Angular spectra of linear dynamical systems in discrete time

DOI10.3934/DCDSB.2024144MaRDI QIDQ7012052zbMATH OpenFDO

Publication date 18 March 2025

Published in Discrete and Continuous Dynamical Systems. Series B (Search for Journal in Brave)

Full work available at URL https://doi.org/10.3934/DCDSB.2024144

angular spectrum ergodic average nonautonomous dynamical systems numerical approximation roughness theorems Sacker-Sell spectrum

Mathematics Subject Classification ID

Nonautonomous smooth dynamical systems (37C60) Attractors and repellers of smooth dynamical systems and their topological structure (37C70) Stability theory for smooth dynamical systems (37C75) Computational methods for ergodic theory (approximation of invariant measures, computation of Lyapunov exponents, entropy, etc.) (37M25) Discrete version of topics in analysis (39A12) Stability theory for difference equations (39A30) Chaotic behavior of solutions of difference equations (39A33)

Review text

In this paper the authors introduce and develop the concept of an angular spectrum for linear, invertible, nonautonomous discrete-time dynamical systems. Building on their earlier work on angular values -- which capture maximal long-time averaged principal angles between successive images of subspaces -- the authors extend the framework by considering all possible accumulation points of these angular averages as the initial subspace varies over the Grassmannian. The resulting outer angular spectrum provides a refined quantitative descriptor of rotational behavior complementary to classical spectral notions such as the Sacker-Sell (dichotomy) spectrum.\N\NThe article establishes fundamental properties of angular spectra, including relationships to previously studied extremal angular values, and shows how the computation can be reduced to a significantly smaller family of subspaces called trace spaces, derived from spectral bundles associated with the dichotomy spectrum. This reduction, based on earlier nonautonomous spectral theory, enables explicit determination of spectral sets in autonomous cases and certain structured nonautonomous examples. The authors compute angular spectra for several model systems, demonstrating that the spectra may consist of isolated points, intervals, or mixtures thereof, and exhibit sensitive dependence on system parameters.\N\NA central theoretical contribution is a roughness theorem showing that the angular spectrum is invariant under summable perturbations for systems possessing a complete exponential dichotomy (CED), a strong form of spectral splitting that generalizes diagonalizability from the autonomous case. The authors prove that summable perturbations preserve both the dichotomy spectrum and angular spectrum, relying on a detailed perturbation analysis for generalized exponential dichotomies and a kinematic similarity argument.\N\NThe paper also develops a numerical method for approximating angular spectra via finite-time averages, using the reduction to trace spaces to make computations tractable. Convergence of these approximations is analyzed, and applications are given to systems with homoclinic orbits and chaotic dynamics.\N\NOverall, the work provides a substantial and well-structured contribution to the qualitative theory of nonautonomous linear dynamics. The notion of angular spectrum enriches the geometric toolkit available for studying rotational behavior, and the combination of rigorous theory, explicit examples, and computational techniques makes the article a valuable reference for researchers in dynamical systems and spectral theory.

Cites work

This page was built for publication: Angular spectra of linear dynamical systems in discrete time

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7012052)