Holomorphic extension of CR functions from quadratic cones (Q927266)

A (real) quadratic cone in \({\mathbb C}^n\) is an irreducible real algebraic subset \(M \subset {\mathbb C}^n\) of pure dimension \(2n-1\) defined as \(M = \{\rho(z) = 0\}\) where \(\rho\) is a real valued homogeneous polynomial of degree \(2\) in \(\Re e (z_j)\) and \(\Im m(z_j)\). A function \( f \in L^1_{\text{loc}}(M)\) on a quadratic cone \(M\) is called CR function if \(\int_{M_{\text{reg}}} f \,\overline\partial \varphi = 0\) for any smooth \((n, n-2)\)-form \(\varphi\) with compact support in \({\mathbb C}^n\). In this paper, the authors tackle the problem of one-sided holomorphic extensions of CR functions \(f\) on hypersurfaces \(M \subset {\mathbb C}^n\) with singularities and they solve it completely when \(M\) is a quadratic cone and \(f\) is defined on a neighborhood of \(0\). The main result is the following. Given a point \(p\) of a real analytic set \( M \), a neighborhood \(\mathcal U\) of \(p\) and a real analytic defining function \(\rho\) so that \( M \cap {\mathcal U} = \{\;z \in {\mathcal U} : \rho = 0\;\}\), let us denote by \({\mathcal U}^{\pm} = \{\;z \in {\mathcal U} : \pm \rho(z) > 0\;\}\) and say that \(M\) admits a two-sided support at \(p\) if there exist germs at \(p\) of complex analytic hypersurfaces \(A^+\) and \(A^-\) such that \(A^{\pm} \subset\overline{ \mathcal U^{\pm}}\). Notice that if \(M\) is \({\mathcal C}^1\), then it admits a two-sided support if and only it is non-minimal around \(p\). Now, if \(M\) is a quadratic cone, then there exists a neighborhood \(\mathcal U\) of the origin so that either all functions of \({\mathcal O}(\mathcal U^+)\) or all functions of \({\mathcal O}(\mathcal U^-)\) extend to a neighborhood of \(0\) if and only if \(M\) does not have a two-sided support at the origin. From this and the jump formula for CR functions, one obtains immediately that in this case every CR function on \({\mathcal U} \cap M\) has a holomorphic extension either to \({\mathcal U}^+\) or \({\mathcal U}^-\). The proof is based on arguments that reduce the problem to questions in complex dimension \(2\) and on an explicit classification of the quadratic cones \(M \subset {\mathbb C}^2\) up to biholomorphisms.