On Distributed Parallelization Strategies for Particle-in-Fourier Schemes

2026-05-11

AI Summary

This study addresses the challenge of efficient distributed parallelization for the Particle-in-Fourier (PIF) method in kinetic plasma simulations by systematically proposing and evaluating three parallelization strategies: domain decomposition, particle decomposition, and a novel space-time decomposition. The latter innovatively integrates temporal parallelism—via the parareal algorithm—into the PIF framework for the first time. All strategies are implemented using MPI within the portable IPPL library. Comprehensive strong and weak scaling analyses are conducted on the Alps and JUWELS Booster supercomputers using 3D-3V Landau damping and Penning trap benchmarks. The experiments elucidate the communication patterns, performance limits, and primary bottlenecks of each approach, offering both theoretical insights and practical guidance for high-performance optimization of the PIF method.

Abstract

We present and compare distributed parallelization strategies for the particle-in-Fourier (PIF) schemes used in kinetic plasma simulations. The different strategies are i) domain decomposition, where both the particles and Fourier modes are split between the MPI ranks ii) particle decomposition, where only the particles are split between the ranks and each rank carries all the modes, and, iii) space-time decomposition, in which time parallelization based on the parareal algorithm is added on top of the particle decomposition. We describe the different communication patterns involved in each of the strategies, the parameter regimes where they work best, and explain their advantages and disadvantages. We implement the strategies within the open-source, performance portable library IPPL and conduct scaling studies with 3D-3V Landau damping and Penning trap benchmark problems on Alps and JUWELS booster supercomputers. We analyze the dominant component timings in each of the strategies and identify areas for future optimizations.

Problem

• Particle-in-Fourier
• distributed parallelization
• kinetic plasma simulations
• MPI
• scalability

Innovation

• Particle-in-Fourier
• distributed parallelization
• parareal algorithm
• kinetic plasma simulation
• performance portability