Speaker
Description
Predictive modelling of high-intensity hadron and lepton beams requires self-consistent, first-principles simulations that remain scalable on heterogeneous HPC systems. This contribution presents OPALX as a unified, performance-portable framework for space-charge modelling across a wide range of accelerator regimes.
OPALX combines grid-based and grid-less methods in one toolchain: PIC with pseudo-spectral FFT Poisson solvers, matrix-free finite-difference PCG solvers, matrix-free finite-element solvers, spectral Particle-in-Fourier methods using distributed NUFFTs, and hierarchical Barnes–Hut tree algorithms. The Particle-in-Fourier implementation includes domain, particle, and space-time decomposition strategies for extreme-scale simulations.
The presentation will compare the mathematical structure, accuracy, conservation behaviour, and scalability of these methods using representative proxy applications: Landau damping for collisionless phase-space dynamics, Penning-trap dynamics for confined charged-particle systems, and disorder-induced heating for collisional relaxation and strongly coupled regimes.