Chapter 8
to isostationary (space isotropic/time stationary) random fields. In this chapter a large number of
spatiotemporal variability functions (covariance, variogram, and structure functions of high order)
are presented. Chapter 9 deals with multivariate and vectorial random fields varying in the spaceetime
domain, including their main mathematical features and differences as regards their interpretation
(mathematical and physical). In Chapter 10, I discuss a selected group of spatiotemporal random fields
with special properties of particular interest to applications (this group includes the frozen random
field and its variations, the plane-wave, the lognormal, the spherical, and the Lagrangian random
fields). Chapter 11 focuses on techniques for constructing multivariate probability density functions
that offer a complete characterization of the spatiotemporal random field in stochastic terms. Due
PREFACE xvii
to their fundamental role in the study of spaceetime heterogeneous random fields, an entire chapter
(Chapter 12) is devoted to the theory of spatiotemporal random functionals. Indeed, the functional
description of randomness naturally involves more complex mathematics, but it has its rewards on
both theoretical and application grounds (e.g., many real-world phenomena and their measurements
need to be expressed in terms of random functionals). Chapter 13 provides a rather detailed account
of the theory of spaceetime heterogeneous (generalized) random fields that is useful in the case of
natural attributes characterized by complex variations and patterns (varying trends, fluctuations of
varying magnitude, coarse-grained measurements, etc.). Interestingly, since the first edition (1992)
of the present book, only certain limited aspects of this theory have been thoroughly discussed in
the literature. Chapter 14 emphasizes the importance of accounting for physical laws, scientific
models, and empirical relationships in the development of a spatiotemporal random field theory.
This valuable core knowledge concerning a phenomenon is usually quantitatively expressed in terms
of stochastic partial differential equations, several of which are reviewed in this chapter. Admittedly, I
only tangentially deal with the solution of these equations and relevant topics. The strongest reason for
excluding the omitted topics is that each would fill another book (I do, of course, give references to the
relevant literature). Chapter 15 presents a series of permissibility criteria for spaceetime covariance
functions (ordinary and generalized) that are widely used in applied stochastics. Certain of these
criteria are necessary and sufficient, whereas some others are only sufficient, but they have the advan-
tage that they refer directly to the covariance function itself. Further, some important practical impli-
cations of permissibility in different kinds of applications are discussed. Chapter 16 presents a rather
large number of techniques for constructing spaceetime covariance models, which can be used in a
variety of scientific applications. Formal and substantive model-building techniques are examined,
each of which has its own merits and limitations. There are many covariance model construction tech-
niques in use and they could not all be discussed in full: my choice of what to present in detail and what
to mention only as a reference simply reflects my personal taste and experience.
The book has benefited by the contributions in the field of my colleagues, collaborators, and stu-
dents during the last few decades. The second edition of the book was written mainly during my leave
of absence year at the Ocean College of Zhejiang University (China). I am grateful for the support of
Zhejiang University and of the CNSF (Grant no. 41671399). I am also grateful to my colleagues at the
Ocean College, particularly Professor Jiaping Wu, who did everything possible to create the right envi-
ronment for writing such a book. Last but not least, this work would not have been completed without
Yongmei’s infinite patience during the long process of writing the book, which is why to Yongmei this
book is dedicated.
George Christakos
CHAPTER
SPACE, TIME, SPACEeTIME,

RANDOMNESS, AND
PROBABILITY
I
CHAPTER OUTLINE
1. Introduction .......................................................................................................................................1
2. SpaceeTime Continuum and Kolmogorov Probability Space ..................................................................4
2.1 SpaceeTime Arguments: Points, Lags, Separations, and Metrics...........................................4
2.2 Transformations and Invariance in SpaceeTime.................................................................17
2.3 SpaceeTime Interpretations.............................................................................................23
2.4 Functions of SpaceeTime Arguments ...............................................................................26
3. Random Variables in SpaceeTime.....................................................................................................31
3.1 Kolmogorov’s Probability Theory .......................................................................................31
3.2 Useful Inequalities ..........................................................................................................35
3.3 Convergence of Random Variable Sequences .....................................................................37
1. INTRODUCTION
Due to its importance in almost any scientific discipline, random field theory is an active area of
ongoing research. Significant work has been done, indeed, in the theory of spatial random field, but
much less so in the theory and applications of spatiotemporal random field, where many important
topics still need to be studied and notions to be advanced. On the other hand, many practitioners argue
that random field remains a tough theory to work with, due to the difficulty of the “nondeterministic”
mathematics involved. This kind of mathematics is also known as stochastics, a term that generally
refers to the mathematical representation of phenomena that vary jointly in space and time under
conditions of in situ uncertainty. In a formal sense, deterministic mathematics can be viewed as a
special case of stochastics under the limiting and rare conditions that the phenomenon under study is
known with certainty. To phrase it in more words, stochastics deals with any topic covered by the
deterministic theory of functions, and, in addition, the presence of uncertainty (technically, sometimes
characterized as randomness) makes stochastics a much larger, considerably more complex and surely
more challenging subject than the deterministic theory of functions. Historically, the development of
stochastics can be traced back in the works of some of the world’s greatest scientists, such as Maxwell
(1860), Boltzmann (1868), Gibbs (1902), Einstein (1905), Langevin (1908), Wiener (1930), Heisen-
berg (1930) Khinchin (1934), Kolmogorov (1941), Chandrasekhar (1943), Lévy (1948), Ito (1954),
Gel’fand (1955), von Neumann (1955), Yaglom (1962), and Bohr (1963), among many others.
Spatiotemporal Random Fields. http://dx.doi.org/10.1016/B978-0-12-803012-7.00001-5

Copyright © 2017 Elsevier Inc. All rights reserved.
1
2 CHAPTER I SPACE, TIME, SPACEeTIME, RANDOMNESS, AND PROBABILITY
It should be pointed out that random field modeling is at the heart of many theoretical advances in
stochastics. It has led to the development of new mathematical concepts and techniques, and, also, it
has raised several interesting theoretical questions worthy of investigation. Computational random
field modeling, on the other hand, deals with computational and numerical aspects of the systematic
implementation of random field theory in the study of complex real-world phenomena, which covers
almost every scientific and engineering discipline. The term “computational” used here should not
create any confusion with computational mathematics and statistics: while computational mathematics
and statistics are concerned with numbers, computational random fields are concerned with physical
quantities.1
In applied sciences, random field modeling deals with spatiotemporal natural attributes, that is,
real-world attributes that develop simultaneously in space and time, and they are measurable or
observable. These natural attributes occur in nearly all the areas of applied sciences, such as ecology
and environment (e.g., concentrations of pollutants in environmental mediadwater/air/soil/biota),
climate predictions and meteorology (e.g., variations of atmospheric temperature, density, moisture
content, and velocity), hydrology (e.g., water vapor concentrations, soil moisture content, and pre-
cipitation data consisting of long time series at various locations in space), oil reservoir engineering
(e.g., porosities, permeabilities, and fluid saturations during the production phase), environmental
health (e.g., human exposure indicators and doseeeffect associations), and epidemiology (e.g., breast
cancer incidence, and Plague mortality). In all these cases, a central issue of random field modeling is
factual accuracy in the informational statements that describe what was observed and experienced.
For sure, the application of random fields in the study of real-world phenomena is not an un-
constrained theoretical exercise. It rather follows certain methodological criteria that involve the
identification of the bounds of the specific application, the evaluation of the context in making sense of
empirical data, a focus on probative evidence from diverse sources, an openness to inductive insights,
and an in-depth analysis justified by the generation of interpretable results. Induction, interpretation,
and abstraction are not competing objectives in this approach, but mutually reinforcing operations.
Random field modeling is concerned, although to varying extends, about both its internal and
external validity. Internal validity relates to whether the findings or results of the random field modeling
relate to and are caused by the phenomena under investigation, and not by other unaccounted for in-
fluences. On the other hand, external validity is assessed by the extent to which these findings or results
can be generalized, and thus applied to other real-world situations. While internal validity is the primary
concern of random field modeling, external validity is also a very important goal.
In this Chapter, I present the fundamentals regarding the conceptual and quantitative characterization
of spaceetime (or space/time, or spacetime) within which random fields will be defined in subsequent
chapters. Arguably, there are many issues surrounding the use and nature of the notion of “spaceetime”
in scientific modeling, and some of them are even controversial. Yet, “space” and “time” are funda-
mental concepts that were invented by humans in their effort to describe Nature, but the map is not the
territory. The formulation of spaceetime introduced in this chapter has the considerable merit of
maintaining close contact between mathematical description and physical reality. Among the central
goals of this formulation are to direct us toward a correct interpretation of spaceetime, and, to the extend
possible, to help us avoid asking the wrong questions and focus on the insignificant issues.
1
Surely, physical quantities are represented by numbers, but they also possess physical meaning and content (which are
ignored by computational mathematics and statistics), and they are also associated with spaceetime arguments.
1. INTRODUCTION 3
In this book, random quantities like the random variable, the random field, and the vector random
field will be studied in both the physical (real) and the frequency domains. Notationally, a random
variable is represented by lowercase Latin letters (x, y, etc.), a random field by uppercase Latin letters
(X, Y, etc.), and a vector random field by uppercase bold Latin letters (X, Y, etc.). Lowercase Greek
letters (c, j, etc.) denote random variable or random field values (realizations), and lowercase bold
Greek letters (c, j, etc.) denote vector random field values. The frequency domain counterparts of the
above random quantities will be denoted by inserting the symbol w above them (e.g., X). e The N
denotes the set of natural numbers (f0; 1; 2; .g or f1; 2; .g, depending on the context). The R1 (or R)
and C denote, respectively, the spaces of real and complex numbers. In the latter case, z ¼ c þ ij ˛ C,
pffiffiffiffiffiffiffi
where i ¼ 1, and the c h Re(z) ˛ R1 and j h Im(z) ˛ R1 denote, respectively, the real and
imaginary parts of z. The complex conjugate of z is denoted by z ¼ c ij ˛ C; also,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
jzj ¼ jz j ¼ c2 þ j2 ˛R1þ;0 (i.e., the positive part of the real line including zero) is the modulus of z,
z. The symbol T is sometimes used to represent the time domain as a subset of R1 (T 4 R1), which is
in agreement with the physical irreversibility of most real-world phenomena. On the real line R1, I use
the convention for closed, open, and half-open intervals written as [c, j], (c, j), [c, j), and (c, j].
Also, Rn is the Euclidean space of dimension n 1.
Scalar, vector, and matrix notation will be used, noticing that scalars can be seen as tensors of rank
n
zero, vectors have rank Pnone, and matrices have rank two. A vector in R will be denoted as
s ¼ ðs1 ; .; sn Þ or s ¼ i¼1 si εi, where εi, i ¼ 1; .; n, are base vectors along the coordinate di-
rections. The simplest choice of an (orthonormal) basis is, of course, the set of unit length vectors εi,
where the ith component is 1 and all others 0. For any two vectors s and s0 ¼ s01 ; .; s0n , their scalar
P 1
product is defined as s$s0 ¼ ni¼1 si s0i . The length of the vector s is jsj ¼ ðs$sÞ2 , and
1
js s0 j ¼ ððs s0 Þ$ðs s0 ÞÞ2 is the distance between s and s0 in Rn. The spaceetime domain is
denoted as Rnþ1, or as Rn,1, if we want to explicitly distinguish space from time (for the same reason,
we may also denote the spaceetime domain as the Cartesian product Rn R1 or Rn T). I.e., in the
case of spaceetime, the domain dimensionality increases to n þ 1 by including the additional term s0
or t representing time. Table 1.1 lists some commonly used symbols. Table 1.2 gives a list of special
Table 1.1 Commonly Used Symbols

