Laurent Lessard, 29 Oct 2014: Difference between revisions
(Created page with "Laurent Lessard, a postdoc at UC Berkeley, will be giving an information CDS seminar on 29 Oct (Wed). He is available to meet with students and faculty during the day. === S...") |
No edit summary |
||
| (4 intermediate revisions by 3 users not shown) | |||
| Line 3: | Line 3: | ||
=== Schedule === | === Schedule === | ||
* 10:00 - John Doyle | * 10:00 - John Doyle, 210 Annenberg | ||
* 11:00 - Seminar, 121 | * 11:00 - Seminar, 121 ANB | ||
* 12:00 - Lunch with faculty (Doyle | * 12:00 - Lunch with faculty (Doyle + ?) | ||
* 1:30 - Open | * 1:30 - Open | ||
* 2:15 - | * 2:15 - Joel Tropp, ANB 307 | ||
* 3:00 - CDS tea | * 3:00 - CDS tea | ||
* 3:30 - | * 3:30 - Adam Wierman, ANB 215 | ||
* 4:15 - Open | * 4:15 - Open | ||
* 5:00 - Done for the day | * 5:00 - Done for the day | ||
Latest revision as of 02:03, 22 October 2014
Laurent Lessard, a postdoc at UC Berkeley, will be giving an information CDS seminar on 29 Oct (Wed). He is available to meet with students and faculty during the day.
Schedule
- 10:00 - John Doyle, 210 Annenberg
- 11:00 - Seminar, 121 ANB
- 12:00 - Lunch with faculty (Doyle + ?)
- 1:30 - Open
- 2:15 - Joel Tropp, ANB 307
- 3:00 - CDS tea
- 3:30 - Adam Wierman, ANB 215
- 4:15 - Open
- 5:00 - Done for the day
(If needed, Laurent is also available on the 30th.)
Talk abstract
Analysis and Design of Optimization Algorithms via Integral Quadratic Constraints
Laurent Lessard, UC Berkeley
ABSTRACT: I will present a new method to analyze and design iterative optimization algorithms, built on the framework of Integral Quadratic Constraints (IQC) from robust control theory. IQCs provide sufficient conditions for the stability of complicated interconnected systems, and these conditions can be checked by semidefinite programming. I will discuss how to adapt IQC theory to study optimization algorithms, proving new inequalities about convex functions. Using these inequalities, I will derive upper bounds on convergence rates for the gradient method, the heavy-ball method, Nesterov's accelerated method, and related variants by solving small, simple semidefinite programs. I will close with a discussion of how these techniques can be used to search for algorithms with desired performance characteristics, establishing a new methodology for algorithm design.
BIO: Laurent Lessard received the BASc in Engineering Science at the University of Toronto and the MS and PhD degrees in Aeronautics and Astronautics from Stanford University. He is currently a postdoctoral scholar in the Berkeley Center for Control and Identification at the University of California, Berkeley. Before that, he was a LCCC postdoc in the Department of Automatic Control at Lund University. His research interests lie at the intersection of optimization and control, with an emphasis on complex or decentralized systems. Dr. Lessard received the 2013 American Automatic Control Council O. Hugo Schuck Best Paper Award.
