The Witt ring of a Brauer-Severi variety (Q1269499)

The paper extends Arason's computation of the Witt ring of a projective space over a field [\textit{J. Kr. Arason}, Math. Ann. 253, 205-212 (1980; Zbl 0431.10011)] to twisted forms of projective spaces, i.e. Severi-Brauer varieties. Namely, it is proven that for a Severi-Brauer variety \(X/k\) the base-change homomorphism \(\sigma^{\ast} : W(k) \rightarrow W(X)\) is surjective. The main idea of the proof is following. Let \(k'\) be an algebraic closure of \(k\) and let \(X' = X \times k'\). Then \(X'\) is a projective space and Arason's method of killing cohomology groups works. It is possible to produce Arason's extension of a bundle on \(X\), computing cohomology on \(X'\) -- a nice and quite surprising idea. The Arason-Elman-Jacob theory of indecomposable self-dual vector bundles [\textit{J. Kr. Arason, R. Elman} and \textit{B. Jacob}, Commun. Algebra 20, No. 5, 1323-1351 (1992; Zbl 0769.14004)] is used to control dependence between bundles on \(X\) and their lifts to \(X'\). The paper depends on Arason's one as well as on Arason-Elman-Jacob: Reference to ``[S] Ch. 7, theorems 10.8 and 10.9(i)'' on page 105 does not point to [S] in the list of references, it points to Scharlau's book on quadratic forms, as in the Arason-Elman-Jacob paper. The paper is written ``didactically'' in the sense that ``everything'' is proved if there is no convenient reference. Nevertheless, triviality of self-dual line bundle is claimed obvious without indication to the fact that the Picard group of a Severi-Brauer variety is torsionfree.