iSMART: An Iterative Sampling-and-Regression Technique for Solving Martingale-Based PDEs

DOI10.48550/arXiv.2607.29470MaRDI QIDQ7284535arXivFDO

Authors Fugui Ma, Tiejun Li, Xiaoguang Li

Publication date 31 July 2026

Published in arXiv

Tldr We propose a novel iterative sampling-and-regression technique (iSMART) to solve high-dimensional martingale-based PDEs with accurate results, bypassing adversarial optimization or nested expectation estimation. iSMART leverages standard SDE path simulation and squared-error loss minimization within an iterative framework that accommodates linear, semi-linear, and fully nonlinear equations with improved convergence behavior for HJB equations.

Full work available at URL https://doi.org/10.48550/arXiv.2607.29470

arXiv classification

math.NA

Cites work

• A deep solver for BSDEs with jumps
• A proof that artificial neural networks overcome the curse of dimensionality in the numerical approximation of Black-Scholes partial differential equations
• Adapted solution of a backward stochastic differential equation
• Convergence of the deep BSDE method for coupled FBSDEs
• Deep backward schemes for high-dimensional nonlinear PDEs
• Deep splitting method for parabolic PDEs
• DeepMartNet -- A Martingale based Deep Neural Network Learning Algorithm for Eigenvalue/BVP Problems and Optimal Stochastic Controls
• DeepMartNet -- A Martingale Based Deep Neural Network Learning Method for Dirichlet BVPs and Eigenvalue Problems of Elliptic PDEs in R^d
• FBSDE based Neural Network Algorithms for High-Dimensional Quasilinear Parabolic PDEs
• Fourier Neural Operator for Parametric Partial Differential Equations
• Machine Learning and Data Sciences for Financial Markets: A Guide to Contemporary Practices Machine Learning and Data Sciences for Financial Markets: A Guide to Contemporary Practices , by Agostino Capponi and Charles-Albert
• MscaleFNO: multi-scale Fourier neural operator learning for oscillatory functions and wave scattering problems
• Solving forward-backward stochastic differential equations explicitly -- a four step scheme
• Solving high-dimensional partial differential equations using deep learning
• The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
• Weak adversarial networks for high-dimensional partial differential equations

This page was built for publication: iSMART: An Iterative Sampling-and-Regression Technique for Solving Martingale-Based PDEs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7284535)