Let f(z) be a holomorphic Hecke eigenform of weight k with respect to the full modular group SL(2,Z), and let Δρ(x;sym2f) be the error term of the Riesz mean of the symmetric square L-function L(s,sym2f). In this paper, using a Voronoï type formula for Δρ(x;sym2f), we consider the third-power moment of Δρ(x;sym2f) and derive the asymptotic formula for
∫T1Δ3ρ(x;sym2f)dx.
Citation: Rui Zhang, Xiaofei Yan. The third-power moment of the Riesz mean error term of symmetric square L-function[J]. AIMS Mathematics, 2021, 6(9): 9436-9445. doi: 10.3934/math.2021548
[1] | Zhen Guo . On mean square of the error term of a multivariable divisor function. AIMS Mathematics, 2024, 9(10): 29197-29219. doi: 10.3934/math.20241415 |
[2] | Bing Long, Zaifu Jiang . Estimation and prediction for two-parameter Pareto distribution based on progressively double Type-II hybrid censored data. AIMS Mathematics, 2023, 8(7): 15332-15351. doi: 10.3934/math.2023784 |
[3] | Gültekin Tınaztepe, Sevda Sezer, Zeynep Eken, Sinem Sezer Evcan . The Ostrowski inequality for s-convex functions in the third sense. AIMS Mathematics, 2022, 7(4): 5605-5615. doi: 10.3934/math.2022310 |
[4] | Alanazi Talal Abdulrahman, Khudhayr A. Rashedi, Tariq S. Alshammari, Eslam Hussam, Amirah Saeed Alharthi, Ramlah H Albayyat . A new extension of the Rayleigh distribution: Methodology, classical, and Bayes estimation, with application to industrial data. AIMS Mathematics, 2025, 10(2): 3710-3733. doi: 10.3934/math.2025172 |
[5] | Zhuoyu Chen, Wenpeng Zhang . A new reciprocity formula of Dedekind sums and its applications. AIMS Mathematics, 2024, 9(5): 12814-12824. doi: 10.3934/math.2024626 |
[6] | Shashi Bhushan, Anoop Kumar, Md Tanwir Akhtar, Showkat Ahmad Lone . Logarithmic type predictive estimators under simple random sampling. AIMS Mathematics, 2022, 7(7): 11992-12010. doi: 10.3934/math.2022668 |
[7] | Refah Alotaibi, Hassan Okasha, Hoda Rezk, Abdullah M. Almarashi, Mazen Nassar . On a new flexible Lomax distribution: statistical properties and estimation procedures with applications to engineering and medical data. AIMS Mathematics, 2021, 6(12): 13976-13999. doi: 10.3934/math.2021808 |
[8] | Jinran Yao, Zhengwei Yin . Numerical contractivity preserving implicit balanced Milstein-type schemes for SDEs with non-global Lipschitz coefficients. AIMS Mathematics, 2024, 9(2): 2766-2780. doi: 10.3934/math.2024137 |
[9] | Azam Zaka, Ahmad Saeed Akhter, Riffat Jabeen . The new reflected power function distribution: Theory, simulation & application. AIMS Mathematics, 2020, 5(5): 5031-5054. doi: 10.3934/math.2020323 |
[10] | Emrah Altun, Mustafa Ç. Korkmaz, M. El-Morshedy, M. S. Eliwa . The extended gamma distribution with regression model and applications. AIMS Mathematics, 2021, 6(3): 2418-2439. doi: 10.3934/math.2021147 |
Let f(z) be a holomorphic Hecke eigenform of weight k with respect to the full modular group SL(2,Z), and let Δρ(x;sym2f) be the error term of the Riesz mean of the symmetric square L-function L(s,sym2f). In this paper, using a Voronoï type formula for Δρ(x;sym2f), we consider the third-power moment of Δρ(x;sym2f) and derive the asymptotic formula for
∫T1Δ3ρ(x;sym2f)dx.
Let H be the upper half plane, i.e., H={z=x+iy| x,y∈R,and y>0}. Let Sk(Γ) be the space of holomorphic cusp forms of even weight k≥2 for the full modular group Γ=SL(2,Z). Then for f∈Sk(Γ) and z∈H, we have the Fourier expansion
f(z)=∞∑n=1af(n)e(nz), |
where af(n) denotes the n-th Fourier coefficient, and e(x)=e2πix. Set λf(n)=af(n)n−k−12, then λf(n) is multiplicative. The analytic properties of λf(n) were studied by many authors (see [5,7,8,9,10,12,13,14,16,20,22,23,24,25]). The symmetric square L-function attached to f can be written as
L(s,sym2f)=ζ(2s)∞∑n=1λf(n2)n−s=∞∑n=1cnn−s |
for ℜs>1. And it can be continued to an entire function on the whole complex plane. Then it is known that for any ϵ>0,
|cn|≤d3(n)≪nϵ, |
where d3(n) is the number of ways to write n as a product of three factors.
