A sum formula for tilting filtrations (Q1582727)

Based on work of \textit{C. M. Ringel} [Math. Z. 208, No. 2, 209-224 (1991; Zbl 0726.16011)], \textit{S. Donkin} [Math. Z. 212, No. 1, 39-60 (1993; Zbl 0798.20035)] developed a theory of tilting modules for algebraic groups. For a reductive algebraic group \(G\) over a field of prime characteristic, a tilting module is a module which admits two finite filtrations: one whose factors are Weyl modules and one whose factors are dual Weyl modules. (The dual Weyl modules can also be thought of as modules induced from a one-dimensional module for a Borel subgroup.) Indecomposable tilting modules in particular have seen extensive study and play an important role in the representation theory of algebraic groups. This work extends previous work of the author. In [Ann. Sci. Éc. Norm. Supér., IV. Sér. 30, No. 3, 353-366 (1997; Zbl 0885.17008)], the author investigated the space of homomorphisms over \(G\) from a Weyl module into an arbitrary tilting module and constructed a filtration of this space. The main result of this current paper is the existence of a sum formula for the dimensions of these filtrations. The formula expresses the sum of the dimensions in terms of a sum of composition multiplicities of Weyl modules in the given tilting module. (The latter can be completely determined from the character of the given tilting module.) The filtration and sum formula are similar to the Jantzen filtration of Weyl modules and the Jantzen sum formula. The author illustrates how this sum formula can potentially be used to decompose tilting modules into indecomposable summands; a task which has proven challenging, especially for large weights. The author further obtains an analogous sum formula for tilting modules over quantum groups. On the basis of similarities in these formulas, he formulates a conjecture about the diagonalizability of a certain homomorphism, the validity of which would allow one to relate characters for modular tilting modules with those for quantum tilting modules for certain weights. Indeed, results of \textit{W. Soergel} [Represent. Theory 1, No. 15, 115-132 (1997; Zbl 0964.17018)] on characters for quantum tilting modules could be transferred to the modular case. Further this could be used to prove Lusztig's conjecture on the characters of simple modules. For more discussion on this and related conjectures, the reader is referred to the author's paper [Algebraic groups and their representations, NATO ASI Ser., Ser. C, Math. Phys. Sci. 517, 25-42 (1998; Zbl 0918.20034)].