JefiGPU: Jefimenko's Equations on GPU

Abstract

We have implemented a GPU version of the Jefimenko's equations -Je-fiGPU. Given the proper distributions of the source terms ρ (charge density) and J (current density) in the source volume, the algorithm gives the electromagnetic fields in the observational region (not necessarily overlaps the vicinity of the sources). To verify the accuracy of the GPU implementation, we have compared the obtained results with that of the theoretical ones. Our results show that the deviations of the GPU results from the theoretical ones are around 5%. Meanwhile, we have also compared the performance of the GPU implementation with a CPU version. The simulation results indicate that the GPU code is significantly faster than the CPU version. Finally, we have studied the parameter dependence of the execution time and memory consumption on one NVIDIA Tesla V100 card. Our code can be consistently coupled to RBG (Relativistic Boltzmann equations on GPUs) and many other GPU-based algorithms in physics.

A GPU-based general numerical framework for plasma simulations in terms of microscopic kinetic equations with full collision terms

“…The paper is organized as follows. In section 2, we will briefly introduce the relevant equations and the proposed numerical framework based on our previous work with JefiGPU [26], ZMCintegral [27][28][29] and RBG [11]. In section 3, we perform a toy model simulation that resembles parts of the AuAu collisions in the STAR experiments [30].…”

“…The drifting and Vlasov terms are discretized by the usual upwind difference for numerical stability. The Maxwell equations are directly solved by JefiGPU [26], which is also scalable on GPU clusters.…”

“…Based on the previous work in [11] and [26], we have developed the RBG-Maxwell framework. Unlike in the previous implementations, we have made several essential changes in the RBG-Maxwell framework such that a stable and highly scalable simulation of the relativistic plasma is possible.…”

“…First of all, as an extension of [26], we have added highscaling functionality in the Jefimenko solver. The source regions and the observable regions are treated differently in the scaling procedure.…”