# Acetylacetone (acac) Proton-Transfer Benchmark: 300 K MD Training Set with PBE-D3 Forces (BOTNet, Batatia et al.), ColabFit Conversion

## Facts

- Provider: Ilyes Batatia, Simon Batzner, David Peter Kovacs, Albert Musaelian, Gregor Simm, Ralf Drautz, Christoph Ortner, Boris Kozinsky, Gabor Csanyi; ColabFit conversion
- License: MIT (https://github.com/davkovacs/BOTNet-datasets/blob/main/LICENSE)
- Size: 4 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/colabfit/BOTnet_ACAC_2022_train_300K_MD
- Access: public, readable without an account
- Category: Chemistry & Materials
- Source: https://doi.org/10.1038/s42256-024-00956-x
- Page: https://scigantic.com/archive/d28b4d5c-ea89-5a73-ac2b-ec98149f9824
- Keywords: Chemistry & Materials, molecular dynamics, DFT, interatomic potentials, benchmark, Parquet, Hugging Face

## About

DFT energies and forces for acetylacetone (acac, pentane-2,4-dione), a small molecule with a strong intramolecular hydrogen bond whose proton hops between the two oxygens, introduced with the BOTNet study of the design space of E(3)-equivariant atom-centered interatomic potentials. ColabFit's card describes the set as 500 decorrelated geometries sampled every 1 ps from a 300 K Langevin molecular dynamics run at the GFN2-xTB semi-empirical level and recomputed with DFT (PBE with D3 dispersion, def2-SVP basis, ORCA). Sibling ColabFit repositories hold the 600 K training set, 300 K and 600 K test sets, an O-C-C-C dihedral scan, the hydrogen-transfer path and isolated atoms, which test extrapolation to bond rotation and proton transfer. ColabFit's card counts 500 configurations and 7,500 atoms, labelled with energy and atomic forces; about 1.3 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 authors' BOTNet-datasets repository (dataset_acac) is released under the MIT License; cite Batatia et al., Nature Machine Intelligence (2025), doi 10.1038/s42256-024-00956-x.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("colabfit/BOTnet_ACAC_2022_train_300K_MD")
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/d28b4d5c-ea89-5a73-ac2b-ec98149f9824
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/d28b4d5c-ea89-5a73-ac2b-ec98149f9824
- Over MCP: `get_schema_card` with id `d28b4d5c-ea89-5a73-ac2b-ec98149f9824` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Benchmark datasets](https://scigantic.com/topics/benchmark)
- [Chemistry & Materials datasets](https://scigantic.com/topics/chemistry-and-materials)
- [DFT datasets](https://scigantic.com/topics/dft)
- [Interatomic potentials datasets](https://scigantic.com/topics/interatomic-potentials)
- [Molecular dynamics datasets](https://scigantic.com/topics/molecular-dynamics)
