Hamiltonian and reversible systems with smooth families of invariant tori

https://doi.org/10.1016/j.indag.2020.12.001Get rights and content

Abstract

For various values of n, d, and the phase space dimension, we construct simple examples of Hamiltonian and reversible systems possessing smooth d-parameter families of invariant n-tori carrying conditionally periodic motions. In the Hamiltonian case, these tori can be isotropic, coisotropic, or atropic (neither isotropic nor coisotropic). The cases of non-compact and compact phase spaces are considered. In particular, for any N3 and any vector ωRN, we present an example of an analytic Hamiltonian system with N degrees of freedom and with an isolated (and even unique) invariant N-torus carrying conditionally periodic motions with frequency vector ω (but this torus is atropic rather than Lagrangian and the symplectic form is not exact). Examples of isolated atropic invariant tori carrying conditionally periodic motions are given in the paper for the first time. The paper can also be used as an introduction to the problem of the isolatedness of invariant tori in Hamiltonian and reversible systems.

Section snippets

Kronecker tori

Finite-dimensional invariant tori carrying conditionally periodic motions are among the key elements of the structure of smooth dynamical systems with continuous time. The importance and ubiquity of such tori stems, in the long run, from the fact that any finite-dimensional connected compact Abelian Lie group is a torus [1], [17], [42]. By definition, given a certain flow on a certain manifold, an invariant n-torus carrying conditionally periodic motions (n being a non-negative integer) is an

Preliminaries

Given non-negative integers a and b, we designate the identity a×a matrix as Ia and the zero a×b matrix as 0a×b. In fact, the symbols I0, 00×b, and 0a×0 correspond to no actual objects and will only be used for unifying the notation.

Let s1, k, and l be non-negative integers and consider a skew-symmetric (2s+2k)×(2s+2k) matrix J of the form J=0s×sZtZL,where Z is an (s+2k)×s matrix of rank s and L is a skew-symmetric (s+2k)×(s+2k) matrix (the superscript “t” denotes transposing). If k=0 then

The system

Now let ζ1,,ζs, ξ1,,ξl, η1,,ηl be arbitrary non-negative real constants and let h:RsR be an arbitrary smooth function. Consider the Hamilton function H(u,p,q)=h(u)+lp1i=1sζiui2+ν=1l(ξνpνqν2+ηνpν33)on the symplectic manifold (5). The term lp1i=1sζiui2 is automatically absent for l=0. According to (6), the equations of motion afforded by H take the form u̇i=0,φ̇α=i=1sZαih(u)ui+2lζiuip1,ṗν=2ξνpνqν,q̇ν=ξνqν2+ηνpν2+δ1νli=1sζiui2,where 1is, 1αs+2k, 1νl, and δ1ν is the Kronecker

Compact phase spaces

Like in the setting of our note [50], the general construction of Sections 2 Preliminaries, 3 The main construction admits an analogue with a compact phase space. Consider the symplectic manifold M̂=Tus×Tφs+2k×Tpl×Tqlwith the same structure matrix (4). Of course, now the corresponding symplectic form Ω is always non-exact. The (s+2k)-tori (7) with u0Ts, p0Tl, q0Tl are again isotropic for k=0, are coisotropic for l=0, and are atropic for kl>0. For any angular variable z introduce the notation

Reversible analogues

Both the Hamiltonian systems (9), (11) are reversible with respect to the phase space involution G˜:(u,φ,p,q)(u,φ,p,q)of type (s+2k+l,s+l), so that dimFixG˜=s+l1, codimFixG˜=s+2k+ldimFixG˜, and codimFixG˜n=l<dimFixG˜, where n=s+2k. However, G˜(Tu0,p0,q0)=Tu0,p0,q0, so that not all the n-tori (7) are invariant under G˜. In the case of the system (9), a torus Tu0,p0,q0 is invariant under G˜ if and only if q0=0. Consequently, the statement “each torus Tu0,p0,q0M is symmetric” is valid if

Compactified reversible analogues

The system (14) can be compactified in the same way as the system (9), cf. [50]. Consider the manifold (3) equipped with the involution G given by the same formula (2) and having the same type (n+l,m). For any u0Tm and q0Tl, consider the n-torus Tu0,q0 given by the same expression (13). Since G(Tu0,q0)=Tu0,q0, a torus Tu0,q0 is invariant under G if and only if q0=q0, i.e., if each component of q0 is equal to either 0 or π.

Now let again ζ1,,ζm, ξ1,,ξl be arbitrary non-negative real

Acknowledgment

I am grateful to B. Fayad for fruitful correspondence and sending me the breakthrough preprint [21] prior to submission to arXiv.

References (54)

  • BarbarescoE.M. et al.

    Involutions whose fixed set has three or four components: a small codimension phenomenon

    Math. Scand.

    (2012)
  • BormanM.S. et al.

    Quasimorphisms on contactomorphism groups and contact rigidity

    Geom. Topol.

    (2015)
  • BounemouraA.

    Some instability properties of resonant invariant tori in Hamiltonian systems

    Math. Res. Lett.

    (2017)
  • BroerH.W. et al.

    Unfoldings of quasi-periodic tori in reversible systems

    J. Dynam. Differential Equations

    (1995)
  • BroerH.W. et al.

    Quasi-Periodic Motions in Families of Dynamical Systems. Order Amidst Chaos

    Lecture Notes in Mathematics

    (1996)
  • BroerH.W. et al.

    Unfoldings and Bifurcations of Quasi-Periodic Tori

    Mem. Amer. Math. Soc.

    (1990)
  • ChenL.

    On invariant tori in some reversible systems

    (2020)
  • de la LlaveR.

    A tutorial on KAM theory

  • DesideriP.E. et al.

    Involutions fixing many components: a small codimension phenomenon

    J. Fixed Point Theory Appl.

    (2017)
  • DuistermaatJ.J. et al.

    Lie Groups

    (2000)
  • DumasH.S.

    The KAM Story. A Friendly Introduction to the Content, History, and Significance of Classical Kolmogorov–Arnold–Moser Theory

    (2014)
  • EliashbergYa.M. et al.

    Introduction to the h-Principle

    (2002)
  • EliassonL.H. et al.

    Around the stability of KAM tori

    Duke Math. J.

    (2015)
  • FarréG. et al.

    Instabilities for analytic quasi-periodic invariant tori

    (2019)
  • FassòF.

    Superintegrable Hamiltonian systems: geometry and perturbations

    Acta Appl. Math.

    (2005)
  • FayadB.

    Lyapunov unstable elliptic equilibria, third version

    (2020)
  • FayadB. et al.

    Some questions around quasi-periodic dynamics

  • Cited by (2)

    View full text