No label defined (Q4528574)

This paper resumes the theory of symmetric power operations on various K-groups, and extends earlier results of the author to a (somewhat) more general setting. More specifically: one considers a finite morphism \(X \to Y\) of Dedekind schemes (that is, regular Noetherian connected schemes of dimension 1) such that the corresponding field extension is \(G\)-Galois and the \(G\)-extension \(X\to Y\) is at most tamely ramified. To this setup one associates the following algebraic objects: \(K_0(G,X)\) the Grothendieck group of locally free \(O_X\)-modules with semilinear \(G\)-action, endowed with Adams operations \(\psi^k\); and the class group \(Cl(O_Y[G])\), which is endowed with a series of operators \(\sigma^k=\psi^{CNT}_{k'}\); the latter having been introduced in the number field case by Cassou-Noguès and Taylor, and \(k'\) being an inverse to \(k\) modulo ord\((G)\) (it is a part of the setup that \(k'\) exists). It is interesting to understand the interplay of \(f_*\) and these operators. Example: For \(k=1\), \(f_*(\sigma^k(O_X))=f_*(O_X)\) is the class of \(O_X\) as an \(O_Y[G]\)-module, and this is the primordial object of tame Galois module theory. In an earlier paper by the same author [\textit{B. Köck}, Math. Ann. 314, No. 4, 667-702 (1999; Zbl 0980.11052), this interplay was described by a kind of Riemann-Roch formula in the number field case. The author now basically goes through the constructions again, giving more algebraic arguments and arriving at versions of this formula which hold for all Dedekind schemes. An important point seems to be Prop. 3.2, where an explicit class involving the inverse different is showed to vanish in a suitable factor group of \(K_0(O_Y[G])\), and this comes down to an entirely local calculation. (Typo: in the statement of Lemma 3.1, the first \(K^{fl}\) group seems to be lacking the affix T).