Integral and Series Representations of Riemann's Zeta Function and Dirichlet's Eta Function and a Medley of Related Results

Abstract

Contour integral representations of Riemann's Zeta function and Dirichlet's Eta (alternating Zeta) function are presented and investigated. These representations flow naturally from methods developed in the 1800s, but somehow they do not appear in the standard reference summaries, textbooks, or literature. Using these representations as a basis, alternate derivations of known series and integral representations for the Zeta and Eta function are obtained on a unified basis that differs from the textbook approach, and results are developed that appear to be new.

Dedicated to the memory of Leslie M. Saunders,
died September 8, 1972, age 29
(Physics Today obituary, issue of December 1972, page 63)

1. Introduction

Riemann's Zeta function 𝜁(𝑠) and its sibling Dirichlet's (alternating zeta) function 𝜂(𝑠) play an important role in physics, complex analysis, and number theory and have been studied extensively for several centuries. In the same vein, the general importance of a contour integral representation of any function has been known for almost two centuries, so it is surprising that contour integral representations for both 𝜁(𝑠) and 𝜂(𝑠) exist that cannot be found in any of the modern handbooks (NIST, [1, Section 25.5(iii)]; Abramowitz and Stegun, [2, Chapter 23]), textbooks (Apostol, [3, Chapter 12]; Olver, [4, Chapter 8.2]; Titchmarsh, [5, Chapter 4]; Whittaker and Watson, [6, Section (13.13)]), summaries (Edwards, [7], Ivić, [8, Chapter 4]; Patterson, [9]; Srivastava and Choi, [10, 11]), compendia (Erdélyi et al., [12, Section 1.12] and Chapter 17), tables (Gradshteyn and Ryzhik, [13], 9.512; Prudnikov et al, [14, Appendix II.7]), and websites [1, 15–19] that summarize what is known about these functions. One can only conclude that such representations, generally discussed in the literature of the late 1800s, have been long-buried and their significance has been overlooked by modern scholars.

It is the purpose of this work to disinter these representations and revisit and explore some of the consequences. In particular, it is possible to obtain series and integral representations of 𝜁(𝑠) and 𝜂(𝑠) that unify and generalize well-known results in various limits and combinations and reproduce results that are only now being discovered and investigated by alternate means. Additionally, it is possible to explore the properties of 𝜁(𝑠) in the complex plane and on the critical line and obtain results that appear to be new. Many of the results obtained are disparate and difficult to categorize in a unified manner but share the common theme that they are all somehow obtained from a study of the revived integral representations. That is the unifying theme of this work. To maintain a semblance of brevity, most of the derivations are only sketched, with citations sufficient enough to allow the reader to reproduce any result for her/himself.

2. Contour Integral Representations of 𝜁(𝑠) and 𝜂(𝑠)

Throughout, I will use 𝑠=𝜎+𝑖𝜌, where 𝑠∈ℭ (complex), 𝜎,𝜌∈ℜ (reals), and 𝑛 and 𝑁 are positive integers.

The Riemann Zeta function 𝜁(𝑠) and the Dirichlet alternating zeta function 𝜂(𝑠) are well known and defined by (convergent) series representations: 𝜁(𝑠)≡^∞_𝑘=1𝑘^−𝑠𝜂(𝑠),𝜎>1,≡−^∞_𝑘=1(−1)^𝑘𝑘^−𝑠,𝜎>0, (1)(2) with 𝜂(𝑠)=1−2^1−𝑠𝜁(𝑠). (3) It is easily found by an elementary application of the residue theorem that the following reproduces (1): 𝜁(𝑠)=𝜋^∞_𝑘=1𝑡residue^−𝑠𝑒^𝑖𝜋𝑡sin(𝜋𝑡)_𝑡=𝑘,𝜎>1 (4) and the following reproduces (2): 𝜂(𝑠)=𝜋^∞_𝑘=1residue𝑡^−𝑠cot(𝜋𝑡)𝑒^𝑖𝜋𝑡_𝑡=𝑘,𝜎>0. (5) Convert the result (4) into a contour integral enclosing the positive integers on the 𝑡-axis in a clockwise direction, and provided that 𝜎>1 so that contributions from infinity vanish, the contour may then be opened such that it stretches vertically in the complex 𝑡-plane, giving 𝜁(𝑠)=𝑖2^𝑐+𝑖∞_{𝑐−𝑖∞}𝑡^−𝑠()1cot𝜋𝑡𝑑𝑡,0<𝑐<1,𝑐∈ℜ,=−2^∞_−∞(𝑐+𝑖𝑡)^−𝑠cot(𝜋(𝑐+𝑖𝑡))𝑑𝑡,𝜎>1, (6)(7) a result that cannot be found in any of the modern reference works cited (in particular [20]). It should be noted that this procedure is hardly novel—it is often employed in an educational context (e.g., in the special case 𝑠=2) to demonstrate the utility of complex integration to evaluate special sums ([10, Section 4.1]), ([21, Lemma 2.1]) and forms the basis for an entire branch of physics [22].

