The Grothendieck-Serre conjecture over semilocal Dedekind rings (Q2674769)

Let \(R\) be a regular local ring, let \(K\) be the fraction field of \(R\) and let \(G\) be a reductive group scheme over \(R\). A famous conjecture of Grothendieck and Serre predicts that the restriction map \(H^1(R,G)\to H^1(K,G)\) between étale cohomology pointed sets has trivial kernel. This conjecture is also commonly considered in the more general case where \(R\) is a regular semilocal domain. The main result of the paper proves the extended conjecture when \(R\) is of Krull dimension \(\leq 1\), i.e., when \(R\) is a semilocal Dedekind domain. In the course of proving this, the author also establishes a decomposition theorem: for every reductive group scheme \(G\) over a semilocal Dedekind domain \(R\), we have \[ \prod_{v} G(K_v)=G(K)\cdot\prod_v G(R_v), \] where \(v\) ranges over the maximal ideals of \(R\), and \(R_v\) and \(K_v\) are the completions of \(R\) and \(K\) at the \(v\)-adic topology, respectively. Separately, the author shows that if the Grothendieck-Serre conjecture holds for a particular regular semilocal domain \(R\), then every reductive group scheme over \(K\) admits at most one reductive model over \(R\). In particular, the latter holds when \(R\) is a semilocal Dedekind domain. The many cases where the Grothendieck-Serre conjecture was previously known are surveyed in the introduction of the paper. Historical note: \textit{Y. A. Nisnevich} [Etale cohomology and Arithmetic of Semisimple Groups. Harvard University (PhD Thesis) (1983); C. R. Acad. Sci., Paris, Sér. I 299, 5--8 (1984; Zbl 0587.14033)] proved the Grothendieck-Serre conjecture when \(R\) is a discrete valuation ring (DVR). His work was based on an unpublished result of Tits which settled the complete DVR case, and in turn relied on Bruhat-Tits theory [\textit{F. Bruhat} and \textit{J. Tits}, Publ. Math., Inst. Hautes Étud. Sci. 60, 1--194 (1984; Zbl 0597.14041)]. However, as the author notes (p.~899), there are some unclear points in the Nisnevich-Tits argument, possibly due to the fact that [loc. cit.] appeared in print two years after Nisnevich's works. The proof of the author's main result follows the outline of the Nisnevich-Tits argument, but with some changes which circumvent the said issues. Thus, in proving the their main result, the author also provides a clear and self-contained proof of Nisnevich's result.