Transition formulae for ranks of abelian varieties (Q261652): Difference between revisions

Let \(E\) be an elliptic curve defined over a number field \(k\) such that it has ordinary good reduction at all places lying above an odd prime \(p\). Assuming that \(E\) has no complex multiplication with some additional conditions, \textit{J. Coates} and \textit{S. Howson} [J. Math. Soc. Japan 53, No. 1, 175--235 (2001; Zbl 1046.11079)] proved that \( \mathrm{rank}\, E(K_n) \leq c\, p^{3n}\) for some constant \(c\) depending on \(E\) and \(p\) where \(K_n=K(E[p^n])\). As a generalization of this result, the authors of the paper under review offer the following conjecture: Let \(K_\infty := \bigcup_{n\geq 1} K_n\) be an arbitrary Lie extension, and \(A\) be an abelian variety of dimension \(g\) over \(k\). If \(K_\infty\) contains the cyclotomic \(\mathbb Z_p\)-extension \(k_\infty\), then \(\mathrm{rank}\, A(K_n) = O(p^{n(d-1)})\) where \(d=\dim \mathrm{Gal} (K_\infty/k)\). The authors prove that if, in addition to the hypothesis of the conjecture, \(\dim \mathrm{Gal} (K_\infty/k)\leq 3\) and \(\mathrm{Gal}(K_\infty/k_\infty)\) is a pro \(p\)-group with no \(p\)-torsion, then for all sufficiently large \(n\), \[ \mathrm{rank}\, A(K_n) \leq C\, p^{n(d-1)} + 4g^2, \] where \(C\) depends on \(A\) and \(\mathrm{Gal} (K_\infty/k)\). Under the condition \(\dim \mathrm{Gal} (K_\infty/k)\leq 3\), the authors use the classification result of \textit{J. González-Sánchez} and \textit{B. Klopsch} [J. Group Theory 12, No. 5, 711--734 (2009; Zbl 1183.20030)] on the analytic pro \(p\)-groups, and under the other condition, they utilize the consequence that the Iwasawa algebra \(\Lambda(\mathrm{Gal} (K_\infty/k))\) contains no zero divisor. In fact, the center piece of their work is a transition formula for a cyclotomic \(\lambda\)-invariant on a certain twist of the \(\Lambda(\mathrm{Gal} (K_\infty/k))\)-torsion module \(\mathrm{Hom}_{\mathrm{cont}}(\mathrm{Sel}_{K_\infty}(A), \mathbb Q_p /\mathbb Z_p)\), and this allows them to find recurrence relations for \(\lambda\)-invariants, and as a consequence, the upper bound on \(\mathrm{rank}\, A(K_n)\).