Quasilinear quadratic forms and function fields of quadrics

From MaRDI portal



Abstract: Let p and q be anisotropic quadratic forms of dimension geq2 over a field F. In a recent article, we formulated a conjecture describing the general constraints which the dimensions of p and q impose on the isotropy index of q after scalar extension to the function field of p. This can be viewed as a generalization of Hoffmann's Separation Theorem which simultaneously incorporates and refines some well-known classical results on the Witt kernels of function fields of quadrics. Using algebro-geometric methods, it was shown that large parts of this conjecture hold in the case where the characteristic of F is not 2. In the present article, we prove similar (in fact, slightly stronger) results in the case where F has characteristic 2 and q is a so-called quasilinear form. In contrast to the situation where mathrmchar(F)eq2, the methods used to treat this case are purely algebraic.


Over the last few decades, one of the major topics in the algebraic theory of quadratic forms has been the study of the isotropy behavior of quadratic forms over function fields of quadrics that has revealed many beautiful and intricate algebro-geometric features of quadrics. Let \(p\) and \(q\) be anisotropic quadratic forms of dimension at least \(2\) over a field \(F\) and let \(F(p)\) be the function field of the projective quadric defined by the equation \(p=0\). Let \(i_0:=i_0(q_{F(p)})\) denote the isotropy index of \(q\) after scalar extension to \(F(p)\), i.e., the maximal dimension of a totally isotropic subspace of \(q_{F(p)}\). For example, the Separation Theorem (due to the reviewer in characteristic not \(2\) [Math. Z. 220, No. 3, 461--476 (1995; Zbl 0840.11017)] and to \textit{A. Laghribi} and the reviewer in characteristic \(2\) [J. Algebra 295, No. 2, 362--386 (2006; Zbl 1138.11012)]) states that if \(\dim(q)\leq 2^n<\dim(p)\) for some positive integer \(n\), then \(i_0=0\), i.e., \(q\) stays anisotropic over \(F(p)\). In [Math. Ann. 372, No. 3--4, 1437--1458 (2018; Zbl 1445.11020)], the present author formulated a very general conjecture that states that if \(s\) is the non-negative integer such that \(2^s<\dim(p)\leq 2^{s+1}\) then \(k:=\dim(q)-2i_0\geq 0\) and \(\dim(q)=a2^{s+1}+\epsilon\) for integers \(a,\epsilon\) with \(a\geq 0\) and \(|\epsilon|\leq k\). Note that this conjecture trivially holds if one already knows that \(i_0=0\). In [loc. cit.], the author proved that the conjecture holds in characteristic not \(2\) in various situations, for example, if \(\dim(q)\leq 2^{s+2}+2^{s+1}\) (from which one can deduce the Separation Theorem) or if \(p\) is a Pfister neighbor. To achieve this, the author used geometric methods from the theory of algebraic cycles, one major ingredient being the action of Steenrod operations on the mod-2 Chow rings of certain homogeneous varieties. In the present paper, the author shows that the conjecture also holds in various cases (including the cases analogous to the above mentioned ones) when the quadratic form \(q\) is quasilinear. In this case, the quadric defined by \(q=0\) is nowhere smooth, and it is not hard to show that for such a \(q\) to become isotropic over \(F(p)\) (the only interesting case for this conjecture) then \(p\) must be quasilinear as well. Since quasilinear forms are amenable to be studied by purely algebraic methods, the tools used in the proofs in the present paper are rather different from those used in [loc. cit.].











This page was built for publication: Quasilinear quadratic forms and function fields of quadrics

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2305680)