https://github.com/tilde-lab/ml-selection

https://github.com/tilde-lab/ml-selection

Last synced: about 2 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/tilde-lab/ml-selection
• Owner: tilde-lab
• License: mit
• Created: 2024-04-10T14:33:41.000Z (9 months ago)
• Default Branch: master
• Last Pushed: 2024-11-01T09:33:43.000Z (2 months ago)
• Last Synced: 2024-11-01T10:25:13.721Z (2 months ago)
• Language: Python
• Size: 4.41 MB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Machine learning for atomistic modeling: descriptor selection

Alina Zhidkovskaya, [Tirtha Vinchurkar](https://orcid.org/0000-0001-5274-3592), and [Evgeny Blokhin](https://orcid.org/0000-0002-5333-3947)


Tilde Materials Informatics and Materials Platform for Data Science LLC

## Intro

The project experiments on predicting the physical properties from the crystal structure by various machine learning methods and descriptors. The properties such as Seebeck coefficient and thermal conductivity are being predicted.

## Technical details

The repository includes Python code for working with the online chemical databases, data processing, and generating chemical descriptors to train machine-learning models. Here one can also find the examples of using the neural networks such as GCN, GAT, PointNet, and Transformer.

The folder `summary` includes experiment metrics. 

## Reproducing this work

Training data is obtained from [MPDS database](https://developer.mpds.io) and compared to [Materials Project](https://materialsproject.org).

## References

- [MPDS](https://doi.org/10.1007/978-3-319-44677-6_62)

- [Materials Project](https://doi.org/10.1063/1.4812323)

## License

MIT

Copyright (c) 2024 Tilde Materials Informatics and Materials Platform for Data Science LLC