Water Phase Diagram from a General-Purpose Atomic Cluster Expansion Potential: Dataset

This dataset contains the training configurations, reference energies and forces, and auxiliary data used to construct the Atomic Cluster Expansion (ACE) potential for water and to compute the water phase diagram using OPES coexistence simulations and Gibbs–Duhem integration. The dataset complements the accompanying GitHub repository, which contains the input files required to reproduce the results. The dataset is provided as a gzip-compressed pandas DataFrame serialized using Python pickle. To load the dataset, it should not be decompressed manually. Instead, it can be read directly in Python using pandas as follows: import pandas as pddf = pd.read_pickle("dataset_water_ACE_revPBE_D3.gzip", compression="gzip") The resulting DataFrame contains the full DFT reference dataset used for training and validating the ACE potential, including atomic configurations, energies, and forces.