Can you toss a coin (fair or unfair) physically maintaining the same probability of “heads p” (0<p<1) in every toss? We can certainly achieve this easily through simulation by drawing uniform variates in the range [0, 1] and then declaring “heads” each time the variate value falls in the range [0, p] whose probability is p. As another example, think of the feeling of weightlessness that astronauts experience going to space. Unless they are accustomed to it, it can well disturb their mental balance if sustained for a long period. The situation can be simulated on earth by taking the astronauts to some high-storied building and bringing them down at top speed through a lift; this should be done repeatedly to give them the desired outcome at no cost at all! The strength of simulation, or imitation (usually artificial) of reality, lies here.
However, selecting the best algorithm to simulate can be an arduous task, complicated by model characteristics and its implementation, the runtime scenario, and synergistic interactions. Roland Ewald takes up the challenge and places before the reader a platform for the aforesaid selection, namely, James II (http://jamesii.org). Its selection mechanism covers three situations: (i) where no previous knowledge is available, (ii) where the dependence of the performance of the simulator on the aspects of the problem is not known explicitly, and (iii) where some relationship between the problem aspects and the simulating algorithm’s performance can be established, but only empirically. The author sums up the book with some interesting experiments on the techniques.
The book has three parts. Part 1, “Background,” covers the selection of a simulation algorithm and its performance analysis. Part 2 covers methods and implementation, and Part 3 provides examples and a conclusion. There are 11 chapters, an appendix, and a helpful bibliography. The work emerged out of the author’s doctoral thesis, and provides good reference material for researchers and post-graduate students of computing and allied fields working in simulation studies.