Proposal of a quantum-annealing-inspired recurrent neural network with optical decision-making for swarm-intelligence-like ground-state searching of an Ising model

摘要

We focus on a sign of swarm intelligence existing in a path-integral quantum Monte Carlo method (QMCM)(also known as simulated quantum annealing) to demonstrate schedule-free swarm-intelligence-like ground-statesearching of an Ising model. Replicas representing different points in imaginary time help mutually to search for groundstatecandidates as if they form a swarm for cooperative work. Replica-replica interactions are local, simple and fluctuateto a certain degree. We propose an optical decision-making method to realize such replica-replica interactions.