Symbol Mathematical Expression
R
Dirac delta dsdðsÞ ¼ 1;
dðss0Þ ¼ 0:

Kronecker delta 1 if i ¼ j;
dij ¼
0 otherwise.
8
< þ1 if ði; j; kÞ ¼ ð1; 2; 3Þ or ð2; 3; 1Þ or ð3; 1; 2Þ;
>
Levi-Chivita
εijk ¼ 1 if ði; j; kÞ ¼ ð3; 2; 1Þ or ð1; 3; 2Þ or ð2; 1; 3Þ;
>
:
0 if i ¼ j or j ¼ k or k ¼ i.
Kronecker/Levi-Chivita εijkεipq ¼ djpdkq djqdkp

relationship
Table 1.2 Special Functions and Polynomials

Name Notation
Gamma function. G(n)

Bessel function of 1st kind and vth order. Jv
Modified Bessel function of the 1st kind. In
Modified Bessel function of 2nd kind and vth order. Kv
Gegenbauer polynomials of degrees l, q. Gl,q
functions and polynomials that will be used in the mathematical expression of several results
throughout the book. For the readers’ convenience, the mathematical definitions and basic properties
of these special functions and polynomials are briefly reviewed in the book’s Appendix.
Although many of the theoretical results presented in each chapter of the book (in the form of
propositions and corollaries) will be repeatedly used in subsequent chapters, most proofs and other
details will not, so they will not be discussed. Instead, examples illustrating the most important
application-related aspects of these proofs will be presented when appropriate. A consistent effort is
made throughout the book to keep a balance between abstract mathematical rigor and real-world science.
In many cases, this means that by suppressing certain strict mathematical conditions, a more realistic
representation of the observed phenomenon is achieved, and, also, a richness of new material is produced
(this is, e.g., the case with spaceetime metric). The remaining of this chapter presents a review of the
basic concepts and principles (regarding space, time, field, uncertainty, and probability) around which
the spatiotemporal random field theory will be developed in the following chapters of the book.
2. SPACEeTIME CONTINUUM AND KOLMOGOROV PROBABILITY SPACE

This section presents an overview of the fundamental notions pertaining to the description of the
spaceetime domain shared by the natural phenomena and the mathematical constructions that
represent them. A general point to be stressed is that although on formal mathematics grounds many
aspects of the analysis in the n þ 1-dimensional (spaceetime) domain are similar to those in the
n-dimensional (spatial) domain, essential differences could exist on physical grounds. On the same
grounds, crucial links may emerge between space and time, in which case the deconstruction of the
concept of spaceetime into separate types of space and time may be an unnecessary conceptualization
often leading to unsatisfactory conclusions.
Having said that, I start with the introduction of the different geometrical notions and arguments
that play a central role in the spatiochronological specification of a phenomenon, an attribute or an
event occurring in physical spaceetime, highlighting similarities and differences as they emerge.
2.1 SPACEeTIME ARGUMENTS: POINTS, LAGS, SEPARATIONS, AND METRICS

