Egor Sedov

Egor Sedov

PhD | Researcher | Machine Learning | AI engineering

Phase computation for the finite-genus solutions to the focusing nonlinear Schrödinger equation using convolutional neural networks

Published in Communications in Nonlinear Science and Numerical Simulation, 2023

A method is developed to retrieve phase parameters of quasi-periodic finite-genus solutions to the focusing NLS equation using convolutional neural networks. The proposed architecture leverages RH problem data generation and Bayesian optimization to learn a direct map from finite-duration waveforms to the spectral phases.

Download paper here or
here

@article{bogdanov2023phase,
  title={Phase computation for the finite-genus solutions to the focusing nonlinear Schr{\"o}dinger equation using convolutional neural networks},
  author={Bogdanov, Stepan and Shepelsky, Dmitry and Vasylchenkova, Anastasiia and Sedov, Egor and Freire, Pedro J and Turitsyn, Sergei K and Prilepsky, Jaroslaw E},
  journal={Communications in Nonlinear Science and Numerical Simulation},
  volume={125},
  pages={107311},
  year={2023},
  publisher={Elsevier}
}