Linearly Stable but Nonlinearly Unstable: Resonances, Krein Signatures, and the Gap Between Spectral and Lyapunov Stability in Hamiltonian Systems

This paper examines the precise relationship between linear stability and Lyapunov stability for equilibria of analytic Hamiltonian systems. Linear stability means that the linearized equations have only bounded solutions, typically because all eigenvalues lie on the imaginary axis and the linearization is diagonalizable. Lyapunov stability is the stronger requirement that solutions starting sufficiently close to the equilibrium remain close for all future time. The central observation is that linear stability does not imply Lyapunov stability. When an elliptic–elliptic equilibrium possesses a low-order resonance of sum type whose resonant coefficient is nonzero, nonlinear terms can drive the actions to infinity even though the linearization itself is stable. Cherry’s classical Hamiltonian provides an explicit illustration: two oscillators in exact two-to-one resonance yield only closed orbits at linear order, yet the cubic resonant term produces solutions that start arbitrarily near the origin and escape to infinity in finite time. The same phenomenon appears in the planar circular restricted three-body problem at the triangular Lagrange points. Below the Routh critical mass ratio the equilibrium is linearly stable, yet at the specific resonant mass ratios corresponding to the two-to-one and three-to-one frequency relations it is Lyapunov-unstable. In both cases the Krein signatures of the modes determine whether a given frequency relation is of sum type (dangerous) or difference type (harmless and confining). The converse implication is more rigid. If the linearization possesses an eigenvalue with positive real part, then the equilibrium is Lyapunov-unstable for any continuously differentiable vector field; the linear divergence dominates near the origin. A borderline exception occurs when the spectrum is pure imaginary but the linearization is non-semisimple, so that linear solutions grow secularly. At the Routh value itself the triangular points furnish such an example: the linear flow is unbounded, yet the nonlinear equilibrium remains Lyapunov-stable because it is enclosed by invariant tori. All of the results are local and finite-order. They concern a neighbourhood of the equilibrium whose size is limited by the distance to the nearest dangerous resonance, and they do not extend automatically to systems with three or more degrees of freedom, strong dissipation, or global phase-space structures. Within these limits, the arithmetic of the frequency vector together with the Krein signatures completely classifies the local possibilities.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-04
DOI
https://doi.org/10.5281/zenodo.23965633
Primary Topic
Advanced Differential Equations and Dynamical Systems
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Linearly Stable but Nonlinearly Unstable: Resonances, Krein Signatures, and the Gap Between Spectral and Lyapunov Stability in Hamiltonian Systems

Doug Doucette
Zenodo (CERN European Organization for Nuclear Research)
Advanced Differential Equations and Dynamical Systems
preprint

Linearly Stable but Nonlinearly Unstable: Resonances, Krein Signatures, and the Gap Between Spectral and Lyapunov Stability in Hamiltonian Systems

Doug Doucette
preprint en

Abstract

This paper examines the precise relationship between linear stability and Lyapunov stability for equilibria of analytic Hamiltonian systems. Linear stability means that the linearized equations have only bounded solutions, typically because all eigenvalues lie on the imaginary axis and the linearization is diagonalizable. Lyapunov stability is the stronger requirement that solutions starting sufficiently close to the equilibrium remain close for all future time. The central observation is that linear stability does not imply Lyapunov stability. When an elliptic–elliptic equilibrium possesses a low-order resonance of sum type whose resonant coefficient is nonzero, nonlinear terms can drive the actions to infinity even though the linearization itself is stable. Cherry’s classical Hamiltonian provides an explicit illustration: two oscillators in exact two-to-one resonance yield only closed orbits at linear order, yet the cubic resonant term produces solutions that start arbitrarily near the origin and escape to infinity in finite time. The same phenomenon appears in the planar circular restricted three-body problem at the triangular Lagrange points. Below the Routh critical mass ratio the equilibrium is linearly stable, yet at the specific resonant mass ratios corresponding to the two-to-one and three-to-one frequency relations it is Lyapunov-unstable. In both cases the Krein signatures of the modes determine whether a given frequency relation is of sum type (dangerous) or difference type (harmless and confining). The converse implication is more rigid. If the linearization possesses an eigenvalue with positive real part, then the equilibrium is Lyapunov-unstable for any continuously differentiable vector field; the linear divergence dominates near the origin. A borderline exception occurs when the spectrum is pure imaginary but the linearization is non-semisimple, so that linear solutions grow secularly. At the Routh value itself the triangular points furnish such an example: the linear flow is unbounded, yet the nonlinear equilibrium remains Lyapunov-stable because it is enclosed by invariant tori. All of the results are local and finite-order. They concern a neighbourhood of the equilibrium whose size is limited by the distance to the nearest dangerous resonance, and they do not extend automatically to systems with three or more degrees of freedom, strong dissipation, or global phase-space structures. Within these limits, the arithmetic of the frequency vector together with the Krein signatures completely classifies the local possibilities.

Zenodo (CERN European Organization for Nuclear Research)
SAIT Polytechnic (CA)
Advanced Differential Equations and Dynamical Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon