Recursos
Resumen
Take a prime like 17, which leaves remainder 8 when divided by 9, and explain why it must be a sum of two rational cubes. The answer comes from a theorem that turns a small Selmer group into a statement about the L-function, the Selmer converse, together with a proof by contradiction built on two incompatible lengths.
The converse, and how it is proved
- Gross–Zagier–Kolyvagin go from analytic rank ≤ 1 to arithmetic. The converse goes the other way: if the p-power Selmer group has Z_p-corank 0 or 1, then analytic rank = Mordell–Weil rank = that corank, and all of Sha is finite. This holds for every curve over Q and every prime p [0:15].
- Reduction: for odd p, auxiliary fields K ⊂ L reduce everything to one point, the conductor-one Heegner trace y. Assume y is torsion [0:54].
- Upper bound (cohomology): the tame-deformed Greenberg Selmer group over R_b = k[t]/(t^b) has length < b*, where b* = 1, or 3 in the strict case [1:20].
- Lower bound (theta lifts): a holomorphic theta family on U(3,1) has small constant terms (because y is torsion), but a positive Fourier–Jacobi coefficient keeps it alive. Lifting to cusp forms, then to 4-dimensional Galois representations, gives extensions with length ≥ b* [2:00].
- The two bounds clash, so y has infinite order [2:50].
From the converse to cube sums
For X³ + Y³ = ℓZ³ at p = 3 (additive reduction), Satgé's descent with its Sha terms kept gives s₃ ≤ 1 [3:15]. The converse makes all ranks agree, and the root number −1 forces L(E,1) = 0, so the rank is 1. A point of infinite order has Z ≠ 0, which gives the cube sum. For example, 17 = (18/7)³ + (−1/7)³ [3:47].
The procedure: bound the full Selmer corank by one, Sha included. Then let the root number pick the rank, and the converse does the rest [4:18].