PROBABILITY, MARKOV CHAINS, QUEUES, AND SIMULATION This page intentionally left blank Copyright(cid:2)c2009byPrincetonUniversityPress PublishedbyPrincetonUniversityPress,41Williamstreet,Princeton,NewJersey08540 IntheUnitedKingdom:PrincetonUniversityPress,6OxfordStreet, Woodstock,OxfordshireOX201TW AllRightsReserved LibraryofCongressCataloging-in-PublicationData Stewart,WilliamJ.,1946– Probability,Markovchains,queuesandsimulation:themathematicalbasisof performancemodeling/WilliamJ.Stewart.–1sted. p.cm. ISBN978-0-691-14062-9(cloth:alk.paper)1.Probability–Computersimulation. 2.Markovprocesses.3.Queueingtheory.I.Title. QA273.S75322009 519.201(cid:2)13–dc22 2008041122 BritishLibraryCataloging-in-PublicationDataisavailable ThisbookhasbeencomposedinTimes Printedonacid-freepaper.∞ press.princeton.edu MATLABisaregisteredtrademarkofTheMathWorks,Inc. TypesetbySRNovaPvtLtd,Bangalore,India PrintedintheUnitedStatesofAmerica 10 9 8 7 6 5 4 3 2 1 Thisbookisdedicatedtoallthose whomIlove,especially Mydearwife, Kathie, andmywonderfulchildren Nicola,Stephanie,Kathryn,andWilliam Myfather,William J.Stewartand the memoryofmymother,Mary(Marshall)Stewart This page intentionally left blank Contents PrefaceandAcknowledgments xv I PROBABILITY 1 1 Probability 3 1.1 Trials,SampleSpaces,andEvents 3 1.2 ProbabilityAxiomsandProbabilitySpace 9 1.3 ConditionalProbability 12 1.4 IndependentEvents 15 1.5 LawofTotalProbability 18 1.6 Bayes’Rule 20 1.7 Exercises 21 2 Combinatorics—TheArtofCounting 25 2.1 Permutations 25 2.2 PermutationswithReplacements 26 2.3 PermutationswithoutReplacement 27 2.4 CombinationswithoutReplacement 29 2.5 CombinationswithReplacements 31 2.6 Bernoulli(Independent)Trials 33 2.7 Exercises 36 3 RandomVariablesandDistributionFunctions 40 3.1 DiscreteandContinuousRandomVariables 40 3.2 TheProbabilityMassFunctionforaDiscreteRandomVariable 43 3.3 TheCumulativeDistributionFunction 46 3.4 TheProbabilityDensityFunctionforaContinuousRandomVariable 51 3.5 FunctionsofaRandomVariable 53 3.6 ConditionedRandomVariables 58 3.7 Exercises 60 4 JointandConditionalDistributions 64 4.1 JointDistributions 64 4.2 JointCumulativeDistributionFunctions 64 4.3 JointProbabilityMassFunctions 68 4.4 JointProbabilityDensityFunctions 71 4.5 ConditionalDistributions 77 4.6 ConvolutionsandtheSumofTwoRandomVariables 80 4.7 Exercises 82 5 ExpectationsandMore 87 5.1 Definitions 87 5.2 ExpectationofFunctionsandJointRandomVariables 92 5.3 ProbabilityGeneratingFunctionsforDiscreteRandomVariables 100 viii Contents 5.4 MomentGeneratingFunctions 103 5.5 MaximaandMinimaofIndependentRandomVariables 108 5.6 Exercises 110 6 DiscreteDistributionFunctions 115 6.1 TheDiscreteUniformDistribution 115 6.2 TheBernoulliDistribution 116 6.3 TheBinomialDistribution 117 6.4 GeometricandNegativeBinomialDistributions 120 6.5 ThePoissonDistribution 124 6.6 TheHypergeometricDistribution 127 6.7 TheMultinomialDistribution 128 6.8 Exercises 130 7 ContinuousDistributionFunctions 