Integral motives, relative Krull-Schmidt principle, and Maranda-type theorems (Q500246)

This article is concerned with Chow motives with integral coefficients that are geometrically split. A Chow motive \(M\) over a field \(F\) is said to be geometrically split if there exists a field extension \(E/F\) such that the base change \(M_E\) is isomorphic to a direct sum of Tate motives. Examples of geometrically split motives are given by the motives of projective homogeneous varieties. A motive \(M\) is said to satisfy the Krull-Schmidt property if any two decompositions of \(M\) into a direct sum of indecomposable objects are equivalent (i.e., isomorphic up to reordering of the direct summands). This property had already been investigated for projective homogeneous varieties: for example, it holds for certain twisted flag varieties satisfying certain conditions [\textit{V. Chernousov} and \textit{A. Merkurjev}, Transform. Groups 11, No. 3, 371--386 (2006; Zbl 1111.14009)] and for projective quadrics [\textit{A. Vishik}, Lect. Notes Math. 1835, 25--101 (2004; Zbl 1047.11033)], but it fails in some cases [Chernousov and Merkurjev, loc. cit.], [\textit{B. Calmès} et al., Compos. Math. 142, No. 4, 1063--1080 (2006; Zbl 1111.14008)]. In this paper, the author ask a slightly different question: suppose that any two decompositions of \(M\) into a direct sum of indecomposable objects become equivalent after base-change to an algebraic closure of \(F\). Is it true that these two decompositions are equivalent? If this is the case, we say that \(M\) satisfies the relative Krull-Schmidt property. The previously known counterexamples to the Krull-Schmidt principle do not provide counterexamples to the relative Krull-Schmidt principle. The main result of this paper is Theorem 3.3. It is a general statement relating decompositions of \(M\) into direct sums of indecomposable objects after base-change and extension of ring of coefficients. Theorem 3.3 is used in Section 5 to show that the relative Krull-Schmidt principle holds for a wide class of projective homogeneous varieties, but fails for some twisted flag variety. The paper also includes results concerning motives with coefficients in a discrete valuation ring (Section 4), and results concerning decomposability of integral motives of generalized Severi-Brauer varieties (Section 5). For instance, it is shown that the classical Severi-Brauer varieties together with the examples previously studied by \textit{N. A. Karpenko} [St. Petersbg. Math. J. 7, No. 4, 649--661 (1996; Zbl 0866.14006); translation from Algebra Anal. 7, No. 4, 196--213 (1995)] are the only generalized Severi-Brauer varieties whose integral motive is indecomposable.