Hafner [3] considered the Riesz mean of the form
Dρ(x;sym2f)=1Γ(ρ+1)∑n≤x′(x−n)ρcn, |
where ρ≥0 is a fixed number and ∑′ indicates that cn is replaced by cn/2 if ρ=0. Also, the Riesz mean can be represented as a sum of "residue function" and "error term":
Dρ(x;sym2f)=L(0,sym2f)Γ(ρ+1)xρ+Δρ(x;sym2f). |
Then let Δρ(x;sym2f) be the error term of the Riesz mean of the symmetric square L-function. Hafner [3] gave that for ρ=0, one has
∑n≤xcn=L(0,sym2f)+Δ0(x;sym2f). |
The symmetric square L-function and the Riesz mean error term have been studied by many authors, for example, see [1,4,6,11,15,18,21,25]. Fomenko [2] considered Δρ(x;symmf) when m=2 and obtained a truncated Voronoï type formula, and we give it in Lemma 2.1. Let N=x12 in Lemma 2.1, one obtain
Δρ(x;sym2f)≪x1+ρ2. |
Using the Voronöi formula, Fomenko got
∫X1Δ2ρ(x;sym2f)dx=CX4ρ+53+O(Xρ+53+ϵ), |
where
C=2−2ρ−1π−2ρ−2(4ρ+5)−1∞∑n=1c2nn−2ρ+43. |
Wang [19] generalized the truncated Voronöi type formula to the case m≥3 under the hypothesis Nice(m,f):
Δρ(x;symmf)=ϵ(symmf)2−ρπ−ρ−1(m+1)−12xmm+1ρ+m2(m+1)×∑n≤Ncnn−1m+1ρ−m+22(m+1)cos(2π(m+1)(nx)1m+1−(A(m,k)+ρ2)π)+O(xmm+1ρ+m−12(m+1)N−ρm+1ρ+m−12(m+1)+ϵ)+O(xmm+1(ρ+1)+ϵN−ρ+1m+1), |
where
A(m,k)={m(m+2)(k−1)8+12, m≡0(mod4);m(m+2)(k−1)8+1, m≡2(mod4);(m+1)2(k−1)8+34, m is odd, |
ϵ(symmf)={+1, if m even ;ϵ(k,m), otherwise , |
with
ϵ(k,m):=i(m+12)2(k−1)+m+12={ik, if m ≡1 (mod 8),−1, if m ≡3 (mod 8),−ik, if m ≡5 (mod 8),+1, if m ≡7 (mod 8). |
Liu and Wang [15] studied the higher-power moments of Δρ(x;sym2f). If exists a real number A0=A0(ρ)>3 such that
∫T1ΔA0ρ(x;sym2f)dx≪T1+2ρ+13A0+ϵ, |
then for any integer 3≤h<A0, they had the following asymptotic formula
∫T1Δhρ(x;sym2f)dx=6Bρ(h,c)(3+(2ρ+1)h)(2π)(ρ+1)h3h2T1+2ρ+13h+O(T1+2ρ+13h+ϵ(T−δρ(h,A0)+T−ρ3)), |
where
Bρ(h;f):=h−1∑l=1(h−1l)sρ(l,h;f)cos(πρ2(h−2l)) |
with
sρ(l,h;f):=∑3√n1+⋯+3√nl=3√nl+1+⋯+3√nhf(n1)⋯f(nh)(n1⋯nh)(ρ+2)/3(1⩽ |
and
\delta_{\rho}\left(h, A_{0}\right): = \sigma_{\rho}\left(h, A_{0}\right) \min \left(\frac{2 \rho}{2 \rho+1}, \frac{1}{3 b_{\rho}\left(H_{0}\right)}\right) |
with
\sigma_{\rho}\left(h, A_{0}\right): = \frac{(2 \rho+1)\left(A_{0}-h\right)}{3\left(A_{0}-2\right)}, \quad 3 \leqslant h < A_{0}, |
b_{\rho}(h): = 3^{h-2}-\frac{1}{3}+\frac{(1-\rho) h}{3} |
and H_{0} is the least even integer such that n \geqslant A_{0} . They also proved the result for \rho = \frac{1}{2} , h = 3, 4, 5 ,
\begin{equation*} \int_{1}^{T}\Delta_{\rho}^{h}(x;sym^{2}f)dx = \frac{6B_{\frac{1}{2}}(h,c)}{(3+2h)(2\pi)^{\frac{3}{2}h}3^{\frac{h}{2}}}T^{1+\frac{2}{3}h} +O\left(T^{1+\frac{2}{3}h-\lambda_{1/2}(h,6)+\epsilon}\right), \end{equation*} |
where \lambda_{1/2}(3, 6) = \frac{1}{22} , \lambda_{1/2}(4, 6) = \frac{1}{87} , \lambda_{1/2}(5, 6) = \frac{1}{498} . Zhang, Han and Zhang [26] studied the power moment of the Riesz mean error term in short intervals. For k\geq3 , T^{\frac{2}{3}+\frac{2\epsilon d_{\rho, k}}{(2\rho+1)\delta}}\leq H\leq T and \epsilon > 0 , they got
\begin{equation*} \int_{T-H}^{T+H}\Delta_{\rho}^{k}(x;sym^{2}f)dx = B_k\int_{T-H}^{T+H}x^{\frac{2\rho+1}{3}k}dx+ O\left(HT^{\frac{2\rho+1}{3}k+\epsilon}(HT^{-\frac{2}{3}})^{-\frac{(2\rho+1)\delta}{d_{\rho,k}}}\right), \end{equation*} |
where
B_k = 2^{-k\rho-k+1}\pi^{-k\rho-k}3^{-\frac{k}{2}}c_k,\ d_{\rho,k} = (3^{k-1}-4+k-\rho k)(k+\delta-2)+(2\rho+1)\delta. |
Their results improved the results in [15] when \rho = \frac{1}{2} , k = 3, 4, 5 and \delta > \frac{50}{31} .
