MultiXscale

The MultiXscale Centre of Excellence will increase performance, productivity and portability (“the Three P's”) across the entire spectrum of scientists active in the domain of multiscale simulation. It couples the scientific expertise of the CECAM network, represented by leading experts in multiscale simulations from different European institutions, with the technical expertise of the EESSI collaboration, and targets the computational laboratories of EuroHPC and beyond.

It will shoulder much of the technical burden of developing and distributing domain-relevant applications for (pre-)exascale through application co-design for exascale technologies, and the provisioning of exascale-oriented libraries and services that nudge the community to adopt battle-tested, future-oriented, scalable workflows and portable technologies. Together, these will allow application developers to pursue domain-relevant scientific innovation without being over-burdened by technical detail, and empower industrial and academic application users to painlessly adopt bleeding-edge technologies from the domain on whatever computational resource they may have access to.

To drive the development of the libraries and services, and to showcase the scientific and industrial potential of truly multiscale approaches, MultiXscale will pursue three pilot use cases of societal and industrial significance:

• helicopter design and certification for civil transport,
• battery applications to support the sustainable energy transition,
• ultrasound for non-invasive diagnostics and biomedical applications.

MultiXscale will extend the applications, user-base and domains actively engaged in the current CoE and EuroHPC ecosystem by addressing specific and critical needs, and advancing the transition towards use of exascale resources by scientific and industrial users in the community of multiscale modelling.

The role of Stuttgart within MultiXscale is the porting of ESPResSo to Tier-0 HPC systems and the development of a pre-exascale demonstrator application to simulate energy materials. The envisioned application will be used to investigate the capacitance and charging dynamics of novel battery designs where the ionic liquid is stored in a nanoporous structure. This effort is carried out in close collaboration with partner institutes to co-design scalable and re-usable community codes: the Jülich Supercomputing Centre for long-range electrostatics solvers, the University of Erlangen–Nuremberg for lattice-Boltzmann fluid solvers, and the Max Planck Institute for Polymer Research for efficient parallel input/output writers with standardized data file formats. The codes produced by this project will be published as free and open-source software libraries that are easily re-used by third-party software. In addition, Stuttgart will collaborate with EESSI to automatically test and deploy scientific software on compute clusters and supercomputers, and CASTIEL to promote HPC and boost digital skills through training.