Convergence of the Chern-Moser-Beloshapka normal forms (Q2195130): Difference between revisions

In [\textit{S. S. Chern} and \textit{J. K. Moser}, Acta Math. 133, 219--271 (1975; Zbl 0302.32015)] and [\textit{N. Tanaka}, J. Math. Soc. Japan 14, 397--429 (1962; Zbl 0113.06303)] the classification problem for Levi-nondegenerate hypersurface in \(\mathbb{C}^n\) was solved. The aim of the present paper is to study the case of higher co-dimension. First, recall that a real submanifold of \(\mathbb{C}^n\) is given locally by \[ \mathrm{Im}\,w=\varphi(z,\,\bar{z},\mathrm{Re}\,w) \] where \(\varphi\) is a germ of a real analytic map and it is assumed to be Levi-nondegenerate. It can be though as a local deformation of a hyperquadric \(\mathrm{Im}\,w= Q(z,\,\bar{z})\) and can be written locally as \[ \mathrm{Im}\,w= Q(z,\,\bar{z})+\Phi_{\geq3}(z,\,\bar{z},\mathrm{Re}\,w)\,. \] In the first theorem it is proved that there exists a formal biholomoprhic change of coordinates \(H(z,w)=(z',w')\) such that \(\Phi_{\geq3}\) can be taken inside a normal form space \(\hat{\mathcal{N}}_f\), which can be described explicitly. Convergence of the formal normal form is guaranteed under the condition \[ \Phi'_{1,1}\,\Phi_{1,2}+\Phi'_{1,2}(Q+\Phi_{1,1})=0 \] where \(\Phi=\sum_{j,\,k}\Phi_{j,\,k}\) (\(\Phi_{j,\,k}(tz,\,s\bar{z},\,u)=t^j\,s^k\,\Phi_{j,\,k}(z,\,\bar{z},\,u)\)) and \[ \Phi'_{j,\,k}(z,\,\bar{z},\,u)=\left(\frac{\Phi_{j,\,k}}{\partial u_1},\dots,\frac{\Phi_{j,\,k}}{\partial u_d}\right)\,. \]