Electromagnetic PIC simulation with highly enhanced energy conservation (1108.5420v2)

Abstract: We have obtained an electromagnetic PIC (EM-PIC) algorithm based on time-space-extended particle in cell model. In this model particles are shaped objects extended over time and space around Lagrangian markers. Sources carried by these particles are weighted completely into centers and faces of time-space cells of simulation-domain. Weighting method is resulted from implication of conservation of charge of shaped particles. By solving Maxwell's equations over source free zones of simulation grid we reduce solution of these equations to finding field values at nods of this grid. Major source of error in this model (and albeit other PIC models) is identified to be mismatching of particle marker location and location of its assigned sources in time and space. Relation of leapfrog scheme for integration of equations of motion with this discrepancy is investigated by evaluation of violation of energy conservation. We come in conclusion that instead of leapfrog we should integrate equations of motion simultaneously. Though equation of particle momentum becomes time implicit, we can solve it using a corrector-predictor method. In this way we obtain excellent improvement in energy conservation compared to existing leapfrog electromagnetic models. The developed theory is tested against results of our two dimensional EM-PIC code.