Motives of quadrics with applications to the theory of quadratic forms (Q4466498)

The present article comprises the lecture notes of a mini-course given by the author at the Université d'Artois, Lens, June 26--28, 2000, plus some additional material. It presents a detailed overview of the author's work on motivic decomposition of quadrics and its relation with problems concerning the isotropy behaviour of quadrics over a field \(k\) of characteristic \(\neq 2\), for example the dimensions of anisotropic quadratic forms in \(I^n\), the splitting pattern, and stably birational equivalence of quadrics. Much of this material has been available previously but only in the form of preprints. These lecture notes therefore represent a welcome and timely addition to the literature. In the first section, the Grothendieck category of Chow motives is introduced. As an example, it is shown how the motive of projective space decomposes into a direct sum of so-called Tate motives \({\mathbb Z}(\ell)[2\ell]\). Section 2 contains proofs of various results due to Rost concerning the motive of an isotropic resp. completely split projective quadric (i.e., the corresponding quadratic form of dimension \(n\) is isotropic resp. has Witt index \([n/2]\)). For example, it is shown that if \(n\) is odd, then the motive of a completely split quadric of dimension \(n\) is isomorphic to the motive of projective \(n\)-space. Using this motivic decomposition, the Chow groups of a competely split quadric are determined. Section 3 deals with the decomposition of the motive \(M(Q)\) of a quadric \(Q\) into a direct sum of indecomposable summands, and criteria for when two such summands are isomorphic. A crucial tool in these considerations is Rost's nilpotence theorem. An important consequence is a criterion for two quadrics \(P\) and \(Q\) to be stably birationally equivalent in terms of having certain isomorphic indecomposable summands in their respective motivic decomposition. The study of indecomposables in the motivic decomposition of quadrics is deepened in section 4, with some important consequences. For example, suppose that the two anisotropic quadrics \(P\) and \(Q\) are stably birationally equivalent, or, in other words, if \(p\) and \(q\) are the associated anisotropic quadratic forms, then \(p\) becomes isotropic over the function field \(k(Q)\) and \(q\) becomes isotropic over the function field \(k(P)\). Then it is shown that \(\dim p-i_1(p)=\dim q-i_1(q)\), where \(i_1(p)\) (resp. \(i_1(q)\)) is the Witt index of \(p\) (resp. \(q\)) over \(F(P)\) (resp. \(F(Q)\)). It should be noted that this has also been proved using different methods by \textit{N. Karpenko} and \textit{A. Merkurjev} [Invent. Math. 153, 361--372 (2003; Zbl 1032.11015)] who also proved conversely that the equality \(\dim p-i_1(p)=\dim q-i_1(q)\) implies that \(q_{k(P)}\) is isotropic iff \(p_{k(Q)}\) is isotropic. A further important consequence of the author's study of motivic decomposition is that two quadrics \(P\) and \(Q\) of the same dimension have isomorphic motives iff the corresponding quadratic forms have equal Witt indices over all field extensions \(E/k\): \(i_W(p_E)=i_W(q_E)\) (for an alternative proof, see \textit{N. Karpenko} [Math. Ann. 317, 585--611 (2000; Zbl 0965.11015)]). Section 5 contains some of the more complicated and technical proofs of results in the preceding sections. Section 6 provides various important applications of the methods developed in this paper to questions in the algebraic theory of quadratic forms. For example, it is shown that anisotropic quadratic forms in \(I^nk\) are either of dimension \(2^n\) or of dimension \(\geq 2^n+2^{n-1}\). There are in the meantime various other proofs of this dimension gap, for example in the paper by Karpenko and Merkurjev [ibid.]. Another application is the complete characterization of quadratic forms of dimension \(9\) in terms of the motivic decomposition of the corresponding quadrics of dimension \(7\), and criteria for when two such quadrics are stably birationally equivalent. These results are attributed to Izhboldin (see also the articles by \textit{O. Izhboldin} resp. by \textit{N. Karpenko} in the present volume [Geometric methods in the algebraic theory of quadratic forms. Proceedings of the summer school, Lens, France, June 2000. Berlin: Springer. Lect. Notes Math. 1835, 143--150 (2004; Zbl 1045.11023), resp. 151--183 (2004; Zbl 1045.11024)]. In the final section 7, the author applies his methods to the determination of splitting patterns. Recall that the splitting pattern of a quadratic form \(q\) over \(k\) is the increasing sequence of all Witt indices \(i_W(q_E)\) where \(E/k\) runs over all field extensions of \(k\). A complete classification of all splitting patterns for forms of odd dimensions \(\leq 21\), and of even dimension \(\leq 12\), is given. For the entire collection see [Zbl 1034.14001].