{"id":"arxiv/2603.13155","url":"https://arxiv.org/abs/2603.13155","tags":["math.OC"],"score":0,"title":"Reinforcement Learning for Discounted and Ergodic Control of Diffusion Processes","domain":"math.OC","source":"arxiv","authors":["Erhan Bayraktar","Ali D. Kara","Somnath Pradhan","Serdar Yuksel"],"reviews":0,"subject":"math.OC","summary":"This paper develops a quantized Q-learning algorithm for the optimal control of controlled diffusion processes on $\\mathbb{R}^d$ under both discounted and ergodic (average) cost criteria. We first establish near-optimality of finite-state MDP approximations to discrete-time discretizations of the diffusion, then introduce a quantized Q-learning scheme and prove its almost-sure convergence to near-optimal policies for the finite MDP. These policies, when interpolated to continuous time, are shown to be near-optimal for the original diffusion model under discounted costs and -- via a…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-03-13","reactions":{}}