search

deepmodeling / DeePTB

DeePTB: A deep learning package for tight-binding approach with ab initio accuracy.
GNU Lesser General Public License v3.0
54 stars 16 forks source link

feat(loss): add new loss options to control number bands to exclude in in dft bands. #181

Closed QG-phy closed 5 months ago

QG-phy commented 5 months ago

  1. Update EigLoss class in loss.py to include new parameters diff_valence and spin_deg

  2. Update loss_options function in argcheck.py to include new parameters diff_valence and spin_deg

  3. Update TrainFullConfigSK dictionary in config_sk.py to include new parameters diff_valence and spin_deg