Tame class field theory for singular varieties over finite fields (Q1676091)

The authors pursue their investigation of the tame class field theory for singular varieties over certain types of fields: algebraically closed fields in [Doc. Math., J. DMV 21, 91--123 (2016; Zbl 1354.14033)], finite fields here. Recall that \textit{K. Kato} and \textit{S. Saito} [Ann. Math. (2) 118, 241--275 (1983; Zbl 0562.14011)] generalized the class field theory for smooth, projective curves over finite fields \(\mathbb{F}\) to smooth, projective varieties \(X\) of arbitrary dimension by constructing a reciprocity isomorphism \(r_X:\mathrm{CH}_0(X) \mathop{\to}\limits^{\sim} \pi_1^{\mathrm{ab}}(X)_W\) between the Chow group of zero cycles and the subgroup of elements of \(\pi_1^{\mathrm{ab}}(X)\) whose images in the absolute Galois group of \(\mathbb{F}\) are integral Frobenius powers (\(W\) stands for Weil). If \(X\) is not necessarily projective but still smooth, \textit{A. Schmidt} and \textit{M. Spieß} [J. Reine Angew. Math. 527, 13--36 (2000; Zbl 0961.14013); \textit{A. Schmidt} [Algebra Number Theory 1, No. 2, 183--222 (2007; Zbl 1184.19002)] showed that an analogous reciprocity isomorphism \(r_X:H_0^S(X, Z) \mathop{\to}\limits^{\sim} \pi_1^{t,\mathrm{ab}}(X)_W\) holds when replacing the Chow group by the \(0\)-th Suslin homology group, and the fundamental group by its tame version. This does not extend to non smooth schemes, for which \(r_X\) is generally neither injective nor surjective. In the present paper, the authors show that the Schmidt-Spiess result can be generalized to singular varieties on replacing the Suslin homology \(H_i^S(X,A)\) by its refined Weil-Suslin version \(H_{i+1}^{\mathrm{WS}}(X,A)\), and the usual abelian tame fundamental group \(\pi_1^{t,\mathrm{ab}}(X)\) by the ``enlarged'' \(\Pi_1^{t,\mathrm{ab}}(X)\) (SGA3, X, \S6), of which \(\pi_1^{t,\mathrm{ab}}(X)\) is the profinite completion. In more detail, for any connected scheme \(X\), separated and of finite type over a finite field \(\mathbb{F}\), they prove two main results: {\parindent=6mm \begin{itemize}\item[1)] The pro-group \(\Pi_1^{t,\mathrm{ab}}(X)_W\) is isomorphic to a (constant) finitely generated abelian group. \item[2)] There exists a refined reciprocity map \(\mathrm{rec}_X : H_1^{\mathrm{WS}}(X, Z) \twoheadrightarrow \Pi_1^{t,\mathrm{ab}}(X)_W\) which is surjective, and such that the composite with the natural map \(H_0^S(X, Z) \to H_1^{\mathrm{WS}}(X, Z)\) is the previous reciprocity map \(r_X\). The kernel of \(\mathrm{rec}_X\) contains the maximal divisible subgroup of \(H_1^{\mathrm{WS}}(X, Z)\), and is equal to it if resolution of singularities for schemes of dimension \(\leq 1+\dim X\) holds over \(\mathbb{F}\). As a corollary, one obtains an isomorphism of profinite completions \(\mathrm{rec}_X^\wedge: H_1^{\mathrm{WS}}(X, Z) ^\wedge \mathop{\to}\limits^{\sim} \pi_1^{t,\mathrm{ab}}(X)\). Under Parshin's conjecture, \(H_1^{\mathrm{WS}}(X, Z)\) is finitely generated, hence \(\mathrm{rec}_X\) would be an isomorphism. \end{itemize}}