Parity sheaves and tilting modules (Q2811179)

Let \(G\) be a split reductive group over a field of characteristic \(p\) with a chosen maximal torus \(T\) and Levi subgroup \(L\). Let \(\Lambda^+\) be a set of dominant weights, where dominance is defined using a fixed Borel subgroup of \(G\) containing \(T\). Additionally, let \(\check{G}\) be the connected complex algebraic group with maximal torus \(\check{T}\) with root datum dual to \(G\). The geometric Satake theorem establishes a equivalence between the category of rational representations of \(G\) and the category \(P(\mathcal{G}r)\) of \(\check{G}(\mathcal{C}[[t]])\)-equivariant perverse sheaves on \(\mathcal{G}r\), the affine Grassmannian for \(\check{G}\) with coefficients on \(k\). If \(p\) is not a torsion prime for \(\check{G}\), then for each \(\lambda\in \Lambda^+\) one can assign a certain indecomposable parity complex \(\mathcal{E}(\lambda)\): these particular parity complexes are called parity sheaves. Call a \(\mathcal{T}\in P(\mathcal{G}r)\) a titling sheaf if it corresponds to a tilting module for \(T\). The aim of the paper being reviewed is to describe a connection between parity sheaves and tilting modules, one which explains the similarities between their properties.NEWLINENEWLINEThe main result is that, provided \(p\) is large enough, \(\mathcal{E}(\lambda) = \mathcal{T}(\lambda)\) for all \(\lambda\in \Lambda^+\). The precise definition of ``large enough'' depends on the type of the root system \(\Phi\) for \(G\), however one can always take \(p>n\) is \(\Phi\) is of type \(C_n\); otherwise one can apply a lower bound \(p> 31\) (a bound which can be further refined).