ABSTRACT
We study the problem of determining that one of k stationary simulated processes which has the largest mean. We adapt for use in the simulation environment a ranking and selection procedure due to Dudewicz and Dalal (1975). In order to implement this procedure, it is necessary to estimate the process variance of each of the k simulated systems; variance estimators arising from the theory of standardized time series are used for this purpose.
- Bechhofer, R.E. (1954). "'A Single-Sample Mu I t i p I e Dec i s i on Procedure for Rank ~ ng Means of Normal Populations wi th Known Vari anoes", Annats Math. Star. , 16--39.Google Scholar
- 13udew{cz, E.J. and S.R. Dalai (1975). "Allocation of Observations in Ranking and Selection wi th Unequal Variances". 3ankhya--8, 28--78.Google Scholar
- Gibbons, J.D. , I. Olkin, and M. Sobel (1977). Selecting and Ordering Populations. Ni }ey, New York. Google ScholarDigital Library
- Go}dsman, D. (1983)."Ranking and Selection in Simulation". Prec. Hinter 5i~ulo Conf. , 386 -393. Google ScholarDigital Library
- Goldsman, O. (1904). On Us/n9 .gtandardlzed Time 3er/es to Analyze Stochastic Processes. Ph.D. D~ ssertation, School of O.R. & I.E. , Cornel 1 Univers~ ty, Ithaca, NY.Google Scholar
- Go}dsman: D. and A. Nozari (1985). "Rank i ng and Se I eot i on Techn i que s i n Computer Simulatlon, I. The Normal Means Approach". Teoh. Report #J--85--ii, School of ISyE, Georgia Tech, Atlanta, GA.Google Scholar
- Gupta, S.S. and S. Panchapakesan (1979). Hut tipte Oeclsion Procedures. 6111 ey, New York.Google Scholar
- Iglehart, D.L. (1977). "Simulating Stab}e Stochastic Systems, VII: Selecting the Best System". TIH3 Studies Management Science, 37--50.Google Scholar
- Kang, K. and D. Goldsman (1985). "The Correlation between Mean and Variance E~timator~". Prec. Aiinter Sfmut. Conf. Google ScholarDigital Library
- Keen i g, L.W. and A.M. Law (1982) . "A Procedure for Selecting a Subset of Size m Containing the | ~est of k Normal Populations, wi th Appl ications to Simulation". Tech. Report #82--9, Dept. of Management Information Systems, Univ. o~ Arizona, Tucson, AZ.Google Scholar
- Law, A.M. and W.D. Kel ton (1982). Simulation Mode/in9 ~nd Analysis. McGraw- H~ I I, New York. Google ScholarDigital Library
- Meketon, M.S. and I~.W. Schme~ser (1S184). "'Overlapplng Batch Means: Something for Nothing?" Prec. Hinter Sfmut, Conf. , 227--230. Google ScholarDigital Library
- Nozar~ , A. and J.S. Morr~ s (t984). "Application of an Optimlzation Procedure to Steady -State 51mulation". Prec. Hinter 3{mul. Conf. , 216--219. Google ScholarDigital Library
- Schruben, L. (1983) . "'Conf~ dence Interval Estimation Using Standardized Time 5er~es". Operations Research, 31, 1090--1108.Google ScholarDigital Library
- 5ul l ivan, O.W. and J.R. Wi I son (1984). "'Restricted Subset Selection for Normal Populations w i th Unknown and Unequal Variances". Proc. HFnter S/rout. ConF. , 266--274. Google ScholarDigital Library
- Ranking and selection procedures using standardized time series
Recommendations
Properties of Standardized Time Series Weighted Area Variance Estimators
We wish to estimate the variance of the sample mean from a continuous-time stationary stochastic process. This article expands on the results of a technical note Goldsman and Schruben 1990 by using the theory of standardized time series to investigate ...
Standardized time series methods: variance estimation using replicated batch means
WSC '01: Proceedings of the 33nd conference on Winter simulationWe present a new method for obtaining confidence intervals in steady-state simulation. In our replicated batch means method, we do a small number of independent replications to estimate the steady-state mean of the underlying stochastic process. In ...
Standardized Time Series LP-Norm Variance Estimators for Simulations
This paper studies a class of estimators for the variance parameter of a stationary stochastic process. The estimators are based on Lp norms of standardized time series, and they generalize previously studied estimators due to Schruben. We show that the ...
Comments