Journal of computational physics vol:181 issue:1 pages:317-337
The adaptation of PIC methods requires the ability to change the number of particles during the calculation. For PIC methods it is not sufficient to adapt the computational grid. It also necessary to control the local number of particles per cell (particle rezoning) by increasing or decreasing its value to control the local accuracy. In the present paper, we describe some general theoretical considerations regarding the accuracy of various particle rezoning methods. Four algorithms are derived and applied to 1D and 2D PIC simulations. The merits and drawbacks of the algorithms are discussed. Particle rezoning is then applied to 1D studies of collisionless shocks and the 2D simulations of charging of dust immersed in a plasma. (C) 2002 Elsevier Science (USA).