The even parity Goldfeld conjecture: congruent number elliptic curves (Q2235727)

The aim of the paper under review is to give an introduction to some recent advances towards Goldfeld's conjecture on the rank of quadratic twists of elliptic curves, and in particular to a recent \(p\)-converse theorem proved by the authors of the paper under review, by looking at the specific family of congruent number elliptic curves. Let us be more precise. A square-free integer \(n \in \mathbb{N}\) is said to be a \textit{congruent number} if there exist \(a,b \in \mathbb{Q}^\times\) such that \(a b = 2n\) and \(\sqrt{a^2 + b^2} \in \mathbb{Q}\). It is a very old and still open problem to determine which natural numbers are congruent. Nevertheless, it is easy to see that a natural number \(n \in \mathbb{N}\) is congruent if and only if the equation \(y^2 = x^3 - n^2 x\) has a solution \((x,y) \in \mathbb{Z}^2\) with \(y \neq 0\). This equation represents a quadratic twist \(E^{(n)}\) of the elliptic curve \(E \colon y^2 = x^3 - x\), and it is easy to see that the torsion points in \(E^{(n)}(\mathbb{Q})\) are precisely those with \(y = 0\). Thus, \(n \in \mathbb{N}\) is congruent if and only if the group \(E^{(n)}(\mathbb{Q})\) has positive rank. These considerations link the congruent number problem to the arithmetic of elliptic curves. In particular, the Birch and Swinnerton-Dyer conjecture predicts that \(E^{(n)}\) has positive rank if and only if the associated \(L\)-function \(L(E^{(n)},s)\) vanishes at \(s = 1\). Moreover, the work of \textit{J. B. Tunnell} [Invent. Math. 72, 323--334 (1983; Zbl 0515.10013)] has shown that \(L(E^{(n)},1) = 0\) if and only if the equation \(n = a(2 a x^2 + y^2 + 8 z^2)\), where \(a = 2\) if \(2 \mid n\) and \(a = 1\) if \(2 \nmid n\), has as many solutions \((x,y,z)\) with \(2 \mid z\) as it has solutions with \(2 \nmid z\). This property of the integer \(n\) can be checked algorithmically, and thus provides a way (conditional on the Birch and Swinnerton-Dyer conjecture) to solve the old congruent number problem. Note moreover that if \(n\) does not satisfy the latter property, then \(n\) is not a congruent number, because the Birch and Swinnerton-Dyer conjecture has been shown to hold by \textit{J. Coates} and \textit{A. Wiles} [Invent. Math. 39, 223--251 (1977; Zbl 0359.14009)] for elliptic curves with complex multiplication, defined over \(\mathbb{Q}\), whose \(L\)-function does not vanish at \(s = 1\). On the other hand, a conjecture of \textit{D. Goldfeld} [Number theory, Proc. Conf., Carbondale 1979, Lect. Notes Math. 751, 108-118 (1979), Zbl 0417.14031] predicts that \(100\%\) of those quadratic twists of an elliptic curve defined over \(\mathbb{Q}\) which have positive root number should have algebraic rank zero, and \(100\%\) of those twists with negative root number should have algebraic rank one. Here, the root numbers are the signs of the functional equations satisfied by the \(L\)-functions of the quadratic twists. In our particular situation, it is easy to see that \(E^{(n)}\) has positive root number if and only if \(n \equiv 1, 2, 3 (8)\). Thus, Goldfeld's conjecture predicts that \(100\%\) of the square-free integers \(n \equiv 1, 2, 3 \, (8)\) are not congruent numbers. This prediction has been recently verified by \textit{A. Smith} [``\(2^\infty\)-Selmer groups, \(2^\infty\)-class groups, and Goldfeld's conjecture'', Preprint, \url{arXiv:1702.02325}]. More precisely, Smith proves that for \(100\%\) of the square-free integers \(n \equiv 1, 2, 3 \, (8)\), the \(2^\infty\)-Selmer group \(\mathrm{Sel}_{2^\infty}(E^{(n)})\) is finite. This implies that \(E^{(n)}\) has algebraic rank zero, because \(E^{(n)}(\mathbb{Q}) \otimes_\mathbb{Z} \mathbb{Q}_2/\mathbb{Z}_2 \subseteq \mathrm{Sel}_{2^\infty}(E^{(n)})\). In order to relate the two aforementioned conjectures, one can employ a so-called \(p\)-converse theorem, i.e., a result which assures that if \(E'\) is an elliptic curve defined over \(\mathbb{Q}\) and \(\mathrm{rank}_{\mathbb{Z}_p}(\mathrm{Sel}_{p^\infty}(E')^\vee) = r\) then the order of vanishing of \(L(E',s)\) at \(s = 1\) is (at least) \(r\). For elliptic curves with complex multiplication, \textit{K. Rubin} [Invent. Math. 103, No. 1, 25--68 (1991; Zbl 0737.11030)] proved such a \(p\)-converse when \(r = 0\) and \(p \nmid \# \mathrm{Aut}_{\overline{\mathbb{Q}}}(E')\). In particular, the case \(p = 2\) remained open until the recent breakthrough provided by work of the authors of the paper under review (see [``A rank zero p-converse to a theorem of Gross-Zagier, Kolyvagin and Rubin'', Preprint]). In particular, as explained in Section 4 of the paper under review, the authors achieve such a \(p\)-converse by proving the main conjecture of Iwasawa theory for modular forms, due to \textit{K. Kato} (see Conjecture 12.10 in [Astérisque 295, 117--290 (2004; Zbl 1142.11336)]), for those modular forms associated to CM elliptic curves defined over \(\mathbb{Q}\). Thus, combining this \(p\)-converse with the aforementioned work of Smith, one sees that for \(100\%\) of square-free integers \(n \equiv 1, 2, 3 \, (8)\), one has that \(\mathcal{L}(n) = 0\), therefore confirming half of the analytic variant of Goldfeld's conjecture for the family of congruent number elliptic curves. To conclude, the paper under review provides a nice introduction to the vast web of conjectures surrounding congruent numbers, and to the recent breakthroughs in these subjects.