Title: Coherent Ising machines – Quantum optics and neural network perspectives [arXiv, PDF]

Authors: Y. Yamamoto, T. Leleu, S. Ganguli, H. Mabuchi

Submitted on 10 June 2020.

Abstract: A non-equilibrium open-dissipative neural network, such as a coherent Ising machine based on mutually coupled optical parametric oscillators, has been proposed and demonstrated as a novel computing machine for hard combinatorial optimization problems. However, there are two challenges in the previously proposed approach: (1) The machine can be trapped by local minima which increases exponentially with problem size and (2) the machine fails to map a target Hamiltonian correctly on the loss landscape of a neural network due to oscillator amplitude heterogeneity. Both of them lead to erroneous solutions rather than correct answers. In this paper, we show that it is possible to overcome these two problems partially but simultaneously by introducing error detection and correction feedback mechanism. The proposed machine achieves efficient sampling of degenerate ground states and low-energy excited states via its inherent migration property during a solution search process.