| In industrial engineering and management science, simulation experiments are usually performed to compare two or more alternative scenarios or systems. We present procedures for selecting the best or near-best of a finite number of simulated systems when “best” is defined by maximum or minimum expected performance. The procedures are appropriate when it is possible to repeatedly obtain small, incremental samples from each simulated system. The motivation for using such a sequential procedure is to eliminate, at an early stage of experimentation, those simulated systems that are clearly inferior, and thereby reduce the overall computational effort required to find the best. First we present procedures that accommodate unequal variances across systems and the use of common random numbers, but assume normally distributed data and no dependence within a system's output. We then extend these procedures to steady-state simulation where a system's outputs may be neither independent nor normally distributed. We prove the asymptotic validity of the procedures, and provide empirical comparisons with some existing indifference-zone selection procedures. |
