Serre's uniformity question and proper subgroups of $C_{ns}^+(p)$
Lorenzo Furio, Davide Lombardo
TL;DR
The paper advances Serre's uniformity question over $\mathbb{Q}$ by proving that for every prime $p>5$ the mod-$p$ Galois image of any non-CM elliptic curve $E/\mathbb{Q}$ is not conjugate to the index-$3$ subgroup $G(p)$ of the normaliser of a non-split Cartan. It develops an effective surjectivity framework, refines local analyses of inertia and canonical subgroups, and conducts a detailed study of modular units on the modular curve $X_{G(p)}$ to bound $|j(E)|$ (via $|\log|q||$). A combination of explicit height comparisons, local constraints, and Abel-summation-based refinements yields $|\log|q||<39$ and an explicit prime bound $p<103{,}000$ in the remaining cases, enabling a finite computational verification that excludes the $G(p)$-image for all $p>5$. Consequently, the only possible non-surjective image for primes $p>37$ remains the normaliser of a Cartan, with the exceptional case $p=5$ already observed in prior work. Together, these results substantially advance the uniformity program by reducing the problem to a finite check and clarifying the landscape of residual images for large primes.
Abstract
Serre's uniformity question asks whether there exists a bound $N>0$ such that, for every non-CM elliptic curve $E$ over $\mathbb{Q}$ and every prime $p>N$, the residual Galois representation $ρ_{E,p}:\operatorname{Gal}(\overline{\mathbb{Q}}/\mathbb{Q}) \to \operatorname{Aut}(E[p])$ is surjective. The work of many authors has shown that, for $p>37$, this representation is either surjective or has image contained in the normaliser of a non-split Cartan subgroup $C_{ns}^+(p)$. Zywina has further proved that, whenever $ρ_{E,p}$ is not surjective for $p>37$, its image is either $C_{ns}^+(p)$ or a certain subgroup $G(p)$ of $C_{ns}^+(p)$ of index $3$. Recently, Le Fourn and Lemos showed that the index-$3$ case cannot arise for $p>1.4 \cdot 10^7$. We strengthen this result by proving that the image of $ρ_{E, p}$ is not conjugate to $G(p)$ for any prime larger than $5$.
