Recursos
Resumen
Take a subvariety X of an abelian variety and a torsion coset T. You can decide whether a piece of X ∩ T is too big, meaning atypical, and you can follow how the paper proves that such pieces sit inside only finitely many maximal ones.
The count you apply
- A special subvariety is a torsion coset B + τ. The special closure S = ⟨X⟩sp is the smallest one that contains X [0:34].
- A component Y of X ∩ T, with T special inside S, is atypical when dim Y > dim X + dim T − dim S [0:51].
- Worked case: a surface, S of dimension 5 and T of dimension 3 give an expected dimension of 0, so a curve is atypical [1:08]. Put the same data inside an S of dimension 4 and the expected dimension is 1, so the curve is typical. Always count inside S, never inside the ambient A [1:42].
The theorem and its reduction
- Over a number field, X has finitely many maximal atypical subvarieties in S. The theorem needs no hypothesis on A, and its bound is not effective [1:59].
- The first step is a non-density theorem: X ∩ A^[dim X+1] is not Zariski dense when X lies in no proper torsion coset. The Barroero–Dill reduction then turns this into finiteness [2:32].
How non-density is proved
- Suppose there is a generic sequence of intersection points. Their heights grow at different rates, so the proof splits A into height scales: at most one bounded level and several unbounded ones, each normalized by its own scale [3:02].
- All scales unbounded: Dill's generalized Vojta–Rémond inequality. Its constant does not depend on the approximation, so you fix η first and only then let the heights grow [4:09].
- A bounded scale: take auxiliary products of bounded degree. Arithmetic positivity comes from comparing two measures. Small sections with positive index follow, and Faltings' product theorem yields a product that contradicts minimality [4:43].