Abstract

I present two research threads applying deep learning to Monte Carlo and quantum mechanics.

Firstly, I discuss Fermionic neural networks and quantum Monte Carlo. This part of the talk will be largely based on [1]. Since the paper is a few years old and has led to quite a bit of follow up work, I will try to offer a perspective both on what it was like to get it working and also comment with the benefit of hindsight.

Second, I will discuss a thread of work accelerating well established Monte Carlo sampling approaches with methods from machine learning. I will use sampling of lattice field theories as a motivating physical example. This part of the talk will be based on [2] with some of [3] if there is time.

[1] Ab initio solution of the many-electron Schrödinger equation with deep neural networks, David Pfau, James S. Spencer, Alexander G. D. G. Matthews, and W. M. C. Foulkes, Phys. Rev. Research. 2020. [2] Continual Repeated Annealed Flow Transport Monte Carlo, Alexander G D G Matthews, Michael Arbel, Danilo Jimenez Rezende, Arnaud Doucet, International Conference on Machine Learning (ICML), 2022. [3] Score-based diffusion meets Annealed Importance Sampling. Arnaud Doucet, Will Grathwohl, Alexander G. D. G. Matthews & Heiko Strathmann, Neural Information Processing Systems (NeurIPS), 2022.