Abstract
Let (n, q) denote the greatest common divisor of positive integers n and q, and let \(f_{r}\) denote the characteristic function of r-full numbers. We consider several asymptotic formulas for sums of the modified square-full (\(r=2\)) and cube-full numbers (\(r=3\)), which is \(\sum _{n\le y}\sum _{q\le x}\sum _{d|(n,q)}df_{r}\left( \frac{q}{d}\right) \log \frac{x}{q}\) with any positive real numbers x and y. Moreover, we derive the asymptotic formula of the above with \(r=2\) under the Riemann Hypothesis.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let \(s=\sigma +it\) be the complex variable, and let \(\zeta (s)\) be the Riemann zeta-function. Let \(r (\ge 2)\) be an integer, we call n an r-full or r-free integer if \(p|n \Rightarrow p^{r}|n\) or \(p|n \Rightarrow p^{r}\not \mid n\), respectively. In the special case when \(r=2\) or 3 integer. We call n a square-full or cube-full numbers, respectively. Let G(r) denote the set of r-full numbers, and let (n, q) denote the greatest common divisor of positive integers n and q. Define
and
It is worth mentioning that the above sum is an analogue of the Ramanujan sum \( c_{q}(n)=\sum _{d|(n,q)}d\mu \left( {q}/{d}\right) , \) with \(\mu \) being the Möbius function. For the case \(r=2\) and \(r=3\), the Dirichlet series of the function \(s_{q}^{(r)}(n)\) is given by
for \(\textrm{Re}~s >\frac{1}{2}\), and
for \(\textrm{Re}~s >\frac{1}{3}\). Here \( \sigma _{1-s}(n)=\sum _{d|n}d^{1-s}, \) and the function \(\kappa _{9}(s)\) is absolutely convergent for \(\textrm{Re}~s >\frac{1}{9}\), that is
(see (1.96), (1.97) and (1.98) in [5]). For any large positive real numbers x and y, and any non-negative integer k, we are interested by studying the double sums
In this paper, we shall consider the asymptotic formulas for \(S_{k}^{(r)}(x,y)\) when \(r=2,3\). In the case \(k=0\) and \(r=2\), the author [6] used the method of Chan and Kumchev [1] (see also [9, 11])Footnote 1 and the theory of exponent pairs (see [3, 5])) to deduce the asymptotic formula to \(S_{0}^{(2)}(x,y)\). It is shown that
holds, where x and y are large real numbers such that \(x \ll y \ll x^{\frac{3}{2}}\).
Recently, the author [7] gave a more precise asymptotic for \(S_{0}^{(2)}(x,y)\) by using Lemma 2.2 below and some properties of the Riemann zeta-function. He proved that
holds, where x and y are large real numbers such that \(x^{\frac{4}{3}}\log x \ll y \ll \frac{x^{\frac{14}{9}}}{\log ^{4}x}\). Moreover, for \(k=0\) and \(r=3\), the author [8] showed that
holds, where x and y denote large real numbers such that \(x \ll y \ll x^{\frac{5}{3}}\). From the above, we notice that it is difficult to improve the error because the term \(O\left( x^{\frac{1}{3}}y\right) \) is absorbed into all error terms. For this reason, in this paper, we consider asymptotic formulas for \(S_{1}^{(r)}(x,y)\) and give the interesting relation between \(S_{0}^{(r)}(x,y)\) and \(S_{1}^{(r)}(x,y)\) for \(r=2,3\). It is the most interesting problem for us to derive asymptotic formulas of (1.4) when \(k=0,1\), and by a similar argument, we may prove that any cases \(k(\ge 2)\). Before going into the statements of our theorems, we denote the Fourier integrals \(\nu (u)\) and \(\xi (u)\) defined by
and
with \(u=\log \frac{x}{y}\), respectively. It follows from (3.6) and (5.5) below that
and
hold. Here the integrals are computable constants, and, strictly speaking, that is enough for the purpose of this paper. Then we have the following results:
Theorem 1
Let the notation be as above. Let x and y be large real numbers such that \(x\log ^{3}x \ll y \ll x^{\frac{14}{9}} \). Then we have
where the function \(\nu (u)\) is given by (1.8) and the error term \(E_{1}^{(2)}(x,y)\) is estimated by
with C being a positive constant.