In this paper, we will use the Voronöi type formula for \Delta_{\rho}(x; sym^{2}f) to study the third-power moment estimates of \Delta_{\rho}(x; sym^{2}f) and obtain the following theorem.
Theorem 1. For any f\in S_{k}(\Gamma)^{+} , x > 1 and \frac{\sqrt{612}}{18}-1 < \rho < \frac{2}{3} , we have
\begin{equation*} \int_{1}^{T}\Delta_{\rho}^{3}(x;sym^{2}f)dx = \left(2^{\rho}\pi^{\rho+1}\sqrt{3}\right)^{-3}\frac{3}{8\rho+8}\cos\left(\frac{\pi\rho}{2}\right)A T^{2\rho+2}+O\left(T^{\frac{6\rho^{2}+21\rho+16}{8+3\rho}+\epsilon}\right), \end{equation*} |
where
\begin{equation*} A = \sum\limits^{\infty}_{\alpha,\ \beta = 1}\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2}\sum\limits^{\infty}_{h = 1\atop h\ is\ cube-free} h^{-\rho-2}c_{\alpha^{3}h}c_{\beta^{3}h}c_{(\alpha+\beta)^{3}h}. \end{equation*} |
Particularly, for \rho = \frac{1}{2} , we can get the following result.
Corollary 2. Let \rho = \frac{1}{2} . For any \epsilon > 0 , we have
\begin{equation*} \int_{1}^{T}\Delta_{\rho}^{3}(x;sym^{2}f)dx = \frac{\sqrt{3}\pi^{-\frac{9}{2}}}{144}B T^{3}+O(T^{\frac{56}{19}+\epsilon}), \end{equation*} |
where
\begin{equation*} B = \sum\limits^{\infty}_{\alpha,\ \beta = 1}\left(\alpha\beta(\alpha+\beta)\right)^{-\frac{3}{2}}\sum\limits^{\infty}_{h = 1\atop h\ is\ cube-free} h^{-\frac{3}{2}}c_{\alpha^{3}h}c_{\beta^{3}h}c_{(\alpha+\beta)^{3}h}. \end{equation*} |
Remark. Note that our results improve the results in [15].
To prove our theorem, we need the following lemmas.
Lemma 2.1. Let x > 1 , N\geq1 . Then for any fixed \rho , 0\leq \rho \leq 1 , we have
\begin{equation} \begin{split} \Delta_{\rho}(x;sym^{2}f) = &2^{-\rho}\pi^{-\rho-1}3^{-\frac{1}{2}}\Sigma(x)+O\left(x^{\frac{4\rho+1}{6}}N^{\frac{-2\rho+1}{6}+\epsilon}\right)\\ &+O\left(x^{\frac{2\rho+2}{3}+\epsilon}N^{-\frac{\rho+1}{3}}\right)+O\left(x^{\epsilon}\right), \end{split} \end{equation} | (2.1) |
where
\begin{equation*} \Sigma(x) = \sum\limits_{n\leq N}c_nn^{-\frac{\rho+2}{3}}x^{\frac{2\rho+1}{3}}\cos\left(6\pi\sqrt[3]{nx}-\frac{\pi\rho}{2}\right). \end{equation*} |
Proof. See Theorem 1.1 in [2].
Lemma 2.2. Let h\geq3 , (i_1, \cdots, i_{h-1})\in \{0, 1\}^{h-1} such that
\begin{equation*} \sqrt[3]{n_1}+(-1)^{i_1}\sqrt[3]{n_2}+(-1)^{i_2}\sqrt[3]{n_3}+\cdots+(-1)^{i_{h-1}}\sqrt[3]{n_h}\neq0. \end{equation*} |
Then
\begin{equation*} \left|\sqrt[3]{n_1}+(-1)^{i_1}\sqrt[3]{n_2}+\cdots+(-1)^{i_{h-1}}\sqrt[3]{n_h}\right|\gg \max(n_1,\cdots,n_h)^{-(3^{h-2}-3^{-1})}. \end{equation*} |
Proof. See for example Lemma 2.3 in [21].
