FBSJNN: A Theoretically Interpretable and Efficiently Deep Learning method for Solving Partial Integro-Differential Equations

TL;DR

This paper introduces FBSJNN, a deep learning framework for solving PIDEs and FBSDEJs that is both interpretable and efficient, utilizing a single neural network to improve accuracy and reduce complexity.

Contribution

The novel FBSJNN framework uses one neural network to approximate solutions and non-local integrals, with theoretical convergence guarantees and improved efficiency over existing methods.

Findings

Achieves relative errors around 10^{-3} in numerical experiments.

Provides theoretical convergence and error estimates.

Reduces parameter count compared to traditional approaches.

Abstract

We propose a novel framework for solving a class of Partial Integro-Differential Equations (PIDEs) and Forward-Backward Stochastic Differential Equations with Jumps (FBSDEJs) through a deep learning-based approach. This method, termed the Forward-Backward Stochastic Jump Neural Network (FBSJNN), is both theoretically interpretable and numerically effective. Theoretical analysis establishes the convergence of the numerical scheme and provides error estimates grounded in the universal approximation properties of neural networks. In comparison to existing methods, the key innovation of the FBSJNN framework is that it uses a single neural network to approximate both the solution of the PIDEs and the non-local integral, leveraging Taylor expansion for the latter. This enables the method to reduce the total number of parameters in FBSJNN, which enhances optimization efficiency. Numerical experiments indicate that the FBSJNN scheme can obtain numerical solutions with a relative error on the scale of $10^{-3}$.