MetaLab is an open-source Python/C++ programming package for the inverse design of nanoscale optics devices by machine learning methods. By using powerful machine learning algorithms (deep neural network, GBDT ...), MetaLab supports quick design of optical structures with high accuracy. The latest version could be accessed and downloaded from https://github.com/closest-git/MetaLab.

Prepare data and save results in the DATA_DIR. Then run the following command:

    cd surface_plasmon
    python Net_sp_design_.py DATA_DIR

We’re especially interested in hearing from and potentially working with those who are studying

• Inverse design of nanophotonic devices by machine learning methods
• Novel machine learning methods for inverse problems

Please use the following bibtex entry:

@article{chen2019smart,
  title={Smart inverse design of graphene-based photonic metamaterials by an adaptive artificial neural network},
  author={Chen, Yingshi and Zhu, Jinfeng and Xie, Yinong and Feng, Naixing and Liu, Qing Huo},
  journal={Nanoscale},
  volume={11},
  number={19},
  pages={9749--9755},
  year={2019},
  publisher={Royal Society of Chemistry}
}

We may release code for more models.

MIT

MetaLAB was written by Yingshi Chen([email protected]).