Remark 1.1
Using (1.5) and (1.10) we deduce the relation
for \(x\log ^{3}x \ll y \ll x^{\frac{3}{2}}\). It follows from (1.10) and (1.11) that
holds. This means that the logarithmic average order of \(s_{q}^{(2)}(n)\) is \( \frac{\zeta (2)\zeta (3)}{\zeta (6)} \) where q and n satisfying the condition \( q\log ^{3}q \ll n \ll q^{\frac{14}{9}}. \)
In fact, it is suspected that there is a deep relationship between a zero-free region of the Riemann zeta-function and the order of magnitude of the error term (1.11). Then we immediately obtain
Conjecture 1
We may conjecture that
holds with an absolute constant \(C>0\).
Theorem 2
Let the notation be as above. Let x and y be large real numbers such that \(x^{\frac{6}{5}}\log ^{3}x \ll y \ll x^{\frac{19}{12}} \). Then we have
where the function \(\xi (u)\) is given by (1.9) and C is a positive constant.
Remark 1.2
Similarly as in Remark 1.1, we have
for \(x^{\frac{6}{5}}\log ^{3}x \ll y \ll x^{\frac{5}{3}}\), and the logarithmic average order of \(s_{q}^{(3)}(n)\) is derived by \( \frac{\zeta (3)\zeta (4)\zeta (5)\kappa _{9}(1)}{\zeta (8)} \) under q and n satisfying the condition \( q^{\frac{6}{5}}\log ^{3}q \ll n \ll q^{\frac{19}{12}}. \)
Next, we assume the truth of the unproved Riemann Hypothesis, that all the complex zeros of the Riemann zeta-function \(\zeta (s)\) lie on the line \(\sigma =\frac{1}{2}\). We consider the precise asymptotic formula concerning \({S}_{1}^{(2)}(x,y)\). Then we derive the following
Theorem 3
Assume that the Riemann Hypothesis is true. Let x and y be large real numbers such that \(x\log ^{3}x \ll y \ll x^{\frac{29}{18}}\textrm{exp}\left( -A\frac{\log x}{\log \log x}\right) \). Then the error term \(E_{1}^{(2)}(x,y)\) of (1.10) is estimated by
with A being a positive constant.
In addition, we assume that all the zeros \(\rho \) of the Riemann zeta-function \(\zeta (s)\) on the critical line are simple, where \(\rho =\frac{1}{2}+i\gamma \) denotes a nontrivial zero of the Riemann zeta-function, and \(\gamma \) denotes the imaginary part of zero on the critical line. Then we may derive a sum involving the zeros \(\rho \) of \(\zeta (s)\) concerning \(E_{1}^{(2)}(x,y)\). To improve the order of magnitude of its sum, we make use of the Gonek-Hejhal Hypothesis (Gonek [2], and Hejhal [4] independently conjectured), namely
for real number \(\lambda < \frac{3}{2}\), where \(\zeta '(s)\) is the first derivative of \(\zeta (s)\), then we may deduce a new estimate of \(E_{1}^{(2)}(x,y)\), which will be done elsewhere.
Notations. Throughout this paper, we use the following notations: The Riemann zeta-function \(\zeta (s)\), defined by \( \sum _{n=1}^{\infty }\frac{1}{n^s} \) for \(\sigma >1\), admits of analytic continuation over the whole complex plane having as its only singularity a simple pole with residue 1 at \(s=1\). In what follows, C donotes any arbitrarily positive number, not necessarily the same ones at each occurrence.