Generally, a continuum E of spaceetime points is associated with the physical spaceetime domain,
and as such, it is assumed to be equipped with the corresponding metric that specifies “distances”
between points in spaceetime. Operationally, a point in E can be defined in different ways, depending
on the situation and the objectives of the analysis.
2. SPACEeTIME CONTINUUM AND KOLMOGOROV PROBABILITY SPACE 5
Definition 2.1
A point in E is denoted by a vector p, which can be defined, either as an element of the n þ 1-
dimensional domain,
p ¼ ðs1 ; .; sn ; s0 Þ ˛ Rnþ1 ; (2.1a)
where, in notational terms, s1 ; .; sn denote space coordinates and time is considered jointly with space
using the convention s0 ¼ t; or, as a pair of elements
p ¼ ðs; tÞ ˛ Rn;1 ; (2.1b)
where s ¼ ðs1 ; .; sn Þ ˛ Rn is the spatial location vector considered separately than the time instant
t ˛ R1. As noted earlier, instead of Rn,1 some authors use the Cartesian domain notation Rn T, i.e., the
time axis T (4R1) is indicated separately from space Rn.
The point vector p plays a key role in physical sciences, not only because it uniquely specifies a point in
the spaceetime domain of interest, but also because of the numerous functions of p (also known as fields,
see Section 2.4) encountered in physical sciences. In Definition 2.1, either the entire real line may be
considered as the time axis (referring, e.g., to situations in which the attribute of interest is reversible), or,
due to physical requirements (i.e., the attribute is irreversible in the real-world), the domain R1þ;f0g
(restricted to the positive part of the real line including zero) may be more appropriate to represent the time
axis. To phrase it in more words, time may be seen as a coordinate (s0) of the vector p in the Rnþ1 domain,
as in Eq. (2.1a). Intuitively, a spaceetime point in Rnþ1 is considered as a fusion of a space point and a
time point (e.g., the “here and now” exists as a unity not specifying the “here” and the “now” separately;
similar is the interpretation of the “there and then”). Alternatively, time may enter the analysis as a distinct
variable via the vectorescalar pair (s, t) of the Rn,1 domain, as in Eq. (2.1b). Both approaches have their
merits and uses, which will be discussed in various parts of the book. Yet, one should be aware of certain
noticeable consequences of the two approaches in applications, as illustrated in the examples below.
Example 2.1
From a modeling viewpoint, in many cases we may treat space and time on essentially the same formal
footing. By setting, e.g., s0 ¼ iat, the wave operator2 can be written as
v2 X X
n n;0
v2 v2
a2 ½$ 1 ½$; (2.2)
vt 2
i¼1
vsi
2
i¼1
vs2i
and we may view the solution of Poisson’s equation in two dimensions and that of the wave equation in
one dimension as analogous problems. The physical differences between space and time need to be
carefully taken into consideration, though, since certain formal analogies between space and time may
be deceptive. Indeed, the boundary conditions and the initial conditions may enter the problem in
different ways, even though the governing equation may look symmetric (as in the wave operator
above). Also, when we study the spaceetime variation of an air pollutant, the way pollution con-
centration changes across space (the distribution of spatial locations in which pollution exceeds a
critical threshold) can be essentially different than concentration changes as a function of time (fre-
quency of threshold exceedances at each location).
2
This operator expresses certain fundamental physical laws like the electromagnetic laws.
Our discussion so far of the issues surrounding the essence and use of the term “spaceetime” lead
to the first postulate.
Postulate 2.1
The vast majority of real-world data are interrelated both in space and time. This spaceetime
connection is ingrained through physical relations and is welcomed in scientific modeling because it
allows the representation of the spaceetime variation of a natural attribute from the limited number of
data usually available.
This postulate is supported by reality, including the fact that spaceetime coupling is known to
remove possibly unphysical divergences from the moments of the corresponding transport processes
(e.g., Shlesinger et al., 1993). Unfortunately, these crucial facts are often ignored in purely technical
treatments of spaceetime phenomena. Indeed, in statistical inferences (i.e., the inductive process of
inferring from a limited sample valid conclusions about the underlying yet unknown population) the
proper assessment of spaceetime correlations is often problematic, since most standard statistics tools of
data analysis and processing have been developed based on the key premise of independent (physical
relation-free) experiments.
Example 2.2
Working in the classical Newtonian conceptual framework, many practitioners find it tempting to
completely separate the space component s from the time component t. Although convenient, indeed,
this approach is often inadequate in real-world studies. A common example is the model-fitting
procedure in which a valid covariance function of space is fitted at any fixed time (or a valid
covariance function of time is fitted at any fixed distance). However, the resulting model is not
necessarily a valid spatiotemporal covariance model, as has been discussed in Ma (2003b). Further-
more, in ocean studies involving underwater acoustics propagation, the travel time t is related to the
horizontal (s1, s2) and vertical (s3) coordinates in R3,1 by means of
1
s21 þ s22 2 cos q þ s3 sin q yt ¼ 0; (2.3)
where q is the angle of a ray element in a refracting medium with sound speed y (Lurton, 2010). Hence,
the acoustics of the phenomenon imply that space s and time t are closely linked through the physical
relation of Eq. (2.5).
The perspective suggested by Eq. (2.1b) enables the introduction of alternative expressions of
spaceetime point determination, while still accounting for the spaceetime connection posited in
Postulate 2.1. These expressions are called conditional, and the reason for this will become obvious
below. Consider a spaceetime domain represented by the nodes i (i ¼ 1; .; m) of a grid or lattice. An
obvious expression of spaceetime at each grid node i is
ðsi ; ti Þ; i ¼ 1; .m; (2.4a)
denoting location si ¼ ðsi1 ; .; sin Þ and time ti. Eq. (2.4a) assigns the same subscript to space and time,
and, hence, it allows the consideration of a unique time instant at each node i. Sometimes, the physics
of the situation may require that more than one time instants need to be considered at each node, in
which case it is convenient to represent spaceetime at each grid node by

si ; tji ; i ¼ 1; .; m; and j ¼ 1; .; m0 ; (2.4b)
where now ji denotes the time instant considered given that we are at spatial position si (e.g., at a given
node 9, ji ¼ 19 means time instant 1 at node 9). Similarly, when more than one nodes need to be
considered during each instant, we can write,

sij ; tj ; i ¼ 1; .; m; and j ¼ 1; .; m0 ; (2.4c)
where now ij denotes the spatial position given that we are at time instant tj (e.g., at a given instant 9,
ij ¼ 19 means spatial node 1 at instant 9). In a sense, Eqs. (2.4b) and (2.4c) introduce conditional
expressions of the spaceetime point (i.e., time conditioned to space and space conditioned to time,
respectively).
In addition to the physical spaceetime domain one may refer, equivalently, to the location-instant
of a physical attribute occurring in the frequency spaceetime domain F .
Definition 2.2
A point in F is denoted by a vector w, which can be defined, either as an element of the n þ 1-
dimensional domain,
w ¼ ðw1 ; .; wn ; w0 Þ ˛ Rnþ1 ; (2.5a)
or, as a pair of elements of the ( n, 1)-dimensional domain,
w ¼ ðk; uÞ ˛ Rn;1 ; (2.5b)
where k ¼ ðk1 ; .; kn Þ denotes spatial frequency (wavevector) and u denotes temporal (wave scalar)
frequency.
In view of Definition 2.2, as was the case with the representations of Eqs. (2.1a) and (2.1b) being
formally equivalent, the same is valid for the representations of Eqs. (2.5a) and (2.5b), but they may be
both affected by physical context too. The attributes of interest in a real-world study satisfy certain
laws of change that may impose some intrinsic links between the location vector s and the time instant t
or, equivalently, between the wavevector k and the wave frequency u. Specifically, the matter of
spaceetime links is discussed in Section 3 of Chapter VII, in which it is shown that physical laws
representing wave phenomena impose a link between k and u, which are not independent but closely
linked by means of a physical dispersion relation.
Example 2.3
The acoustic pressure X(s, t) of a wave propagating in spaceetime is governed by the classical law

2 v
2
y V Xðs; tÞ ¼ 0;
2
(2.6a)
vt2
where y is the local sound speed. The dispersion relation of this underwater acoustics law in R3,1 that
links k and u is
k2 y2 u2 ¼ 0; (2.6b)
P
where k2 ¼ 3i¼1 ki2 .3 Similarly to Eq. (2.3), underwater acoustics laws representing wave phenomena
impose a strong link between k and u, which are not independent anymore, but they are closely
connected physically by the dispersion relation of Eq. (2.6b). As a result of this dependence, the
spaceetime frequency domain is restricted, thus affecting the form of the spaceetime covariance
function (more details in Chapter VII).
A real (or complex) function of the spaceetime vector p (including the random fields to be
considered, Chapter II) is termed continuous or discrete parameter according to whether its argument p
takes continuous or discrete values. Unless stated otherwise, in the following continuous-parameter
functions will be considered, which is most often the case in applied stochastics. Spaceetime varia-
tion analysis requires the simultaneous consideration of pairs of points defined by the vectors p and p0 ,
in which case the notion of spaceetime lag emerges naturally.
Definition 2.3
The spaceetime lag, Dp ¼ p0 p, between a pair of points p and p0 is defined, either as the single
vector
Dp ¼ ðDs1 ; .; Dsn ; Ds0 Þ ˛ Rnþ1 ; (2.7a)

where Dsi ¼ s0i si (i ¼ 1; .; n; 0) with the convention that the vector components with i ¼ 1; .; n
denote space lags and the component with i ¼ 0 denotes time separation; or, as a pair of vectorescalar
components
Dp ¼ ðh; sÞ ˛ Rn;1 ; (2.7b)

where the spatial vector lag h ¼ ðh1 ; .; hn Þ ˛ Rn , with hi ¼ s0i si (i ¼ 1; .; n), and the time scalar
lag, s ¼ t0 t, are represented separately.4
A direct correspondence can be established between the representations of Eqs. (2.7a) and (2.7b)
by observing that
Dsi ¼ hi ði ¼ 1; .; nÞ;
Ds0 ¼ h0 ¼ s ði ¼ 0Þ.
For simplicity, in the following the notations Dsi and hi (i ¼ 1; .; n; 0) will be used inter-
changeably. As in Definition 2.1, Eq. (2.7a) allows an intrinsic mixing of space and time lags, whereas
Eq. (2.7b) considers them explicitly. As we will see in various parts of the book, this distinction may
have considerable consequences in the study of certain aspects of random field theory and its physical
applications.
Remark 2.1
At this point, it is worth noticing that the above spaceetime notation is sometimes termed the Eulerian
spaceetime coordinate representation, where s and t are allowed to vary independently. Another
3
Here, the dispersion relation results from the requirement that X ðs; tÞ ¼ eiðk$sutÞ is a solution to the physical law.
4
The lag sign, i.e., the order of the spaceetime points p, p0 in the lag can affect mathematical manipulations and also can be
consequential in certain applications (Chapter VII).
possibility is the Lagrangian spaceetime coordinate representation, in which case s and t do not vary
independently, but, instead, s is considered a function of t, which is expressed as s(t). The Eulerian
notation has been used, e.g., in turbulence studies to represent correlations between physical attributes
(pressure or flow velocity) where s ¼ ðs1 ; .; sn Þ specifies a measurement location, s þ h denotes
locations at varying distances h from s, and s is the time increment between times t and t þ s. The
Lagrangian setting has been used to determine correlations for properties of fluid particles passing
through locations s at times t0 (different for each particle), traveling along certain trajectories and
arriving at locations s þ h(t0 þ s) at times t0 þ s. The displacement vector h(t0 þ s) is a random
variable describing the locations at times t þ s of the particles in the ensemble averaging with respect
to the initial locations s at times t0. In this book, Eulerian spaceetime coordinates will be predomi-
nantly considered, whereas the Lagrangian notation will be used only in special cases.
Spaceetime points and lags viewed as vectors, standard vector operations can be applied on them.
A partial list of such operations that are useful in the following is given in Table 2.1 (see also
Appendix). The spaceetime point vector p can be expressed in terms of the unit vectors along the
(orthogonal) coordinate directions (base vectors) εi, i ¼ 1; .; n. The notation for integer powers
involving spaceetime coordinates, pl, where l ¼ (l1,., ln, l0) is a multi-index of nonnegative
Table 2.1 Standard Vector Operations on SpaceeTime Points and Lags

