DPG Phi
Verhandlungen
Verhandlungen
DPG

Berlin 2014 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

P: Fachverband Plasmaphysik

P 27: Theory of nonideal Plasmas

P 27.6: Talk

Friday, March 21, 2014, 12:15–12:30, SPA HS201

Adaptive-Particle-Refinement for PIC Simulations — •Nils Moschüring and Hartmut Ruhl — Ludwig-Maximilians-Universität, Munich, Germany

Particle-in-cell (PIC) codes commonly employ a finite element method in order to solve Vlasov's equation and a finite-difference-time-domain method to solve Maxwell's equations. The number of employed quasi-elements in the finite element solver is significant for the reliability of the obtained results. In most cases it is also the defining property when evaluating the computational demands of a given simulation. Thus it is regarded to be very useful to dynamically adapt this number. We have developed an adaptive-particle-refinement (APR) algorithm which dynamically adapts the number of quasi-particles during a simulation while conserving the total momentum, total energy and total charge. It introduces no unphysical divergence in the electromagnetic fields. It can either increase the number of quasi-particles by splitting them or decrease the number by merging them. This can be useful in various situations: 1. The number of quasi-particles is an important property which determines the signal-to-noise ratio for various derived quantities. 2. In cases where the charge density is not conserved, numerical algorithms can lead to a steep increase in quasi-particles and thus computational demand. Quasi-particle merging can counteract this. 3. When performing Monte-Carlo-Collisional (MCC) simulations it is imperative to have quasi-particles with comparable or equal weight in order to achieve statistical significance. 4. It represents an important step in order to implement Adaptive-Mesh-Refinement-PIC (AMR-PIC).

100% | Mobile Layout | Deutsche Version | Contact/Imprint/Privacy
DPG-Physik > DPG-Verhandlungen > 2014 > Berlin