A similar application to (5), with some trivial trigonometric simplification, gives 𝜂(𝑠)=−𝑖2^𝑐+𝑖∞_{𝑐−𝑖∞}𝑡^−𝑠()=1sin𝜋𝑡𝑑𝑡2^∞_−∞(𝑐+𝑖𝑡)^−𝑠sin(𝜋(𝑐+𝑖𝑡))𝑑𝑡. (8)(9) Equation (9) reduces, with 𝑐=1/2, to a Jensen result [23], presented (without proof) in 1895, and, to the best of my knowledge, reproduced, again, but only in the special case 𝑐=1/2, only once in the modern reference literature [17]. In particular, although Lindelof's 1905 work [24] arrives at a result equivalent to that obtained here, attributing the technique to Cauchy (1827), it is performed in a general context, so that the specific results (7) and (9) never appear, particularly in that section of the work devoted to 𝜁(𝑠). In modern notation, for a meromorphic function 𝑓(𝑧), Lindelof writes ([24, Equation III(4)]) in general^∞_𝜈=−∞𝑓(𝜈)1=−()𝑓(𝑧)2𝜋𝑖𝜋cot𝜋𝑧𝑑𝑧, (10) where the contour of integration encloses all the singularities of the integrand but never specifically identifies 𝑓(𝑧)=𝑧^−𝑠. Much later, after the integral has been split into two by the invocation of identities for cot(𝜋𝑧), he does make this identification ([24, Equation 4.III(6)]), arriving (with (3)) at ((26)—see below) a now-well-known result, thereby bypassing the contour integral representation (7). In a similar vein, Olver ([4, page 290]) reproduces (10), employing it to evaluate remainder terms, the Abel-Plana formula and a Jensen result but again never writes (7) explicitly. Perhaps the omission of an explicit statement of (7) and (9) from Chapter IV of Lindelof [24] is why these forms seem to have vanished from the historical record, although citations to Jensen and Lindelof abound in modern summaries. (Recently, Srivastava and Choi [11, pages 169–172], have reproduced Lindelof's analysis, also failing to arrive explicitly at (7); although they do write a general form of (8) ([11, Section 2.3, Equation (38)]), it is studied only for the case 𝑐=1/2.) (Note added in proof: after a draft copy of this paper was posted, I was made aware that Ruijsenaars has previously obtained a generalized form of (7) for the Hurwitz zeta function ([25, Equation (4.22)]). He also notes (private communication) that he could not find prior references to his result in the literature).

At this point, it is worthwhile to quote several related integral representations that define some of the relevant and important functions of complex analysis, starting with Riemann's original 1859 contour integral representation of the zeta function [26]: 𝜁(𝑠)=Γ(1−𝑠)𝑡2𝜋𝑖𝑒^−𝑡−^𝑠−11𝑑𝑡, (11) where the contour of integration encloses the negative 𝑡-axis, looping from 𝑡=−∞−𝑖0 to 𝑡=−∞+𝑖0 enclosing the point 𝑡=0.

The equivalence of (11) and (1) is well described in texts (e.g., [11]) and is obtained by reducing three components of the contour in (11). A different analysis is possible however (e.g., [7, Section 1.6])—open and translate the contour in (11) such that it lies vertically to the right of the origin in the complex 𝑡-plane (and thereby vanishes) with 𝜎<0 and evaluate the residues of each of the poles of the integrand lying at 𝑡=±2𝑛𝜋𝑖 to find 𝜁(𝑠)=−𝑖Γ()2𝜋1−𝑠)(2𝜋^𝑠×exp𝑖𝜋𝑠2−exp−𝑖𝜋𝑠2^∞_𝑘=1𝑘^𝑠−1. (12) This can be further reduced, giving the well-known functional equation for 𝜁(𝑠): 𝜁()=()Γ(𝑠)1−𝑠2cos𝜋𝑠/2(2𝜋)^𝑠𝜁(𝑠), (13) which is valid for all 𝑠 by analytic continuation from 𝜎<0. This demonstrates that (11) and (6) are fundamentally different representations of 𝜁(𝑠), since further deformations of (11) do not lead to (7).

