Skip to content

Latest commit

 

History

History
20 lines (13 loc) · 608 Bytes

README.md

File metadata and controls

20 lines (13 loc) · 608 Bytes

magnetic-ordering

Classify magnetic orderings using the current version of e3nn. This is an updated version of the code used by Merker et al.

The most recent run was at 71.7% accuracy overall; more information in statistics.txt.

Organization

Data preprocessing and model training/testing are in two separate files. Both files can be run separately, or both at once using run_magnetic_ordering.py.

Dependencies

Non-extensive list of notable dependencies:

  • e3nn >= 0.4.4
  • torch==1.10.2
  • torch-geometric (installed with cuda version 11.3)
  • numpy==1.22.1
  • pymatgen==2022.3.29
  • ase==3.22.1