ABSTRACT
Suppose that we want to compare k different systems, where /spl mu//sub i/ denotes the steady state mean performance of system i. Our goal is to use simulation to pick the "best" system (i.e., the one with the largest or smallest steady state mean). To do this, we present some two stage procedures based on the method of batch means. Our procedures also construct multiple comparisons with the best (MCB) confidence intervals for /spl mu//sub i/-max/sub j/spl ne/i//spl mu//sub j/, i=1,...,k. Under the assumption of an indifference zone of (absolute or relative) width /spl delta/, we can show that asymptotically (as /spl delta//spl rarr/0 with the size of the batches proportional to 1//spl delta//sup 2/), the joint probability of correctly selecting the best system and of the MCB confidence intervals simultaneously containing /spl mu//sub i/-max/sub j/spl ne/i//spl mu//sub j/, i=1,...,k, is at least 1-/spl alpha/, where /spl alpha/ is prespecified by the user.
- Bechhofer, R. E. 1954. A single-sample multiple decision procedure for ranking means of normal populations with known variances. Annals of Mathematical Statistics 25:16-39.Google ScholarCross Ref
- Billingsley, P. 1968. Convergence o} Probability Measures. New York: John Wiley.Google Scholar
- Damerdji, H., P. W. Glynn, M. K. Nakayama, and J. R. Wilson. 1995. Selecting the Best System in Steady-State Simulations. Forthcoming technical report.Google Scholar
- Ethier, S. N. and T. C. Kurtz. 1986. Markov Processes: Characterization and Convergence. New York: John Wiley.Google ScholarCross Ref
- Glynn, P. W. 1990. Diffusion approximations. Chapter 4 of Handbooks in Operations Research and Management Science, Vol. 2, Stochastic Models, ed. D. Heyman and M. Sobel. Elsevier Science Publishers B. V. (North-Holland).Google Scholar
- Glynn, P. W. and W. Whitt. 1987. Sufficient conditions for functional-limit-theorem versions of L = )~W. Queueing Systems: Theory and Applications 1:279-287. Google ScholarDigital Library
- Goldsman, D. and B. L. Nelson. 1994. Ranking, selection and multiple comparisons in computer simulation. In Proceedings o} the 199~ Winter Simulation Conference, ed. J. D. Tew, S. Manivannan, D. A. Sadowski, and A. F. Seila, 192- 199. Piscataway, New Jersey: Institute of Electrical and Electronics Engineers. Google ScholarDigital Library
- Hsu, J. C. 1984. Constrained simultaneous confidence intervals for multiple comparisons with the best. Annals of Statistics 12:1136-1144.Google ScholarCross Ref
- Matejcik, F. J. and B. L. Nelson. 1992. Two-stage multiple comparisons with the best for computer simulation. Operations Research, forthcoming.Google Scholar
- Nakayama, M. K. 1994. Two-stage stopping procedures based on standardized time series. Management Science 40:1189-1206. Google ScholarDigital Library
- Nakayama, M. K. 1995. Procedures for simultaneous selection and multiple comparisons in steady-state simulations. Unpublished manuscript.Google Scholar
- Newman, C. M. and A. L. Wright. 1981. An invariance priciple for certain dependent sequences. Annals of Probability 9:671-675.Google ScholarCross Ref
- Rinott, Y. 1978. On two-stage selection procedures and related probability-inequalities. Communications in Statistics--Theory and Methods A7:799- 811.Google ScholarCross Ref
- Whitt, W. 1989a. Planning queueing simulations. Management Science 35:1341-1366. Google ScholarDigital Library
- Whitt, W. 1989b. Simulation run length planning. In Proceedings o} the 1989 Winter Simulation Conference, ed. E. A. MacNair, K. J. Musselman, and P. Heidelberger, 106-112. Piscataway, New Jersey: Institute of Electrical and Electronics Engineers. Google ScholarDigital Library
- Wilcox, R. R. 1984. A table for Rinott's selection procedure. Journal o} Quality Technology 16:97- 100.Google ScholarCross Ref
Index Terms
- Selecting the best system in steady-state simulations using batch means
Recommendations
Two-stage procedures for multiple comparisons with a control in steady-state simulations
WSC '96: Proceedings of the 28th conference on Winter simulationSuppose that we have k different stochastic systems, where /spl mu/i denotes the steady-state mean of system i. We assume that the system labeled k is a control and want to compare the performance of the other sys tems, labeled 1,2,...,k - 1, relative ...
On the Average-Case Complexity of Selecting the kth Best
Let $\bar V_k (n)$ be the minimum average number of pairwise comparisons needed to find the kth largest of n numbers $(k \geqq 2)$, assuming that all $n!$ orderings are equally likely. D. W. Matula proved that, for some absolute constant c, $\bar V_k (n) -...
Comments