[Paper Review] An analytic KAM-Theorem

Joachim Albrecht|arXiv (Cornell University)|May 2, 2007

Quantum chaos and dynamical systems8 references3 citations

TL;DR

This paper presents a new analytic KAM-theorem tailored for the differential aspects of KAM theory, establishing the existence of quasi-periodic solutions in Hamiltonian systems under Diophantine frequency conditions. By solving a linearized equation iteratively and proving convergence via estimates on analytic functions and symplectic transformations, the theorem ensures the persistence of invariant tori under small analytic perturbations.

ABSTRACT

We prove an analytic KAM-Theorem, which is used in [1], where the differential part of KAM-theory is discussed. Related theorems on analytic KAM-theory exist in the literature (e. g., among many others, [7], [8], [13]). The aim of the theorem presented here is to provide exactly the estimates needed in [1].

Motivation & Objective

• To establish a precise analytic KAM-theorem tailored for the differential formulation of KAM theory as used in [1].
• To provide sharp estimates on the convergence of the iterative KAM process in the context of analytic Hamiltonian systems.
• To ensure the persistence of quasi-periodic solutions under small analytic perturbations of integrable systems.
• To prove that symplectic transformations generated by analytic vector fields preserve the structure of the system over time.
• To derive quantitative bounds on the size of the domain and the size of the perturbation for which the KAM iteration converges.

Proposed method

• Formulates the Hamiltonian as a sum of a normal form $ N = a + \langle \omega, y \rangle + \frac{1}{2}\langle y Q(x), y \rangle $ and a remainder $ R $, with $ R $ analytic and small.
• Imposes Diophantine conditions on the frequency vector $ \omega \in \Omega(\gamma, \tau) $ to ensure small divisors are controlled.
• Solves the linearized equation for the homological equation using estimates on analytic functions in complex domains $ \mathcal{D}(r,s) $.
• Applies an iterative scheme based on solving a time-dependent Hamiltonian flow generated by a vector field $ F $, with $ F $ affine-linear in $ y $.
• Uses estimates on the flow of $ F $, including bounds on the derivative $ Z_\zeta $, to prove convergence of the transformation sequence.
• Establishes that the resulting maps are symplectic and preserve analyticity and periodicity in the angle variables $ x $.

Experimental results

Research questions

• RQ1 Under what conditions does a quasi-periodic solution persist under small analytic perturbations of an integrable Hamiltonian system?
• RQ2 How can one derive sharp estimates on the size of the domain and the size of the perturbation for convergence of the KAM iterative scheme?
• RQ3 What are the necessary and sufficient conditions on the frequency vector $ \omega $ to avoid small divisor problems in the analytic setting?
• RQ4 How can one ensure that the symplectic transformations generated by the iterative process preserve analyticity and periodicity?
• RQ5 What quantitative bounds are required on the vector field $ F $ to guarantee the convergence of the KAM iteration in the analytic category?

Key findings

• The KAM iterative process converges under the assumption that the frequency vector $ \omega $ satisfies Diophantine conditions $ |\langle \omega, k \rangle| \geq \gamma / |k|^\tau $ for $ \tau > n-1 $.
• The resulting transformation maps the original domain $ \mathcal{D}(\varrho, \sigma) $ into a smaller domain $ \mathcal{D}(\varrho - 2\delta, \sigma/2) $, with convergence guaranteed for $ t < \sigma\delta / (2K) $.
• The maps generated by the flow of $ F $ are proven to be symplectic transformations, preserving the canonical structure of the system.
• The transformation $ Z(t, \cdot) - \text{id} $ is shown to be in $ \mathcal{P}_{2n}(\varrho - 2\delta, \sigma/2) $, ensuring analyticity and periodicity in $ x $.
• The estimates on the derivative $ |Z_\zeta(t, \zeta)| $ are bounded by $ \frac{2nK}{\delta\sigma} \exp\left(\frac{2nK}{\delta\sigma} t\right) $, controlling the growth of the transformation.
• The theorem guarantees that the final transformation is a simple canonical transformation when $ F $ is affine-linear in $ y $, ensuring the structure of the Hamiltonian is preserved.