Book Review of Landau & Paez by W. Press 

Reviewed by William H. Press, Harvard University, Cambridge, Massachusetts, 
  Physics Today, p 71, July 1998.
 
Computational Physics: Problem Solving with Computers Rubin H. Landau
and Manuel J. Paez Mejia Wiley, New York, 1997. 520 pp. $49.95 hc
ISBN 0, Includes diskette

Come, let us parse together and consider that deceptively simple noun
phrase ``computational physics.'' Adjective specializing the noun, it
ought to mean something like ``a subfield of physics based on, or
heavily utilizing, computation.'' Indeed, at the graduate and
professional levels, this is exactly what it does mean. It is a lively
field of science, whose practitioners are typically the high priests
of large, sophisticated numerical codes for use in hydrodynamics,
plasma dynamics, gravitational N body, lattice quantum chromodynamics
and so forth. Professional computational physicists might engage in a
nitty gritty discussion on some fine point of parallel hardware memory
caching or a heated debate on high-performance FORTRAN versus message
passing interface. In undergraduate teaching, however, computational
physics has come to mean something different Physicists are by nature
problem solvers, and we have never been shy about ``ours''), this may
be it.

Rubin Landau and Manuel Paez Mejia's Computational Physics is one of a
spate of recently published texts for this kind of useful course.
Others include Nicholas Giordano's Computational Physics (Prentice
Hall, 1997) and Tao Pang's An Introduction to Computational Physics
(Cambridge U. P. 1997). (To find a dozen more titles going back to
Steven Koonin and Dawn Meredith's Computational Physics (Addison
Wesley, 1990), search the Website http://www.amazon.com for the joint
subject fields ``physics'' AND ``data processing.'') The present book,
developed for a course at Oregon State University is distinguished by
the breadth of material that it contains. This is not, however, always
a good thing, since the coherence of the book as a systematic text
suffers. Think of this as a good source book for instructors rather
than as a text for students. The book is divided into a large number
(something like 300) of very short sections. The authors categorize
these as theory, method, exploration, problem (meaning a statement of
a topic), assessment (meaning an exercise for the student),
implementation (meaning a computer program supplied on the book's
accompanying diskette) and so on, but this classification scheme
breaks down.

There are a lot of good things in the inchoate mass, however:
discussion of computer programming languages and methods, error
analysis, numerical methods (including differential equations and
Fourier methods), data fitting, Monte Carlo methods, summary and
assessment of public domain software, a bit on parallel computing and
on and on. Among the sampled topics of physics are quantum
eigenproblems, bound states and scattering; anharmonic oscillations,
nonlinear dynamics and chaos; the Ising model; fractals and pattern
formation; and solitons.

Students in a course like this can learn a lot of physics. One might
worry, however, that it is a kind of disorganized and opportunistic
physics-the ``best bits''. That's not a sin in the context of a
comprehensive undergraduate physics curriculum precisely because a
course like this is so useful as a general problem solving course and
is thus potentially so useful to students majoring in other areas of
science, the individual instructor has to make a tough choice: Do you
want to teach physics, or do you want to use physics to teach broad
based computer and analytic problem solving skills? This book can be 
useful either way.