PEPC: An Open-Source Multi-Physics Framework for Mesh-free Simulation of N-Body Systems

Gibbon, Paul

Conference Presentation (Invited)

FZJ-2025-04795

PEPC: An Open-Source Multi-Physics Framework for Mesh-free Simulation of N-Body Systems

Gibbon, P. (Corresponding author)FZJ*

2025

SIAM Conference on Computational Science and Engineering, SIAM CSE25, Fort Worth, TX, USA, 3 Mar 2025 - 7 Mar 2025 [10.34734/FZJ-2025-04795]

This record in other databases:

Please use a persistent id in citations: doi:10.34734/FZJ-2025-04795

Abstract: PEPC is a multi-physics community tree-code developedand hosted by JSC over the last two decades [P. Gibbonet al. (2024), DOI: 10.5281/zenodo.11035167.]. At itscore the code utilises a hierarchical tree structure originallybased on the Salmon-Warren Hashed Oct Tree scheme toperform rapid force summation of a dynamical N-body sys-tem interacting via long-range potentials on heterogeneoussuperomputer architectures. Depending on the physicalchoice of potential, the algorithm can be applied to di-verse fields such as plasma, gravitational systems and vor-tex fluids [Durante et al., Math. Comp. in Simulation225, 528 (2024)]. In this talk we highlight some of the re-cent key developments in the PEPC framework, coveringalgorithmic and physical aspects and such as GPU port-ing, time-integration schemes, management of dynamicalparticle populations [Chew et al., Plasma Phys. Contr.Fusion 63, 045012 (2024)] and inclusion of self-generatedmagnetic fields [Siddi et al., Phys. Plasmas 24, 082103(2017)] in charged-particle systems.

Note: Minisymposium on Advances in Particle MethodsOrganizers: Andrew J. Christlieb, Jingwei Hu, Lee RicketsonDescription:Particle methods have been an essential tool in the simulation of problems that are dominated by kinetic effects. Typically these methods are at best second order and have dominated engineering and physics due to their ease of implementation and direct interpretability. Over the past 10 years, advances in particle methods have created a renaissance and how we approach solving these problems, offering more flexibility, greater accuracy and rigorous structure preservation. This minisymposium is designed to highlight advances in particle methods. Topics include fully implicit methods, mesh free representations, structure preserving algorithms, asymptotic preserving methods, adjoint methods for particle representations. This survey of recent advances is meant to engage the mathematical community in a discussion about value of these methods and the importance of this renaissance in method development.

Contributing Institute(s):