Review by Choice Review
The original edition of these works, Numerical Recipes: The Art of Scientific Computing, by William H. Press et al. (CH, Sep'86) contained procedures in FORTRAN and PASCAL translated from FORTRAN. It was warmly received, and Press and colleagues went on to produce versions in PASCAL, BASIC, and C, also well received (e.g., Numerical Recipes in Pascal, by William H. Press et al., CH, May'90). FORTRAN and C seem to represent a steady state, at least for the next few years; leaving aside the object-oriented languages, C has become the language of choice among professional programmers, displacing PASCAL and cutting out MODULA-2. FORTRAN lives, despite the fact that it is disdained in computer science circles. Its adherents remain committed despite its quirks (and possibly because it contains the GO TO command). As far as this reviewer knows, the "structured" FORTRAN revision that appeared imminent (in 1987) still has not been issued, and one can only conjecture that there is resistance by unreconstructed FORTRAN-ers, who like their language the way it is. There is no denying that there is an immense amount of FORTRAN software available. In the review of the first edition, this reviewer stated that "users will want to solve linear systems and eigensystems, interpolate/extrapolate, evaluate integrals and functions, find roots, determine extrema, analyze data by Fourier and/or statistical methods, solve differential equations, etc. Since users might very well want to understand the procedures that they employ, and perhaps even modify them, the authors work very hard at clarity, both in the programs they provide and in exposition. Nor do they hesitate to make choices and recommendations." That remains true, and there is also much new material, including a chapter on integral equations and inverse methods, more on numerical linear algebra (band matrices, sparse systems, more on the Cholesky or QR methods), multigrid methods in solving elliptic partial differential equations, more on functional approximation (and applications to numerical differentiation and integration), fast transform methods, etc. It is still true that the methods offered reflect (mostly) informed current practice, that the selection of references for further reading is excellent, that the index is at least adequate, and that the books are well-produced physically. To quote again: "This reviewer knows of no other single source of so much material of this nature. Highly recommended." Advanced undergraduate through professional. R. J. Wernick; emeritus, State University of New York at Oswego
Copyright American Library Association, used with permission.
Review by Choice Review
