A segment-wise dynamic programming algorithm for BSDEs|  

作　　者：Christian Bender Steffen Meyer 

机构地区：[1]Department of Mathematics,Saarland University,Campus E 24,D-66123 Saarbrücken,Germany

出　　处：《Probability, Uncertainty and Quantitative Risk》2025年第1期103-134,共32页概率、不确定性与定量风险(英文)

摘　　要：We introduce and analyze a family of linear least-squares Monte Carlo schemesfor backward SDEs, which interpolate between the one-step dynamic programmingscheme of Lemor, Warin, and Gobet (Bernoulli, 2006) and the multi-step dynamicprogramming scheme of Gobet and Turkedjiev (Mathematics of Computation, 2016). Ouralgorithm approximates conditional expectations over segments of the time grid. Wediscuss the optimal choice of the segment length depending on the 'smoothness' of theproblem and show that, in typical situations, the complexity can be reduced compared tothe state-of-the-art multi-step dynamic programming scheme.

关 键 词：Backward stochastic differential equations Empirical regression Dynamic programming Monte Carlo methods 

分 类 号：O211[理学—概率论与数理统计]