Development of a Scalable PIC Simulator and Its Application to Spacecraft–Plasma Interaction Problems

The particle-in-cell (PIC) method is widely used for various problems in space plasma physics. Spacecraft-plasma interaction problem is one of such extensive topics. With increasing space development and exploration, a strong demand should arise regarding better understanding of plasma disturbance around a spacecraft. For the study, we have developed an electromagnetic PIC code EMSES [1] for the study of spacecraft-plasma interactions. EMSES is assumed to run on modern supercomputers, and thus various optimizations are made for distributed-memory systems. Among the optimizations, load balancing is a very common issue for parallelization of the PIC simulations. In the PIC method, a huge number of particles are mapped onto a simulation domain that is segmented with a mesh, and the particle population potentially takes arbitrary distributions. Hence, it is generally difficult to distribute computation loads equally in terms of both the number of particles and domain size to many computation nodes. We have proposed a dynamic load-balancing algorithm OhHelp [2] for the PIC simulations, and successfully implemented EMSES with it.