Journal of computational physics vol:115 issue:1 pages:213-227
An algorithm for the dynamic control of the number of particles in a particle-in-cell (PIC), plasma simulation is presented. The algorithm selectively splits and coalesces particles to control the number of particles in each grid cell. It is designed for multiple-length scale problems where an adaptive grid can be applied and for PIC simulations on parallel computers where a constant number of particles per cell is useful for computational efficiency. The algorithm preserves the charge assignment at grid points while splitting one particle into two, or while coalescing two particles into one. Errors in momentum or energy conservation are controlled by a simple, a priori test. The accuracy of the algorithm is demonstrated in several simulations in one dimension, including a collisionless slow-shock, where an adaptive grid calculation with dynamic number control gives comparable accuracy to a uniformly zoned calculation without dynamic control with just 50% of the effort. (C) 1994 Academic Press, Inc.