Discriminating Chaotic Time Series with Visibility
Graph Eigenvalues
Vincenzo Fioriti
Alberto Tofani
Antonio Di Pietro
Italian National Agency for New Technologies, Energy and Sustainable
Economic Development (ENEA), CR Casaccia Labs
S. Maria in Galeria
301,00130 Rome, Italy
Abstract
Time series can be transformed into graphs called horizontal visibility graphs (HVGs) in order to gain useful insights. Here, the maximum eigenvalue of the adjacency matrix associated to the HVG derived from several time series is calculated. The maximum eigenvalue methodology is able to discriminate between chaos and randomness and is suitable for short time series, hence for experimental results. An application to the United States gross domestic product data is given.
