Events
Colloquium
Data-Driven Perspectives on First-Order Methods for Convex Optimization
 |
Speaker: Bartolomeo Stellato (Princeton) Assistant Professor at the Department of Operations Research and Financial Engineering at Princeton University Princeton University Wednesday, April 1, 2026 11:30AM - 1:00PM Lunch at 11:30am in 1307
Talk 12:00-1:00pm in 1327 Location: Yale Institute for Foundations of Data Science & Webcast, 219 Prospect Street, 13th Floor, New Haven, CT 06511 and via Webcast: https://yale.hosted.panopto.com/Panopto/Pages/Viewer.aspx?id=b817dfe8-7a86-4757-804e-b3ca014a0803 |
Abstract: First-order methods are widely used in large-scale convex optimization, yet providing sharp guarantees on their convergence behavior remains a key challenge. In many practical settings, the same optimization problem is solved repeatedly with varying parameters, naturally modeled as draws from an unknown distribution. In this talk, I present two complementary approaches that use the observed convergence on a limited number of problem instances to improve both the analysis and the design of first-order methods. The first combines the performance estimation problem (PEP) with Wasserstein distributionally robust optimization into a convex semidefinite program that produces probabilistic convergence bounds, unifying worst-case and average-case analysis. The second uses PAC-Bayes theory to learn algorithm parameters, such as step-sizes and warm-starts, with provable generalization guarantees rooted in the convergence properties of the underlying operators. Together, these lines of work connect classical tools from convex analysis and operator theory with ideas from statistical learning, offering both tighter performance certificates and a principled approach to algorithm tuning.
Speaker Bio: Bartolomeo Stellato is an Assistant Professor in the Department of Operations Research and Financial Engineering at Princeton University. Previously, he was a Postdoctoral Associate at the MIT Sloan School of Management and Operations Research Center. He holds a DPhil (PhD) in Engineering Science from the University of Oxford, a MSc in Robotics, Systems and Control from ETH Zürich, and a BSc in Automation Engineering from Politecnico di Milano. He developed OSQP, a widely used solver in mathematical optimization. His awards include a Sloan Research Fellowship, the 2024 Beale — Orchard-Hays Prize, an ONR Young Investigator Award, an NSF CAREER Award, the 2024 Princeton SEAS Howard B. Wentz Jr. Faculty Award, the 2022 Franco Strazzabosco Young Investigator Award from ISSNAF, a Princeton SEAS Innovation Award in Data Science, the 2021 Best Paper Award in Mathematical Programming Computation, and the 2018 First Place Prize Paper Award in IEEE Transactions on Power Electronics. His research focuses on data-driven computational tools for mathematical optimization, machine learning, and optimal control.
Add To: Google Calendar | Outlook | iCal File
- Colloquium
Submit an Event
Interested in creating your own event, or have an event to share? Please fill the form if you’d like to send us an event you’d like to have added to the calendar.