Last week I attended the excellent workshop of tutorials on “Understanding Many-Particle Systems with Machine Learning.” It’s part of a longer program on “Understanding Many-Particle Systems with Machine Learning.” In general, I think this is a fascinating subject. Modern computational many-body physics comes with heaps of approximations and functionals derived from heuristics or data–exactly the sort of space where modern machine learning can make a difference. The workshop was a good intro to the subject (although most talks were either ‘many body physics’ or ‘machine learning’–looking forward to future workshops where hopefully more of the talks will combine the disciplines)–IPAM tends to post slides and videos of the lectures a few weeks after.

Matthias Rupp gave a very good introduction to the subject. The paper version of the lecture topics can be found here; it’s a very good read. The paper also comes with a sample dataset for predicting atomization energies–looking forward to playing around with this!