Speaker
Description
RHybrid (paRallel Hybrid) is a highly parallel hybrid space plasma simulation platform based on the macroscopic particle cloud-in-cell (CIC) technique for kinetic ions and the staggered Cartesian mesh grid method for electric and magnetic fields and charge-neutralizing fluid electrons. RHybrid has been used extensively for interpretation of spacecraft observations from plasma environments of Mercury, Venus, Mars and in studies of other solar system objects as well as extrasolar objects. Typically, RHybrid is applied to the solar wind interactions (space weather) of planets, where the size of the obstacle to the solar wind is smaller than Earth's magnetosphere.
Contrary to its non-parallel predecessor the HYB model, RHybrid does not yet support spatial adaptivity - adaptive mesh refinement (AMR) - nor temporal adaptivity i.e. substepping. In AMR, spatial resolution is increased (grid cell size decreased) in selected regions in order to better capture relevant phenomena while lowering computational requirements in other regions with less resolution. Temporal substepping also lowers computational requirements by solving certain regions only when necessary instead of solving everything whenever a solution is required anywhere in the simulated volume. AMR and substepping can each potentially reduce the time to solution by an order of magnitude, hence implementing both could allow running e.g. 100 times larger simulations than before. We present our progress towards a novel hybrid particle-in-cell/CIC model via adding AMR and substepping to RHybrid.
We discuss existing algorithms for AMR and substepping and our contributions as well as show various tests in 1, 2 and 3 dimensions in order to verify the new code. This development enables us to prepare for the interpretation of upcoming observations from the BepiColombo mission and to resolve Mercury's space weather self-consistently with ion kinetic effects at an unprecedented accuracy in terms of both the spatial resolution and the resolution of particle velocity distributions.
