Efficient calculation of optical properties of suspended-core fiber via a machine learning algorithm

Area of Science:

• Materials Science
• Optical Engineering
• Computational Physics

Background:

• Mid-infrared applications require advanced fiber designs due to increasing nonlinearity demands.
• Chalcogenide suspended-core fibers (SCFs) offer advantages in nonlinearity and dispersion management for mid-infrared applications.
• Traditional numerical methods for SCF modeling face limitations in speed and complexity.

Purpose of the Study:

• To develop and validate a machine learning model for calculating optical properties of chalcogenide SCFs.
• To accurately predict effective mode area, nonlinear coefficient, and dispersion.
• To provide a computationally efficient alternative to traditional modeling techniques.

Main Methods:

• An artificial neural network (ANN) model was established to predict optical properties.
• The model was trained and tested using parameters for As2S3 SCFs within specific wavelength and structural ranges.
• Performance was compared against traditional numerical calculation models.

Main Results:

• The ANN model accurately predicted effective mode area, nonlinear coefficient, and dispersion for As2S3 SCFs.
• The model demonstrated significant advantages in computational speed and resource utilization compared to traditional methods.
• Precise optical property predictions were achieved within the specified parameter ranges (1.0-4.0 µm wavelength, 0.4-0.6 µm core radius, 0.06-0.09 µm cantilever width).

Conclusions:

• The proposed ANN model provides a cost-effective and efficient solution for modeling and optimizing chalcogenide SCFs.
• Machine learning accelerates the design process for advanced optical fibers.
• This approach facilitates precise optical property predictions crucial for mid-infrared applications.