Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An...

28
MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications

Transcript of Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An...

Page 1: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

MODELING UNCERTAINTYAn Examination of Stochastic Theory,Methods, and Applications

Page 2: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Vanderbei, R. / LINEAR PROGRAMMING: Foundations and ExtensionsJaiswal, N.K. / MILITARY OPERATIONS RESEARCH: Quantitative Decision MakingGal, T. & Greenberg, H. / ADVANCES IN SENSITIVITY ANALYSIS AND

PARAMETRIC PROGRAMMINGPrabhu, N.U. / FOUNDATIONS OF QUEUEING THEORY

Fang, S.-C., Rajasekera, J.R. & Tsao, H.-S.J. / ENTROPY OPTIMIZATIONAND MATHEMATICAL PROGRAMMING

Yu, G. / OPERATIONS RESEARCH IN THE AIRLINE INDUSTRYHo, T.-H. & Tang, C. S. / PRODUCT VARIETY MANAGEMENTEl-Taha, M. & Stidham , S. / SAMPLE-PATH ANALYSIS OF QUEUEING SYSTEMSMiettinen, K. M. / NONLINEAR MULTIOBJECTIVE OPTIMIZATIONChao, H. & Huntington, H. G. / DESIGNING COMPETITIVE ELECTRICITY MARKETSWeglarz, J. / PROJECT SCHEDULING: Recent Models, Algorithms & ApplicationsSahin, I. & Polatoglu, H. / QUALITY, WARRANTY AND PREVENTIVE MAINTENANCETavares, L. V. / ADVANCED MODELS FOR PROJECT MANAGEMENTTayur, S., Ganeshan, R. & Magazine, M. / QUANTITATIVE MODELING FOR SUPPLY

CHAIN MANAGEMENTWeyant, J./ ENERGY AND ENVIRONMENTAL POLICY MODELINGShanthikumar, J.G. & Sumita, U./APPLIED PROBABILITY AND STOCHASTIC PROCESSESLiu, B. & Esogbue, A.O. / DECISION CRITERIA AND OPTIMAL INVENTORY PROCESSESGal, T., Stewart, T.J., Hanne, T./ MULTICRITERIA DECISION MAKING: Advances in MCDM

Models, Algorithms, Theory, and ApplicationsFox, B. L./ STRATEGIES FOR QUASI-MONTE CARLOHall, R.W. / HANDBOOK OF TRANSPORTATION SCIENCEGrassman, W.K./ COMPUTATIONAL PROBABILITYPomerol, J-C. & Barba-Romero, S./MULTICRITERION DECISION IN MANAGEMENTAxsäter, S./ INVENTORY CONTROLWolkowicz, H., Saigal, R., Vandenberghe, L./ HANDBOOK OF SEMI-DEFINITE

PROGRAMMING: Theory, Algorithms, and ApplicationsHobbs, B. F. & Meier, P. / ENERGY DECISIONS AND THE ENVIRONMENT: A Guide

to the Use of Multicriteria MethodsDar-El, E./ HUMAN LEARNING: From Learning Curves to Learning OrganizationsArmstrong, J. S./ PRINCIPLES OF FORECASTING: A Handbook for Researchers and

PractitionersBalsamo, S., Personé, V., Onvural, R./ ANALYSIS OF QUEUEING NETWORKS WITH BLOCKINGBouyssou, D. et al/ EVALUATION AND DECISION MODELS: A Critical PerspectiveHanne, T./ INTELLIGENT STRATEGIES FOR META MULTIPLE CRITERIA DECISION MAKINGSaaty, T. & Vargas, L./ MODELS, METHODS, CONCEPTS & APPLICATIONS OF THE ANALYTIC

HIERARCHY PROCESSChatterjee, K. & Samuelson, W./ GAME THEORY AND BUSINESS APPLICATIONSHobbs, B. et al/ THE NEXT GENERATION OF ELECTRIC POWER UNIT COMMITMENT MODELSVanderbei, R.J./ LINEAR PROGRAMMING: Foundations and Extensions, 2nd Ed.Kimms, A./ MATHEMATICAL PROGRAMMING AND FINANCIAL OBJECTIVES FOR

SCHEDULING PROJECTSBaptiste, P., Le Pape, C. & Nuijten, W./ CONSTRAINT-BASED SCHEDULINGFeinberg, E. & Shwartz, A./ HANDBOOK OF MARKOV DECISION PROCESSES: Methods

and ApplicationsRamík, J. & Vlach, M. / GENERALIZED CONCAVITY IN FUZZY OPTIMIZATION

