Research Topic

Runtime Models of Particle In Cell Algorithms

Modern high performance computing (HPC) systems provide diverse processor architectures making efficient parallel computing a difficult task. Keeping the physical limitations with high clock speed rates and energy consumptions of processors in mind, the attractiveness of modern multicore processors becomes perspicuous. Thus, leveraging their benefits with hybrid parallelisation strategies becomes necessary. The variety of possible compositions for computing nodes within a heterogeneous computing system motivates investigations for realistic performance and scalability models letting us predict potentials for code optimizations and load balancing strategies.

To explore the research in performance modelling a parallel particle-in-cell (PIC) code is being developed, simulating charged particles accelerated by electromagnetic fields flying through space in time. The numerical solver leverages the finite integration method for calculations of the discretised Maxwell equations and a method proposed by Buneman for calculating currents that result from the motion of the charged particles in a relativistic context. A 3D benchmark will be developed simulating a linear particle accelerator with perfect electric conducting boundaries. The code considers hybrid parallelisation with MPI for inter-node communication and OpenMP for intra-node communication.

Key Research Area

HPC, computational electrodynamics


Prof. Dr.-Ing. Thomas Weiland (TEMF, Computational Electromagnetic Laboratory)

Prof. Dr. Christian Bischof (Scientific Computing)


Grischa Jacobs


Dolivostraße 15

D-64293 Darmstadt



+49 6151 16 - 24377


+49 6151 16 - 24404


jacobs (at) gsc.tu...

 Print |  Impressum |  Sitemap |  Search |  Contact |  Privacy Policy
zum Seitenanfangzum Seitenanfang