Machine learning for fusion simulations – predicting energy and entropy

What is it about?

Machine learning can be used for computer simulations of fusion experiments. We present a neural network based model to provide information about energy, entropy, and pressure based on temperature and density. We find that adding the material's phase information (e.g., solid, liquid) improves prediction quality.

Featured Image

Photo by Joan Gamell on Unsplash

Photo by Joan Gamell on Unsplash

Why is it important?

Machine learning offers opportunities to speed up and unify many models used in inertial confinement fusion simulations. We demonstrate that machine learning can perform well on this type of prediction task, a promising sign for future developments.