AND DECISION ANALYSISSong, J. & Yao, D. / SUPPLY CHAIN STRUCTURES: Coordination, Information and

OptimizationKozan, E. & Ohuchi, A./ OPERATIONS RESEARCH/ MANAGEMENT SCIENCE AT WORKBouyssou et al/ AIDING DECISIONS WITH MULTIPLE CRITERIA: Essays in

Honor of Bernard RoyCox, Louis Anthony, Jr./ RISK ANALYSIS: Foundations, Models and Methods

INTERNATIONAL SERIES INOPERATIONS RESEARCH & MANAGEMENT SCIENCEFrederick S. Hillier, Series Editor Stanford University

Page 3: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

MODELING UNCERTAINTYAn Examination of Stochastic Theory,Methods, and Applications

Edited byMOSHE DRORUniversity of Arizona

PIERRE L’ECUYERUniversité de Montréal

FERENC SZIDAROVSZKYUniversity of Arizona

KLUWER ACADEMIC PUBLISHERSNEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW

Page 4: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

eBook ISBN: 0-306-48102-2Print ISBN: 0-7923-7463-0

©2005 Springer Science + Business Media, Inc.

Print ©2002 Kluwer Academic Publishers

All rights reserved

No part of this eBook may be reproduced or transmitted in any form or by any means, electronic,mechanical, recording, or otherwise, without written consent from the Publisher

Created in the United States of America

Visit Springer's eBookstore at: http://ebooks.springerlink.comand the Springer Global Website Online at: http://www.springeronline.com

Dordrecht

Page 5: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contents

Professor Sidney J. YakowitzD. S. Yakowitz

Preface

Contributing Authors

Part I

Stability of Single Class Queueing NetworksHarold J. Kushner

1

2

12345

IntroductionThe ModelStability: IntroductionPerturbed Liapunov FunctionsStability

3Sequential Optimization Under UncertaintyTze Leung Lai

12

IntroductionBandit Theory2.1 Nearly optimal rules based on upper confidence bounds and

Gittins indicesA hypothesis testing approach and block experimentationApplications to machine learning, control and scheduling ofqueues

2.22.3

3 Adaptive Control of Markov Chains3.13.2

Parametric adaptive controlNonparametric adaptive control

4 Stochastic Approximation

4Exact Asymptotics for Large Deviation Probabilities, withApplications

xvii

xxi

1

13

13

1315222328

35

3537

3742

4444454749

57

Page 6: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

vi MODELING UNCERTAINTY

Iosif Pinelis1. Limit Theorems on the last negative sum and applications to non-

parametric bandit theory1.11.2

Condition (4)&(8): exponential and superexponential casesCondition (4)&(8): exponential (beyond (14)) and subexpo-nential casesThe conditional distribution of the initial segmentof the sequence of the partial sums givenApplication to Bandit Allocation AnalysisTest-times-only based strategyMultiple bandits and all-decision-times based strategy

1.3

1.41.4.11.4.2

23

Large deviations in a space of trajectoriesAsymptotic equivalence of the tail of the sum of independent randomvectors and the tail of their maximum3.13.2

3.3

IntroductionExponential inequalities for probabilities of large deviationof sums of independent Banach space valued r.v.’sThe case of a fixed number of independent Banach space val-ued r.v.’s. Application to asymptotics of infinitely divisibleprobability distributions in Banach spacesTails decreasing no faster than power onesTails, decreasing faster than any power ones

Tails, decreasing no faster than

3.43.53.6

Part II

5Stochastic Modelling of Early HIV Immune Responses Under Treatment

by Protease InhibitorsWai-Yuan Tan and Zhihua Xiang

12

IntroductionA Stochastic Model of Early HIV Pathogenesis Under Treatment bya Protease Inbihitor2.12.2

2.3

Modeling the Effects of Protease InhibitorsModeling the Net Flow of HIV From Lymphoid Tissues toPlasmaDerivation of Stochastic Differential Equations for The StateVariables

3

4

Mean Values of

A State Space Model for the Early HIV Pathogenesis Under Treat-ment by Protease Inhibitors4.1

4.2

Estimation of

Estimation of and

56

An Example Using Real DataSome Monte Carlo Studies

5962

63

6668687072

7777

81

838688

89

95

95

96

9798

99

100

103

104106

107

108113

given

Given with

Page 7: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contents

6The impact of re-using hypodermic needlesB. Barnes and J. Gani

12345678

IntroductionGeometric distribution with variable success probabilityValidity of the distributionMean and variance of IIntensity of epidemicReducing infectionThe spread of the Ebola virus in 1976Conclusions

