\(p\)-filtrations and the Steinberg module (Q5952414): Difference between revisions

Let \(G\) be a connected and simply connected reductive algebraic group defined over an algebraically closed field of characteristic \(p>0\). Recall that a finite dimensional \(G\)-module \(M\) is said to have good filtration if it has a filtration whose layers are induced modules \(\nabla(\lambda)\). Now the author considers modules that have a filtration with layers of the form \(L(\lambda_0)\otimes\nabla(\lambda_1)^{(p)}\), where \(L(\lambda_0)\) denotes a simple \(G\)-module whose highest weight \(\lambda_0\) lies in the restricted region and \(\nabla(\lambda_1)^{(p)}\) is the first Frobenius twist of \(\nabla(\lambda_1)\). He says that a module with such a filtration has good \(p\)-filtration. It has been conjectured by Donkin that a module \(M\) has good \(p\)-filtration if and only if \(M\otimes\text{St}\) is \(G\)-acyclic. Here \(\text{St}\) is the Steinberg module. The author now proves the `only if' part when \(p>2h-2\) and both parts when \(G=\text{SL}_2\). There are several related results. Say \(p>2h-2\). A module with good filtration has good \(p\)-filtration. The tensor product of a module with good filtration and a module with good \(p\)-filtration is one with good \(p\)-filtration. There are analogues with higher Steinberg modules and with quantum groups.