Torsion of abelian varieties, Weil classes and cyclotomic extensions (Q4237709)

Let \(K\) be a finitely generated extension of \(\mathbb{Q}\), \(K(a) \subset \mathbb{C}\) its algebraic closure, \(K(c)\) the subfield of \(K(a)\) generated by \(K\) and all the roots of unity in \(K(a)\) and let \(K(ab)\) be the maximal abelian extension of \(K\). Let \(X\) be an abelian variety over \(K\). The pair \((X,K)\) satisfy the hypothesis (H) if it verifies one of the following conditions: 1) There exists a discrete valuation \(v\) of \(K\) such that \(X\) has potential purely multiplicative reduction at \(v\). 2) \(K\) does not contain a CM-field. 3) The Hodge group of \(X\) is semisimple. 4) The centre, \(F\), of \(\text{End}^0(X)= \text{End}(X) \otimes_{\mathbb{Z}} \mathbb{Q}\) is a CM-field and the couple \((X,F)\) is of Weil type. The main result proved in the paper is: If the pair \((X,K)\) satisfies the hypothesis (H) and \(L\) is a Galois extension of \(K\) contained in \(K(a)\) whose intersection with \(K(c)\) is finite over \(K\), then the group \(\text{TORS} (X(L))\) of torsion points of \(X(L)\) is finite. This is a refinement of a result of Bogomolov (Séminaire Delange-Pisot-Poitou, mai 1982, Paris).