Low-noise particle simulations of plasma physics using particle methods

Astronomy Unit Seminars
Ben McMillan
Sijme-Jan Paardekooper
February 19th, 2016 at 14:30
GO Jones Room 610

The Particle-In-Cell method solves the Vlasov-Maxwell equations numerically by following a reduced number of marker particles along their physical trajectories. Although integrating trajectories accurately in a smooth field is relatively straightforward, inaccuracies arise when computing moments of the distribution functions, due to uneven sampling. This ‘numerical noise’ is particularly problematic when simulating weak instabilities, such as drift instabilities in magnetic confinement fusion or beam driven instabilities in space physics. Noise can be significantly reduced by using a control variates scheme, where most of the integration can be performed analytically, and only the fluctuating component needs to be evaluated using the Monte-Carlo method. A simple variant of this method has been implemented in the PIC code EPOCH: I will present an example of its use for a beam problem.