Jiatong Li L&S Math & Physical Sciences
Optical Ising Machine with Nonlinear Optical Microdevice Array
Photons, with their natural parallelism as bosons, low heat generation, and extremely high speed, are an ideal medium for large-scale linear operations. Therefore, photonic computing can be a promising alternative for AI-related hardware because of the neural networks’ heavy reliance on matrix manipulations . However, one major challenge is implementing efficient nonlinear operations in photonic systems. This picture changed in 2025, when Qixin Feng’s group at UC Berkeley successfully developed the Nonlinear Optical Microdevice Array(NOMA), an electro-optical hybrid device that realizes nonlinear responses with femtojoule-level energy consumption using incoherent light.
Our project aims to build a photonic Ising machine with this cheap non-linearity from NOMA. This system simulates the evolution of electron spin states toward their ground state under the competing impacts from ferromagnetic and antiferromagnetic couplings. By combining photonic linear computation with NOMA-based nonlinearity, the system may enable energy-efficient optical computing and provide a new platform for solving NP-class optimization problems.
Message To Sponsor
Thank you for funding my exploration of this photonic computing approach. Photonics has been my core research interest, and this marks a meaningful starting point for me to apply its parallel and energy-efficient nature to computational hardware—one of the key challenges of our time. I am excited to see where this idea will lead and what it can contribute to the contemporary technological landscape.