Equivalence of Conjectures on Heavenly Elliptic Curves

Cam McLeman, Christopher Rasmussen · 2025

Read Equivalence of Conjectures on Heavenly Elliptic Curves by Cam McLeman, Christopher Rasmussen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-17474.txt.

Introduction

Cam McLeman and Christopher Rasmussen wrote a 2025 paper tying together two conjectures on heavenly elliptic curves. Heavenly varieties show tightly constrained Galois behavior at some prime. Two finiteness claims concern curves over quadratic number fields. One counts curve types, the other counts fields that admit such curves.

The pair presented early evidence at a 2024 workshop. Here they prove the two claims stand or fall together. That equivalence lets future work attack whichever form is easier. Finiteness would mean heavenly curves are rare gems, not a flood. The result sharpens a corner of arithmetic geometry.

Catalog record

CREATORCam McLeman, Christopher Rasmussen
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE20 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-17474

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Equivalence of Conjectures on Heavenly Elliptic Curves Cam McLeman, Christopher Rasmussen Abstract . Heavenly abelian varieties are abelian varieties defined over number fields that exhibit constrained ℓ \ell -adic Galois representations for some rational prime ℓ \ell . At the ICMS Workshop held in November 2024, we presented evidence for two finiteness conjectures around the distribution of heavenly elliptic curves over quadratic number fields, one in terms of isomorphism classes of curves, and a second in terms of quadratic fields which admit heavenly elliptic curves. We prove these two conjectures are equivalent. 1. Introduction Let us fix an algebraic closure \overline{\mathbb{Q}} of \mathbb{Q} , let \ell be a rational prime, and let K\subseteq\overline{\mathbb{Q}} be a number field. Recall that in the Ihara program to study Galois groups through the natural pro- \ell quotients of fundamental groups, one encounters the tower of fields K\subseteq\mbox{{山}}\subseteq\mbox{{天}}\subseteq\overline{\mathbb{Q}} , where 山 (“yama”) and 天 (“ten”) are defined as follows: \mbox{{天}}:=\mbox{{天}}(K,\ell) is the maximal pro- \ell extension of K(\mathbf{\upmu}_{\ell^{\infty}}\!) unramified away from \ell , and \mbox{{山}}:=\mbox{{山}}(K,\ell) is the fixed field of the outer pro- \ell Galois representation attached to \mathbb{P}^{1}_{K}\smallsetminus\{0,1,\infty\} . It is known that \mbox{{山}}\subseteq\mbox{{天}} for any K and \ell , and that \mbox{{山}}(\mathbb{Q},\ell)=\mbox{{天}}(\mathbb{Q},\ell) for odd regular primes \ell . (More precisely, Sharifi [ Sha02 ] proved this is equivalent to the conjecture of Deligne and Ihara, which was later proved by Brown [ Bro12 ] .) An abelian variety A/K is called heavenly at \ell if K(A[\ell^{\infty}])\subseteq\mbox{{天}} . It follows from Serre-Tate theory that such a variety necessarily has good reduction away from \ell . Thus, for fixed K , fixed dimension g , and fixed prime \ell , there can be only finitely many abelian varieties A/K of dimension g which are heavenly at \ell . For fixed K and g , let \mathcal{H}(K,g) be the set of pairs ([A]_{K},\ell) where [A]_{K} is the K -isomorphism class of a g -dimensional abelian variety A/K which is heavenly at \ell . In [ RT08 ] , the second author and Tamagawa conjectured that \mathcal{H}(K,g) is finite for any choice of K and g ; equivalently, for \ell\gg_{K,g}0 , there should be no abelian varieties A/K in dimension g which are heavenly at \ell . Many partial results towards