DIGITAL COMPUTER SIMULATION: THE ALLOCATION OF COMPUTER TIME IN COMPARING SIMULATION EXPERIMENTS,

摘要

The report describes an improved step-by-step procedure for minimizing the computer time needed to obtain a specified statistical precision in the comparison of simulation experiments. The simulations are considered as covariance stationary stochastic processes. Well-known time-saving methods for reducing variance by inducing positive correlations between the experiments and high negative correlations between replications are included in the procedure. Results show that for a given level of accuracy,significantly less computer time is required when sample sizes are determined by the method suggested in this study than when they are equal. Also,small differences in the autocorrelation functions are important when each process is highly correlated.