AdvancedMarkovChainMonteCarloMethods:LearningfromPastSamples

MarkovChainMonteCarlo(MCMC)methodsarenowanindispensabletoolinscientificcomputing.ThisbookdiscussesrecentdevelopmentsofMCMCmethodswithanemphasisonthosemakinguseofpastsampleinformationduringsimulations.Theapplicationexamplesaredrawnfromdiversefieldssuchasbioinformatics,machinelearning,socialscience,combinatorialoptimization,andcomputationalphysics.KeyFeatures:ExpandedcoverageofthestochasticapproximationMonteCarloanddynamicweightingalgorithmsthatareessentiallyimmunetolocaltrapproblems.AdetaileddiscussionoftheMonteCarloMetropolis-Hastingsalgorithmthatcanbeusedforsamplingfromdistributionswithintractablenormalizingconstants.Up-to-dateaccountsofrecentdevelopmentsoftheGibbssampler.Comprehensiveoverviewsofthepopulation-basedMCMCalgorithmsandtheMCMCalgorithmswithadaptiveproposals.Thisbookcanbeusedasatextbookorareferencebookforaone-semestergraduatecourseinstatistics,computationalbiology,engineering,andcomputersciences.Appliedortheoreticalresearcherswillalsofindthisbookbeneficial.