The dynamic renormalization method for finding the maximum Lyapunov exponent of a chaotic attractor (Q2567817)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | The dynamic renormalization method for finding the maximum Lyapunov exponent of a chaotic attractor |
scientific article |
Statements
The dynamic renormalization method for finding the maximum Lyapunov exponent of a chaotic attractor (English)
0 references
13 October 2005
0 references
This paper is concerned with the explicit computation of the higher Lyapunov exponent of a chaotic attractor. In the classical algorithm proposed by \textit{Benettin} et al. [Phys. Rev. A 14, 2338--2345 (1976)] based on the formula \( \lambda_{\max}= \lim_{t \to + \infty} t^{-1} \log \| h(t) \| \) where \( \lambda_{\max}\) is the higher exponent of Lyapunov of an attractor \(\mathcal{A}\) of \( x' = f(x), (1)\) and \( h(t)\) is the solution of the variational equation of (1) along the trajectory \( \varphi^t (x_0)\) starting from \( x_0 \in \mathcal{A}\) with \( h(0)=h_0,\) \(\| h_0 \| =1.\) The solution of the variational equation is computed numerically and renormalized at fixed times \( n T, n=1,2,\ldots \) to approximate the above limit. The present authors propose an alternative algorithm in which the renormalization times are selected dynamically in the sense that e.g. in the case of \( \lambda_{\max}>0 \), for a given fixed number \( u_{\max} >1\), the variational equation is integrated until the first \( t_1\) in which \( \| h(t_1) \| = u_{\max}\), then after renormalization integration is continued until some \( t_2>t_1\) in which again \( \| h(t_2) \| = u_{\max}\) and so on. In order to justify this algorithm two theorems on the limit relation of the higher Lyapunov exponent (regardless of its sign) are given. The paper also includes some numerical experiments of the new proposed technique for a number of problems integrated numerically with the classical fourth order Runge-Kutta method with a fixed step size showing that the proposed dynamical approach may have some advantages over the fixed interval algorithm of Benettin et al. [loc. cit.].
0 references
chaotic attactors
0 references
Lyapunov exponents
0 references
renormalization method
0 references
comparison of methods
0 references
algorithm
0 references
numerical experiments
0 references
Runge-Kutta method
0 references