Lemma 2.3. If g(x) and h(x) are continuous real-valued functions of x and g(x) is monotonic, then
\begin{equation*} \int_{a}^{b}g(x)h(x)dx\ll \left(\max\limits_{a\leq x\leq b}|g(x)|\right)\left(\max\limits_{a\leq u < v\leq b}\left|\int_{u}^{v}h(x)dx\right|\right). \end{equation*} |
Proof. See the Lemma 1 in [17].
Lemma 2.4. Suppose k\geq3 , (i_1, \ldots, i_{k-1})\in\{0, 1\}^{k-1} , (i_1, \ldots, i_{k-1})\neq(0, \ldots, 0) and
N_1, \ldots, N_k > 1,\ \ 0 < \Delta\ll E^{\frac{1}{3}},\ \ E = \max(N_1, \ldots, N_k). |
Let
\begin{equation*} \mathcal{A} = \mathcal{A}(N_1, \ldots, N_k;i_1, \ldots, i_{k-1};\Delta) \end{equation*} |
denote the number of solutions of the inequality
\begin{equation} \left|\sqrt[3]{n_{1}}+(-1)^{i_1}\sqrt[3]{n_2}+\cdots+(-1)^{i_{k-1}}\sqrt[3]{n_k}\right| < \Delta \end{equation} | (2.2) |
with N_j < n_j\leq2N_j , 1\leq j\leq k . Then
\begin{equation*} \mathcal{A}\ll \Delta E^{-\frac{1}{3}}N_1\cdots N_k+E^{-1}N_1\cdots N_k. \end{equation*} |
Proof. It can be proved similarly as Lemma 2.4 of [21]. Without loss of generality, suppose E = N_{k} . If n_{1}, \ldots, n_{k} satisfy (2.2), then
\left|\sqrt[3]{n_{1}}+(-1)^{i_1}\sqrt[3]{n_2}+\cdots+(-1)^{i_{k-2}}\sqrt[3]{n_{k-1}}\right| = (-1)^{i_{k-1}+1}\sqrt[3]{n_k}+\theta\Delta |
for some |\theta| < 1 . We can get
\left(\sqrt[3]{n_{1}}+(-1)^{i_1}\sqrt[3]{n_2}+\cdots+(-1)^{i_{k-2}}\sqrt[3]{n_{k-1}}\right)^{3} = (-1)^{i_{k-1}+1}n_k+O\left(\Delta N_{k}^{\frac{2}{3}}\right). |
Hence for fixed n_{1}, \ldots, n_{k-1} , the number of n_k is \ll 1+\Delta N_{k}^{\frac{2}{3}} and thus
\mathcal{A}\ll \Delta N_{k}^{\frac{2}{3}}N_1\cdots N_{k-1}+N_1\cdots N_{k-1}. |
Throughout this paper, n , m and k denote natural numbers. The constants implied in the symbols \ll and O may depend on \epsilon .
By (2.1), we know that to establish the asymptotic formula in our theorem, we shall prove, for H\geq1 , that
\begin{equation} \int^{2H}_{H}\Sigma(x)^{2}dx = \frac{1}{2}\sum\limits_{n = 1}^{\infty}c_n^{2}n^{-\frac{2\rho+4}{3}}\int^{2H}_{H}x^{\frac{4\rho+2}{3}}dx +O\left(H^{\frac{4\rho+5}{3}+\epsilon}N^{-\frac{2\rho+1}{3}}\right), \end{equation} | (3.1) |
and take N = H^{\frac{6\rho+8}{(8+3\rho)(\rho+1)}} ,
\begin{equation} \int^{2H}_{H}\Sigma(x)^{3}dx = \frac{3}{4}\cos\left(\frac{\pi\rho}{2}\right)A\int^{2H}_{H}x^{2\rho+1}dx +O\left(H^{\frac{6\rho^{2}+21\rho+16}{8+3\rho}+\epsilon}\right). \end{equation} | (3.2) |
Then, since (a+b)^{3} = a^{3}+O\left(|b|a^2+|b|^3\right) , it follows from (2.1) that
\begin{equation} \begin{split} \int_{H}^{2H}\Delta_{\rho}^{3}(x;sym^{2}f)dx = &\left(2^{\rho}\pi^{\rho+1}\sqrt{3}\right)^{-3}\int^{2H}_{H}\Sigma(x)^{3}dx\\ &+O\left(H^{\frac{2\rho+2}{3}+\epsilon}N^{-\frac{\rho+1}{3}}\int^{2H}_{H}\Sigma(x)^{2}dx+H^{2\rho+3+\epsilon}N^{-\rho-1}\right)\\ = &\left(2^{\rho}\pi^{\rho+1}\sqrt{3}\right)^{-3}\frac{3A}{4}\cos\left(\frac{\pi\rho}{2}\right)\int^{2H}_{H}x^{2\rho+1}dx +O\left(H^{\frac{6\rho^{2}+21\rho+16}{8+3\rho}+\epsilon}\right) \end{split} \end{equation} | (3.3) |
by (3.1) and (3.2). Adding this for H = T/2, T/2^2, \ldots, we see that the asymptotic formula in our theorem follows.
