/DinAE_4DVarNN_torch

Primary LanguageJupyter NotebookOtherNOASSERTION

Pytorch code for the joint learning of variational models and associated solvers

Associated preprints: https://arxiv.org/abs/2006.03653 and https://arxiv.org/abs/2007.12941

Content of the repository:

  • Directory TrainedModels: trained models for reproducing MNIST and Lorenz-96 results reported in the preprint
  • Notebooks notebookPyTorch_VarModelNN_*_Preprint2020: notebooks to repdoduce the results and figures in the preprint
  • Notebooks notebookPyTorch_DinAE_4DVarNN_*_GitOceaniX.ipynb: notebooks to the proposed framework on Lorenz_63, Lorenz-96 and MNIST datasets

License: CECILL-C license, see attached licence file.