Pytorch code used for the paper Recurrent Localization Networks applied to the Lippmann-Schwinger Equation. All software is contained in the rln directory, and a sample driver file is provided in main.py. Future development focused on generalizing this approach will be conducted in a separate repository.
The data for this paper is freely available on Mendeley data. A temporary version that is more shell-friendly is also available on Dropbox, but may not be permanently available. A script to automatically download the Dropbox version and prepare it for evaluation is available in this repository's data directory.
A brief discussion of this repository's software design is available in the rln readme file
FLN:
python main.py --eval --load models_trained/FLN_c10.model --model_type FLN --CR 10
RLN-t:
python main.py --eval --load models_trained/RLN_t_c10.model --model_type RLN-t --CR 10
RLN:
python main.py --eval --load models_trained/RLN_full_c10.model --model_type RLN --CR 10
Train FLN using same configuration as paper:
python main.py --model_type FLN --CR 10