摘要
Thestateequationsofstochasticcontrolproblems,whicharecontrolledstochasticdifferentialequations,areproposedtobediscretizedbytheweakmidpointruleandpredictor-correctormethodsfortheMarkovchainapproximationapproach.Localconsistencyofthemethodsareproved.NumericaltestsonasimplifiedMerton'sportfoliomodelshowbettersimulationtofeedbackcontrolrulesbythesetwomethods,ascomparedwiththeweakEuler-MaruyamadiscretisationusedbyKrawczyk.ThissuggestsanewapproachofimprovingaccuracyofapproximatingMarkovchainsforstochasticcontrolproblems.
出版日期
2010年01月11日(中国期刊网平台首次上网日期,不代表论文的发表时间)