Tag
This paper explores the deep connection between reinforcement learning and potential theory, viewing core RL representations and algorithms through Laplace's, Poisson's, and diffusion equations to potentially enhance sample efficiency and formal constraints.