Power comparisons for the multivariate batch-means method

Abstract

The multivariate batch-means (MBM) method of analyzing simulation output has the potential to be a widely-used tool by practitioners. This paper presents a power comparison of selected statistics that can be used for choosing the batches, which is one of the key elements in the successful use of the MBM method. Explicit procedures are given for five batch-selection criteria, and some guidelines for the proper use of the MBM method are stated.

A Comprehensive Review of Methods for Simulation Output Analysis

“…of the above inequality is equal to 0. If the overall confidence level must be at least 1−α, then the α i can be chosen so that m i=1 α i = α. Multivariate estimation methods are described in Charnes (1989Charnes ( , 1990Charnes ( , 1991 and Chen and Seila (1987).…”

A review of advanced methods for simulation output analysis

“…If the overall confidence level must be at least l O O ( 1 -a ) percent, then the a,'s can be chosen so that xi"=, at = C Y . The existing multivariate estimation methods include Charnes (1989Charnes ( , 1990Charnes ( , 1991 and Clien and Seila (1987).…”

Advanced methods for simulation output analysis

“…If the overall confidence level must be at least lOO(1 -C Y ) percent, then the ai's can be chosen so that CY^ = a. The existing multivariate estimation methods include Charnes (1989Charnes ( , 1990Charnes ( , 1991 and Chen and Seila (1987).…”