Invariant manifolds of degenerate tori and double parabolic orbits to infinity in the \((n+2)\)-body problem (Q6560728)

This paper considers vector fields of the form \N\[\NX(x,y,\theta ) = \begin{pmatrix} f^N(x,y,\theta , \lambda) + O(\|(x,y) \|)^{N+1}\\\Ng^M(x,y,\theta , \lambda) + O(\|(x,y) \|)^{M+1}\\\N\omega + h^{\geq P}(x,y,\theta , \lambda) \end{pmatrix},\tag{1}\N\]\Nand maps that have the form \N\[\NF_\lambda \left(\begin{array}{c} x \\\Ny \\\N\theta \end{array} \right) = \left(\begin{array}{c} x+f^{\geq N}(x,y,\theta , \lambda) \\\Ny+g^{\geq M}(x,y,\theta , \lambda) \\\N\theta + \omega + h^{\geq P}(x,y,\theta , \lambda) \end{array} \right),\tag{2}\N\]\Nwhere \((x, y)\in B\subset {\mathbb R}^n\times {\mathbb R}^m\), B is a ball around the origin, \(\theta\in {\mathbb T}^d =({\mathbb R}/2\pi {\mathbb Z})^d\), \(\lambda\in {\mathbb R}^p\), \(f^N\) , \(g^M\), and \(h^P\) are functions of orders \(N\), \(M\), and \(P\) in \((x, y)\), respectively.\N\NThe authors present a general theorem which, under suitable conditions, provides the existence of invariant manifolds of the invariant torus \({\mathcal T}\) for vector fields of the form (1) (and for maps with equivalent conditions given in (2)). In addition, it shows the existence of new type of orbits in the \(N\)-body problem, defined for all time either in the future or in the past, with a prescribed final behaviour. These orbits are called \textit{double parabolic orbits to infinity}.\N\NIn order to find the invariant manifolds of \({\mathcal T}\) in (1), the authors apply the parametrization method given, among others, in [\textit{X. Cabré} et al., Indiana Univ. Math. J. 52, No. 2, 283--328 (2003; Zbl 1034.37016); Indiana Univ. Math. J. 52, No. 2, 329--360 (2003; Zbl 1034.37017)].\N\NThe existence of the manifolds is formulated as an \textit{a posteriori}, that is, in Theorem 2.7, for maps, or Theorem 2.14, for flow. Separately, Theorem 2.8 (Theorem 2.16, in the case of flows) provides such approximation. Finally, Theorems 2.9 and 2.16 simply join the \textit{a posteriori} and the approximation results into an existence result, to ease their application in practice.\N\NThe authors claim that the main differences among the results in the present paper and those in their previous papers [J. Differ. Equations 268, No. 9, 5516--5573 (2020; Zbl 1437.37022); J. Differ. Equations 268, No. 9, 5574--5627 (2020; Zbl 1437.37023)] are the following. First, instead of considering parabolic fixed points, here they consider parabolic tori. This is a non-trivial extension that widens the field of application of the results. Also, they are interested in particular in the case where the dynamics on the manifold synchronizes with the one on \({\mathcal T}\). This fact, that always happens in the hyperbolic case, may not occur in the parabolic one. Their theorem is also valid even when this synchronization does not take place, and conditions under which it happens are given. In this sense, the authors improve their previous results in [Commun. Math. Phys. 374, No. 1, 63--110 (2020; Zbl 1465.37098)], where only the cases where the synchronization occurs where considered. One of the consequences of synchronization is that then the invariant manifolds are foliated by the stable leaves of the points in the torus and this foliation is regular in the base.\N\NSecond, the vector field need not be defined in a whole neighborhood of the torus, not even at a formal level. It only requires some kind of regularity in sectorial domains with the torus at their vertex, expressed in terms of homogeneous functions. It is required that the main terms be defined and regular around the torus, although the authors believe that this requirement may be relaxed and they impose it for convenience, since it holds in the examples they consider. Third, to simplify the proof, in this article only the analytical case is considered.\N\NAs an example, in the last section (Section 5), the main theorem is used to prove the existence of double parabolic motions to infinity in the (\(n + 2\))-body problem in which the distances among the first \(n\) bodies remain bounded for all time, while the relative distances between the first \(n\) bodies and the last two and the distances between the last bodies tend to infinity, when time goes to infinity. Moreover, it proves that the final motion of the first \(n\) bodies corresponds to a KAM torus of the \(n\)-body problem. It is done by finding suitable coordinates, which include a normal form procedure and blown-up, in which the general theorem applies.