P
n;0
p ¼ ðs1 ; .; sn ; s0 Þ ¼ s i εi
i¼1
Y
n;0 Y
n
pl ¼ sli i ¼ sli i tl0 ; l ¼ ðl1 ; .; ln ; l0 Þ
i¼1 i¼1
X
l
ðs þ hÞl ¼ k! k lk
Cl! s h ; k ¼ jkj; l ¼ jlj; k l
k¼0
X
l
ðt þ sÞl ¼ m! m lm
Cl! t s (2.8aeg)
m¼0

X n k
k 2 2
jsj ¼ si ðk ¼ 1; 2; .Þ
i¼1
vjsj
¼ si jsj1 ¼ zi
vsi
vzk
¼ ðdik zi zk Þjsj1
vsi
vp
¼ εi
vsi
vð p$p0 Þ vp 0 vp0
¼ $p þ p$ (2.9aec)
vsi vsi vsi
vð p p0 Þ vp vp0
¼ p0 þ p
vsi vsi vsi
Continued
Table 2.1 Standard Vector Operations on SpaceeTime Points and Lagsdcont’d

X
n;0
X
n;0
Dp ¼ p p0 ¼ si s0i εi ¼ h i εi
i¼1 i¼1
vDp (2.10aec)
¼ εi ði ¼ 1; .; n; 0Þ
vsi
vDp
¼ εi
vs0i
V$p ¼ V$Dp ¼ n þ 1
(2.11aeb)
V p ¼ V Dp ¼ 0
P
integers such that jlj ¼ n;0i¼1 li and l! ¼ Pi li (Table 2.1), is essential in the study of random fields
representing the spaceetime heterogeneous variation pattern of a physical attribute, and can be
combined with other functions, such as, e.g., pl eg$p (Chapter XIII).5 Also, spaceetime vector dif-
ferentiation operators in terms of p and Dp are useful in the study of derivative random fields, their
covariance, variogram, and structure functions (Chapters VI and VII), as well as the physical laws they
obey (usually expressed in terms of partial differential equations, PDE, Chapter XIV). In such and
similar cases, the proper representation of spaceetime geometry (in terms of coordinates, distances,
lags etc.) allows an efficient approach to random field modeling, which is the subject of this book.
More specifically, if the random field represents a spaceetime heterogeneous attribute distribution
(i.e., one exhibiting a spaceenonhomogeneous/timeenonstationary, or, simply, spaceetime hetero-
geneous pattern),6 the following space/time polynomial functions appear in the analysis (Chapter XIII)
m X
X n
r
hn=m ðs; tÞ ¼ cr;z s11 .srnn tz ; (2.12a)
z¼0 jrj¼0
where n and m are integers representing, respectively, the spatial and temporal orders of the phenomenon
heterogeneity, and cr,z are known coefficients. The notation used in Eq. (2.12a) shows explicitly the
degrees of all spatial and temporal monomials, but it is cumbersome, since it requires keeping track of all
the power exponents (ri, i ¼ 1; .; n, and z). An alternative notation, which turns out to be more effi-
cient, is based on the spaceetime monomial functions pl introduced earlier. In this setting, a spaceetime
polynomial of spaceetime degrees n=m can be expressed as (Christakos and Hristopulos, 1998)
ðn=mÞ
NnX
hn=m ðs; tÞ ¼ ca ha ðs; tÞ; (2.12b)
a¼1
P
n
In these applications, g ¼ (a, b), l ¼ (r, z), and g$p ¼
5
ai si þ bs0 .
i¼1
6
Etymologically, the words homogeneous and heterogeneous come from Ancient Greek. The Greek words mogεn 2 (homo-
genes) and sεrogεn 2 (heterogenes) come from m 2 (homos, “same”) and sεro2 (heteros, “other, different”) respectively,
followed by g no2 (genos, “kind”). The “-ous” is an adjectival suffix.
Table 2.2 Number of Monomials for n [ 2, 3, and Different Combinations

