Shifted convolution sums of coefficients of symmetric power L L -functions with k k -full kernels over sums of squares in arithmetic progressions

Jewel Mahajan, Arnab Mitra · 2026

Read Shifted convolution sums of coefficients of symmetric power L L -functions with k k -full kernels over sums of squares in arithmetic progressions by Jewel Mahajan, Arnab Mitra (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-30618.txt.

Introduction

Jewel Mahajan and Arnab Mitra return to symmetric power L-functions in a second study of their coefficients over sums of squares. This time they restrict to sums of m squares that land in a fixed residue class modulo q, an arithmetic progression condition, for even m between two and twelve. They study partial sums of the coefficients and of their squares over that restricted set.

As an application, they investigate shifted convolution sums, which compare the coefficient sequence against a shifted copy of itself, weighted by a k-full kernel function for any k of at least two. Kernel functions of this type, introduced by Ivić and Tenenbaum, isolate the part of each integer built from high prime powers. The pair also count sign changes of the coefficients twisted by such a kernel, again over sums of squares. The work extends their moment results to finer arithmetic structure.

Catalog record

CREATORJewel Mahajan, Arnab Mitra
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE207 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.30618

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Shifted convolution sums of coefficients of symmetric power L L -functions with k k -full kernels over sums of squares in arithmetic progressions Jewel Mahajan, Arnab Mitra Abstract . Let q q be an integer and let f f be a normalised Hecke eigenform of integral weight for the full modular group. Let L ⁡ ( s , sym j ​ f ) L(s,\mathrm{sym}^{j}f) denote the j j -th symmetric power L L -function associated to f f , and let λ sym j ​ f ​ ( n ) \lambda_{\mathrm{sym}^{j}f}(n) denote its n n -th coefficient. We study the behaviour of the partial sum of λ sym j ​ f ​ ( n ) \lambda_{\mathrm{sym}^{j}f}(n) , and of its second moment, taken over those sums of m m squares that are congruent to 1 1 modulo q q . As an application, we investigate the shifted convolution sum of λ sym j ​ f ​ ( n ) \lambda_{\mathrm{sym}^{j}f}(n) against a k k -full kernel function, for any k ≥ 2 k\geq 2 . We also study the number of sign changes of λ sym j ​ f ​ ( n ) \lambda_{\mathrm{sym}^{j}f}(n) twisted with a k k -full kernel function, again over sums of m m squares. Throughout, m m is even with m ∈ { 2 , 4 , 6 , 8 , 10 , 12 } m\in\{2,4,6,8,10,12\} . 