Almost-periodic perturbations of non-hyperbolic equilibrium points via Pöschel-Rüssmann KAM method (Q2332319)

The article is focused in the generalization of two previous results of quasi-linear ODE systems to the almost periodic framework. Firstly, the authors recall a result of \textit{À. Jorba} and \textit{C. Simó} [SIAM J. Math. Anal. 27, No. 6, 1704--1737 (1996; Zbl 0863.34043)], which proves the existence of quasi-periodic solutions of the perturbed linear system: \[ x'=Ax+h(t,x)+f(t,x,\epsilon), \quad x\in \mathbb{R}^{n},\tag{1} \] where \(A\) is a nonsingular matrix with purely imaginary and different eigenvalues \(\lambda_{i}\), the functions \(f\) and \(h\) are quasi-periodic in \(t\) with frequency \(\omega\in \mathbb{R}^{d}\). Moreover, \(f\) is small for small values of \(\epsilon\) and \(h=\mathcal{O}(x^{2})\) when \(x\to 0\). Secondly, the authors recall a result of \textit{W. Si} and \textit{J. Si} [Nonlinearity 31, No. 6, 2361--2418 (2018; Zbl 1401.37067)], which gives sufficient conditions ensuring the existence of quasi-periodic solutions of the system \[ x'=\phi(x)+h(t,x)+f(t,x,\epsilon), \quad x\in \mathbb{R}^{n},\tag{2} \] where \(\phi(x)=(a_{1}x_{1}^{l_{1}},\ldots,a_{n-1}x_{n-1}^{l_{n-1}},a_{n}x_{n}^{l_{n}})^{T}\) with \(a_{i}\in \mathbb{R}\setminus\{0\}\) and \(l_{i}\in \mathbb{Z}^{+}\setminus\{0\}\), \(l_{n}\) satisfies the property \(l_{n}\geq \bar{l}=\max\{l_{1},\ldots,l_{n-1}\}\), \(h=\mathcal{O}(x^{l_{n}+1})\) and \(f(\theta,x,0)=0\). In addition, the authors assume that either \(\det D\phi(0)=0\) (partially degenerate condition) or \(D\phi(0)=0\) (degenerate condition). The generalization of the above results to the almost periodic framework arise the following questions: \textbf{(Q1)} For the almost-periodically forced system (1), can we get the existence of almost-periodic response functions with the same frequency \(\omega\) as forcing? \textbf{(Q2)} For the almost-periodically forced degenerate system (2), can we get the existence of almost-periodic solutions with the same frequency \(\omega\) as forcing? \textbf{(Q3)} Can the almost-periodically forced degenerate system (2) admit almost-periodic solutions with the same frequency \(\omega\) as forcing under non-resonant conditions (weaker of some previous ones) or without any non resonant condition? The question \textbf{(Q1)} is partially addressed in Section 3, which considers the system \[ x'=Ex+h(t,x)+f(t,x,\epsilon)\tag{3} \] where \(E\) is a matrix with purely imaginary and different eigenvalues \(\lambda_{i}\), \(f\) and \(h\) are almost periodic in \(t\). Moreover, \(f\) is small for small values of \(\epsilon\) and \(C^{1}\)-smooth with respect to \(\varepsilon\), \(h=\mathcal{O}(x^{2})\) when \(x\to 0\). Finally, \(h\) and \(f\) admits the same spatial series expansion. The authors state that by adapting the Pöschlel-Rüssmann KAM method [\textit{J. Pöschel}, Regul. Chaotic Dyn. 16, No. 1--2, 17--23 (2011; Zbl 1219.37044)]. System (3) can be transformed into \[ \left\{\begin{array}{rcl} \theta' & = & \omega, \\ x' & = & (E+Q(\theta,\epsilon))x+G(\theta,x,\epsilon)+F(\theta,\epsilon) \end{array}\right.\tag{4} \] with \[ Q(\theta,\epsilon)=\frac{\partial f}{\partial x}(\theta,0,\epsilon), \quad F(\theta,\epsilon)=f(\theta,0,\epsilon) \] and \[ G(\theta,x,\epsilon)=h(\theta,x)+f(\theta,x,\epsilon)-f(\theta,0,\epsilon)-\frac{\partial f}{\partial x}(\theta,0,\epsilon)x, \] which are analytic on the set \(\triangle_{s,r}=\mathbb{T}_{s}^{\mathbb{Z}}\times B_{r}\), where \(B_{r}\subset \mathbb{C}^{n}\) is a closed ball centered at the origin with radius \(r\) and \[ \mathbb{T}_{s}^{\mathbb{Z}}=\left\{\theta=(\cdots,\theta_{\lambda},\cdots)\in\mathbb{C}^{\mathbb{Z}}/(2\pi\mathbb{Z})^{\mathbb{Z}} \colon \sup_{\lambda\in \mathbb{Z}}|Im\,\theta_{\lambda}|\leq s\right\} . \] The main result of Section 3 is Theorem 3.1, which formulates several conditions (non-resonants conditions and non-degeneracy conditions) ensuring the existence of an analytic transformation, which for small values of \(\epsilon\) transforms (4) in \[ \left\{\begin{array}{rcl} \theta' & = & \omega \\ z' & = & E_{*}(\epsilon)+h_{*}(\theta,z,\varepsilon). \end{array}\right.\tag{5} \] The questions \textbf{(Q2)}--\textbf{(Q3)} are considered in Section 4, where the authors state that System (2) is equivalent to \[ \left\{\begin{array}{rcl} \dot{\theta} & = & \omega, \\ \dot{x} & = & \phi(x)+h(\theta,x)+f(\theta,x,\epsilon), \end{array}\right.\tag{6} \] where \(\phi\), \(h\) and \(f\) have the same properties as in (2). In particular, if \(\epsilon=0\) it follows that \(f(\theta,x,0)=0\) and System (6) has the origin as equilibrium and has the invariant torus \[ \mathcal{T}_{0}=\mathbb{T}^{\mathbb{Z}}\times \{0_{n}\}. \] The authors state [p. 564] that the existence of almost periodic response solutions System (2) is equivalent to the persistence of such a degenerate torus \(\mathcal{T}_{0}\) under a sufficiently small perturbation \(f\) or equivalently that System (6) has a degenerate response torus. Theorem 4.1, Theorem 4.2 and Theorem 4.3 provide a set of technical conditions ensuring that System (6) has a degenerate response torus. This three results impose smallness assumptions on \(h\) and technical assumptions on the partial derivatives of \(f\) with respect to \(\epsilon\). Nevertheless, only Theorems 4.1 and 4.2 consider non-resonance conditions.