On Selecting the Best of K Simulated Systems: An Expository Survey.
Abstract
This research has focused on simulation analysis using two new techniques, frequency domain experiments and standardized time series. The research investigates the problem of selecting the 'best' one of k arbitrary systems or alternatives. Consider one observation from each of the k systems. By 'best,' we mean that system which has the highest probability of yielding the 'most desirable' of the k observations. The term 'most desirable' is defined according to some criterion of goodness determined by the experimenter. We show that this problem can be formulated as a multinomial selection problem. Hence, multinomial selection procedures are, in a sense, nonparametric procedures. An up-to-date survey of such multinomial procedures is given. Further, we describe how some of these procedures can be adapted for use in the simulation environment.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 15, 1985
- Accession Number
- ADA161903
Entities
People
- David Goldsman
- Lee Schruben
Organizations
- Cornell University College of Engineering