Criteria of motivic equivalence for quadratic forms and central simple algebras (Q1579112)

Let \(F\) be a field of characteristic \(\neq 2\) and let \(\phi_1\) and \(\phi_2\) be two quadratic forms over \(F\) of the same dimension. They are called splitting-equivalent if they have the same Witt index over every field extension \(E\) of \(F\), and motivic-equivalent if the motives of the associated quadrics are isomorphic. \textit{A. Vishik} proved in his unpublished thesis [Max-Planck Institut für Mathematik, Preprint 13 (1998)] that if the characteristc of \(F\) is \(0\), then two forms are splitting-equivalent if and only if they are motivic-equivalent. The main theorem of the present paper is to prove this result in any characteristic. This is achieved by simply working in a different category of motives, namely the category \(CV^0\) of Chow-correspondences of degree \(0\), whereas Vishik works in Voevodsky's triangulated category of motivic complexes, \(DM_{-}^{\text{eff}}\) (which contains \(CV^0\) as a full subcategory if \(\text{char}(F)=0\)). The easier part is to show that motivic equivalence implies splitting equivalence. For the converse, there is a rather elementary proof by \textit{O. T. Izhboldin} [Doc. Math. 3, 341-351 (1998; Zbl 0957.11019)] that in the case of odd dimension, splitting equivalence implies that the two forms are similar and thus their quadrics are isomorphic, clearly implying motivic equivalence. Therefore, the author focusses on the difficult case of forms of even dimension. The final part of the paper deals with motivic/splitting equivalence of finite-dimensional central simple algebras over \(F\). Two such algebras \(A_1\) and \(A_2\) of the same dimension over \(F\) are called motivic-equivalent if the motives of their associated Severi-Brauer varieties are isomorphic, and splitting-equivalent if their Schur indices coincide over every field extension \(E\) of \(F\). It is shown that the algebras \(A_1\) and \(A_2\) are motivic-equivalent if and only if the classes \([A_1]\) and \([A_2]\) generate the same subgroup in the Brauer group of \(F\), whereas they are splitting-equivalent if and only if \([A_1]=\pm [A_2]\). Hence, splitting equivalence generally does not imply motivic equivalence.