Dimensionality Reduction in Stochastic Complex Dynamical Networks

Abstract

Complex systems are ubiquitous in nature and engineering, but their analysis and control are hampered by their high dimensionality and the influence of various factors on their dynamics. Dimensionality reduction aims to find a low-dimensional representation of the complex system that preserves its essential features and reveals its underlying mechanisms and long-term dynamics. However, most existing methods for dimensionality reduction assume deterministic systems, while many real-world systems exhibit stochasticity. Here, we develop a general analytical framework for dimensionality reduction of stochastic complex dynamical networks that can map a high-dimensional system with stochastic terms to a low-dimensional effective system with a single effective state variable and few effective parameters. The effective parameters are those that determine the network's dynamical behavior and are associated with specific system states. The effective equation is a low-dimensional representation of the original stochastic complex dynamical network that preserves its essential dynamical features. The framework also allows us to analyze the dynamic behavior and potential convergence of the stochastic complex dynamical network by using the standard deviation of the effective equation.