Base de données de configurations atomiques sur l’eau

The dataset contains a water-based database generated through quantum mechanical calculations using the Quantum ESPRESSO package. In addition to water, the database includes other chemically related compounds such as hydrogen peroxide, orthosilicic acid, pyrosilicic acid, and hydrogarnet defects. Each file with the .out extension provides essential information, including the simulation box lattice parameters, the number of atoms, the total energy, the atomic forces, and the stress tensor. These data are extracted to enable the training of machine learning-based interatomic potentials. This database can be used either to train a machine learning potential or to extract atomic coordinates for re-running quantum mechanical calculations with different exchange-correlation (XC) functionals of choice. For direct training, it is advisable to optimize the dataset in advance by selecting a representative subset that aligns with the specific requirements of the model or application.