Convergence of Proximal Policy Gradient Method for Problems with Control Dependent Diffusion Coefficients

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Davey, Ashley, Zheng, Harry
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916756374683648
author Davey, Ashley
Zheng, Harry
author_facet Davey, Ashley
Zheng, Harry
contents We prove convergence of the proximal policy gradient method for a class of constrained stochastic control problems with control in both the drift and diffusion of the state process. The problem requires either the running or terminal cost to be strongly convex, but other terms may be non-convex. The inclusion of control-dependent diffusion introduces additional complexity in regularity analysis of the associated backward stochastic differential equation. We provide sufficient conditions under which the control iterates converge linearly to the optimal control, by deriving representations and estimates of solutions to the adjoint backward stochastic differential equations. We introduce numerical algorithms that implement this method using deep learning and ordinary differential equation based techniques. These approaches enable high accuracy and scalability for stochastic control problems in higher dimensions. We provide numerical examples to demonstrate the accuracy and validate the theoretical convergence guarantees of the algorithms.
format Preprint
id arxiv_https___arxiv_org_abs_2505_18379
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Convergence of Proximal Policy Gradient Method for Problems with Control Dependent Diffusion Coefficients
Davey, Ashley
Zheng, Harry
Optimization and Control
68Q25, 93E20, 49M05, 35C05, 65C30
We prove convergence of the proximal policy gradient method for a class of constrained stochastic control problems with control in both the drift and diffusion of the state process. The problem requires either the running or terminal cost to be strongly convex, but other terms may be non-convex. The inclusion of control-dependent diffusion introduces additional complexity in regularity analysis of the associated backward stochastic differential equation. We provide sufficient conditions under which the control iterates converge linearly to the optimal control, by deriving representations and estimates of solutions to the adjoint backward stochastic differential equations. We introduce numerical algorithms that implement this method using deep learning and ordinary differential equation based techniques. These approaches enable high accuracy and scalability for stochastic control problems in higher dimensions. We provide numerical examples to demonstrate the accuracy and validate the theoretical convergence guarantees of the algorithms.
title Convergence of Proximal Policy Gradient Method for Problems with Control Dependent Diffusion Coefficients
topic Optimization and Control
68Q25, 93E20, 49M05, 35C05, 65C30
url https://arxiv.org/abs/2505.18379