Full EPIC-GOD: An energy-conserving full particle-in-cell code for GPU acceleration using OpenACC

Abstract

We present Full EPIC-GOD, a fully implicit, energy-and charge-conserving electromagnetic particle-in-cell (PIC) code. Unlike conventional full-PIC approaches that often suffer from numerical violations of conservation laws, Full EPIC-GOD tightly couples particle dynamics with Maxwell's equations via an iterative predictor-corrector scheme. Charge conservation is rigorously enforced through Esirkepov's method, while total energy conservation is achieved using consistent field interpolation and residual-based iterative convergence. The algorithm employs second-order accurate discretization in both space and time and supports adaptive time stepping to enhance numerical stability and efficiency. The code is parallelized with OpenACC and demonstrates strong scaling on multi-GPU systems, enabling large-scale kinetic simulations. We validate the code using standard benchmarks involving kinetic waves, electromagnetic instabilities, and magnetic reconnection. The results show excellent agreement with theory and prior simulations, confirming the method's accuracy, stability, and conservation properties. Designed for fully kinetic plasma simulations, Full EPIC-GOD enables high-fidelity modeling of collisionless plasma dynamics across microscopic to relativistic regimes in space and astrophysical environments.