On a noncommutative Iwasawa main conjecture for varieties over finite fields (Q2441483): Difference between revisions

The noncommutative Iwasawa main conjecture over totally real fields has been shown recently by \textit{J. Ritter} and \textit{A. Weiss} [J. Am. Math. Soc. 24, No. 4, 1015--1050 (2011; Zbl 1228.11165)] and by \textit{M. Kakde} [Invent. Math. 193, No. 3, 539--626 (2013; Zbl 1300.11112)] independently. Here the author formulates and proves an analogue for \(\ell\)-adic Lie extensions of a separated scheme \(X\) of finite type over a finite field \(\mathbb F_q\) of characteristic prime to \(\ell\). Assume first that \(X\) is geometrically connected and let \(G\) be a factor group of the fundamental group of \(X\) such that \(G\) is the semi-direct product of a compact \(\ell\)-adic Lie group \(H\) and of \(\Gamma = \mathrm{Gal}(\mathbb F_(q^\infty)/\mathbb F_q) = \mathbb Z_\ell\). Write \(\Lambda_G\) for the complete algebra \(\mathbb Z_\ell[[G]]\). Let \(S = \{f \in \Lambda_G : \Lambda_G/\Lambda_Gf\text{ is of finite type over }\Lambda_H\}\) be the canonical Ore set. Recall the exact localisation sequence of algebraic K-groups \(K_1(\Lambda_G) \to K_1((\Lambda_G)_S) \to^d K_0(\Lambda_G,(\Lambda_G)_S) \to 0\). Let us turn \(\Lambda_G\) into a smooth \(\Lambda_G\)-sheaf \(\mathcal M(G)\) on \(X\) via the contragredient action of the fundamental group of \(X\) on \(\Lambda_G\). Every continuous \(\mathbb Z_\ell\)-representation \(\rho\) of \(G\) induces a homomorphism \(\rho\) of \(K_1(\Lambda_G)\) into the units \(Q(\Lambda_G)^{\times}\) of the field of fractions of \(\Lambda_G\). On the other hand, \(\rho\) gives rise to a flat and smooth \(\mathbb Z_\ell\)-sheaf \(\mathcal M(\rho)\) on \(X\). Let \(\mathcal F\) be the compact cohomology of a flat constructible \(\mathbb Z_\ell\)-sheaf \(\mathcal F\) on \(X\). The Grothendieck trace formula expressing the \(L\)-function \(L(\mathcal F,T)\) as the alternate product of the determinants \(\det(1-T\phi : H_c^i (\bar{X}, \mathcal F))\), where \(\phi\) is the geometric Frobenius, implies that \(L(\mathcal F,T)\) is in fact a rational function. The author's main result then reads: {\parindent=6mm\begin{itemize}\item[1)] \(R\Gamma_c(X, \mathcal M(G)\otimes\mathcal F)\) is a perfect complex of \(\Lambda_G\)-modules whose cohomology is S-torsion. Moreover, \(id-\phi\) is a quasi-automorphism of the complex \(R\Gamma_c(\bar{X}, \mathcal M(G)_S\otimes\mathcal F)\), and hence gives rise to an element \(\mathcal L_G (X/\mathbb F_q, \mathcal F) = [id-\phi]^{-1}\) in \(K_1(\Lambda_G)_S)\). \item[2)] \(d(\mathcal L_G (X/\mathbb F_q, \mathcal F)) =[R\Gamma_c (X,\mathcal M(G)\otimes\mathcal F)]^{-1}\). \item[3)] If the representation is continuous, then \(\rho(\mathcal L_G (X/\mathbb F_q, \mathcal F)) = L(\mathcal M(\rho)\otimes\mathcal F, [\phi]^{-1})\) in \(Q(\Lambda_G)^{\times}\). \end{itemize}} This theorem implies the expected interpolation property of \(\mathcal L_G (X/\mathbb F_q, \mathcal F)\) with respect to special values of \(L\)-functions. In the final section of the paper, the limitation to geometrically connected schemes is overcome by allowing \(G\) (which is then only required to be a virtual pro-\(\ell\)-group) to be the covering group of any suitable principal covering of \(X\). Note that using the results of this paper and of \textit{M. Emerton} and \textit{M. Kisin} [Ann. Math. (2) 153, No. 2, 329--354 (2001; Zbl 1076.14027)], \textit{D. Burns}, ``On main conjectures of geometric Iwasawa theory and related conjectures'' Preprint (2011)] has recently constructed, in the case when \(\ell\) divides \(q\), a modification of \(\mathcal L_G (X/\mathbb F_q, \mathcal F)\) which has the right interpolation property. Also in the case when \(\ell = p\), Trihan and Vauclair have announced proofs of general main conjectures for varieties over function fields.