In this talk I will detail my experience working at Los Alamos National Lab (LANL) this summer and explain how we tried to use neural networks to predict features of magnetohydrodynamics (MHD) simulations. MHD, while having a very unwieldy name, is more friendly than it perhaps originally seems! I will explain the basics of MHD and how we used encoder-decoder convolutional neural networks to approach tasks such as surrogate modeling. National Lab gods willing, I will have many pretty pictures.

Ari Bormanis will present his talk at 1:00, and speaker Andrew Arnold will present at 1:45.