The NOMAD CoE aims at providing exascale libraries for all major code families that can be used by the entire community, having a focus on advanced methodology (Pillar 1). This is a particular challenge in view of the unfavorable scaling of these methods with system size. The developed libraries will be first linked to above mentioned demonstrator codes to highlight our achievements.
List of codes and libraries developed under the scope of the NOMAD CoE:
Library of code-agnostic as well as code-specific components of Green-function based methods | |
Library of workflows for common and complex materials simulation tasks | |
Dense, symmetric (hermitian) scalable eigenvalue solver for DFT applications | |
|
Linear-scaling library for the evaluation of non-local Fock exchange |
Quantum Mechanics & Interatomic Potentials / Gaussian Approximation Potential | |
Atomic Cluster Expansion | |
Higher-order equivariant message passing neural networks | |
Data-driven electronic structure models | |
NOMAD data infrastructure |