Algebraic twists of GL(2) automorphic forms

TL;DR

This paper establishes new bounds on the correlation of Hecke eigenvalues of GL(2) automorphic forms with algebraic trace functions over finite fields, generalizing previous results to number fields and employing adelic methods.

Contribution

It introduces power-saving bounds for correlations involving trace functions, extending prior work to number fields with a more natural adelic framework.

Findings

Proves a 1/8 power saving for forms with level coprime to p

Achieves a 1/12 power saving for forms with level divisible by p

Generalizes results of Fouvry, Kowalski, and Michel to number fields

Abstract

In this article, we consider the problem of estimating the correlation of Hecke eigenvalues of GL2 automorphic forms with a class of functions of algebraic origin defined over finite fields called trace functions. The class of trace functions is vast and includes many standard exponential sums like Gauss sums, Klostermann sums, Hyperklostermann sums etc. In particular, we prove a Burgess type power saving (of 1/8) over the trivial bound for forms with level coprime to the prime p over whose residue field, the trace function is defined. For forms whose level is divisible by at most one power of p we obtain a saving of 1/12. This generalizes the results of E. Fouvry, E. Kowalski and Ph. Michel to the case of number fields, with a slightly more restrictive assumption on the Fourier-M\"obius group attached to the trace function. Moreover, the implied constant for the 1/12 power saving bound has polynomial dependence on the archimedean part of the conductor of the form. We work using the language of adeles which makes the analysis involved softer and makes the generalization to number fields more natural. The proof proceeds by studying the amplified second moment spectral average of the correlation sum using the relative trace formula.