sipkit.pdf (290.47 kB)
Silicon Photonics Toolkit
preprintposted on 2023-07-19, 14:54 authored by Aycan Vit, Kazim Gorgulu, Ali Amiri, Emir Salih Magden
In integrated optics, accessing optical parameters of waveguides is one of the most important yet recurring tasks in the process of designing photonic devices. Silicon Photonics Toolkit (sipkit) is a python package that provides computationally efficient access to waveguide parameters as functions of key optical and physical variables, to aid the design of silicon photonic systems and scientific discovery through integrated optics. In addition to its state-of-the-art data mapping for rapid parameter access, sipkit also allows for full automatic-differentiation capability through its compatibility with JAX. With efficient gradient computation for optimization algorithms, sipkit can therefore be used in the design of photonic systems using modern machine learning methods. With this combination of streamlined data access and support for automatic differentiation, sipkit enables researchers and engineers to design complex photonic systems with greater efficiency and flexibility.