Adjoint Method-based Fourier Neural Operator Surrogate Solver for Wavefront Shaping in Tunable Metasurfaces

Data preparation

Our metalens dataset used in the paper can be accessed on GoogleDrive.

The train-test data should be placed in data directory. :

(repository)
└───data
      └───dataset
            ├───train
            └───valid

Training

After preparing the dataset in data/DRMI_dataset directory, use

python fno_train.py

for our model, or

python train.py

for comparison group.

To change the settings for training, modify arguments in train python files.

Evaluation

python iteration.py

You should set the log path in the iteration python file in order to evaluate.