On tensoring with the Steinberg representation (Q2210563)

The authors work in the following setting. Let \(k\) be an algebraically closed field of positive characteristic \(p\), and let \(G\) be a simple, simply connected algebraic group scheme over \(k\). Relative to a maximal torus \(T\), let \(X\) be the weight lattice with fundamental weights \(\varpi_i\), \(X_+\) the dominant weights, and \[ X_r =\{ \lambda \in X_+ \mid 0 \leq \langle \lambda, \alpha^\vee\rangle < p^r, \; \forall \alpha \text{ a simple root}\} \] For each \(\lambda \in X_+\) there are \(G\)-modules \(\nabla(\lambda)=\text{ind}^G_B\lambda\), the Weyl module \(\Delta(\lambda)\), and a finite dimensional simple module \(L(\lambda)\). The authors are primarily focused on when a \(G\)-module has \emph{good filtration}, meaning that it has a \(G\)-filtration whose sections are of the form \(\nabla(\lambda)\), and a related concept of \emph{good \((p,r)\)-filtration} defined in Section 1.2. Their focus on good filtrations is motivated by a conjecture of S.~Donkin. Conjecture 1. Let \(M\) be a finite dimensional \(G\)-module. Then \(M\) has a good \((p,r)\)-filtration if and only if \(\text{St}_r\otimes M\) has a good filtration. where \(\text{St}_r = L((p^r-1)\sum_i \varpi_i)\) is the \emph{\(r^\text{th}\) Steinberg module}. In particular, the authors note that the implication \[ M \text{ has a good } (p,r)\text{-filtration} \Rightarrow \text{St}_r\otimes M \text{ has a good filtration.} \] is equivalent to \(\text{St}_r\otimes L(\lambda)\) having a good filtration for all \(\lambda \in X_+\). Hence, they show that such good filtrations of \(\text{St}_r\otimes L(\lambda)\) exist when \begin{itemize} \item[(i)] \(p\geq 2h-4\) where \(h\) is the Coxeter number of \(G\), \item[(ii)] \(G\) is rank 2, \item[(iii)] \(p\geq 3\) and \(\lambda=\varpi_i\) is a fundamental weight, \item[(iv)] and in various cases when \(\text{rank}(G)\leq 5\). \end{itemize} Result (i) is an improvement on the previous known bound, which is that the statement held for \(p\geq 2h-2\). After a brief recollection of definitions and known results in Section 2, Section 3 of the paper develops some sufficient conditions for \(\text{St}_r\otimes M\) to have a good filtration in terms of \emph{generalized Frobenius contraction functors}, which the authors define, and projection functors from [\textit{J. C. Jantzen}, Representations of algebraic groups. Providence, RI: American Mathematical Society (AMS) (2003; Zbl 1034.20041)]. These conditions are applied in Section 4 to obtain results (i) and (ii). Also in Section 4, the authors establish result (iii), which in fact also holds for most group when \(p=2\). The cases where it may not hold are types \(E_7\) and \(E_8\) when \(p=2\), and the authors list the fundamental weights which may cause an obstruction. The results of point (iv) span Sections 5 and 6. In Section 5, the authors use the computer algebra systems LiE and Magma to verify the result for small primes and groups of rank 3, 4, and 5. Section 6 focuses on the cases when \(p=2\) and \(G\) is of type \(A_5\) or \(E_7\), with the \(A_5\) case being the primary focus. Detailed analysis of internal submodule structures are used in these cases since the previous techniques of the paper are insufficient. The authors are optimistic that further study of their techniques in these cases will give insight into the general case of Conjecture 1. The paper concludes with Section 7, which defines a \(G\)-module structure on the Hopf algebra \(k[G_r]\) of \(G_r\), the \(r^\text{th}\) Frobenius kernel of \(G\), and considers when \(\text{St}_r\otimes k[G_r]\) has a good filtration.