Analysis and design of rhythmic neuromorphic networks through dominance and bifurcations

Omar Juarez-Alvarez, Alessio Franci

Published 2024-01-09Version 1

The control of oscillator networks capable of exhibiting complex rhythmic behaviors is a fundamental engineering problem motivated by the analysis and design of a variety of rhythmic biological and artificial systems. This work aims at introducing new theoretical tools, grounded in dominance analysis and bifurcation theory, to analyze and design biological and bio-inspired rhythmic networks. We derive constructive conditions under which the spectral properties of the network adjacency matrix fully and explicitly determine both the emergence of a network rhythm and its detailed profile (oscillator amplitudes and phases). The derived conditions can be used for analysis, prediction, and control of the rhythmic behavior of an existing network or for the design of a rhythmic network with a desired rhythmic behavior. The modeling framework under which we develop our theory is motivated by neuromorphic engineering, which makes our approach compatible with both the architecture of rhythmic biological networks and with the technological constraints needed to design bio-inspired rhythmic networks in compact and energy-efficient neuromorphic electronics.