134 7.1 TheUniformDistribution 134 7.2 TheExponentialDistribution 136 7.3 TheNormalorGaussianDistribution 141 7.4 TheGammaDistribution 145 7.5 ReliabilityModelingandtheWeibullDistribution 149 7.6 Phase-TypeDistributions 155 7.6.1 TheErlang-2Distribution 155 7.6.2 TheErlang-r Distribution 158 7.6.3 TheHypoexponentialDistribution 162 7.6.4 TheHyperexponentialDistribution 164 7.6.5 TheCoxianDistribution 166 7.6.6 GeneralPhase-TypeDistributions 168 7.6.7 FittingPhase-TypeDistributionstoMeansandVariances 171 7.7 Exercises 176 8 BoundsandLimitTheorems 180 8.1 TheMarkovInequality 180 8.2 TheChebychevInequality 181 8.3 TheChernoffBound 182 8.4 TheLawsofLargeNumbers 182 8.5 TheCentralLimitTheorem 184 8.6 Exercises 187 II MARKOVCHAINS 191 9 Discrete-andContinuous-TimeMarkovChains 193 9.1 StochasticProcessesandMarkovChains 193 9.2 Discrete-TimeMarkovChains:Definitions 195 9.3 TheChapman-KolmogorovEquations 202 9.4 ClassificationofStates 206 9.5 Irreducibility 214 9.6 ThePotential,Fundamental,andReachabilityMatrices 218 9.6.1 PotentialandFundamentalMatricesandMeanTimetoAbsorption 219 9.6.2 TheReachabilityMatrixandAbsorptionProbabilities 223 Contents ix 9.7 RandomWalkProblems 228 9.8 ProbabilityDistributions 235 9.9 Reversibility 248 9.10 Continuous-TimeMarkovChains 253 9.10.1 TransitionProbabilitiesandTransitionRates 254 9.10.2 TheChapman-KolmogorovEquations 257 9.10.3 TheEmbeddedMarkovChainandStateProperties 259 9.10.4 ProbabilityDistributions 262 9.10.5 Reversibility 265 9.11 Semi-MarkovProcesses 265 9.12 RenewalProcesses 267 9.13 Exercises 275 10 NumericalSolutionofMarkovChains 285 10.1 Introduction 285 10.1.1 SettingtheStage 285 10.1.2 StochasticMatrices 287 10.1.3 TheEffectofDiscretization 289 10.2 DirectMethodsforStationaryDistributions 290 10.2.1 IterativeversusDirectSolutionMethods 290 10.2.2 GaussianEliminationandLUFactorizations 291 10.3 BasicIterativeMethodsforStationaryDistributions 301 10.3.1 ThePowerMethod 301 10.3.2 TheIterativeMethodsofJacobiandGauss–Seidel 305 10.3.3 TheMethodofSuccessiveOverrelaxation 311 10.3.4 DataStructuresforLargeSparseMatrices 313 10.3.5 InitialApproximations,Normalization,andConvergence 316 10.4 BlockIterativeMethods 319 10.5 DecompositionandAggregationMethods 324 10.6 TheMatrixGeometric/AnalyticMethodsforStructuredMarkovChains 332 10.6.1 TheQuasi-Birth-DeathCase 333 10.6.2 BlockLowerHessenbergMarkovChains 340 10.6.3 BlockUpperHessenbergMarkovChains 345 10.7 TransientDistributions 354 10.7.1 MatrixScalingandPoweringMethodsforSmallStateSpaces 357 10.7.2 TheUniformizationMethodforLargeStateSpaces 361 10.7.3 OrdinaryDifferentialEquationSolvers 365 10.8 Exercises 375 III QUEUEINGMODELS 383 11 ElementaryQueueingTheory 385 11.1 IntroductionandBasicDefinitions 385 11.1.1 ArrivalsandService 386 11.1.2 SchedulingDisciplines 395 11.1.3 Kendall’sNotation 396 11.1.4 GraphicalRepresentationsofQueues 397 11.1.5 PerformanceMeasures—MeasuresofEffectiveness 398 11.1.6 Little’sLaw 400