2 Some Lemmas
Lemma 2.1
Suppose that the Dirichlet series \( \alpha (s):=\sum _{n=1}^{\infty }\frac{a_n}{n^s} \) converges for \(\textrm{Re}~s >\sigma _{c}\). If \(\sigma _{0}>\max (0,\sigma _{c})\) and \(x>1\), then
Proof
This is Riesz typical means of Perron’s formula. For more details, see (5.20)–(5.22) in [10]. \(\square \)
Lemma 2.2
Let \(\textrm{Re}~z \le 0\), and let \( \sigma _{z,b}(n) \) denote the generalization of the divisor function defined by \( \sigma _{z,b}(n) =\sum _{d^{b}|n}d^{bz}. \) Then we have
where \(\sum {}^{'}\) indicates that the last term is to be halved if x is an integer, and
uniformly for \(b\ge 1\) and \(D_{z,b}(x)\) is given by the following
-
(i)
If \(b=1,2\) and \(-\frac{2}{3b^{2}}<\textrm{Re}~z \le 0\), then
$$\begin{aligned} D_{z,b}(x) = \zeta (b(1-z))x +\frac{1}{1+bz}\zeta \left( z+\frac{1}{b}\right) x^{z+\frac{1}{b}}. \end{aligned}$$ -
(ii)
If \(b\ge 3\) and \(-1<\textrm{Re}~z \le 0\), then
$$\begin{aligned} D_{z,b}(x) = \zeta (b(1-z))x. \end{aligned}$$
Proof
The proof of this result can be found in [Theorem 1.4, [11]]. \(\square \)
Lemma 2.3
There is an absolute constant \(C>0\) such that \(\zeta (s)\ne 0\) for
Proof
This lemma is given by Theorem 6.1 in [5]. \(\square \)
Lemma 2.4
For \(|t|\ge 2\) and \(\sigma \ge 1-C(\log t)^{-\frac{2}{3}}(\log \log t)^{-\frac{1}{3}}\) we have
Proof
The first term of this lemma is a well-known result. The second term of this lemma is given by Lemma 12.3 in [5]. \(\square \)
Lemma 2.5
For \(t\ge t_{0} >0\) uniformly in \(\sigma \), we have
Proof
The proof of this lemma follows from Theorem II.3.8 in [12] (see also [5, 13]). \(\square \)
Lemma 2.6
For any positive number \(T>1\) we have
Proof
The second and third terms of (2.1) are due to Theorem 5.1 and Theorem 8.5 in [5]. We use (2.2) below and the formula \( \int _{1}^{T}|\zeta (\sigma +it)|^{4}dt=O(T) \) for \(\frac{1}{2}<\sigma \le 1\) to deduce (2.1). \(\square \)
Lemma 2.7
Assume that the Riemann hypothesis is true. Then we have
for every \(\sigma \)\((\frac{1}{2}+\delta \le \sigma \le 2)\) and \(t\ge t_{0}\) being a sufficiently large real number.
Proof
The first and second terms of this lemma are given by (14.2.5), (14.2.6), (14.14.1) and (14.16.2) in [13], respectively. \(\square \)
The next lemma is a well-known result (see [5, 13]), that is
Lemma 2.8
The functional equation of the Riemann zeta-function is given by
where \( \chi (s)=2^{s}\pi ^{s-1}\sin \left( \frac{\pi s}{2}\right) \Gamma (1-s). \) Thus in any bounded vertical strip, we have
3 Proof of Theorem 1
We assume that \(1 \le y \le x^M\) for some constant M. Without loss of generality we can assume that \(x,\ y\in \mathbb {Z}+\frac{1}{2}\). We apply Lemma 2.1 with (1.2) to get
where \(\alpha = 1 +\frac{1}{\log x}\). Let T be a real parameter at our disposal. We have
Taking \(b=1\) and \(z=1-s\) into Lemma 2.2 and using the estimate \( \sum _{n\le y}\sigma _{0}(n)\ll y\log y \) we have
where
and
Define
with C being the same as that in Lemma 2.3 and \( T=x^{2}. \) Let \(\Gamma (\alpha , \beta ,T)\) denote the contour consisting of the line segments \([\alpha -iT, \beta -iT]\), \([\beta -iT, \beta +iT]\) and \([\beta +iT, \alpha +iT]\).
In \(K_{1}\), we move the integration, with respect to s, to \(\Gamma (\alpha ,\beta ,T)\) with \(\beta =\frac{1}{6}-\varepsilon (T)\). We denote the integrals over the horizontal line segments by \(K_{1,1}\) and \(K_{1,3}\), and the integral over the vertical line segment by \(K_{1,2}\), respectively. We use Lemmas 2.3–2.5 to deduce
and
It remains to evaluate the residues of the poles of the integrand, and there exist three simple poles at \(s=1\), \(\frac{1}{2}\) and \(\frac{1}{3}\) with residues \( \frac{\zeta (2)\zeta (3)}{\zeta (6)}x, \) \( \frac{2\zeta (\frac{1}{2})\zeta (\frac{3}{2})}{\zeta (3)}x^{\frac{1}{2}}, \) and \( \frac{3\zeta (\frac{1}{3})\zeta (\frac{2}{3})}{\zeta (2)}x^{\frac{1}{3}}, \) respectively. Therefore, we have
by setting \(T=x^2\), where C is a positive constant.
