0HPRLUVRIWKH2VDND,QVWLWXWH RI7HFKQRORJ\6HULHV$ 9RO1RáâSSØ
Dirichlet Series Associated with Hyperharmonic Numbers
by
Ken KAMANO
1Department of General Education, Faculty of Engineering
(Manuscript received Sep 30, 2011)
Abstract
We investigate a complex variable function hr(s) defined as a Dirichlet series associated with hyper-harmonic numbers. This function is a specialization of the non-strict multiple zeta function and it can be meromorphically continued to the whole complex plane. We represent the function hr(s) in terms of h1(s) and the Riemann zeta function, and this result gives information of poles of hr(s).
keywords; Dirichlet series, harmonic numbers, multiple zeta function
1partially supported by The Sumitomo Foundation (100601).
,
Ken KAMANO
1. Introduction and main results Harmonic numbers Hn(n ≥ 0) are rational numbers defined by
Hn= 1 + 1 2+ 1 3+· · · + 1 n (n ≥ 1)
and H0= 0. These numbers appear in various areas in mathematics and have been classically investigated. Harmonic numbers have been generalized in many directions. For example, Conway and Guy [4, p. 258] defined hyperharmonic numbers Hn(r) for integers r ≥ 1 by the following recurrence relations:
Hn(1)= Hn, Hn(r)= n i=1 Hi(r−1) (r ≥ 2).
Benjamin-Gaebler-Gaebler [3] investigated these numbers from a combinatorial point of view. They proved that Hn(r) are expressed in terms of generalized Stirling numbers, called r-Stirling numbers. As stated in [4], hyperharmonic numbers can be expressed in terms of the ordinary harmonic numbers:
(1.1) Hn(r)= n + r − 1 r − 1 (Hn+r−1− Hr−1)
for r ≥ 1. Recently this identity was generalized by Mez˝o-Dil [9, Theorem 1]. They also considered the infinite sum (1.2) ∞ n=1 Hn(r) nm
for an integer m ≥ r + 1, and gave some identities involving this sum (see [9, Sect. 4 and 6]).
In this paper, as a natural extension of (1.2), we consider the following Dirichlet series associated with hyperharmonic numbers: (1.3) hr(s) := ∞ n=1 Hn(r) ns (s ∈ C).
By (1.1) and the well-known estimate Hn∼ log n, we have Hn(r)= O(nr−1log n) as n tends to infinity for a fixed integer r ≥ 1. Therefore the right-hand side of (1.3) is absolutely convergent for (s) > r and it defines holomorphic function in this region. Since the derivative of the Riemann zeta function ζ(s) can be written as−∞n=1log n/ns for(s) > 1, the function h1(s) can be regarded as an analogue of ζ(s). Matsuoka [8] proved the meromorphic continuation of h1(s) to the whole s-plane and gave the location of poles and residues (see Theorem 2.2 in the next section). This function has been investigated more precisely by Apostol-Vu [2] and Mez˝o-Dil [9].
As stated in Proposition 2.1 below, the function hr(s) is a specialization of a multiple zeta function: hr(s) = ζr+1 (s,
r−1 0, . . . , 0, 1). Here ζr(s1, . . . , sr) are non-strict multiple zeta functions defined by
(1.4) ζr(s1, . . . , sr) =
m1≥m2≥···≥mr≥1 m−s1
1 m−s2 2. . . m−sr r.
When r = 1, the function ζ1(s) is nothing but the Riemann zeta function ζ(s). It is known that the right-hand side of (1.4) is absolutely convergent when(s1) is sufficiently large. Probably strict multiple zeta functions, defined by
(1.5) ζr(s1, . . . , sr) =
m1>m2>···>mr≥1 m−s1
1 m−s2 2. . . m−sr r,
are better known than non-strict ones. Values of strict multiple zeta functions at positive integers are called Multiple Zeta Values (MZVs) and many relations among MZVs have been given (e.g. see [5]
Dirichlet Series Associated with Hyperharmonic Numbers
and references therein). The multiple sum (1.5) is also absolutely convergent when (s1) is sufficiently large, and Akiyama-Egami-Tanigawa [1], Zhao [10] and Matsumoto-Tanigawa [7] proved the meromorphic continuation of ζr(s1, . . . , sr) to the whole Cr space by different methods. Since the function ζr can be expressed in terms of ζ1, ζ2, . . . , ζr, it can be also meromorphically continued to the wholeCr space. As a special case, we obtain the meromorphic continuation of hr(s) to the whole s-plane.
The following is the main result of this paper. We note that the statement (i) gives another proof of the meromorphic continuation of hr(s).
Theorem 1.1. (i). For a complex number s with (s) > r, we have
hr(s) = 1 r! r k=1 r k h1(s − k + 1) + r k=1 k r k + 1 − r k Hr−1 ζ(s − k + 1) , (1.6)
whererk are Stirling numbers of the first kind. The function hr(s) can be meromorphically continued to the whole s-plane by this expression.
(ii). The function hr(s) has double poles at s = 1, 2, . . . , r and has (possible) simple poles at s = −a (a = 0, 1, 2, 3, . . .). The residues at each poles are given as follows:
(1.7) Res s=a hr(s) = 1 r! a r a + 1 + r a (γ − Hr−1)− r k=a+1 r k Bk−a k − a (a = 1, 2, . . . r). (1.8) Res s=−a hr(s) = − 1 r! r k=1 r k Ba+k a + k (a = 0, 1, 2, . . .). Here γ is the Euler constant and Bn is the n-th Bernoulli number defined by
t et− 1 = ∞ n=0 Bn n! t n.
