Fundamental domains for quaternionic S-arithmetic groups over totally real fields

Basic Information

• Paper ID: 2510.09356
• Title: Fundamental domains for quaternionic S-arithmetic groups over totally real fields
• Authors: Marc Masdeu, Eloi Torrents
• Classification: math.NT (Number Theory)
• Publication Date: October 13, 2025
• Paper Link: https://arxiv.org/abs/2510.09356

Abstract

Let $B$ be a totally definite quaternion algebra over a totally real field $F$, $\mathfrak{p}$ a prime ideal of $F$, and $\Gamma$ the group of elements of reduced norm one in an Eichler $\mathcal{O}_F[1/\mathfrak{p}]$-order $R$ contained in $B$. This paper presents an algorithm for computing fundamental domains for the action of $\Gamma$ on the Bruhat-Tits tree of $\operatorname{GL}_2(F_\mathfrak{p})$. Using this algorithm, the authors tabulate all Shimura curves of genus at most 3 over totally real fields that can be further uniformized by some prime $\mathfrak{p}$.

Research Background and Motivation

1. Core Problem: The fundamental problem addressed in this paper is computing fundamental domains for the action of quaternionic S-arithmetic groups on the Bruhat-Tits tree. This is an important computational problem in algebraic number theory and the theory of automorphic forms.
2. Significance:
   • The structure of bad special fibers of Shimura curves with bad reduction is encoded by these fundamental domains
   • Computing fundamental domains is crucial for understanding p-adic uniformization of Shimura curves
   • These computational results can be used to determine all Shimura curves within a specific genus range
3. Limitations of Existing Methods:
   • Previous algorithms were primarily designed for the case over the rational numbers $\mathbb{Q}$
   • Effective computational methods for the general totally real field case were lacking
   • Existing methods are inefficient when handling high-genus cases
4. Research Motivation:
   • Generalize the Franc-Masdeu algorithm to the case of totally real fields
   • Provide systematic methods to compute and classify low-genus Shimura curves
   • Supply computational tools for p-adic theory of Shimura curves

Core Contributions

1. Algorithmic Innovation: Generalize the Franc-Masdeu algorithm to quaternion algebras over totally real fields, providing a complete algorithm for computing fundamental domains
2. Theoretical Development:
   • Provide effective methods to determine whether two vertices or edges on the Bruhat-Tits tree are equivalent under group action
   • Give standardized methods for representing vertices and edges of the Bruhat-Tits tree using matrices
3. Computational Results: Systematically compute and tabulate all p-adic uniformizable Shimura curves of genus at most 3 over totally real fields of degree 2-7
4. Software Implementation: Provide a SageMath implementation, enabling the algorithm to be used and verified by other researchers

Detailed Methodology

Task Definition

Given a totally definite quaternion algebra $B$ over a totally real field $F$, a prime ideal $\mathfrak{p}$, and an Eichler order $R$, compute the fundamental domain for the action of $\Gamma = \iota(R[1/\mathfrak{p}]^{\times 1}) \subseteq \text{SL}_2(F_\mathfrak{p})$ on the Bruhat-Tits tree $T_\mathfrak{p}$.

Core Algorithm Architecture

1. Representation of the Bruhat-Tits Tree

• Vertex Representation: Homotopy classes $[Λ]$, where $Λ \subseteq F_\mathfrak{p}^2$ is a lattice
• Normalized Representation: Using $2 \times 2$ matrices over $\mathcal{O}_F$ to represent vertices and edges

Lemma 2.2: Vertices can be represented as matrices of the form $\begin{pmatrix} \pi^m & 0 \\ r & \pi^n \end{pmatrix}$ or $\begin{pmatrix} 0 & \pi^m \\ \pi^n & r \end{pmatrix}$, where $r \in S_n$.

2. Equivalence Determination Algorithm

The core problem is determining whether two vertices/edges $u, v$ are $Γ$-equivalent, i.e., whether there exists $γ \in Γ$ such that $\iota(γ)u = v$.

Lemma 3.2: Let $2m = a + b$, where $a = \text{val}_\pi(\det u)$ and $b = \text{val}_\pi(\det v)$. Then $\text{Hom}_Γ(u,v) = Γ \cap π^{-m}v^*Λ_0u$

Algorithm Core Steps:

1. Construct the lattice $Λ_{u,v} = \iota^{-1}(π^h v^* Λ_0 u) \cap R + \mathfrak{p}^{d⌈m/d⌉+1}R$
2. Search for elements with reduced norm $δ^{2⌈m/d⌉}$
3. Use the LLL algorithm to solve the shortest vector problem

3. Fundamental Domain Computation Algorithm

Algorithm 2 main workflow:

1. Start from a base point $v_0$
2. For each vertex to be processed, check all its outgoing edges
3. Use equivalence determination to check whether new edges/vertices are already in the fundamental domain
4. Record boundary relations and compute genus
5. Repeat until all vertices are processed