of Continuity Orders n=m
v m N2 (v/m) N3 (v/m)
0 0 1 1
1 0 3 4
0 1 2 2
1 1 6 8
2 0 6 10
2 1 12 20
1 2 9 12
2 2 18 30
n r
where ha ðs; tÞ ¼ pðr;zÞ ¼ sr tz ¼ P si i tz , and Nn ðn=mÞ is the number of monomials that depends on
i¼1
the spatial dimension n and the orders n, m. This notation is surely more compact than the one pre-
sented in Eq. (2.12a), and it involves only a single index a instead of the n þ 1 spaceetime indices
ðr1 ; .; rn ; zÞ. The Nn ðn=mÞ for any n and n=m can be then determined by
X n
ðjrj þ n 1Þ!
Nn ðn=mÞ ¼ ðm þ 1Þ ; (2.12c)
jrj¼0
jrj!ðn 1Þ!
i.e., for jrj n number of monomials it is equal to the permutations of n integers from f0; 1; .; ng that
add up to jrj. Table 2.2 displays the monomials for the n=m combinations that are most commonly used
in applications.
Example 2.4
For illustration, in this example, I chose to focus on the R2,1 domain, where (s1, s2) denote the Cartesian
coordinates associated with the locational vector s, and t is the time scalar. Also, I assume that n=m ¼
2=1 are the space/time orders of attribute nonhomogeneity/nonstationarity. Under these conditions, the
corresponding monomials (12 in number) are listed in Table 2.3.
In light of Definition 2.3, a relevant notion of spaceetime modeling is the spaceetime “distance”
or metric. In applied sciences, metrics are mathematical expressions of the concept of distance in the
spaceetime continuum E (an in depth treatment of the metric notion is presented in Chapter III). In
general, a spaceetime metric would be defined as a function of the pair ð p; p0 Þ. In many cases,
however, the value of the metric function is invariant for each given lag Dp ¼ p0 p, i.e., the metric
is defined as a function of the spaceetime lag (which is more restrictive, yet satisfactory in many
physical situations). These metrics are sometimes referred to as “lag-based metrics” and are denoted
by the absolute value of Dp, i.e., jDpj.7 Using the correct metric plays a crucial role in the study of
7
In a more conventional mathematical sense, jDpj is a real-valued function on R1 satisfying three basic conditions:
(1) jDpj 0 for all Dp (jDpj ¼ 0 only if Dp ¼ 0); (2) jaDpj ¼ jajjDpj for all Dp, a ˛ R1; and
(3) jDp1 þ Dp2j jDp1j þ jDp2j for all jDp1j, jDp2j.
Another random document with
no related content on Scribd:
schemes at a favorable moment, and the equally prompt recession
when conditions alter; the investment of great resources in
enterprises which yield no immediate return; the decision and
secrecy important in overcoming competitors; the unhesitating
sacrifice of workmen and their families when the market calls for a
shut-down of production—such traits as these are of the utmost
importance to commercial success, and belong to arbitrary control
rather than to anything of a more popular sort. On the other hand, it
would be easy to show at any length desired that such control is
accompanied by a widespread disaffection of spirit on the part of the
working classes, which, expressed in unwilling labor, strikes and
agitation, is a commercial disadvantage, and a social problem so
urgent as to unsettle the whole economic system.
The autocratic system has evidently a special advantage in a time
of rapid and confused development, when conditions are little
understood or regulated, and the state of things is one of somewhat
blind and ruthless warfare; but it is quite possible that as the new
industries become established and comparatively stable, there will
be a commercial as well as a social demand for a system that shall
invite and utilize more of the good-will and self-activity of the
workman. “The system which comes nearest to calling out all the
self-interests and using all the faculties and sharing all the benefits
will outcompete any system that strikes a lower level of motive
faculty and profit.”[120] And the penetrating thinker who wrote this
sentence believed that the function of the autocratic “captain of
industry” was essentially that of an explorer and conqueror of new
domains destined to come later under the rule of a commonwealth.
Indeed the rise, on purely commercial grounds, of a more humane
and individualizing tendency, aiming in one way or another to
propitiate the self-feeling of the workman and get him to identify
himself with his work, is well ascertained. Among the familiar phases
of this are the notable growth of coöperative production and
exchange in Belgium, Russia and other European countries, the
increasing respect for labor unions and the development by large
concerns of devices for insurance, for pensions, for profit-sharing
and for the material and social comfort of their employees. “As a
better government has come up from the people than came down
from the kings, so a better industry appears to be coming up from
the people than came down from the capitalists.”[121]
In some form or other the democratic principle is sure to make its
way into the economic system. Coöperation, labor unions, public
regulation, public ownership and the informal control of opinion will
no doubt all have a part; the general outcome being that the citizen
becomes a more vital agent in the life of the whole.
Before discussing further the power of the capitalist-manager
class, we ought to think out clearly just what we mean by social
power, since nowhere are we more likely to go astray than in
vagueness regarding such notions.
Evidently the essence of it is control over the human spirit, and the
most direct phases of power are immediately spiritual, such as one
mind exercises over another by virtue of what it is, without any
means but the ordinary symbols of communication. This is live,
human power, and those who have it in great degree are the prime
movers of society, whether they gain any more formal or
conventional sort or not. Such, for instance, are the poets, prophets,
philosophers, inventors and men of science of all ages, the great
political, military and religious organizers, and even the real captains
of industry and commerce. All power involves in its origin mental or
spiritual force of some sort; and so far as it attaches to passive
attributes, like hereditary social position, offices, bank-accounts, and
the like, it does so through the aid of conventions and habits which
regard these things as repositories of spiritual force and allow them
to exercise its function.
In its immediate spiritual phase power is at a maximum of vitality
and a minimum of establishment. Only a few can recognize it. Its
possessors, then, strive to establish and organize it, to give it social
expression and efficacy, to gain position, reputation or wealth. Since
power is not apparent to the common mind until it takes on these
forms, they are, to superficial observation and in all the conventional
business of life, the only valid evidence of it. And yet by the time
these symbols appear, the spiritual basis has often passed away.
Primary power goes for the most part unseen, much of it taking on
no palpable form until late in life, much yielding only posthumous
reputation, and much, and that perhaps the finest sort, having never
any vulgar recognition whatever.
Regarding money-value we may say, in general, that it is one
expression of the conventional or institutional phase of society, and
exhibits all that mixture of grandeur and confusion with which nature
usually presents herself to our understanding. I mean that its
appraisal of men and things is partly expressive of great principles,
and partly, so far as we can see, unjust, trivial or accidental. Some
gains are vital or organic, springing from the very nature of life and
justified as we come to understand that life; some are fanciful,
springing from the tastes or whims of the rich, like the value of
diamonds or first editions, and some parasitical, like those of the
legally-protected swindler. In general the values of the market are
those of the habitual world in all its grossness; spiritual values,
except those that have become conventional, being little felt in it.
These appeal to the future. The detailed working of market value has
no ascertainable connection with moral worth, and we must not
expect it to have. If a man’s work is moral, in the higher sense, it is in
its nature an attack upon the habitual world which the latter is more
likely to resent than reward. One can only take up that useful work
that seems best suited to him, trying to be content if its value is
small, and, if large, to feel that the power over money it gives him is
rightly his only in so far as he uses it for the general good.
The more tangible kind of social power—so far as it is intrinsic to
the man and not adventitious like inherited wealth—depends chiefly
upon organizing capacity, which may be described as the ability to
build and operate human machinery. It has its roots in tact and skill
in dealing with men, in tenacity, and in a certain instinct for
construction. One who possesses it sees a new person as social
material, and is likely to know what can be made of him better than
he knows himself.[122]
Of all kinds of leadership this has the readiest recognition and the
highest market value; and naturally so, since it is essential to every
sort of coöperative achievement. Its possessors understand the
immediate control of the world, which they will exercise no matter
what the apparent forms of organization may be. In all ages they
have gained and held the grosser forms of power, whenever these
were at all open to competition. Thus, during the early Middle Age,
men of energy and management, more or less favored by situation,
built up for themselves local authority and estate, or perhaps
exploited the opportunities for still wider organization, like the