the conjecture have been obtained, and a description of recent results is given in [ MR25 , §3] . The article [ MR25 ] studies heavenly elliptic curves over quadratic fields. Based on striking connections between the traces of Frobenius elements for curves with complex multiplication and for heavenly elliptic curves, we conjectured the following: excluding infinite families arising with \ell<7 , or from base change constructions, there are only finitely many \overline{\mathbb{Q}} -isomorphism classes of elliptic curves which are defined over a quadratic field K and which are heavenly at some prime \ell . We present the conjecture in two ways, one in terms of the isomorphism classes of heavenly elliptic curves, and the other in terms of the quadratic fields admitting new examples of heavenly elliptic curves. The purpose of the present article is to state both forms precisely and demonstrate their equivalence. For any number field K , integer g>0 , and prime \ell , let \mathcal{H}(K,g,\ell) denote the set of K -isomorphism classes of abelian varieties which are defined over K , have dimension g , and which are heavenly at \ell . For any B\geq 1 , set \mathcal{H}_{B}:=\bigcup_{\ell\geq B\vphantom{[}}\bigcup_{\begin{subarray}{c}K\\ [K:\mathbb{Q}]=2\end{subarray}}\left\{\bigl([A]_{K},\ell\bigr):[A]_{K}\in\mathcal{H}(K,1,\ell)\right\}. We also define \overline{\mathcal{H}}_{B} , which tracks analogous pairs, but only up to \overline{\mathbb{Q}} -isomorphism: \overline{\mathcal{H}}_{B}:=\left\{\bigl([A\times_{K}\overline{\mathbb{Q}}]_{\overline{\mathbb{Q}}},\ell\bigr):([A]_{K},\ell)\in\mathcal{H}_{B}\right\}. Note \overline{\mathcal{H}}_{B} is in obvious bijection with the set of equivalence classes of \mathcal{H}_{B} under the equivalence relation ([A]_{K},\ell)\sim([A^{\prime}]_{K^{\prime}},\ell^{\prime})\ \quad\text{precisely if}\quad\ell=\ell^{\prime}\ \text{and}\ A\times_{K}\overline{\mathbb{Q}}\cong_{\overline{\mathbb{Q}}}A^{\prime}\times_{K^{\prime}}\overline{\mathbb{Q}}. Note that if E/\mathbb{Q} is heavenly at some prime \ell_{0} , then the infinite family \{([E\times_{\mathbb{Q}}K]_{K},\ell_{0})\}_{K} , indexed by all quadratic fields K , is a subset of \mathcal{H}_{1} . To exclude the redundancy of such examples, we further define \displaystyle\overline{\mathcal{H}}(K,g) \displaystyle:=\{([A\times_{K}\overline{\mathbb{Q}}]_{\overline{\mathbb{Q}}},\ell):([A]_{K},\ell)\in\mathcal{H}(K,g)\}, \displaystyle\mathcal{H}^{\circ}_{B} \displaystyle:=\left\{([A]_{K},\ell)\in\mathcal{H}_{B}:([A\times_{K}\overline{\mathbb{Q}}]_{\smash{\overline{\mathbb{Q}}}},\ell)\not\in\overline{\mathcal{H}}(\mathbb{Q},1)\right\}, \displaystyle\overline{\mathcal{H}}^{\circ}_{B} \displaystyle:=\left\{([A]_{\smash{\overline{\mathbb{Q}}}},\ell)\in\overline{\mathcal{H}}_{B}:([A]_{\smash{\overline{\mathbb{Q}}}},\ell)\not\in\overline{\mathcal{H}}(\mathbb{Q},1)\right\}. However, \overline{\mathcal{H}}^{\circ}_{1} is not finite, either. By [ RT17 , Prop. 5.4] , there are infinitely many pairs ([E]_{\overline{\mathbb{Q}}},\ell_{0})\in\overline{\mathcal{H}}_{1}^{\circ} with \ell_{0}=2 ; we expect a similar result for \ell_{0}\in\{3,5\} (see [ MR25 , Rmk. 3.5] ). But for each \ell_{0}>5 , \overline{\mathcal{H}}_{1} contains only finitely many pairs with \ell=\ell_{0} [ MR25 , Cor. 3.4] , suggesting the following. Conjecture 1 . The set \overline{\mathcal{H}}_{7} is finite. Let us now describe the distribution of heavenly elliptic curves from a different perspective, by considering the following question: for which choices of a quadratic field K and a rational prime \ell do there exist