WSC 2006 Abstracts

Bayesian Ideas and Discrete Event Simulation: Why, What and How
Stephen E. Chick (INSEAD)

Abstract:
Bayesian methods are useful in the simulation context for several reasons. They provide a convenient and useful way to represent uncertainty about alternatives (like manufacturing system designs, service operations, or other simulation applications) in a way that quantifies uncertainty about the performance of systems, or about inputs parameters of those systems. They also can be used to improve the efficiency of discrete optimization with simulation and response surface methods. Bayesian methods work well with other decision theoretic tools, and can therefore provide a link from traditional operations-level experiments to higher-level managerial decision-making needs, in addition to improving the efficiency of computer experiments.

White Noise Assumptions Revisited: Regression Metamodels and Experimental Designs in Practice
Jack P.C. Kleijnen (Tilburg University)

Abstract:
Classic linear regression metamodels and their concomitant experimental designs assume a univariate (not multivariate) response and white noise. By definition, white noise is normally (Gaussian), independently (implying no common random numbers), and identically (constant variance) distributed with zero mean (valid metamodel). This advanced tutorial tries to answer the following questions: (i) How realistic are these classic assumptions in simulation practice? (ii) How can these assumptions be tested? (iii) If assumptions are violated, can the simulation's I/O data be transformed such that the analysis becomes correct? (iv) If such transformations cannot be applied, which alternative statistical methods (for example, generalized least squares, bootstrapping, jackknifing) can then be applied?

Inside Discrete-Event Simulation Software: How it Works and Why it Matters
Thomas J. Schriber (Ross School of Business (Wyly 5733)) and Daniel T. Brunner (Kiva Systems, Inc.)

Abstract:
This paper provides simulation practitioners and consumers with a grounding in how discrete-event simulation software works. Topics include discrete-event systems; entities, resources, control elements and operations; simulation runs; entity states; entity lists; and entity-list management. The implementation of these generic ideas in AutoMod, SLX, and Extend is described. The paper concludes with several examples of “why it matters” for modelers to know how their simulation software works, including coverage of SIMAN (Arena), ProModel, and GPSS/H as well as the other three tools.

Black-Box Algorithms for Sampling from Continuous Distributions
Josef Leydold (University of Economics and B.A. Vienna) and Wolfgang Hörmann (University for Economics and B.A. Vienna)

Abstract:
For generating non-uniform random variates, black-box algorithms are powerful tools that allow drawing samples from large classes of distributions. We give an overview of the design principles of such methods and show that they have advantages compared to specialized algorithms even for standard distributions, e.g., the marginal generation times are fast and depend mainly on the chosen method and not on the distribution. Moreover these methods are suitable for specialized tasks like sampling from truncated distributions and variance reduction techniques. We also present a library called UNU.RAN that provides an interface to a portable implementation of such methods.

Splitting for Rare-Event Simulation
Pierre L'Ecuyer and Valérie Demers (DIRO, Université de Montréal) and Bruno Tuffin (IRISA/INRIA)

Abstract:
Splitting and importance sampling are the two primary techniques to make important rare events happen more frequently in a simulation, and obtain an unbiased estimator with much smaller variance than the standard Monte Carlo estimator. Importance sampling has been discussed and studied in several articles presented at the Winter Simulation Conference in the past. A smaller number of WSC articles have examined splitting. In this paper, we review the splitting technique and discuss some of its strengths and limitations from the practical viewpoint. We also introduce improvements in the implementation of the multilevel splitting technique. This is done in a setting where we want to estimate the probability of reaching B before reaching (or returning to) A when starting from a fixed state not in B, where A and B are two disjoint subsets of the state space and B is very rarely attained. This problem has several practical applications.

Model Composability
Hessam S. Sarjoughian (Arizona State University)

Abstract:
Composition of models is considered essential in developing heterogeneous complex systems and in particular simulation models capable of expressing a system's structure and behavior. This paper describes model composability concepts and approaches in terms of modeling formalisms. These composability approaches along with some of the key capabilities and challenges they pose are presented in the context of semiconductor supply chain manufacturing systems.

Gradient-Based Simulation Optimization
Sujin Kim (Purdue Univeristy)

Abstract:
We present a review of methods for simulation optimization. In particular, we focus on gradient-based techniques for continuous optimization. We demonstrate the concepts and techniques using the multidimensional newsvendor problem. We also discuss mathematical techniques and results that are useful in verifying and analyzing the simulation optimization procedures.

A Comprehensive Review of Methods for Simulation Output Analysis
Christos Alexopoulos (Georgia Institutre of Technology)

Abstract:
This paper reviews statistical methods for analyzing output data from computer simulations. Specifically, it focuses on the estimation of steady-state system parameters. The estimation techniques include the replication/deletion approach, the regenerative method, the batch means method, and methods based on standardized time series.

Ranking and Selection Procedures for Simulation
Kirk C. Benson (Center for Army Analysis) and David Goldsman and Amy R. Pritchett (Georgia Institute of Technology)

Abstract:
We present sequential ranking and selection statistical procedures that determine the best simulated model configuration among competing alternatives. The best in this context denotes the largest expected value of a given performance metric. In order to run the procedures efficiently, we give algorithms using batched observations, which under certain conditions, exhibit the characteristics necessary for the appropriate application of ranking and selection procedures. We present empirical results that indicate that the sequential procedures are quite parsimonious, in terms of the number of required observations.