A Multi Level Multi Domain method for Particle In Cell Plasma simulations: algorithm and first benchmarks (Maria Elena Innocenti)

by Maria Elena Innocenti (CPA, KU Leuven)

Room C200 B02.18

Abstract

The computational requirements of current Particle In Cell plasma physics simulations are such that the design of adaptive techniques, which allow to simulate with increased resolution only the domain areas where that level of resolution is actually needed, has received a notable level of attention in recent years. However, evolving Particle In Cell techniques into adaptive methods entails two fundamental challenges. First, the necessity arises to adapt a method which is based on the equality between the grid and the particle shape function size (which is assumed to be fixed in time) to progressively smaller grid spacings and to control the development of self forces on particles traversing regions with different refinement. Second, the need to comply with the strict stability constraints of the explicit PIC methods severely limits the choices regarding the refined grids spacings. In this talk, a possible solution to these challenges is presented. The problem of adapting the particle shape function size to changing grid spacings is solved with the use of a Multi Level Multi Domain approach (the different, progressively more refined levels are simulated fully with fields and particles and particle shape functions are sized on the local grid spacing), while the stability constraints are relaxed by employing an Implicit Moment PIC algorithm instead of an explicit one. This novel implementation will be explained in details (with particular attention to the continuity of particle structures across the grid boundaries) and then benchmarked against a series of 1D test simulations, aimed at assessing the reaction of fields and particles to the exchange of information between the grid levels.