A comparison of correlation and Lyapunov dimensions (Q707180)

The authors provide a detailed comparison of the correlation and the Kaplan-Yorke dimensions based on the intensive computational study of 46 different 3D chaotic systems with fractional dimensions which span the entire range from 2 to 3. For all these cases, the correlation dimension and the Lyapunov spectrum are calculated using the best methods available. By fitting these results with different forms of regression, the authors construct new Lyapunov dimensions that are found to correlate better to the correlation dimension \(D_{2}\) than the Kaplan-Yorke dimension \(D_{\text{KY}}\) does. A new Lyapunov dimension \(D_{\Sigma}\) that uses a quadratic interpolation instead of the linear one used for the derivation of \(D_{\text{KY}}\) is introduced and tested. This new Lyapunov dimension is found to correlate less well to \(D_{2}\) than \(D_{\text{KY}}\) or the Lyapunov dimension \(D_{\text{fit}}\) calculated using a least-squares linear regression and the Lyapunov dimension \(D_{\text{fit-C}}\) calculated using a regression with a constant term.