Remark 1.2. It is easy to see that r 2 − r 1 Hr−1= 0 (r ≥ 1)
(by Lemma 2.4 below, for example). Therefore the term for k = 1 vanishes in the second sum of (1.6). Remark 1.3. We consider the Barnes type multiple zeta function:
Zr(s) =
m1≥0,...,mr≥0 (m1,...,mr)=(0,...,0)
(m1+· · · + mr)−s
for (s) > r. The function Zr(s) can be meromorphically continued to the whole plane, and the special values of Zr(s) essentially coincide with residues of hr(s):
Zr(−a) = r Ress=−ahr(s) (a ≥ 1).
Ken KAMANO
2. Proof of Theorem 1.1
First we show that hr(s) is a specialization of the non-strict multiple zeta function: Proposition 2.1. For a complex number s with (s) > r, we have
hr(s) = ζr+1 (s, r−1 0, . . . , 0, 1).
Proof. By the repeated use of the equation Hn(r)=ni=1Hi(r−1), we have hr(s) = n≥m1≥...≥mr≥1 1 nsmr = n≥m1≥...≥mr≥1 1 nsm10· · · m0r−1mr = ζr+1 (s, r−1 0, . . . , 0, 1). To prove Theorem 1.1, we need the following theorem proved by Matsuoka (note that he considered the function h(s) = h1(s) − ζ(s + 1)). Our main theorem (Theorem 1.1) is a generalization of this theorem. Theorem 2.2 (Matsuoka [8]). (i). The function h1(s) can be meromorphically continued to the whole plane and is holomorphic except for s = 1, 0, −1, −3, −5, . . .. The function h1(s) has a double pole at s = 1 and simple poles at s = 0, −1, −3, −5, . . ..
(ii). The residues of h1(s) at each poles are given as follows: Res s=1 h1(s) = γ, (2.1) Res s=0 h1(s) = 1 2, (2.2) Res s=1−2ah1(s) = − B2a 2a (a ≥ 1). (2.3)
Remark 2.3. Equations (2.2) and (2.3) can be expressed as
(2.4) Res
s=1−ah1(s) = − Ba
a (a ≥ 1).
Since Ba = 0 for all odd integers a ≥ 3, Equation (2.4) implies that h1(s) is holomorphic at s = −2, −4, −6, . . ..
For integers r ≥ 1 and k ≥ 0, we denote by rk the number of permutations of r elements with k disjoint cycles (e.g. [6, Chapter 6]). These numbers are called Stirling numbers of the first kind and appear in the coefficients of the expansion of the rising factorial:
(n + 1) · · · (n + r − 1) = r k=1 r k nk−1 (r ≥ 1). (2.5)
Lemma 2.4. The following identities hold: (2.6) n + r − 1 r − 1 = 1 (r − 1)! r k=1 r k nk−1.
Dirichlet Series Associated with Hyperharmonic Numbers (2.7) n + r − 1 r − 1 1 n + 1+· · · + 1 n + r − 1 = 1 (r − 1)! r−1 k=1 r k + 1 nk−1.
Proof. It is clear that (2.6) follows from (2.5). The left-hand side of (2.7) is equal to d dx x + r − 1 r − 1 x=n. By (2.6), this equals 1 (r − 1)! r k=2 r k (k − 1)xk−2 x=n = 1 (r − 1)! r−1 k=1 r k + 1 knk−1 and (2.7) follows.
Proof of Theorem 1.1. (i). By (1.1) and Lemma 2.4, we have Hn(r)= n + r − 1 r − 1 Hn+ 1 n + 1+· · · + 1 n + r − 1− Hr−1 = 1 (r − 1)! r k=1 r k Hn n1−k + r−1 k=1 r k + 1 k n1−k − r k=1 r k Hr−1 n1−k . Therefore we obtain hr(s) = 1 (r − 1)! r k=1 r k h1(s − k + 1) + r k=1 k r k + 1 − r k Hr−1 ζ(s − k + 1)
and this proves (1.6). Since h1(s) and ζ(s) are meromorphically continued to the whole plane, the function hr(s) can be also continued.
(ii). First, from the first part of (1.6), double poles appear at s = 1, 2, . . . , r and simple poles appear at s = 0, −1, −2, . . .. Next, from the latter part of (1.6), only simple poles appear at s = 1, 2, . . . , r. By Theorem 2.2 and the fact the Riemann zeta function has only a simple pole at s = 1 with residue 1, we
obtain residues of hr(s) as (1.7) and (1.8).
References
[1] S. Akiyama, S. Egami and Y. Tanigawa, Analytic continuation of multiple zeta-functions and their values at
non-positive integers. Acta Arith.98 (2001), 107–116.
[2] T. M. Apostol and T. H. Vu, Dirichlet series related to the Riemann zeta function. J. Number Theory 19 (1984), 85–102.
[3] A. T. Benjamin, D. Gaebler and R. Gaebler, A combinatorial approach to hyperharmonic numbers. Integers3 (2003), A15.
[4] J. H. Conway and R. K. Guy, The Book of Numbers. Springer, (1976). [5] M. Eie, Topics in Number Theory. World Scientific, (2009).
[6] R. L. Graham, D. E. Knuth and O. Patashnik, Concrete Mathematics : A Foundation for Computer Science (2nd
Edition). Addison-Wesley Professional, (1994).
[7] K. Matsumoto and Y. Tanigawa, The analytic continuation and the order estimate of multiple Dirichlet series. J. Th´eor. Nombres Bordeaux15 (2003), 267–274.
[8] Y. Matsuoka, On the values of a certain Dirichlet series at rational integers. Tokyo J. Math.5 (1982), 399–403. [9] I. Mez˝o and A. Dil, Hyperharmonic series involving Hurwitz zeta function. J. Number Theory130 (2010), 360–369. [10] J. Zhao, Analytic continuation of multiple zeta functions. Proc. Amer. Math. Soc.128 (2000), 1275–1283.
