Learning Euler Factors of Elliptic Curves

We apply transformer models and feedforward neural networks to predict Frobenius traces ap from elliptic curves given other traces aq. We train further models to predict ap bmod 2 from aq bmod 2, and cross-analysis such as ap bmod 2 from aq. Our experiments reveal that these models achieve high accuracy, even in the absence of explicit number-theoretic tools like functional equations of L-functions. We also present partial interpretability findings.