New Developments in Statistical Computing

text screens require substantial background, whereas others are very intuitive, requiring hardly any background. Users move from one type of text screen to the other without warning. I would hesitate to require the use of GASP outside the classroom. One thing I like to have students do when studying stochastic processes is to start asking “what if” about the processes. With GASP, users can interact with the displays by changing various probabilities or rates to investigate the effects of such changes on the process being studied. The quantities reported in GASP are primarily things like average displacement, empirical arrival rate, or average time to be serviced. The key to understanding stochastic processes lies in the understanding of the variability in the process. For users to understand how these quantities vary under the same initial conditions, the simulation needs to be repeated several times with users recording the displayed statistics. To investigate the influence of the model parameters on the process, the simulations would need to be repeated with a new set of initial conditions. The simulations are slow because of the animation, and this approach to studying the behavior of stochastic processes is tedious. I believe that students would benefit more by creating and studying their own simulations. Of course, these will most likely not be animated! In terms of using GASP for in-class demonstrations, a word of caution about a few of the displays is in order. Depending on the choice of probabilities or rates, some of the simulations are very slow to settle down. Therefore, it is possible that the punch line may not occur within a class period. This problem may not only be due to the simulation being slowed down for the animation; it may also be a result of the behavior of the particular process being studied.