Classical Iwasawa theory and infinite descent on a family of abelian varieties (Q2036276)

Let \(q \equiv 7 \mod 16\) be a prime and set \(K := \mathbb{Q}(\sqrt{-q})\). \textit{B. H. Gross} [Arithmetic on elliptic curves with complex multiplication. Berlin etc.: Springer-Verlag (1980; Zbl 0433.14032)] constructed an interesting elliptic curve \(A\) defined over the Hilbert class field \(H = K(j(\mathcal{O}_K))\) of \(K\) with complex multiplication by \(\mathcal{O}_K\), minimal discriminant \((-q^3)\), and isogenous to all of its conjugates. This paper is concerned with the Iwasawa theory of this family of elliptic curves for the prime \(p=2\). This includes new cases of a meanwhile classical conjecture of \textit{R. Greenberg} [Adv. Stud. Pure Math. 30, 335--385 (2001; Zbl 0998.11054)] on pseudo-nullity of Iwasawa modules and a new proof of a theorem of \textit{H. Hasse} [J. Number Theory 1, 231--234 (1969; Zbl 0167.32302)]. The latter answers the question when the class number of \(\mathbb{Q}(\sqrt{-2q})\) is exactly divisible by \(4\). \par Let us introduce a little more notation. The prime \(2\) splits in \(K\). Let \(\mathfrak{p}\) and \(\mathfrak{p}^{\ast}\) be the primes above \(2\). Hence there are unique \(\mathbb{Z}_2\)-extensions \(K_{\infty}\) and \(K_{\infty}^{\ast}\) unramified outside \(\mathfrak{p}\) and \(\mathfrak{p}^{\ast}\), respectively. Let \(\alpha\) be a fixed root of the polynomial \(x^4+q\). Then \(J := K(\alpha, \sqrt{-1})\) is its splitting field. We set \(J_{\infty} := JK_{\infty}\) and likewise \(J_{\infty}^{\ast} := JK_{\infty}^{\ast}\). Let \(X(J_{\infty})\) be the Galois group over \(J_{\infty}\) of the maximal abelian pro-\(2\)-extension which is unramified outside the primes above \(\mathfrak{p}\). We define \(X(J_{\infty}^{\ast})\) similarly. The first main result asserts that both Iwasawa modules, \(X(J_{\infty})\) and \(X(J_{\infty}^{\ast})\), are free \(\mathbb{Z}_2\)-modules of rank \(1\). \par The relation to the Gross family is as follows. We first recall a few facts about this family. Let \(h\) be the class number of \(K\) which is known to be odd. Let \(B\) be the \(h\)-dimensional abelian variety which is obtained by restriction of scalars of \(A\) from \(H\) to \(K\). Let \(\mathcal{B} = \mathrm{End}_K(B)\). Then \(\mathcal{B}\) is an order in the CM field \(\mathcal{T} = \mathcal{B} \otimes \mathbb{Q}\) which is ramified over \(\mathcal{O}_K\) precisely at the primes dividing \(h\). The action of \(\mathcal{B}\) upon the torsion subgroup of \(B(K)\) induces a surjection from \(\mathcal{B}\) onto \(\mathcal{O}_K / 2 \mathcal{O}_K\) whose kernel is the product of two primes \(\mathfrak{P}\) and \(\mathfrak{P}^{\ast}\) lying above \(\mathfrak{p}\) and \(\mathfrak{p}^{\ast}\), respectively. \par Then \(J_{\infty}\) is obtained from \(D := K(\sqrt{-1})\) by adjoining all \(\mathfrak{P}\)-power division points, and similarly for \(J_{\infty}^{\ast}\). Let \(\Sha(B/J_{\infty})\) be the Tate-Shafarevich group of \(B/J_{\infty}\). Then the above result for \(X(J_{\infty})\) can be reformulated as follows. Either \(B(J_{\infty}) \otimes \mathbb{Q}\) has \(\mathbb{Q}\)-dimension \(2h\) and \(\Sha(B/J_{\infty})(\mathfrak{P})\) vanishes or \(B(J_{\infty}) \otimes \mathbb{Q}\) vanishes and \(\Sha(B/J_{\infty})(\mathfrak{P}) \simeq \mathbb{Q}_2 / \mathbb{Z}_2\). Of course, there is again an analogous version with asterisks. The authors indeed show a stronger result over the corresponding \(\mathbb{Z}_2\)-extensions of \(D\). \par So far all results have been obtained by elementary methods (in the authors' words) based only on class field theory and classical Iwasawa theory. Chevalley's formula on class numbers in finite cyclic extensions of number fields is a crucial ingredient. Using more elaborate analytic arguments, the authors finally prove that \(B(D) \otimes \mathbb{Q}\) is \(2h\)-dimensional and that the Tate-Shafarevich group \(\Sha(B/D)\) is finite. They largely follow an argument due to \textit{S. Miller} and \textit{T. Yang} [Math. Res. Lett. 7, No. 2-3, 263--277 (2000; Zbl 1025.11011)]. \par As a further application, it is shown that the composite \(K_{\infty} K_{\infty}^{\ast}\) of all \(\mathbb{Z}_2\)-extension of \(K\) has no non-trivial abelian unramified \(2\)-extension. This in particular verifies Greenberg's conjecture in a special case.