Representation ring of Levi subgroups versus cohomology ring of flag varieties (Q330852)

It is a classical result in representation theory that to each partition \(\lambda\) one may attach a finite-dimensional representation \(V(\lambda)\) of the general linear group \(\mathrm{GL}_r({\mathbb C})\). The tensor product of any two such representations decomposes into the direct sum of irreducible representations whose multiplicities are precisely those prescribed by the Littlewood-Richardson rule. This fact implies a non-trivial and beautiful relationship with Schubert Calculus for the complex Grassmannians \(\mathrm{Gr}(r,n)\), because the structural constants of the singular cohomology \(H^*(\mathrm{Gr}(r,n),{\mathbb Q})\) of \(\mathrm{Gr}(r,n)\) are precisely the Littlewood-Richardson coefficients. This occurrence can be expressed by saying that there is an explicit surjective homomorphism \[ \xi: \mathrm{Rep}_{\mathrm{poly}}^{{\mathbb C}}(L)\rightarrow H^*(\mathrm{Gr}(r,n)),\leqno{(*)} \] where \(\mathrm{Rep}^{\mathbb C}_{\mathrm{poly}}\) denotes the complexification of the polynomial representation ring of \(\mathrm{GL}_r\), mapping \(V(\lambda)\) to \(\sigma_\lambda\), the Schubert cycle associated to the partition \(\lambda\). For more on this relationship, from a different point of view, the nice survey by \textit{H. Tamvakis} [Enseign. Math. (2) 50, No. 3--4, 267--286 (2004; Zbl 1131.14311)] is recommended. The purpose of the paper under review is to explore how much such a phenomenology can be generalized for other flag varieties of the form \(G/P\), by replacing the general linear group \(\mathrm{GL}_r({\mathbb C})\) with any Levi subgroup \(L\) of any reductive group \(G\) and the corresponding generalized flag variety. More precisely, one starts from the datum \((G,B, P, L)\), where \(G\) is a connected reductive group \(G\) over \({\mathbb C}\), \(B\) a Borel subgroup of it with maximal torus \(T\), \(P\) a parabolic subgroup and \(L\) a Levi subgroup of \(P\) containing \(T\). Then, the role of the partitions in the map \((*)\) is now played by the set \(D\) of the dominant characters of \(T\), so that the main achievement of the paper, summarized in Theorem 5, consists in the construction of a surjective \({\mathbb C}\)-algebra homomorphism \[ \mathrm{Rep}_{\lambda-\mathrm{poly}}^{{\mathbb C}}(L)\rightarrow H^*(G/P, {\mathbb C})\leqno{(**)} \] where \(\mathrm{Rep}_{\lambda-\mathrm{poly}}^{{\mathbb C}}(L)\) is the \(\lambda\)-polynomial subring of \(\mathrm{Rep}^{\mathbb C}(L)\), the complexification of the representation ring \(\mathrm{Rep} (L)\) of \(L\), which maps any almost faithful irreducible \(G\)-module \(V(\lambda)\) to a cohomology class of the flag variety \(G/P\). The reviewer is aware that many of the terms used in the review are not explained. To keep it to a reasonable length the non expert but interested reader is necessarily invited to look at the paper for the background. The \(\lambda\)-polynomial subring occurring in the map \((**)\), and the map itself, is defined via a morphism \(\theta_\lambda:G\rightarrow {\mathfrak g}\), where \({\mathfrak g}\) is the Lie algebra of \(G\), known in the literature as Springer morphism, well described in the paper under review or in the original source [\textit{T. A. Springer}, in: Algebr. Geom., Bombay Colloq. 1968, 373--391 (1969; Zbl 0195.50803)], or in Section 9 of the important paper by \textit{P. Bardsley} and \textit{R. W. Richardson} [Proc. Lond. Math. Soc. (3) 51, 295--317 (1985; Zbl 0604.14037)]. The paper is very well organized. After the introduction and the preliminaries occupying up to Section 3, Section 4 is devoted to the statement and the proof of the main Theorem 5, concerned with the map \((**)\). The remaining sections are devoted to apply the theorem in a number of concrete situations. First of all to the case where \(G=\mathrm{GL}_r\), so recovering the known results regarding the representations of the general linear group (surjection \((*)\)). Section 6 sets a few preliminaries necessary to study the case when \(G\) is chosen to be another classical group other than \(\mathrm{GL}_r\): the interplay with the theory of \(\lambda\)-rings is highlighted, here. Finally sections 7, 8, 9 are concerned with the specializations of the main Theorem 5, by fixing the group \(G\) to be respectively \(\mathrm{Sp}_{2n}\), \(\mathrm{SO}_{2n+1}\) and \(\mathrm{SO}_{2n}\). Section 10 studies the case in which \(G\) can be arbitrarily chosen among the classical groups, but assuming that the parabolic subgroup \(P\) coincides with the Borel subgroup \(B\). This difficult but exciting paper ends with a complete, though essential, reference list.