Array demonstrates path to photonic Ising computers

Photonic Ising computers have the potential to solve optimization problems at the speed of light that would take a traditional computer millions of years, if ever, to complete. These theoretical computers show strong potential for encoding spin states through orthogonal lasing polarization states. Vertical cavity surface-emitting lasers (VCSELs) have been proposed as light sources for photonic computing, but their inherent polarization preference and limited switching has kneecapped their polarization encoding ability.

Yuan et al. aimed to create a VCSEL that could switch freely between two orthogonal polarization states by engineering the aperture geometry to balance the array’s intrinsic anisotropy. They settled on a VCSEL array with a tailored oxidation aperture. This design had a more balanced modal gain between orthogonal polarization states, making it able to enhance polarization switching probability. Experiments showed the array could achieve polarization locking — the physical mechanism that enables interactions between Ising spins — at significantly lower power, making it scalable for optical computing systems.

“We are most excited that our tailored VCSEL arrays demonstrate clear potential for future Ising computing schemes, where polarization states can directly encode spin information,” said author Yuan Zifeng. “Our results show that polarization-switchable VCSELs can serve as the fundamental building blocks for energy-efficient photonic Ising computers, optical neural networks, and ultrafast signal processors.”

Zifeng notes that since the platform can be scaled up to large arrays, it can enable on-chip computing systems, which could solve optimization problems or perform AI-like computations at the speed of light. The group plans on continuing to work on a VCSEL-based Ising computer and also exploring other photonic computing architectures.

Source: “Tailored polarization-switchable VCSEL arrays for photonic ising computing,” by Zifeng Yuan, Tingting Chen, Dewen Zhang, Yuan Gao, Wenkai Shan, Beibei Lin, and Aaron Danner, Applied Physics Letters (2025). The article can be accessed at https://doi.org/10.1063/5.0291349 .