Machine Learning for Symmetries and Conservation Laws

by 子鸣 (Ziming) 刘 (Liu) (MIT)

Wednesday 27 Oct 2021, 10:00 → 11:00 Asia/Shanghai

Online (West Building)

https://cern.zoom.us/j/390597013?pwd=cUlSVGltUnVjd04vQjZ1bnpVVEE2UT09 meeting ID: 390-597-013 pwd:446459 Although machine learning and artificial intelligence facilitate data analysis, speed up simulations and achieve good performance on various applications, it is largely unexplored whether machine learning can deepen our understanding of physics, just as physicists do. In this talk, I will demonstrate that machine learning can auto-discover symmetries, conservation laws, non-conservation and can exploit many more simplifying properties via physics-augmented learning. Ziming Liu is a second-year PhD student at Massachusetts Institute of Technology (MIT) and Institute for Artificial Intelligence and Fundamental Interactions (IAIFI), advised by Prof. Max Tegmark. Before that, he received his bachelor’s degree in physics from Peking University. His research interests lie on the intersection of AI/ML and physics in general, including (not limited to): (1) AI for physics: automating scientific discoveries using AI; (2) physics for AI: use toolkits from physics to better analyze and understand AI.

Slides ziming_talk_PKU.pdf