All-optical scalable and programmable VCSEL-based Ising annealer with parallel feedback

Area of Science:

• Photonics
• Computational Science
• Nanotechnology

Background:

• Ising machines offer a promising approach for tackling large-scale combinatorial optimization problems.
• Existing electrical bottlenecks limit the scalability and efficiency of current computational methods.

Purpose of the Study:

• To propose and demonstrate an all-optical parallel feedback system for Ising machines.
• To overcome electrical bottlenecks by leveraging optical components for computation.

Main Methods:

• Utilizing mutual injection locking of vertical-cavity surface-emitting lasers (VCSELs).
• Employing programmable spatial light modulators (SLMs) for optical computation.
• Analyzing simulation results for a 16-bit square-lattice MAXCUT optimization problem.

Main Results:

• Successfully demonstrated optical injection locking between single-mode master and slave VCSELs.
• Achieved polarization switching with minimal master VCSEL injection power (<1%).
• Validated the potential for solving MAXCUT optimization problems via simulation.

Conclusions:

• The proposed all-optical system effectively addresses electrical bottlenecks in Ising machines.
• The proof-of-concept demonstrates feasibility for advanced optical computation.
• Future integration with nanophotonic devices like metasurface-enabled SLMs and VCSELs promises enhanced scalability and programmability.