7Nonparametric Frequency Detection and Optimal Coding in Molecular

BiologyDavid S. Stoffer

1234

IntroductionThe Spectral EnvelopeSequence AnalysesDiscussion

Part III

8An Efficient Stochastic Approximation Algorithm for Stochastic Saddle

Point ProblemsArkadi Nemirovski and Reuven Y. Rubinstein

1 Introduction1.1 Classical stochastic approximation

2 Stochastic saddle point problem2.12.1.12.1.22.22.32.4

2.5

The problemStochastic settingThe accuracy measureExamplesThe SASP algorithmRate of convergence and optimal setup: off-line choice ofthe stepsizesRate of convergence and optimal setup: on-line choice of thestepsizes

3 Discussion3.13.2

Comparison with Polyak’s algorithmOptimality issues

4 Numerical Results4.14.2

A Stochastic Minimax Steiner problemA simple queuing model

5 ConclusionsAppendix: A: Proof of Theorems 1 and 2

vii

117

117118119120122123124128

129

129133140152

155

155

155155157157158159159162

163

164167167168172172174178179

Page 8: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

viii MODELING UNCERTAINTY

Appendix: B: Proof of the Proposition 182

9Regression Models for Binary Time SeriesBenjamin Kedem, Konstantinos Fokianos

12

IntroductionPartial Likelihood Inference2.12.22.32.4

Definition of Partial LikelihoodAn Assumption Regarding the CovariatesPartial Likelihood EstimationPrediction

34

Goodness of FitLogistic Regression4.1 A Demonstration

5 Categorical Data

10Almost Sure Convergence Properties of Nadaraya-Watson Regression

EstimatesHarro Walk

123

IntroductionResultsLemmas and Proofs

11Strategies for Sequential Prediction of Stationary Time SeriesLászló Györfi, Gábor Lugosi

1234

IntroductionUniversal prediction by partitioning estimatesUniversal prediction by generalized linear estimatesPrediction of Gaussian processes

Part IV

12The Birth of Limit Cycles in Nonlinear Oligopolies with Continuously

Distributed Information LagCarl Chiarella and Ferenc Szidarovszky

1234567

IntroductionNonlinear Oligopoly ModelsThe Dynamic Model with Lag StructureBifurcation Analysis in the General CaseThe Symmetric CaseSpecial Oligopoly ModelsConclusions

185

185187187188188190191192194196

201

201203205

225

225228236240

249

249

249251251253259263267

Page 9: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contents ix13A Differential Game of Debt Contract ValuationA. Haurie and F. Moresino

1234

IntroductionThe firm and the debt contractA stochastic gameEquivalent risk neutral valuation4.14.2

Debt and Equity valuations when bankrupcy is not consideredDebt and Equity valuations when liquidation may occur

567

Debt and Equity valuations for Nash equilibrium strategiesLiquidation at fixed time periodsConclusion

14Huge Capacity Planning and Resource Pricing for Pioneering ProjectsDavid Porter

123

IntroductionThe ModelResults3.13.23.3

Cost and Performance UncertaintyCost Uncertainty and FlexibilityPerformance Uncertainty and Flexibility

4 Conclusion

15Affordable Upgrades of Complex Systems: A Multilevel, Performance-

Based ApproachJames A. Reneke and Matthew J. Saltzman and Margaret M. Wiecek

12

IntroductionMultilevel complex systems2.12.2

An illustrative exampleComputational models for the example

3 Multiple criteria decision making3.13.23.3

Generating candidate methodsChoosing a preferred selection of upgradesApplication to the example

4 Stochastic analysis4.14.2

Random systems and riskApplication to the example

5 ConclusionsAppendix: Stochastic linearization12

Origin of stochastic linearizationStochastic linearization for random surfaces

16On Successive Approximation of Optimal Control of Stochastic Dynamic

SystemsFei-Yue Wang, George N. Saridis

269

269270273275276278280281282

285

285287291292297298298

301

301306309312313314315317320321321322327327327

333

Page 10: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

x MODELING UNCERTAINTY

12345

IntroductionProblem StatementSub-Optimal Control of Nonlinear Stochastic Dynamic SystemsThe Infinite-time Stochastic Regulator ProblemProcedure for Iterative Design of Sub-optimal Controllers5.15.2

