Selmer ranks under quadratic twists satisfying the Heegner hypothesis

Alexandros Konstantinou · 2025

Read Selmer ranks under quadratic twists satisfying the Heegner hypothesis by Alexandros Konstantinou (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-2510-23291.txt.

Introduction

Alexandros Konstantinou, the sole author, studies how Selmer ranks change when an elliptic curve is twisted in a quadratic way. Selmer ranks measure hard arithmetic facts about a curve and bound the size of its rational point group. In this 2025 work, he uses only twists that meet the Heegner hypothesis, a named condition in number theory.

He starts from a rational curve with partial two-torsion and a common relaxed Selmer group shared across the twists. From there, he derives clear formulas that show the effect of each twist on the rank. The formulas use matrices whose entries lie in the field with two elements. As a check, he shows the formulas agree with what the parity conjecture predicts. That match supports both the formulas and the picture of how ranks should behave under twisting.

Catalog record

CREATORAlexandros Konstantinou
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE133 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-23291

Browse the full library · About the Tower of Babel

Full text

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

Selmer ranks under quadratic twists satisfying the Heegner hypothesis Alexandros Konstantinou Abstract . We investigate variations of Selmer ranks under quadratic twists satisfying the Heegner hypothesis. In particular, starting with an elliptic curve E / ℚ E/\mathbb{Q} with partial 2 2 -torsion and a common relaxed Selmer group, we derive explicit formulae describing the effect of twisting on Selmer ranks in terms of matrices over 𝔽 2 \mathbb{F}_{2} . As an application, we show that these formulae are compatible with predictions made by the parity conjecture. † † margin: 2010 Mathematics Subject Classification 11G05 (11G07, 11G40) 1. Introduction 1.1. Elliptic curves over quadratic extensions A classical problem in the arithmetic of elliptic curves is to understand how the Mordell–Weil rank changes over quadratic extensions. To study this, consider an elliptic curve E/\mathbb{Q} and a quadratic field K=\mathbb{Q}(\sqrt{d}) . The action of the nontrivial element \sigma\in\Gal(K/\mathbb{Q}) naturally splits the group E(K) into (\pm 1) -eigenspaces, which correspond to E(\mathbb{Q}) and the quadratic twist E^{d}(\mathbb{Q}) . This decomposition yields the formula: \rk E(K)=\rk E(\mathbb{Q})+\rk E^{d}(\mathbb{Q}). Consequently, the problem of studying ranks over quadratic extensions is entirely encoded in the ranks of the quadratic twists E^{d}/\mathbb{Q} . The study of quadratic twists lies at the heart of modern arithmetic statistics. Prominent among the conjectures in this area is that of Goldfeld, which predicts that within the quadratic twist family of an elliptic curve over \mathbb{Q} , half of the twists have analytic rank 0 and half have analytic rank 1 . In a recent trilogy of papers [ 24 , 25 , 26 ] , it has been established that the Birch and Swinnerton-Dyer conjecture implies Goldfeld’s conjecture, thereby linking two of the deepest conjectures in the arithmetic of elliptic curves. All investigations of this kind typically proceed via the more accessible Selmer groups, since the computation of the Mordell–Weil rank is obscured by the mysterious nature of the Tate–Shafarevich group. Early work includes [ 7 ] , which shows that for the congruent number curve y^{2}=x^{3}-x , the 2 -Selmer ranks of its quadratic twists follow a remarkably regular distribution. Subsequent results [ 8 , 20 ] extend this to elliptic curves with full rational 2 -torsion and no cyclic 4 -isogeny