简介:GivenanewDouble-MarkovriskmodelDM=(μ,Q,ν,H;Y,Z)andDouble-MarkovriskprocessU={U(t),t≥0}.Theruinorsurvivalproblemisaddressed.Equationswhichthesurvivalprobabilitysatisfiedandtheformulasofcalculatingsurvivalprobabilityareobtained.Recursionformulasofcalculatingthesurvivalprobabilityandanalyticexpressionofrecursionitemsareobtained.TheconclusionsareexpressedbyQmatrixforaMarkovchainandtransitionprobabilitiesforanotherMarkovChain.
简介:InthispaperareversibleMarkovprocessasachemicalpolymersreactionoftwotypesofmonomersisdefined.Byanalyzingthepartitionfunctionsoftheprocessweobtainthreedifferentdistributionsoftheaveragemolecularweight,dependingofthevalueofstrengthofthefragmentationreaction,andprovethatagelationoftheprocesswilloccurinthethermodynamiclimit.
简介:TheauthorpresentssomestraightforwardproofsfortwocomparisontheoremsforGreenfunctionsofMarkovchains,whichslightlyimprovethepreviousresultsbyVaropoulos,DurrettandYanandChen.ArecentresultbyRogersandWilliamsaboutinstantaneousMarkovchainsisalsoimprovedbyusingthesameidea.
简介:<正>Inthispaper.wegivecharacterizationsofNashinequalitiesforbirth-deathprocessanddiffusionprocessontheline.Asaby-product.weprovethatfortheseprocesses.transienceimpliesthatthesemigroupsP(t)decayas‖P(t)‖1--x≤Ct-1.SufficientconditionsforgeneralMsrkovchainsarealsoobtained.
简介:1.IntrodnctionTheweightedMarkovdecisionprocesses(MDP’s)havebeenextensivelystudiedsince1980’s,seeforinstance,[1-6]andsoon.ThetheoryofweightedMDP’swithperturbedtransitionprobabilitiesappearstohavebeenmentionedonlyin[7].Thispaperwilldiscussthemodelsofwe...
简介:Weraiseandpartlyanswerthequestion:whetherthereexistsaMarkovsystemwithrespecttowhichthezerosoftheChebyshevpolynomialsaredense,butthemaximumlengthofazerofreeintervalofthenthChebyshevpolynomialdoesnottendstozero.Wealsodrawtheconclu-tionthataMarkovsystem,underanadditionalassumption,isdenseifandonlyifthemaxi-mumlengthofazerofreeintervalofthenthassociatedChebyshevpolynomialtendstozero.
简介:在这篇论文,我们考虑Markov进程(X~∈(t),Z~∈(t))相应于有小参数∈的一个微弱地联合的椭圆形的PDE系统>0。我们首先证明那(X~∈(t),Z~∈(t))由联合方法有Feller连续性,然后证明那(X~∈(t),Z~∈t))由收养的Lyapunov不平等有不变的措施μ~∈(·)。最后,当小参数∈趋于到零,我们为μ~∈(·)建立一个大偏差原则。
简介:SupposethatCisafinitecollectionofpatterns.ObserveaMarkovchainuntiloneofthepatternsinCoccursasarun.Thistimeisdenotedbyτ.Inthispaper,weaimtogiveaneasywaytocalculatethemeanwaitingtimeE(τ)andthestoppingprobabilitiesP(τ=τA)withA∈C,whereτAisthewaitingtimeuntilthepatternAappearsasarun.
简介:Inordertoovercomethedisadvantagesoflowaccuracyrate,highcomplexityandpoorrobustnesstoimagenoiseinmanytraditionalalgorithmsofcloudimagedetection,thispaperproposedanovelalgorithmonthebasisofMarkovRandomField(MRF)modeling.Thispaperfirstdefinedalgorithmmodelandderivedthecorefactorsaffectingtheperformanceofthealgorithm,andthen,thesolvingofthisalgorithmwasobtainedbytheuseofBeliefPropagation(BP)algorithmandIteratedConditionalModes(ICM)algorithm.Finally,experimentsindicatethatthisalgorithmforthecloudimagedetectionhashigheraverageaccuracyratewhichisabout98.76%andtheaverageresultcanalsoreach96.92%fordifferenttypeofimagenoise.
简介:Regimeswitching,whichisdescribedbyaMarkovchain,isintroducedinaMarkovcopulamodel.Weprovethatthemarginals(X,H~i),i=1,2,3oftheMarkovcopulamodel(X,H)arestillMarkovprocessesandhavemartingaleproperty.Inthisproposedmodel,apricingformulaofcreditdefaultswap(CDS)withbilateralcounterpartyriskisderived.更多还原
简介:这篇论文为随机的中立Markov处理全球指数的稳定性问题跳有不明确的参数和多重时间延期的系统(MJS)。延期分别地被看作经常,时间变化盒子,和不确定性被假定是围住的标准。由选择适当Lyapunov-Krasovskii功能,它给足够的条件以便不明确的中立MJS是全球性为所有可被考虑的不确定性的指数地随机的联盟者马厩。稳定性标准在线性矩阵不平等(LMI)形式被提出,它能容易在实践被检查。最后,二个数字例子被利用说明发达技术的有效性。
简介:LetXbeanm-symmetricMarkovprocessandMamultiplicativefunctionalofXsuchthattheM-subprocessofXisalsom-symmetric.TheauthorcharacterizestheDirichletformassociatedwiththesubprocessintermsofthatassociatedwithXandthebivariateRevuzmeasureofM.
简介:TheauthorsestablishtheHilbertianinvarianceprinciplefortheempiricalprocessofastationaryMarkovprocess,byextendingtheforward-backwardmartingaledecompositionofLyons-Meyer-ZhengtotheHilbertspacevaluedadditivefunctionalsassociatedwithgeneralnon-reversibleMarkovprocesses.
简介:ThispaperdevelopsanewlowerboundmethodforPOMDPsthatapproximatestheupdateofabeliefbytheupdateofitsnon-zerostates.ItusestheunderlyingMDPtoexploretheoptimalreachablestatespacefrominitialbeliefandselectactionsduringvalueiterations,whichsignificantlyacceleratestheconvergencespeed.Also,analgorithmwhichcollectsandprunesbeliefpointsbasedontheupperandlowerboundsispresented,andexperimentalresultsshowthatitoutperformssomeofthestate-of-artpoint-basedalgorithms.