Results and Impact

HPDDM in PETSc 3.25

Robust Schwarz and GenEO-type preconditioners available in CPU-GPU PETSc workflows

B7 Exascale algorithms B9 Resilience, robustness and accuracy B10 Scientific productivity B11 Reproducibility and replicability

PFDA-BS VQE on 8000 GPUs

Black-box optimization for chemistry workloads, validated on LUMI from 3 to 560 parameters

B7 Exascale algorithms B8 Discovery, design, and decision algorithms B10 Scientific productivity

50× Speedup with GenEO

GenEO preconditioning on large-scale problems (45M DoFs, 2916 cores)

B7 Exascale algorithms B9 Resilience, robustness and accuracy

8000 GPU Scale

Bayesian optimization on LUMI supercomputer with fractal decomposition

B7 Exascale algorithms B8 Discovery, design, and decision algorithms

80-200% CPU Savings

Neural operators accelerating Newton iterations, domain-decomposed PINNs

B7 Exascale algorithms B10 Scientific productivity

Urban Meshing

From GIS data to watertight meshes for city-scale energy simulation

B7 Exascale algorithms B10 Scientific productivity B11 Reproducibility and replicability

HPC Benchmarking

Automated CI/CD pipeline: code → EuroHPC runs in < 24h

B9 Resilience, robustness and accuracy B10 Scientific productivity B11 Reproducibility and replicability