On the number of even values of an eta-quotient

TL;DR

This paper establishes a general lower bound on the number of even Fourier coefficients of eta-quotients in arithmetic progressions, extending Serre's classical results and connecting to recent modular form research.

Contribution

It provides a broad lower bound for even coefficients of eta-quotients, generalizing Serre's theorem and incorporating recent advances in modular form lacunarity.

Findings

The ratio of even coefficients to x is unbounded for large x.

The bound approaches the best known for the partition function p(n).

The work links eta-quotients' properties to lacunarity results in modular forms.

Abstract

The goal of this note is to provide a general lower bound on the number of even values of the Fourier coefficients of an arbitrary eta-quotient $F$, over any arithmetic progression. Namely, if $g_{a,b}(x)$ denotes the number of even coefficients of $F$ in degrees $n\equiv b$ (mod $a$) such that $n\le x$, then we show that $g_{a,b}(x) / \sqrt{x}$ is unbounded for $x$ large. Note that our result is very close to the best bound currently known even in the special case of the partition function $p(n)$ (namely, $\sqrt{x}\log \log x$, proven by Bella\"iche and Nicolas in 2016). Our argument substantially relies upon, and generalizes, Serre's classical theorem on the number of even values of $p(n)$, combined with a recent modular-form result by Cotron \emph{et al.} on the lacunarity modulo 2 of certain eta-quotients. Interestingly, even in the case of $p(n)$ first shown by Serre, no elementary proof is known of this bound. At the end, we propose an elegant problem on quadratic representations, whose solution would finally yield a modular form-free proof of Serre's theorem.