Reciprocity for Kato-Saito idele class group with modulus (Q2161159)

The paper under review contains a refinement of the higher-dimensional class field theory of [\textit{K. Kato} and \textit{S. Saito}, Contemp. Math. 55, 255--331 (1986; Zbl 0614.14001)] in the case of varieties over finite fields. Consider a \(d\)-dimensional integral normal scheme \(X\) of finite type over a finite field and a regular dense open subscheme \(U\subset X\) whose complement is of pure codimension 1. For a closed subscheme \(D\) supported in \(X\setminus U\) Kato and Saito introduced the Milnor \(K\)-theory sheaf \(\mathcal{K}^M_{d, (X,D)}\) and constructed a so-called reciprocity map between the direct limit of the Nisnevich cohomology groups \(H^d_{\mathrm{Nis}}(X,\mathcal{K}^M_{d, (X,D)})\) over all \(D\) supported in \(U\setminus X\) and the abelianized étale fundamental group \(\pi_1^{\mathrm{ab}}(U)\) of \(U\); this map is injective with dense image. In the present paper the authors descend the above construction to the level of a single group \(H^d_{\mathrm{Nis}}(X\mathcal{K}^M_{d, (X,D)})\) with \(D\) fixed. To do so, they introduce a new notion of abelianized fundamental group with modulus associated with the pair \((X,D)\) that they denote by \(\pi_1^{\mathrm{adiv}}(X,D)\). This is the main new construction on the paper. Afterwards, using the presentation of \(H^d_{\mathrm{Nis}}(X,\mathcal{K}^M_{d, (X,D)})\) via higher idèles due to \textit{M. Kerz} [Math. Res. Lett. 18, No. 4, 699--713 (2011; Zbl 1284.19006)], they construct a reciprocity map \(H^d_{\mathrm{Nis}}(X,\mathcal{K}^M_{d, (X,D)})\to \pi_1^{\mathrm{adiv}}(X,D)\) with dense image which after taking the direct limit over all possible \(D\) gives back that of Kato and Saito. Thus the important improvement with respect to the work of Kato and Saito is that in defining their reciprocity map Gupta and Krishna keep control of the ramification along \(X\setminus U\). Such an improvement has already been proposed by \textit{M. Kerz} and \textit{S. Saito} [Duke Math. J. 165, No. 15, 2811--2897 (2016; Zbl 1401.14148)] but instead of the \(K\)-cohomology groups of Kato and Saito they were using a new notion of Chow groups of zero-cycles with modulus. In the companion paper [\textit{R. Gupta} and \textit{A. Krishna}, Adv. Math. 404, Part A, Article ID 108376, 75 p. (2022; Zbl 07537674)] the authors show that the above reciprocity map is actually injective and moreover their abelianized fundamental group \(\pi_1^{\mathrm{adiv}}(X,D)\) identifies with the related group \(\pi_1^{\mathrm{ab}}(X,D)\) considered earlier by Kerz and Saito. Both groups are defined using abelian coverings étale over \(U\) whose ramification along \(X\setminus U\) is bounded by \(D\) in a sense made precise through the use the ramification filtrations of Abbes-Saito and Matsuda. However, whereas Kerz and Saito bound ramification by pulling back to curves, Gupta and Krishna pull back to the generic points of \(D\). Thus the two papers together build a bridge between the reciprocity isomorphisms of Kato-Saito and Kerz-Saito and establish (in a somewhat roundabout way) a Bloch type cohomological formula for the Chow groups with modulus considered in [\textit{M. Kerz} and \textit{S. Saito}, Duke Math. J. 165, No. 15, 2811--2897 (2016; Zbl 1401.14148)] for pairs \((X,D)\) defined over finite fields.