Physics-Informed LeaRnIng for Multiscale Systems (PILgRIMS)

The DARPA Physics of Artificial Intelligence ( PAI ) program is part of a broad initiative by DARPA to develop robust, interpretable machine learning frameworks that incorporate physical principles to solve challenging, large scale problems in applied math, science, engineeering. At the Predictive Science Lab, this involves:

  1. Developing deep learning models to quantify uncertainty in complex high-dimensional systems described by stochastic partial differential equations.

  2. Leveraging the properties of Lie groups to learn potentially chaotic dynamical systems.

  3. Learning the dynamics of complex flows through assimilation of particle image velocimetry data with deep neural networks.

Physics-informed machine learning solving ultra-high-dimensional stochastic partial differentions without the need of a finite element simulator. From Karumuri et al. 2019.


Funding Source:

Dr. Rohit Tripathy

Rohit received his Bachelors degree in Mechanical Engineering from VIT University in Vellore, India and is currently a Ph.D. candidate at the Predictive Science Lab. His research interests are in uncertainty quantification, physics-informed machine learning and probabilistic numerics.