On the local-global principle for twists of abelian varieties
Nirvana Coppola, Lorenzo La Porta, Matteo Longo
TL;DR
The paper develops a cohomological framework for local-global questions about twists of abelian varieties over number fields, introducing Selmer and Tate--Shafarevich sets $ ext{Sel}_m(G_K, D_L^ imes)$ and $ ext{X}_m(G_K, D_L^ imes)$ to measure obstructions. It proves finiteness of $ ext{X}_m(G_K, D_L^ imes)$ under mild assumptions leveraging Faltings-type results, and provides explicit criteria ensuring $ ext{X}_m(G_K, D_L^ imes)=1$, notably in cases where $A$ is geometrically simple with commutative endomorphism algebra or when $ ext{μ}_m subseteq D_K^ imes$ together with coprimality conditions involving $d=rac{2 ext{dim}(A)}{[Z:bQ]}$. The work yields broad positive results for odd $m eq 1$, including CM scenarios and rational roots of unity, and applies to concrete families such as CM elliptic curves and Jacobians, clarifying when local twists glue to global twists. Overall, the results offer a robust cohomological toolkit for determining when a locally prescribed twisting behavior arises globally, with implications for the arithmetic of abelian varieties and their endomorphism structures.
Abstract
This paper investigates the existence of a local-global principle for certain twists of abelian varieties defined over number fields. Our main focus is to determine when, for $m$ a positive integer, locally $m$-atic twists of an abelian variety $A$ over a number field $K$ are globally $m$-atic. We define and study a "Tate-Shafarevich cohomology set" that governs the obstruction to the local-global principle for $m$-atic twists. We prove that, under some mild assumptions, this set is finite, and give criteria for it to be trivial, i.e. for the local-global principle to be satisfied.