over \mathbb{Q} , while [ 12 ] establishes an analogous theorem over arbitrary number fields in the generic 2 -torsion case. In sharp contrast, when E(\mathbb{Q})[2]\cong\mathbb{Z}/2\mathbb{Z} , corresponding to the intermediate case of partial 2 -torsion, no distribution function exists for the 2 -Selmer ranks within quadratic twist families [ 11 ] . More recently, similar questions have been investigated for higher-dimensional abelian varieties [ 15 , 29 , 30 ] . Beyond statistical questions, quadratic extensions satisfying the Heegner hypothesis , meaning imaginary quadratic fields K/\mathbb{Q} in which every prime of bad reduction for E splits, offer a particularly rich setting for studying rational points. For such K , the theory of Heegner points provides a systematic supply of rational points on E , arising from the modular parametrisation, and among these one distinguishes a canonical point P_{K}\in E(K) . The work of Gross–Zagier and Kolyvagin show that if P_{K} is non-torsion, then \operatorname{rk}(E/K)=1 and \Sha(E/K) is finite. To descend these results to \mathbb{Q} , one appeals to analytic work on L -functions in quadratic twist families [ 1 , 17 , 28 ] , yielding the strongest known case of the Birch and Swinnerton–Dyer conjecture: if the analytic rank of E/\mathbb{Q} satisfies \mathrm{rk}_{\mathrm{an}}(E/\mathbb{Q})\leq 1 , then \mathrm{rk}_{\mathrm{an}}(E/\mathbb{Q})=\operatorname{rk}E(\mathbb{Q})\quad\text{and}\quad\Sha(E/\mathbb{Q})\ \text{is finite.} This leads us to a natural and delicate intersection: considering quadratic extensions that satisfy the Heegner hypothesis for elliptic curves with partial 2 -torsion. In this specific scenario, the behaviour of the full 2 -Selmer group is governed by the interaction between the \varphi – and \hat{\varphi} –Selmer groups. 1.2. Effect of twists satisfying the Heegner hypothesis In what follows, we write E/\mathbb{Q} for an elliptic curve with E(\mathbb{Q})[2]\cong\mathbb{Z}/2\mathbb{Z} . We may and do assume that after a change of variables E is given by an integral Weierstrass model y^{2}=x(x^{2}+ax+b) with a,b\in\mathbb{Z} . The point (0,0) generates E(\mathbb{Q})[2] and defines a 2 –isogeny \varphi:E\to E\textquoteright , with dual isogeny \hat{\varphi}:E\textquoteright\to E . Our main result focuses on twists by integers d arising from \mathcal{D}_{E/\mathbb{Q}} (see Definition 2.8 below). These all define imaginary quadratic fields satisfying the Heegner hypothesis for E/\mathbb{Q} . In particular, we give an explicit formula describing how the \varphi – and \hat{\varphi} –Selmer ranks change under such twists in terms of matrices over \mathbb{F}_{2} built from the Legendre symbols of the primes dividing d . The variation in Selmer ranks depends only on the sign of the discriminant \Delta_{E} , the coefficients a,b defining E , and on whether the “relaxed” Selmer group \Sel_{\mathcal{S}_{d}}^{\varphi}(E/\mathbb{Q}) (see § 3.1 ) contains a suitable integer – a condition that is straightforward to verify in practice (see Remark 5.3 ). Theorem 1.1 (= Thm. 5.2 , Cor. 5.7 ) . Let E/\mathbb{Q} be given by the equation y^{2}=x(x^{2}+ax+b) , and let d\in\mathcal{D}_{E/\mathbb{Q}} . Define \delta_{\varphi}(d)=\mathrm{rk}_{\varphi}(E^{-d}/\mathbb{Q})-\mathrm{rk}_{\varphi}(E/\mathbb{Q}),\qquad\delta_{\hat{\varphi}}(d)=\mathrm{rk}_{\hat{\varphi}}(E^{\prime-d}/\mathbb{Q})-\mathrm{rk}_{\hat{\varphi}}(E\textquoteright/\mathbb{Q}). The values of \delta_{\varphi}(d) and \delta_{\hat{\varphi}}(d) are given in the following table. \begin{array}[]{c|c|l}\hline\cr\delta_{\varphi}(d)&\delta_{\hat{\varphi}}(d)&\text{Conditions}\\ \hline\cr 0&\omega(d)&b<0\\[3.0pt] \omega(d)&0&\Delta_{E}<0,\ b>0\\[4.0pt] \omega(d)-1-\operatorname{rank}_{\mathbb{F}_{2}}(\mathbf{A}_{d})&\omega(d)-\operatorname{rank}_{\mathbb{F}_{2}}(\mathbf{A}_{d})&\Delta_{E}>0,\ b>0,\ a<0,\ \eta_{\varphi}(d)=1\\[4.0pt] \omega(d)-\operatorname{rank}_{\mathbb{F}_{2}}(\widehat{\mathbf{A}_{d}})&\omega(d)+1-\operatorname{rank}_{\mathbb{F}_{2}}(\widehat{\mathbf{A}_{d}})&\Delta_{E}>0,\ b>0,\ a<0,\ \eta_{\varphi}(d)=0\\[4.0pt] \omega(d)+1-\operatorname{rank}_{\mathbb{F}_{2}}(\widetilde{\mathbf{A}_{d}})&\omega(d)-\operatorname{rank}_{\mathbb{F}_{2}}(\widetilde{\mathbf{A}_{d}})&\Delta_{E}>0,\ b>0,\ a>0,\ \eta_{\varphi}(d)=1\\[4.0pt] \omega(d)-\operatorname{rank}_{\mathbb{F}_{2}}(\overline{\mathbf{A}_{d}})&\omega(d)-1-\operatorname{rank}_{\mathbb{F}_{2}}(\overline{\mathbf{A}_{d}})&\Delta_{E}>0,\ b>0,\ a>0,\ \eta_{\varphi}(d)=0\\[2.0pt] \hline\cr\end{array} Here, \omega(d) denotes the number of prime factors of d , and all matrices \mathbf{A}_{d} , \widetilde{\mathbf{A}_{d}} , \widehat{\mathbf{A}_{d}} and \overline{\mathbf{A}_{d}} are defined over \mathbb{F}_{2} in Notation 5.1 . 1.3. Structure of the argument Given a 2 –isogeny \varphi:E\to E^{\prime} , the Selmer group \Sel^{\varphi}(E/\mathbb{Q}) is defined as the subgroup of H^{1}(\mathbb{Q},E[\varphi]) defined by the certain local conditions at each place of \mathbb{Q} . In particular, for each place v , let \delta_{\varphi,v}\colon E^{\prime}(\mathbb{Q}_{v})/\varphi(E(\mathbb{Q}_{v}))\to H^{1}(\mathbb{Q}_{v},E[\varphi]) be the local Kummer map, and write \mathcal{L}_{\varphi,v}(E)\ :=\ \operatorname{im}\bigl(\delta_{\varphi,v}\bigr)\ \subseteq\ H^{1}(\mathbb{Q}_{v},E[\varphi]). Then the \varphi –Selmer group is the subgroup of H^{1}(\mathbb{Q},E[\varphi] cut out by these local subgroups: \Sel^{\varphi}(E/\mathbb{Q})=\left\{\,c\in H^{1}(\mathbb{Q},E[\varphi]):\operatorname{res}_{v}(c)\in\mathcal{L}_{\varphi,v}(E)\text{ for all }v\,\right\}. Thus, the effect of twisting on \Sel^{\varphi}(E/\mathbb{Q}) is determined entirely by how the local subgroups \mathcal{L}_{\varphi,v}(E) change under quadratic twisting. To study this, we consider certain relaxed versions of Selmer groups, obtained by omitting the conditions at the primes dividing d and at infinity: \Sel_{\mathcal{S}_{d}}^{\varphi}(E/\mathbb{Q})=\{\,c\in H^{1}(\mathbb{Q},E[\varphi]):\operatorname{res}_{v}(c)\in\mathcal{L}_{\varphi,v}(E)\text{ for all }v\nmid d\infty\,\}, and similarly for E^{-d} . Under the natural isomorphism E[\varphi]\cong E^{-d}[\varphi] , the groups \Sel^{\varphi}(E/\mathbb{Q}) , \Sel^{\varphi}(E^{-d}/\mathbb{Q}) , and their relaxed variants all embed in the same cohomology space H^{1}(\mathbb{Q},E[\varphi]) . As subgroups of this cohomology group, one has \Sel_{\mathcal{S}_{d}}^{\varphi}(E/\mathbb{Q})=\Sel_{\mathcal{S}_{d}}^{\varphi}(E^{-d}/\mathbb{Q}), which is entirely natural: since E/\mathbb{Q}_{p} and E^{-d}/\mathbb{Q}_{p} are isomorphic for all p\mid 2N_{E} , and have good reduction at every prime not dividing d and the real place, their local conditions coincide at all places defining the relaxed Selmer groups. Consequently, any discrepancy between \Sel^{\varphi}(E/\mathbb{Q}) and \Sel^{\varphi}(E^{-d}/\mathbb{Q}) can only arise from the primes dividing d and infinity. To this end, we take the common relaxed Selmer group \Sel_{\mathcal{S}_{d}}^{\varphi}(E/\mathbb{Q})=\Sel_{\mathcal{S}_{d}}^{\varphi}(E^{-d}/\mathbb{Q}) as a common starting point, and successively impose the remaining local conditions at the places v\mid d\infty . By tracking how these local conditions change under twisting (see Lemmas 2.13 and 2.16 ), we deduce the effect that the quadratic twist has on the Selmer group. To achieve this, we introduce the homomorphisms \varepsilon_{\infty},\ \lambda_{p},\ \gamma_{p}:\mathbb{Q}^{\times}/\mathbb{Q}^{\times 2}\to\mathbb{F}_{2}, which capture the effect of imposing the

Read the complete text