On the combinatorics of \(B\times B\)-orbits on group compactifications (Q1812168)

The authors prove a conjecture of Springer by giving a combinatorial proof of the fact that a certain map is an involution [\textit{T. A. Springer}, Intersection cohomology of \(B\times B\)-orbit closures in group compactifications. J. Algebra 258, No. 1, 71--111 (2002; Zbl 1110.14047)]. The problem arose in the study of intersection cohomology of orbit closures for the \(B\times B\) action on a wonderful compactification of a semi-simple adjoint group \(G\) with Weyl group \(W\). There is a poset \(V\) that encodes the closure relation between the orbits and Springer constructed Kazhdan-Lusztig polynomials for \(V\), in analogy with the usual ones for the Chevalley-Bruhat order on \(W\). This involves an involution on a bimodule for the Iwahori-Hecke algebra of \(W\). The definition of the involution was combinatorial, but the proof used Verdier duality. Springer conjectured a generalization to arbitrary Coxeter groups \(W\). Positive evidence was obtained by computer experiments of the reviewer. The authors now prove the conjecture by reinterpreting the poset \(V\) as an interval in a ``twisted'' Chevalley-Bruhat order on a larger Coxeter group. Curiously there is quite a bit of choice for this larger group.