Performance of a Wavelet-Based Spectral Procedure for Steady-State Simulation Analysis

A summary and an analysis are given for an experimental performance evaluation of WASSP, an automated wavelet–based spectral method for constructing an approximate confidence interval on the steady–state mean of a simulation output process such that the delivered confidence interval satisfies user–specified requirements on absolute or relative precision as well as coverage probability. The experimentation involved three difficult test problems, each with an output process exhibiting some combination of the following characteristics: a long warm–up period, a persistent autocorrelation structure, or a highly nonnormal marginal distribution. These problems were used to compare the performance of WASSP with that of the Heidelberger–Welch algorithm and ASAP3, two sequential procedures based respectively on the methods of spectral analysis and nonoverlapping batch means. Concerning efficiency (required sample sizes) and robustness against the statistical anomalies commonly encountered in simulation studies, WASSP outperformed the Heidelberger–Welch procedure and compared favorably with ASAP3.