Abstract
An annealing machine in which the Ising model is used to solve combinatorial optimization problems is proposed. In the annealing machine, combinatorial optimization problems are mapped to an Ising model that expresses the magnetic spins, and the problems are solved using the Ising model’s convergence property. A CMOS annealing machine, which implements annealing machines with CMOS semiconductor circuits, effectively solves the problems through a combination of deterministic and stochastic operations. An examination of a prototype chip confirmed that the CMOS annealing machine effectively yielded an approximate solution, and the power efficiency of its calculation was better than that of conventional von Neumann computers. In this article, a whole system for the realization of the technique is also described.
