Torsion points of abelian varieties with values in infinite extensions over a \(p\)-adic field (Q847654)

Fix a rational prime \(p\). Let \(K\) be a finite extension of the \(p\)-adic number field \(\mathbb Q_p\). Let \(A\) be an abelian variety defined over \(K\). Let \(L\) be an extension of \(K\). As usual, we write \(A(L)\) to denote the group of \(L\)-rational points on \(A\), with \(A(L)[p^{\infty}] = \bigcup_{n}A(L)[p^n]\) its \(p\)-primary torsion subgroup. If \(L\) is a finite extension of \(K\), then a classical theorem due to \textit{A. Mattuck} [Ann. Math. (2) 62, 92--119 (1955; Zbl 0066.02802)] shows that the torsion subgroup of \(A(L)\) is finite, and hence that \(A(L)[p^{\infty}]\) is finite. This interesting and nicely presented article explores the analogous problem in the setting where \(L\) is an infinite extension of \(K\). In particular, the author obtains the following main results, generalizing earlier works of \textit{H. Imai} [Proc. Japan Acad. 51, 12--16 (1975; Zbl 0323.14010)] and \textit{N. M. Katz, S. Lang} and \textit{K. A. Ribet}, Enseign. Math., II. Sér. 27, 285--314, 315--319 (1981; Zbl 0495.14011)] in this direction. To begin, let us say that an algebraic extension \(M\) of a field \(F\) is {\textit{prime to \(p\)}} if it is given by a union of finite extensions of \(F\), each having degree prime to \(p\). We then say that an algebraic extension \(M\) of \(F\) is {\textit{potentially prime-to-\(p\)}} if it is a prime-to-\(p\) extension of a finite extension of \(F\). Fixing an algebraic extension \(L\) of the \(p\)-adic field \(K\), let \(L(\mu_{p^{\infty}}) = \bigcup_{n\geq0} L(\mu_{p^n})\) denote the extension obtained by adjoining to \(L\) all \(p\)-power roots of unity. Let \(k_K\) denote the residue field of \(K\), with \(k_L\) that of \(L\), and \(k_{L_{\infty}}\) that of \(L(\mu_{p^{\infty}})\). If \(A\) over \(K\) has potentially good reduction, then the first main result in Theorem 1.1 asserts that following statements are true. (1) If \(k_{L_{\infty}}\) is potentially prime-to-\(p\) over \(k_K\), then \(A(L)[p^{\infty}]\) is finite. (2) If \(L\) contains \(K(\mu_{p^{\infty}})\) and the extension \(K(A[p])\) obtained by adjoining the coordinates of all \(p\)-torsion points of \(A\) to \(K\), then \(k_L\) is potentially prime-to-\(p\) over \(k_K\) if and only if \(A(L)[p^{\infty}]\) is finite. (3) If \(L(\mu_{p^{\infty}})\) is Galois over \(K\) with finite residue field \(k_{L_{\infty}}\), then the torsion subgroup of \(A(L)\) is finite. The main technical ingredient used to show this result is the matrix characterization of the Galois representation associated to the \(p\)-adic Tate module of \(A\) given in Proposition 2.6, following the analogous deduction of Theorem 1.1 in [\textit{B. Conrad}, in: Modular forms and Fermat's last theorem. Papers from a conference, Boston, MA, USA, August 9--18, 1995. New York, NY: Springer. 373--420 (1997; Zbl 0927.11037)]. A global analogue of this first main result is also presented in Corollary 2.2-2.3, as well as an interesting discussion about the setting of elliptic curves with complex multiplication in \S 2.3. To state the second main result, let \(B\) denote any semiabelian variety defined over \(K\). Let us then write \(K_{B, p} = K(B[p^{\infty}])\) to denote the extension obtained by adjoining to \(K\) all coordinates of \(p\)-power torsion points of \(B\). It is natural to ask whether or not the torsion subgroup of \(A(K_{B,p})\) is finite. If \(A=B\) is an elliptic curve without complex multiplication, then the analogous torsion subgroup of \(A(K_{A,p})\) for \(K\) a global number field is known to be finite by a relatively simple argument in Galois cohomology, as explained in [\textit{J. Coates}, in: Arithmetic theory of elliptic curves. Lect. Notes Math. 1716, 1--50 (1999; Zbl 1029.11016)]. In the more general setting described above with \(A=E_1\) and \(B=E_2\) a pair of elliptic curves, as considered by the author in Theorem 1.2, the behaviour turns out to be more involved. In particular, the behaviour depends on the reduction types of \(E_1\) and \(E_2\). For instance, Proposition 3.4 asserts that \(E_1(K_{E_2, p})[p^{\infty}]\) is infinite if both \(E_1\) and \(E_2\) have good ordinary reduction. On the other hand, this is not the case if either \(E_1\) or \(E_2\) has multiplicative or supersingular reduction, as shown by arguments throughout the final two sections of the article.