Taming a non-convex landscape with dynamical long-range order: memcomputing the Ising spin-glass
Recent work on quantum annealing has emphasized the role of collective behavior in solving optimization problems. By enabling transitions of large clusters of variables, such solvers are able to navigate their state space and locate solutions efficiently despite having only local connections between elements. However, collective behavior is not exclusive to quantum annealers, and classical solvers that display collective dynamics should also possess an advantage in navigating a non-convex landscape. Here, we propose a simple model that demonstrates this effect, based on the recently suggested digital memcomputing machines (DMMs), which utilize a collection of dynamical components with memory connected to represent the structure of the underlying optimization problem. This model, when applied to finding the ground state of the Ising spin glass, undergoes a transient phase of avalanches which can span the entire lattice. We then show that a full implementation of a DMM exhibits superior scaling compared to other methods when tested on the same problem class. These results establish the advantages of computational approaches based on collective dynamics.