2020 Mathematics Subject Classification 11M36, 11M41, 11M06 1. Introduction Let k\geq 2 be any integer. An integer valued function a(n) is called a k -full function if p^{k}|a(n) whenever p|a(n) . Note that any positive integer n1 can be uniquely decomposed into n=a_{1}(n)a_{2}(n) with (a_{1}(n),a_{2}(n))=1 , where a_{1}(n) is k -free and a_{2}(n) is k -full. A non-negative integer valued function a(n) is called a k -full kernel function if a(n)=a(a_{2}(n)) for all n\geq 1 and a(n)\ll n^{\epsilon} for all \epsilon>0 . The notion of a k -full kernel function is given by Ivić and Tenenbaum [ 7 ] . Note that k -full kernel functions are not necessarily multiplicative. Let S_{\kappa} be the space of all holomorphic cusp forms of integer weight \kappa for the full modular group \mathrm{SL}(2,\mathbb{Z}) and f\in S_{\kappa} . Let \lambda_{f}(n) be the normalised n -th Fourier coefficient of the Fourier expansion of f(z) at the cusp \infty , i.e., f(z)=\sum_{n=1}^{\infty}\lambda_{f}(n)n^{\frac{\kappa-1}{2}}e^{2\pi inz}, where \Im(z)>0 . Then the L -function attached to \lambda_{f}(n) is defined as L(s,f)=\sum_{n=1}^{\infty}\frac{\lambda_{f}(n)}{n^{s}} for \Re(s)>1 , where \lambda_{f}(n) are the eigenvalues of all the Hecke operators T_{n} . In 1974, Deligne [ 2 ] proved that for any prime p , there exist complex numbers \alpha(p) and \beta(p) such that (1) \displaystyle\alpha(p)\beta(p) \displaystyle=1, (2) \displaystyle|\alpha(p)|=|\beta(p)| \displaystyle=1. Then L(s,f) can be written as L(s,f)=\prod_{p}\left(1-\frac{\alpha(p)}{p^{s}}\right)^{-1}\left(1-\frac{\beta(p)}{p^{s}}\right)^{-1}, where the product runs over all primes p . Also, |\lambda_{f}(n)|\leq d(n) , where d(n) is the divisor function. The symmetric square L -function is defined as L(s,\mathrm{sym}^{2}f):=\sum_{n=1}^{\infty}\frac{\lambda_{\mathrm{sym}^{2}f}(n)}{n^{s}}=\prod_{p}\left(1-\frac{\alpha^{2}(p)}{p^{s}}\right)^{-1}\left(1-\frac{\beta^{2}(p)}{p^{s}}\right)^{-1}\left(1-\frac{1}{p^{s}}\right)^{-1} for \Re(s)>1 , where \lambda_{\mathrm{sym}^{2}f}(n) is multiplicative. In 1987 , Erdős and Ivić [ 3 ] proved that \displaystyle\sum_{n\leq x}a(n)d(n+1)=C_{1}x\log x+C_{2}x+O(x^{\frac{8}{9}+\epsilon}) \displaystyle\sum_{n\leq x}a(n)w(n+1)=D_{1}x\log\log x+D_{2}x+O\left(\frac{x}{\log x}\right), where d(n) is the divisor function, and w(n) is the number of different prime factors of n , and C_{1}>0 , D_{1}>0 , C_{2} , D_{2} are constants that can be evaluated explicitly. In [ 13 ] , Lü and Wang investigated the shifted convolution sums of squares of Fourier coefficients with a 2 -full kernel function a(n) and obtained an asymptotic formula for the sum \displaystyle\sum_{n\leq x}a(n)\lambda^{2}_{f}(n+1). Later, Venkatasubbareddy and Sankaranarayanan [ 12 ] extended the problem to fourth moments, and generalised to k -full kernel functions. For q\geq 100 and q\ll x^{\frac{23}{181}-\epsilon} , they proved that (3) \displaystyle\sum_{\begin{subarray}{c}n\leq x+1\\ n\equiv 1~(\mathrm{mod}~q)\end{subarray}}\lambda_{f}^{4}(n)=c_{1}x\log x\frac{\phi(q)}{q^{2}}+O\left(\frac{x^{\frac{158}{181}+\epsilon}q^{1+\epsilon}}{\phi(q)}\right) and for q\geq 100 and q\ll x^{\frac{3}{23}-\epsilon} , they proved that (4) \displaystyle\sum_{\begin{subarray}{c}n\leq x+1\\ n\equiv 1~(\mathrm{mod}~q)\end{subarray}}\lambda_{f}^{4}(n)=c_{1}x\log x\frac{\phi(q)}{q^{2}}+O\left(\frac{x^{\frac{20}{23}+\epsilon}q^{1+\epsilon}}{\phi(q)}\right). As an application, for a k -full kernel function a(n) , they proved that (5) \displaystyle\sum_{n\leq x}a(n)\lambda_{f}^{4}(n+1)=c_{2}x\log x+O\left(x^{\frac{520k+23}{543k}+\epsilon}\right). On the other hand, Wang [ 14 ] proved that for a k -full kernel function a(n) and f\in S_{\kappa} , (6) \displaystyle\sum_{n\leq x}a(n)\lambda^{2}_{\mathrm{sym}^{j}f}(n+1)=C_{f,j}x+O\left(x^{1-\frac{2k-2}{3k(j+1)^{2}}+\epsilon}\right), and (7) \displaystyle\sum_{n\leq x}a(n)\lambda^{3}_{\mathrm{sym}^{2}f}(n+1)=Dx+O\left(x^{\frac{1805k+46}{1851k}+\epsilon}\right). This article investigates the behaviour of the partial sum of \lambda_{\mathrm{sym}^{j}f}(n) and its second moment for a sequence of integers n which are expressible as a sum of m squares, which are 1 modulo q , where q is an integer. As an application, we examine their shifted convolution sum with a weight function, which is a k -full kernel function for k\geq 2 . We also examine the number of sign changes of a(n)\lambda_{\mathrm{sym}^{j}f}(n+1) , where n+1 runs over the sum of m squares. Here m=2,4,6,8,10,12 . In particular, we prove Theorem 1 . Let f\in S_{\kappa} and let q\geq 100 be any integer. Then for any \epsilon>0 and q\ll x^{\frac{1}{j+1}-\epsilon} , where j\geq 2 is a fixed integer, we have \sum_{\begin{subarray}{c}a_{1}^{2}+a_{2}^{2}\leq x+1\\ a_{1}^{2}+a_{2}^{2}\equiv 1~(\mathrm{mod}~q)\\ (a_{1},a_{2})\in\mathbb{Z}^{2}\end{subarray}}\lambda_{\mathrm{sym}^{j}f}(a_{1}^{2}+a_{2}^{2})=O\left(\frac{x^{1+\epsilon-\frac{1}{j+1}}q^{1+\epsilon}}{\phi(q)}\right). Theorem 2 . Let f\in S_{\kappa} and let q\geq 100 be any integer. Then for any \epsilon>0 and q\ll x^{\frac{2}{j+1}-\epsilon} , where j\geq 2 is a fixed integer and for m\in\{4,6,8,10,12\} , we have \sum_{\begin{subarray}{c}a_{1}^{2}+\cdots+a_{m}^{2}\leq x+1\\ a_{1}^{2}+\cdots+a_{m}^{2}\equiv 1~(\mathrm{mod}~q)\\ (a_{1},\ldots,a_{m})\in\mathbb{Z}^{m}\end{subarray}}\lambda_{\mathrm{sym}^{j}f}(a_{1}^{2}+\cdots+a_{m}^{2})=O\left(\frac{x^{\frac{m}{2}+\epsilon-\frac{2}{j+3}}q^{\frac{j+1}{j+3}+\epsilon}}{\phi(q)}\right). Theorem 3 . Let f\in S_{\kappa} and let q\geq 100 be any integer. Then for any \epsilon>0 and q\ll x^{\frac{1}{(j+1)^{2}}-\epsilon} , where j\geq 2 is a fixed integer, we have \sum_{\begin{subarray}{c}a_{1}^{2}+a_{2}^{2}\leq x+1\\ a_{1}^{2}+a_{2}^{2}\equiv 1~(\mathrm{mod}~q)\\ (a_{1},a_{2})\in\mathbb{Z}^{2}\end{subarray}}\lambda_{\mathrm{sym}^{j}f}^{2}(a_{1}^{2}+a_{2}^{2})=\frac{c_{j,f}}{q}x+O\left(\frac{x^{1+\epsilon-\frac{1}{(j+1)^{2}}}q^{1+\epsilon}}{\phi(q)}\right), where c_{j,f} is a constant that depends on j and f . Theorem 4 . Let f\in S_{\kappa} and let q\geq 100 be any integer. Then for any \epsilon>0 and q\ll x^{\frac{2}{(j+1)^{2}}-\epsilon} , where j\geq 2 is a fixed integer