CHAOS: An octree-based PIC-DSMC code for modeling of electron kinetic properties in a plasma plume using MPI-CUDA parallelization

摘要: Abstract A new computational framework for a coupled PIC-DSMC tool using multiple GPUs to model the kinetic behavior of electrons in plasma plumes is presented this work. The disparate length scales Debye and collisional mean free path are resolved by separate, independent linearized Morton Z-ordered forest trees. 2:1 restraint imposed PIC module solve partial differential equations context an AMR/Octree framework. MPI-CUDA parallelization strategies used implement preconditioned conjugate gradient method solving electrostatic Poisson's equation on octree discussed scaling code near ideal speedup as function number demonstrated. validated analytical test cases, octree-based simulations found be ten times more efficient compared uniform grid especially plume which have large density variations. then demonstrated with simulation collisionless, mesothermal approach both ions electrons. effect ion mass electron source location analyzed comparing dynamics velocity distribution functions. It shown that confined observed heavier xenon ions, present electric propulsion devices, protons. confinement traps resulting higher temperatures proton case. In simulations, however, temperature anisotropic. Finally, when shifted case considered distributions all three directions unequal non-Maxwellian, contrary co-located beam attract electrons, initially oscillate between radial edges confirmed bi-modal at early times. As evolves, it electrostatically these within beam, allowing them thermalize, from single-peak functions later These results demonstrate can efficiently accelerate computations fully enable study kinetics neutralization highest physical fidelity.