The moduli spaces of Jacobians isomorphic to a product of two elliptic curves (Q5963285)

Let \(K\) be an algebraically closed field, let \(C/K\) denote a curve of genus \(2\), and let \(E_k\) for \(k=1,2\) denote elliptic curves over \(K\). If the Jacobian variety \(J_C\) is isomorphic to \(E_1\times E_2\), then there is an isogeny \(: E_1 \to E_2\) with degree \(d\), and the curve \(C\) is said to \textit{have type \(d\)}. Let \(M_2\) be the moduli space of curves of genus \(2\) over \(K\), and let \(T(d)\) denote the subset of \(M_2\) consisting of curves of type \(d\). Given an integral binary quadratic form \(q\) of \textit{type \(d\)}, the author of the paper under review defines a \textit{generalized Humbert variety} \(H(q)\) to be the set of curves in \(M_2\) whose \textit{refined Humbert invariant} primitively represents \(q\), and proves that \(T(d)\) is a finite union of \(H(q)\) where \(q\) varies over certain \(\mathrm{GL}_2(\mathbb Z)\)-equivalent classes of forms of type \(d\). Moreover the author shows that if \(\mathrm{char}(K)\nmid d\), each \(H(q)\) is birationally isomorphic to either the modular curve \(X_0(d)^+\) or the quotient of the modular curve by a suitable Atkin-Lehner involution. The author asks for which \(d\) the subset \(T(d)\) is empty, and offers the result: \(T(d)\) is empty if and only if \(d=1\) or if \(d\) is an even idoneal number which is not divisible by \(8\). It was conjectured by Euler and Gauss that the largest idoneal number is \(1848\), and it remains to be unsolved while Chowla (unconditionally) proved that there are finitely many idoneal numbers, and Weinberger showed that the conjecture is true under the Generalized Riemann Hypothesis. Under this conjecture, there are \(21\) idoneal numbers not divisible \(8\), and hence, the author offers a conjecturally complete list of \(d\) for which \(T(d)\) is empty.