Experimental investigations on Lehmer's conjecture for elliptic curves

Basic Information

• Paper ID: 2510.08871
• Title: Experimental investigations on Lehmer's conjecture for elliptic curves
• Authors: Sven Cats, John Michael Clark, Charlotte Dombrowsky, Mar Curcó Iranzo, Krystal Maughan, Eli Orvis
• Classification: math.NT (Number Theory)
• Publication Date: October 10, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.08871

Abstract

This paper presents a method for computing non-torsion points of minimal positive canonical height on a given elliptic curve $E/\mathbb{Q}$ over all number fields of fixed degree. The authors employ this method to collect extensive data and utilize it to investigate conjectures related to Lehmer and Lang.

Research Background and Motivation

Core Problems

This research addresses the canonical height problem for non-torsion points on elliptic curves, particularly the verification of two important conjectures in number theory:

1. Lehmer's Conjecture (Elliptic Curve Version): For the constant $C_E := \inf\{\hat{h}(P) \cdot [K(P) : K]\}$, where the infimum ranges over all non-torsion points $P \in E(\overline{K}) - E(\overline{K})_{\text{tors}}$, this constant satisfies $C_E > 0$.
2. Lang's Conjecture: For the constant $C_{K,d} := \inf\{\hat{h}(P)/M_{E'}\}$, where the infimum ranges over all elliptic curves $E'/K$ and corresponding non-torsion points, this constant satisfies $C_{K,d} > 0$.

Research Significance

• These conjectures are central problems in algebraic geometry and number theory, involving arithmetic properties of points on elliptic curves
• While theoretical progress has been made on abelian varieties, experimental verification work remains scarce
• Understanding the distribution of minimal height points is crucial for comprehending the arithmetic structure of elliptic curves

Limitations of Existing Work

• Prior work by Elkies, Taylor, and others focused primarily on searches over specific families of elliptic curves
• Lack of large-scale, systematic experimental data to verify these conjectures
• Absence of a universal algorithmic framework for computing minimal height points on arbitrary elliptic curves

Core Contributions

1. Algorithmic Contribution: Proposes a systematic method for computing non-torsion points of minimal canonical height on elliptic curves over fixed-degree number field extensions
2. Dataset Construction: Establishes a large-scale database containing quadratic point data for 17,834 elliptic curves, with 728 cases provably finding minimal height points
3. Computational Optimization: Improves the height bounds of Cremona-Prickett-Siksek, making the algorithm more efficient
4. Experimental Verification: Provides the first large-scale experimental verification of Lehmer's and Lang's conjectures

Methodology Details

Problem Definition

Given an elliptic curve $E/K$ and a collection of number field extensions $\mathcal{F}$, compute: $C_{E,\mathcal{F}} := \inf_{F\in\mathcal{F}, P\in E(F)-E(F)_{\text{tors}}} \{\hat{h}(P) \cdot [F : K]\}$

Core Algorithmic Framework

Step One: Finiteness Reduction

Employing discriminant bounds to reduce the infinite search problem to a finite one:

Lemma 2.3: For $D \in \mathbb{R}_{\geq 0}$, $F \in \mathcal{F}$, $d = [F : K]$, define: $\Delta(D,E,F) := \exp(d\delta_K \log d + d(2d-2)B_E + (2d-2)D)$

If the discriminant $|\Delta_F| \geq \Delta(D,E,F)$, then for all non-torsion points $P \in E(F)$ satisfying $K(P) = F$, we have $\hat{h}(P) \geq D/d$.

Theorem 2.4: Let $D' \in \mathbb{R}_{\geq 0}$ such that $C_{E,\mathcal{F}} \leq D'$, and define: $\mathcal{F}' = \{F \in \mathcal{F} : |\Delta_F| \leq \Delta(D', E, F)\}$ Then $\mathcal{F}'$ is finite and $C_{E,\mathcal{F}'} = C_{E,\mathcal{F}}$.

Step Two: Height Bound Computation

Using a modified CPS bound to estimate the difference between Weil height and canonical height: $|h(P) - \hat{h}(P)| \leq B_E$

For a collection of number field extensions $\mathcal{F}$, take: $B_E = \max\{B_{E,F} : F \in \mathcal{F}\}$

where $B_{E,F} = \frac{1}{[F:K]} \sum_v M_v$, with the sum ranging over Archimedean places of $F$ and prime ideals where $E$ has bad reduction.

Step Three: Finite Search

For each $F \in \mathcal{F}'$, search for points $P \in E(F)$ satisfying: $h(P) \leq \frac{D'}{[F:K]} + B_E$

Technical Innovations

1. Discriminant Bound Method: First systematic application of discriminant bounds to elliptic curve height computations
2. Adaptive Search Strategy: Dynamically adjusts search ranges based on initial search results
3. Optimized Height Bounds: Improves existing bounds for applicability to number field extension families

Experimental Setup

Dataset

• Elliptic Curve Source: All elliptic curves in the Cremona database with conductor not exceeding 3,000
• Number Field Extensions: All quadratic fields $\{F/\mathbb{Q} : [F : \mathbb{Q}] \leq 2\}$
• Total Scale: 17,834 elliptic curves

Computational Resources

• Computation Time: Over 800 CPU hours
• Computing Environment: Red Hat Enterprise Linux 8.10 server
• Software Tools: Magma 21.2-2 and SageMath 10.6