founders of Burgundy and Brittany and the early kings of France;
very much in the same manner as men of our own day build up
commercial and industrial systems and become senators and railway
presidents.
Indeed, this type of ability was never in such demand as it now is,
for the conduct of the vast and diverse social structures rising about
us—industrial enterprises, political parties, labor unions,
newspapers, universities and philanthropies.
It has its high money value partly because of its rarity and partly
because there is a regular market for it; the need being so urgent
and obvious as to create a steady and intelligent demand. In this
latter respect it contrasts with services, like moral leadership, which
people need but will seldom pay for. A third reason is that its
possessors are almost always clever enough to know their own
value and secure its recognition.
In discussing the power of the capitalist class there is no question
of the finer and higher forms of power. We shall rarely find among
the rich any pregnant spiritual leadership, theirs being a pedestrian
kind of authority which has a great deal to do with the every-day
comfort of their contemporaries but does not attempt to sway the
profounder destinies of the race. Nor does the world often accord
them enduring fame: lacking spiritual significance their names are
writ in water. Even in industry the creative thought, the inventions
which are the germs of a new era, seldom come from money-
winners, since they require a different kind of insight.
The capitalist represents power over those social values that are
tangible and obvious enough to have a definite standing in the
market. His money and prestige will command food, houses, clothes,
tools and all conventional and standard sorts of personal service,
from lawn-mowing to the administration of a railroad, not genius or
love or anything of that nature. That wealth means social power of
this coarser sort is apparent in a general way, and yet merits a
somewhat closer examination.
We have, first, its immediate power over goods and services: the
master of riches goes attended by an invisible army of potential
servitors, ready to do for him anything that the law allows, and often
more. He is in this way, as in so many others, the successor of the
nobleman of mediæval and early modern history, who went about
with a band of visible retainers eager to work his will upon all
opposers. He is the ruler of a social system wherever he may be.
The political power of wealth is due only in part to direct
corruption, vast as that is, but is even more an indirect and perfectly
legal pressure in the shape of inducements which its adroit use can
always bring to bear—trade to the business man, practice to the
lawyer and employment to the hand-worker: every one when he
thinks of his income wishes to conciliate the rich. Influence of this
sort makes almost every rich man a political power, even without his
especially wishing to be. But when wealth is united to a shrewd and
unscrupulous political ambition, when it sets out to control legislation
or the administration of the laws, it becomes truly perilous. We
cannot fail to see that a large part of our high offices are held by men
who have no marked qualification but wealth, and would be
insignificant without it; also that our legislation—municipal, state and
national—and most of our administrative machinery, feel constantly
the grasp of pecuniary power. Probably it is not too much to say that
except when public opinion is unusually aroused wealth can
generally have its way in our politics if it makes an effort to do so.
As to the influence of the rich over the professional classes—
lawyers, doctors, clergymen, teachers, civil and mechanical
engineers and the like—we may say in general that it is potent but
somewhat indirect, implying not conscious subservience but a moral
ascendency through habit and suggestion. The abler men of this sort
are generally educated and self-respecting, have a good deal of
professional spirit and are not wholly dependent upon any one
employer. At the same time, they get their living largely through the
rich, from whom the most lucrative employment comes, and who
have many indirect ways of making and marring careers. The ablest
men in the legal profession are in close relations with the rich and
commonly become capitalists themselves; physicians are more
independent, because their art is not directly concerned with
property, yet look to wealthy patients for their most profitable
practice; clergymen are under pressure to satisfy wealthy
parishioners, and teachers must win the good will of the opulent
citizens who control educational boards.
Now there is nothing in social psychology surer than that if there is
a man by whose good will we desire to profit, we are likely to adapt
our way of thinking to his. Impelled to imagine frequently his state of
mind, and to desire that it should be favorable to our aims, we are
unconsciously swayed by his thought, the more so if he treats us
with a courtesy which does not alarm our self-respect. It is in this
way that wealth imposes upon intellect. Who can deny it?
Newspapers are generally owned by men of wealth, which has no
doubt an important influence upon the sentiments expressed in
them; but a weightier consideration is the fact that they depend for
profit chiefly upon advertisements, the most lucrative of which come
from rich merchants who naturally resent doctrines that threaten their
interest. Of course the papers must reach the people, in order to
have a value for advertising or any other purpose, and this requires
adaptation to public opinion; but the public of what are known as the
better class of papers are chiefly the comparatively well-to-do. And
even that portion of the press which aims to please the hand-working
class is usually more willing to carry on a loud but vague agitation,
not intended to accomplish anything but increase circulation, than to
push real and definite reform.
All phases of opinion, including the most earnest and honest
inquiry into social questions, finds some voice in print, but—leaving
aside times when public opinion is greatly aroused—those phases
that are backed by wealthy interests have a great advantage in the
urgency, persistence and cleverness with which they are presented.
At least, this has been the case in the past. It is a general feeling of
thoughtful men among the hand-working class that it is hard to get a
really fair statement of their view of industrial questions from that
portion of the newspaper and magazine press that is read by well-to-
do people. The reason seems to be mainly that the writers live
unconsciously in an atmosphere of upper-class ideas from which
they do not free themselves by thorough inquiry. Besides this, there
is a sense of what their readers expect, and also, perhaps, a vague
feeling that the sentiments of the hand-working class may threaten
public order.
Since the public has supplanted the patron, a man of letters has
least of all to hope or fear from the rich—if he accepts the opinion of
Mr. Howells that the latter can do nothing toward making or marring
a new book.
The power of wealth over public sentiment is exercised partly
through sway over the educated classes and the press, but also by
the more direct channel of prestige. Minds of no great insight, that is
to say the majority, mould their ideals from the spectacle of visible
and tangible success. In a commercial epoch this pertains to the
rich; who consequently add to the other sources of their influence
power over the imagination. Millions accept the money-making ideal
who are unsuited to attain it, and run themselves out of breath and
courage in a race they should never have entered; it is as if the thin-
legged and flat-chested people of the land should seek glory in foot-
ball. The money-game is mere foolishness and mortification for most
of us, and there is a madness of the crowd in the way we enter into
it. Even those who most abuse the rich commonly show mental
subservience in that they assume that the rich have, in fact, gotten
what is best worth having.
As hinted above, there is such a thing as an upper-class
atmosphere, in the sense of a state of mind regarding social
questions, initiated by the more successful money-winners and
consciously or unconsciously imposed upon business and
professional people at large. Most of us exist in this atmosphere and
are so pervaded by it that it is not easy for us to understand or fairly
judge the sentiment of the hand-working classes. The spokesmen of
radical doctrines are, in this regard, doing good service to the public
mind by setting in motion counterbalancing, if not more trustworthy,
currents of opinion.
If any one of business or professional antecedents doubts that he
breathes a class atmosphere, let him live for a time at a social
settlement in the industrial part of one of our cities—not a real
escape but as near it as most of us have the resolution to achieve—
reading working-class literature (he will be surprised to find how well
worth reading it is), talking with hand-working people, attending
meetings, and in general opening his mind as wide as possible to the
influences about him. He will presently become aware of being in a
new medium of thought and feeling; which may or may not be
congenial but cannot fail to be instructive.
FOOTNOTES:
[119] The Spirit of Laws, book v, chap. 6.
[120] Henry D. Lloyd, Man the Social Creator, 255.
[121] Idem, 246. Lloyd was rather a prophet than a man of
science, but there is a shrewd sense of fact back of his visions.
[122] Such a one
“Lässt jeden ganz das bleiben was er ist;

