« DIMACS Workshop on Entropy and Optimization
DIMACS Workshop on Entropy and Optimization
May 18, 2022 - May 19, 2022
Location:
Online Event
Organizer(s):
Mohit Singh, Georgia Institute of Technology
Nisheeth Vishnoi, EPFL
Entropy-maximizing probability distributions subject to observed marginal constraints have been central to many scientific disciplines since the work of Boltzmann. These distributions, over both discrete and continuous domains, have recently played an important role in both computer science and mathematics. In particular, entropy-maximizing probability distributions have been used in interesting ways in applications such as algorithms for counting problems, algorithms for NP-hard optimization problems, and interior-point methods for convex programming. There has also been a history of using entropy-maximizing distributions in combinatorics, Markov chains, geometric functional analysis, and learning.
The goal of this workshop is to understand these connections systematically and explore further possibilities by bringing together researchers from computer science and mathematics.
Wednesday, May 18, 2022
Welcome from Organizers
Mohit Singh, Georgia Institute of Technology
Nisheeth Vishnoi, EPFL
Optimal Mixing of Glauber Dynamics for Spin Systems via Spectral Independence
Zongchen Chen, Massachusetts Institute of Technology
Entropic Independence: Optimal Sampling from Combinatorial Distributions
Nima Anari, Stanford University
Localization Schemes: A Framework for the Analysis of Sampling Algorithms
Ronen Eldan, Weizmann Institute of Science
Break + Q & A/Discussion
Statistically Near-Optimal Hypothesis Selection
Mark Braverman, Princeton University
Evolving Entropic Regularization
Sebastien Bubeck, Microsoft Research
Q & A/Discussion
Open Problem Session
Thursday, May 19, 2022
Many Facets of Bethe Approximation
Péter Csikvári, Alfréd Rényi Institute of Mathematics
A Quick Estimate for the Volume of a Polyhedron
Alexander Barvinok, University of Michigan
On the Maximum-entropy Sampling Problem
Jon Lee, University of Michigan
Break + Q & A/Discussion
Max Entropy Distributions in (Network) Optimization, Counting and Probability
Shayan Oveis-Gharan, University of Washington
Strong Expansion of Sparse Distributions
Tali Kaufman, Bar-Ilan University
Q & A/Discussion
Open Problem Session
Presentations at the workshop are by invitation, but we invite all registered participants to contribute topics for the Open Problem sessions that will take place at the end of each day.
The workshop is being held online and is open to all who register. Please register using the link at the bottom of the page. Registration is required for admission to the online venue. Please allow some time for your registration to be processed (which will only occur during business hours). We will invite you to contribute to the open problem session once you are registered.
Once your registration is processed, you can click here (or the picture) to enter the virtual venue. You will be taken to a Landing Page that contains information on the virtual venue, how to navigate within it, and how to get help during the event. You will enter the event from the Landing Page. Please feel free to review this Participant Guide prepared by our event hosts at Virtual Chair especially for this event.
Presented in association with the Special Focus on Bridging Continuous and Discrete Optimization.