Implementing Distributed-Weight Particle Loading for BARS with PIC Codes

To implement the mini-BARS algorithm [1] for improved sampling of velocity-partitioned phase space, we use a methodology for generating mixed-Maxwellian-weighted electron velocity distribution functions (VDF) as initial conditions. We test these methods on kinetic, nonlinear initial value plasma waves. We show how to down-sample from a master VDF (e.g. ≥10⁸ particles) so that low order moments are conserved as is entropy per velocity partition. From many samples those with the least pathological features are selected. Some advantages of this approach are observed over traditional PIC sampling, including the so called quiet start or random sampling strategy that Monte Carlo techniques such as PIC most naturally adhere to. This work is performed using the versatile and sophisticated LANL DPIC code [2] together with mini-BARS.