An Improved Bound Towards a Conjecture of Serre on Surjective Galois Representations

Suppose that $E$ is an elliptic curve defined over $\mathbb{Q}$ without complex multiplication and with conductor $N$. For each positive integer $m$, the action of the absolute Galois group $G_{\mathbb{Q}}=\operatorname{Gal}(\bar{\mathbb{Q}}/\mathbb{Q})$ on the torsion points over $\bar{\mathbb{Q}}$ gives rise to a representation of $G_\mathbb{Q}$. A celebrated paper of Serre shows that this representation is surjective for all sufficiently large primes; the other primes are termed \emph{exceptional}. Serre conjectures that there are no exceptional primes $\ell>37$ for any non CM elliptic curve over $\mathbb{Q}$. The best result in this direction is due to Cojocaru, who proves that the largest exceptional prime $\ell_0\ll_{\epsilon} N^{1+\epsilon}$. In this paper we lower the exponent on the bound to obtain \ell_0\ll_{\epsilon} N_0^{1/4}+\epsilon}, where $N_0$ is the product of primes of bad reduction. If $E$ has no places of multiplicative reduction, then we have \ell_0\ll_{\epsilon}N^{1/8}+\epsilon}. Assuming the Frey-Szpiro conjecture, we have that \ell_0\ll_{\epsilon}N^{1/8}+\epsilon} in general. Our main methods include the Rankin-Selberg method and the classical work on distribution of quadratic residues.