Gold nanoparticle AuNP1000 dataset

This is version 1 of the AuNP1000 dataset for the 1st stage of the AIS25 AI4Materials Nanoparticle (NP) Challenge. The competition is hosted at https://team-capex.github.io/ais25-challenge/. The dataset contains the total energy and interatomic forces of 1000 gold structures evaluated with Density Functional Theory (DFT) at the PBE+D3 level. The dataset is composed ofMaterials Project structures of Au (mp-81, mp-1008634, mp-2647062, mp-1238808).BulkGrain boundariesLow Miller index surface facetsBulk Au (fcc, bcc, hcp).Volume scans around the equilibrium unit cell volumeMolecular dynamics (MD) snapshots from ML MD trajectories of strained (-10%, -5%, 0%, 5%, 10%) 2x2x2 supercells at three different temperatures: 300K, 500K, 700K.Au nanoparticles in vacuum ranging in size from 42 atoms to 561 atoms. All NPs have been sampled with an active learning workflow with ML potentials. The NPs were chosen to maximize the ensemble uncertainty in energy and force prediction (query by committee). The NPs fall into three categories:Relaxed NPs from a minima hopping routineHigh temperature MD out of equilibriumSnapshots from metadynamics simulations where the minimum, mean, and maximum coordination number has been biased along with the radius of gyration.<br><b>DFT settings</b><br>The DFT was performed using the Vienna Ab initio Simulation Package (VASP) with the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional and Grimme's D3 dispersion correction. A plane wave energy cutoff of 350 eV was used along with a k-point density of 50 Å in periodic directions. The used INCAR is uploaded alongside the dataset.<br>