Exact Design ProcedureApproximate Design Procedures for the Regulator Problem

6 Closing Remarks by Fei-Yue Wang

17Stability of Random Iterative MappingsLászló Gerencsér

123

IntroductionPreliminary resultsThe proof of Theorem 1.1

Appendix

Part V

18’Unobserved’ Monte Carlo Methods for Adaptive AlgorithmsVictor Solo

1234

56

El SidIntroductionOn-line Binary ClassificationBinary Classification with Noisy Measurements of Classifying Variables-OfflineBinary Classification with Errors in Classifying Variables -OnlineConclusions

19Random Search Under Additive NoiseLuc Devroye and Adam Krzyzak

1234567891011

Sid’s contributions to noisy optimizationFormulation of search problemRandom search: a brief overviewNoisy optimization by random search: a brief surveyOptimization and nonparametric estimationNoisy optimization: formulation of the problemPure random searchStrong convergence and strong stabilityMixed random searchStrategies for general additive noiseUniversal convergence

334335337346349349353356

359

359364367368

373

373

373374375

376378380

383

383384385390393394394398399400410

Page 11: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contents xi20Recent Advances in Randomized Quasi-Monte Carlo MethodsPierre L’Ecuyer and Christiane Lemieux

123

IntroductionA Closer Look at Low-Dimensional ProjectionsMain Constructions3.13.23.2.13.2.23.2.33.2.43.33.43.53.6

Lattice RulesDigital NetsSobol’ SequencesGeneralized Faure SequencesNiederreiter SequencesPolynomial Lattice RulesConstructions Based on Small PRNGsHalton sequenceSequences of Korobov rulesImplementations

4 Measures of Quality4.14.2

Criteria for standard lattice rulesCriteria for digital nets

5 Randomizations5.15.25.35.45.5

Random shift modulo 1Digital shiftScramblingRandom Linear ScramblingOthers

6 Error and Variance Analysis6.16.26.2.16.2.2

Standard Lattices and Fourier ExpansionDigital Nets and Haar or Walsh ExpansionsScrambled-type estimatorsDigitally shifted estimators

789

Transformations of the IntegrandRelated MethodsConclusions and Discussion

Appendix: Proofs

Part VI

21Singularly Perturbed Markov Chains and Applications to Large-Scale Sys-

tems under UncertaintyG. Yin, Q. Zhang, K. Yin and H. Yang

12

IntroductionSingularly Perturbed Markov Chains2.12.2

Continuous-time CaseTime-scale Separation

3 Properties of the Singularly Perturbed Systems3.13.23.3

Asymptotic ExpansionOccupation MeasuresLarge Deviations and Exponential Bounds

419

420423425426428431431432433435438439439440441444448449449450451452452453455455457461462464464

475

475

476480481483485485487492

Page 12: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

xii MODELING UNCERTAINTY

3.3.13.3.2

Large DeviationsExponential Bounds

4 Controlled Singularly Perturbed Markovian Systems4.14.2

Continuous-time Hybrid LQGDiscrete-time LQ

56

Further RemarksAppendix: Mathematical Preliminaries6.16.26.3

Stochastic ProcessesMarkov chainsConnections of Singularly Perturbed Models: ContinuousTime vs. Discrete Time

22Risk–Sensitive Optimal Control in Communicating Average Markov

Decision ChainsRolando Cavazos–Cadena, Emmanuel Fernández–Gaucherand

12345678

IntroductionThe Decision ModelMain ResultsBasic Technical PreliminariesAuxiliary Expected–Total Cost Problems: IAuxiliary Expected–Total Cost Problems: IIProof of Theorem 3.1Conclusions

Appendix: A: Proof of Theorem 4.1Appendix: B: Proof of Theorem 4.2

23Some Aspects of Statistical Inference in a Markovian and Mixing FrameworkGeorge G. Roussas

12

IntroductionMarkovian Dependence2.12.2

Parametric Case - The Classical ApproachParametric Case - The Local Asymptotic Normality Approach

2.3 The Nonparametric Case3 Mixing

3.13.23.33.43.4.13.4.23.4.33.4.43.4.53.4.63.4.7

Introduction and DefinitionsCovariance InequalitiesMoment and Exponential Probability BoundsSome Estimation ProblemsEstimation of the Distribution Function or Survival FunctionEstimation of a Probability Density Function and its DerivativesEstimating the Hazard RateA Smooth Estimate of F andRecursive EstimationFixed Design RegressionStochastic Design Regression

492493494495498504505505506

508

515

516518521526529538542544547551

555

556557558561

567576576578581582582584586588589591592

Page 13: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contents xiii

Part VII

24Stochastic Ordering of Order Statistics IIPhilip J. Boland, Taizhong Hu, Moshe Shaked and J. George Shanthikumar

