Transfer of quadratic forms and of quaternion algebras over quadratic field extensions (Q1656512): Difference between revisions

Let \(F\) be a field (of any characteristic), \(K\) an étale quadratic extension of \(F\) (i.e., either \(F\times F\) or a separable quadratic field extension of \(F\)) and \(Q\) a quaternion \(K\)-algebra. The main result of this paper (for which two different proofs are provided) is that the following conditions are equivalent: (1) \(Q\) contains a quadratic \(F\)-algebra linearly disjoint from \(F\). (2) \(Q\) contains an étale quadratic \(F\)-algebra linearly disjoint from \(F\). (3) The corestriction of \(Q\) from \(K\) to \(F\) is not a division algebra. In the special case of \(K=F\times F\), the algebra \(Q\) decomposes as \(Q_1 \times Q_2\) where \(Q_1\) and \(Q_2\) are quaternion \(F\)-algebras. In this case, the corestriction of \(Q\) to \(F\) is \(Q_1 \otimes Q_2\). Therefore, the theorem above recovers the known fact (proven by Albert and Draxl in the cases of characteristic not 2 and 2, respectively) that a tensor product of quaternion algebras is not a division algebra if and only if there exists a quadratic étale extension of \(F\) that embeds into both algebras. This theorem in the case of characteristic not 2 appeared as Corollary 16.28 in [\textit{M.-A. Knus} et al., The book of involutions. With a preface by J. Tits. Providence, RI: American Mathematical Society (1998; Zbl 0955.16001)]. The authors manage to provide characteristic free proofs by some clever tricks with the Scharlau transfer.