Statistical Analysis of Discrete-Event Simulations.

摘要

The paper surveys past and ongoing work by the authors in statistical analysis of simulations. An earlier developed technique for analyzing simulations of GI/G/S queues and Markov chains is shown to apply to discrete-event simulations which can be modeled as regenerative processes. It is possible to address questions of simulation run duration and of starting and stopping simulations because of the existence of a random grouping of observations which produces independent identically distributed blocks in the course of the simulation. This grouping allows one to obtain confidence intervals for a general function of the steady-state distribution of the process being simulated. The technique is illustrated with a simulation of a retail inventory distribution system. (Modified author abstract)