Er wacht nur drüber das er’s immer sei
Am rechten Ort; so weiss er aller Menschen
Vermögen zu dem seinigen zu machen.”
“He lets every one remain just what he is, but takes care that
he shall always be it in the right place: thus he knows how to
make all men’s power his own.” Schiller, Wallenstein’s Lager, I, 4.
CHAPTER XXIV
ON THE ASCENDENCY OF A CAPITALIST CLASS—
Continued
The Influence of Ambitious Young Men—Security of the

Dominant Class in an Open System—Is there Danger of
Anarchy and Spoliation?—Whether the Sway of Riches is
Greater now than Formerly—Whether Greater in
America than in England.
In any society where there is some freedom of opportunity
ambitious young men are an element of extreme importance. Their
numbers are formidable and their intelligence and aggressiveness
much more so: in short, they want an opening and are bound to get
it.
As the members of this class are mainly impecunious, it might be
supposed that they would be a notable offset to the power of wealth;
and in a sense they are. It is their interest to keep open the
opportunity to rise, and they are accordingly inimical to caste and
everything which tends toward it. But it by no means follows that they
are opposed to the ascendency of an upper class based on wealth
and position. This becomes evident when one remembers that their
aim is not to raise the lower class, but to get out of it. The rising
young man does not identify himself with the lowly stratum of society
in which he is born, but, dissatisfied with his antecedents, he strikes
out for wealth, power or fame. In doing so he fixes his eyes on those
who have these things, and from whose example he may learn how
to gain them; thus tending to accept the ideals and standards of the
actual upper class. He gives a great deal of attention to the points of
view of A, a railroad president, B, a senator, and even of C, head of
a labor organization, but to a mere farmer or laborer, whose hand is
on no levers, he is indifferent.
The students of our universities are subject to a conflict between
the healthy idealism of youth, which prevails with the more generous,
and the influences just indicated, which become stronger as
education draws closer to practical affairs. On the whole, possessed
of one great privilege and eager to gain others, they are not so close
in spirit to the unprivileged classes as might be imagined.
Thus the force of ambitious youth goes largely to support the
ascendency of the money-getting class; directly, in that it accepts the
ideals of this class and looks forward to sharing its power; indirectly,
in that it is withdrawn from the resources of the humbler class. How
long will the rising lawyer retain his college enthusiasm for social
reform if the powers that be welcome him and pay him salaries?
We have then the fact, rather paradoxical at first sight, that the
dominant class in a competitive society, although unstable as to its
individual membership, may well be more secure as a whole than
the corresponding class under any other system—precisely because
it continually draws into itself most of the natural ability from the
other classes. Throughout English history, we are told, the salvation
of the aristocracy has been its comparative openness, the fact that
ability could percolate into it, instead of rising up behind it like water
behind a dam, as was the case in pre-revolutionary France. And the
same principle is working even more effectually in our own economic
order. A great weakness of the trades-union movement, as of all
attempts at self-assertion on the part of the less privileged classes, is
that it is constantly losing able leaders. As soon as a man shows that
marked capacity which would fit him to do something for his fellows,
it is ten to one that he accepts a remunerative position, and so
passes into the upper class. It is increasingly the practice—perhaps
in some degree the deliberate policy—of organized wealth to win
over in this way the more promising leaders from the side of labor;
and this is one respect in which a greater class-consciousness and
loyalty on the part of the latter would add to its strength.
Thus it is possible to have freedom to rise and yet have at the
same time a miserable and perhaps degraded lower class—
degraded because the social system is administered with little regard
to its just needs. This is more the case with our own industrial
system, and with modern society in general, than our self-
satisfaction commonly perceives. Our one-sided ideal of freedom,
excellent so far as it goes, has somewhat blinded us to the
encroachments of slavery on an unguarded flank. I mean such
things as bad housing, insecurity, excessive and deadening work,
child labor and the lack of any education suited to the industrial
masses—the last likely to be remedied now that it is seen to threaten
industrial prosperity.
It is hard to say how much of the timidity noticeable in the
discussion of questions of this sort by the comfortable classes is due
to a vague dread of anarchy and spoliation by an organized and self-
conscious lower class; but probably a good deal. If power, under
democracy, goes with numbers, and the many are poor, it would
seem at first glance that they would despoil the few.
To conservative thinkers a hundred, or even fifty, years ago this
seemed almost an axiom, but a less superficial philosophy has
combined with experience to show that anarchy, in Mr. Bryce’s
words, “is of all dangers or bugbears the one which the modern
world has least cause to fear.”[123]
The most apparent reason for this is the one already discussed,
namely, that power does not go with mere numbers, under a
democracy more than under any other form of government; a
democratic aristocracy, that is, one whose members maintain their
position in an open struggle, being without doubt the strongest that
can exist. We shall never have a revolution until we have caste;
which, as I have tried to show, is but a remote possibility. And as an
ally of established power we have to reckon with the inertia of social
structure, something so massive and profound that the loudest
agitation is no more than a breeze ruffling the surface of deep
waters. Dominated by the habits which it has generated, we all of us,
even the agitators, uphold the existing order without knowing it.
There may, of course, be sudden changes due to the fall of what has
long been rotten, but I see little cause to suppose that the timbers of
our system are in this condition: they are rough and unlovely, but far
from weak.
Another conservative condition is that economic solidarity which
makes the welfare of all classes hang together, so that any general
disturbance causes suffering to all, and more to the weak than to the
strong. A sudden change, however reasonable its direction, must in
this way discredit its authors and bring about reaction. The hand-
working classes may get much less of the economic product than
they ought to; but they are not so badly off that they cannot be
worse, and, unless they lose their heads, will always unite with other
classes to preserve that state of order which is the guaranty of what
they have. Anarchy would benefit no one, unless criminals, and
anything resembling a general strike I take to be a childish expedient
not likely to be countenanced by the more sober and hard-headed
leaders of the labor movement. All solid betterment of the workers
must be based on and get its nourishment from the existing system
of production, which must only gradually be changed, however
defective it may be. The success of strikes, and of all similar tactics,
depends, in the nature of things, on their being partial, and drawing
support from the undisturbed remainder of the process. It is the
same principle of mingling stability with improvement which governs
progress everywhere.
And, finally, effective organization on the part of the less privileged
classes goes along with intelligence, with training in orderly methods
of self-assertion, and with education in the necessity of patience and
compromise. The more real power they get, the more conservatively,
as a rule, they use it. Where free speech exists there will always be
a noisy party advocating precipitate change (and a timid party who
are afraid of them), but the more the people are trained in real
democracy the less will be the influence of this element.
Whatever divisions there may be in our society, it is quite enough
an organic whole to unite in casting out tendencies that are clearly
anarchic. And it is also evident that such tendencies are to be looked
for at least as much among the rich as among the poor. If we have at
one extreme anarchists who would like to despoil other people, we
have, at the other, monopolists and financiers who actually do so.
It is a common opinion that the sway of riches over the human
mind is greater in our time than previously, and greater in America
than elsewhere. How far is this really the case?
To understand this matter we must not forget that the ardor of the
chase—as in a fox hunt—may have little to do with the value of the
quarry. The former, certainly, was never so great in the pursuit of
wealth as here and now; chiefly because the commercial trend of the
times, due to a variety of causes, supplies unequalled opportunities
and incitements to engage in the money-game. In this, therefore, the
competitive zeal of an energetic people finds its main expression.
But to say that wealth stands for more in the inner thought of men,
that to have or not to have it makes a greater intrinsic difference, is
another and a questionable proposition, which I am inclined to think
opposite to the truth. Such spiritual value as personal wealth has
comes from its power over the means of spiritual development. It is,
therefore, diminished by everything which tends to make those
means common property: and the new order has this tendency.
When money was the only way to education, to choice of
occupation, to books, leisure and variety of intercourse, it was
essential to the intellectual life; there was no belonging to the
cultured class without it. But with free schools and libraries, the
diffusion of magazines and newspapers, cheap travel, less
stupefying labor and shorter hours, culture opportunity is more and
more extended, and the best goods of life are opened, if not to all,
yet to an ever-growing proportion. Men of the humblest occupations
can and do become gentlemen and scholars. Indeed, people are
coming more and more to think that exclusive advantages are
uncongenial to real culture, since the deepest insight into humanity
can belong only to those who share and reflect upon the common
life.
The effect is that wealth is shorn of much of that prestige of
knowledge, breeding and opportunity which always meant more than
its material power. The intellectual and spiritual centre of gravity, like
the political, sinks down into the masses of the people. Though our
rich are rich beyond the dreams of avarice, they mean less to the
inner life of the time, exercise less spiritual authority, perhaps, than
the corresponding class in any older society. They are the objects of
popular curiosity, resentment, admiration or envy, rather than the
moral deference given to a real aristocracy. They are not taken too
seriously. Indeed, there could be no better proof that the rich are no
overwhelming power with us than the amount of good-natured
ridicule expended upon them. Were they really a dominant order, the
ridicule, if ventured at all, would not be good-natured. Their
ascendency is great when compared with a theory of equality—and
in this sense the remarks in the last chapter should be understood—
but small compared with that of the ruling classes of the Old World.
Over a class of frenzied gold-seekers, rich or poor, chiefly in the
towns, the money-idea is no doubt ascendant; but if you approach
the ordinary farmer, mechanic or sober tradesman you are likely to
find that he sets no high rate on wealth beyond what is necessary for
the frugal support of a family, and that he neither admires nor envies
the rich, but looks at the millionaire and thinks: “After all, it isn’t life.
What does he get out of it more than the rest of us?” The typical
American is an idealist, and the people he looks up to are those who
stand in some way for the ideal life—or whom he supposes to do so
—most commonly statesmen, but often writers, scientists or
teachers. Education and culture, as Mr. Bryce and others have
noticed, is cherished by plain people all over the land, often to a
degree that puts to shame its professed representatives.
We find, then, that agitators who strive to incite the people against
the rich encounter with disgust an idealism which refuses to believe
that their advantages are extravagantly great; and one of the main
grievances of such men is what they look upon as the folly or lack of
spirit of the poor in this regard.
Never before, probably, was there so large a class of people who,
having riches, feel that they are a doubtful blessing, especially in
relation to the nurture of children. Many a successful man is at his
wits’ end to give his children those advantages of enforced industry,
frugality and self-control which he himself enjoyed. One of the richest
men of the day holds that accumulations are generally bad for the
children, as well as for society, and favors almost unlimited
graduated taxation of inheritances.[124] According to the philosophy
which he supports by practice as well as theory, the man who finds
himself rich is to live modestly and use his surplus as a trust fund for
the benefit of the public.
What would a man wish for his own son, if he could choose? First,
no doubt, some high and engrossing purpose, which should fill his
life with the sense of worthy striving and aspiration. After this he
would wish for health, friends, peace of mind, the enjoyment of
books, a happy family life and material comfort. But the last, beyond
that degree which even unskilled labor should bring, he would regard
as of secondary importance. Not a straitened house and table but a
straitened soul is the real evil, and the two are more separable now
than formerly. The more a real democracy prevails, the less is the
spiritual ascendency of riches.
There is, for instance, no such settled and institutional deference
to wealth in the United States as there seems to be in England; the
reason being, in part, that where there are inherited classes there
are also class standards of living, costly in the upper class, to which
those who would live in good company are under pressure to
conform. In England there is actually a ruling order, however ill
defined, which is generally looked up to and membership in which is
apparently the ambition of a large majority of all aspiring men who do
not belong to it by birth. Its habits and standards are such that only
the comparatively rich can be at home in it. There is nothing
corresponding to this with us. We have richer men and the pursuit of
riches is an even livelier game, but there is no such ascendency in
wealth, no such feeling that one must be rich to be respectable. With
us, if people have money they enjoy it; if not, they manage with what
they have, neither regarding themselves nor regarded by others as
essentially inferior.
It is also a general feeling here that wealth should not be a
controlling factor in marriage, and it is not common for American
parents to object seriously to a proposed son-in-law (much less a
daughter-in-law) on the mere ground of lack of means, apart from his
capacity to earn a living. The matter-of-fact mercenariness in this
regard which, as we are led to believe by the novelists, prevails in
the upper circles of England, is as yet somewhat shocking to the
American mind.
Hereditary titles, sometimes imagined to be a counterpoise to the
ascendency of wealth, are really, in our time at least, a support and
sanction to it, giving it an official standing and permanence it cannot
have in democracy. We understand that in England wealth—with
tact, patience and maybe political services—will procure a title,
which, unlike anything one can get for money in America, is
indestructible by vice and folly, and can be used over and over to
buy wealth in marriage. “Nothing works better in America than the
promptness with which the degenerate scions of honored parents
drop out of sight.”[125] Rank is not an offset but a reward and bribe to
wealth; perhaps the only merit that can be claimed for it in this
connection being that the desire and deference for it imposes a
certain discipline on the arrogance of newly acquired riches.
The English idea that those in high offices should have a
magnificent style of living, “becoming to their station,” is also one that
goes with caste feeling. It makes it hardly decent for the poor to hold
such offices, and is almost absent here, where, if riches are
important to political success, the condition is one of which the
people do not approve and would gladly dispense with.
I doubt whether the whole conception which imputes merit to
wealth and seeks at least the appearance of the latter in modes of
dress, attendance and the like, is not stronger everywhere in Europe
than in the United States.
FOOTNOTES:
[123] The American Commonwealth, Chapter 94.
[124] Andrew Carnegie.
[125] T. W. Higginson, Book and Heart, 145.
CHAPTER XXV
THE ORGANIZATION OF THE ILL-PAID CLASSES
The Need of Class Organization—Uses and Dangers of

