Event-driven, hybrid particle-in-cell simulation: a new paradigm for multi-scale plasma modeling

  • Authors:
  • Y. A. Omelchenko;H. Karimabadi

  • Affiliations:
  • SciberQuest, Inc., Computational Physics, Solana Beach, CA;SciberQuest, Inc., Computational Physics, Solana Beach, CA

  • Venue:
  • Journal of Computational Physics
  • Year:
  • 2006

Quantified Score

Hi-index 31.46

Visualization

Abstract

Particle-in-cell models have become standard computational tools for studying complex nonlinear phenomena in space and laboratory plasmas. These simulations are normally very compute-intensive since they require time integration of strongly coupled equations governing the field and particle dynamics. As a result, despite a significant progress in hardware technology, particle-in-cell codes are rarely used to simulate long-time evolution of large-scale systems with strongly varying temporal and spatial scales. We propose an alternative paradigm to time stepping, which is traditionally used for time integration of such systems. This new approach is based on explicit discrete-event simulation technology. It offers distinct advantages over synchronous time stepping: (i) updates of individual macro-particles and discrete field elements are performed asynchronously, (ii) local time increments are determined and self-adaptively adjusted in time through scheduling and execution of physically meaningful local updates ("events"). The event-driven time advance is accurate, free of the global Courant condition, stable, parallelizable, extendable to multiple dimensions and well suited for nonuniform spatial meshes. We demonstrate the new method on a one-dimensional hybrid particle-in-cell model with applications to several plasma discontinuities, including a high-Mach-number fast magnetosonic shock and the associated plasma turbulence.