Jacobian deformation ellipsoid and Lyapunov stability analysis revisited
Franz Waldner (1), Rainer Klages (2) ((1) Physics Institute,, University of Zurich, Switzerland, (2) Queen Mary University of London,, School of Mathematical Sciences, UK)

TL;DR
This paper revisits Lyapunov stability analysis, addressing its limitations in local instability detection and short-term analysis by introducing the Jacobian deformation ellipsoid and comparing it with existing methods.
Contribution
It introduces the Jacobian deformation ellipsoid and its orthogonal submatrix to improve local stability analysis without long-term integration, enhancing short-term and origin-specific instability detection.
Findings
The Jacobian deformation ellipsoid provides accurate local exponents.
Compared methods show improved detection of instability origins.
Short-term exponents closely match long-term results.
Abstract
The stability analysis introduced by Lyapunov and extended by Oseledec is an excellent tool to describe the character of nonlinear n-dimensional flows by n global exponents if these flows are stable in time. However, there are two main shortcomings: (a) The local exponents fail to indicate the origin of instability where trajectories start to diverge. Instead, their time evolution contains a much stronger chaos than the trajectories, which is only eliminated by integrating over a long time. Therefore, shorter time intervals cannot be characterized correctly, which would be essential to analyse changes of chaotic character as in transients. (b) Moreover, although Oseledec uses an n dimensional sphere around a point x to be transformed into an n dimensional ellipse in first order, this local ellipse has yet not been evaluated. The aim of this contribution is to eliminate these two…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
