Boltzmann Generators: Deep Learning of Thermodynamics and Efficient Monte Carlo

ORAL

Abstract

We introduce Boltzmann Generators, a machine learning method to efficiently sample Boltzmann distributions of complex multibody physics systems that give rise naturally to thermodynamic properties of the system, such as free energies, entropies and temperature dependence.

Presenters

  • Frank Noe

    Free University of Berlin

Authors

  • Frank Noe

    Free University of Berlin