The code used to generate the numerical insights in the paper "Optimal adaptive control with separable drift uncertainty" https://arxiv.org/abs/2309.07091 authored by Samuel Cohen, Christoph Knochenhauer and Alexander Merkel.
This repository contains a basic implementation of the Deep Galerkin Method for a stochastic optimal control problem in a 1+3 dimensional space. It includes the proposed DGM architecture and is used to solve the fully nonlinear HJB equation of Section 6 where a problem is numerically.
The "outputs" folder needs to be specified before use.