Leveraging Machine Learning and a Photonic Integrated Chip for the Spectro-Temporal Tailoring of Supercontinuum Generation

Programmable photonic integrated circuits (PICs) [1] have recently gathered great interest due to their capacity to add flexibility and versatility in optical systems to a degree that is not generally possible with either free-space or fibered setups. In particular, it has recently been shown that such programmable PICs can be used to control the spectral profile of supercontinuum generation (SC) [2], which can provide extra flexibility to several applications such as microscopy, metrology and hyperspectral imaging [3]. However, in many applications, there is an interest not only in controlling the optical spectrum, but also in tailoring the temporal profile of different wavelength components [4]. Yet, a platform that provides such flexibility on SC spectro-temporal control remains elusive.