Quantum neural network for combinatorial optimization and boltzmann sampling

Joint Event on 6^th International Conference on Photonics & 7^th International Conference on Laser Optics

July 31- August 02, 2017 Milan, Italy

Yoshihisa Yamamoto

Japan Science and Technology Agency, Japan

Posters & Accepted Abstracts: J Laser Opt Photonics

Abstract :

Many combinatorial optimization problems can be mapped onto ground-state-search problems of the Ising model or more general neural network. However, physical Ising machines are often based on localized spins, resulting in limited number of spin-spin couplings and severe scalability problems. We report 100�2,000 spin networks with all-to-all connections using time-multiplexed optical parametric oscillator pulses and measurement and feedback control. The developed machine can outperform simulated annealing in terms of accuracy and computation time for dense graphs.