Ephemerally, at one time it was noted in [17] that the following complex integral representation 𝜁(𝑠)=𝜋2(𝑠−1)^∞_−∞(1/2+𝑖𝑡)^1−𝑠cosh²(𝜋𝑡)𝑑𝑡,𝜎>0 (14) can be obtained by the invocation of the Cauchy-Schlömilch transformation [27]. In fact, the representation (14) follows by the simple expedient of integrating (7) by parts in the special case 𝑐=1/2 (see (35)), as also noted in [17]. Another result 𝜁(𝑠)=2^𝑠−11−2^1−𝑠Γ(𝑠+1)^∞₀𝑡^𝑠cosh²(𝑡)𝑑𝑡,𝜎>−1 (15) obtained in [27] on the basis of the Cauchy-Schlömilch transformation does not follow from (14) as erroneously claimed at one time, but later corrected, in [17].

Based on (11), a number of other representations of 𝜁(𝑠) are well known. From the tables cited, the following are worth noting here:

([12, Equation 1.12(6)]) 𝜁(𝑠)=12Γ(𝑠)(1−2^−𝑠)^∞₀𝑡^𝑠−1sinh(𝑡)𝑑𝑡,𝜎>1, (16) ([12, Equation 1.12(4)]) 𝜁(𝑠)=1Γ(𝑠)^∞₀𝑡^𝑠−1exp(𝑡)−1𝑑𝑡,𝜎>1, (17) and ([13, Equation 3.527(1)]) 𝜁(𝑠)=14(2𝑎)^(𝑠+1)Γ(𝑠+1)^∞₀𝑡^𝑠sinh²(𝑎𝑡)𝑑𝑡,𝜎>1 (18) By way of comparison, Laplace’s representation ([13, Equation 8.315(2)]) 1Γ(𝑠)=12𝜋𝑖^∞_−∞𝑒^{(𝑐+𝑖𝑡)}(𝑐+𝑖𝑡)^𝑠𝑑𝑡 (19) defines the inverse Gamma function.

