This repository contains source code for our machine learning model for predicting self-consistent Hubbard parameters, as presented in this work:
Uhrin, M., Zadoks, A., Binci, L., Marzari, N., & Timrov, I. (2025). Machine learning Hubbard parameters with equivariant neural networks. Npj Computational Materials, 11(1), 19. https://doi.org/10.1038/s41524-024-01501-5i
The experiments carried out in this work can be found in the experiments/ folder along with all the notebooks to generate the plots.
As an example, from experiments you can use:
python run.py experiment=predict_hp model=u
to run an experiment that trains a model to predict Hubbard U values from a linear-response dataset.
Additional experiments can be found in the experiments/experiment/ folder.