Scalable simulations of 3D LTSTMR with a novel plasma statistical physics algorithm

Abstract:

Magnetic reconnection (MR) is recognized to be a key driver in controlling behavior of the solar-earth system (e.g., flare, solar wind, aurora), which involves complex multi temporal-spatial scale and fully coupled magnetic field & plasma motion (B & U ) physical processes. Substantial research efforts have shown that there are many kinetic scale fractal and turbulence magnetic structures. We define these typical 3D continuous kinetic-dynamic-hydro (KDH) fully coupled MR on the solar-terrestrial space environment as 3D large temporal-spatial scale turbulent MR (3D LTSTMR). Until now, there are still many uncertainties (e.g., turbulent acceleration, turbulence by plasmas motion and magnetic field collective interaction) due to the limitation of current observation technologies and the incompleteness of present theoretical system. Numerical simulation, as an independent way from theory and observations, provides an important means for exploring the above phenomena. While conventional simulation methods, based on directly solving magnetohydrodynamics partial differential equation (e.g., ATHENA, NIRVANA, ZEUS, FLASH), can not attain the extreme range of scale of the 3D LTSTMR in the complex solar-earth system. Here we present a parallel lattice Boltzmann algorithm based on plasma statistical physics to describe the continuous features of plasma from macro flow scale to micro particle scale.With the scalable and robust simulation up to 100,000 cores on Tianhe-2 supercomputer, the largest simulation ever run, we analyze the basic features of the turbulence in the magnetic fluctuation-induced self-generating-organization (MF-ISGO), the turbulence in the plasma turbulence-induced self-feeding-sustaining (PT-ISFS), and the interaction of turbulence between MF-ISGO and PT-ISFS in the 3D LTSTMR for the first time, and all results agree well with latest observational data and theories.