{"id":"arxiv/2605.02653","url":"https://arxiv.org/abs/2605.02653","tags":["math.OC"],"score":0,"title":"Mirror Descent for Deterministic Optimal Control","domain":"math.OC","source":"arxiv","authors":["Ye Feng","Jianfeng Lu"],"reviews":0,"subject":"math.OC","summary":"We study an explicit mirror-descent method for finite-horizon deterministic optimal control problems. The method is motivated by Pontryagin's maximum principle: at each iteration, one solves the state and adjoint equations and updates the control by maximizing a first-order approximation of the regularized Hamiltonian penalized by a Bregman divergence. In the Euclidean case, the update reduces to a projected gradient step in the control variable. Under global smoothness assumptions and uniform convexity of the mirror map, we prove a relative smoothness estimate for the cost functional and…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-05-04","reactions":{}}