On a family of real hypersurfaces in a complex quadric (Q2636745)

The paper finishes the question whether the unit sphere \(S^{2n-1}\) is up to CR-equivalence the only connected and simply-connected compact real-analytic hypersurface in \(\mathbb C^n\) which is homogeneous with respect to a Lie group of CR-transformations, \(n\geq 2\). By the classification theorems of \textit{E. Cartan} [Ann. Sc. Norm. Super. Pisa, II. Ser. 1, 333--354 (1932; Zbl 0005.37401)] and \textit{A. Morimoto} and \textit{T. Nagano} [J. Math. Soc. Japan 15, 289--300 (1963; Zbl 0119.06701)] the answer to this question is affirmative for \(n\not\in\{3, 7\}\). The results of the article say that the answer is also positive for \(n=7\), but negative for \(n=3\). The hypersurface orbits of the natural action of \(SO(n+1,\mathbb R)\) on the affine quadric \[ Q^n=\bigg\{(z_1,\dots,z_{n+1})\in\mathbb C^{n+1} \;\bigg|\; \sum_{k=1}^{n+1}z_k^2=1\bigg\}, \] i.e. the members of the family \[ M_t^n=\bigg\{(z_1,\dots,z_{n+1})\in Q^n\;\bigg|\; \sum_{k=1}^{n+1}|z_k|^2=t\bigg\},\quad t>1, \] are the only possible examples for hypersurfaces in \(\mathbb C^n\) with the above properties, and they are pairwise CR-inequivalent. The author excludes these candidates for \(n\not=3\) by showing that any real-analytic CR-embedding of \(M_t^n\) into \(\mathbb C^n\) for some \(t>1\) extends to a biholomorpic embedding of the Stein domain \[ \bigg\{(z_1,\dots,z_{n+1})\in Q^n\;\bigg|\; \sum_{k=1}^{n+1}|z_k|^2<t\bigg\}. \] This implies that \(S^n\) admits a real-analytic totally real embedding into \(\mathbb C^n\) and therefore is parallizable, hence \(n\in\{3,7\}\). But a real-analytic totally real embedding \(S^7\rightarrow \mathbb C^7\) is impossible, see [\textit{E. L. Stout} and \textit{W. R. Zame}, Adv. Math. 60, 154--160 (1986; Zbl 0589.32026)]. \textit{P. Ahern} and \textit{W. Rudin} [Proc. Am. Math. Soc. 94, 460--462 (1985; Zbl 0567.32006)] constructed a totally real embedding of \(S^3\) into \(\mathbb C^3\), induced by the map \(\mathbb C^2\rightarrow \mathbb C^3\), \((z,w)\mapsto (z,w,\overline{z}w \overline{w}^2+iz\overline{z}^2\overline{w})\). The analysis of this map leads to the result that there exists \(\delta>1\) such that \(M_t^3\) admits a real-analytic CR-embedding into \(\mathbb C^4\) for \(1<t<\delta\). The author conjectures that \(\delta\geq\sqrt{(2+\sqrt{2})}\Big/3\).