Recursos
Resumen
Suppose someone hands you an elliptic curve over Q with a rational point of order two, and its 2-power Selmer corank. From that you can work out its analytic rank, its Mordell–Weil rank and whether Sha is finite. You can also say when the theorem has nothing to tell you.
The rule
The rational-two-torsion 2-converse ([0:59]) states: if E(Q)[2] ≠ 0 and s₂(E) ∈ {0, 1}, then ord_{s=1} L(E,s) = rank E(Q) = s₂(E) and Sha(E/Q) is finite. Reduction at 2 can be anything, CM curves are included, and one rational 2-torsion line is enough. The theorem does not prove the BSD leading-coefficient formula.
Keep in mind why the input must be the corank ([0:24]): s₂ = r + corank Sha[2^∞]. So s₂ = 1 does not by itself give you a point of infinite order, and knowing only the rank does not satisfy the hypothesis.
Reading cases
- 2-torsion and s₂ = 1: a simple zero, rank 1, Sha finite ([1:35]).
- 2-torsion and rank 1, with s₂ unknown: the theorem is silent.
- No rational 2-torsion: outside this theorem. A companion paper handles that case.
- 2-torsion and s₂ = 0: L(E,1) ≠ 0, rank 0, Sha finite, root number +1 ([2:16]).
Why it works
The curve sits at corner 0 of a binary cube of quadratic twists indexed by F₂ᵇ ([2:42]). Every other corner is built to have the required analytic order and a valuation bound that stays uniform. Expanding the test values gives terms weighted by 2^{|I|}. Once b > r(2^M − 1), some nonzero corner agrees with corner 0 modulo 2^M (a Chevalley–Warning count). If the value at your curve were zero, that corner would break its bound ([3:25]). For corank 0 the inputs are Kato classes and central values. For corank 1 they are Heegner sums, a fixed companion k with s₂(E/K) = 1 + 0, and ring-class interpolation ([4:30]).