Quadratic forms over dyadic valued fields. I: The graded Witt ring (Q1084434)

From the author's introduction: ''This paper is devoted to the concrete calculation of the graded Witt ring of a 2-Henselian dyadic, discretely valued field. Section 1 is devoted to some computations necessary to test for the isotropicity of \(n\)-fold Pfister forms over such fields. These results, which are quite general and are proved for arbitrary dyadic valued fields, are of interest in their own right. Section 2 is the computational section, where we find bases for some specific ideal quotients needed in {\S} 3. (...) Section 3 contains the main result, which is the analogue of Springer's theorem for the graded Witt ring of a 2-Henselian dyadic, discretely valued field [cf. \textit{A. R. Wadsworth}, ibid. 105, 473--496 (1983; Zbl 0506.12024), for the non-dyadic version]. In particular we compute \(GW(F)\) for such a field \(F\) in terms of \(GW(\bar F)\bar F\) and \(v(2)\in\mathbb{Z}\), where \(\bar F\) is the residue class field of \(F\). (...) The final {\S} 4 is devoted to some specific applications of {\S} 3, answering questions concerning the 'amenability problem'.'' The results of {\S} 2 can be viewed as the ideal-theoretic analogues of the cohomological results of \textit{K. Kato} [Invent. Math. 66, 493--512 (1982; Zbl 0497.18017)]. For a multiquadratic extension \(M\) of the field \(F\) let \(h_2(M/F)\) and \(h_3(M/F)\) be the homology groups of the Witt ring complex studied by \textit{R. Elman}, \textit{T.-Y. Lam}, \textit{J.-P. Tignol}, and \textit{A. R. Wadsworth} [Am. J. Math. 105, 1119--1170 (1983; Zbl 0492.10014)]. It was shown there that \(h_ 2(M/F)\neq 0\) for some triquadratic extensions \(M/F\) (while \(h_2(M/F)=0\) for all quadratic and biquadratic extensions \(M/F\)). In the present paper for any \(n\geq 3\) examples are given of multiquadratic extensions \(M/F\) of degree \(2^n\) with \(h_2(M/F)\) any finite elementary 2-group. Also the first explicit example of a triquadratic extension \(M/F\) is given with the property that \(h_3(M/F)\neq 0\).