607

607

1234567

IntroductionLikelihood Ratio Orders ComparisonsHazard and Reversed Hazard Rate Orders ComparisonsUsual Stochastic Order ComparisonsStochastic Comparisons of SpacingsDispersive Ordering of Order Statistics and SpacingsA Short Survey on Further Results

608609611615615618620

25Vehicle Routing with Stochastic Demands: Models & Computational

MethodsMoshe Dror

625

12

IntroductionAn SVRP Example and Simple Heuristic Results2.1 Chance Constrained Models

3 Modeling SVRP as a stochastic programming with recourse problem3.13.23.3

The modelThe branch-and-cut procedureComputation of a lower bound on and on Q(x)

4 Multi-stage model for the SVRP4.1 The multi-stage model

5678

Modeling SVRP as a Markov decision processSVRP routes with at most one failure – a more ‘practical’ approachThe Dror conjectureSummary

26Life in the Fast Lane: Yates’s Algorithm, Fast Fourier and Walsh TransformsPaul J. Sanchez, John S. Ramberg and Larry Head

12

IntroductionLinear Models2.12.1.12.1.22.1.3

Factorial AnalysisDefinitions and BackgroundThe ModelThe Coefficient EstimatorWalsh AnalysisDefinitions and BackgroundThe ModelDiscrete Walsh TransformsFourier AnalysisDefinitions and BackgroundThe Model

2.22.2.12.2.22.2.32.32.3.12.3.2

3 An Example

625627630631633635636638640641643645646

651

652653654654656658658658662662663663665666

Page 14: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

xiv MODELING UNCERTAINTY

4 Fast Algorithms4.14.24.3

Yates’s Fast Factorial AlgorithmFast Walsh TransformsFast Fourier Transforms

5 ConclusionsAppendix: A: Table of Notation

27Uncertainty Bounds in Parameter Estimation with Limited DataJames C. Spall

123

IntroductionProblem FormulationThree Examples of Appropriate Problem Settings3.1

3.23.3

Example 1: Parameter Estimation in Signal-Plus-Noise Modelwith Non-i.i.d. DataExample 2: Nonlinear Input-Output (Regression) ModelExample 3: Estimates of Serial Correlation for Time Series

4 Main Results4.14.24.3

Background and NotationOrder Result on Small-Sample ProbabilitiesThe Implied Constant of Bound

5 Application of Theorem for the MLE of Parameters in Signal-Plus-Noise Problem5.15.2

BackgroundTheorem Regularity Conditions and Calculation of ImpliedConstantNumerical Results5.3

6 Summary and ConclusionsAppendix: Theorem Regularity Conditions and Proof (Section 4)

28A Tutorial on Hierarchical Lossless Data CompressionJohn C. Kieffer

1 Introduction1.11.21.3

Pointer Tree RepresentationsData Flow Graph RepresentationsContext-Free Grammar Representations

2 Equivalences Between Structures2.12.2

Equivalence of Pointer Trees and Admissible GrammarsEquivalence of Admissible Grammars and Data Flow Graphs

345

Design of Compact StructuresEncoding MethodologyPerformance Under Uncertainty

Part VIII

29Eureka! Bellman’s Principle of Optimality is valid!

670671676679682684

685

686688689

690691692693693695695

697697

698699702703

711

711715716718721721723725727729

735

735

Page 15: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contents xv

Moshe Sniedovich1234567

IntroductionRemediesThe Big FixThe Rest is MathematicsRefinementsNon-Markovian Objective functionsDiscussion

30Reflections on Statistical Methods for Complex Stochastic SystemsMarcel F. Neuts

123

The Changed Statistical SceneMeasuring Teletraffic Data StreamsMonitoring Queueing Behavior

735738739740744746748

751

751754757

Author Index 761

Page 16: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Preface

This volume titled MODELING UNCERTAINTY: An Examination of Stochas-tic Theory, Methods, and Applications, has been compiled by the friends andcolleagues of Sid Yakowitz in his honor as a token of love, appreciation, andsorrow for his untimely death. The first paper in the book is authored by Sid’swife – Diana Yakowitz – and in it Diana describes Sid the person, his drivefor knowledge and his fascination with mathematics, particularly with respectto uncertainty modelling and applications. This book is a collection of paperswith uncertainty as its central theme.

