JefiGPU: Jefimenko's equations on GPU

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

“…We have simulated a toy model quark-gluon system in relativistic heavy-ion collisions as an example to verify our code, and given the corresponding evolution of the charge and current densities as well as the induced EM fields. For a mimic of the STAR AuAu experiments using the RBG-Maxwell framework, readers can refer to [40] for more details. More examples and accurate comparisons will be conducted in our future work.…”

“…In the evolution of QGP, several stages are considered: pre-equilibrium, relativistic hydrodynamics and hadronization. By combining our framework with the hadronization process, we can calculate the charge-odd directed flow Dv , 1 which is the difference between the directed flows of charged particles and their anti-particles (one can refer to [40] for more details). In [40], the obtained results in the slopes of v 1 and Dv 1 in mid-rapidity are in qualitative agreement with the STAR data, and the obtained electric and magnetic fields have opposite contributions to v 1 and Dv 1 but with the same magnitude.…”

“…By combining our framework with the hadronization process, we can calculate the charge-odd directed flow Dv , 1 which is the difference between the directed flows of charged particles and their anti-particles (one can refer to [40] for more details). In [40], the obtained results in the slopes of v 1 and Dv 1 in mid-rapidity are in qualitative agreement with the STAR data, and the obtained electric and magnetic fields have opposite contributions to v 1 and Dv 1 but with the same magnitude. The results are insensitive to the values of the coupling constant and can be understood by a simple sum rule in a naive coalescence picture of hadronization.…”

Jefipic: A 3-D Particle-in-Cell Method Based on Jefimenko's Equations on Gpu

“…However, the numerical simulation of Jefimenko's equations is extremely time consuming, since the integral is performed over the entire source volume with retarded time. For a performance enhancement, we use the GPU-based package JefiGPU [40]. JefiGPU performs the Jefimenko's equation directly on GPU with the help of the high dimensional integration package ZMCintegral [41].…”