3. Simple Reductions

3.1. The Case 0≤𝑐≤1

The simplest reductions of (7) and (9) are obtained by uniting the two halves of their range corresponding to 𝑡<0 and 𝑡>0. With the use of −(𝑐+𝑖𝑡)^𝑠+(𝑐−𝑖𝑡)^𝑠𝑖=2𝑐²+𝑡^2(𝑠/2)𝑡×sin𝑠arctan𝑐,(𝑐+𝑖𝑡)^𝑠+(𝑐−𝑖𝑡)^𝑠=2𝑐²+𝑡^2(𝑠/2)𝑡cos𝑠arctan𝑐, (20)(21) one finds (with 0<𝑐<1 and 0≤𝑐≤1, resp.) 𝜁(𝑠)=−^∞₀𝑡cos𝑠arctan𝑐()()()()𝑡cos𝜋𝑐sin𝜋𝑐−sinh𝜋𝑡cosh𝜋𝑡sin𝑠arctan𝑐×𝑐²+𝑡^2𝑠/2cosh²(𝜋𝑡)−cos²(𝜋𝑐)⁻¹𝜂(𝑠)=𝑑𝑡,𝜎>1,_∞0𝑡−sin𝑠arctan𝑐()()𝑡cos𝜋𝑐sinh𝜋𝑡+cos𝑠arctan𝑐()()×𝑐sin𝜋𝑐cosh𝜋𝑡₂+𝑡^2𝑠/2cosh²(𝜋𝑡)−cos²(𝜋𝑐)⁻¹𝑑𝑡, (22)(23) the latter valid for all 𝑠. In the case that 𝑐=1/2, (22) reduces to 𝜁(𝑠)=2^𝑠∞₀sinh(𝜋𝑡)sin(𝑠arctan(2𝑡))1+4𝑡^2𝑠/2cosh(𝜋𝑡)𝑑𝑡,𝜎>1, (24) equivalent to the well-known result (attributed to Jensen by Lindelof, and a special case of Hermite's result for the Hurwitz zeta function ([6, Section 13.2)]) (which also appears as an exercise in many textbooks—for example, [11, Equation 2.3(41)]): 𝜁(𝑠)=2^(𝑠−1)𝑠−1−2^𝑠∞₀sin(𝑠arctan(2𝑡))𝑒^−𝜋𝑡1+4𝑡^2𝑠/2cosh(𝜋𝑡)𝑑𝑡. (25) Similarly, (23) reduces to the known ([24, IV.III(6)]) result: 𝜂(𝑠)=2^𝑠−1∞₀cos(𝑠arctan(𝑡))1+𝑡^2𝑠/2cosh(𝜋𝑡/2)𝑑𝑡. (26) Comparison of (24) and (25) immediately yields the known result ([13, Equation 3.975(2)]): 𝜁(𝑠)=2^(𝑠−1)1−𝑠+2^𝑠∞₀sin(𝑠arctan(2𝑡))𝑒^𝜋𝑡1+4𝑡^2𝑠/2cosh(𝜋𝑡)𝑑𝑡,𝜎>1. (27) Notice that (25) and (26) are valid for all 𝑠. Simple substitution of 𝑐→0 in (23) gives 𝜂(𝑠)=−sin𝑠𝜋2^∞₀𝑡^−𝑠sinh(𝜋𝑡)𝑑𝑡,𝜎<0, (28) which, after taking (3) into account, leads to 𝜁(𝑠)=sin(𝑠𝜋/2)2^1−𝑠−1^∞₀𝑡^−𝑠sinh(𝜋𝑡)𝑑𝑡,𝜎<0, (29) displaying the existence of the trivial zeros of 𝜁(𝑠) at 𝑠=−2𝑁, since the integral in (29) is clearly finite at each of these values of 𝑠. The result (29) is equivalent to the known result (16) after taking (13) into account. Other possibilities abound. For example, set 𝑐=1/3 in (22) to obtain 𝜁(𝑠)=3^𝑠−1∞₀2sin(𝑠arctan(𝑡))sinh2𝜋𝑡3−√((𝑡×3cos𝑠arctan))1+𝑡^2𝑠/22cosh2𝜋𝑡3+1⁻¹𝑑𝑡,𝜎>1. (30) Further, a novel representation can be found by noting that (9) is invariant under the transformation of variables 𝑡→𝑡+𝜌 together with the choice 𝑐=𝜎 along with the requirement that 0≤𝜎≤1. This effectively allows one to choose 𝑐=𝑠, yielding the following complex representation for 𝜂(𝑠), as well as the the corresponding representation for 𝜁(𝑠) because of (3), which is valid in the critical strip 0≤𝜎≤1: 𝜂(𝑠)=−𝑖2^𝑠+𝑖∞_{𝑠−𝑖∞}𝑡^−𝑠()=1sin𝜋𝑡𝑑𝑡2^∞_−∞(𝑠+𝑖𝑡)^−𝑠sin(𝜋(𝑠+𝑖𝑡))𝑑𝑡. (31)(32) Because (20) and (21) are valid for 𝑐∈ℭ, it follows that (23) with 𝑐=𝑠 is also a valid representation for 𝜂(𝑠), as well as the the corresponding representation for 𝜁(𝑠) with recourse to (3), again only valid in the critical strip 0≤𝜎≤1. Another interesting variation is the choice 𝑐=1/𝜌 in (23), valid for 𝜌>1, again applicable to the corresponding representation for 𝜁(𝑠) because of (3). Such a representation may find application in the analysis of 𝜁(𝜎+𝑖𝜌) in the asymptotic limit 𝜌→∞. Finally, the choice 𝑐=𝜎,0≤𝜎≤1 leads to an interesting variation that is discussed in Section 9.

3.2. Simple Recursion via Partial Differentiation

As a function of 𝑐, 𝜁(𝑠) and 𝜂(𝑠) are constant, as are their integral representations over a range corresponding to 0<𝑁−𝑐<1; however the integrals change discontinuously at each new value of 𝑐 that corresponds to a new value of 𝑁. Thus, the operator 𝜕/𝜕𝑐 acting on (7) and (9) vanishes, except at nonnegative integers 𝑐=𝑁, when it becomes indefinite. Therefore, it is of interest to investigate this operation in the neighbourhood of 𝑐=𝑁/2 as well as the limit 𝑐→𝑁.

