Deep Learning Subgrid Models of Magnetohydrodynamic Turbulence

We explore the suitability of deep learning to capture the physics of subgrid-scale ideal magnetohydrodynamics turbulence of simulations of the magnetized Kelvin-Helmholtz instability (KHI). We produce simulations at different resolutions to systematically quantify the performance of neural network models to reproduce the physics of these complex simulations. We then implement these deep learning models in low resolution KHI simulations to examine their ability to reproduce the magnetic field amplification observed at high resolutions. We discuss the feasibility of using such models to reproduce the physics of magnetohydrodynamics turbulence in numerical relativity simulations of binary neutron star mergers.