A Generative Residual Enhanced Neural Operator Based on the Boundary Element Method for Accurate Metasurface Parameter Analysis

Key Points

• This research aims to enhance the accuracy of metasurface parameter analysis using a generative residual enhanced neural operator.
• Developed a neural operator based on the boundary element method for field distribution prediction.
• Tested the operator across various structural parameters and frequencies.
• Evaluated the performance against conventional methods for speed and accuracy.
• The neural operator significantly reduced prediction time compared to traditional methods.
• Achieved improved accuracy in field distribution across different parameters and frequencies.