Multivariate inference in stationary simulation using batch means

ABSTRACT

Methods are presented for computing a joint confidence region and simultaneous confidence intervals for the mean of multivariate observations from a simulation operating in steady-state. These methods place observations into batches and use the batch means to compute the confidence region or confidence intervals. As sequential method to determine the batch size that assures independence among the multivariate batch means is presented and empirical studies are used to demonstrate the performance and properties of these methods.

REFERENCES

• 1: Fisb3uan, G. S. (1978). Groupir~ Observations tn Digital Experiments. l~na~ement Science 24, 510-521.
• 2: George S. Fishman, Principles of Discrete Event Simulation, John Wiley & Sons, Inc., New York, NY, 1978
• 3: Hannan. E.J. (1970).Multiple Time Series. John Wiley, New York.
• 4: Law, A. M. (1977). Confidence Intervals in Discrete Event Simulation: A Comparison o1 Replication and Batch Means. 1Naval ResearchLoF_jstics ~arterly 2.~, 667-678.
• 5: Morrison, D. F. (1976). Multivariate St a.t.istical Methods. McCraw-Hi ii, New York.

INDEX TERMS

Primary Classification:
  G. Mathematics of Computing
  G.3 PROBABILITY AND STATISTICS
      Subjects: Multivariate statistics

Additional Classification:
  I. Computing Methodologies
  I.6 SIMULATION AND MODELING

General Terms:
Design, Performance, Theory