Characteristic ideals and Selmer groups (Q2517149)

Let \(A\) be an abelian variety defined over a global field of prime characteristic \(p\) and let \(\mathcal F/F\) be a \(\mathbb Z_p^N\)-extension, unramified outside a finite set of places of \(F\). The main goal of this article is to construct a pro-characteristic ideal associated to the Pontryagin dual of the \(p\)-primary Selmer group of \(A\), in view of an Iwasawa Main Conjecture (IMC) in this non-Noetherian commutative setting. Recall that for any finite extension \(E/F\), the \(p\)-part of the Selmer group of \(A\) over \(E\) is defined as \[ \mathrm{Sel}(E):=\mathrm{Sel}_A(E)_p:=\mathrm{Ker}\{H_{fl}^1(X_E, A[p^\infty]) \to\prod_v H_{fl}^1(X_{E_v}, A[p^\infty])\}, \] where \(H_{fl}^1\) denotes flat cohomology, \(X_E:=\mathrm{Spec}(E)\) and the map is the product of the natural restrictions at all places \(v\) of \(E\). For an infinite algebraic extension \(K/F\), define \(\mathrm{Sel}(K)\) by taking inductive limits over all the finite subextensions, and let \(\mathcal S(K)\) denote the Pontryagin dual of \(\mathrm{Sel}(K)\). To define the pro-characteristic ideal of \(\mathcal S(\mathcal F)\), take a \(\mathbb Z_p\)-basis \(\{\gamma_i\}\) for \(\mathrm{Gal}(\mathcal F/F)\) and, for any \(d \geq 0\), let \(\mathcal F_d\) be the fixed field of \(\gamma_1, \dots, \gamma_d\); denote by \(\pi_{d-1}^d\) the natural projection between Iwasawa algebras \(\Lambda(\mathcal F_d) \to \Lambda(\mathcal F_{d-1})\). Since \(\Lambda(\mathcal F) = \mathrm{proj.lim } \Lambda(\mathcal F_d)\) and \(\mathcal S(\mathcal F) = \mathrm{proj.lim } \Lambda(\mathcal S_d)\), it is natural to define \(\widetilde{{\mathrm{Char}}}_{\Lambda(\mathcal F)}(\mathcal S(\mathcal F))\) as being \(\mathrm{proj.lim } {\mathrm{Char}}_{\Lambda(\mathcal F_d)} (\mathcal S(\mathcal F_d))\), where \({\mathrm{Char}}_{\Lambda(\mathcal F_d )} (. )\) denotes the characteristic ideals classically defined in \(\Lambda(\mathcal F_d ) \cong \mathbb Z_p [[t_1,\dots,t_d]]\), \(t_i = \gamma_{i-1}\). But one needs first check that the ideals \({\mathrm{Char}}_{\Lambda(\mathcal F_d)} (\mathcal S(\mathcal F_d ))\) form a projective system with respect to the maps \(\pi_{d-1}^d\). In the present context, a result of \textit{T. Ochiai} [Ann. Inst. Fourier 55, No. 1, 113--146 (2005; Zbl 1112.11031)], can be applied directly to show this is the case modulo two simple conditions, say (C), which we do not recall. Assuming that all ramified places in \(\mathcal F/ F\) are of ordinary reduction and that there exists an \(e > 0\) such that \(\mathcal S (\mathcal F_e)\) is \(\Lambda(\mathcal F_e)\)-torsion, the authors first compute a precise relation, say (R) (which we do not recall either), between \({\mathrm{Char}}_{\Lambda(\mathcal F_{d-1})} (\mathcal S(\mathcal F_{d-1} ))\) and \(\pi_{d-1}^d({\mathrm{Char}}_{\Lambda(\mathcal F_d)} (\mathcal S(\mathcal F_d ))\). Then, assuming that the ramified places of \(\mathcal F/F\) are totally ramified (examples are the extensions obtained from \(F\) by adding the \(\mathfrak A^n\)-torsion points of a normalized rank 1 Drinfeld module over \(F\)), they show that (R) implies (C) and, even more precisely, that the equality \(\pi_{d-1}^d({\mathrm{Char}}_{\Lambda(\mathcal F_d)} (\mathcal S(\mathcal F_d )) = {\mathrm{Char}}_{\Lambda(\mathcal F_{d-1})} (\mathcal S(\mathcal F_{d-1} ))\) holds for \(d \gg 0\), which allows to define \(\widetilde{{\mathrm{Char}}}_{\Lambda(\mathcal F)} (\mathcal S(\mathcal F))\). As an application, using a deep result of \textit{K. F. Lai}, \textit{I. Longhi}, \textit{K.-S. Tan} and \textit{F. Trihan} [``The Iwasawa Main conjecture of constant ordinary abelian varieties over function fields'', Preprint, \url{arXiv:1406.6125}], they prove the following non-Noetherian commutative IMC: ``Suppose that \(A/F\) is a constant abelian variety and let \(\mathcal F/F\) be totally ramified as above. There exists an element \(\theta_{A,\mathcal F}\) interpolating the classical \(L\)-function \(L(A, \chi, 1)\) (where \(\chi\) varies among characters of \(\mathrm{Gal} (\mathcal F/F ))\) such that \(\widetilde{{\mathrm{Char}}}_{\Lambda(\mathcal F)} (\mathcal S(\mathcal F)) = (\theta_{A,\mathcal F})\).'' Actually, Lai et al. [loc. cit.] defined \(\theta_{A,M}\) for any abelian extension \(M/F\) unramified outside a finite set of primes, they showed that \(\pi_{d-1}^d(\theta_{A,\mathcal F_d} ) = \theta_{A,\mathcal F_{d-1}}\), and proved the interpolation formula as well as the equality \({\mathrm{Char}}_{\Lambda(\mathcal F_d)} (\mathcal S(\mathcal F_d ))=( \theta_{A,\mathcal F_d} )\) for all \(d\). The authors' contribution here consists in taking the limit.