Sequential spin-logic algorithmics for Ising machines.

Abstract

Ising machines are considered as promising hardware solvers. The idea is to map the solution of a problem to the ground state of an Ising spin glass and find this minimum by performing annealing in an Ising machine that emulates spin physics. However, previous approaches to deterministic sequential algorithmics have been plagued by the hiearchy problem: couplings become exponentially small along the causal direction. Here, we present a method to map sequential algorithms to an Ising problem with only a linear decay of the couplings and illustrate this with n-bit binary number adders. The resulting computing system is thus scalable, opening a way for efficient Ising machines for sequential algorithms.