Implementation Details

• Provable Cases: Complete search performed when $\Delta < 10^5$ and $B < 50$
• Approximate Cases: For other cases, search only quadratic fields with discriminant $|\Delta_K| \leq 1000$
• Verification Standard: 728 cases yielded provably minimal height points

Experimental Results

Main Findings

Minimal Height Records

The minimal height point discovered is:

• Elliptic Curve: $y^2 + xy + y = x^3 + x^2 - 2990x + 71147$ (Cremona label 1470l1)
• Coordinates: $(27, -119, 1)$
• Height: $0.0099641079999...$

Number Field Distribution Statistics

In all data, the distribution of defining fields for minimal height points:

• Rational field $\mathbb{Q}$: 2,199 curves
• Gaussian field $\mathbb{Q}(\sqrt{-4})$: 1,191 curves
• Eisenstein field $\mathbb{Q}(\sqrt{-3})$: 1,610 curves

Among the 728 provable cases:

• $\mathbb{Q}(\sqrt{-3})$: 20 curves
• $\mathbb{Q}(\sqrt{-4})$: 14 curves
• $\mathbb{Q}$: 11 curves

Algorithm Validity Verification

• Consistency Check: All provable cases show final results consistent with initial search results
• Coverage Range: 86 curves obtained complete verification, 542 curves had initial search failures
• Computational Limits: Curves with conductor 11 (11a1, 11a2) begin to exceed computational bounds

Conjecture Verification Results

• Preliminary verification of Lehmer's and Lang's conjectures
• No apparent relationship found between minimal height and elliptic curve conductor or discriminant
• All computational results support predictions that $C_E > 0$ and $C_{K,d} > 0$

Related Work

Theoretical Background

• Lehmer's Original Conjecture (1933): Concerning lower bounds on logarithmic heights of algebraic numbers
• Elliptic Curve Generalization: Theoretical work by Hindry-Silverman and others
• Abelian Variety Case: Research on related conjectures in more general settings

Computational Methods

• Silverman Bounds: Classical bounds on differences between Weil and canonical heights
• CPS Method: Improved bounds by Cremona-Prickett-Siksek
• Bruin Optimization: Further refinements for the rational field case

Experimental Precedents

• Elkies' Work: Searches for small height points on specific elliptic curves
• Taylor Computations: Unpublished results on quadratic fields
• This Paper's Contribution: First large-scale systematic experimental study

Conclusions and Discussion

Main Conclusions

1. Successfully developed a practical algorithm for computing minimal height points on elliptic curves
2. Established the largest-scale database of small height points on elliptic curves to date
3. Experimental data supports the correctness of Lehmer's and Lang's conjectures
4. Small height points tend to be defined over number fields with small discriminants

Limitations

1. Computational Complexity: Complete verification remains difficult for elliptic curves with large conductors
2. Number Field Restrictions: Currently considers only quadratic field extensions
3. Theoretical Analysis: Lacks deep theoretical explanations for observed phenomena
4. Statistical Patterns: No explicit relationship found between heights and curve invariants

Future Directions

1. Extension to Higher Degree Fields: Consider cubic, quartic, and higher-degree number field extensions
2. Algorithm Optimization: Improve initial searches using precomputed LMFDB data
3. Theoretical Analysis: Seek theoretical explanations for observed statistical patterns
4. Parallel Computing: Develop more efficient parallel algorithms for large-scale data processing

In-Depth Evaluation

Strengths

1. Methodological Innovation: First systematic framework for computing minimal height points on elliptic curves
2. Experimental Scale: Data from 17,834 curves is unprecedented in this field
3. Theoretical Foundation: Algorithm built on rigorous number-theoretic foundations
4. Openness: Code and data are fully open-source, facilitating subsequent research

Weaknesses

1. Limited Theoretical Insights: Contributions are primarily computational; theoretical novelty is relatively limited
2. Insufficient Statistical Analysis: Statistical analysis of experimental data could be more thorough
3. Generalization Issues: Method currently applicable only to quadratic field cases
4. Computational Bottlenecks: Complex cases still face computational feasibility challenges

Impact

1. Field Contribution: Provides important experimental tools and data resources for arithmetic geometry of elliptic curves
2. Practical Value: Algorithm directly applicable to related number-theoretic research
3. Reproducibility: Complete open-source implementation ensures result verifiability
4. Inspirational Significance: Provides methodological reference for similar computational number theory problems

Application Scenarios

1. Number Theory Research: Experimental verification of elliptic curve height theory
2. Algorithm Development: Reference for algorithm design in related computational problems
3. Educational Applications: Case studies for computational number theory and elliptic curve theory
4. Subsequent Research: Data foundation for deeper theoretical investigations

References

The paper cites 12 important references, including:

• Lehmer (1933): Original conjecture formulation
• Silverman (1984, 1990): Foundational work on height theory
• Cremona-Prickett-Siksek (2006): Improvements to height bounds
• Hindry-Silverman (2000): Systematic treatment of Diophantine geometry
• LMFDB: Modern number theory database resources

Summary: This paper makes important contributions to computational verification of height theory on elliptic curves. While theoretical innovations are limited, its systematic methodology and large-scale experimental data provide valuable resources for the field's development. The open-source implementation of the algorithm and detailed experimental reports ensure high credibility and practical utility of the results.