Fast Forward Design of Randomly Coupled Multi-Core Fibers by Machine Learning

A neural network and algorithmic-assisted method is proposed to quickly design randomly coupled multicore fibers. Compared with traditional design methods, this method has good design accuracy and significantly saves time and computing resources.