Extended Fourier Neural Operator

Requirements

• PyTorch >=1.8.0. Pytorch 1.8.0 starts to support complex numbers and it has a new implementation of FFT
• torchinfo
• yaml
• numpy
• scipy
• ray

files

• run: containing excutable python files
  • efno_field_2d.py is the EFNO for predicting electirc and magnetic field
  • efno_2d.py is the EFNO for predicting apparent resistivity and phase
  • efno_2d_invariant.py is the EFNO for super-resolution
  • fourier_2d.py is the Fourier Neural Operator(Li et al., 2021)
  • ``ECNN_2d.py` is the extended dense convolutional encoder-decoder network (modified from Zhu et. al, 2019)
  • cofigure.yml is the configuration for efno_2d.py, efno_2d_invariant.py and fourier_2d.py
  • cofigure.yml is the configuration for efno__field2d.py
  • cofigure_CNN.yml is the configuration for ECNN_2d.py
• scripts: some auxiliary python fiels
• model: trained model file
• Log: log file
• temp: if stop early, you can file model file here.

Data

shared in google drive: https://drive.google.com/drive/folders/12nnzinkdz84tAYOOEqsJTzKOgy1MUqpo?usp=sharing

or you can generate training and testing samples by using python files in data_gen.

• gaussian_random_fields.py:using gaussian random filed with different length scale.

• MT2D_secondary.py is compute 2D MT response by using secondary filed method (SFM), Parallel version

• model_random.pygenerate conductivity structures and compute the apparent resistivity and phase by using finite difference method.

Usage

cd ./run
python efno_2d.py random_best

citing

@article{peng2022rapid,
  title={Rapid surrogate modeling of electromagnetic data in frequency domain using neural operator},
  author={Peng, Zhong and Yang, Bo and Xu, Yixian and Wang, Feng and Liu, Lian and Zhang, Yi},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  volume={60},
  pages={1--12},
  year={2022},
  publisher={IEEE}
}