Applied to (9) (and equivalent to integration by parts), the requirement that 𝜕𝜂(𝑠)𝜕𝑐=0,𝑐≠𝑁 (33) and identification of one of the resulting terms yields 𝜂()𝜋𝑠+1=−2𝑠^∞_−∞(𝑐+𝑖𝑡)^−𝑠cos(𝜋(𝑐+𝑖𝑡))sin²(𝜋(𝑐+𝑖𝑡))𝑑𝑡0≤𝑐<1, (34) a new result with improved convergence at infinity. Similarly operating with 𝜕/𝜕𝑐 on (7) yields 𝜁()=𝜋𝑠+12𝑠^∞_−∞(𝑐+𝑖𝑡)^−𝑠sin²(𝜋(𝑐+𝑖𝑡))𝑑𝑡0<𝑐<1. (35) (It should be noted that Srivastava et al. [20] study an incomplete form of (35) in order to obtain representations for sums both involving 𝜁(𝑛), and 𝜁(𝑛) by itself, without commenting on the significance of the contour integral with infinite bounds (35).) Using (20) and (21) to reduce the integration range of (34) gives a fairly lengthy result for general values of 𝑐. If 𝑐=1/2, (34) becomes 𝜂()=2𝑠+1^𝑠−1𝜋𝑠×^∞₀1+𝑡^2−𝑠/2sinh𝜋𝑡2sin(𝑠arctan(𝑡))×cosh⁻²𝜋𝑡2𝑑𝑡, (36) which, by letting 𝑠→𝑠−1, can be rewritten as 𝜂()=2()𝜂(𝑠)𝑠+1𝑠−1𝑠+2^𝑠−1𝜋𝑠×^∞₀cos(𝑠arctan(𝑡))1+𝑡^2−𝑠/2sinh𝜋𝑡2×cosh⁻²𝜋𝑡2𝑡𝑑𝑡, (37) and, with regard to (3), 𝜁()=2(𝑠+1𝑠−1)1−2^1−𝑠𝜁(𝑠)𝑠(1−2^−𝑠)+2^𝑠−1𝜋𝑠(1−2^−𝑠)×^∞₀cos(𝑠arctan(𝑡))1+𝑡^2−𝑠/2sinh𝜋𝑡2×cosh⁻²𝜋𝑡2𝑡𝑑𝑡. (38) Similarly, ((37) and (38) are the first two entries in a family of recursive relationships that will be discussed elsewhere (in preparation)). In the case that 𝑐=1/2, (35) reduces to the well-known result ([24, Equation 4,III(5)] (attributed to Jensen [23])) 𝜁(𝑠+1)=𝜋2^𝑠−1𝑠^∞₀cos(𝑠arctan(𝑡))1+𝑡^2𝑠/2cosh²(𝜋𝑡/2)𝑑𝑡. (39) If 𝑐→0 in (34) (for the case 𝑐=0 in (35) see the following section), one finds a finite result (also equivalent to integrating (28) by parts): 𝜂(𝑠)=𝜋sin(𝑠𝜋/2)𝑠−1^∞₀𝑡^1−𝑠cosh(𝜋𝑡)sinh²(𝜋𝑡)𝑑𝑡,𝜎<0, (40) and, when 𝑐=1, one finds the unusual forms: 𝜂(𝑠+1)={0,1}+𝑖𝜋𝑟2𝑠×^𝜋₀1±𝑟𝑖𝑒^{𝑖𝜃−𝑠}cosh𝜋𝑟𝑒^𝑖𝜃𝑒^𝑖𝜃sinh²𝜋𝑟𝑒^𝑖𝜃−𝑑𝜃𝑟𝜋𝑠^∞₁1+𝑟²𝑡^2−𝑠/2((()cos𝑠arctan𝑟𝑡))×cosh𝜋𝑟𝑡sinh⁻²(𝜋𝑟𝑡)𝑑𝑡, (41) which is valid for any 0<𝑟<1 and all 𝑠 (particularly 𝑟=1/𝜋 or 𝑟=1/2), after deformation of the contour of integration in (34) either to the left or to the right of the singularity at 𝑡=1, respectively, belonging to a choice of ± where the symbol {0,1} means zero or one as the case may be.

