Coupled oscillator networks for von Neumann and non von Neumann computing (2012.12386v1)

Abstract: The frenetic growth of the need for computation performance and efficiency, along with the intrinsic limitations of the current main solutions, is pushing the scientific community towards unconventional, and sometimes even exotic, alternatives to the standard computing architectures. In this work we provide a panorama of the most relevant alternatives, both according and not the von Neumann architecture, highlighting which of the classical challenges, such as energy efficiency and/or computational complexity, they are trying to tackle. We focus on the alternatives based on networks of weakly coupled oscillators. This unconventional approach, already introduced by Goto and Von Neumann in the 50s, is recently regaining interest with potential applications to both von Neumann and non von Neumann type of computing. In this contribution, we present a general framework based on the phase equation we derive from the description of nonlinear weakly coupled oscillators especially useful for computing applications. We then use this formalism to design and prove the working principle and stability assessment of Boolean gates such as NOT and MAJORITY, that can be potentially employed as building blocks for both von Neumann and non-von Neumann architectures.