elliptic curves E/K which are heavenly at \ell ? For some choices of K and \ell , it holds that every heavenly E/K is \overline{\mathbb{Q}} -isomorphic to some heavenly E_{0}/\mathbb{Q} . Analogous to the restriction placed on \overline{\mathcal{H}}^{\circ}_{B} , we refine the question to exclude such pairs. So for any B\geq 1 , define the set \mathcal{R}_{B} as follows: an ordered pair (\ell,K)\in\mathcal{R}_{B} precisely if (i) \ell\geq B is prime, (ii) there exists an elliptic curve E/K , heavenly at \ell , such that (iii) E\times_{K}\overline{\mathbb{Q}}\not\cong A\times_{\mathbb{Q}}\overline{\mathbb{Q}} for any A/\mathbb{Q} which is heavenly at \ell over \mathbb{Q} . Conjecture 2 . The set \mathcal{R}_{7} is finite. Remark 3 . Condition (iii) in the definition of \mathcal{R}_{B} is quite mild. The finite set \mathcal{H}(\mathbb{Q},1) was computed explicitly in [ RT08 ] , and spans a set J of only 16 rational j -invariants. So (iii) is satisfied automatically if j(E)\not\in J . Remark 4 . Theorem 8 below implies that Conjectures 1 and 2 are equivalent. Let us explain briefly why the equivalence is not immediate. The concern is the following: given a single pair ([\overline{E}]_{\overline{\mathbb{Q}}},\ell)\in\overline{\mathcal{H}}_{7} , there could exist an infinite sequence A_{i}/K_{i} of elliptic curves defined over quadratic fields, each heavenly at \ell and satisfying A_{i}\times_{K_{i}}\overline{\mathbb{Q}}\cong\overline{E} . In this case, \mathcal{R}_{7} possibly contains the infinite family \{(\ell,K_{i})\} . However, as demonstrated in the proof of Theorem 8 , this is possible only if \overline{E} can also be represented by a heavenly elliptic curve defined over \mathbb{Q} , and consequently the existence of such a pair in \overline{\mathcal{H}}_{7} does not automatically imply \mathcal{R}_{7} is infinite. 2. Equivalence of Conjectures For any prime \ell>2 , set \ell^{\star}:=(-1)^{(\ell-1)/2}\ell , so that \mathbb{Q}(\sqrt{\ell^{\star}})\subseteq\mathbb{Q}(\mathbf{\upmu}_{\ell}) . Within \GL_{2}(\mathbb{F}_{\ell}) , we let \mathcal{U} and \mathcal{B} denote the subgroup of unit upper triangular matrices, and the Borel subgroup of upper triangular matrices, respectively, so that \mathcal{U}\leq\mathcal{B}\leq\GL_{2}(\mathbb{F}_{\ell}) and \sharp\mathcal{U}=\ell . Let \chi\colon G_{\mathbb{Q}}\to\mathbb{F}_{\ell}^{\times} be the \ell -adic cyclotomic character modulo \ell . We recall the following fact, which is a specific case of [ RT17 , Lemma 3.3] . Lemma 5 . Let K be a number field and let A/K be an elliptic curve which is heavenly at \ell . With respect to some basis for A[\ell] , the representation \rho_{A}\colon G_{K}\to\GL_{2}(\mathbb{F}_{\ell}) has the form \rho_{A}=\begin{pmatrix}\chi^{i_{1}}&\star\\ 0&\chi^{i_{2}}\end{pmatrix},\qquad 0\leq i_{1},i_{2}<\ell-1. In particular, with respect to this basis, \rho_{A}(G_{\mathbb{Q}})\leq\mathcal{B} and \rho_{A}(G_{\mathbb{Q}(\mathbf{\upmu}_{\ell})})\leq\mathcal{U} . We define maps \alpha and \beta as follows: \displaystyle\alpha \displaystyle\colon\mathcal{H}^{\circ}_{B}\to\overline{\mathcal{H}}^{\circ}_{B}, \displaystyle([A]_{K},\ell) \displaystyle\mapsto([A]_{\overline{\mathbb{Q}}},\ell), \displaystyle\beta \displaystyle\colon\mathcal{H}_{B}^{\circ}\to\mathcal{R}_{B}, \displaystyle([A]_{K},\ell) \displaystyle\mapsto(\ell,K). It is evident that both maps are surjective. Further, it is straightforward to see that \beta has finite fibers: Suppose (\ell,K)\in\mathcal{R}_{B} . If ([A]_{K},\ell)\in\beta^{-1}((\ell,K)) , then necessarily A has good reduction away from \ell .

Read the complete text