Event-Driven Temporal Refinement of Multi-Scale Plasma Simulations

Summary form only given. We present an adaptive approach to temporal refinement of particle-in-cell and fluid (MHD, two-fluid) simulations of space and laboratory plasmas. It is based on a novel methodology for explicit integration of inhomogeneous nonlinear systems with multiple time scales: discrete-event simulation (DES). DES enables asynchronous time integration on arbitrary meshes by adaptively distributing CPU resources in accordance with local time scales. This removes computational penalties incurred in explicit time-stepping codes due to the Courant-Friedrich-Levy (CFL) restriction on a global time-step size. DES stands apart from a broad family of local (multiple) time-stepping algorithms in that it requires neither selecting a global synchronization time step nor predetermining a sequence of time-integration operations for individual parts of a nonlinear system. Instead, solution elements self-adaptively predict and synchronize their temporal trajectories by enforcing causality constraints, which are formulated in terms of local changes to the evolving solution. Together with flux-conservative propagation of asynchronous information, this remarkable property of DES ensures stable and efficient asynchronous runs. In addition, DES automatically eliminates idle computation in inactive regions of the global domain. DES examples, ranging from electrostatic particle-in-cell to hybrid to fluid simulations, demonstrate the superior performance of event-driven computation in terms of accuracy, efficiency and robustness (stability). We also report our progress in developing a new, 3D hybrid-DES code, HYDES, designed for modeling multi-scale low-frequency processes in laboratory (FRC, z-pinch, ICF) and space (magnetospheric, heliospheric) plasmas.