4. Special Cases

4.1. The Case 𝑐=0

With respect to (35), setting 𝑐=0 together with some simplification gives 𝜁(𝑠)=−𝜋^𝑠−1sin(𝜋𝑠/2)𝑠−1^∞₀𝑡^1−𝑠sinh²(𝑡)𝑑𝑡,𝜎<0, (42) reproducing (18), with 𝑎=1, after taking (13) into account. This same case (𝑐=0) naturally suggests that the integral in (6) be broken into two parts: 𝜁(𝑠)=𝜁₀(𝑠)+𝜁₁(𝑠), (43) where 𝜁₀(𝑠)=𝑖2^𝑖_−𝑖𝑡^−𝑠()𝜁cot𝜋𝑡𝑑𝑡,₁(𝑠)=𝑖2^𝑖_−𝑖∞𝑡^−𝑠()𝑖cot𝜋𝑡𝑑𝑡+2_𝑖^𝑖∞𝑡^−𝑠cot(𝜋𝑡)𝑑𝑡 (44)(45) such that the contour in (44) bypasses the origin to the right. It is easy to show that the sum of the two integrals in (45) reduces to 𝜁₁(𝑠)=sin𝜋𝑠2^∞₁𝑡^−𝑠()=()coth𝜋𝑡𝑑𝑡,𝜎>1,sin𝜋𝑠/2𝑠−1+sin𝜋𝑠2^∞₁𝑡^−𝑠𝑒^{(−𝜋𝑡)}sinh(𝜋𝑡)𝑑𝑡, (46)(47) the latter being valid for all 𝑠. One, of several commonly used possibilities of reducing (44), is to apply ([28, Equation (54.3.2)]) ()2𝑡cot𝜋𝑡=−𝜋^∞_𝑘=0𝑡^2𝑘𝜁(2𝑘),|𝑡|<1, (48) giving 𝜁₀(𝑠)2=−𝜋sin𝜋𝑠2^∞_𝑘=0(−1)^𝑘𝜁(2𝑘),2𝑘−𝑠 (49) being convergent for all 𝑠. Such sums appear frequently in the literature (e.g., [10, 29] Section (3.2)) and have been studied extensively when 𝑠=−𝑛 [10, 30]. In the case 𝑠=2𝑛, (49) reduces to 𝜁₀(𝑠)=𝜁(𝑠)   and (47) vanishes.

Along with (42) the integral in (35) can be similarly split, using the convergent series representation: 1sin²()=2𝜋𝑣𝜋^2∞_𝑘=0(2𝑘−1)𝜁(2𝑘)𝑣^2𝑘−2,|𝑣|<1, (50) where the Bernoulli numbers 𝐵_2𝑘 appearing in the original ([28, Equation (50.6.10)]) are related to 𝜁(2𝑘) by 𝜁()12𝑘=−2(−1)^𝑘(2𝜋)^2𝑘𝐵_2𝑘Γ().2𝑘+1 (51) Thus the representation 𝜁(𝑠)1=−2𝑖𝜋(𝑠−1)^𝑖_−𝑖𝑡^1−𝑠sin²()−()𝜋𝑡𝑑𝑡𝜋sin𝜋𝑠/2𝑠−1^∞₁𝑡^1−𝑠sinh²(𝜋𝑡)𝑑𝑡 (52) becomes (formally) 𝜁(𝑠)()=−sin𝜋𝑠/2×2𝑠−1𝜋^∞_𝑘=0(−1)^𝑘(2𝑘−1)𝜁(2𝑘)2𝑘−𝑠+𝜋^∞₁𝑡^1−𝑠sinh²(𝜋𝑡)𝑑𝑡 (53) after straightforward integration of (50) with 𝜎<0. In (53), the integral converges, and the formal sum conditionally converges or diverges according to several tests. Howver, further analysis is still possible. Following the method of [10, Section 3.3] applied to (49), together with [10, Equation 3.4(15)], we have the following general result for |𝑣|<1: ^∞_𝑘=0𝜁(2𝑘)𝑣^2𝑘=12𝑘−𝑠+2𝑠^𝑣₀𝑡^∞−𝑠_𝑘=1𝜁(2𝑘)𝑡^2𝑘−1=1𝑑𝑡+12𝑠2𝑣^𝑠𝑣₀𝑡^−𝑠(Ψ(1+𝑡)−Ψ(1−𝑡))𝑑𝑡,𝜎<2 (54)(55) so that (formally), after premultiplying (55) with 1/𝑣, then operating with 𝑣²(𝜕/𝜕𝑣), ^∞_𝑘=0𝜁(2𝑘)(2𝑘−1)𝑣^2𝑘12𝑘−𝑠=−+12𝑠2𝑣^𝑠×^𝑣₀𝑡^1−𝑠Ψ^′(1+𝑡)+Ψ^′(11−𝑡)𝑑𝑡=−+12𝑠2𝑣^𝑠×^𝑣₀𝑡^1−𝑠𝜋²1+cot²()−1𝜋𝑡𝑡²𝑑𝑡,𝜎<2. (56)(57) Notice that the right-hand side of (55) provides an analytic continuation of the left-hand side for |𝑣|>1 and the right-hand side of (57) gives a regularization of the left-hand side. Putting together (53) and (57), incorporating the first term of (57) into the second term of (53), setting 𝑣=𝑖, and constraining 𝜎>0 eventually yields a new integral representation (compare with (121), [31, Equation (2.8)] and [25, Equation (4.80)]): 𝜁(𝑠)𝜋=−^𝑠−1sin(𝜋𝑠/2)𝑠−1^∞₀𝑡^1−𝑠1sinh²𝑡−1𝑡²𝑑𝑡, (58) which is valid for 0<𝜎<2, a nonoverlapping analytic continuation of (42) that spans the critical strip. Additionally, using (13) followed by replacing 𝑠→1−𝑠 gives a second representation: 𝜁(𝑠)=2^𝑠−1Γ(1+𝑠)^∞₀𝑡^𝑠1sinh²𝑡−1𝑡²𝑑𝑡 (59) valid for −1<𝜎<1, a non-overlapping analytic continuation of (18). Over the critical strip 0<𝜎<1 representations (58) and (59) are simultaneously true and augment the rather short list of other representations that share this property ([32], as cited in [12] (unnumbered equations immediately following 1.12(1), also reproduced in [11, Equations 2.3(44)−(46)].)). See Section 9 for further application of this result.