Fifty authors from all over the world collectively contributed 30 papers tothis volume. Each of these papers was reviewed and in the majority of casesthe original submission was revised before being accepted for publication inthe book. The papers cover a great variety of topics in probability, statistics,economics, stochastic optimization, control theory, regression analysis, simula-tion, stochastic programming, Markov decision process, application in the HIVcontext, and others. Some of the papers have a theoretical emphasis and othersfocus on applications. A number of papers have the flavor of survey work in aparticular area and in a few papers the authors present their personal view of atopic. This book has a considerable number of expository articles which shouldbe accessible to a nonexpert, say a graduate student in mathematics, statistics,engineering, and economics departments, or just anyone with some mathemat-ical background who is interested in a preliminary exposition of a particulartopic. A number of papers present the state of the art of a specific area orrepresent original contributions which advance the present state of knowledge.Thus, the book has something for almost anybody with an interest in stochasticsystems.

The editors have loosely grouped the chapters into 8 segments, accordingto some common mathematical thread. Since none of us (the co-editors) is anexpert in all the topics covered in this book, it is quite conceivable that the pa-pers could have been grouped differently. Part 1 starts with a paper on stabilityin queuing networks by H.J. Kushner. Part 1 also includes a queuing related

Page 17: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

paper by T.L. Lai, and a paper by I. Pinelis on asymptotics for large deviationprobabilities. Part 2 groups together 3 papers related to HIV modelling. Thefirst paper in this group is by W.-Y. Tan and Z. Xiang about modelling earlyimmune responses, followed by a paper of B. Barnes and J. Gani on the impactof re-using hypodermic needs, and closes with a paper by D.S. Stoffer. Part 3groups together optimization and regression papers. It contains 4 papers startingwith a paper by A. Nemirovski and R.Y. Rubinstein about classical stochasticapproximation. The next paper is by B. Kedem and K. Fokianos on regressionmodels for binary time series, followed with a paper by H. Walk on properties ofNadarya - Watson regression estimates, and closing with a paper on sequentialpredictions of stationary time series by L. Györfi and G. Lugosi. Part 4’s 6 pa-pers are in the area of economics analysis starting with a nonlinear oligopoliespaper by C. Chiarella and F. Szidarovszky. The paper by A. Haurie and F.Moresino examines a differential game of debt contract valuation. Next comesa paper by D. Porter, followed by a paper about complex systems in relation toaffordable upgrades by J.A. Reneke, M.J. Saltzman, and M.M. Wiecek. The 5thpaper in this group, by F.-Y. Wang and G.N. Sardis, concerns optimal controlin stochastic dynamic systems, and the last paper is by L. Gerencsér is aboutstability of random iterative mappings. Part 5 loosely groups 3 papers startingwith a paper by V. Solo on Monte Carlo methods for adaptive algorithms, fol-lowed by a paper on random search with noise by L. Devroye and A. Krzyzak,and closes with a survey paper on randomized quasi-Monte Carlo methods byP. L’Ecuyer and C. Lemieux. Part 6 is a collection of 3 papers sharing a focuson Markov decision analysis. It starts with a paper by G. Yin, Q. Zhang, K.Yin, and H. Yang on singularly perturbed Markov chains. The second paper, onrisk sensitivity in average Markov decision chains, is by R. Cavazos–Cadenaand E. Fernández–Gaucherand. The 3rd paper, by G.G. Roussas, is on statis-tical inference in a Markovian framework. Part 7 includes a paper on orderstatistics by P.J. Boland, T. Hu, M. Shaked, and J.G. Shanthikumar, followedby a survey paper on routing with stochastic demands by M. Dror, a paper onfast Fourier and Walsh transforms by P.J. Sanchez, J.S. Ramberg, and L. Head,a paper by J.C. Spall on parameter estimation with limited data, and a tuto-rial paper on data compression by J.C. Kieffer. Part 8 contains 2 ‘reflections’papers. The first paper is by M. Sniedovich – an ex-student of Sid Yakowitz.It reexamines Bellman’s principle of optimality. The last paper in this volumeon statistical methods for complex stochastic systems is reserved to M.F. Neuts.

The efforts of many workers have gone into this volume, and would not havebeen possible without the collective work of all the authors and reviewers whoread the papers and commented constructively. We would like to take this op-portunity to thank the authors and the reviewers for their contributions. Thisbook would have required a more difficult ’endgame’ without Ray Brice’s ded-

xviii MODELING UNCERTAINTY

Page 18: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

PREFACE xix

ication and painstaking attention for production details. We are very gratefulfor Ray’s help in this project. Paul Jablonka is the artist who contributed the artwork for the book’s jacket. He was a good friend to Sid and we appreciate hiscontribution. We would also like to thank Gary Folven, the editor of KluwerAcademic Publishers, for his initial and never fading support throughout thisproject. Thank you Gary !

