Kazhdan-Lusztig combinatorics in the moment graph setting (Q1950627)

Kazhdan-Lusztig polynomials play an important role in representation theory. Let \((W,S)\) be a Coxeter system and \(J \subset S\). For every ordered pair \(y \leq w \in W^J = W / \langle J \rangle\) there is a polynomial \(P^J_{y,w}(q)\). Their importance was first manifested in deep connections to the local intersection cohomology of the Schubert variety corresponding to \(w\) called \(X^J_w\) in case \(W\) is a Weyl group. Kazhdan and Lusztig also conjectured that they play a very important role in other aspects of Lie Theory. On a different note, Goresky, Kottwitz and MacPherson [\textit{M. Goresky} et al., Invent. Math. 131, No. 1, 25--83 (1998; Zbl 0897.22009)] introduced the concept of moment graphs in order to study equivariant cohomology of some complex algebraic varieties with torus actions. The connection is that if \(W\) is a Weyl group, the Bruhat graph is such a moment graph. The Bruhat graph has vertices given by \(W\) and there is an arrow \(x \to y\) if \(X^J_x \subset X^J_y\) and no Schubert variety is strictly in between them. In this case there is something called the canonical- or BMP-sheaf for each \(w \in W^J\) on the Bruhat graph. They contain some information about the local intersection cohomology of the corresponding Schubert variety \(X^J_w\). In characteristic zero and for large prime characteristics Kazhdan-Lusztig polynomials can be computed as graded ranks of the stalks of the BMP-sheaf on the Bruhat graph. The article lifts a series of equalities between Kazhdan-Lusztig polynomials to this setting by showing that they also hold for BMP-sheaves. Only minor restrictions have to be made on the field involved. The arguments are purely combinatorial. The article recalls all the necessary notions such as moment graphs, sheaves on moment graphs and BMP-sheaves before dealing with the technical details of the proofs.