Deep Learning and Computations of PDEs

Abstract: Deep neural networks are rapidly emerging as a key tool in en-hancing computations of PDEs. In this talk, we survey recent developmentsin my research group on two related themes. First, we use supervised learningto approximate observables for PDEs efficiently with applications to UQ andPDE constrained optimization. Second, we discuss recently introduced physicsinformed neural networks (PINNs) as an efficient discretization tool for bothforward as well as inverse problems for certain classes of PDEs.

analysis of PDEsdynamical systemsfunctional analysisoptimization and controlspectral theory

Audience: general audience

Webinar on PDE and related areas

Series comments: The webinar is organised jointly from IIT-Kanpur, TIFR-CAM,Bangalore, IISER-Pune and IISER-Kolkata.

Zoom Link for the talk will be available on iitk.ac.in/math/weekly-webinar-on-pde-and-related-areas

If you would like to receive webinar announcements(including zoom ID and password to attend the talk), please send a blank email with the subject line "WWOPARA_sub" to pdewebinar@iitk.ac.in. In order to unsubscribe, please send us a blank email with the subject "WWOPARA_unsub".