Coherent Ising machine: a photonic Ising model solver based on degenerate optical parametric oscillator network
Friday, Jan 18, 12 – 1 pm, Elings 1605
As various systems and networks in our society grow larger and more complex, analysis and
optimization of such systems are becoming increasingly important. Such tasks are classified as combinatorial optimization problems, which are generally difficult to solve with current digital computers. It is well known that combinatorial optimization problems can be converted to ground-state-search problems of the Ising model, a theoretical model for the interacting spins. Recently, several approaches to find solutions to the Ising model using artificial spin systems have been studied intensively. A coherent Ising machine (CIM) is one of such systems in which degenerate optical parametric oscillators (DOPO) pulses are used as artificial spins. By using a long-distance (typically 1 km) fiber cavity that contains a phase sensitive amplifier based on a periodically poled lithium niobate waveguide, we can generate thousands of DOPO pulses multiplexed in time domain. Since a DOPO phase only takes either 0 or p at above threshold, we can stably express an Ising spin with a DOPO by allocating phase 0 (p) as spin up (down). The “spin-spin interaction” can be implemented by using a measurement-feedback scheme, with which we can effectively realize mutual injection of lights among thousands of DOPO pulses. The networked DOPOs are most likely to oscillate at a phase configuration that best stabilize the whole network, which gives the solution to the given Ising problem. Based on this scheme, we realized a CIM with all-to-all-coupled 2000 DOPO pulses, by which we could find good solutions to 2000-node combinatorial optimization problems in less than 100 microseconds. In the talk, I will describe the basic principle and the experimental details of the CIM, as well as our effort for finding its applications.