Applied Math Colloquium: Benjamin Grimmer (JHU)
Location
Mathematics/Psychology : 104
Date & Time
October 10, 2025, 12:00 pm – 1:00 pm
Description
Title: Beyond Minimax Optimality: A Subgame Perfect Gradient Method
Abstract: This talk will take up
the task of designing the provably best possible gradient method for
smooth convex optimization. Methods with big-O optimal worst-case
guarantees were (famously) discovered
in the 80s by Nesterov. Methods with exactly minimax optimal worst-case
guarantees were developed in the last decade. We will present a
"subgame perfect" method that is not only optimal against a worst-case
problem instance but also optimally leverages all
gradient information revealed at each step. This corresponds to being
dynamically minimax optimal, or in game theory terms, provides us with a
subgame perfect strategy for optimization. Besides attaining this high
standard (beyond minimax optimality), our
subgame perfect gradient method is also very fast at solving actual
problems.
Bio: Ben Grimmer is an
assistant professor of applied mathematics and statistics at Johns
Hopkins University, supported by AFOSR, NSF, and as a Sloan Fellow.
Prior to joining Hopkins, Ben did his PhD at Cornell,
advised by James Renegar and Damek Davis, spending a couple of
semesters with Google Research and Simons. Ben's work primarily focuses
on novel methods for the design and analysis of first-order methods,
recently receiving the INFORMS Optimization Societies
Young Researcher Prize. Some of his recent computer-assisted works have
received substantial interest, being featured in popular mathematics
venues like Quanta.
Tags: