Eric Vanden-Eijnden, New York University

Learning to sample better
Date
Nov 1, 2022, 4:30 pm5:30 pm
Location
101 - Sherrerd Hall

Details

Event Description

Abstract: Sampling high-dimensional probability distributions is a common task in computational chemistry, Bayesian inference, etc. Markov Chain Monte Carlo (MCMC) is the method of choice to perform these calculations, but it is often plagued by slow convergence properties. I will discuss how methods from deep learning (DL) can help enhance the performance of MCMC via a feedback loop in which we simultaneously use DL to learn better samplers based e.g. on  generative models such as normalizing flows, and MCMC to obtain the data for the training of these models. I will draw connection between these methods and score-based diffusion models that have proven successful for image generation. I will also illustrate these techniques via several examples, including the calculation of free energies and Bayes factors. 

Short bio: Eric Vanden-Eijnden is a Professor of Mathematics at the Courant Institute of Mathematical Sciences, New York University.  His research focuses on the mathematical and computational aspects of statistical mechanics, with applications to complex dynamical systems arising in molecular dynamics,  materials science, atmosphere-ocean science, fluids dynamics, and neural networks. He is also interested in the mathematical foundations of machine learning (ML) and the applications of ML in scientific computing. He is known for the development and analysis of multiscale numerical methods for systems whose dynamics span a wide range of spatio-temporal scales. He is the winner of the Germund Dahlquist Prize and the J.D. Crawford Prize, and a recipient of the Vannevar Bush Faculty Fellowship.  He was a plenary speaker at the 2015 International Congress of Industrial and Applied Mathematics (ICIAM) in Beijing and an invited speaker at the 2022 International Congress of Mathematics (ICM).

Event Category
ORFE Department Colloquia