Description: In Part II we discuss some of the issues involved with physics informed machine learning, such as how to represent physical space and time using techniques such as dilated causal convolutional layers, and how to enforce physical laws in the loss function of the deep neural network. We step through a basic example of solving Burgers equation and show how to penalize the loss function using differential equation constraints according to the method of Raissi et al. (J. Comp. Phys., 2019).
Presenter(s): Dr. Wes Brewer - PET GDIT
Location: Webcast
Date & Time: March 31, 2020, 3:00p - 4:30p ET
Additional Notes: Part of the HPDA Seminar Series (2020).
Distribution Statement D. Distribution authorized to the Department of Defense and U.S. DoD contractors only, Administrative or Operational Use, 31 March 2020. Other requests for this document shall be referred to the High Performance Computing Modernization Office, 3909 Halls Ferry Road, Vicksburg, MS 39180.