Unions—General Disposition of the Hand-Working
Classes.
It is not the purpose of this book to add anything to the merely
controversial literature of the time; and in treating the present topic I
intend no more than to state a few simple and perhaps obvious
principles designed to connect it with our general line of thought.
It is quite apparent that an organized and intelligent class-
consciousness in the hand-working people is one of the primary
needs of a democratic society. In so far as this part of the people is
lacking in a knowledge of its situation and in the practice of orderly
self-assertion, a real freedom will also be lacking, and we shall have
some kind of subjection in its place; freedom being impossible
without group organization. That industrial classes exist—in the
sense already explained[126]—cannot well be denied, and existing
they ought to be conscious and self-directing.
The most obvious need of class-consciousness is for self-
assertion against the pressure of other classes, and this is both most
necessary and most difficult with those who lack wealth and the
command over organized forces which it implies. In a free society,
especially, the Lord helps those who help themselves; and those
who are weak in money must be strong in union, and must also exert
themselves to make good any deficiency in leadership that comes
from ability deserting to more favored classes.
That the dominant power of wealth has an oppressive action, for
the most part involuntary, upon the people below, will hardly be
denied by any competent student. The industrial progress of our time
is accompanied by sufferings that are involved with the progress.
These sufferings—at least in their more tangible forms—fall almost
wholly upon the poorer classes, while the richer get a larger share of
the increased product which the progress brings. By sufferings I
mean not only the physical hardship and liability to disease, early
decay, and mutilation or death by accident, which fall to the hand-
worker; but also the debasement of children by premature and
stunting labor, the comparative lack of intellectual and social
opportunities, the ugly and discouraging surroundings, and the
insecurity of employment, to which he and his are subject. There is
no purpose to inflict these things; but they are inflicted, and the only
remedy is a public consciousness, especially in the classes who
suffer from them, of their causes and the means by which they can
be done away with.
The principal expressions of class-consciousness in the hand-
working classes in our day are labor unions and that wider, vaguer,
more philosophical or religious movement, too various for definition,
which is known as socialism. Regarding the latter I will only say at
present that it includes much of what is most vital in the
contemporary working of the democratic spirit; the large problems
with which its doctrines deal I prefer to discuss in my own way.
Labor unions are a simpler matter. They have arisen out of the
urgent need of self-defence, not so much against deliberate
aggression as against brutal confusion and neglect. The industrial
population has been tossed about on the swirl of economic change
like so much sawdust on a river, sometimes prosperous, sometimes
miserable, never secure, and living largely under degrading,
inhuman conditions. Against this state of things the higher class of
artisans—as measured by skill, wages and general intelligence—
have made a partly successful struggle through coöperation in
associations, which, however, include much less than half of those
who might be expected to take advantage of them.[127] That they are
an effective means of class self-assertion is evident from the
antagonism they have aroused.
Besides their primary function of group-bargaining, which has
come to be generally recognized as essential, unions are performing
a variety of services hardly less important to their members, and
serviceable to society at large. In the way of influencing legislation
they have probably done more than all other agencies together to
combat child-labor, excessive hours, and other inhuman and
degrading kinds of work; also to provide for safeguards against
accident, for proper sanitation of factories, and the like. In this field
their work is as much defensive as aggressive, since employing
interests, on the other side, are constantly influencing legislation and
administration to their own advantage.
Their function as spheres of fellowship and self-development is
equally vital and less understood. To have a we-feeling, to live
shoulder to shoulder with one’s fellows, is the only human life; we all
need it to keep us from selfishness, sensuality and despair, and the
hand-worker needs it even more than the rest of us. Usually without
pecuniary resource and insecure of his job and his home, he is, in
isolation, miserably weak and in a way to be cowed and unmanned
by misfortune or mere apprehension. Drifting about in a confused
society, unimportant, apparently, to the rest of the world, it is no
wonder if he feels
“I am no link of Thy great chain,”[128]
and loses faith in himself, in life and in God. The union makes him
feel that he is part of a whole, one of a fellowship, that there are
those who will stand by him in trouble, that he counts for something
in the great life. He gets from it that thrill of broader sentiment, the
same in kind that men get in fighting for their country; his self is
enlarged and enriched and his imagination fed with objects,
comparatively, “immense and eternal.”
Moreover, the life of labor unions and other class associations,
through the training which it gives in democratic organization and
discipline, is perhaps the chief guaranty of the healthy political
development of the hand-working class—especially those imported
from non-democratic civilizations—and the surest barrier against
recklessness and disorder. That their members get this training will
be evident to anyone who studies their working, and it is not
apparent that they would get it in any other way. Men learn most in
acting for purposes which they understand and are interested in, and
this is more certain to be the case with economic aims than with any
other.
Thus, if unions should never raise wages or shorten hours, they
would yet be invaluable to the manhood of their members. At worst,
they ensure the joy of an open fight and of companionship in defeat.
Self-assertion through voluntary organization is of the essence of
democracy, and if any part of the people proves incapable of it it is a
bad sign for the country. On this ground alone it would seem that
patriots should desire to see organization of this sort extend
throughout the industrial population.
The danger of these associations is that which besets human
nature everywhere—the selfish use of power. It is feared with reason
that if they have too much their own way they will monopolize
opportunity by restricting apprenticeship and limiting the number of
their members; that they will seek their ends through intimidation and
violence; that they will be made the instruments of corrupt leaders.
These and similar wrongs have from time to time been brought home
to them, and, unless their members are superior to the common run
of men, they are such as must be expected. But it would be a
mistake to regard these or any other kinds of injustice as a part of
the essential policy of unions. They are feeling their way in a human,
fallible manner, and their eventual policy will be determined by what,
in the way of class advancement, they find by experience to be
practicable. In so far as they attempt things that are unjust we may
expect them, in the long run, to fail, through the resistance of others
and through the awakening of their own consciences. It is the part of
other people to check their excesses and cherish their benefits.
In general no sort of persons mean better than hand-laboring men.
They are simple, honest people, as a rule, with that bent toward
integrity which is fostered by working in wood and iron and often lost
in the subtleties of business. Moreover, their experience is such as to
develop a sense of the brotherhood of man and a desire to realize it
in institutions. Not having enjoyed the artificial support of
accumulated property, they have the more reason to know the
dependence of each on his fellows. Nor have they any great hopes

ebook download Spatiotemporal Random Fields, Second Edition: Theory and Applications George Christakos - eBook PDF all chapter

Uploaded by

Copyright:

Available Formats

You might also like

ebook download Spatiotemporal Random Fields, Second Edition: Theory and Applications George Christakos - eBook PDF all chapter

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ebook download Spatiotemporal Random Fields, Second Edition: Theory and Applications George Christakos - eBook PDF all chapter

Uploaded by

Copyright:

Available Formats

Spatiotemporal Random Fields, Second

Edition: Theory and Applications

(eBook PDF) Turbomachinery: Basic Theory and

Kappa Distributions. Theory and Applications in Plasmas

Solutions for Biot's Poroelastic Theory in Key

Geophysical potential fields: geological and

Schaum's Outline of Probability, Random Variables, and

(eBook PDF) Psycholinguistics: Introduction and

(eBook PDF) Psycholinguistics Introduction and

(eBook PDF) Probability, Statistics, and Random Signals

Copyright © 2017 Elsevier Inc. All rights reserved.

For Information on all Elsevier publications visit our

Publisher: Candice Janco

CHAPTER I Space, Time, SpaceeTime, Randomness, and Probability .................. 1

7.3 Third Classification: Probability Law Shapes .................................................... 74

6. Correlation Theory of Complex Spatiotemporal Random Fields ............................146

5. The Central Limit Theorem......................................................................................233

6. Scales of Spatiotemporal Dependence and the Uncertainty Principle ....................329

2.2 The Spherical Symmetry Technique................................................................. 412

4. Discrete Linear Representations of Spatiotemporal Random Fields .......................490

3.2 Normalized Angular Spectrum Decomposition.............................................. 554

SPACE, TIME, SPACEeTIME,

Spatiotemporal Random Fields. http://dx.doi.org/10.1016/B978-0-12-803012-7.00001-5

Table 1.1 Commonly Used Symbols

Kronecker/Levi-Chivita εijkεipq ¼ djpdkq djqdkp

Table 1.2 Special Functions and Polynomials

Gamma function. G(n)

2. SPACEeTIME CONTINUUM AND KOLMOGOROV PROBABILITY SPACE

2.1 SPACEeTIME ARGUMENTS: POINTS, LAGS, SEPARATIONS, AND METRICS

Dp ¼ ðDs1 ; .; Dsn ; Ds0 Þ ˛ Rnþ1 ; (2.7a)

Dp ¼ ðh; sÞ ˛ Rn;1 ; (2.7b)

Table 2.1 Standard Vector Operations on SpaceeTime Points and Lags

Table 2.1 Standard Vector Operations on SpaceeTime Points and Lagsdcont’d

Table 2.2 Number of Monomials for n [ 2, 3, and Different Combinations

“Lässt jeden ganz das bleiben was er ist;

The Influence of Ambitious Young Men—Security of the

The Need of Class Organization—Uses and Dangers of

“I am no link of Thy great chain,”[128]

You might also like