Explicit Heegner points: Kolyvagin's conjecture and non-trivial elements in the Shafarevich-Tate group (Q999708)

Let \(E/\mathbb{Q}\) be an elliptic curve and consider its analytic rank \(r_{\text{an}}(E/\mathbb{Q})\) (the order of vanishing of the \(L\)-function \(L(E/\mathbb{Q},s)\) at \(s=1\)) and its Mordell-Weil rank \(r_{\text{MW}}(E/\mathbb{Q})\). Conjecturally \(r_{\text{an}}(E/\mathbb{Q})=r_{\text{MW}}(E/\mathbb{Q})\) and this result is known for curves of analytic rank \(\leq 1\) by the works of \textit{V.A. Kolyvagin} [Proc. Int. Congr. Math., Kyoto/Japan 1990, Vol. I, 429--436 (1991; Zbl 0749.14012)] and \textit{B. Gross -D. Zagier} [Invent. Math. 84, 225--320 (1986; Zbl 0608.14019)]. In particular the work of Kolyvagin provides bounds for the Selmer group via the construction of a system of cohomology classes arising from Heegner points and is based on the assumption that such a system contains a non-trivial element. The paper under review presents an algorithm to (almost) compute explicit Heegner points in ring class fields and then obtain Kolyvagin's cohomology classes from them. The Heegner points are approximated via the modular parametrization of the elliptic curve \(E\) (i.e., via truncated Fourier expansion of the modular form associated to \(E\)), the accuracy in the computation is provided by upper bounds for the heights of Heegner points presented in the Appendix of the paper. The authors give several examples of computations producing non-trivial cohomology classes (in particular the first known example for elliptic curves of rank 2) and provide interesting consequences. For some curves of analytic rank 1 they show how to construct explicit points in the Shafarevich-Tate group from the Heegner points. For curves of analytic rank greater than 1 they show that the presence of a non-trivial cohomology class (plus some mild hypotheses on a prime \(p\)) implies that the \(\mathbb{Z}_p\)-corank of the \(p\)-Selmer group of \(E\) is at least 2 (resp. 3) if \(r_{\text{an}}(E/\mathbb{Q})\) is even and \(\geq 2\) (resp. odd and \(\geq 3\)).