PIMD data for training effective potential incorporating nuclear quantum effects

<p>The dataset contains training data to generate machine-learned effective potentials reproducing correct nuclear quantum statistics. The generation procedure is described in I. Zaporozhets, F. Musil, V. Kapil, &amp; C. Clementi (2024). Accurate nuclear quantum statistics on machine-learned classical effective potentials. <a href="https://arxiv.org/abs/2407.03448">[arXiv: 2407.03448]</a></p> <p>The code required to generate the dataset can be found in the repository: &nbsp;<a href="https://github.com/ClementiGroup/accurate_nuclear_quantum_statistics_on_machine_learned_classical_effective_potentials">cg_nuclear_quantum_statistics</a>.</p> <h2>Dataset structure</h2> <p>The archive contains datasets for four systems: a particle in 1D Morse potential, a single water molecule in a vacuum, a Zundel cation, and a box of 256 water molecules. For the particle in Morse potential, separate .npy files (loadable with numpy in Python) for harmonic coupling (spring) forces and coordinates are given for temperatures 100, 300, and 600 K. For other systems,&nbsp; the data are provided in HDF5 format (see below).</p> <p>Directory structure:</p> <p>CG_quantum_statistics<br>├── 0_morse_potential<br>│ &nbsp; ├── temp_100_spring_forces.npy<br>│ &nbsp; ├── temp_100_total_coordinates.npy<br>│ &nbsp; ├── temp_300_spring_forces.npy<br>│ &nbsp; ├── temp_300_total_coordinates.npy<br>│ &nbsp; ├── temp_600_spring_forces.npy<br>│ &nbsp; └── temp_600_total_coordinates.npy<br>├── 1_h2o_molecule<br>│ &nbsp; └── h2o.h5<br>├── 2_zundel_cation<br>│ &nbsp; └── h5o2+.h5<br>└── 3_bulk_h2o<br>&nbsp; &nbsp; └── h2o_256.h5</p>