The geometry of flex tangents to a cubic curve and its parameterizations (Q765860)

This article's concern is the problem of producing points in elliptic curves over finite fields. One way is to find a so-called pseudo-parametrization, that is, a parametrization using field operations and extraction of cubic roots. Here this is achieved by studying the dual of the elliptic curve (take a nonsingular plane cubic model), which itself is an algebraic curve. The idea is as follows: one can cut the elliptic curve \(C\) by a family \(L_t\) of lines depending rationally on a parameter \(t\), producing triples of points whose coordinates are calculated by solving cubic equations. This solution involves a square root extraction, so one looks for lines such that the discriminant of the intersection is actually a square. Now, dual space is the set of lines of the projective plane, and those lines intersecting \(C\) in multiple points (that is, tangents and flexes) make up the dual curve \(\widehat{C}\); given a family of lines, the discriminant of \(L_t\cdot C\) describes the intersections between the rational curve \(L\) (in dual space) and \(\widehat{C}\). But this discriminant is a square precisely when every point of \(L\cdot\widehat{C}\) has even multiplicity. The geometric aspect is that the nine flexes of an elliptic cube correspond to nine cusps on the dual curve. The game is then to find rational curves in dual space that pass through some of these cusps and have only even intersections. The conclusion is that several pseudo-parametrizations shown in other articles can be produced with this geometric method, indeed the authors show how infinitely many inequivalent ones arise from different choices of rational curves. There is also a connection to rational curves on \(K3\) surfaces.