Étale neighbourhoods and the normal crossings locus (Q533115)

One of the results in this article may be considered more or less known to specialists, but apparently it has no rigorous proof in the existing literature: Theorem. The normal crossings locus \(X_{nc}\) of an algebraic variety \(X\) is open in \(X\). It should be mentioned, that the corresponding analytic case is not that difficult, but here \(X\) is an algebraic variety over an algebraically closed field \(k\) of any characteristic. Applications of the above theorem can be found in the resolution theory of singularities. It is a typical way of reasoning to transfer analytic to algebraic results relating the algebraic local ring of \(x\in X\) and its completion or Henselization (the ``intermediate object''). They are obtained via étale topology and the celebrated Artin approximation theorem. This paper shows, there may still be surprises in that direction. The properties studied are local, thus it is sufficient to consider subvarieties \(X\) of the affine \(n\)-space. \(X\) is said to be algebraically normal crossings (anc) at \(p\in X(k)\), if it is defined near \(p\) by an equation \(y_1 \cdot \dots \cdot y_l\) (\(l\leq n\)), where \(y_i\) are local coordinates (i.e. a regular system of parameters for the local ring of \(X\) at \(p\)). \(X\) is said to be normal crossings (nc) at \(p\), if the completion \(\hat{X}_p\) is anc at \(p\). First of all it is shnown, that \(X_{anc}\) is open when \(X\) is a finite union of hypersurfaces. Then -- more generally -- sets of points \(q\in X\) are discussed, for which \(\hat{\mathcal O}_{X,q}\) has some given property. Are these subsets of \(X\) Zariski-open (resp. closed or locally closed)? Again the appropriate techniques arise from étale und formal neighbourhoods, related as mentioned above. This leads to the theorem on openness of \(X_{nc}\); as a corollary, the openness of the monomial locus \(X_{mon}\) of a hypersurface is shown. The article is illustrated with several instructive examples: It is natural to ask for other local properties of schemes which give rise to results similar as in the cases before. This is discussed here for Mikado points (as considered by H. Hauser) and for formally irreducible points. The examples worked out in detail show that such expectations fail in general.