Confidence Intervals for Steady-State Simulations II: A Survey of Sequential Procedures

We continue our survey of methods for constructing confidence intervals for steady-state means via simulation by studying sequential procedures which determine the length of the simulation during the course of the run. Our goal is to provide the simulation practitioner with some guidance as to which published procedures might actually perform well in practice. Empirical results for a variety of stochastic models with known steady-state means suggest that sequential procedures by Fishman and by Law and Carson provide good performance relative to the criterion probability of coverage.