In \(K_{2}\), we move the integration, with respect to s, to \(\Gamma (\alpha ,\frac{5}{2},T)\). We denote the integrals over the horizontal line segments by \(K_{2,1}\) and \(K_{2,3}\), and the integral over the vertical line segment by \(K_{2,2}\), respectively. Using Lemmas 2.5 and 2.8 we have
and
where \(\nu (u)\) is given by
with \(u=\log \frac{x}{y}\). We use Lemma 2.5 and the inequality \( |\frac{1}{\zeta (s)}|\le \frac{\zeta (\sigma )}{\zeta (2\sigma )}\) for \( \sigma >1 \) to obtain the absolute value of \(\nu (u)\), that is
hence \(|\nu (u)|\) is an upper bound. Next, there exists a simple pole at \(s=2\) of the integral \(K_2\) with residue \( \frac{\zeta (4)\zeta (6)}{8\zeta (12)}\left( \frac{x}{y}\right) ^{2} \) by using the value \( \zeta (0)=-\frac{1}{2}. \) Hence, we have
with \(T=x^2\).
Combining (3.3) and (3.7) with (3.1) and taking \(T=x^2\), we obtain the formula (1.10).
4 Proof of Theorem 3
In this section we assume that the Riemann Hypothesis is true, and \(1 \le y \le x^M\) for some constant M. Without loss of generality we can assume that \(x,\ y\in \mathbb {Z}+\frac{1}{2}\). The proof of this theorem follows by the same method as in Theorem 1, in addition to the Riemann Hypothesis. We start with (3.1), and set \(\beta =\frac{1}{12}+\delta \) with \(\delta =\frac{10}{\log \log T}\), and \(\varepsilon =\frac{1}{\log \log T}\) in Lemma 2.7 and \(T=x^2\).
In \(K_{1}\), we move the integration, with respect to s, to \(\Gamma (\alpha ,\beta ,T)\) with \(\beta =\frac{1}{12}+\delta \). We denote the integrals over the horizontal line segments by \(K_{1,1}\) and \(K_{1,3}\), and the integral over the vertical line segment by \(K_{1,2}\), respectively. We use Lemmas 2.5, 2.7, and 2.8 to deduce
with A being a positive constant, and \(T=x^2\). We use Lemmas 2.5, 2.7, 2.8, and the estimates of \(K_{1,1}\) and \(K_{1,3}\) in the proof of Theorem 1 to deduce
Therefore, by using Cauchy’s residue theorem we have
by setting \(T=x^2\), where A is a positive constant.
As for \(K_{2}\), we make use of the same result in the proof of Theorem 1, that is
with \(T=x^2\).
Combining (4.1) and (4.2) with (3.1) and taking \(T=x^2\), we obtain the formula (1.13).
5 Proof of Theorem 2
Assume that \(1 \le y \le x^M\) for some constant M. Without loss of generality we can assume that \(x,\ y\in \mathbb {Z}+\frac{1}{2}\). The proof of this theorem follows by the same method as in Theorem 1. We apply Lemma 2.1 with (1.3) and substitute \(b=1\) and \(z=1-s\) into Lemma 2.2 to deduce
where
and
Here \(\alpha = 1 +\frac{1}{\log x}\) and T is a real parameter at our disposal.
We move the integration, with respect to s, to \(\Gamma (\alpha ,\beta ,T)\) with \(\beta =\frac{1}{8}-\varepsilon (T)\), where \(\varepsilon (T)\) is given by (3.2). We denote the integrals over the horizontal line segments by \(L_{1,1}\) and \(L_{1,3}\), and the integral over the vertical line segment by \(L_{1,2}\), respectively. Since the estimate \(\kappa _{9}(\sigma +iT)\ll 1\) for \(\sigma \ge \beta \), it follows from Lemmas 2.3–2.5 that
For \(L_{1,2}\), we use integration by parts, Cauchy–Schwarz’s inequality twice, Lemmas 2.3–2.6, and the estimate \(\kappa _{9}(\frac{1}{8}+it)\ll 1\) to deduce
It remains to evaluate the residues of the poles of the integrand, and there exist four simple poles at \(s=1\), \(\frac{1}{3}\), \(\frac{1}{4}\) and \(\frac{1}{5}\) with residues \( \frac{\zeta (3)\zeta (4)\zeta (5)\kappa _{9}(1)}{\zeta (8)}x, \) \( \frac{3\zeta (\frac{1}{3})\zeta (\frac{4}{3})\zeta (\frac{5}{3})\kappa _{9}\left( \frac{1}{3}\right) }{\zeta (\frac{8}{3})}x^{\frac{1}{3}}, \) \( \frac{4\zeta (\frac{1}{4})\zeta (\frac{3}{4})\zeta (\frac{5}{4})\kappa _{9}\left( \frac{1}{4}\right) }{\zeta (2)}x^{\frac{1}{4}}, \) and \( \frac{5\zeta (\frac{1}{5})\zeta (\frac{3}{5})\zeta (\frac{4}{5})\kappa _{9}\left( \frac{1}{5}\right) }{\zeta (\frac{8}{5})}x^{\frac{1}{5}}, \) respectively. Therefore, we have
by setting \(T=x^2\), with C being a positive constant.
