Machine-learning the Sato-Tate conjecture (Q2066955)

In this paper, some of the latest techniques from machine-learning to the arithmetic of hyperelliptic curves are applied. More precisely, it is observed that a Bayes classifier can be trained to distinguish the Sato-Tate groups for genus 2 curves efficiently with high precision, given a small number of Euler factors for the \(L\)-function. These results are consistent with the Sato-Tate conjecture for curves of low genus, and provide evidence that machine-learning can be used to classify curves according to their Sato-Tate groups. Note that the approach of using Euler factors is in accordance with the setup of the Langlands program. Further, the observations of the paper are verified using known results from the literature and the data available in the LMFDB.