A self-similar sine–cosine fractal architecture for multiport interferometers

Abstract Multiport interferometers based on integrated beamsplitter meshes have recently captured interest as a platform for many emerging technologies. In this paper, we present novel architecture multiport the sine–cosine fractal decomposition of unitary matrix. Our is unique in that it self-similar, enabling construction modular multi-chiplet devices. Due to modularity, our design enjoys improved resilience hardware imperfections compared conventional interferometers. Additionally, structure circuit enables systematic truncation, which key reducing footprint chip well compute time training optical neural networks, while maintaining full connectivity. Numerical simulations show truncation these gives robust performance even under large fabrication errors. This step forward large-scale programmable photonics, removing major hurdle scaling up practical machine learning and quantum computing applications.