Publikationsrepositorium - Helmholtz-Zentrum Dresden-Rossendorf

This talk will introduce the basic concepts of how particle-in-cell codes model plasma dynamics and discuss their implementation in the open-source code PIConGPU, focusing on how parallelism can be exploited to enable efficient scaling on today's largest HPC systems. Furthermore, the problem of IO limitations with larger simulations is discussed and the plugin method for in-situ data analysis in PIConGPU is presented to overcome these limitations. Finally, an overview of different physics cases simulated with PIConGPU is presented, ranging from small-scale laser-plasma accelerators to plasma jets in astrophysics.