Discrete-state phasor neural networks

Abstract

An associative memory network with local variables assuming one of $q$ equidistant positions on the unit circle ($q$-state phasors) is introduced, and its recall behavior is solved exactly for any $q$ when the interactions are sparse and asymmetric. Such models can describe natural or artificial networks of (neuro-)biological, chemical, or electronic limit-cycle oscillators with $q$-fold instead of circular symmetry, or similar optical computing devices using a phase-encoded data representation.