The ExaFF (Exascale-ready machine learning Force Fields) consortium sets out with the objective to enable the transition of Gaussian approximation potentials (GAPs) to the new GPU-based pre-exascale HPC architectures, LUMI in particular. The transition from CPUs to GPUs represents a major challenge for computational scientists because the existing codes need to be adapted to a different computational logic. ExaFF is a concerted effort between computational physicists and software experts to port parts of the GAP and TurboGAP codes to hybrid architectures. We will also develop the methodologies required to extend the GAP formalism to handle electrostatic interactions efficiently and accurately, and deal with the coupling between ionic and electronic degrees of freedom. These new advances will be used to study the interaction between ions and nanoporous carbon materials for energy-storage applications, and the degradation of semiconductors under heavy radiation environments.

2022

2024