Technical Innovations

1. Standardized Matrix Representation: Through Lemma 2.2, converting p-adic matrix computations to computations over the integer ring, greatly simplifying implementation complexity
2. Application of Lattice Methods: Converting the equivalence determination problem into finding elements of specific norm in a specific lattice, efficiently solved using the LLL algorithm
3. Precomputation of Boundary Data: By precomputing reduction information of the boundary layer, reducing the reduction complexity of arbitrary vertices to $O(\text{dist}(v))$
4. Genus Control: The algorithm can set maximum genus limits, avoiding computation of overly complex cases

Experimental Setup

Data Range

• Number Field Degree: Totally real fields of degree 2-7
• Genus Range: 0-3
• Prime Conditions: Unramified primes with inertia degree 1

Computational Environment

• Hardware: Apple M4 processor, 16GB RAM
• Software: SageMath implementation
• Algorithm Optimization: Using LLL algorithm and lattice reduction techniques

Evaluation Metrics

1. Computational Efficiency: Runtime of Algorithm 1 on vertices at different distances
2. Coverage Completeness: Ensuring completeness of classification through the Selberg-Zograf bound
3. Result Verification: Verifying correctness of computational results through multiple methods

Experimental Results

Main Computational Achievements

Table 1: Statistics of discovered Shimura curves

Performance Analysis

• Algorithm Efficiency: The runtime of Algorithm 1 grows slowly with vertex distance, demonstrating good scalability
• Large-Scale Computation: Successfully computed fundamental domains of genus 65 (16 vertices, 80 edges) in approximately 30 seconds

Concrete Examples

1. Quadratic Field $\mathbb{Q}(\sqrt{97})$:
   • Prime $\mathfrak{p} = (10-\sqrt{97})$
   • Relatively simple fundamental domain structure
2. Quadratic Field $\mathbb{Q}(\sqrt{5})$:
   • Prime $\mathfrak{p} = (\frac{3\sqrt{5}-1}{2})$
   • Quaternion algebra discriminant $N^- = (21)$
   • Yields a Shimura curve of genus 65

Completeness Verification

Using the Selberg-Zograf bound: $A < \frac{64}{3}(g+1)$ where the area formula is: $A = \frac{4}{(2π)^{2n}} d_F^{3/2} ζ_F(2) Φ(\mathfrak{p}N^-) Ψ(N^+)$

This bound ensures the completeness of the classification.

Related Work

Main Research Directions

1. Bruhat-Tits Tree Theory: Serre's foundational theory provides the mathematical framework for this paper
2. p-adic Uniformization of Shimura Curves: The Čerednik-Drinfel'd theorem provides the theoretical foundation
3. Computational Algebraic Geometry: The Franc-Masdeu algorithm serves as the starting point for this work

Relationship with Related Work

• Generalization: This paper generalizes the FM14 algorithm from the rational numbers to general totally real fields
• Systematicity: Compared to Voight's classification work, this paper provides more systematic computational methods
• Practicality: Provides usable software implementation, facilitating use by other researchers

Conclusions and Discussion

Main Conclusions

1. Successfully generalize the Bruhat-Tits tree fundamental domain computation algorithm to the case of totally real fields
2. Systematically compute complete classifications of low-genus Shimura curves
3. Verify the efficiency and scalability of the algorithm

Limitations

1. Computational Complexity: For high-genus cases, computational complexity remains high
2. Prime Restrictions: The algorithm requires unramified primes with inertia degree 1
3. Memory Requirements: Large-scale computations require substantial memory support

Future Directions

1. Generalize to the case of arbitrary primes
2. Optimize the algorithm to handle higher-genus cases
3. Apply to explicit computation of Shimura curve equations

In-Depth Evaluation

Strengths

1. Theoretical Contribution: Successfully solves the important problem of computing fundamental domains for quaternionic S-arithmetic groups over totally real fields
2. Algorithmic Innovation: Cleverly transforms p-adic computations into computations over the integer ring, greatly simplifying implementation
3. Systematicity: Provides complete classification results with important reference value
4. Practicality: Open-source implementation allows results to be verified and further utilized

Weaknesses

1. Prime Restrictions: Strong restrictions on primes limit the generality of the method
2. Complexity Analysis: Lacks detailed theoretical complexity analysis
3. High-Genus Limitations: Efficiency for high-genus cases needs improvement

Impact

1. Academic Value: Provides important computational tools for Shimura curve theory
2. Application Prospects: Applicable to cryptography, coding theory, and related fields
3. Reproducibility: Complete open-source implementation facilitates verification and extension

Applicable Scenarios

1. Explicit computation and classification of Shimura curves
2. Computational research on quaternion algebras
3. Numerical computation of p-adic automorphic forms
4. Concrete computational problems in algebraic number theory

References

This paper primarily references the following important works:

• FM14 Franc-Masdeu's original work on fundamental domain computation for $\text{GL}_2(\mathbb{Q}_p)$
• Ser03 Serre's classical theory on trees
• Voi09 Voight's classification work on low-genus Shimura curves
• Mil15 Milione's theory on p-adic uniformization of Shimura curves

Overall Assessment: This is a high-quality computational number theory paper that successfully solves an important theoretical computational problem and provides systematic classification results. The algorithm design is ingenious, the implementation is complete, and it makes significant contributions to related fields.