Moshe Dror Pierre L’Ecuyer Ferenc Szidarovszky

Page 19: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contributing Authors

B. BarnesSchool of Mathematical Sciences

Australian National University

Canberra, ACT 0200

Australia

Philip J. BolandDepartment of Statistics

University College Dublin

Belfield, Dublin 4

Ireland

Rolando Cavazos–CadenaDepartamento de Estadística y Cálculo

Universidad Auténoma Agraria Antonio Narro

Buenavista, Saltillo COAH 25315

MÉXICO

Carl ChiarellaSchool of Finance and Economics

University of Technology

Sydney

P.O. Box 123, Broadway, NSW 2007

Australia

[email protected]

Luc DevroyeSchool of Computer Science

McGill University

Montreal, Canada H3A 2K6

Page 20: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

xxii MODELING UNCERTAINTY

Moshe DrorDepartment of Management Information Systems

The University of Arizona

Tucson, AZ 85721, USA

[email protected]

Emmanuel Fernández–GaucherandDepartment of Electrical & Computer Engineering

& Computer Science

University of Cincinnati

Cincinnati, OH 45221-0030

USA

Konstantinos FokianosDepartment of Mathematics & Statistics

University of Cyprus

P.O. Box 20537 Nikosia, 1678, Cyprus

J. GaniSchool of Mathematical Sciences

Australian National University

Canberra, ACT 0200

Australia

László GerencsérComputer and Automation Institute

Hungarian Academy of Sciences

H-1111, Budapest Kende u 13-17

Hungary

László GyörfiDepartment of Computer Science and Information Theory

Technical University of Budapest

1521 Stoczek u. 2,

Budapest, Hungary

[email protected]

A. HaurieUniversity of Geneva

Geneva Switzerland

Page 21: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contributing Authors xxiii

Larry HeadSiemens Energy & Automation, Inc.

Tucson, AZ 85715

Taizhong HuDepartment of Statistics and Finance

University of Science and Technology

Hefei, Anhui 230026

People’s Republic of China

Benjamin KedemDepartment of Mathematics

University of Maryland

College Park, Maryland 20742, USA

John C. KiefferECE Department

University of Minnesota

Minneapolis, MN 55455

Department of Computer Science

Concordia University

Montreal, Canada H3G 1M8

Harold J. KushnerApplied Mathematics Dept.

Lefschetz Center for Dynamical Systems

Brown University

Providence RI 02912

Tze Leung LaiStanford University

Stanford, California

Adam Krzyzak

Page 22: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

xxiv MODELING UNCERTAINTY

Pierre L’EcuyerDépartement d’Informatique et de Recherche Opérationnelle

Université de Montréal, C.P. 6128, Succ. Centre-Ville

Montréal, H3C 3J7, Canada

[email protected]

Christiane LemieuxDepartment of Mathematics and Statistics

University of Calgary, 2500 University Drive N.W.

Calgary, T2N 1N4, Canada

[email protected]

Gábor LugosiDepartment of Economics,

Pompeu Fabra University

Ramon Trias Fargas 25-27,

08005 Barcelona, Spain

[email protected]

F. MoresinoCambridge University

United Kingdom

Arkadi NemirovskiFaculty of Industrial Engineering and Management

Technion—Israel Institute of Technology

Haifa 32000, Israel

Marcel F. NeutsDepartment of Systems and Industrial Engineering

The University of Arizona

Tucson, AZ 85721, U.S.A.

[email protected]

Iosif PinelisDepartment of Mathematical Sciences

Michigan Technological University

Houghton, Michigan 49931

[email protected]

Page 23: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contributing Authors xxv

David PorterCollage of Arts and Sciences

George Mason University

John S. RambergSystems and Industrial Engineering

University of Arizona

Tucson, AZ 85721

James A. RenekeDept. of Mathematical Sciences

Clemson University

Clemson SC 29634-0975

George G. RoussasUniversity of California, Davis

Reuven Y. RubinsteinFaculty of Industrial Engineering and Management

Technion—Israel Institute of Technology

Haifa 32000, Israel

Matthew J. SaltzmanDept. of Mathematical Sciences

Clemson University

Clemson SC 29634-0975

Paul J. SanchezOperations Research Department

Naval Postgraduate School

Monterey, CA 93943

George N. SaridisDepartment of Electrical, Computer and Systems Engineering

Rensselaer Polytechnic Institute

Troy, New York 12180

Page 24: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

xxvi MODELING UNCERTAINTY

Moshe ShakedDepartment of Mathematics

University of Arizona

Tucson, Arizona 85721

USA

J. George ShanthikumarIndustrial Engineering & Operations Research

University of California

Berkeley, California 94720

USA

Moshe SniedovichDepartment of Mathematics and Statistics

The University of Melbourne

Parkville VIC 3052, Australia

[email protected]

