Abstract:The modeling and prediction of the ultrafast nonlinear dynamics in the optical fiber are essential for the studies of laser design, experimental optimization, and other fundamental applications. The traditional propagation modeling method based on the nonlinear Schrödinger equation (NLSE) has long been regarded as extremely time-consuming, especially for designing and optimizing experiments. The recurrent neural network (RNN) has been implemented as an accurate intensity prediction tool with reduced complexity and good generalization capability. However, the complexity of long grid input points and the flexibility of neural network structure should be further optimized for broader applications. Here, we propose a convolutional feature separation modeling method to predict full-field ultrafast nonlinear dynamics with low complexity and strong generalization ability with high accuracy, where the linear effects are firstly modeled by NLSE-derived methods, then a convolutional deep learning method is implemented for nonlinearity modeling. With this method, the temporal relevance of nonlinear effects is substantially shortened, and the parameters and scale of neural networks can be greatly reduced. The running time achieves a 94% reduction versus NLSE and an 87% reduction versus RNN without accuracy deterioration. In addition, the input pulse conditions, including grid point numbers, durations, peak powers, and propagation distance, can be generalized accurately during the predicting process. The results represent a remarkable improvement in ultrafast nonlinear dynamics prediction and this work also provides novel perspectives of the feature separation modeling method for quickly and flexibly studying the nonlinear characteristics in other fields.

AI-enabled Fast and Accurate Modeling for Femtosecond Chirped-pulse Amplification

Enhanced bi-LSTM for Modeling Nonlinear Amplification Dynamics of Ultra-Short Optical Pulses

Numerical Model of Chirped Pulse Amplification in Pulsed Synchronously Pumped Low-Repetition-Rate Fiber Amplifiers

Fast Predicting the Complex Nonlinear Dynamics of Mode‐Locked Fiber Laser by a Recurrent Neural Network with Prior Information Feeding

CNN-based Neural Network Model for Amplified Laser Pulse Temporal Shape Prediction with Dynamic Requirement in High-Power Laser Facility

Simulating the Spiking Response of VCSEL-based Optical Spiking Neuron

Low-complexity full-field ultrafast nonlinear dynamics prediction by a convolutional feature separation modeling method

Gain Effect and Amplification Characteristics Analysis in Fiber Chirped Pulse Amplification Systems

Compact and stable chirped pulse amplifier system based on nonlinear amplifying loop mirror oscillator and chirped-volume Bragg grating compression

396 MHz CPA femtosecond laser system based on a single crystal fiber rod-type amplifier

1720nm-1800nm tunable chirped pulse amplification based on thulium-doped fiber

Accelerating LSTM-based High-Rate Dynamic System Models

High-power 0.4-mJ picosecond CPA system based on an extra-large-mode-area triple-clad fiber

Amplification of femtosecond pulses with AI-assisted spectral phase modulation

Development of a Femtosecond High-Power Ytterbium-doped Fiber Laser System and Study on Spiky Spectral Modulation in Highly Nonlinear Chirped-Pulse Fiber Amplifiers

C-LSTM: Enabling Efficient LSTM Using Structured Compression Techniques on FPGAs

Photonic Neural Activation Function Based on High-Speed Electro-Absorption Modulated Laser

A Pulse Compressor for A Chirped-Pulse Amplification System Based on A Frequency-Doubling Crystal and Grating

Femtosecond pulse amplification on a chip

Wavelength-Tunable Chirped Pulse Amplification System (1720 nm–1800 nm) Based on Thulium-Doped Fiber

Dynamics of Yb-doped Double Clad Pulsed Fiber Amplifier Based on Slow Motion Dynamic Rate Equations