Programmable sub-pulse control and LSTM-PINN prediction in a mid-infrared SOPO.

Area of Science:

• Laser Physics
• Nonlinear Optics
• Materials Science

Background:

• Generating mid-infrared (MIR) pulse bursts with controllable sub-pulse features is a significant challenge in laser technology.
• Existing methods often lack the flexibility to independently tailor pulse number, spacing, and amplitude.

Purpose of the Study:

• To demonstrate a novel self-optical parametric oscillator (SOPO) for programmable MIR pulse burst generation.
• To achieve independent control over sub-pulse characteristics (number, spacing, amplitude).
• To develop an AI-driven framework for predicting laser parameters and simplifying experimental modeling.

Main Methods:

• Utilized a 3.8 μm SOPO based on a periodically poled lithium niobate (Nd:MgO:PPLN) crystal.
• Employed dynamic stepwise Q-switching for precise control over pulse generation.
• Developed and applied a physics-informed Long Short-Term Memory-Physics-Informed Neural Network (LSTM-PINN) framework.

Main Results:

• Achieved independent control of sub-pulse number, spacing, and amplitude.
• Delivered sub-pulse energy of 77.3 μJ at 10 kHz repetition rate.
• Obtained minimum sub-pulse width of 23.5 ns and interval of 150 ns.
• LSTM-PINN framework accurately predicted pulse width, amplitude, and energy with high R² values (0.99 for training, 0.93 for validation).

Conclusions:

• Demonstrated a flexible and programmable MIR pulse burst generation system.
• The AI-driven prediction framework significantly simplifies laser experimentation and modeling.
• This work paves the way for intelligent laser control and advanced applications in the mid-infrared spectrum.