Wave Computing based on Dynamical Networks: Applications in Optimization Problems

TL;DR

This paper introduces a wave-based dynamical network computing framework that exploits wave propagation and manipulation to efficiently solve complex NP-hard optimization problems through intrinsic parallelism and multidimensional processing.

Contribution

It presents a novel wave computing architecture that extends parallelism into spatial, temporal, and frequency domains, validated by SPICE simulations for NP-hard problems.

Findings

Successfully simulated the architecture using SPICE

Demonstrated potential in solving NP-hard problems

Extended parallelism to multidimensional spaces

Abstract

We develop a computing framework that leverages wave propagation within an interconnected network, where nodes and edges possess wave manipulation capabilities, such as frequency mixing or time delay. This computing paradigm can not only achieve intrinsic parallelism like existing works by the exploration of an exponential number of possibilities simultaneously with very small number of hardware units, but also extend this unique characteristic to a multidimensional space including spatial, temporal and frequency domains, making it particularly effective for addressing NP-hard problems. The proposed architecture has been validated through SPICE simulations, demonstrating its potential capability in solving several NP-hard problems, such as the Number Partitioning Problem, the 0/1 Knapsack Problem, and the Traveling Salesman Problem.