Nico Hoffmann (HZDR): AI for Advanced Photon Sciences. A surrogate modelling perspective.

How to connect:

Please see the email announcement

Abstract:

Fast simulation techniques are becoming more and more important for advanced understanding of complex physical processes involved in photon sciences, e.g. What configuration(s) of our system are explaining certain experimental diagnostics of some Laser-Plasma accelerator? Modern numerical code typically requires discretization of the underlying equation, tailored implementations for distributed computing while parameter studies don’t take leverage on previous runs. We will be introducing state-of-the-art machine learning techniques for surrogate modeling & reduced order modeling that tackle any of the aforementioned challenges ranging from purely data-driven methods to physics-informed neural networks.