The Gorenstein locus of minuscule Schubert varieties (Q960574)

An important task in the study of a homogeneous varieties is that of describing the singular locus of its Schubert varieties. There are many results regarding such a problem, but also many aspects needing further investigations. Recall that a homogeneous variety, i.e. a complete variety of the form \(G/P\), where \(P\) is a parabolic subgroup of an algebraic group \(G\), is said of type \(A\) if its Weyl group \(W_P\) is of type \(A_{n}\) for some positive integer \(n\). In case of homogeneous spaces of type \(A\) there are several results regarding the singular locus of Schubert varieties and many guesses and/or conjectures have been formulated in the recent literature. The paper under review is concerened with a recent conjecture by Woo and Yong regarding the Gorenstein locus of a Schubert variety \(X(w)\) of a type \(A\) homogeneous space. Recall that a point of an algebraic variety is Gorenstein if and only if its local ring is Gorenstein, i.e. if it has finite injective dimension when thought of as a module over itself. Gorenstein singularities are milder than other singularities, because they simulate many behaviours occurring for smooth varieties. The conjecture by Yong and Woo claims that the non-Gorenstein locus of \(X(w)\) is the union of irreducible components whose generic point is non Gorenstein. This amounts to say, in other words, that once one knows the irreducible components of the singular locus, one has enough information to detect the Gorenstein locus. The author studies the aforementioned conjecture by using a combinatorial tool introduced in [\textit{N.~Perrin}, Compos. Math. 143, No. 5, 1255--1312 (2007; Zbl 1129.14069)]. In fact it is possible to associate to each minuscule Schubert variety a quiver which generalizes Young diagrams. The key notion to be defined is that of holes, virtual holes and essential holes. A quiver is a graph, and a hole is any vertex satisfying certain properties, which is not the case to explain here. However, even in the introduction the author makes very nice and illustrative examples already for the case of grassmannians, the most popular kind of minuscule homogeneous space. This allows the author to prove the aforementioned conjecture by Yong and Woo for minuscule Schubert varieties. It is indeed a consequence of the following important theorem, the main result of the paper: The generic point of a Schubert subvariety \(X(w')\) of a minuscule Schubert variety \(X(w)\) is in the Gorentsein locus if and only if the quiver of \(X(w')\) contains all the non non-Gorentein holes of the quiver \(X(w)\). The paper is very well written and very well organized, and the way it adresses to the readers is quite friendly, in the usual author's style. Not only in Section 1 there are all the basics one needs to follow the path drawn by the author, but Section 2 shows through examples how to obtain informations from the quivers associated to Schubert varieties, including the use of the holes introduced in Definition 1.8. Very interesting is the description of the relative canonical model of a Schubert variety in terms of quivers: this is done in Section 3, the final section of the paper, which is richely equipped with illuminating examples as well.