GPU Accelerated MagnetoHydroDynamics for Astrophysics & Testing for Exascale Codes

Magnetic fields play a significant role in galaxy evolution because they have similar energy densities to the fluid. Magnetic fields would also enable anisotropic conduction and cosmic ray transport to be accurately simulated. Unfortunately, magnetohydrodynamical (MHD) simulations of galaxies are incredibly computationally expensive to run to the point where MHD simulations of entire galaxies have been impossible to perform at high resolution.

Cholla is a massively parallel, GPU accelerated, code for modeling astrophysical fluid dynamics. Cholla can harness the incredible power of new GPU accelerated supercomputers like Summit and Frontier to deliver simulations at unprecedented resolutions. My current work is to add magnetic fields to Cholla. Galaxies, and most other systems with plasma, contain significant magnetic fields that may affect their dynamics, so we need to accurately simulate both the magnetic fields and their interaction with matter. Magnetic fields present unique challenges to simulate since they must maintain nearly perfectly zero divergence or the simulation will produce incorrect results, we address this issue using the Constrained Transport algorithm which enforces zero divergence via translating magnetic fluxes to electric fields and then uses those to update the magnetic field.

A testing framework was required for Cholla as the complexity and number of contributors grew. I built a testing framework for Cholla using a mixture of GoogleTest and custom code. This framework works with both automated testing systems (Jenkins & GitHub Actions) and on any system that Cholla itself will run on, including the Summit supercomputer and testbed systems for Frontier. This new framework has enabled faster and more confident development of Cholla and is designed with exascale code in mind.