No 17-torsion on elliptic curves over cubic number fields (Q558170): Difference between revisions

For an integer \(d\), let \(S(d)\) denote the set of prime numbers \(p\) for which there exists an elliptic curve \(E\) over a number field \(K\) with \([K: \mathbb{Q}]= d\) and a point \(P\) in \(E(K)\) of order \(p\). In the author's previous paper [Ann. Inst. Fourier 50, No. 3, 723--749 (2000; Zbl 0971.11030)], it is shown that \(S(3)= \{2,3,5,7,11,13\) and maybe \(17\}\). The goal of the present article is to show that 17 does not belong to \(S(3)\). Suppose that there exists an elliptic curve over a cubic number field \(K\) endowed with a \(K\)-rational point of order 17. Then one can associate with it a point \(P= (p_1 ,p_2, p_3)\in X_1(17)^{(3)}(\mathbb{Z}[1/17])\) (symmetric power) such that \(p_i\) are generically non-cuspidal points, but \(P\) coincides in the fiber at 2 with a triplet of cusps \(P_2\in X_1(17)^{(3)}\) above the cusp \(3.\infty\in X_0(17)^{(3)}\). He focuses on a morphism \(F_{P_0}: X_1(17)^{(3)}\to J_1(17)\) defined by \(F_{P_0}(Q)= t(Q- P_0)\), where \(t\) is an element of the Hecke algebra \(\mathbb{T}_{\Gamma_1(17)}\) which kills the 2-torsion of \(J_1(17)\). In order to derive a contradiction, he appeals to [loc. cit., 1.5] by which he is reduced to showing that 1) \(F_{P_0}\) is a formal immersion at \(P_0(\mathbb{F}_2)\) and 2) no non-cuspidal point of \(X_1(17)^{(3)}(\mathbb{Z})\) is mapped by \(F_{P_0}\) to the nontrivial section of a \(\mu_2\)-subscheme of \(J_1(17)_{/\mathbb{Z}}\). The point 1) is already shown to hold in [loc.cit., 4.3], and he finishes the proof by showing in this paper the validity of the assertion 2) using elementary theory of formal groups.