Penalty-Uniform Localization for State-Constrained Policy Iteration

DOI10.48550/arXiv.2609.16829MaRDI QIDQ7345476arXivFDO

Authors Yeongjong Kim, Ji-Woong Jang, Yeoneung Kim, Jiwoong Jang

Publication date 15 September 2026

Published in arXiv

Tldr We study the trade-off between penalty parameter $\varepsilon$ and spatial discretization step $h$ for state-constrained optimal control in the deterministic discounted setting, showing that even with a small penalty, localization can be achieved as $O(h)$ in terms of occupation measure error, independent of $\varepsilon$, when $h\le\varepsilon$.

Full work available at URL https://doi.org/10.48550/arXiv.2609.16829

Mathematics Subject Classification ID

Viscosity solutions to Hamilton-Jacobi equations in optimal control and differential games (49L25) Discrete approximations in optimal control (49M25) Stability and convergence of numerical methods for boundary value problems involving PDEs (65N12) Nonlinear systems in control theory (93C10)

arXiv classification

math.NAmath.OC

This page was built for publication: Penalty-Uniform Localization for State-Constrained Policy Iteration

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7345476)