Fine Selmer groups of elliptic curves over \(p\)-adic Lie extensions (Q1772094): Difference between revisions

Authors' introduction: The aim of this paper is to discuss variants for elliptic curves of some deep conjectures of classical cyclotomic Iwasawa theory. Let \(F\) be a finite extension of \(\mathbb Q\), \(p\) an odd prime number, and \(F^{\text{cyc}}\) the cyclotomic \(\mathbb Z_p\)-extension of \(\mathbb Q\). Let \(K(F^{\text{cyc}})\) denote the maximal unramified abelian \(p\)-extension of \(F^{\text{cyc}}\), in which every prime of \(F^{\text{cyc}}\) above \(p\) splits completely, and define \(Y(F^{\text{cyc}})\) to be the Galois group of \(K(F^{\text{cyc}})\) over \(F^{\text{cyc}}\). Put \(\Gamma = G(F^{\text{cyc}}/F)\), and write \(\Lambda(\Gamma)\) for the Iwasawa algebra of \(\Gamma\). Iwasawa proved that \(Y(F^{\text{cyc}})\)is always a finitely generated torsion \(\Lambda(\Gamma)\)-module, and he conjectured [\textit{K. Iwasawa}, Number Theory, Algebraic Geometry, Commutative Algebra, in Honor of Y. Akizuki, 1--11 (1973; Zbl 0281.12005)] that in fact \(Y(F^{\text{cyc}})\) is always a finitely generated \(\mathbb Z_p\)-module. At present, this conjecture has only been proven when \(F\) is abelian over \(\mathbb Q\) [see \textit{B. Ferrero} and \textit{L. C. Washington}, Ann. Math. (2) 109, 377--395 (1979; Zbl 0443.12001), \textit{W. Sinnott}, Invent. Math. 75, 273--282 (1984; Zbl 0531.12004)]. Perhaps surprisingly, it does not seem to have been pointed out in the literature that there is a precise analogue of Iwasawa's conjecture for elliptic curves over \(F^{\text{cyc}}\). Let \(E\) be an elliptic curve over \(F\), and let \(R(E/F^{\text{cyc}})\) be the fine Selmer group of \(E\) over \(F^{\text{cyc}}\). We denote by \(Y(E/F^{\text{cyc}})\) the Pontryagin dual of \(R(E/F^{\text{cyc}})\). An equivalent form of the so called weak Leopoldt Conjecture for \(E\) over \(F^{\text{cyc}}\) is the assertion that the finitely generated \(\Lambda(\Gamma)\)-module \(Y(E/F^{\text{cyc}})\) is \(\Lambda(\Gamma)\)-torsion (see Lemma 3.1).We make the stronger conjecture (Conjecture A) that, parallel to Iwasawa's conjecture, \(Y(E/F^{\text{cyc}})\) is always a finitely generated \(\mathbb Z_p\)-module. Of course, it has long been known [\textit{B. Mazur}, Invent. Math. 18, 183--266 (1972; Zbl 0245.14015)] that the dual of the Selmer group, as opposed to the fine Selmer group, of \(E\) over \(F^{\text{cyc}}\) is not, in general, a finitely generated \(\mathbb Z_p\)-module. Our second conjecture concerns the mysterious phenomenon of certain arithmetic Iwasawa modules for \(p\)-adic Lie extensions of dimension strictly greater than 1 being much smaller than one intuitively expects, and was first noted by \textit{R. Greenberg} [Adv. Stud. Pure Math. 30, 335--385 (2001; Zbl 0998.11054)]. A natural question arising from Greenberg's conjecture is the following. Let \(F_\infty\) be a Galois extension of \(F\) satisfying (i) \(G = G(F_\infty/F)\) is a pro-\(p\) \(p\)-adic Lie group of dimension \(> 1\), and having no element of order \(p\), (ii) \(F_\infty\) contains \(F^{\text{cyc}}\), and (iii) only finitely many primes of \(F\) ramify in \(F_\infty\). Write \(K(F_\infty)\) for the maximal unramified abelian \(p\)-extension of \(F_\infty\) in which every prime above \(p\) splits completely. Is it true that the Galois group \(Y(F_\infty)\) of \(K(F_\infty)\) over \(F_\infty\) is always pseudo-null as a \(\Lambda(G)\)-module (see [\textit{O. Venjakob}, J. Eur. Math. Soc. (JEMS) 4, No. 3, 271--311 (2002; Zbl 1049.16016)] for a discussion of the notion of a pseudo-null module for Iwasawa algebras)? Surprisingly,\textit{Y. Hachimori} and \textit{R. Sharifi} [J. Algebr. Geom. 14, No. 3, 567--591 (2005; Zbl 1085.11054)], have given a series of remarkable counter-examples to the assertion that \(Y(F_\infty)\) is pseudo-null as a module over \(\Lambda(G)\) when \(G\) has dimension bigger than 1, and \(F_\infty\) is a CM-field. Nevertheless, our second conjecture asserts that the analogue of this question for elliptic curves should have an affirmative answer. Specifically, let \(E\) be an elliptic curve defined over \(F\), and let \(Y(E/F_\infty)\) be the dual of the fine Selmer group of \(E\) over \(F_\infty\). Then, recalling that we are assuming that \(G\) has dimension \(> 1\), we conjecture (see Conjecture B) that \(Y(E/F_\infty)\) is pseudo-null as a \(\Lambda(G)\)-module. It seems that a general proof of our pseudo-nullity conjecture is quite inaccessible at present, and one can only hope to investigate it in interesting numerical examples. We do discuss some intriguing numerical examples in \S 4, including one case (see Example 4.8) in which \(G\) is an open subgroup of \(\text{GL}_2(\mathbb Z_5)\), and thus of dimension 4. A natural question arising from our study of these examples (in particular, see Example 4.8) is whether or not every elliptic curve over \(\mathbb Q\) has a point of infinite order in the field generated over \(\mathbb Q\) by its \(p\)-power division points, for any fixed prime \(p\).