Electro-optic Modulation Programmable Optical Frequency Comb based on Deep Learning

【Objective】To meet the diverse application demands for high-performance Optical Frequency Comb (OFC), especially in terms of independently adjustable parameters like bandwidth, flatness, central wavelength, and spectral line spacing, a method based on secondary coupled Radio Frequency (RF) signals to drive a single Dual Drive Mach-Zehnder Modulator (DDMZM) for OFC generation is proposed.【Methods】Utilizing a single multiplier to generate the secondary RF coupled signals not only increases the number of comb lines produced by the OFC but also offers the advantages of a simple structure and low cost. Additionally, to further enhance the optimization efficiency and performance of the OFC, a deep learning-based inverse design and analysis approach is adopted.【Results】The study shows that the inverse design based on the constructed cascaded network can identify the corresponding parameters for the target OFC in less than one second. This rapid parameter determination method enables programmability of the number of comb lines, OFC power, and line spacing. It can also generate a 13-line OFC with a flatness of 1.769 dB. This efficient design method provides robust support for the rapid preparation and application of OFCs.【Conclusion】The proposed solution in this study demonstrates significant advantages in OFC generation technology, particularly in performance, flexibility, and optimization efficiency. The method of generating OFC through DDMZM driven by secondary coupled RF signals not only simplifies the system structure and reduces costs but also significantly improves design efficiency through the reverse design approach of deep learning. These characteristics make this solution suitable for a wide range of applications, especially in scenarios requiring quick, efficient, and flexible adjustment of OFC parameters.