4000^3 cells is extremely large for a PiC simulation and the dispersion free Maxwell solvers are indeed state of the art (without them this simulation would be

4000^3 cells is extremely large for a PiC simulation and the dispersion free Maxwell solvers are indeed state of the art (without them this simulation would be impossible and scaling those solvers is quite hard), but the 2 particles per cell is worryingly low, even for a 3d simulation with high order particles. Right now I can not read the supplemental material so I am not sure to which degree this was validate with more particles in2d, but if nature does not behave like this simulation, that is going to be the reason. Source: I write PiC codes for a living.