Schubert patches degenerate to subword complexes (Q1024597)

Let \(G\) be a reductive complex Lie group, with a chosen maximal torus \(T\), associated Weyl group \(W\), Borel subgroup \(B_+=T_+N_+\) and opposite Borel \(B_-=TN_-\). For \(w \in W\), \(X_w=\overline{B_-wB_+}/B_+\) is the associated Schubert variety. For \(v \in W\), \(X_o^v=N_+vB_+/B_+\) is the opposite Schubert cell, and \(X^v_{wo}=X_w \cap X^v_o\) is the corresponding Kazhdan--Lusztig variety. Now suppose that \(v\) is expressed as a minimal length product of reflections associated to simple roots. The sequence \(Q\) of simple roots is called a reduced word for \(v\). Then if \(w \in W\), the author defines a simplicial complex \(\Delta(Q,w)\) called the subword complex. To any simplicial complex \(\Delta\) with vertex set \(V\), and any field \(F\), the Stanley--Reisner scheme \(SR(\Delta) \subseteq F^V\) is the union of the hyperplanes \(F^S\) for \(S \in \Delta\); there is a natural action of the the torus \((\mathbb G_m)^V\) on \(SR(\Delta)\). The author's main result asserts that for \(w \leq v \in W\) and reduced word \(Q\) for \(v\), there is a sequence of flat \(T\)--equivariant degenerations from \(X^v_{wo}\) to \(SR(\Delta(Q,w))\). This implies a number of previously known forumlae in equivariant \(K\)--theory of Schubert classes and geometry of Schubert cells.