Divergent CR-equivalences and meromorphic differential equations (Q502149): Difference between revisions

Consider two real-analytic generic CR-submanifolds \(M\) and \(M'\) in \(\mathbb{C}^N\), \(N\geq 2\). Both manifolds are assumed to pass through the origin. Two germs \((M,0)\) and \((M',0)\) are called \textit{holomorphically equivalent} if there exists a germ of an invertible holomorphic \(F: (\mathbb{C}^N,0)\to (\mathbb{C}^N,0)\) such that \(F(M)\subset M'\). This map is called a \textit{holomorphic equivalence between \((M,0)\) and \((M',0)\).} A \textit{formal mapping} \(F: (\mathbb{C}^N,0)\to (\mathbb{C}^N,0)\) is one whose components are formal power series. When this is the case and \(F\) is invertible, \((M,0)\) and \((M',0)\) are called \textit{formally equivalent}. It is clear that holomorphic equivalence implies formal equivalence but the converse is not a priori immediate. It is of great interest to understand the relationship between holomorphic and formal classifications. For example it is known that, under the hypothesis of Levi-nondegeneracy, formal equivalences must in fact be convergent. In [\textit{M. S. Baouendi} et al., J. Geom. Anal. 12, No. 4, 543--580 (2002; Zbl 1039.32021)] it was conjectured that the groups \(\mathrm{Aut}(M,p)\) and \(\mathcal{F}(M,p)\) of holomorphic and formal self-equivalences of the germ \((M,p)\) coincide for a holomorphically nondegenerate hypersurface \(M\). A related question in CR-geometry asks whether the local holomorphic classification of real-analytic CR-manifolds of CR-dimenion \(n\) and CR-codimension \(k\) coincide with the formal classification. In [\textit{N. Mir}, Methods Appl. Anal. 21, No. 4, 481--502 (2014; Zbl 1302.32017)] a positive answer was conjectured. The main results of this paper give counterexamples to these conjectures. The methods of the paper involve tools from a broad range of topics in complex analysis and dynamical systems, in particular, the analytic theory of differential equations.