We shall now give the details of the proof of (3.2). The proof of (3.1), which we shall omit, employs similar arguments and is simpler.
For simplicity, put
\begin{equation} r = r(n,m,k): = (nmk)^{-\frac{\rho+2}{3}}c_nc_mc_k,\ \ \ \ \ n,m,k\leq N, \end{equation} | (3.4) |
and r = 0 otherwise. Taking the third power of both sides of \Sigma(x) and for each element {\bf{i}} = (i_1, i_2)\in\mathbb{I}^{2}, \mathbb{I} = \{0, 1\} , using the elementary formula
\begin{equation*} \cos a_1 \cos a_2 \cos a_3 = \frac{1}{4}\sum\limits_{{\bf{i}}\in\mathbb{I}^{2}}\cos\left(a_1+(-1)^{i_1}a_2+(-1)^{i_2}a_3\right), \end{equation*} |
we can write
\begin{equation} \begin{split} \Sigma(x)^{3} = &\frac{3}{4}\sum rx^{2\rho+1}\cos\left(6\pi\left(\sqrt[3]{n}+\sqrt[3]{m}-\sqrt[3]{k}\right)\sqrt[3]{x}-\frac{\pi\rho}{2}\right)\\ &+\frac{1}{4}\sum rx^{2\rho+1}\cos\left(6\pi\left(\sqrt[3]{n}+\sqrt[3]{m}+\sqrt[3]{k}\right)\sqrt[3]{x}-\frac{3\pi\rho}{2}\right)\\ = &S_0(x)+S_1(x)+S_2(x), \end{split} \end{equation} | (3.5) |
where
\begin{equation} S_0(x): = \frac{3}{4}\cos\left(-\frac{\pi\rho}{2}\right)\sum\limits_{\sqrt[3]{n}+\sqrt[3]{m} = \sqrt[3]{k}} rx^{2\rho+1}, \end{equation} | (3.6) |
\begin{equation} S_1(x): = \frac{3}{4}\sum\limits_{\sqrt[3]{n}+\sqrt[3]{m}\neq\sqrt[3]{k}}rx^{2\rho+1}\cos\left(6\pi\left(\sqrt[3]{n}+\sqrt[3]{m} -\sqrt[3]{k}\right)\sqrt[3]{x}-\frac{\pi\rho}{2}\right), \end{equation} | (3.7) |
\begin{equation} S_2(x): = \frac{1}{4}\sum rx^{2\rho+1}\cos\left(6\pi\left(\sqrt[3]{n}+\sqrt[3]{m} +\sqrt[3]{k}\right)\sqrt[3]{x}-\frac{3\pi\rho}{2}\right). \end{equation} | (3.8) |
The main term in the right-hand side of (3.2) comes from S_0(x) . Indeed, integrating both sides of (3.6) with respect to x over the interval (H, 2H) , we have
\begin{equation} \int_{H}^{2H}S_0(x)dx = \frac{3}{4}\cos\left(-\frac{\pi\rho}{2}\right)\sum\limits_{\sqrt[3]{n}+\sqrt[3]{m} = \sqrt[3]{k}} r\int_{H}^{2H}x^{2\rho+1}dx. \end{equation} | (3.9) |
For natural numbers n , m and k , the condition \sqrt[3]{n}+\sqrt[3]{m} = \sqrt[3]{k} holds if and only if n , m and k all have the same cube-free part h , such that n = \alpha^3h , m = \beta^3h , k = \gamma^3h and \alpha+\beta = \gamma . Hence
\begin{equation*} \begin{split} \sum\limits_{\sqrt[3]{n}+\sqrt[3]{m} = \sqrt[3]{k}}r = \sum\limits_{h\leq N\atop h\ is\ cube-free}h^{-\rho-2}\sum\limits_{\alpha+\beta\leq\sqrt[3]{\frac{N}{h}}}\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2} c_{\alpha^3h}c_{\beta^3h}c_{(\alpha+\beta)^3h}. \end{split} \end{equation*} |
Since c_{uv}\leq c_{u}c_{v} and c_{w}\ll w^{\epsilon} , we have
