On certain root number 1 cases of the cube sum problem

DOI10.1016/J.JPAA.2025.108145MaRDI QIDQ6873528zbMATH OpenFDO

Publication date 16 January 2026

Published in Journal of Pure and Applied Algebra (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1016/J.JPAA.2025.108145

primes represented by binary forms Selmer groups Sylvester's cube sum problem

Cubic and quartic Diophantine equations (11D25) Elliptic curves over global fields (11G05) Class numbers, class groups, discriminants (11R29)

Review text

A classical problem originating from Sylvester asks for which integers are sums of two rational cubes; there are some precise conjectures as well as many known results. It is well known that \(n>2\) is a sum of two rational cubes if and only if the elliptic curve \(E_{-432n^2} : y^2 = x^3 - 432n^2\) over \(\mathbb{Q}\) has positive rank. An important invariant associated to this curve is its global root number \(w(n)\); it is \((-1)^{d}\) where \(d\) is the order of the zero of the corresponding \(L\)-function at \(s=1\). Thus, the root number being \(1\) leaves both possibilities open -- \(n\) may or may not be a sum of two rational cubes. The \(2\)-Selmer group and its relation with the \(2\)-rank of the class group of the cubic field corresponding to \(x^3 - 432n^2\) is one of the tools available to compute the elusive algebraic (Mordell-Weil) rank. In addition to this Selmer group, the \(3\)-Selmer group given by the natural rational degree \(3\) isogeny \(E_{-432n^2} \rightarrow E_{16n^2}\), proves also helpful.\N\NThe authors use the relation between this \(3\)-Selmer group and the \(3\)-rank of a suitable cubic field to formulate their results. As they point out in their remark 0.2, even though the \(3\)-torsion points of \(E_{-432n^2}\) are contained in \(\mathbb{Q}(\zeta_3, \sqrt[3]{n})\), the condition for being expressible as a sum of two rational cubes is not captured by subfields of the above field, and they need to choose their cubic fields carefully. One of the main results proved in this paper gives necessary conditions for a cube-free \(n>2\) to be a sum of two rational cubes under the condition that the global root number is \(1\). More precisely, they prove:\N\NLet \(\ell\) be a prime. Then we have: (i) If \(\ell \equiv 7\) modulo \(9\), such that \(3 \ell\) is a sum of two rational cubes, then the \(3\)-rank of \(\mathbb{Q}(\sqrt[3]{12 \ell})\) is \(2\). Further, there is a positive proportion of primes \(\ell \equiv 7\) modulo \(9\), such that \(3 \ell\) is not a sum of two rational cubes. For primes \(\ell \equiv 4\) modulo \(9\), such that \(3 \ell^2\) is a sum of two rational cubes, then the \(3\)-rank of \(\mathbb{Q}(\sqrt[3]{18 \ell})\) is \(2\). Further, there is a positive proportion of primes \(\ell \equiv 4\) modulo \(9\), such that \(3 \ell^2\) is not a sum of two rational cubes.\N\NIn fact, the authors have the stronger conclusion as a corollary:\N\NIf \(\ell \equiv 7\) modulo \(9\) is a prime such that \(3 \ell\) is a sum of two rational cubes, then the class group of \(\mathbb{Q}(\sqrt[3]{12 \ell})\) contains \(\mathbb{Z}/6 \mathbb{Z} \times \mathbb{Z}/3 \mathbb{Z}\)). An analogous statement holds for primes \(\ell \equiv 4\) modulo \(9\) for which \(3 \ell^2\) is a sum of two rational cubes.\N\NThe above stronger consequence is derived from the first result above and the one described below:\N\NSuppose \(n > 2\) is a cube-free integer which is a sum of two rational cubes. Further, assume the following: (i) the cube-free part of \(4n\) is different from \(\pm{1}\) modulo \(9\); and (ii) the global root number of \(E_{-432n^2}\) is \(1\). Then the class number of the cubic field \(\mathbb{Q}(\sqrt[3]{4n})\) is even. In particular, any prime \(\ell \equiv 1\) modulo \(9\) such that the \(2\)-rank of \(\mathbb{Q}(\sqrt[3]{4\ell})\) is zero, fails to be a sum of two rational cubes. A similar statement holds for \(\ell^2\) if the \(2\)-rank of \(\mathbb{Q}(\sqrt[3]{2\ell})\) is zero.\N\NSimilar to the first main result mentioned above, the authors also prove a version for primes \(\ell \equiv 1\) modulo \(9\). The precise statement is:\N\NLet \(\ell \equiv 1\) modulo \(9\) be a prime such that either \(2 \ell\) or \(2 \ell^2\) is a sum of two rational cubes. Then the \(3\)-rank of \(\mathbb{Q}(\sqrt[3]{2\ell})\) is \(2\). Further, there is a positive proportion of primes \(\ell \equiv 1\) modulo \(9\) such that neither \(2 \ell\) nor \(2 \ell^2\) is a sum of two rational cubes.\N\NThe positive density results mentioned in the first and the last results above essentially follow classical results. For instance, one relates the cubic residue symbol of \(2\) or \(3\) modulo a prime \(\ell \equiv 1\) modulo \(3\) with the representation of this prime by a certain integral binary quadratic form like \(x^2+27y^2\). One needs a refined version where the representation by a binary forms is studied for primes in a certain arithmetic progression defined by a residue modulo \(9\). We mention in passing that using binary cubic forms, it has been shown (see [\textit{S. Jha} et al., Proc. Am. Math. Soc. 153, No. 11, 4657--4668 (2025; Zbl 08105042)]) that there are infinitely many primes \(\ell \equiv 1\) modulo \(9\) which are sums of two rational cubes.

Cites work

This page was built for publication: On certain root number 1 cases of the cube sum problem

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6873528)