Guardado en:
Detalles Bibliográficos
Autores principales: Liang, Zongxia, Luo, Xiaodong, Yu, Xiang
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2506.22203
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910214853230592
author Liang, Zongxia
Luo, Xiaodong
Yu, Xiang
author_facet Liang, Zongxia
Luo, Xiaodong
Yu, Xiang
contents We develop a continuous-time reinforcement learning framework for a class of singular stochastic control problems without entropy regularization. The optimal singular control is characterized as the optimal singular control law, which is a pair of regions of time and the augmented states. The goal of learning is to identify such an optimal region via the trial-and-error procedure. In this context, we generalize the existing policy evaluation theories with regular controls to learn our optimal singular control law and develop a policy improvement theorem via the region iteration. To facilitate the model-free policy iteration procedure, we further introduce the zero-order and first-order q-functions arising from singular control problems and establish the martingale characterization for the pair of q-functions together with the value function. Based on our theoretical findings, some q-learning algorithms are devised accordingly and a numerical example based on simulation experiment is presented.
format Preprint
id arxiv_https___arxiv_org_abs_2506_22203
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Reinforcement Learning Framework for Some Singular Stochastic Control Problems
Liang, Zongxia
Luo, Xiaodong
Yu, Xiang
Optimization and Control
93E20, 93B47, 49K45
We develop a continuous-time reinforcement learning framework for a class of singular stochastic control problems without entropy regularization. The optimal singular control is characterized as the optimal singular control law, which is a pair of regions of time and the augmented states. The goal of learning is to identify such an optimal region via the trial-and-error procedure. In this context, we generalize the existing policy evaluation theories with regular controls to learn our optimal singular control law and develop a policy improvement theorem via the region iteration. To facilitate the model-free policy iteration procedure, we further introduce the zero-order and first-order q-functions arising from singular control problems and establish the martingale characterization for the pair of q-functions together with the value function. Based on our theoretical findings, some q-learning algorithms are devised accordingly and a numerical example based on simulation experiment is presented.
title A Reinforcement Learning Framework for Some Singular Stochastic Control Problems
topic Optimization and Control
93E20, 93B47, 49K45
url https://arxiv.org/abs/2506.22203