Analyticité séparée et prolongement analytique. (Separate analyticity and analytic extension) (Q913998)

This work deals with the old and always interesting question of separate analyticity. Definition. A subset E of a polydisc \(\Delta^ n\subset {\mathbb{C}}^ n\) is said to be separately analytic if for each hyperplane H parallel the axes \(E\cap H\) is empty or it is analytic in \(\Delta^ n\), with all irreducible components of dimension \(\geq 1.\) Definition. A function f: \(U\to X\) mapping a domain \(U\subset {\mathbb{C}}^ n\) into a complex reduced analytic space X is said to be separately analytic if for each hyperplane H parallel to axes the restriction of f to \(U\cap H\) is analytic. The author generalizes a theorem of Rothstein by getting rid of additional assumptions and obtains the following: Theorem 1.4. Each separately analytic subset of \(\Delta^ n\) is analytic in \(\Delta^ n.\) The question is, as always, more delicate for functions. Here the author gets: Proposition 1.8. A separately analytic mapping from a domain \(U\subset {\mathbb{C}}^ n\) to a complex analytic variety, f: \(U\to M\), such that \(\dim U\geq \dim M+2,\) is analytic. The methods of proof, inspired partly by the work of \textit{H. Alexander}, \textit{B. A. Taylor} and \textit{J. L. Ullman} [Invent. Math. 16, 335-341 (1972; Zbl 0238.32007)] and Rothstein [the author and \textit{W. Rothstein}, Lect. Notes Math. 683, 179-202 (1978; Zbl 0409.32010)] are very much likeable (especially to the reviewer) for being direct, clear and geometrical despite the fact that the proofs are not easy.