4.2. The Case 𝑐≥1

Interesting results can be obtained by translating the contour in (6) and (9) and adding the residues so-omitted. From (1) and (2) define the partial sums 𝜁_𝑁(𝑠)≡^𝑁_𝑘=1𝑘^−𝑠𝜂_𝑁(𝑠)≡−^𝑁_𝑘=1(−1)^𝑘𝑘^−𝑠, (60) let 𝑐→𝑐_𝑁, 𝑁=1,2,3,… with 𝑁≤𝑐_𝑁<𝑁+1, (61) and adjust all the above results accordingly by redefining 𝑐→𝑐_𝑁. (If 𝑐_𝑁=𝑁, an extra half-residue is removed and the contour deformed to pass to the right of the point 𝑡=𝑁.) This gives, for example, the generalization of (6): 𝜁(𝑠)=𝜁_𝑁(𝑠)−12𝑁^𝑠𝛿_{𝑁,𝑐𝑁}+𝑖2^𝑐𝑁_𝑐^+𝑖∞_{𝑁−𝑖∞}𝑡^−𝑠cot(𝜋t)𝑑𝑡,𝜎>1, (62) and (9): 𝜂(𝑠)=𝜂_𝑁(𝑠)+(−1)^𝑁2𝑁^𝑠𝛿_{𝑁,𝑐𝑁}−𝑖2^𝑐𝑁_𝑐^+𝑖∞_{𝑁−𝑖∞}𝑡^−𝑠sin(𝜋𝑡)𝑑𝑡, (63) so (22) and (23) with 𝑐→𝑐_𝑁 become integral representations of the remainder of the partial sums 𝜁(𝑠)−𝜁_𝑁(𝑠) and 𝜂(𝑠)−𝜂_𝑁(𝑠), the former corresponding to a special case of the Hurwitz zeta function 𝜁(𝑠,𝑁).

