Dynamics of two-resonant biholomorphisms (Q2449424)

Let \(F\) be the germ of a holomorphic self-map of \(\mathbb C^n\) fixing the origin \(0\) and such that \(dF_0\) is diagonal. One says that the germ is \(m\)-resonant provided the resonances among the eigenvalues of \(dF_0\) are generated by \(m\) linearly independent over \(\mathbb Q\) natural multi-indices. The \(1\)-resonant germs have been introduced and studied by \textit{D. Zaitsev} and the reviewer [J. Eur. Math. Soc. (JEMS) 15, No. 1, 179--200 (2013; Zbl 1351.37080)], and the \(m\)-resonant germs by \textit{J. Raissy, D. Zaitsev} and the reviewer [``Dynamics of multi-resonant biholomorphisms'', Int. Math. Res. Not. IMRN, no. 20, 4772--4797 (2013)]. It was proved that, under some generic condition, there exists a basin of attraction for \(F\) of parabolic type. The strategy used in those papers, and somewhat in the paper under review, is the following. Using the Poincaré-Dulac normal form, an \(m\)-resonant germ can be formally conjugated to a germ with only resonant terms which admits an invariant holomorphic foliation given by the submersion over \(\mathbb C^m\) defined by the generators of the resonances. Such a foliation is thus ``almost'' invariant under the original germ. The action of the map on the leaves space is parabolic. Therefore, under some conditions, one can use the classical Leau-Fatou flower theorem if \(m=1\) or Hakim's theory if \(m>1\) to find a basin of attraction in the leaf space and then, given some control on the fibers, one can obtain a basin of attraction (of parabolic type) for the \(m\)-resonant germ. There are two places in the previous argument where some additional conditions are required in order to make the strategy work: on the map defined by the action on the leaves space (the so-called ``parabolic shadow'' of \(F\)) and on the fibers direction. The hypotheses required on the parabolic shadow of \(F\) were essentially those which allowed to apply Hakim's theory. In the paper under review, the authors consider the case \(m=2\) and essentially release the hypotheses on the parabolic shadow of \(F\) in order to have a parabolic basin of attraction. Indeed, they allow the parabolic shadow of \(F\) to have a so-called ``irregular-non degenerate'' characteristic direction, so that they can use the results of the second named author [J. Geom. Anal. 22, No. 2, 352--382 (2012; Zbl 1281.32014)] in order to construct a basin of attraction for the parabolic shadow, which can be lifted to a parabolic basin of attraction for \(F\), providing control on the fibers is given. They also show that if the parabolic shadow has a ``degenerate'' characteristic direction, then there is no control along the fibers, so that the previous strategy, in such a case, cannot be exploited. And, in fact, as shown with an example both in the paper of Zaitsev and the reviewer [loc. cit.] and in this paper, under such a condition there exist \(m\)-resonant maps with no basins of attraction.