Published December 19, 2022 | Version v1
Dataset Open

ML-ready Curie temperatures and descriptors extracted from the JuHemd database

  • 1. Peter Grünberg Institut and Institute for Advanced Simulation, Forschungszentrum Jülich and JARA, D-52425 Jülich, Germany

* Contact person

Description

The uploaded archive provides a ML-ready data set extracted from the juHemd database (see references) augmented with supplemental data for atomic descriptors. Descriptors provided in this data set include structural, magnetic, atomic quantities as well as derived (summed) quantities. In total, 118 possible descriptors are included of which 12 are DFT generated. For each simulation type (LDA/GGA) there is also a data set cleaned from DFT data available. After data cleaning and preprocessing we extracted 387 LDA calculated magnetic Heusler structures as well as 408 GGA structures which have a full structural and magnetic data set. As we only aim at magnetic compounds, we chose to filter out compounds from the original JuHemd which have at least 0.1 Bohr magneton as total absolute magnetic moment. For each data file there is an existing descriptor file naming all the descriptors included in the data set.

Files

File preview

files_description.md

All files

Files (162.2 KiB)

Name Size
md5:a72424082c179085d38ce2a1ee2232e3
322 Bytes Preview Download
md5:ae38da79addeb5a36daf16f19f853132
161.9 KiB Download

References

Journal reference (PhD thesis from which this data set emerged)
R. Hilgers, PhD Thesis, RWTH Aachen (2024), doi: 10.18154/rwth-2024-09243

Website
R. Kováčik, P. Mavropoulos, S. Blügel, Materials Cloud (2022), doi: 10.24435/materialscloud:ww-pv

Software (Software which generated this dataset from the JuHemd database)
R. Hilgers, D. Wortmann, S. Blügel, Zenodo (2022), doi: 10.5281/zenodo.7452755

Preprint
R. Hilgers, D. Wortmann, S. Blügel, Machine Learning-based estimation and explainable artificial intelligence-supported interpretation of the critical temperature from magnetic ab initio Heusler alloys data, arXiv (2023), doi: 10.48550/arXiv.2311.15423