Visibility of Mordell-Weil groups (Q2469678)

Let \(A,B,C/K\) be abelian varieties over a number field \(K\) such that there exists an exact sequence \[ 0 \to C \to B \to A\to 0. \] The quotient map \(B \to A\) is optimal, in the sense that the kernel is connected. Mazur introduced the visible subgroup \(\text{Vis}_B H^1(K,C)\) of \(H^1(C,K)\) relative to the embedding \(C\hookrightarrow B\) and defined this to be the co-kernel of \(B(K)\to A(K)\), which is a natural subgroup of \(H^1(C,K)\) by the connecting homomorphism \(\delta: A(K) \to H^1(C,K)\) in Galois cohomology. This subgroup has been studied by many people, where the focus has been on the intersection of \(\text{Vis}_B H^1(K,C)\) with the Tate-Shafarevich group \(\text Ш(C/K)\). For an optimal covering \(B\to A\) the author introduces the visible quotient \(\text{Vis}^B(A(K))\) of \(A(K)\) relative to be the co-kernel of \(B(K) \to A(K)\). In particular, \(\text{Vis}^B(A(K))\) is isomorphic to \(\text{Vis}_B H^1(K,C)\). The author expects that abelian varieties with non-trivial Mordell-Weil groups give rise to many visible Tate-Shafarevich groups. He poses the following conjecture: Let \(A/K\) be an abelian variety over a number field \(K\). Then for each \(m\) there exists an exact sequence \(0\to C\to B\to A\to 0\) such that 1. The image of \(B(K)\) in \(A(K)\) is contained in \(mA(K)\), so \(A(K)/mA(K)\) is a quotient of \(\text{Vis}^B(A(K))\). 2. If \(K=\mathbb Q\) and \(A\) is modular then \(B\) is modular. (An abelian variety \(A\) is called modular if it is a quotient of \(J_1(N)= \text{Jac}(X_1(N))\).) 3. The rank of \(C\) is zero. 4. We have \(\text{Coker} (B(K)\to A(K)) \subset \text Ш(C/K)\subset H^1(C,K) \) via the connecting homomorphism. The following evidence for the conjecture is given: Let \(A=E\) be the rank 1 elliptic curve \(y^2-y=x^3-x\) of conductor 37. The author proves that the conjecture holds for all primes \(m=p<25000\) with \(p\neq 2, 37\).