Estimating Effective Collision Frequency and Kinetic Entropy Uncertainty in Particle-in-Cell Simulations

A kinetic entropy diagnostic was systematically developed for fully kinetic collisionless particle-in-cell (PIC) simulations in Liang et al., Phys. Plasmas 26, 082903 (2019). Here, we first show that kinetic entropy can be used to quantitatively evaluate numerical dissipation in the PIC simulation. Assuming numerical effects can be treated using a relaxation time approximation collision operator, the rate of increase of the kinetic entropy is related to the kinetic entropy. The effective collision frequency due to numerical effects is then easy to evaluate in a collisionless PIC simulation. We find an effective collision frequency of approximately a tenth of the ion cyclotron frequency. This could have important implications for collisionless PIC simulation studies of magnetic reconnection, plasma turbulence, and collisionless shocks. Then, we analyze the uncertainty of the local kinetic entropy density at different locations as a function of the chosen velocity space grid. We find that although the numerically obtained kinetic entropy density varies significantly for small or large velocity space grids, there is a range for which the kinetic entropy density is only weakly sensitive to the velocity space grid. Our analysis of the uncertainty suggests a velocity space grid close to the thermal velocity is optimal, and the uncertainty introduced is significantly less than the physical change in kinetic entropy density.