Matching Algorithms for High-Dimensional Stochastic Optimal Control
Published in In Preparation, 2025
This work develops algorithms for high-dimensional stochastic optimal control, exploring connections between flow matching, score-based models, mean field games, and control theory. The project involves implementing and comparing multiple SOC solvers including adjoint matching, finite element methods, neural ODEs, and neural-FBSDE solvers.
Recommended citation: R. Leburu, L. Nurbekyan, L. Ruthotto, & G. Zhang. (in preparation). "Matching Algorithms for High-Dimensional Stochastic Optimal Control."