DXMag HeuslerDB has been updated with machine-learning predictions for quaternary and all-d Heusler compounds
We have updated DXMag HeuslerDB and added machine-learning prediction results for quaternary and all-d Heusler compounds.
The newly added datasets were generated through a machine-learning-accelerated high-throughput workflow using machine-learning interatomic potentials and transfer-learned regression models. This update includes 131,544 quaternary Heusler compounds and 105,763 all-d Heusler compounds.
Each entry contains ground-state structural and physical-property data, including relaxed structures, formation energies, energies above the convex hull, local magnetic moments, minimum phonon frequencies, Curie/Néel temperatures, and magnetic anisotropy energies.
The database is available at:
For methodological details, please see the following paper:
- E. Xiao and T. Tadano, “Accurate Screening of Functional Materials with Machine-Learning Potential and Transfer-Learned Regressions: Heusler Alloy Benchmark”, npj Computational Materials (2026)