DeepAI AI Chat
Log In Sign Up

Neural network architectures using min plus algebra for solving certain high dimensional optimal control problems and Hamilton-Jacobi PDEs

by   Jérôme Darbon, et al.

Solving high dimensional optimal control problems and corresponding Hamilton-Jacobi PDEs are important but challenging problems in control engineering. In this paper, we propose two abstract neural network architectures which respectively represent the value function and the state feedback characterisation of the optimal control for certain class of high dimensional optimal control problems. We provide the mathematical analysis for the two abstract architectures. We also show several numerical results computed using the deep neural network implementations of these abstract architectures. This work paves the way to leverage efficient dedicated hardware designed for neural networks to solve high dimensional optimal control problems and Hamilton-Jacobi PDEs.


page 20

page 21

page 25

page 30


Exploring Deep and Recurrent Architectures for Optimal Control

Sophisticated multilayer neural networks have achieved state of the art ...

Imbedding Deep Neural Networks

Continuous depth neural networks, such as Neural ODEs, have refashioned ...

Machine Learning for semi linear PDEs

Recent machine learning algorithms dedicated to solving semi-linear PDEs...

A control method for solving high-dimensional Hamiltonian systems through deep neural networks

In this paper, we mainly focus on solving high-dimensional stochastic Ha...

Lessons from AlphaZero for Optimal, Model Predictive, and Adaptive Control

In this paper we aim to provide analysis and insights (often based on vi...