Can Nonlinear Parametric Oscillators Solve Random Ising Models?

We study large networks of parametric oscillators as heuristic solvers random Ising models. In these networks, known coherent machines, the model to be solved is encoded in coupling between oscillators, and a solution offered by steady state network. This approach relies on assumption that mode competition steers network ground-state model. By considering broad family frustrated models, we show most-efficient does not correspond generically ground infer close threshold are intrinsically solvers. Nevertheless, can find correct if driven sufficiently above threshold, regime where nonlinearities play predominant role. for all probed instances model, converges with finite probability.