ThesenotesprovideanintroductiontoMarkovchainMonteCarlomethods
thatareusefulinbothBayesianandfrequentiststatisticalinference.Such
methodshaverevolutionizedwhatcanbeachievedcomputationally,primarily
butnotonlyintheBayesianparadigm.Theaccountbeginsbydescribingordi-
naryMonteCarlomethods,which,inprinciple,haveexactlythesamegoalsas
theMarkovchainversionsbutcanrarelybeimplemented.Subsequentsections
describebasicMarkovchainMonteCarlo,foundedontheHastingsalgorithm
andincludingboththeMetropolismethodandtheGibbssamplerasspecial
cases,andgoontodiscussmorerecentdevelopments.TheseincludeMarkov
chainMonteCarlo p–values,theLangevin–Hastingsalgorithm,auxiliaryvari-
ablestechniques,perfectMarkovchainMonteCarloviacouplingfromthe
past,andreversiblejumpsmethodsfortargetspacesofvaryingdimensions.
Specimenapplications,drawnfromseveraldi?erentdisciplines,aredescribed
throughoutthenotes.Severaloftheseappearforthefirsttime.Allcompu-
tationsuseAPLastheprogramminglanguage,thoughthisisnotnecessarily
a recommendation! The author welcomes comments and criticisms.

好东西，8快行不行？

能不能传给我一份，我的邮件是s040855@sparc20c.sta.cuhk.edu.hk!

谢谢！当然我会付给你8元！

我有钱了1

怎么两个一样的！