Modelling subsonic turbulence with SPH-EXA

Abstract: The numerical simulation of subsonic turbulence with SPH has traditionally been hindered by E0 errors, inaccurate gradient evaluations, and excessive dissipation. In this work, we present numerical results of SPH simulations of subsonic turbulence and compare to state-of-the-art codes such as AREPO and GIZMO. For the first time, SPH can reproduce the results of such codes, showing a similar interval of wavenumber in the inertial range of the Kolmogorov cascade in the subsonic regime. We use the SPH-EXA code to perform these simulations, a GPU-based state-of-the-art SPH code with high performance and extreme scalability at its core. SPH-EXA is coupled with a modern SPH solver based on methods such as an integral approach to gradient calculation, artificial viscosity switches that include a linear field cleaner, a flexible family of pairing-resistant interpolation kernels, generalized volume elements, and a controlled handling of density jumps which maximizes Lagrangian compatibility. In addition, it includes a novel and extremely scalable gravity solver for astrophysical applications.