The Heterogeneous Exascale Particle-In-Cell

Financiers

Matter in the state of plasma is currently used in many technological processes and is a crucial part of modern industrial and scientific undertakings. The use of plasma technology ranges from material deposition in chip production to space technology and future fusion power plants. However, plasma is inhospitable in terms of characterisation and is difficult to model due to physical processes in plasma happening on many different lengths and time scales. That often leads to balancing the modelling efforts between precision and rigour on one side and speed and accessibility on the other side. The former properties are native to the kinetic description of plasma and the latter to a fluid description. With the arrival of exascale computational capabilities, there should not be a compromise between the two approaches.

The main representative of the kinetic approach is the particle-in-cell codes, which combine Maxwell's field equations with Newtonian mechanics. This method is a first-principle method since it self-consistently describes the plasma with particle-particle, particle-field and particle-boundary interactions from basic equations. However, on one hand, it allows for a very natural description of plasma on a particle level, but on the other hand, the computational requirements are significantly higher than in the Eulerian approach of the fluid codes. This is because the kinetic codes follow each computational particle trajectory during the simulation, while the fluid code assumes that plasma is in thermodynamic equilibrium and treats it as a homogeneous matter - a fluid - flowing past fixed measuring/computational points. In many technological processes, deviation from the thermodynamic equilibrium is a crucial part of the problem, so a kinetic description is needed to gain insight into the processes and to be able to optimize them.

With Heterogeneous Exascale Particle-In-Cell (HEXAPIC), we intend to significantly shorten the simulation time of the fully kinetic approach and, at the same time, increase the level of complexity of plasma processes' description. We shall do that by developing optimized PIC algorithms for exascale computing and utilizing modern features of exascale infrastructure with a tailor-made PIC code framework. This will take advantage of state-of-the-art approaches in memory optimization, load balancing, use of accelerators, and optimized communication protocols of computer systems.

The HEXAPIC project aims to enable high-fidelity kinetic simulations of various technological processes involving plasma on reasonable time scales by introducing improved PIC algorithms, tailoring PIC workflow for exascale, and optimizing code framework using state-of-the-art software (programming models) and hardware approaches.

Project Partners

University of Luxembourg