{"id":2417,"date":"2025-02-20T12:42:08","date_gmt":"2025-02-20T12:42:08","guid":{"rendered":"https:\/\/meyn.ece.ufl.edu\/?page_id=2417"},"modified":"2026-03-17T12:13:38","modified_gmt":"2026-03-17T17:13:38","slug":"representation-based-reinforcement-learning-and-control-for-dynamical-systems","status":"publish","type":"page","link":"https:\/\/faculty.eng.ufl.edu\/meyn\/c3\/c3-9\/representation-based-reinforcement-learning-and-control-for-dynamical-systems\/","title":{"rendered":"Online Optimization-based Control via Two-Time-Scale Dynamics"},"content":{"rendered":"\n

Emiliano Dall\u2019Anese<\/a> (Boston University)<\/h3>\n\n\n\n

<\/strong><\/p>\n\n\n\n

Abstract<\/strong>: Optimization-based control laws, arising in model predictive control, feedback optimization, and control Lyapunov and barrier function frameworks, induce closed-loop dynamics that are most naturally described as interconnections between continuous-time physical systems and discrete-time algorithmic iterations. In practice, these interconnections are implemented through sampling and zero-order holds, and rely on iterative solvers that may be terminated after a finite number of steps due to computational bottlenecks. This structure fundamentally departs from the assumptions under which most optimization-based controllers are analyzed.<\/p>\n\n\n\n

Motivated by this, the first part of this talk studies the stability of interconnected continuous-time (CT) systems \u2013 modeling the physical plant \u2013 and discrete-time (DT) systems \u2013 modeling iterative optimization algorithms. A continuous-time \u201creduced model\u201d is introduced to capture the idealized closed-loop behavior induced by exact or infinitely fast optimization. A two-time-scale CT-DT model is then presented to explicitly represent sampling and a finite number of algorithmic iterations. Under a contractivity assumption on the reduced model, the main result is that there exists a sampling time threshold below which the resulting CT\u2013DT interconnection is exponentially stable, the threshold is parameterized by the number of solver iterations, including the case of a single iteration. The analysis yields explicit bounds and enables a comparison with classical small-gain\u2013type conditions. The second part of the talk addresses the challenging case in which the reduced model is not globally contractive. Rather than asymptotic stability, trajectory-based bounds are presented to characterize the difference between ideal and implementable closed-loop behaviors, providing insight into the performance and robustness of two-time-scale optimization-based controllers beyond the contractive regime. The technical findings are illustrated through examples of single-iteration model predictive control and safety filters. <\/strong> <\/p>\n\n\n

\n
\"\"<\/figure>\n<\/div>\n\n\n

Speaker Bio: <\/strong>Emiliano Dall\u2019Anese is an Associate Professor in the Department of Electrical and Computer Engineering at Boston University, where he is also the Associate Head of the Division of Systems Engineering. He received the Ph.D. in Information Engineering from the Department of Information Engineering, University of Padova, Italy, in 2011. He was with the University of Minnesota as a postdoc (2011-2014), the National Renewable Energy Laboratory as a senior researcher (2014-2018), and the Department of Electrical, Computer, and Energy Engineering at the University of Colorado Boulder as a faculty (2018-2024). His research interests span the broad areas of optimization, control, and systems theory; current applications include power systems and autonomous systems. He received the National Science Foundation CAREER Award in 2020, the IEEE PES Prize Paper Award in 2021, the IEEE Transactions on Control of Network Systems Best Paper Award in 2023, the IEEE PES ISGT Europe best paper award in 2024, and the Outstanding Associate Editor recognition from IEEE LCSS in 2025. <\/p>\n","protected":false},"excerpt":{"rendered":"

Emiliano Dall\u2019Anese (Boston University) Abstract: Optimization-based control laws, arising in model predictive control, feedback optimization, and control Lyapunov and barrier function frameworks, induce closed-loop dynamics that are most naturally described as interconnections between continuous-time physical systems and discrete-time algorithmic iterations. In practice, these interconnections are implemented through sampling and zero-order holds, and rely on iterative […]<\/p>\n","protected":false},"author":1347,"featured_media":0,"parent":2631,"menu_order":3,"comment_status":"closed","ping_status":"closed","template":"page-templates\/page-section-nav.php","meta":{"_acf_changed":false,"inline_featured_image":false,"featured_post":"","footnotes":"","_links_to":"","_links_to_target":""},"class_list":["post-2417","page","type-page","status-publish","hentry"],"acf":[],"_links":{"self":[{"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/pages\/2417","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/users\/1347"}],"replies":[{"embeddable":true,"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/comments?post=2417"}],"version-history":[{"count":2,"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/pages\/2417\/revisions"}],"predecessor-version":[{"id":2717,"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/pages\/2417\/revisions\/2717"}],"up":[{"embeddable":true,"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/pages\/2631"}],"wp:attachment":[{"href":"https:\/\/faculty.eng.ufl.edu\/meyn\/wp-json\/wp\/v2\/media?parent=2417"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}