\(W\)-translated Schubert divisors and transversal intersections (Q2082276)

The partial flag variety \(X = \mathcal F\mathcal l_{n_1,\dots,n_k;n}\), parametrizing the space of partial flags \(\{V_{n_1}\subset \ldots \subset V_{n_k}\subset \mathbb C^n\}\) is a smooth projective \(\mathrm{SL}(n, \mathbb C)\)-homogeneous variety. The cohomology ring \(H^*(X,\mathbb Z)\) is torsionfree, and has a \(\mathbb Z\)-additive basis of Schubert classes \({\sigma}^u\) labeled by a subset of the permutation group \(S_n\) of \(n\) objects. The structure constants in the cup product are nonnegative integers. One of the most central problems in Schubert calculus is finding a manifestly positive combinatorial formula of structure constants which determines a manifestly positive formula for general structure coefficients. However, this problem is widely open except for two-step flag varieties \(\mathcal F\mathcal l_{n_1,n_2;n}\) beyond the famous Littlewood-Ricardson rules for complex Grassmannians \(Gr(m, n) = \mathcal F\mathcal l_{m;n}\). In addition, there is a recent work on three-step flag varieties, and there has been a nice algorithm for the Schubert structure constants even for flag varieties of general Lie type. In the present paper, the authors propose a geometric approach towards a manifestly positive formula of structure coefficients \(\mathcal F\mathcal l_{n_1,\dots,n_k;n}\). They start with the geometrical meaning of the Schubert structure constants. In this paper, they consider the finitely many subvarieties \(uX_v\)'s where \(u \in S_n\), and call them \(W\)-translated Schubert varieties. They study the transversality of intersections among them. Here these transversal intersections of \(W\)-translated Schubert varieties uses toric degenerations and Gelfand-Cetlin polytopes. The \(W\)-translated Schubert varieties are also expected to enjoy similar nice degeneration properties. This leads to the strategy that intersection problems in the generic fiber are likely transferred to intersection problems of toric subvarieties of \(X_{\Delta}\), and hence to that of faces of the Gelfand-Cetlin polytope \(\Delta\). Overall, they propose a conjecture that Schubert varieties of appropriate dimensions intersect transversally up to translation by Weyl group elements, and verify it in various cases, including the complex Grassmannian \(\mathrm{Gr}(2, n)\) and the complete flag variety \(\mathcal F_{1,2,3;4}\).