Comparing Galois representations in the residually reducible case

Basic Information

• Paper ID: 2510.12956
• Title: Comparing Galois representations in the residually reducible case
• Authors: Nuno Freitas, Ignasi Sánchez-Rodríguez
• Classification: math.NT (Number Theory)
• Publication Date: October 14, 2025
• Paper Link: https://arxiv.org/abs/2510.12956v1

Abstract

Let $n \geq 2$ and $p$ be a prime, with $K$ a number field. Consider two Galois representations $\rho_1, \rho_2 : \text{Gal}(\overline{K}/K) \to \text{GL}_n(\mathbb{Z}_p)$ whose residual images are $p$-groups. This paper explains and implements an algorithm that makes Loïc Grenié's results effective for determining whether the semisimplifications of $\rho_1$ and $\rho_2$ are isomorphic. As an application, it is proved that an irreducible representation $\rho: G_{\mathbb{Q}(\sqrt{-3})} \to \text{GL}_2(\mathbb{Z}_3)$ unramified outside 3 is completely determined by the characteristic polynomials of Frobenius elements at five primes of small norm. As additional verification, the algorithm is applied to the 2-adic example studied by Grenié, recovering Grenié's results in a completely automated manner.

Research Background and Motivation

Problem Background

This research concerns the comparison problem for Galois representations, a central problem in algebraic number theory. According to Faltings' foundational work, two continuous representations $\rho_1, \rho_2 : G_K \to \text{GL}_n(E)$ unramified outside a set $S$ are isomorphic if and only if they have the same traces at Frobenius elements in a finite set $T$.

Limitations of Existing Methods

1. Faltings-Serre method: Applies to representations with values in $\text{GL}_2(\mathbb{Q}_2)$ and absolutely irreducible residual image
2. Livné's extension: Handles 2-adic 2-dimensional representations with 2-group residual image
3. Grenié's generalization: Extends to arbitrary $p$-adic $n$-dimensional representations, but still requires $p$-group residual image

The main limitation of these methods is that even for small $n$, $p$, and $S$, they require constructing large number fields, making computation infeasible.

Research Motivation

The core motivations of this paper are:

1. Algorithm Implementation: Transform Grenié's theoretical results into computable algorithms
2. Generality: Design algorithms applicable to general cases without relying on additional information about specific representations
3. Database Potential: Run the algorithm once for each parameter set, potentially creating databases
4. Practical Applications: Prove modularity of abelian surfaces

Core Contributions

1. Algorithm Implementation: Implements a variant of Grenié's theorem (Theorem 4.1), providing a fully automated algorithm
2. Theoretical Improvement: Proposes a method for constructing smaller extension fields $K_S$, significantly reducing computational complexity
3. Concrete Application Results:
   • Proves specific results in the 3-adic case (Corollary 1.1)
   • Recovers Grenié's 2-adic results (Corollary 1.2)
4. Modularity Proof: First successful application of 3-adic Faltings-Serre type methods to prove modularity in the residually reducible case
5. Open Source Implementation: Provides complete Magma code implementation

Detailed Methodology

Task Definition

Input: Two continuous Galois representations $\rho_1, \rho_2 : G_K \to \text{GL}_n(E)$ unramified outside a set $S$, with $p$-group residual images Output: Determine whether the semisimplifications of $\rho_1$ and $\rho_2$ are isomorphic Constraints: $n \geq 2$, $p$ prime, $K$ a number field

Core Algorithm Architecture

1. Improved Extension Field Construction

The traditional $K_S$ construction requires $\lambda + \varepsilon + m$ layers, where:

• $\lambda$ satisfies $2^\lambda \geq r = N^2(1+\varepsilon)^{N(N-1)/2}$
• $\varepsilon = 0$ (if $p \neq 2$) or $\varepsilon = 1$ (if $p = 2$)
• $m$ satisfies $p^m \geq n$

The improved method proposed in this paper:

K₀ = K
For i ≥ 0:
  Kᵢ₊₁ = maximal p-elementary extension of Kᵢ satisfying:
    1. Unramified outside S
    2. Kᵢ₊₁/K is Galois
    3. Exponent of Gal(Kᵢ₊₁/K) divides pᵐ
If Kᵢ₊₁ = Kᵢ for some i < λ + ε + m, stop and set K_S = Kᵢ

2. Main Theorem (Theorem 4.1)

Let $\Sigma \subset G_K$ be a set, and define $\Sigma' = \{\sigma^k : \sigma \in \Sigma, k \in \mathbb{Z}_{\geq 0}\}$. If:

1. $\Sigma'$ covers generators of each maximal cyclic subgroup of $\text{Gal}(K_S/K)$
2. For all $\sigma \in \Sigma$, $\text{CharPoly}(\rho_1(\sigma)) = \text{CharPoly}(\rho_2(\sigma))$

Then the semisimplifications of $\rho_1$ and $\rho_2$ are isomorphic.

