Cohomology and torsion cycles over the maximal cyclotomic extension (Q2001381)

Let \(k\) be a number field, and \(K\) the field obtained by adjoining all roots of unity to \(k\) in a fixed algebraic closure \(\overline{k}\). Let \(A\) be an abelian variety defined over \(k\). Ribet proved in the appendix of [\textit{N. M. Katz} and \textit{S. Lang}, Enseign. Math. (2) 27, 285--314, 315--319 (1981; Zbl 0495.14011)] that the torsion subgroup of \(A(K)\) is finite. In the present paper, the authors offer two kinds of generalizations of Ribet's theorem. Let \(X\) be a smooth proper geometrically connected variety definined over \(k\). The first one shows that for all odd \(i\) and all \(j\), the groups \(H_{\text{ét}}^i(X\times_k\overline{k},\mathbf{Q}/\mathbf{Z}(j))^G\) are finite. Ribet's theorem is the special case \(i=j=1\) of this result, since there is a Galois-equivariant isomorphism between \(H_{\text{ét}}^1(A^*\times_k\overline{k},\mathbf{Q}/\mathbf{Z}(1))\) and the torsion subgroup of \(H_{\text{ét}}^1(A^*\times_k\overline{k}, \mathbf{G}_m)\), which can be identified with that of \(A(\overline{K})\). The second one generalizes Ribet's theorem from a motivic viewpoint. They propose a conjecture which asserts that for all \(i>0\) the Chow groups \(\mathrm{CH}^i(X_K)\) have finite torsion subgroup. As an evidence, they prove the following positive result: Assume that \(H_{\mathrm{Zar}}^2(X,\mathcal{O}_X)\) vanishes and that \(H_{\text{ét}}^3(X\times_k\overline{k},\mathbf{Z}_{\ell})\) are torsion free for all \(\ell\). Then the torsion subgroup of \(\mathrm{CH}^2(X_K)\) is finite. The assumptions of the result are satisfied, for instance, by smooth complete intersections of dimension \(>2\) in the projective space.