Table Of ContentSTOCHASTIC
SIMULATION
OPTIMIZATION
An Optimal Computing Budget Allocation
SERIES ON SYSTEM ENGINEERING AND OPERATIONS
RESEARCH
Editor-in-Chief: Yu-Chi Ho (Harvard University, USA)
Associate Editors: Chun-Hung Chen (George Mason University, USA)
Loo Hay Lee (National University of Singapore,
Singapore)
Qianchuan Zhao (Tsinghua University, China)
About the Series
The series provides a medium for publication of new developments and advances in
high level research and education in the fields of systems engineering, industrial
engineering, and operations research. It publishes books in various engineering areas
in these fields. The focus will be on new development in these emerging areas with
utilization of the state-of-the-art technologies in addressing the critical issues.
The topics of the series include, but not limited to, simulation optimization,
simulation process and analysis, agent-based simulation, evolutionary computation/
soft computing, supply chain management, risk analysis, service sciences, bioinformatics/
biotechnology, auctions/competitive bidding, data mining/machine learning, and robust
system design.
Vol. 1 Stochastic Simulation Optimization: An Optimal Computing Budget Allocation
Authors: C.-H. Chen and L. H. Lee
Gregory - Stochastic Simulation Optimization.pm2d 8/2/2010, 6:20 PM
System Engineering and Operations Research – Vol. 1
STOCHAST IC
SIMULAT ION
OPT IMIZAT ION
An Optimal Computing Budget Allocation
Chun-Hung Chen
George Mason Univ., USA
National Taiwan Univ.
Loo Hay Lee
National Univ. of Singapore
World Scientific
NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TAIPEI • CHENNAI
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library.
System Engineering and Operations Research — Vol. 1
STOCHASTIC SIMULATION OPTIMIZATION
An Optimal Computing Budget Allocation
Copyright © 2011 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN-13 978-981-4282-64-2
ISBN-10 981-4282-64-2
Typeset by Stallion Press
Email: enquiries@stallionpress.com
Printed in Singapore.
Gregory - Stochastic Simulation Optimization.pm1d 8/2/2010, 6:20 PM
May11,2010 6:45 SPI-B930 9inx6in b930-fm
Foreword to the WSP Series
on System Engineering
and Operation Research
Advancement in science and technology often blurs traditional dis-
ciplinary boundary. Control system theory and practice, operations
research, and computational intelligence combine to contribute to
moderncivilization inmyriadways.From trafficcontrol onland,sea,
and air, to manufacturing automation, to social and communication
networks, these knowledge-based and human-made systems rely on
research results in the above disciplinary topics for their smooth and
efficient functioning.
TheWorldScientificPublishingSeriesonSystemEngineeringand
Operations Research is launched to fill this niche for students and
scholars doing research in these areas. The first book in this series is
by two leading scientists in the area of efficent simulation and mod-
eling of complex systems. They articulate clearly the computational
burden involved in such study and device innovative methodology to
overcome the difficulties. I welcome their contribution in inaugurat-
ing this series and look forward to additional books in this genre.
Yu-Chi Ho
Editor-in-Chief
WSP Series on System Engineering
and Operations Research
January 20, 2010
v
May11,2010 6:45 SPI-B930 9inx6in b930-fm
TThhiiss ppaaggee iinntteennttiioonnaallllyy lleefftt bbllaannkk
May11,2010 6:45 SPI-B930 9inx6in b930-fm
Preface
“Simulation” and “optimization” are two very powerful tools in sys-
tems engineering and operations research. With the advance of new
computing technology, simulation-based optimization is growing in
popularity. However, computational efficiency is still a big concern
because (i) in the optimization process, many alternative designs
must be simulated; (ii) to obtain a sound statistical estimate, a large
number of simulation runs (replications) is required for each design
alternative. A usermay beforced to compromise on simulation accu-
racy, modeling accuracy, and the optimality of the selected design.
There have been several approaches developed to address such an
efficiency issue. This book is intended to offer a different but ambi-
tious approach by trying to answer the question “what is an optimal
(or the most efficient) way to conduct all the simulations in order
to find a good or optimal solution (design)?” The ultimate goal is
to minimize the total simulation budget while achieving a desired
optimality level, or to maximize the probability of finding the best
design using a fixed computing budget. The primary idea is called
Optimal Computing Budget Allocation (OCBA).
This book aims at providing academic researchers and industrial
practitioners a comprehensive coverage of the OCBA approach for
stochastic simulation optimization. Chapter 1 introduces stochas-
vii
May11,2010 6:45 SPI-B930 9inx6in b930-fm
viii SSO: An Optimal Computing Budget Allocation
tic simulation optimization and the associated issue of simulation
efficiency. Chapter 2 gives an intuitive explanation of computing
budget allocation and discusses its impact on optimization perfor-
mance. Then a series of OCBA approaches developed for various
problems are presented, from selecting the best design (Chapter 3),
selecting a set of good enough designs (Chapter 5), to optimization
with multiple objectives (Chapter 6). Chapter 4 provides numeri-
cal illustrations, showing that the computation time can be reduced
significantly. Chapter 4 also offers guidelines for practical implemen-
tation of OCBA. Chapter 7 extends OCBA to large-scale simulation
optimization problems. The OCBA technique is generic enough that
it can be integrated with many optimization search algorithms to
enhance simulation optimization efficiency. Several potential search
techniques are explored. Finally, Chapter 8 gives a generalized view
of the OCBA framework, and shows several examples of how the
notion of OCBA can be extended to problems beyond simulation
and/or optimization such as data envelopment analysis, experiments
of design, and rare-event simulation. To help those readers without
much simulation background, in the appendix, we offer a short but
comprehensive presentation of stochastic simulation basics. We also
include a basic version of OCBA source code in the appendix.
OCBA has several strengths: it is effective, easy to understand,
simpletoimplement,andcanbeeasilygeneralized orintegratedwith
other methods to extend its applicability. We believe that this book
is highly useful for different purposes.
1. For researchers, this book offers a series of promising approaches
for efficiency enhancement in computer simulation, stochastic
optimization, statistical sampling, and ranking and selection.
The generalized framework may lead to numerous new lines of
researches.
2. For courses, this book could serve as a textbook for advanced
stochastic simulation or simulation optimization courses. They
can cover Appendix A, Chapters 1 and 2 for introductions; Chap-
ters 3 through 4, and parts of Chapters 5 through 8 for advanced
materials.
May11,2010 6:45 SPI-B930 9inx6in b930-fm
Preface ix
3. For practioners, this book offers a simple but effective approach
to enhance the computational efficiency of simulation optimiza-
tion. Simulation practioners from industries, governments, and
the military should find this book useful and relatively easy to
read and apply because it gives numerous intuitive illustrations,
well-structured algorithms, and practical implementation guide-
lines.
