Reusability report: Predicting spatiotemporal nonlinear dynamics in multimode fibre optics with a recurrent neural network
This repository contains code used for the manuscript title as Reusability report: Predicting spatiotemporal nonlinear dynamics in multimode fibre optics with a recurrent neural network by Tegin and Dinc et al. (2021) Nature Machine Intelligence, 1-5.
BPM Notebooks folder contains a time-dependent beam propagation simulation for graded-index multimode fiber. The code is written in Python with the CuPy library to perform GPU parallelized simulations.
RNN Notebooks folder contains the training and testing procedures of the recurrent neural network used in the manuscript. The training and testing tasks are performed with these codes written in Python with Tensorflow library and Keras API.
This study is arising from the work of Predicting ultrafast nonlinear dynamics in fibre optics with a recurrent neural network. Salmela et al. Nat Mach Intell (2021) to further investigate the possibilities on modeling nonlinear dynamics in an optical fibre with a recurrent neural network.
Reusability report: Predicting spatiotemporal nonlinear dynamics in multimode fibre optics with a recurrent neural network.
Uğur Teğin, Niyazi Ulaş Dinç, Christophe Moser, and Demetri Psaltis, 2021.
https://doi.org/10.1038/s42256-021-00347-6
@article{teugin2021reusability,
title={Reusability report: Predicting spatiotemporal nonlinear dynamics in multimode fibre optics with a recurrent neural network},
author={Te{\u{g}}in, U{\u{g}}ur and Din{\c{c}}, Niyazi Ula{\c{s}} and Moser, Christophe and Psaltis, Demetri},
journal={Nature Machine Intelligence},
pages={1--5},
year={2021},
publisher={Nature Publishing Group}
}
Google Colab services
CuPy based on Cuda 11.0
Tensorflow 2.4.0
Python 3
Please contact Uğur Teğin or Niyazi Ulaş Dinç for questions.
This project is covered under the Creative Common (CC BY NC) License. The data and code are avaiable for non-commercial research purposes only with proper citation to aforementioned manuscript.