Covering data and higher dimensional global class field theory (Q838462): Difference between revisions

From the introduction: ``The aim of global class field theory is the description of abelian extensions of arithmetic schemes (i.e. regular schemes \(X\) of finite type over Spec\((\mathbb Z)\)) in terms of arithmetic invariants attached to \(X\). The solution of this problem in the case \(\dim X = 1\) was one of the major achievements of number theory in the first part of the previous century. In the 1980s, mainly due to \textit{K. Kato} and \textit{S. Saito} [Contemp. Math. 55, 255--331 (1986; Zbl 0614.14001)], a generalization to higher dimensional schemes has been found. The description of the abelian extensions is given in terms of a generalized idèle class group, whose rather involved definition is based on Milnor \(K\)-sheaves. In the course of the last years, G. Wiesend developed a new approach to higher dimensional class field theory which only uses data attached to points and curves on the scheme. The central and new idea was to consider data which describe not necessarily abelian Galois coverings of all curves on the scheme, together with some compatibility condition. Then one investigates the question whether these data are given by a single Galois covering of the scheme. The essential advantage of the nonabelian approach is that one can use the topological finite generation of the tame fundamental groups of smooth curves over separably closed fields as an additional input. The restriction to abelian coverings is made at a later stage. One obtains an explicitly given class group \({\mathcal C}_X\) together with a reciprocity homomorphism \(\rho_X: \mathcal C_X \to \pi_1^{ab}(X)\) to the abelianized fundamental group, which has similar properties like the classical reciprocity homomorphism of one-dimensional class field theory. As a result of the method, the full abelian fundamental group can be described only if \(X\) is flat over Spec\((\mathbb Z)\) and for varieties over finite fields which are proper over a curve. For a general variety over a finite field, the method only yields a description of the tame part \(\pi_1^{t,ab}(X)\) (this description is equivalent to that given by \textit{A. Schmidt} and \textit{M. Spieß} [J. Reine Angew. Math. 527, 13--36 (2000; Zbl 0961.14013)]). Wiesend's approach is independent from and easier than the original approach of \textit{K. Kato} and \textit{S. Saito} [loc. cit.]. Although it fails to describe the wild part in positive characteristic, it should be seen as a substantial progress in the theory. For example, it provides an easier proof of the finite generation of the Chow group of zero cycles modulo rational equivalence of arithmetic schemes (first proved by Kato and Saito). Furthermore, the explicit definition of the class group will hopefully make this theory more suited for applications. \textit{G. Wiesend} published his results in [J. Number Theory 121, No. 1, 118--131 (2006; Zbl 1120.14014)] and [Math. Z. 256, No. 4, 717--729 (2007; Zbl 1115.14016)]. It is, however, not easy to follow his arguments, and his papers contain a number of gaps and mistakes. As a result, it was not clear whether Wiesend's theorems should be considered as proven. The aim of this article is to provide a complete account of the theory which is more accessible, corrects the mistakes and fills the gaps in Wiesend's papers. We use the same key ideas but have introduced quite a number of improvements. A more direct approach to the reciprocity map for flat arithmetic schemes can be found in [\textit{M. Kerz}, \url{arXiv:0711.4485v2}].'' As the goal of the authors was to clarify the ideas of Wiesend on this subject, they should be commended for having achieved it. The algebraic geometry is elegantly presented, and the exposition makes the article highly readable.