SciML Turbulence Simulation Surrogate Model

As part of final project for the Scientific Machine Learning (S&DS 689) class at Yale University, worked on a machine learning model to more efficiently simulate plasma turbulence.

Full Writeup/Paper: Here

Code Repository: Here

I specifically worked on the development and implementation of the convolutional variational autoencoder (CVAE) and the coupled dense neural network (DNN) in PyTorch—which are summarized in the final paper—and the training of these respective models.

We hope to continue our work and explore other multi-physics models with my specific focus being the exploration of using DeepM&Mnets.