Accelerating the simulation of kinetic shear Alfvén waves with a dynamical low-rank approximation.

We propose a dynamical low-rank algorithm for a gyrokinetic model that is used to describe strongly magnetized plasmas. The low-rank approximation is based on a decomposition into variables parallel and perpendicular to the magnetic field, as suggested by the physics of the underlying problem. We show that the resulting scheme exactly recovers the dispersion relation even with rank 1. We then perform a simulation of kinetic shear Alfvén waves and show that using the proposed dynamical low-rank algorithm a drastic reduction (multiple orders of magnitude) in both computational time and memory consumption can be achieved. We also compare the performance of robust first and second-order projector splitting, BUG (also called unconventional), and augmented BUG integrators as well as a FFT-based spectral and Lax–Wendroff discretization. • First dynamical low-rank framework for strongly magnetized plasmas. • Drastic reduction in the memory and computational cost due to a decomposition motivated by the underlying physical problem. • We show that exact dispersion relation is recovered with rank 1. • We provide a comparison of novel robust integrators and two space/time discretization strategies. [ABSTRACT FROM AUTHOR]