Identifying the twist factor of twisted partially coherent optical beams

Area of Science:

• Optics and Photonics
• Machine Learning Applications
• Statistical Optics

Background:

• Twisted partially coherent light offers unique control over light's statistical properties.
• Recognizing the twist factor is challenging due to low coherence and beam stochasticity.
• Current methods for twist factor recognition are complex and require optimization.

Purpose of the Study:

• To introduce a novel method for recognizing the twist factor of twisted partially coherent beams.
• To simplify the identification process by leveraging a characteristic hollow intensity structure.
• To develop a deep learning model for efficient twist factor detection.

Main Methods:

• Utilizing a circular aperture at the source plane to generate a hollow intensity structure.
• Training a deep learning model to identify the twist factor from the generated intensity signature.
• Employing a simplified model structure without complex optimization layers.

Main Results:

• The circular aperture successfully produces a characteristic hollow intensity structure.
• The trained deep learning model effectively identifies the twist factor of the beams.
• The simplified model streamlines the recognition process, eliminating the need for complex optimization.

Conclusions:

• The proposed method provides a promising solution for enhanced detection of twisted partially coherent light.
• This approach simplifies twist factor recognition, paving the way for future research.
• The study highlights the potential of deep learning in analyzing complex optical phenomena.