Exascale simulation software for systems with three-body interactions

The 3xa project will develop scalable methods for the simulation of three-body interactions in particle systems, applying vectorized kernels, dynamic load balancing approaches and adaptive resolution schemata.

Particle simulation has established itself as an essential instrument for basic science and applied research. It enables fundamental discoveries about materials and helps to optimize processes and products, for example in the chemical industry.

Interaction potentials that describe the interactions between various kinds of particles form the foundation of particle simulations. So far, models of interactions between pairs of particles are well established. A key argument in favor of using interaction potentials between pairs of particles is their efficient computability using highly scalable algorithms. Because they are only capable of additive pair-wise interactions, however, their ability to approach physical reality is limited. This leads, for example, to inaccurate substance data for real systems of polar substances, which occur in many application cases in the chemical industry. However, the necessity to observe real multi-body interactions in particle simulations is seldom taken into account today because of their extremely high computational costs.

The 3xa project is developing interdisciplinary and holistically scalable methods for the computation of three-body interactions, and will demonstrate them in the context of particle simulations. In order to simulate the largest possible systems, HLRS is working with project partners to optimize scalability at the core and node as well as at the internode level.

Further, to optimize resource usage, a highly scalable adaptive resolution method that merges two- and three-body modeling will be developed for scenarios that require only locally accurate three-body modeling. These advances will deliver insights in the broad field of three-body interaction models. The project also aims to make significant progress on exemplary applications from the chemical industry using heterogeneous exascale computing systems.


01. November 2022 -
31. October 2025



Project partners

  • Helmut-Schmidt-Universität/Universität der Bundeswehr Hamburg
  • Technische Universität München
  • Technische Universität Berlin


BMBF Logo: Federal Ministry of Education and Research