XenonPy is a Python library with implemented machine learning tools for materials informatics. It provides a library to compute material descriptors using compositional, structural, and RDKit descriptors. The library also has a heat map visualization function to help users easily understand the correlation between the descriptors and the desired properties. Pre-trained models using neural networks are available to predict various compound properties, and a framework for transition learning is also provided. Combining with iqspr, it is also possible to search for unknown materials with desired properties.


Official site https://github.com/yoshida-lab/XenonPy
Openness ★★★

BSD 3-Clause License

Core Developers

Liu Chang(NIMS), Ryo Yoshida(ISM), Yukinori Koyama(NIMS)


Python>=3.5, PyTorch>=0.4

Forum https://ma.issp.u-tokyo.ac.jp/en/app/1435

External Link