Average behaviour of higher moments of cusp form coefficients (Q2080519)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Average behaviour of higher moments of cusp form coefficients
scientific article

    Statements

    Average behaviour of higher moments of cusp form coefficients (English)
    0 references
    0 references
    0 references
    9 October 2022
    0 references
    Let \(H_k\) be the space of all normalized primitive holomorphic cusp forms of even integral weight \(k\) for the full modular group, which consists of the eigenfunctions for all Hecke operators. \(f\in H_k\) admits the Fourier expansion \[ f(z)=\sum_{n=1}^{\infty}\lambda_f(n)n^{\frac{k-1}{2}}e^{2\pi inz}, \quad \operatorname{Im}z>0, \] where we normalize \(\lambda_f(1) = 1\) and \(\lambda_f(n) \) is the \(n\)th normalized Fourier coefficient (Hecke eigenvalue) of \(f\). The Ramanujan-Petersson conjecture (proved by Deligne) implies that there exist two complex numbers \(\alpha_f(p),\beta_f(p)\) such that \[ \lambda_f(p)=\alpha_f(p)+\beta_f(p), \quad |\alpha_f(p)|=| \beta_f(p)|=\alpha_f(p)\beta_f(p)=1.\tag{1.2} \] The Hecke \(L\)-function associated with \(f\) is defined by (\(\operatorname{Re}s>1\)) \[ L(f,s)=\sum_{n=1}^{\infty}\frac{\lambda_f(n)}{n^s}=\prod_{p}^{}{\left( 1-\frac{\alpha_f(p)}{p^s}\right)}^{-1}{\left( 1-\frac{\beta_f(p)}{p^s}\right)}^{-1}. \] The Rankin-Selberg \(L\)-function associated with \(f\) is defined by \begin{align*} L(f\times f,s)&=\sum_{n=1}^{\infty}\frac{\lambda_f(n)}{n^s}=\zeta(2s)\prod_{p}^{}{\left( 1-\frac{\alpha_f(p)^2}{p^s}\right)}^{-1}{\left( 1-\frac{\beta_f(p)^2}{p^s}\right)}^{-1}\\ & =\zeta(2s)\sum_{n=1}^{\infty}\frac{\lambda_f(n)^2}{n^s}:=\sum_{n=1}^{\infty}\frac{\lambda_{f\times f}(n)}{n^s}. \end{align*} For any given integer \(w\ge 1\), consider the \(w\)th convolution of power of \(\lambda_{f\times f}\) generated by \(L(f\times f,s)\): \[ L(f\times f,s)^w=\sum_{n=1}^{\infty}\frac{\lambda_{w,f\times f}(n)}{n^s}, \quad \operatorname{Re}s>1. \] Let \(\boldsymbol{i}=(i_1,\dots,i_l), \boldsymbol{w}=(w_1,\dots,w_l)\in \mathbb{N}^l\) be \(l\)-dimensional vectors and let \[ \Lambda_{\boldsymbol{w}}^{\boldsymbol{i}}(n)=\lambda_{w_1,f\times f}(n)^{i_1}\cdots \lambda_{w_l,f\times f}(n)^{i_l}. \] The main purpose of the paper is to obtain an asymptotic formula for the circle-divisor problem type sum \begin{align*} S_{\boldsymbol{w}}^{\boldsymbol{i}}(x)&:= \sum_{n=a^2+b^2\le x} \Lambda_{\boldsymbol{w}}^{\boldsymbol{i}}(n)= \sum_{n=a^2+b^2\le x}\lambda_{w_1,f\times f}(n)^{i_1}\cdots \lambda_{w_l,f\times f}(n)^{i_l}\\ &=4\sum_{n\le x}\lambda_{w_1,f\times f}(n)^{i_1}\cdots \lambda_{w_l,f\times f}(n)^{i_l}r(n), \end{align*} where \(r(n)=\sum_{d|n}\chi_4(d)\) with \(\chi_4\) denoting the real primitive character associated with the Gaussian field. Then from the \(L\)-function \[ L_{\boldsymbol{w}}^{\boldsymbol{i}}(s):=\frac{1}{4}\sum_{n=1}^{\infty}\frac{ \Lambda_{\boldsymbol{w}}^{\boldsymbol{i}}(n)}{n^s}, \quad \operatorname{Re}s>1 \tag{2.4} \] it is possible to factor out the power of \(\zeta(s)\) and the product of some symmetric \(L\)-functions and the twisted \(L\)-functions with the remaining factor being a Dirichlet series which converges absolutely for \(\operatorname{Re}s>\frac{1}{2}\) by the results of \textit{Y.-K. Lau} and \textit{G. Lü} [Q. J. Math. 62, No. 3, 687--716 (2011; Zbl 1269.11044)]. Then by the well-known procedure and many known results on \(L\)-functions, the author proves an asymptotic formula for \(S_{\boldsymbol{w}}^{\boldsymbol{i}}(x)\) similar to that for the divisor problem.
    0 references
    0 references
    0 references
    0 references
    0 references
    Fourier coefficients
    0 references
    holomorphic cusp forms
    0 references
    Rankin-Selberg \(L\)-functions
    0 references
    0 references