Schur algebras and global bases: New proofs of old vanishing theorems (Q1357565): Difference between revisions

Let \(U\) be the quantized enveloping algebra associated to a finite-type root datum. Put \({\mathcal A}=\mathbb{Z}[q,q^{-1}]\). \textit{G. Lusztig} [Adv. Math. 70, 237-249 (1988; Zbl 0651.17007)] has defined an \(\mathcal A\)-form \(U_{\mathcal A}\) in \(U\). In this paper, by base change the author obtains an algebra \(U_k\) for any \(\mathcal A\)-algebra \(k\). He studies the category \(U_k\)-Int (resp. \(U^b_k\)-Int) of integrable modules for \(U_k\) (resp. \(U^b_k)\). The author shows how one can prove in a rather straightforward manner certain cohomological results for the categories \(U_k\)-Int and \(U^b_k\)-Int. He starts by showing that these two categories have structures akin to highest weight categories. The key idea is that the injective modules should have good filtrations. These results are extended to the categories \(U_k\)-Int and \(U^b_k\)-Int, which structures are shown to be closely connected. The vanishing theorems are deduced from this connection. These results are proved under technical restrictions on \(k\) in [\textit{H. H. Andersen, P. Polo} and \textit{K. Wen}, Invent. Math. 120, 409-410 (1995; Zbl 0870.17008)] by reduction to the classical case. An important aspect of this subject is the observation that a highest weight category is patched together in a very satisfactory manner from the module categories of a collection of rather special finite-dimensional algebras. Using the global bases, the author shows that the algebras patch together nicely among themselves, recovering results on the full categories \(U_k\)-Int and \(U^b_k\)-Int by limiting arguments. The definition of a highest weight category given in [\textit{E. Cline}, \textit{B. Parshall} and \textit{L. Scott}, J. Reine Angew. Math. 391, 85-99 (1988; Zbl 0657.18005)] is unsuitable for working over an arbitrary ground ring because of its heavy reliance on local finiteness. To overcome this, the author works with comodule categories over flat \(k\)-coalgebras, and introduces weight spaces in an explicit manner. His point of view is that \(U_k\)-Int and \(U^b_k\)-Int are comodule categories for the associated coordinate (co)algebras. The framework introduced is rather specifically tailored to this situation.