L0 regularization-based compressed sensing with quantum-classical hybrid approach

L0-regularization-based compressed sensing (L0-RBCS) is capable of outperforming L1-RBCS, but it is difficult to solve an optimization problem for L0-RBCS that cannot be formulated as a convex optimization. To achieve the optimization for L0-RBCS, we propose a quantum-classical hybrid system consisting of a quantum machine and a classical digital processor. Because forming a densely-connected network on a quantum machine is required for solving this problem, the coherent Ising machine (CIM) is one of suitable quantum machines for composing this hybrid system. To evaluate theoretically the performance of the CIM-classical hybrid system, a truncated Wigner stochastic differential equation (W-SDE) is obtained from the master equation for the density operator of the network of degenerate optical parametric oscillators, and macroscopic equations are derived from the W-SDE using statistical mechanics. We show that the system performance in principle approaches the theoretical limit of compressed sensing and in practical situations this hybrid system can exceed L1-RBCS's estimation accuracy.