Ising machines (IMs) are attracting huge attention as next-generation hardware for combinatorial optimization and machine learning. Photonic IMs exploit optics and promise ultrafast solutions to problems hard for digital processors. Realizing IMs is a major challenge in current research. A recent study in Physical Review Letters, led by Prof. Davide Pierangeli in collaboration with the group of Prof. Claudio Conti, proposes a novel class of IMs based on the polarization of light. Unlike existing paradigms, the polarization machine is high-dimensional and exhibits enhanced computing capabilities. Leveraging polarization effects in nonlinear media, the setup operates by an unprecedented computing mechanism called dimensional collapse. This concept enables high-dimensional nonlinear systems for Ising computing, paving the way for novel algorithms and processors for AI. The research is supported by the ERC StG 2024 project LOOP Optical polarization for ultrafast computing.
 

Open-access article: https://doi.org/10.1103/qs29-2xqc 
Contacts: Prof. Davide Pierangeli davide.pierangeli@roma1.infn.it