Solving PDEs with Deep Learning

Monday, November 26, 2018 - 1:25pm - 2:25pm
Lind 305
Lexing Ying (Stanford University)
In this talk, I will discuss some recent work on using deep neutral networks in solving high dimensional PDE problems. Examples include molecular dynamics, density functional theory, and inverse scattering problems. In each case, we propose novel neural network architectures based on the physical properties and sparse structures of the problem under investigation.