Filtrations of principal indecomposable modules of Frobenius kernels of reductive groups (Q2266064)

Let G be a simply connected, semi-simple algebraic group over an algebraically closed field K of characteristic \(p>0\), and \(u_ n\) the hyperalgebra of the n-th Frobenius kernel (it is a \(p^{n\cdot \dim G}\)- dimensional subalgebra of the hyperalgebra \(U_ k\) of G). Let T be a maximal torus of G. Let X be the character group of T, and W the Weyl group of G. J. Ballard proved for \(p\geq 3h-3\) (where h is the Coxeter number of G) that each principal indecomposable \(u_ n\)-module \(Q(\lambda,n)\) with \(\lambda \in X_ n\subset X\) admits a G-structure. This bound was improved to \(p\geq 2h-2\) by J. Jantzen. He also proved that each \(u_ n\)-PIM Q(\(\lambda\),n) admits a Weyl filtration, i.e. a G- filtration with subquotients of the form Weyl module \(V(\mu)\), \(\mu \in X^+\) (the set of dominant weights). More generally such filtrations exist also for all \(Q(\lambda,n)\otimes V(\nu)^{(p^ n)}\) with \(\nu \in X^+\), where the exponent \((p^ n)\) denotes a Frobenius twist of a module. The purpose of this paper is to deal with the multiplicity of \(V(\mu)\) occurring in these Weyl filtrations. This problem is related to the question of multiplicities of the simple \(u_ n\)-T-modules \(\hat M(\lambda,n)\), \(\lambda\in X\), in composition series of the universal \(u_ n\)-T-modules \(\hat Z(\mu,n)\) with highest weight \(\mu\in X.\) In Section 1 we introduce the affine Weyl group \(W_{a,p}\) of G, discuss its alcove geometry and a partial order on \(W_{a,p}\) which induces an order \(\uparrow \uparrow\) on X. In Section 2 we mainly discuss the Weyl filtration of Q(\(\lambda\),n). First we define for each \(\lambda \in X_ n\) the set \({\mathcal S}_{\lambda,n}=\{\mu \in X^+|\) \(\mu \uparrow \uparrow \lambda^ 0-\delta \}\) where \(\lambda^ 0=\sigma_ 0\lambda - \delta +p^ n\delta\) (here \(\sigma_ 0\in W\) is the longest element, and \(\delta\) is half the sum of all positive roots). For each \(\mu\in {\mathcal S}_{\lambda,n}\) set \(b^{\lambda}_{\mu}\) equal to the multiplicity of \(\hat M(\)\(\lambda\),n) as composition factor of \(\hat Z(\)\(\mu\) \(+p^ n\delta,n)\). We then prove for \(p\geq 2h-2\), for all \(\sigma\in W\), \(\lambda \in X_ n\), \(\mu\in {\mathcal S}_{\lambda,n}\), \(\nu \in X^+\) with \(\sigma \cdot \mu +p^ n(\nu +\delta)\in X^+\) that the multiplicity of \(V(\sigma \cdot \mu +p^ n(\nu +\delta))\) in a Weyl filtration of \(Q(\lambda,n)\otimes V(\nu)^{(p^ n)}\) is equal to \[ \sum_{\tau \in W}\det (\tau)[\hat Z(\sigma \cdot \mu +p^ n(\nu +\delta),n):\hat M(\lambda +p^ n\tau \cdot \mu,n)] \] (here we use the dot notation \(\sigma \cdot \mu =\sigma (\mu +\delta)-\delta)\). This is applied in Section 3 in the case \(n=1\) to prove the following main theorem: Suppose \(p\geq 2h-2\). Let \(\lambda \in X_ 1\) be p-regular. Then (1) \(b^{\lambda}_{\lambda^ 0-\delta}=1\); (2) If \(\mu\),\(\mu\) '\(\in {\mathcal S}_{\lambda,1}\) with \(\mu\) '\(\uparrow \uparrow \mu\), then \(b^{\lambda}_{\mu}\leq b^{\lambda}_{\mu '}\). As a corollary we get \(b^{\lambda}_{\mu}>0\) for all \(\mu\in {\mathcal S}_{\lambda,1}\).