A COLLECTION of recipes could be called a cookbook, but Daan Frenkel and Berend Smit's book achieves more than that. By explaining the physics behind the algorithms, the authors let you learn how molecular dynamics and Monte Carlo simulation methods work, how to apply these methods in a sensible way and what information can be extracted from them. Although computer simulation, lying between analytical theory and experiment, is now regarded as the third branch of science, it is still viewed with scepticism by some researchers precisely because of this interdisciplinary character.
However, this book brilliantly lays down the scientific foundations of the simulational approach, mostly in the framework of equilibrium classical statistical mechanics. After the introductory chapters in part 1, which explain the basic aspects of Monte Carlo and molecular dynamic simulations of simple fluid-like systems, the book emphasizes the possibility of carrying out these simulations in various statistical ensembles, and discusses how to study phase equilibria. These chapters, which are in parts 2 and 3, are the core of the book, and here the emphasis of subjects differs from classical texts such as Mike Allen and Dominic Tildesley's well established Computer Simulation of liquids (1987 Clarendon,Oxford).
Frenkel and Smit give a concise, comparative discussion of the various ways of calculating those quantities, such as free energy, entropy and chemical potential, that are not direct observables of simulations' Particular emphasis is devoted to the "Gibbs ensemble", a recently developed technique that can be used to study the co-existence of phases in the absence of interfaces between the phases. Applications of this new ensemble were pioneered by the authors, who are therefore able to give an authoritative presentation.
I believe these first three parts of the book should be obligatory reading for any beginner to molecular simulation. They will also particularly suit those researchers who use commercial "molecular modelling" software, by allowing them to understand more fully what they can and cannot do with such packages. But the book also contains valuable material for the experienced specialist, in particular the advanced techniques covered in part 4. This is a compilation of miscellaneous topics, such as how to deal with rigid constraints, how to sample rare events and so on. Again, the authors place particular emphasis on their own research strengths, namely polymers and other complex fluids of that nature. In combination with the techniques of part 3, such as the Gibbs ensemble and chemical potential sampling, this gives a unique and very impressive review.
The usefulness of this book is enhanced by three special features. First, it contains a concise sketch of more than 40 algorithms (with comments) that indicate how to begin the process of translating the various methods explained into computer code. Second, there are nine short appendices that deal with various technical aspects (such as Verlet tables, cell lists and Ewald sums) and background information on statistical mechanics (such as linear response theory). Finally, there are several hundred references to the original literature, including titles, thereby giving a much more useful guide than other reviews in this area.
In summary, this book is strongly biased towards the scientific interests of the authors, and this bias is both a strength and a weakness. The material that is covered is extremely well digested and is presented in a concise fashion with fine humeur and critical judgement whenever appropriate. However, there has been much recent progress in simulation that one does not find anywhere in this book. There is, for example, no mention of finite size effects at phase transitions, nor of path integral Monte Carlo simulations of solids at low temperatures, applications of Monte Carlo methods to simulate growth phenomena far from equilibrium, or the Car-parrinello approach, which "marries" molecular dynamics and electronic structure calculations.
Although these subjects are clearly beyond the scope of this book, I believe that there should at least have been a brief look at these related problems. They do have some connections to the physics of simple and complex fluids that are, after all, the main topics of the book. Also, many of the examples shown refer only to simulations of 108 or 256 Lennard-Jones atoms, and the authors do not provide enough warning that for some characteristics - particularly long wavelength properties - this would simply not be enough atoms for a proper simulation.
Despite these concerns, this book fills a gap in the literature on computer simulation. It will substantially contributie to the recognition of the fact that molecular simulation is serious science that promises a great deal.