Victor SoloSchool of Electrical Engineering and Telecommunications

University of New South Wales

Sydney NSW 2052, Australia

[email protected]

James C. SpallThe Johns Hopkins University

Applied Physics Laboratory

Laurel, MD 20723-6099

[email protected]

David S. StofferDepartment of Statistics

University of Pittsburgh

Pittsburgh, PA 15260

Ferenc SzidarovszkyDepartment of Systems and Industrial Engineering

University of Arizona

Tucson, Arizona, 85721-0020, USA

[email protected]

Page 25: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Contributing Authors xxvii

Wai-Yuan TanDepartment of Mathematical Sciences

The University of Memphis

Memphis, TN 38152-6429

[email protected]

Harro WalkMathematisches Institut A

Universität Stuttgart

Pfaffenwaldring 57, D-70569

Stuttgart, Germany

Fei-Yue WangDepartment of Systems and Industrial Engineering

University of Arizona

Tucson, Arizona 87521

Margaret M. WiecekDept. of Mathematical Sciences

Clemson University

Clemson SC 29634-0975

Zhihua XiangOrganon Inc.

375 Mt. Pleasant Avenue

West Orange, NJ 07052

[email protected]

D. S. YakowitzTucson, Arizona

H.YangDepartment of Wood and Paper Science

University of Minnesota

St. Paul, MN 55108

[email protected]

Page 26: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

xxviii MODELING UNCERTAINTY

G. YinDepartment of Mathematics

Wayne State University

Detroit, MI 48202

[email protected]

K. YinDepartment of Wood and Paper Science

University of Minnesota

St. Paul, MN 55108

[email protected], [email protected]

Q. ZhangDepartment of Mathematics

University of Georgia

Athens, GA 30602

[email protected]

Page 27: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

This book is dedicated to thememory of Sid Yakowitz.

Page 28: Modeling Uncertainty - An Examination of Stochastic Theory … · MODELING UNCERTAINTY An Examination of Stochastic Theory, Methods, and Applications Edited by MOSHE DROR University

Chapter 1

PROFESSOR SIDNEY J. YAKOWITZ

D. S. YakowitzTucson, Arizona

Sidney Jesse Yakowitz was born in San Francisco, California on March8, 1937 and died in Eugene, Oregon on September 1, 1999. Sid’s parents,Morris and MaryVee, were chemists with the Food and Drug Administrationand encouraged Sid to be a life-long learner. He attended Stanford Universityand after briefly toying with the idea of medicine, settled into engineering (“Isaved hundreds of lives with that decision!”). Sid graduated from Stanford witha B.S in Electrical Engineering in 1960.

His first job out of Stanford was as a design engineer with the Universityof California’s Lawrence Radiation Laboratory (LRL) at Berkeley. Sid wasunhappy after college but claimed that he learned the secret to happiness fromhis office mate at LRL, Jim Sherwood, who told him he was being paid to becreative. Sid decided that “Good engineering design is a synonym for ‘invent-ing’.”

For graduate school, Sid chose Arizona State University. By this time, hisbattle since childhood with acute asthma made a dry desert climate a manda-tory consideration. In graduate school he flourished. He received his M.S. inElectrical Engineering in 1965, an M.A. in Mathematics in 1966, and Ph.D. inElectrical Engineering in 1967. His new formula for happiness in his work ledhim to consider each topic or problem that he approached as an opportunity to“invent”.

In 1966 Sid was hired as an Assistant Professor in the newly founded De-partment of Systems and Industrial Engineering at the University of Arizona inTucson. This department remained his “home” for 33 years with the exceptionof brief sabbaticals and leaves such as a National Academy of Science Post-doctoral Fellowship at the Naval Postgraduate School in Monterey, Californiain 1970-1971.

In 1969 Sid’s book Mathematics of Adaptive Control Processes (Yakowitz,1969) was published as a part of Richard Bellman’s Elsevier book series. Thisbook was essentially his Ph.D. dissertation and was the first of four published