This book is offered as a comprehensive and up-to-date guide to the various techniques for statisticians, operations researchers, and others who use stochastic simulation methods in engineering, in business, and in various branches of science. It offers explicit recommendations for the use of techniques and algorithms. After a short introduction, there are chapter-length treatments of pseudorandom numbers, random variables, stochastic models, variance reduction, analysis of simulation output, and uses of simulation. Many algorithms are described in detail sufficient for programming. FORTRAN 77 programs for eight techniques are included. Each chapter concludes with a number of exercises.
According to the author’s Preface, “The aim of this volume is to display [the simulator’s] tools in their most useful form, with guidance about their use.” The volume succeeds well in this endeavor. It makes no attempt to address other, more substantive issues in designing and using a simulation, such as model validity or the handling of time as a variable. To do so would have required a volume at least twice the size.
The book combines a lucid discussion of each technique with formal statements of algorithms and examples. The bibliographic citations are quite appropriately used to guide the reader to other sources dealing specifically with particular techniques. A complete bibliography on stochastic simulation was outside the author’s purview; the approach is expository rather than encyclopedic. The inclusion of exercises makes the book appropriate for classroom use, but only in conjunction with other materials that deal with the logic of the use of the tools that are so clearly described here.
Although the author’s Preface says that the general mathematical level of the book is “. . . elementary, involving no more than a first course in probability and statistics. . . ,” the inclusion of theorems at various points in the text, the frequent use of integrals, and the occasional use of matrix notation belie this claim.
One of the book’s nice features is the inclusion of FORTRAN 77 routines for such tasks as generating normal and gamma variates. Since these are intended for reading by humans (if for no other reason than to translate them into other languages such as Pascal or C), it would have been helpful if higher programming standards had been used. The programs would be far more useful to the reader if adequate commenting had been done, if variable names had been chosen more often for their information-carrying content than for their brevity, if output had always been kept separate from computation, if structured programming concepts involving loops, instead of backward GOTOs, had been used consistently, etc. It’s about time statisticians learned what is now taught in most introductory programming courses]
Common sense pervades the recommendations. For example, in the discussion of programming for the generation of random variables, the author says, “. . . if you have a program available to you which will do what you want, use it. It is very unlikely that any lack of speed on its part will compensate for your time in programming and testing a preferred method.”
The book is nicely printed and well bound, and the Index is accurate. There appears to be no more than the usual number of typos. Overall, the book is a valuable resource for those seriously interested in stochastic simulation modeling, but it is not for the beginner.
