Spectrally\nProgrammable Optical Frequency Comb Generation

Optical frequency combs (OFCs) are critical components\nin several\nfields, such as optical communications, microwave photonics, ranging,\natomic clocks, and photonic neural networks, which meet diverse application\nrequirements through precise spectral manipulations. Conventional\nOFC spectral manipulation methods use Fourier transform pulse shapers\nfor postprocessing. However, they face constraints in terms of spectral\nresolution and operational bandwidth. In this study, a deep-learning-assisted\napproach is introduced to enable efficient spectral shaping of OFCs\nthrough nonlinear broadening, achieving broader spectral shaping by\ncontrolling a narrower-bandwidth seed comb. By training a convolutional\nneural network, we model complex nonlinear interactions within a highly\nnonlinear fiber to predict and control the spectral output of a broadened\ncomb. Experimental validation confirms the system’s ability\nto generate, with high precision, diverse spectral shapes such as\nGaussian, parabolic, Cauchy, Laplace, and Gaussian mixture models\nwithin seconds, with a relative error of 10<sup>–1</sup>–10<sup>–2</sup>. Our study presents a flexible, rapid, and efficient\nmethod for the spectral shaping of OFCs using deep learning, marking\na substantial advancement in the programmability and utility of OFC\ntechnologies.