Normalizing Flow (NF) is a special type of generative model that can offer invertible mapping, and has gained significant attention in the machine learning community. In contrast with the other generative models, NF it particularly useful for solving a variety of computational science problems. This talk will survey our recent progresses in applying NF to sampling of Boltzmann distributions in many-body systems, model reduction of complex stochastic dynamics, and modeling of stochastic fields. Those works indicate that NF has great potential to be a powerful tool in scientific computing beyond machine learning.
Hao Wu is an associate professor in Institute of Natural Science, Shanghai Jiao Tong University. He obtained his PhD in Computer Science from Tsinghua University in 2007, worked as a postdoc in Institute of Mathematics, Free University of Berlin from 2007 to 2018, and was a PI at Zuse Institute Berlin from 2017 to 2018. Before moving to Shanghai Jiao Tong University, he was a faculty member in the School of Mathematical Sciences at Tongji University. His main interests are in developing numerical and machine learning methods for modeling and analysis of simulation data of many-body systems.