Evolution of the marker distribution in gyrokinetic δf particle-in-cell simulations

Abstract

The evolution of the particle weight in a [Formula: see text] particle-in-cell simulation depends on the marker distribution that can evolve in a turbulent field due to turbulent diffusion. When Monte Carlo methods are used to implement the test particle collision operator, or when the particle motion is not strictly Hamiltonian in a collisionless simulation, the marker distribution will evolve along the particle trajectory and, in general, cannot be known exactly. A two-dimensional numerical marker distribution is proposed as an approximation. It is shown to be advantageous over other common methods for evaluating the marker distribution in long-time turbulence simulations. A generalized two-weight [Formula: see text]-method is proposed to mitigate the marker evolution problem.