@article{PhysRevX.11.021060,
  title = {Learning and Avoiding Disorder in Multimode Fibers},
  author = {Matth\`es, Maxime W. and Bromberg, Yaron and de Rosny, Julien and Popoff, S\'ebastien M.},
  journal = {Phys. Rev. X},
  volume = {11},
  issue = {2},
  pages = {021060},
  numpages = {12},
  year = {2021},
  month = {Jun},
  publisher = {American Physical Society},
  doi = {10.1103/PhysRevX.11.021060},
  url = {https://link.aps.org/doi/10.1103/PhysRevX.11.021060}
}

Global requirements:

• Numpy
• Matplotlib

Contain the raw and processed data required to generate the figures and to run the demo codes.

• param.json: json file containing the parameters of the experiment
• TM_modes_X.npz: transmission matrix in the mode basis after correction for the deformation
• TM5_0.npy and TM5_1.npy: full transmission matrix in the pixel basis for no deformation (). Because of the 100 Mo restriction of Github, the file is split into two, can be recombined with:

import numpy as np
part1 = np.load('TM5_0.npy')
part2 = np.load('TM5_1.npy')
TM_ref_pix = np.concatenate([part1, part2], axis = 0)

• TM17_0.npy and TM17_1.npy: full transmission matrix in the pixel basis for no deformation ()
• TM25_0.npy and TM25_1.npy: full transmission matrix in the pixel basis for
• TM35_0.npy and TM35_1.npy: full transmission matrix in the pixel basis for
• TM50_0.npy and TM50_1.npy: full transmission matrix in the pixel basis for
• TM52_0.npy and TM52_1.npy: full transmission matrix in the pixel basis for the maximum deformation ().
• conversion_matrices.npz: contains the matrices modes_in and modes_out mode matrices computed for a system without aberration. They are generated using Generate_theoretical_modes/Generate_modes.ipynb.
• modes_in_after_correction.npy: Change of basis matrix between the input mode basis and the input pixel basis after aberration correction optimization.
• modes_out_after_correction.npy: Change of basis matrix between the output mode basis and the output pixel basis after aberration correction optimization.
• mask_near_degenerate.npy: a mask of the same size as the mode basis TM that represents the blocks of quasi-degenerate modes.
• TM_XX_optimization_results.npz with XX = 17, 25, 35 and 50. Results of the aberration correction relative to the corresponding full pixel basis transmission matrix TMXX.

Calculation of the theoretical fiber modes.

Requires: pyMMF

See section 2.1 of the Supplementary Information.

• Generate_theoretical_modes/Generate_modes.ipynb: Jupyter notebook containing an sample code to compute the theoretical mode profiles knowing the properties of the multimode fiber.
• Generate_theoretical_modes/functions.py: Some useful functions to generate the plots.

See section 2.2 of the Supplementary Information.

Requires: PyTorch

• Aberration_correction/Demo_correction_aberration.ipynb: Demo code to use the aberration correction model based on PyTorch framework. It requires a TM measured in the pixel basis and the theoretical modes. It learns the aberrations and misalignments of the optical system and compensate for them. It outputs a TM in the basis of the fiber modes.

• Aberration_correction/Compare_optimization_results.ipynb: Code to compare the results of the optimization for different values of the deformation applied. Uses the TM_XX_optimization_results.npz data files. Corresponds to the results presented in the Section S5 of the Supplementary Information of manuscript.

• Aberration_correction/functions.py: Some useful functions to generate the plots.

• PyTorchAberrations/aberration_models.py: The PyTorch model to apply a set of aberrations and a deformation to the change of basis matrices.

• PyTorchAberrations/aberration_layers.py: Individual custom layers corresponding to each aberration/deformation we can introduce in the model.

• PyTorchAberrations/aberration_functions.py: Useful PyTorch functions, in particular to handle complex linear operations using two layers for the real and imaginary parts.

Processing of the results and creation of the plots.

• Analysis_Deformation.ipynb: Data processing and creation of the figures in the main text of the article.

• Figures_SI.ipynb: Data processing and creation of the figures for the Supplementary Information.

• functions.py: Some useful functions to generate the plots.

Generation of input mask on the digital micro-mirror device (DMD)

See section 1.2 of the Supplementary Information.

Requires: SLMLayout

• Demo_layout.ipynb: A sample code to generate input patterns to send on the DMD.

Visit our website on Wavefrontshaping.net for more information, codes and tutorials.