For \(L_{2}\), we move the integration, with respect to s, to \(\Gamma (\alpha ,\frac{5}{2},T)\). We denote the integrals over the horizontal line segments by \(L_{2,1}\) and \(L_{2,3}\), and the integral over the vertical line segment by \(L_{2,2}\), respectively. Following the same method as in (3.4) and (3.5) we have
and
where \(\xi (u)\) is given by
with \(u=\log \frac{x}{y}\). Similarly as in (3.6), we have
There exists a simple pole at \(s=2\) of the integral \(L_2\) with residue \( \frac{\zeta (6)\zeta (8)\zeta (10)\kappa _{9}(2)}{8\zeta (16)}\left( \frac{x}{y}\right) ^2. \) Hence, we have
by using \(T=x^2\).
Combining (5.2) and (5.6) with (5.1) and taking \(T=x^2\), the formula (1.12) is proved.
References
Chan, T.H., Kumchev, A.V.: On sums of Ramanujan sums. Acta Arith. 152(1), 1–10 (2012)
Gonek, S.M.: On negative moments of the Riemann zeta-function. Mathematika 36, 71–88 (1989)
Graham, S.W., Kolesnik, G.P.: Van der Corput’s Method of Exponential Sums. London Mathematical Society Lecture Note Series, vol. 126. Cambridge University Press, Cambridge (1991)
Hejhal, D.P.: On the distribution of \(\log {\left|\zeta ^{\prime }(\frac{1}{2}+it)\right|}\). In: Aubert, K.E., Bombieri, E., Goldfeld, D. (eds.) Number Theory, Trace Formula and Discrete Groups, pp. 343–370. Academic Press, San Diego (1989)
Ivić, A.: The Riemann Zeta-Function. Dover Publications, New York (2003)
Kiuchi, I.: On sums of sums involving squarefull numbers. Acta Arith. 200(2), 197–211 (2021)
Kiuchi, I.: On a sum involving squarefull numbers. Rocky Mt. J. Math. 52, 1713–1718 (2022)
Kiuchi, I.: On sums of sums involving cube-full numbers. Ramanujan J. 59, 279–296 (2022)
Kühn, P., Robles, N.: Explicit formulas of a generalized Ramanujan sum. Int. J. Number Theory 12, 383–408 (2016)
Montgomery, H.L., Vaughan, R.C.: Multiplicative Number Theory I. Cambridge Studies in Advanced Mathematics, Classical Theory. Cambridge University Press, Cambridge (2007)
Robles, N., Roy, A.: Moments of averages of generalized Ramanujan sums. Monatsh. Math. 182, 433–461 (2017)
Tenenbaum, G.P.: Introduction to Analytic and Probabilistic Number Theory. Garduate Studies, vol. 163. AMS, Providence, RI (2008)
Titchmarsh, E.C.: The Theory of the Riemann Zeta-Function, 2nd edn. Oxford University Press, Oxford (1986)
Acknowledgements
The author would like to thank the referee for his/her careful reading of the earlier version of this paper, giving him many valuable suggestions. The author is supported by JSPS Grant-in-Aid for Scientific Research(C)(21K03205).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Kiuchi, I. Sums of logarithmic weights involving r-full numbers. Ramanujan J 64, 1045–1059 (2024). https://doi.org/10.1007/s11139-024-00891-w
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11139-024-00891-w
Keywords
- Square full numbers
- Cube full numbers
- Riemann zeta-function
- Divisor function
- Riemann hypothesis
- Asymptotic results on arithmetical functions