Credibility of the Final Results from Quantitative Stochastic Simulation
G. Ewing, D. McNickle and K. Pawlikowski
Department of Computer Science
University of Canterbury
Abstract
The credibility of estimated confidence intervals for mean values produced by quantitative stochastic simulation is considered. Basic rules of proper experimental studies of quality of such interval estimators is formulated and applied in evaluation of two methods of confidence-interval construction: Batch Means and Spectral Analysis, in the case of simulations run on single and multiple processors.