# Liquid Water and Ice Ih/Ic at Hybrid DFT (revPBE0-D3): Neural Network Training Set for Ab Initio Thermodynamics (Cheng et al. 2019, Materials Cloud), ColabFit Conversion

## Facts

- Provider: Bingqing Cheng, Edgar A. Engel, Jörg Behler, Christoph Dellago and Michele Ceriotti; ColabFit conversion
- License: CC-BY-4.0 (https://archive.materialscloud.org/records/eg3pn-1fw83)
- Size: 4 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/colabfit/HO_PNAS_2019
- Access: public, readable without an account
- DOI: https://doi.org/10.24435/materialscloud:2018.0020/v1
- Category: Chemistry & Materials
- Source: https://doi.org/10.1073/pnas.1815117116
- Page: https://scigantic.com/archive/37aa722d-be84-5c5f-ab40-de492229041f
- Keywords: Chemistry & Materials, DFT, quantum chemistry, Parquet, Hugging Face

## About

Training configurations of liquid water and hexagonal (Ih) and cubic (Ic) ice with total energies and forces at the hybrid-functional revPBE0-D3 DFT level, used to fit the neural network potential behind an ab initio prediction of water thermodynamics. Combining free energy methods with machine learning, the study accounts for quantum nuclear motion, anharmonicity and proton disorder, estimates melting points of light and heavy water, and finds nuclear quantum effects make ice Ih more stable than ice Ic by about 0.2 meV per H2O. ColabFit's card counts 1,588 configurations and 304,896 atoms (H, O), labelled with energy, atomic forces; the repository is about 25.3 MB. Files are the ColabFit Exchange conversion: Parquet files under co/ hold one row per configuration (atomic numbers, positions, cell, computed properties and metadata), ds.parquet holds dataset-level information, and cs/ groups configurations into named sets where defined; read them with pyarrow, pandas, polars or Hugging Face datasets. The original deposit (DOI 10.24435/materialscloud:2018.0020/v1) is released under CC-BY-4.0; cite Cheng et al., PNAS (2019), doi:10.1073/pnas.1815117116.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("colabfit/HO_PNAS_2019")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `co/co_0.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/37aa722d-be84-5c5f-ab40-de492229041f
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/37aa722d-be84-5c5f-ab40-de492229041f
- Over MCP: `get_schema_card` with id `37aa722d-be84-5c5f-ab40-de492229041f` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Chemistry & Materials datasets](https://scigantic.com/topics/chemistry-and-materials)
- [DFT datasets](https://scigantic.com/topics/dft)
- [Quantum chemistry datasets](https://scigantic.com/topics/quantum-chemistry)
