DFT dataset supplied for publication `Data-efficient machine-learning of complex Fe–Mo intermetallics using domain knowledge of chemistry and crystallography`

DFT dataset supplied for publication `Data-efficient machine-learning of complex Fe–Mo intermetallics using domain knowledge of chemistry and crystallography` --- Overview This dataset contains results of density functional theory (DFT) calculations for binary Fe-Mo alloy structures in topologically close-packed (TCP) crystal phases, together with pre-computed BOPfox bond-order potential (BOP) moment descriptors. It accompanies the publication: > Mariano Forti, Alesya Malakhova, Yury Lysogorskiy, Wenhao Zhang, Jean-Claude Crivello,> Jean-Marc Joubert, Ralf Drautz, Thomas Hammerschmidt,> *Data-efficient machine-learning of complex Fe–Mo intermetallics using domain knowledge> of chemistry and crystallography*, npj Computational Materials (2025).> DOI: **10.1038/s41524-026-02070-5** The dataset is used to train and validate machine-learning models (Kernel RidgeRegression, Random Forest, MLP) that predict formation energies of TCP phases fromBOP moment descriptors encoding chemical, crystallographic and local bondingdomain knowledge. --- DFT Methodology - Code : VASP (Vienna Ab initio Simulation Package)- Exchange–correlation : PBE-PAW (GGA)- ENCUT : 450 eV- k-point spacing : 0.02 Å⁻¹ (Monkhorst-Pack)- Magnetic configurations : ferromagnetic (FM) and non-magnetic (NM) spin initialisation.- Relaxation : full ionic and cell relaxation followed by E–V curve fitting (Birch-Murnaghan equation of state) to obtain equilibrium volume V₀, bulk modulus B₀, and total energy E₀ --- ## How to Load the Data (Python) Dataset download is integrated into the notebook-based worflow in github repository --- ### Related Resources - GitHub repository (code, notebooks, Tools package): https://github.com/AIIMProject/MLFeMoTCPs- **BOPfox** (required to recompute BOP descriptors): available upon request from the authors.