\begin{equation*} \begin{split} \sum\limits_{\alpha+\beta > \sqrt[3]{\frac{N}{h}}}&\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2} c_{\alpha^3h}c_{\beta^3h}c_{(\alpha+\beta)^3h}\\ &\ll c_{h}^{3}\sum\limits_{\alpha > \frac{1}{2}\sqrt[3]{\frac{N}{h}}}\sum\limits_{\beta\leq \alpha}\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2}c_{\alpha^3}c_{\beta^3}c_{(\alpha+\beta)^3}\\ &\ll c_{h}^{3}\sum\limits_{\alpha > \frac{1}{2}\sqrt[3]{\frac{N}{h}}}\alpha^{-2\rho-4+\epsilon}\sum\limits_{\beta\leq \alpha} \beta^{-\rho-2}c_{\beta}^3\\ &\ll c_{h}^{3}\left(\frac{N}{h}\right)^{-\frac{2\rho+3}{3}+\epsilon}. \end{split} \end{equation*} |
Thus
\begin{equation*} \begin{split} \sum\limits_{\sqrt[3]{n}+\sqrt[3]{m} = \sqrt[3]{k}}r = &\sum\limits_{h\leq N\atop h\ is\ cube-free}h^{-\rho-2}\sum\limits_{\alpha,\beta = 1}^{\infty}\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2} c_{\alpha^3h}c_{\beta^3h}c_{(\alpha+\beta)^3h}\\ &+O\left(\sum\limits_{h\leq N}h^{-\rho-2}c_{h}^{3}\left(\frac{N}{h}\right)^{-\frac{2\rho+3}{3}+\epsilon}\right)\\ = &\sum\limits_{\alpha,\beta = 1}^{\infty}\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2}\sum\limits_{h = 1\atop h\ is\ cube-free}^{\infty} h^{-\rho-2}c_{\alpha^3h}c_{\beta^3h}c_{(\alpha+\beta)^3h}\\ &+O\left(\sum\limits_{\alpha,\beta = 1}^{\infty}\left(\alpha\beta(\alpha+\beta)\right)^{-\rho-2}c_{\alpha^3}c_{\beta^3} c_{(\alpha+\beta)^3}\sum\limits_{h > N}h^{-\rho-2}c_h^3\right)\\ &+O\left(N^{-\frac{2\rho+3}{3}+\epsilon}\sum\limits_{h\leq N}h^{-\frac{\rho+3}{3}}\right)\\ = &A+O(N^{-\rho-1+\epsilon}). \end{split} \end{equation*} |
Substituting this into (3.9), we have
\begin{equation} \int_{H}^{2H}S_0(x)dx = \frac{3}{4}\cos\left(\frac{\pi\rho}{2}\right)A\int_{H}^{2H}x^{2\rho+1}dx+O\left(H^{2\rho+2}N^{-\rho-1+\epsilon}\right). \end{equation} | (3.10) |
This yields the main term in (3.2) and hence that in the formula in our theorem.
We now proceed to show that the contributions from S_1(x) and S_2(x) are bounded by H^{\frac{6\rho^{2}+21\rho+16}{8+3\rho}+\epsilon} . Applying the Lemma 2.3 we find that for any real numbers p(\neq0) and q ,
\begin{equation} \begin{split} \int_{H}^{2H}x^{2\rho+1}&\cos\left(p\sqrt[3]{x}+q\right)dx\\ & = \int_{H}^{2H}3p^{-1}x^{2\rho+\frac{5}{3}}\left(\frac{p}{3x^{\frac{2}{3}}} \cos\left(p\sqrt[3]{x}+q\right)\right)dx\\ &\ll H^{2\rho+\frac{5}{3}}|p|^{-1}|\int_{u}^{v}\left(\frac{p}{3x^{\frac{2}{3}}} \cos\left(p\sqrt[3]{x}+q\right)\right)dx|\ll H^{2\rho+\frac{5}{3}}|p|^{-1}. \end{split} \end{equation} | (3.11) |
Let us examine S_2(x) first. Integrating both sides of (3.8) and then using (3.11) and (3.4), we have
\begin{equation} \begin{split} \int_{H}^{2H}S_2(x)dx&\ll \sum rH^{2\rho+\frac{5}{3}}\left(\sqrt[3]{n}+\sqrt[3]{m} +\sqrt[3]{k}\right)^{-1}\ll H^{2\rho+\frac{5}{3}}\sum\limits_{n\leq m\leq k\leq N} rk^{-\frac{1}{3}}\\ &\ll H^{2\rho+\frac{5}{3}+\epsilon}\sum\limits_{n\leq m\leq k\leq N}n^{-\frac{\rho+2}{3}}m^{-\frac{\rho+2}{3}} k^{-\frac{\rho+3}{3}}\ll H^{2\rho+\frac{5}{3}+\epsilon}N^{\frac{4}{3}-\rho}. \end{split} \end{equation} | (3.12) |
