Independence of points on elliptic curves arising from special points on modular and Shimura curves. I: Global results (Q1006537)

Given a morphism \(\Phi\) between a modular curve \(S\) and an elliptic curve \(A\), the authors prove in the present paper that whenever \(\Gamma\) is a finite rank subgroup of \(A(\overline{\mathbb{Q}})\), and \(\text{CM}\) the set of CM points an \(S\), then \(\Phi({\text{CM}})\cap \Gamma\) is finite. This result strengthens a theorem of \textit{J. Nekovář} and \textit{N. Schappacher} [``On the asymptotic behaviour of Heegner points.'' Turk. J. Math. 23, No. 4, 549--556 (1999; Zbl 0977.11025)], among others, who proved it under the hypothesis that \(\Gamma\) be finitely generated. Here, the subgroup \(\Gamma\) is thus allowed to contain points of unbounded degree. Actually, the authors prove a stronger statement, where \(\Phi\) is replaced by a correspondence, and \(S\) (resp. \(A\)) can be any Shimura curve (resp. any abelian variety); and a sequel of the paper contains local versions of the present one. One of the central point of the proof is to take a ``fatten'' version of the subgroup \(\Gamma\), namely, for \(\varepsilon >0\): \[ \Gamma_\varepsilon := \{\gamma+a\; :\; \gamma\in\Gamma, h(a)\leq \varepsilon\} \] where \(h\) is a height function on \(A\). The authors prove that for \(\varepsilon\) small enough, \(\Phi(\text{CM})\cap\Gamma_\varepsilon \) is still finite. The proof is based on previous works on the equidistribution of special points, such as \textit{W. Duke}'s [``Hyperbolic distribution problems and half-integral weight Maass forms.'' Invent. Math. 92, No. 1, 73--90 (1988; Zbl 0628.10029)] and \textit{S.-W. Zhang}'s [``Equidistribution of CM-points on quaternion Shimura varieties.'' Int. Math. Res. Not. 2005, No. 59, 3657--3689 (2005; Zbl 1096.14016)] on one hand, and on estimations of measures on complex manifolds with singularities. Let's explain roughly the use of equidistribution in this context. \textit{L. Szpiro} et al. [``Equirépartition des petits points.'' Invent. Math. 127, No. 2, 337--347 (1997; Zbl 0991.11035)] and subsequent refinements prove that Galois orbits of ``special'' points of an abelian variety \(A\) are equidistributed for the Haar probability of \(A(\mathbb{C})\): the proof uses Arakelov theory. On the other hand, if the intersection \(\Phi(\text{CM})\cap\Gamma_\varepsilon \) were infinite, one could construct a subsequence of ``special'' points weakly converging to the hyperbolic measure of \(S\) (Duke's theorem and generalizations, built on analytic theory of \(L\)-functions). But the pushforward of the hyperbolic measure by the morphism \(\Phi\) is distinct from the Haar probability of \(A(\mathbb{C})\) (this can be easily seen if \(S\) has cusps): this leads to a contradiction. Of course, working with a correspondence induces complications, and a geometric input is necessary. Part II, see Compos. Math. 145, No. 3, 566--602 (2009; Zbl 1232.11073).