A PARALLEL PARTICLE-IN-CELL MODEL FOR THE MASSIVELY PARALLEL PROCESSOR

A two-dimensional parallel particle-in-cell (PIC) algorithm was developed for the Massively Parallel Processor (MPP), which has 16,384 processors configured in a 128 × 128 array with nearest-neighbor communication. The developed parallel PIC algorithm loads particles in a cell randomly to a row of processors and fills only half of the processors with particles. Because the vacant processors and the random particle sequence in a row of processors simplify the communication among the processors, the algorithm efficiently sorts particles and performs gather/scatter procedures for collecting charge density according to their cells. From the cell charge density, the algorithm calculates electric fields at the cells by fast Fourier transform. With faster I/O or more array memory, the parallel PIC algorithm can be greatly improved. © 1990.