Next we consider S_1(x) . For convenience, let
\begin{equation*} \Delta = \sqrt[3]{n}+\sqrt[3]{m}-\sqrt[3]{k}. \end{equation*} |
Using the trivial bound
\begin{equation*} \int_{H}^{2H}x^{2\rho+1}\cos\left(6\pi\Delta\sqrt[3]{x}-\frac{\pi\rho}{2}\right)dx\ll H^{2\rho+2}, \end{equation*} |
where |\Delta|\leq H^{-\frac{8}{24+9\rho}} and applying (3.11) when |\Delta| > H^{-\frac{8}{24+9\rho}} , we deduce from (3.7) that
\begin{equation} \begin{split} \int_{H}^{2H}S_1(x)dx&\ll H^{2\rho+2}\sum\limits_{0 < |\Delta|\leq H^{-\frac{8}{24+9\rho}}\atop n\leq m}r+H^{2\rho+\frac{5}{3}}\sum\limits_{|\Delta| > H^{-\frac{8}{24+9\rho}}\atop n\leq m}r|\Delta|^{-1}\\ &: = H^{2\rho+2}W_1+H^{2\rho+\frac{5}{3}}W_2. \end{split} \end{equation} | (3.13) |
Now we consider first the sum W_1 . For given n\leq m\leq N there is most one natural number k for which |\Delta|\leq H^{-\frac{8}{24+9\rho}} , since \Delta = \sqrt[3]{n}+\sqrt[3]{m}-\sqrt[3]{k} and \left|\Delta\sqrt[3]{N^{2}}\right|\leq H^{-\frac{8}{24+9\rho}}\sqrt[3]{N^{2}} = o(1) . Such k , if it exists, must be greater than m . By Lemma 2.2, 0 < |\Delta|\leq H^{-\frac{8}{24+9\rho}} implies m^{-\frac{8}{3}}\ll H^{-\frac{8}{24+9\rho}} , so m\gg H^{\frac{1}{8+3\rho}} . Hence, by (3.4),
\begin{equation*} W_1\ll H^{\epsilon}\sum\limits_{n\leq m\leq N\atop m\gg H^{\frac{1}{8+3\rho}}}(nm)^{-\frac{\rho+2}{3}}m^{-\frac{\rho+2}{3}}\ll H^{\epsilon}\sum\limits_{m\gg H^{\frac{1}{8+3\rho}}}m^{-\rho-1}\ll H^{-\frac{\rho}{8+3\rho}+\epsilon}. \end{equation*} |
Similarly, in the sum W_2 , we have k = \left(\sqrt[3]{n}+\sqrt[3]{m}\right)^{3}+O\left(|\Delta|\sqrt[3]{m^{2}}\right) > m and there are \ll 1+|\Delta|\sqrt[3]{m^{2}} such k . Hence
\begin{equation*} \begin{split} W_2&\ll H^{\epsilon}\sum\limits_{n\leq m\leq N\atop |\Delta| > H^{-\frac{8}{24+9\rho}}}n^{-\frac{\rho+2}{3}}m^{-\frac{\rho+2}{3}}|\Delta|^{-1}\left(1+|\Delta|\sqrt[3]{m^{2}}\right) m^{-\frac{\rho+2}{3}}\\ &\ll H^{\epsilon}\sum\limits_{n\leq m\leq N}n^{-\frac{\rho+2}{3}}m^{-\frac{2\rho+4}{3}}H^{\frac{8}{24+9\rho}}+ H^{\epsilon}\sum\limits_{n\leq m\leq N}n^{-\frac{\rho+2}{3}}m^{-\frac{2\rho+2}{3}}\\ &\ll H^{\frac{8}{24+9\rho}+\epsilon}+H^{\epsilon}N^{\frac{2}{3}-\rho}\\ &\ll H^{\frac{8}{24+9\rho}+\epsilon}. \end{split} \end{equation*} |
Collecting these estimates we find that
\begin{equation*} \int_{H}^{2H}S_1(x)dx\ll H^{\frac{6\rho^2+21\rho+16}{8+3\rho}+\epsilon}. \end{equation*} |
Combining this with (3.10) and (3.12), we deduce from (3.2) the formula (3.5). Then according to (3.3), our theorem is proved.
Zhang is supported by the National Natural Science Foundation of China [Grant No. 11771256]. Yan is supported by the National Natural Science Foundation of China [Grant No. 11801327] and Natural Science Foundation of Shandong Province [Grant No. ZR201709280100].
The authors declare no conflict of interest.
