Skip to content

GLAD-RUC/GeoMix

Repository files navigation

Geometric Mixture Models for Electrolyte Conductivity Prediction (NeurIPS 2025)

Anyi Li, Jiacheng Cen, Songyou Li, Mingze Li, Yang Yu, Wenbing Huang

License: MIT

[OpenReview] [Paper] [Poster] [arXiv]

Installation

Key Requirements

e3nn==0.5.1
matplotlib==3.7.2
numpy==1.26.4
scipy==1.8.1
sympy==1.12
torch==2.1.0+cu118
torch_geometric==2.6.1
torch_scatter==2.1.2+pt21cu118
torch_sparse==0.6.18+pt21cu118

To install the required packages, you can use the provided environment file:

conda env create -f geomix-env.yml
conda activate geomix

Alternatively, you can install the packages manually with pip:

pip install -r requirements.txt

Data

The datasets used in this project (CALiSol and DiffMix) are not included in this repository due to their large size. Please follow the instructions in the dataset directory to prepare the data.

Main Experiments

To reproduce the main experiments from the paper, you can run the training script with appropriate arguments:

bash ./run.sh

For other models and configurations, please refer to the scripts in the scripts directory.

Acknowledgements

The benchmark is based on the CALiSol-23: Experimental electrolyte conductivity data for various Li-salts and solvent combinations and Differentiable modeling and optimization of non-aqueous Li-based battery electrolyte solutions using geometric deep learning. We appreciate to the original authors for their contributions to the field of electrolytes modeling and optimization.

Citation

If you find this repository helpful for your research, please consider citing these papers:

# main
@inproceedings{li2025geometric,
  title={Geometric Mixture Models for Electrolyte Conductivity Prediction},
  author={Li, Anyi and Cen, Jiacheng and Li, Songyou and Li, Mingze and Yu, Yang and Huang, Wenbing},
  year={2025},
  booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems}
}

# dataset
@article{de2024calisol,
  title={CALiSol-23: Experimental electrolyte conductivity data for various Li-salts and solvent combinations},
  author={de Blasio, Paolo and Elsborg, Jonas and Vegge, Tejs and Flores, Eibar and Bhowmik, Arghya},
  journal={Scientific Data},
  volume={11},
  number={1},
  pages={750},
  year={2024},
  publisher={Nature Publishing Group UK London}
}

@article{zhu2024differentiable,
  title={Differentiable modeling and optimization of non-aqueous Li-based battery electrolyte solutions using geometric deep learning},
  author={Zhu, Shang and Ramsundar, Bharath and Annevelink, Emil and Lin, Hongyi and Dave, Adarsh and Guan, Pin-Wen and Gering, Kevin and Viswanathan, Venkatasubramanian},
  journal={Nature Communications},
  volume={15},
  number={1},
  pages={8649},
  year={2024},
  publisher={Nature Publishing Group UK London}
}

License

This project is licensed under the MIT License - see the LICENSE file for details.

About

[NeurIPS 2025] Official implementation of GeoMix: A geometry-aware model for mixture property prediction.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors