{"id":"arxiv/2610.03242","url":"https://arxiv.org/abs/2610.03242","tags":["math.OC"],"score":0,"title":"A Newton--Kleinman Iteration for Infinite-Dimensional Linear-Quadratic Optimal Control","domain":"math.OC","source":"arxiv","authors":["Timo Reis"],"reviews":0,"subject":"math.OC","summary":"We develop a Newton--Kleinman iteration for linear-quadratic optimal control of infinite-dimensional systems with possibly unbounded control and observation operators. In this setting, the classical algebraic Riccati equation need not be meaningful and therefore cannot serve as the basis of the iteration. We instead linearize a Lur'e identity formulated directly on the system-node domain. Starting from any stabilizing feedback, the resulting iteration is well defined, preserves stability, and yields a monotonically decreasing sequence of cost operators. A resolvent-based normalization makes…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-02","reactions":{}}