[1] |
D. Bump, D. Ginzburg, Symmetric square L-functions on {\rm GL}(r), Ann. Math., 136 (1992), 137–205. doi: 10.2307/2946548
![]() |
[2] |
O. Fomenko, The behavior of Riesz means of the Coefficients of a symmetric square L-function, J. Math. Sci., 143 (2007), 3174–3181. doi: 10.1007/s10958-007-0201-7
![]() |
[3] | J. Hafner, On the representation of the summatory functions of a class of arithmetical functions, Springer Berlin Heidelberg, (1981), 148–165. |
[4] |
X. Han, X. Yan, D. Zhang, On fourier coefficients of the symmetric square L-function at piatetski-shapiro prime twins, Mathematics, 9 (2021), 1254. doi: 10.3390/math9111254
![]() |
[5] | J. Huang, H. Liu, Divisor problems related to Hecke eigenvalues in three dimensions, J. Math., 2021 (2021), 1–12. |
[6] |
J. Huang, H. Liu, F. Xu, Two-dimensional divisor problems related to symmetric L-functions, Symmetry, 13 (2021), 359. doi: 10.3390/sym13020359
![]() |
[7] |
H. Iwaniec, W. Luo, P. Sarnak, Low lying zeros of families of L-functions, Publ. Math. IHES, 91 (2000), 55–131. doi: 10.1007/BF02698741
![]() |
[8] |
Y. J. Jiang, G. S. Lü, X. F. Yan, Mean value theorem connected with Fourier coefficients of Hecke-Maass forms for SL(m, \mathbb{Z}), Math. Proc. Cambridge Philos. Soc., 161 (2016), 339–356. doi: 10.1017/S030500411600027X
![]() |
[9] |
H. Liu, S. Li, D. Zhang, Power moments of automorphic L-function attached to Maass forms, Int. J. Number Theory, 12 (2016), 427–443. doi: 10.1142/S1793042116500251
![]() |
[10] |
H. Lao, On comparing Hecke eigenvalues of cusp forms, Acta Math. Hungar., 160 (2020), 58–71. doi: 10.1007/s10474-019-00996-5
![]() |
[11] |
H. Lao, On the fourth moment of coefficients of symmetric square L-function, Chin. Ann. Math. Ser. B, 33 (2012), 877–888. doi: 10.1007/s11401-012-0746-8
![]() |
[12] |
H. Lao, M. McKee, Y. Ye, Asymptotics for cuspidal representations by functoriality from GL(2), J. Number Theory, 164 (2016), 323–342. doi: 10.1016/j.jnt.2016.01.008
![]() |
[13] |
H. Liu, Mean value estimates of the coefficients of product L-functions, Acta Math. Hungar., 156 (2018), 102–111. doi: 10.1007/s10474-018-0839-2
![]() |
[14] |
H. Liu, R. Zhang, Some problems involving Hecke eigenvalues, Acta Math. Hungar., 159 (2019), 287–298. doi: 10.1007/s10474-019-00913-w
![]() |
[15] |
K. Liu, H. Wang, Higher power moments of the Riesz mean error term of symmetric square L-function, J. Number Theory, 131 (2011), 2247–2261. doi: 10.1016/j.jnt.2011.05.015
![]() |
[16] |
P. Song, W. Zhai, D. Zhang, Power moments of Hecke eigenvalues for congruence group, J. Number Theory, 198 (2019), 139–158. doi: 10.1016/j.jnt.2018.10.006
![]() |
[17] | K. Tsang, Higher-power moments of \Delta(x), E(t) and P(x), Proc. London Math. Soc., 65 (1992), 65–84. |
[18] |
Y. Tanigawa, D. Zhang, W. Zhai, On the Rankin-Selberg problem: Higher power moments of the Riesz mean error term, Sci. China Ser. A, 51 (2008), 148–160. doi: 10.1007/s11425-007-0130-4
![]() |
[19] |
H. Wang, On the Riesz means of coefficients of mth symmetric power L-functions, Lith. Math. J., 50 (2010), 474–488. doi: 10.1007/s10986-010-9100-6
![]() |
[20] |
Y. Ye, D. Zhang, Zero density for automorphic L-functions, J. Number Theory, 133 (2013), 3877–3901. doi: 10.1016/j.jnt.2013.05.012
![]() |
[21] |
W. Zhai, On higher-power moments of \Delta(x) (II), Acta Arith., 114 (2004), 35–54. doi: 10.4064/aa114-1-3
![]() |
[22] |
D. Zhang, Y. Lau, Y. Wang, Remark on the paper "On products of Fourier coefficients of cusp forms", Arch. Math., 108 (2017), 263–269. doi: 10.1007/s00013-016-0996-x
![]() |
[23] |
D. Zhang, Y. Wang, Higher-power moments of Fourier coefficients of holomorphic cusp forms for the congruence subgroup \Gamma_0(N), Ramanujan J., 47 (2018), 685–700. doi: 10.1007/s11139-018-0051-6
![]() |
[24] |
D. Zhang, Y. Wang, Ternary quadratic form with prime variables attached to Fourier coefficients of primitive holomorphic cusp form, J. Number Theory, 176 (2017), 211–225. doi: 10.1016/j.jnt.2016.12.018
![]() |
[25] | D. Zhang, W. Zhai, On the distribution of Hecke eigenvalues over Piatetski-Shapiro prime twins, Acta Math. Sin. (Engl. Ser.), 2021. DOI: 10.1007/s10114-021-0174-3. |
[26] |
R. Zhang, X. Han, D. Zhang, Power moments of the Riesz mean error term of symmetric square L-function in short intervals, Symmetry, 12 (2020), 2036. doi: 10.3390/sym12122036
![]() |
1. | Yanjie Hou, Xiaofei Yan, Deyu Zhang, ON HIGHER MOMENTS OF HECKE EIGENVALUES FOR THE CONGRUENCE GROUP, 2023, 53, 0035-7596, 10.1216/rmj.2023.53.1833 |