On the series expansion of k-free Dirichlet series and its analytical continuation
Artur Kawalec · 2026
Read On the series expansion of k-free Dirichlet series and its analytical continuation by Artur Kawalec (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-2603-22775.txt.
Introduction
Kawalec develops series expansions for k-free Dirichlet series. He expands the k-free zeta series as a Laurent series with a simple pole. For the regular part, he proves a Stieltjes-like formula for its coefficients. He also studies a second analytic continuation of these series. That route yields a formula for the zeta value at one over k. The formula uses the k-free indicator function and holds for integers k of at least two.
Catalog record
| CREATOR | Artur Kawalec |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20299 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.22775
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On the series expansion of k-free Dirichlet series and its analytical continuation Artur Kawalec Abstract In this article, we develop a k-free zeta Dirichlet series into a Laurent series with a simple pole, and prove a Stieltjes like formula for the expansion coefficients of the regular part. We also investigate another analytical continuation of these series and develop a formula for ζ ( 1 k ) \zeta(\tfrac{1}{k}) for positive integer k ≥ 2 k\geq 2 in terms of the k-free indicator function. 1 Introduction In the previous article [2], we developed several results concerning the square-free Dirichlet series, such as its Laurent series expansion and analytical continuation, and also a formula for \zeta(\frac{1}{2}) represented by a certain Dirichlet series with |\mu(n)| coefficients. In this article, we generalize the same results but for the k-free case for positive integer k\geq 2 . We have the Dirichlet series \frac{\zeta(s)}{\zeta(ks)}=\sum_{n=1}^{\infty}\frac{\mu^{(k)}(n)}{n^{s}} (1) which is absolutely convergent for \Re(s)>1 . Let \mu(n) be the standard Möbius function. We recall that an integer n is k-free (for k\geq 2 ) if it is not divisible by any k^{th} power equal to k [3][5][6]. We write an indicator function \mu^{(k)}(n)=\begin{cases}1&\text{if n is k-free}\\ 0&\text{otherwise}\end{cases} (2) to distinguish between the k-free numbers. The k-free indicator function can be defined in terms of the Möbius function by the relation \mu^{(k)}(n)=\sum_{d^{k}|n}\mu(d) (3) For the square-free integer case k=2 one has \mu^{(2)}(n)=\mu^{2}(n)=|\mu(n)| . And for k=3 case \mu^{(3)}(n) is referred to as cube-free and so forth. We define the Laurent series expansion \frac{\zeta(s)}{\zeta(ks)}=\frac{1}{\zeta(k)(s-1)}+\sum_{n=0}^{\infty}\gamma_{n}^{M,k}\frac{(s-1)^{n}}{n!} (4) about the pole at s=1 and residue \frac{1}{\zeta(k)} . This series is analytic on a disk centered at s=1 and radius of convergence R=1+\frac{2}{k} due being limited by the next closest to the pole at the first trivial zero when ks=-2 for s=-\frac{2}{k} from the center at s=1 . The main result of this article is we derive a general log-limit formula for these expansion coefficients as Theorem 1 . For n\geq 0 , the coefficients are: \gamma^{M,k}_{n}=\lim_{x\to\infty}(-1)^{n}\Bigg\{\sum_{j=1}^{x}\frac{\mu^{(k)}(j)\log^{n}(j)}{j}-\frac{\log^{n+1}(x)}{\zeta(k)(n+1)}\Bigg\} (5) The limit formula (5) was discovered by Wolf in [7] for k=2 case, and as we showed in [2]. In this article, we extend the same Wolf formula for the general k-free k\geq 2 case. We now consider the case for n=0 , where the growth of this Dirichlet series \sum_{n\leq x}\frac{\mu^{(k)}(n)}{n}=O(\log x) (6) gives a crude log bound; the reason because it’s always less than the harmonic series \sum_{n\leq x}\frac{\mu^{(k)}(n)}{n}<\sum_{n\leq x}\frac{1}{n}=O(\log x). (7) The harmonic series also has a sharper growth estimate \sum_{n\leq x}\frac{1}{n}=\gamma+\log x+O(\frac{1}{x}) (8) where is \gamma is the Euler-Mascheroni constant, and the extraction of this constant is by the limit \displaystyle\gamma \displaystyle=\lim_{x\to\infty}\left(\sum_{n\leq x}\frac{1}{n}-\log x\right) (9) \displaystyle=0.57721566490153286060\ldots. In [7] Wolf considered a similar square-free analogue case \sum_{n\leq x}\frac{|\mu(n)|}{n}=\gamma^{M}+\frac{6}{\pi^{2}}\log x+O(\frac{1}{x}) (10) where the new constant \displaystyle\gamma^{M} \displaystyle=\lim_{x\to\infty}\left(\sum_{n\leq x}\frac{|\mu(n)|}{n}-\frac{6}{\pi^{2}}\log x\right) (11) \displaystyle=1.04389451571193829740\ldots is an analogue of the Euler’s constant, where it is associated with k=2 square-free case. We now extend these results to the general k-free case and write an estimate for the k-free harmonic sum \sum_{n\leq x}^{x}\frac{\mu^{(k)}(n)}{n}=\gamma^{M,k}+\frac{1}{\zeta(k)}\log x+O(\frac{1}{x}) (12) where the generalized k-free analogue of Euler constant \gamma^{M,k} is extracted the same way \gamma^{M,k}=\lim_{x\to\infty}\left(\sum_{n=1}^{x}\frac{\mu^{(k)}(n)}{n}-\frac{1}{\zeta(k)}\log x\right)\\ (13) For n=0 case only, we assume the notation \gamma^{M,k}=\gamma^{M,k}_{0} . In Table 1, we compute the first 10 constants for k=2 to 11 and observe that its value is approaching the Euler’s constant as k\to\infty , hence one has \lim_{k\to\infty}\gamma^{M,k}=\gamma (14) the reason being is that the k-free Dirichlet series approaches the harmonic series \lim_{k\to\infty}\sum_{n\leq x}\frac{\mu^{(k)}(n)}{n}\to\sum_{n\leq x}\frac{1}{n} (15) as k\to\infty , where only the higher order k divisors are missing, and one also has \frac{1}{\zeta(k)}\to 1 . Table 1: A high-precision computation k-free analogue constant k \gamma^{M,k} 2 1.043894515711938297404563438509 3 0.891541580568652696893988105635 4 0.768619483359480556610282699590 5 0.689533694428236052333652469748 6 0.641881881833269966325319431727 7 0.613974332885360734840225581982 8 0.597899163505840636385864829837 9 0.588751517287173610371480648659 10 0.583598636665716799342777885260 11 0.580722048488931420177890261674 We next consider an alternative representation for these coefficients. If we take the Laurent series expansion of zeta about s=1 of \zeta(s)=\frac{1}{s-1}+\gamma+O(|s-1|) (16) to the 0^{th} order, and another expansion of \frac{1}{\zeta(ks)}=\frac{1}{\zeta(k)}-k\frac{\zeta^{\prime}(k)}{\zeta(k)^{2}}(s-1)+O(|s-1|^{2}) (17) to the 1^{st} order, then by multiplying them together yields \frac{\zeta(s)}{\zeta(ks)}=\frac{1}{\zeta(k)(s-1)}+\frac{\gamma}{\zeta(k)}-k\frac{\zeta^{\prime}(k)}{\zeta(k)^{2}}+O(|s-1|). (18) As a result, the 0^{th} order coefficient in (18) is collected in the limit s\to 1 as \gamma^{M,k}=\frac{\gamma}{\zeta(k)}-k\frac{\zeta^{\prime}(k)}{\zeta(k)^{2}} (19) gives a better closed-form formula in terms of other known constants. And the value for the zeta derivative \zeta^{\prime}(k) can be computed by \zeta^{\prime}(k)=-\sum_{n=1}^{\infty}\frac{\log(n)}{n^{k}}. (20) In Tables 2-5, we compute the first 10 higher order \gamma^{M,k}_{n} coefficients to high-precision ( 30 decimal places) by the n^{th} differentiation of the (rhs) of (4) by subtracting the pole term in the limit as \gamma^{M}_{n,k}=\frac{d^{n}}{ds^{n}}\Bigg[\frac{\zeta(s)}{\zeta(ks)}-\frac{1}{\zeta(k)(s-1)}\Bigg]_{s\to 1} (21) which is easily possible to do in software packages such as Mathematica or Pari/GP [4]. Table 2: n \gamma_{n}^{M,2} 0 1.043894515711938297404563438509 1 -0.236152886477122974860578286060 2 0.319384120408014249249465207074 3 -0.501294458741649566645935631332 4 1.010739722784850417039579626049 5 -2.544030257932552280334481508980 6 7.666100995112318690725728704276 7 -26.88797470534219199661349019865 8 107.6566910334506652692812639473 9 -484.6934692784684121614213582581 10 2424.080089640181055133479838894 Table 3: n \gamma_{n}^{M,3} 0 0.891541580568652696893988105635 1 -0.245232425193549088584910660338 2 0.478859132334302048001657659233 3 -1.001749939953870715953016999262 4 2.605658110908472598918158429694 5 -8.280181525858129142262381587973 6 31.06687418066902031238934836818 7 -134.2881733346003220009976922544 8 657.6022908453450809684814267259 9 -3601.372879333978339048317645190 10 21824.75413023318146861252990975 Table 4: n \gamma_{n}^{M,4} 0 0.768619483359480556610282699590 1 -0.181272887555709074717349472246 2 0.482385593391270339523799620921 3 -1.217994351127279734636351368008 4 3.711024193639538830737880019661 5 -13.61003278671187974663949651367 6 58.48358113969641160388773799724 7 -287.8121812713812744669879624051 8 1596.153337468935239880936537014 9 -9857.607981656058667385016103321 10 67147.71869990596717226860751284 Table 5: n \gamma_{n}^{M,5} 0 0.689533694428236052333652469748 1 -0.110662327010032813709101796076 2 0.415273869629961542853619205015 3 -1.216297879807816360207457417664 4 4.165641604898467635390667588548 5 -16.90556080242211587404036520880 6 79.86174202644288489695500264961 7 -430.5047021058354385867335267643 8 2605.890311019817862275989909315 9 -17505.48225950875316901409735157 10 129339.3988398803463345898998687 2 Proof of Theorem 1 To show that, let us recap some basic definitions. The k-free counting function is defined by Q_{k}(x)=\sum_{n\leq x}\mu^{(k)}(n) (22) for positive integer argument x>1 , and by the average value Q_{k}(x)=\frac{1}{2}\left[\sum_{n<x}\mu^{(k)}(n)+\sum_{n\leq x}\mu^{(k)}(n)\right] (23) for positive real argument x>0 , so whenever x is integer an average value of left and right side of the step is taken. One has the asymptotic Q_{k}(x)\sim\frac{x}{\zeta(k)} (24) and the limit \lim_{x\to\infty}\frac{Q_{k}(x)}{x}=\frac{1}{\zeta(k)} (25) which implies that the probability of choosing a k-free integer at random from a set of integers is \frac{1}{\zeta(k)} . The k-free counting function can be expressed as Q_{k}(x)=\frac{x}{\zeta(k)}+f_{k}(x) (26) where the bound f_{k}(x)=O(x^{\frac{1}{k}}) (27) which it suffices to prove by the prime number theorem. But if assuming (RH), there is a much better bound f_{k}(x)=O(x^{\frac{1}{2k}}) (28) as shown in [3]. Also, the m^{th} derivatives of (1) are given by (-1)^{m}\left[\frac{\zeta(s)}{\zeta(ks)}\right]^{(m)}=\sum_{n=1}^{\infty}\frac{\mu^{(k)}(n)\log^{m}(n)}{n^{s}} (29) for \Re(s)>1 . So we proceed proving Theorem 1 by carrying out appropriate Stieljtes integration. For m\geq 0 case, we have: \displaystyle\sum_{n\leq x}\frac{\mu^{(k)}(n)\log^{m}(n)}{n} \displaystyle=\int_{1^{-}}^{x}\frac{\log^{m}(t)}{t}dQ_{k}(t)=\int_{1^{-}}^{x}\frac{\log^{m}(t)}{t}d\left(\frac{t}{\zeta(k)}\right)+\int_{1^{-}}^{x}\frac{\log^{m}(t)}{t}df_{k}(t) \displaystyle=\frac{1}{\zeta(k)}\int_{1^{-}}^{x}\frac{\log^{m}(t)}{t}\,dt+\left[\frac{\log^{m}(t)}{t}f_{k}(t)\right]_{1^{-}}^{x}-\int_{1^{-}}^{x}\frac{d}{dt}\left[\frac{\log^{m}(t)}{t}\right]f_{k}(t)\,dt \displaystyle=A(x)+\frac{\log^{m}(x)}{x}f_{k}(x)+B_{m}+\int_{1^{-}}^{x}\frac{\log^{m}(t)-m\log^{m-1}(t)}{t^{2}}f_{k}(t)\,dt (30) Here we note there is a subtle detail concerning the lower integration variable where we must consider the value Q_{k}(1^{-})=0 (31) before the step, since Q_{k}(1^{+})=1 after the step. As a result, the value f_{k}(1^{-})=-\frac{1}{\zeta(k)} (32) is the right solution. We define a constant B_{m} by B_{m}=\begin{cases}\frac{1}{\zeta(k)}&\text{if }m=0\\ 0&\text{if }m>0\end{cases} (33) which only arises for m=0 case. The reason is that it is assumed the convention that 0^{0}=1 , which implies [\log(1)]^{0}=1 , thus yielding a nonzero term, otherwise for m>1 , we have [\log(1)]^{m}=0^{m}=0 . And this integral identity can be generated as \displaystyle A(x) \displaystyle=\frac{1}{\zeta(k)}\int_{1}^{x}\frac{\log^{m}(t)}{t}\,dt (34) \displaystyle=\frac{\log^{m+1}(x)}{\zeta(k)(m+1)} for m\geq 1 . Then, it is readily seen that the limit \sum_{n\leq x}\frac{\mu^{(k)}(n)\log^{m}(n)}{n}-\frac{\log^{m+1}(x)}{\zeta(k)(m+1)}=(-1)^{m}\gamma_{k,m}^{M}+O\left(x^{\tfrac{1}{2k}-1}\log^{m}(x)\right) (35) exists as x\to\infty and converges to a constant (-1)^{m}\gamma_{k,m}^{M}=B_{m}+\int_{1^{-}}^{\infty}\frac{\log^{m}(t)-m\log^{m-1}(t)}{t^{2}}f_{k}(t)\,dt (36) for m\geq 0 by the estimate (27) or (28) for k\geq 2 . Now the next step is to connect this constant to the series expansion in (4) as