Time-series analysis of networks: exploring the structure with random walks

Area of Science:

• Network science
• Complex systems analysis
• Time series analysis

Background:

• Scale-free networks are ubiquitous in nature.
• Understanding network topology and function is crucial.
• Existing methods may not capture unified dynamical mechanisms.

Purpose of the Study:

• To develop a method for generating time series from networks.
• To investigate the relationship between network structure and time series properties.
• To explore a unified dynamical mechanism governing network organization.

Main Methods:

• Generating time series using a finite-memory random walk on scale-free networks.
• Analyzing temporal correlations and self-similar characteristics of generated time series.
• Applying multiscale analysis to classify diverse physical networks.

Main Results:

• Time series generated from networks reveal topological and functional properties.
• Assortative networks yield time series with long-range correlation.
• Disassortative networks yield time series with anticorrelation.
• Multiscale analysis successfully classifies networks based on functional origin.

Conclusions:

• Network node-degree mixing patterns directly influence time series correlations.
• A unified dynamical mechanism likely governs the structure of diverse networks.
• Time series analysis offers a powerful lens for network characterization and classification.