and for m\in\{4,6,8,10,12\} , we have \sum_{\begin{subarray}{c}a_{1}^{2}+\cdots+a_{m}^{2}\leq x\\ a_{1}^{2}+\cdots+a_{m}^{2}\equiv 1~(\mathrm{mod}~q)\\ (a_{1},\ldots,a_{m})\in\mathbb{Z}^{m}\end{subarray}}\lambda_{\mathrm{sym}^{j}f}^{2}(a_{1}^{2}+\cdots+a_{m}^{2})=\frac{c_{m,j,f}}{q}x^{\frac{m}{2}}+O\left(\frac{x^{\frac{m}{2}+\epsilon-\frac{2}{(j+1)^{2}}}q^{1+\epsilon}}{\phi(q)}\right), where c_{m,j,f} is a constant that depends on j , f and m . Theorem 5 . For any integer k\geq 2 , let a(n) be a k -full kernel function and f\in S_{\kappa} . Let j\geq 2 be an integer. Then for any \epsilon>0 and m\in\{4,6,8,10,12\} , we have \sum_{\begin{subarray}{c}n\leq x\\ (a_{1},\ldots,a_{m})\in\mathbb{Z}^{m}\\ n=a_{1}^{2}+\cdots+a_{m}^{2}-1\end{subarray}}a(n)\lambda_{\mathrm{sym}^{j}f}(n+1)=O\left(x^{\frac{m}{2}-\frac{2k-2}{3k(j+1)}+\epsilon}\right), and also \sum_{\begin{subarray}{c}n\leq x+1\\ (a_{1},a_{2})\in\mathbb{Z}^{2}\\ n=a_{1}^{2}+a_{2}^{2}-1\end{subarray}}a(n)\lambda_{\mathrm{sym}^{j}f}(n+1)=O\left(x^{1-\frac{k-1}{3k(j+1)}+\epsilon}\right). Theorem 6 . For any integer k\geq 2 , let a(n) be a k -full kernel function and f\in S_{\kappa} . Let j\geq 2 be an integer. Then for any \epsilon>0 and m\in\{4,6,8,10,12\} , we have \sum_{\begin{subarray}{c}n\leq x+1\\ (a_{1},\ldots,a_{m})\in\mathbb{Z}^{m}\\ n=a_{1}^{2}+\cdots+a_{m}^{2}-1\end{subarray}}a^{2}(n)\lambda_{\mathrm{sym}^{j}f}^{2}(n+1)=D_{j,f,m}x^{\frac{m}{2}}+O\left(x^{\frac{m}{2}-\frac{2k-2}{3(j+1)^{2}k}+\epsilon}\right), and also \sum_{\begin{subarray}{c}n\leq x+1\\ (a_{1},a_{2})\in\mathbb{Z}^{2}\\ n=a_{1}^{2}+a_{2}^{2}-1\end{subarray}}a^{2}(n)\lambda_{\mathrm{sym}^{j}f}^{2}(n+1)=D_{j,f,2}x+O\left(x^{1-\frac{k-1}{3(j+1)^{2}k}+\epsilon}\right), where D_{j,f,m} and D_{j,f,2} are constants that can be evaluated explicitly. Theorem 7 . Let f\in S_{\kappa} and j\geq 2 be a fixed integer and a(n) be a k -full kernel function for any integer k\geq 2 . Then, for sufficiently large x , the sequence \{a(n)\lambda_{\mathrm{sym}^{j}f}(n+1)~|~n=a_{1}^{2}+a_{2}^{2}-1,a_{i}\in\mathbb{Z}\} has at least x^{1-\delta_{j}} sign changes between x and 2x , for any \delta_{j} with 1-\frac{k-1}{3k(j+1)^{2}}<\delta_{j}<1 . Theorem 8 . Let f\in S_{\kappa} and j\geq 2 be a fixed integer, m\in\{4,6,8,10,12\} and a(n) be a k -full kernel function for any integer k\geq 2 . Then for sufficiently large x , the sequence \{a(n)\lambda_{\mathrm{sym}^{j}f}(n+1)~|~n=\sum_{i=1}^{m}a_{i}^{2}-1,a_{i}\in\mathbb{Z}\} has at least x^{1-\delta_{j}} sign changes between x and 2x , for any \delta_{j} with 1-\frac{2k-2}{3k(j+1)^{2}}<\delta_{j}<1-\frac{k-1}{3k(j+1)^{2}} . Organisation of the article This article is organised as follows. In Section 2 , we discuss the preliminaries, important lemmas, and bounds necessary to prove our results. The proofs of our main results, related to the partial sums of the symmetric power L -function attached to Hecke eigenforms, are given in

Read the complete text