International Series in Operations Research & Management Science Volume 133 For other titles published in this series, go to www.springer.com/series/6161 George Samuel Fishman Christos Alexopoulos · David Goldsman · James R. Wilson Editors Advancing the Frontiers of Simulation A Festschrift in Honor of George Samuel Fishman With Contributions by Russell Cheng, Soumyadip Ghosh, Shane G. Henderson, Peter W. Glynn, Eunji Lim, James O. Henriksen, Wolfgang Hormann,¨ Josef Leydold, Jack P. C. Kleijnen, Pierre L’Ecuyer, Franc¸ois Panneton, Andrew F. Seila, Sally Brailsford, Steffen Straßburger, Thomas Schulze, Richard Fujimoto, Shing Chih Tsai, Barry L. Nelson, Jeremy Staum, Tubaˆ Aktaran-Kalaycı 123 Editors Christos Alexopoulos David Goldsman Georgia Institute of Technology Georgia Institute of Technology H. Milton Stewart School of Industrial H. Milton Stewart School of Industrial and Systems Engineering and Systems Engineering 765 Ferst Drive NW 765 Ferst Drive NW Atlanta GA 30332-0205 Atlanta GA 30332-0205 USA USA [email protected] [email protected] James R. Wilson Edward P. Fitts Department of Industrial and Systems Engineering North Carolina State University 111 Lampe Drive Raleigh NC 27695-7906 USA [email protected] ISSN 0884-8289 ISBN 978-1-4419-0816-2 e-ISBN 978-1-4419-0817-9 DOI 10.1007/b110059 Springer Dordrecht Heidelberg London New York Library of Congress Control Number: 2009932145 c Springer Science+Business Media, LLC 2009 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com) Preface This Festschrift honors George Samuel Fishman, one of the founders of the field of computer simulation and a leader of the disciplines of operations research and the management sciences for the past five decades, on the occasion of his seventieth birthday. The papers in this volume span the theory, methodology, and application of computer simulation. The lead article is appropriately titled “George Fishman’s Professional Career.” In this article we discuss George’s contributions to operations research and the man- agement sciences, with special emphasis on his role in the advancement of the field of simulation since the 1960s. We also include a brief personal biography together with comments by several individuals about the extraordinary effect that George has had on all his students, colleagues, and friends. The second article, titled “A Conversation with George Fishman,” is the transcript of an extended interview with George that we conducted in October 2007. In the article titled “Computer Intensive Statistical Model Building,” Russell Cheng studies resampling methods for building parsimonious multiple linear regres- sion models so as to represent accurately the behavior of the dependent variable in terms of the smallest possible subset of explanatory (independent) variables. The author shows how bootstrap resampling can be used not only for rapid identification of good models but also for efficient comparison of competing models. The next article is titled “Patchwork Distributions.” In this article, Soumyadip Ghosh and Shane Henderson consider a class of multivariate probability distribu- tions that can be used to model a finite-dimensional random vector when the user has specified all the marginal distributions of the random vector, the covariance matrix of the random vector, and the probabilities that the random vector lies in certain regions. Peter Glynn and Eunji Lim examine the foundations of the batch-means method for steady-state simulation analysis in their article titled “Asymptotic Validity of Batch Means Steady-State Confidence Intervals.” Although the large-sample valid- ity of many implementations of the batch-means method requires that the target out- put process must satisfy a functional central limit theorem, in this article the authors establish the validity of the batch-means method for Harris-recurrent Markov pro- cesses that satisfy a weaker (nonfunctional) central limit theorem. v vi Preface In the article titled “Efficient Modeling of Delays in Discrete-Event Simula- tion,” Jim Henriksen considers the problem of efficiently implementing four types of delays that are commonly encountered in simulation modeling, with a detailed discussion of the solution approach provided by the delay-modeling algorithms that have been developed for the SLX simulation language. In the article titled “Sampling from Linear Multivariate Densities,” Wolfgang Hormann¨ and Josef Leydold develop an efficient acceptance-rejection algorithm for generating a vector of dependent random variables whose joint density is linear with a domain that is bounded and symmetric about a point; and ultimately the authors extend their algorithm to generate random vectors from concave differentiable den- sities over point-symmetric domains. Jack Kleijnen’s article, “Factor Screening in Simulation Experiments: Review of Sequential Bifurcation,” provides an overview of factor-screening methods for simulation experiments with special emphasis on sequential bifurcation, a method that is particularly effective in certain types of large-scale simulation studies. “F2-Linear Random Number Generators,” by Pierre L’Ecuyer and Franc¸ois Panneton, is the ninth article in this volume. This article reviews various construc- tion methods for random-number generators based on linear recurrence modulo 2, examines their theoretical properties, describes the relevant computational tools and algorithms, and ends with comparisons based on various qualitative criteria. The F2- linear class contains many long-period random-number generators—including the Tausworthe, generalized feedback shift register (GSFR), twisted GSFR, Mersenne twister, WELL, and xorshift generators. The tenth article is titled “Opportunities and Challenges in Health Care Simula- tion.” In this article Andrew F. Seila and Sally Brailsford review successful applica- tions of simulation in health care, examine the suitability of simulation models and methods for the analysis of health-care systems, and explore the reasons for the lack of adoption of simulation as a “routine” tool for health-care systems analysis. The authors end with an insightful list of ideas aimed at promoting wider adoption of simulation in health care. In the eleventh article, “Future Trends in Distributed Simulation and Distributed Virtual Environments,” Steffen Straßburger, Thomas Schulze, and Richard Fujimoto report the main results of a peer study of current trends in distributed simulation and distributed virtual environments. The survey assesses the current state of this methodology, its relevance to simulation practice, and the research challenges that must be addressed so as to facilitate the widespread use of this methodology in industry and government as well as in research organizations. In “Combined Screening and Selection of the Best with Control Variates,” Shing Chih Tsai, Barry L. Nelson, and Jeremy Staum formulate ranking-and-selection pro- cedures with screening that exploit point estimators based on the method of control variates to gain greater statistical efficiency. Compared with previous ranking-and- selection procedures that incorporate screening and selection of the best alternative, substantial improvements in performance are achieved by the new procedures. In the final article, “Optimal Linear Combinations of Overlapping Variance Esti- mators for Steady-State Simulation,” Tubaˆ Aktaran-Kalaycı, Christos Alexopoulos, David Goldsman, and James R. Wilson seek to estimate the variance parameter of a simulation output process using optimal linear combinations of variance estimators Preface vii based on the methods of overlapping batch means and standardized time series. From the latter estimators they derive asymptotically valid confidence intervals for both the mean and variance parameter of the target process. We thank the authors who have contributed to this volume, especially for their forbearance throughout the lengthy process of completing this book. Special thanks go to Yvonne Smith (Georgia Tech), who produced the transcription of our interview with George, and to Carolyn Ford at Springer, who so ably supervised the production of this book. We are also indebted to Fred Hillier (Stanford University) and Gary Folven (Springer) for patiently guiding us through the entire process. Atlanta, GA, USA Christos Alexopoulos Atlanta, GA, USA David Goldsman Raleigh, NC, USA James R. Wilson March 2009 Contents George Fishman’s Professional Career ................................ 1 Christos Alexopoulos, David Goldsman, and James R. Wilson A Conversation with George Fishman ................................ 21 Christos Alexopoulos, David Goldsman, and James R. Wilson Computer Intensive Statistical Model Building ........................ 43 Russell Cheng Patchwork Distributions