并行montecarlo方法研究及其在期权定价中的应用-research on parallel monte carlo method and its application in option pricing.docxVIP

郑州大学硕士学位论文并行MonteCarlo方法研究及其在期权定价中的应用AbstractFinancialanalysisistheimportantapplicationofcalculationscienceatpresent．Itreceivesthemoreandmorewidespreadattention．Butalongwiththetechnicaldevelopment,thefinancialanalysisproposedthemoreandmorecomplexrandomquestion,80itisverydifficultofevenimpossibletogiveaapproximateresdtwiththedeft／ritemethod．TheMonteCarlomethodisthemostcommonlyusedmethodtothefinancialanalysis,sometimesevenistheonlymethod．Atthe$anlotimeitisaveryeffectivespecialnumericalanalysismethodforEuropeanoptionpricing．However,duetothehugenumberofexperiments，sometimesonmillionsofsimulation,thereisaconsiderableamountofcomputationontheMonteCarlomethodforalleffectivepricingprocess．En锄ol玛computationslcosthasserioushinderedtheapplicationoftheMonteCarlomethod,thereforeitisurgentneedstosolvetheproblemofmajorcomputationquantity．Theparallelmachineprovidedtheparallelcomputingmethodwhichsolveslargeamountofcomputationandlongtimeforcalculatingproblem．ThroughthestudyofthebasicprincipleoftheMonteCarlomethodandthespecialtyoftheEuropeanoptionpricing,thisarticlecarriedonthesimulationwilhMonteCarlomethodforEuropeanoptionpricing．InviewofthecomplexityofthefinancialanalysisandthehugecomputationqtlantityquestionofMonteCarlomethod,thisarticleusedthepsendo-randomnumbersoflognormaldistributioninsteadofrandomnumbers，handedoverthetmgenumberofpseudo—randomtothecomputercompleted．ThisarticleproposedtheparallelMonteCarlomethodtosovletheproblem，basedonthedeepanalysisofthesimulationprocessandthecharacteristicsofparallelcomputation．ThroughtheresearchandanalysisofcommunicationtimebetweenclusternodesinCOW，theparallelalgorithmhasebeenimprovedagainandagain．Thebestparallelalgorithmusedmaster-slaveII宴堕垡兰翌型兰坠型笙l一一兰堑翌竺塑!型!查鲨墅塞墨苎童塑壑塞丝主堕壁旦programmingmodel，realizedload-balancing,improvedtheutilizationrateofpfD。日so$incow,effectivelysolvedtheproblemthattherunningtimewaslongoneno曲0璐computationforserialalgorithmsThroughtheconfirmation,theparallelalgorithmobtainedahighspeedupandparallelefficiency,∞llancedtlleeffic