Regret Lower Bounds for Learning Linear Quadratic Gaussian Systems

• 成果类型: Article
• 署名作者: Ziemann, Ingvar; Sandberg, Henrik
• 署名单位: University of Pennsylvania; Royal Institute of Technology
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2024.3439132
• 发表日期: 2025
• 页码: 159-173
• 关键词: costs; observability; Adaptation models; CONTROLLABILITY; Adaptive control; uncertainty; State feedback; closed loop identification; fundamental limits; Statistical learning

摘要

In this article, we establish regret lower bounds for adaptively controlling an unknown linear Gaussian system with quadratic costs. We combine ideas from experiment design, estimation theory, and a perturbation bound of certain information matrices to derive regret lower bounds exhibiting scaling on the order of magnitude root T in the time horizon T . Our bounds accurately capture the role of control-theoretic parameters and we are able to show that systems that are hard to control are also hard to learn to control; when instantiated to state feedback systems we recover the dimensional dependency of earlier work but with improved scaling with system-theoretic constants, such as system costs and Gramians. Furthermore, we extend our results to a class of partially observed systems and demonstrate that systems with poor observability structure also are hard to learn to control.