Wednesday, 17th of October
14h30 (Shannon amphitheatre
, 660 building) (see location
(Institute of Theoretical Physics, Chinese Academy of Sciences)
Title: Solving Statistical Mechanics using Variational Autoregressive Networks
Statistical mechanics problems in physics concern about how to estimate free energy of the system, how to compute macroscopic properties of the system such as magnetizations and correlations, and how to sample from the Boltzmann distribution efficiently. In this talk I will introduce a general framework for solving such statistical physics problems using neural networks. The approach extends the celebrated variational mean-field approaches using autoregressive neural networks which support direct sampling and exact calculation of normalized probability of configurations. The network computes variational free energy, estimates physical quantities such as entropy, magnetizations and correlations, and generates uncorrelated samples all at once. Training of the network employs the policy gradient approach in reinforcement learning, which unbiasedly estimates the gradient of variational parameters. We apply our approach to several classical systems, including 2-d Ising models, Hopfield model, Sherrington--Kirkpatrick spin glasses, and the inverse Ising model, for demonstrating its advantages over exiting variational mean-field methods.
Contact: guillaume.charpiat at inria.fr
All TAU seminars: here