Journal
NONLINEAR DYNAMICS
Volume 111, Issue 14, Pages 13343-13355Publisher
SPRINGER
DOI: 10.1007/s11071-023-08531-6
Keywords
Extended physics-informed neural network with twin subnets; Coupled mixed derivative NLSE; Vector optical solitons; Model parameters
Categories
Ask authors/readers for more resources
Using the extended physics-informed neural network, predictions for seven types of vector solitons in the coupled mixed derivative nonlinear Schrodinger equation are made. The results confirm the effectiveness of the physical neural network in solving the NLSE and reveal specific error patterns in different soliton solutions. The study also explores methods to improve parameter prediction for the model. These findings provide valuable references for the study of optical soliton transmission processes through machine learning.
Using the extended physics-informed neural network with twin subnets to study the coupled mixed derivative nonlinear Schrodinger equation (NLSE), seven types of vector solitons including vector single soliton, vector double solitons, anti-dark vector single soliton, anti-dark vector double solitons, vector rogue wave, vector bright-dark single soliton and vector bright-dark double solitons are predicted. The prediction results from seven types of vector solitons with different angles confirm that the physical neural network can be used to effectively solve the coupled mixed derivative NLSE. The error on the two sides and the falling point of the vector rogue wave solution is significantly greater than the middle, while the error of other six types of vector soliton solution mainly reflects in the peak or valley value of bright or dark soliton, and increases along the transmission distance. In addition, how to improve the prediction of model parameters from two aspects of data set and loss function is also studied. These results have certain reference value for studying the optical soliton transmission process by the machine learning.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available