简介:P.M.Djuric,etc.(1992)researchedonthesegmentationofnonstationarystochasticprocessintopiecewisestationarystochasticprocessbyBayesiancriterion,andgaveadynamicequationaboutthenumberofsegments,theirboundariesandARmodelordersforeachsegment,butdidnotgivedetailedsolutionfortheequation.Becausethesolutionfortheequationisverycomplex,thispaperinvestigatesthesolution,derivessomerecursiverelations,simplifiestheproblem,savescomputationtimeandgoesfurtherintothesegmentationofnonstationarystochasticprocessintopiecewisestationarystochasticprocess.
简介:在这篇论文,作者首先作为噪音来源与Lévy过程学习二种随机的微分方程(SDE)。基于这些SDE和Lévy进程驾驶的多维的向后的随机的微分方程(BSDE)的解决方案的存在和唯一,作者继续学习一随机线性二次(LQ)有Lévy进程的最佳的控制问题,在状态和控制的费用weighting矩阵被允许不定的地方。包含平等和不平等限制的一个种新随机的Riccati方程从方形的结束的想法被导出,它的解决之可能性被证明为well-posedness和能具有州的反馈或LQ问题的开环的形式的最佳的控制的存在足够。而且,作者获得一些特殊情况的Riccati方程的答案的存在和唯一。最后,二个例子被举说明这些理论结果。
简介:Theaimofthispaperistostudythepracticalф0-stabilityinprobability(Pф0SiP)andpracticalф0-stabilityinpthmean(Pф0SpM)ofswitchedstochasticnonlinearsystems.Sufficientconditionsonsuchpracticalpropertiesareobtainedbyusingthecomparisonprincipleandthecone-valuedLyapunovfunctionmethods.Also,basedonanextendedcomparisonprinciple,aperturbationtheoryofswitchedstochasticsystemsisgiven.
简介:Itisknownthatnearlyuncoupledirreduciblestochasticmatricesmustpossesssub-dominanteigenvaluesnearλ=1.Itisnaturetoaskwhethertheconverseistrue.HortfielandMeyer[2]gaveapositiveanswer.TheyintroducedthenotionofuncouplingmeasureofStochasticmatrices.Forann×nstochasticmatrixPtheuncouplingmeasureofPisde-finedasσ(p)=min((sumfromi∈M1,j∈M1(Pij))+(sumfromi∈M1,j∈M1(Pij)),wheretheminimumistakenoverall
简介:Animportantfunctioningmechanismofbiologicalmacromoleculesisthetransitionbetweendifferentconformedstatesduetothermalfluctuation.Inthepresentpaper,abiologicalmacromoleculeismodeledastwostrandswithsidechainsfacingeachother,anditsstochasticdynamicsincludingthestatisticsofstationarymotionandthestatisticsofconformationaltransitionisstudiedbyusingthestochasticaveragingmethodforquasiHamiltoniansystems.ThetheoreticalresultsareconfirmedwiththeresultsfromMonteCarlosimulation.
简介:Brownian运动和泊松过程(BDSDEP)在随机的时间间隔上与non-Lipschitz系数驾驶的向后的二倍地随机的微分方程被学习。为quasilinear的一个类的答案的概率的解释随机的部分微分积分的方程(SPDIE)与BDSDEP被对待。在non-Lipschitz条件下面,BDSDEP的可测量的答案的存在和唯一结果经由变光滑的技术被建立。然后,为BDSDEP的答案的连续依赖被导出。最后,为quasilinearSPDIE的一个班的答案的概率的解释被给。
简介:Thispaperdiscussestherobustquadraticstabilizationcontrolproblemforstochasticuncertainsystems,wheretheuncertainmatrixisnormbounded,andtheexternaldisturbanceisastochasticprocess.Twokindsofcontrollersaredesigned,whichincludestatefeedbackcaseandoutputfeedbackcase.Theconditionsfortherobustquadraticstabilizationofstochasticuncertainsystemsaregivenvialinearmatrixinequalities.Thedetaileddesignmethodsarepresented.Numericalexamplesshowtheeffectivenessofourresults.
简介:
简介:Amaximumprincipleisprovedforsemilinearstochasticevolutionsystems.Themaincontributionofthisworkisthatinourproblem,theinfinitesimalgeneratorofthesemigroupofthesystemsneednottobeelliptic.ThisgeneralizesaresultofA.Bensoussanin1983.