Flexibility of Schubert classes (Q386019)

For a homogeneous variety \(X:=G/P\), where \(G\) is a linear algebraic group and \(P\) is a parabolic subgroup, it is well known that the cohomology ring of \(X\) admits a \(\mathbb{Z}\)-basis given by the classes of Schubert varieties. A natural question in this setting, first asked in [\textit{A. Borel} and \textit{A. Haefliger}, Bull. Soc. Math. Fr. 89, 461--513 (1961; Zbl 0102.38502)], is whether there are other projective subvarieties of \(X\) representing a given Schubert class \(\sigma\). The notions of a \textit{rigid}, \textit{multi-rigid} and \textit{flexible} Schubert class were introduced to describe this behavior. More precisely, in the first case \(\sigma\) does not have projective variety representatives other than Schubert varieties and, in the second, every positive integer multiple \(m\sigma\) of \(\sigma\) can only be represented by a union of \(m\) Schubert varieties. The class \(\sigma\) is otherwise called flexible. Further, in [Sel. Math., New Ser. 18, No. 3, 717--777 (2012; Zbl 1250.32022)], \textit{C. Robles} and \textit{D. The} introduced a space of fist order obstructions to the multi-rigidity of \(\sigma\) in the Lie algebra cohomology. In the paper under review, such a class is called \textit{obstructed} if its obstruction space is nontrivial. Also, recall that a \textit{cominuscule} homogeneous variety can be seen as a group theoretic generalization of a Grassmannian. For instance, the notion of irreducible cominuscule complex homogeneous variety is equivalent to that of compact Hermitian symmetric space. The main result of the paper under review states that, for an obstructed Schubert class \(\sigma\) in the cohomology of a cominuscule homogeneous variety \(X\) and for \(m\) a positive integer, there exists an irreducible subvariety of \(X\) representing \(m\sigma\). This entails the flexibility of any obstructed Schubert class in \(H^*(X,\mathbb{Z})\). In order to prove this statement, the paper is organized along the lines of a classification of irreducible cominuscule homogeneous varieties. More in detail, the following main tools are used in combination with other specific results from the literature. The case of \(X\) a smooth quadric hypersurface is addressed by means of some Bertini type theorems. The authors then settle the case when \(X\) equals the ordinary, Lagrangian or orthogonal Grassmannian by considering the subfunctors of points corresponding to the irreducible subvarieties to be constructed in each case. Finally, they apply all of the above methods, together with some results from the theory of buildings (Tits correspondences, specifically), to deal with the case of the Cayley plane and that of the Freudenthal variety. The authors add that, in each of the above instances, their construction of the representing irreducible subvarieties does not depend on \(X\) being cominuscule. Based on this, they ask whether a certain dichotomy following from the main Theorem holds for any homogeneous variety.