On ring class eigenspaces of Mordell-Weil groups of elliptic curves over global function fields (Q958717): Difference between revisions

Let \(F\) be a function field of transcendence degree 1 over its finite constant field of odd characteristic and \(E_{/F}\) a non-isotrivial elliptic curve. One can always assume that there exists a place of \(F\) (called \(\infty\)) where \(E\) has split multiplicative reduction: write \(\mathfrak{n}\infty\) for the conductor of \(E\) with \(\mathfrak{n}\) an ideal of \(A\) (the subring of \(F\) consisting of elements that are regular outside \(\infty\)). Consider a quadratic extension \(K/F\) in which \(\infty\) is inert (i.e., \(K/F\) is ``imaginary'') and any prime dividing \(\mathfrak{n}\) splits (i.e., \(K\) satisfies the Heegner hypothesis). In this setting (and with the essential ingredient of the modularity of \(E\)), one can associate to any ideal \(\mathfrak{c}\) of \(A\) such that \((\mathfrak{c},\mathfrak{n})=1\), a point \(\alpha_{\mathfrak{c}}\in E(H_{\mathfrak{c}})\) (a ``Drinfeld-Heegner'' point), where \(H_{\mathfrak{c}}\) is the ring class field of \(K\) of conductor \(\mathfrak{c}\). Let \(G_{\mathfrak{c}}=\text{Gal}(H_{\mathfrak{c}}/K)\), then the action of \(G_{\mathfrak{c}}\) on \(E(H_{\mathfrak{c}})\otimes_{\mathbb{Z}} \mathbb{C}\) produces a decomposition in eigenspaces \[ E(H_{\mathfrak{c}})\otimes_{\mathbb{Z}} \mathbb{C} = \bigoplus_{\chi\in \widehat{G}_{\mathfrak{c}}} E(H_{\mathfrak{c}})^\chi \] where \(\widehat{G}_{\mathfrak{c}}=\text{Hom}(G_{\mathfrak{c}},\mathbb{C}^*)\). Let \(\alpha_{\mathfrak{c},\chi}\) be the projection of \(\alpha_{\mathfrak{c}}\) on the \(\chi\)-component, then the main result of the paper is the following: if \(\alpha_{\mathfrak{c},\chi}\neq 0\), then \(\dim_{\mathbb{C}} E(H_{\mathfrak{c}})^\chi = 1\) (Theorem 1.1). The author follows closely the strategy of the proof of the analogous result for number fields given in \textit{M. Bertolini} and \textit{H. Darmon} [J. Reine Angew. Math. 412, 63--74 (1990; Zbl 0712.14008)], adapting the arguments to the function field setting. The first step of the proof is the study of the torsion \(E[\ell]\) (in particular of \(E[\ell](H_{\mathfrak{c}})\,\)) as a Galois module for an infinite set of auxiliary primes \(\ell \neq p\) (which basically uses Igusa's ``open image'' theorem and Chebotarev density theorem). This allows to reduce the main theorem to a mod \(\ell\) statement like: if \(\alpha_{\mathfrak{c},\chi}\neq 0\) in \(E(H_{\mathfrak{c}})/\ell E(H_{\mathfrak{c}})\), then \(\dim_{\mathbb{F}_\ell} \left( E(H_{\mathfrak{c}})/\ell E(H_{\mathfrak{c}}) \right)^\chi = 1\). Then the main ingredient is the Euler system of Heegner points (the \(\alpha_{\mathfrak{c}+\mathfrak{d}}\) above, with their compatibility relations as \(\mathfrak{d}\) varies through the ideals of \(A\) coprime with \(\mathfrak{c}\)), which produces classes in the cohomology groups \(H^1( G_{\mathfrak{c}+\mathfrak{d}}, E(H_{\mathfrak{c}+\mathfrak{d}}))[\ell]\). Via Tate's duality, those classes enable the author to control the size of the classical Selmer groups and then the rank of the \(\chi\)-components of the Mordell-Weil groups (with some extra effort required to show that they are all of the same dimension). The special case \(\chi =\mathbf{1}\) produces a different proof (with more ``classical'' techniques) of a result of \textit{M. L. Brown} [Heegner modules and elliptic curves. Lecture Notes in Mathematics 1849. Berlin: Springer (2004; Zbl 1146.11029)] which reads as: if \(Tr^{H_{\mathfrak{c}}}_K(\alpha_{\mathfrak{c}}) \in E(K)\) is not a torsion point, then the rank of \(E(K)\) is 1 and the \(\ell\)-part of the Tate-Shafarevich group of \(E_{/K}\) is trivial for infinitely many primes \(\ell \neq p\). This is known to imply the Birch and Swinnerton-Dyer conjecture for \(E_{/K}\) by the work of \textit{K. Kato} and \textit{F. Trihan} [Invent. Math. 153, No.~3, 537--592 (2003; Zbl 1046.11047)].