You are on page 1of 2

BOOK REVIEWS 293

the subject by authors such as Zeeman, Poston and Stewart, there are however several
new items of interest. The best of these is the thirty-page Chapter Four on The Analysis of
Downloaded 12/03/14 to 130.113.86.233. Redistribution subject to SIAM license or copyright; see http://www.siam.org/journals/ojsa.php

Discontinuous Processes by J. Q. Smith, P. J. Harrison, and E. C. Zeeman, concentrating


on problems in Bayesian decision theory. The authors formulate a theory to explain how it
is that decision-makers whose information and utilities evolve in time can experience
sudden qualitative changes in the minima of expected loss functions (Bayes decisions).
Even in uncomplicated situations which appear to be very stable, sudden changes can
occur which can be understood using a cusp catastrophe model. Examples include
behavior of elected officials with two constituencies and drivers making adjustments
under the influence of alcohol.
The remainder of the book consists of short articles illustrating applications of
catastrophe theory to phase transformations, competition and games, earth sciences, and
biomechanical, biochemical, and biophysical problems. An interesting article on the
nature of language has little to do with the subject of the book. The bibliography, by
Poston and Stewart, contains over 400 items dating up to 1977.
In summary, what appears to have been a stimulating seminar has not produced a
book with the same effectiveness.
THOMAS F. BANCHOFF
Brown University

Solitons and Nonlinear Wave Equations. By R. K. DODD, J. C. EILBECK, J. D. GIBBON


AND H. C. MORRIS. Academic Press, London, 1982. x + 630 pp. ISBN
0-12-2191200X.
Singly authored scientific books often project particular points of view which limit
their perspectives, while conference proceedings, even with contributors of the highest
level, necessarily lack the cohesion that makes a real book. An interesting compromise
between these two extreme cases is Solitons and Nonlinear Wave Equations by Roger
Dodd, Chris Eilbeck, John Gibbon and Hedley Morris. Each of these four authors is a
substantial contributor to the growing body of soliton research, but their areas of interest
are sufficiently diverse that the book covers an unusually wide range of subject matter in a
unified manner. The results is a rich and informative book that, as the authors have
intended, will be of great value to the reader "who has no prior knowledge of solitons and
nonlinear wave equations" and will be equally useful to those working in the field.
It is particularly pleasing that the first figure is a direct reproduction of a solitary
wave in John Scott Russell’s wave tank, the second is a plot of some solutions for the
Korteweg-de Vries equation and the third is a reproduction from the numerical
experiments on nonlinear resonance by Fermi, Pasta and Ulam. Such a juxtaposition of
components in soliton research is an example of the way that the authors have organized
and coordinated their material throughout the book to assist the reader with no prior
knowledge. The second chapter is specifically designed to introduce students from applied
mathematics and from hydrodynamics to the scattering problems of physics while the
first chapter shows how the Korteweg-de Vries equation arises from a variety of fluid
problems. The third and fourth chapters discuss isospectral theory and the inverse
scattering method in considerable detail while chapter six through nine present a wide
spectrum of results on the Zakharov and Shabat inverse scattering problem, sine-Gordon
kinks, wave resonances and the nonlinear Schrtidinger equation, and unstable systems. A
final chapter on numerical studies has the effect of leaving a reader at the edge of some
exciting current research.
Numerically generated figures are used liberally to great advantage and problems
294 BOOK REVIEWS

are provided at the ends of most chapters. Although all the authors are in departments of
mathematics, they display current knowledge of activities in soliton physics; thus recent
Downloaded 12/03/14 to 130.113.86.233. Redistribution subject to SIAM license or copyright; see http://www.siam.org/journals/ojsa.php

developments in solid state physics are well treated. The book could serve well as a text for
the introductory course in "solitonics" which this reviewer would like to see become
standard fare in graduate studies of applied science.
ALWYN SCOTT
Technical University of Denmark

Statistical Software: A Comparative Review. By IVOR FRANCIS. North-Holland, New


York, 1981. xx + 542 pp. $70.00.
This book compiles the results of a survey of developers and users of statistical
software concerning the capabilities, portability, ease of use, and reliability of 117
statistical programs and packages. The survey was conducted during 1980 and the book
was published in late 1981. Each entry contains a short description of the product
containing a statement of purpose, availability, cost, hardware requirements, and
proposed enhancements. The meat is contained in responses to a detailed questionnaire
concerning areas of statistical coverage, data management, editing capabilities, and
intended audience. Each package developer rated his own product. Developers also
nominated users of their system, who were queried on the same items to produce "user
ratings" to which the "developer ratings" were compared. In no case were the user ratings
based on returns from more than three individual users.
Since the book itself is a review (and not an evaluation) of the statistical software
literature, this article is a second-order or meta-review. As such, it is less a review of the
book and its contents than it is a discussion of the whole enterprise of software evaluation,
not only in the context of statistical applications, but also more broadly. The impetus for
Francis’s work came from the activities of a committee of the American Statistical
Association in the mid-seventies. Statisticians, then, may be the first professional society
to go about software evaluation in a systematic way, largely divorced from commercial
interests. The efforts expended have been large, and they deserve some reflection and
comment.
Efforts at software review are plagued by three difficulties: the rapid obsolescence of
today’s software, the elusiveness of objective criteria and reviewers, and the expense of
comprehensive evaluation. First, software is always in a state of flux. Unlike many
consumer products, in which changes from one year to the next are largely cosmetic,
statistical software changes both rapidly and radically. The economics of the ,industry
dictate this, since the main source of income for major producers is the annual licensing
and update fee, not the new-installation charge. Software developers are sensitive to the
shortcomings of their products, and often criticisms have been remedied between the time
developers were shown a draft of a review and the time that the review appeared in print.
For an example, see Berk and Francis (J. Amer. Statist. Assoc., (1978) pp. 65-98).
Publication time is so slow when compared to the half-life of major statistical software
that any attempt to review and evaluate software using the traditional printed media must
be greeted with skepticism. This book presents a cross-sectional snapshot of the state of
statistical software circa 1980. Does it have any relevance today? Who should own it?
Who should consult it? A corollary question: was it worth the immense effort required to
produce a compilation that was arguably destined to become obsolescent so quickly?
Perhaps the most challenging difficulty is that of objectivity in the review process. To
integrate the information from scores of packages, running on tens of computers, requires
skill, expertise, knowledge, persistence, dedication, and no small measure of wisdom. In

You might also like