We determine all the possible torsion groups of elliptic curves over cyclic cubic fields, over non-cyclic totally real cubic fields and over complex cubic fields.
IntroductionAn important problem in the theory of elliptic curves is to determine the possible torsion groups of elliptic curves over number fields of degree d. Mazur solved the problem for d = 1 [14] and Kamienny [12], building on the work of Kenku and Momose [13], solved the problem for d = 2. Recently, Derickx, Etropolski, van Hoeij, Morrow and Zuerick-Brown announced the solution of the problem for d = 3 [5], building on the work of Parent [17, 18]. For d > 3 the problem remains unsolved at the moment.A question that naturally arises is which of the groups that arise as torsion groups of elliptic curves over number fields of degree d arise over some natural subset of the set of number fields of degree d. These subsets of the set of all number fields of degree d can be chosen to be, perhaps most naturally, the subset of real (or totally real) number fields, the subset of complex number fields, and the subset of number fields whose normal closure over Q has Galois group isomorphic to some prescribed group G. Throughout the paper, by abuse of language we say that a number fields K has Galois group G over Q if the Galois group over Q of the normal closure of K over Q is G.These problems are of course meaningless for the case d = 1, and for d = 2 one can only consider the subdivision of quadratic fields into real and imaginary quadratic fields. The possibilities in each of these cases (for d = 2) follow directly from [3] and are summed up in Theorem 3.1. Thus the problem has been solved completely for d = 2, so it is natural to consider the case d = 3, which is the last case where all the possible torsion groups are known. The main result of the paper is the determination of all possible torsion groups of elliptic curves over a) All cubic fields with Galois group Z/3Z. b) All complex cubic fields. c) All totally real cubic fields with Galois group S 3 .Parts of some proofs of our results are based on computations in Magma [1]. Some of these computations depend on a magma package writen by Solomon Vishkautsan and the first author [7]. All computations done in Magma for this paper can be found at https://github.com/wishcow79/chabauty/tree/master/pWe would like to thank Solomon Vishkautsan for making his Chabauty code public and his help in explaining on how to use it.