3. Algorithm Implementation Details

Part A: Construction of first m layers

K₀ = K
For 0 ≤ i ≤ m-1:
  Kᵢ₊₁ = Kᵢ(ᵖ√α₁, ..., ᵖ√αₐ)
  where αᵢ are generators of Selₚ(Kᵢ, S)

Part B: Subsequent layer construction (i ≥ m)

1. Compute Gal(Kᵢ/K)-submodules: Calculate all Galois submodules of Selₚ(Kᵢ, S)
2. Eliminate 1-dimensional submodules: Check residue degree condition $f(Q|Q \cap O_K) \leq p^m$
3. Eliminate higher-dimensional submodules: Apply Lemma 3.4 recursively
4. Construct field Kᵢ₊₁: Select the unique submodule satisfying the conditions

Technical Innovations

1. Memory Optimization: Avoid storing all submodules by using inductive computation of relevant submodules
2. Residue Degree Bound: Limit extension field size by controlling residue degree $\leq p^m$
3. Early Stopping Criterion: Theorem 8.1 provides natural stopping conditions for the algorithm
4. GRH Avoidance: Under specific conditions, can avoid the Generalized Riemann Hypothesis

Experimental Setup

Computational Environment

• Software: Magma V2.28-20
• Hardware: 4-thread configuration
• Code: Complete implementation available on GitHub

Test Cases

Case 1: Grenié's Example

• Parameters: $n = 3$, $p = 2$, $K = \mathbb{Q}$, $S = \{2\}$
• Extension field layers: $K_S = K_3$ (degree 64 extension)
• Runtime: Approximately 30 hours (assuming GRH)

Case 2: 3-adic Application

• Parameters: $n = 2$, $p = 3$, $K = \mathbb{Q}(\sqrt{-3})$, $S = \{q_3\}$
• Extension field layers: $K_S = K_2$ (degree 54 extension)
• Runtime: Approximately 17 hours (assuming GRH)

Evaluation Metrics

1. Correctness: Consistency with known theoretical results
2. Efficiency: Computation time and memory usage
3. Practicality: Ability to handle real modularity problems

Experimental Results

Main Results

Corollary 1.1 (3-adic case)

Let $K = \mathbb{Q}(\sqrt{-3})$, and $\rho_1, \rho_2 : G_K \to \text{GL}_2(\mathbb{Z}_3)$ be continuous representations unramified outside $q_3$ with the same determinant, which is trivial modulo 3. Then $\rho_1$ and $\rho_2$ have isomorphic semisimplifications if and only if for all $t \in T$, where $T = \{2O_K, p_7^1, p_7^2, p_{19}^*, p_{73}^*\}$$\rho_1(\text{Frob}_t)$ and $\rho_2(\text{Frob}_t)$ have the same trace.

Corollary 1.2 (Recovering Grenié's Results)

For $n = 3$ or $n = 4$, let $\rho_1, \rho_2 : G_{\mathbb{Q}} \to \text{GL}_n(\mathbb{Z}_2)$ be continuous representations unramified outside 2 with 2-group residual image. The prime sets $T = \{5,7,11,17,23,31\}, \quad T' = \{5,7,11,17,19,23,31,73,137,257,337\}$ can be used to determine isomorphism.

Modularity Applications

Abelian Surface of Conductor 37

The algorithm verifies the modularity of the Jacobian of the genus 2 curve with LMFDB label 2187.a.6561.1.

Abelian Surface of Conductor $3^{10}$

Verifies the modularity of the Jacobian of the genus 2 curve with label 59049.a.177147.1.

Computational Efficiency

1. Memory Usage: Reduced from 52GB to 600MB (through optimized submodule computation)
2. Extension Field Size: Significantly smaller than theoretical bounds
3. Practical Feasibility: First successful automated application of the 3-adic Faltings-Serre method

Related Work

Historical Development

1. Faltings (1983): Establishes foundational theoretical framework
2. Serre: Proposes computational methods for the GL₂(ℚ₂) case
3. Livné (1987): Extends to 2-adic 2-dimensional residually reducible case
4. Grenié (2007): Generalizes to arbitrary $p$-adic $n$-dimensional case
5. Duan (2021): Variant methods exploiting special structure

Position of This Paper

This paper is the first complete automated implementation of Grenié's general theory and successfully applies it to real modularity problems.

Conclusions and Discussion

Main Conclusions

1. Theoretical Contribution: Successfully implements an automated algorithm for comparing Galois representations in the residually reducible case
2. Computational Breakthrough: Significantly reduces computational complexity through improved extension field construction
3. Practical Application: First successful application of 3-adic methods to prove modularity of abelian surfaces

Limitations

1. GRH Dependence: Most computations still require assuming the Generalized Riemann Hypothesis
2. Parameter Restrictions: Only applies to cases where residual image is a $p$-group
3. Computational Complexity: Still computationally intensive for large parameters

Future Directions

1. GRH Avoidance: Find more cases that do not depend on GRH
2. Algorithm Optimization: Further reduce computational complexity
3. Application Extension: Apply to more modularity problems
4. Database Construction: Build precomputed databases for commonly used parameters

In-Depth Evaluation

Strengths

1. Theory-Practice Integration: Transforms deep theoretical results into practical algorithms
2. Technical Innovation: Important innovations in memory optimization and algorithm design
3. Practical Value: Solves concrete modularity problems, demonstrating method practicality
4. Open Source Contribution: Provides complete code implementation, promoting research reproducibility

Weaknesses

1. Computational Limitations: Still infeasible for larger parameters
2. Theoretical Assumptions: Heavily relies on unproven assumptions like GRH
3. Limited Scope: Restricted to specific types of Galois representations

Impact

1. Academic Value: Provides new tools for computational problems in algebraic number theory
2. Practical Value: Can be used to verify modularity of elliptic curves and abelian varieties
3. Methodological Contribution: Demonstrates how to algorithmize theoretical results

Applicable Scenarios

1. Modularity Verification: Modularity proofs for elliptic curves and abelian varieties
2. Galois Representation Comparison: Determining representation isomorphism for specific parameters
3. Theoretical Verification: Computational verification of known results

References

The paper cites 17 important references, including Faltings' foundational work, Grenié's theoretical contributions, and related computational method developments. Particularly important are Grenié (2007) for the original theory and van Geemen-Top (1994) for providing test examples.