Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improve the Fitting Accuracy of Deep Learning for the Nonlinear Schrödinger Equation Using Linear Feature Decoupling Method

Yunfan Zhang, Zekun Niu, Minghui Shi, Weisheng Hu, Lilin Yi

Abstract

We utilize the Feature Decoupling Distributed (FDD) method to enhance the capability of deep learning to fit the Nonlinear Schrodinger Equation (NLSE), significantly reducing the NLSE loss compared to non decoupling model.

Improve the Fitting Accuracy of Deep Learning for the Nonlinear Schrödinger Equation Using Linear Feature Decoupling Method

Abstract

We utilize the Feature Decoupling Distributed (FDD) method to enhance the capability of deep learning to fit the Nonlinear Schrodinger Equation (NLSE), significantly reducing the NLSE loss compared to non decoupling model.

Paper Structure

This paper contains 9 sections, 6 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Principle
  3. The Nonlinear Schrödinger Equation (NLSE)
  4. Principle of utilizing FDD model to fit NLSE
  5. Simulation and results discussion
  6. Simulation and training setup
  7. Accuracy of non decoupling model and FDD model
  8. NLSE loss of FDD model system and non decoupling model system
  9. Conclusion

Figures (4)

  • Figure 1: Two methods for fitting the NLSE.
  • Figure 2: The model accuracy of FDD model and non decoupling model.
  • Figure 3: NMSE of non decoupling systems and FDD model.
  • Figure 4: NLSE loss training curve of non decoupling model and FDD model.