In the case that 𝑐_𝑁=𝑁=1, one finds 𝜁(𝑠)=12+^∞₀cosh(𝜋𝑡)sin(𝑠arctan(𝑡))1+𝑡^2𝑠/2()=sinh𝜋𝑡𝑑𝑡,𝜎>1,𝑠+12()+𝑠−1^∞₀𝑒^−𝜋𝑡sin(𝑠arctan(𝑡))1+𝑡^2𝑠/2()𝜂(𝑠)=1sinh𝜋𝑡𝑑𝑡,2+^∞₀sin(𝑠arctan(𝑡))sinh(𝜋𝑡)1+𝑡^2𝑠/2𝑑𝑡, (64)(65)(66) with the latter two results being valid for all 𝑠. Neither (64) nor (66) appear in references; (65) corresponds to a limiting case of Hermite's representation of the Hurwitz zeta function ([12, Equation 1.10(7)]). In analogy to the developments leading to (27), equations (64), (66), and (3), together with the following modified form of [13, Equation (9.51.5)] 𝜁(𝑠)=2^𝑠−1𝑠(2^𝑠)+2−1)(𝑠−12^𝑠−1𝑠×−1_∞0sin(𝑠arctan(𝑡))1+𝑒^−𝜋𝑡1+𝑡^2𝑠/2sinh(𝜋𝑡)𝑑𝑡, (67) can be used to obtain the possibly new result: 𝜁(𝑠)=𝑠−32()+𝑠−1^∞₀sin(𝑠arctan(𝑡))𝑒^𝜋𝑡1+𝑡^2𝑠/2sinh(𝜋𝑡)𝑑𝑡,𝜎>1. (68) For 𝑐_𝑁=1/2+𝑁 corresponding to general partial sums, two possibilities are apparent. Based on (20), (21), (22), and (23) (for comparison, see also [4, Section 8.3], and [7, Section (6.4)] and Chapter 7), we find𝜁(𝑠)−^𝑁_𝑘=11𝑘^𝑠=1212+𝑁^1−𝑠∞₀sinh(𝜋𝑡(1+2𝑁))1+𝑡^2−𝑠/2sin(𝑠arctan(𝑡))cosh²((=()𝜋𝑡1/2+𝑁))𝑑𝑡𝜎>11/2+𝑁^1−𝑠1𝑠−1−22+𝑁^1−𝑠∞₀1+𝑡^2−𝑠/2sin(𝑠arctan(𝑡))1+𝑒^{𝜋𝑡(1+2𝑁)}𝜂(𝑠)+𝑑𝑡,_𝑁𝑘=1(−1)^𝑘𝑘^𝑠=(−1)^𝑁12+𝑁^1−𝑠∞₀1+𝑡^2−𝑠/2cos(𝑠arctan(𝑡))cosh(𝜋𝑡(1/2+𝑁))𝑑𝑡 (69)(70)(71)which together extend (24) and (26), respectively. Application of (34) and (35) with (20) and (21), also with 𝑐_𝑁=1/2+𝑁, gives the remainders𝜁(𝑠+1)−^𝑁_𝑘=11𝑘^𝑠+1=𝜋(1/2+𝑁)^1−𝑠𝑠^∞₀1+𝑡^2−𝑠/2cos(𝑠arctan(𝑡))cosh²((𝜂()+𝜋𝑡1/2+𝑁))𝑑𝑡,𝑠+1^𝑁_𝑘=1(−1)^𝑘𝑘^𝑠+1=𝜋(−1)^𝑁(1/2+𝑁)^1−𝑠𝑠^∞₀sinh(𝜋𝑡(1/2+𝑁))1+𝑡^2−𝑠/2sin(𝑠arctan(𝑡))cosh²(𝜋𝑡(1/2+𝑁))𝑑𝑡 (72) which are valid for all 𝑠. For the case 𝑐_𝑁=𝑁 applying (62) and (63) gives the remainders𝜁(𝑠)−^𝑁_𝑘=11𝑘^𝑠1=−2𝑁^𝑠+𝑁^1−𝑠∞₀cosh(𝜋𝑁𝑡)1+𝑡^2−𝑠/2sin(𝑠arctan(𝑡))()1sinh𝜋𝑁𝑡𝑑𝑡,𝜎>1=−2𝑁^𝑠+𝑁^1−𝑠𝑠−1+2𝑁^1−𝑠∞₀1+𝑡^2−𝑠/2sin(𝑠arctan(𝑡))𝑒^{2𝑁𝜋𝑡}𝜂(𝑠)+−1𝑑𝑡,_𝑁𝑘=1(−1)^𝑘𝑘^𝑠=(−1)^𝑁2𝑁^𝑠+(−1)^1+𝑁𝑁^1−𝑠∞₀1+𝑡^2−𝑠/2sin(𝑠arctan(𝑡))sinh(𝜋𝑁𝑡)𝑑𝑡, (73)(74)(75)which extend (64) and (66), the latter being valid for all 𝑠. It should be noted that (74), valid for 𝑠∈ℭ, generalizes well-established results (e.g., [13, Equation 3.411(12)]) that are only valid in the limiting case 𝑠=𝑛. (Although [13, Equation 3.411(21)] gives (with a missing minus sign) an integral representation for finite sums such as those appearing in (74), specified to be valid only for integer exponents (𝑘^𝑛), that particular result in [13] also appears to apply after the generalization 𝑛→𝑠.) Omitting the integral term from (70) gives an upper bound for the remainder; similarly, omitting the integral term in (74) gives a lower bound for the remainder, provided that 𝑠∈ℜ in both cases. In addition, many of the above results, in particular (70) and (74), appear to be new and should be compared to classical results such as [8, Theorem 1.8, page 23 and pages 99-100], and asymptotic approximations such as [33, Equation (8.2.1)].

5. Series Representations

5.1. By Parts

By simple manipulation of the integral representations given, it is possible to obtain new series representations. Starting from the integral representation (29), integrate by parts, giving 𝜁(𝑠)=sin(𝜋𝑠/2)𝜋²2^(1−𝑠)−1(−𝑠/2+1)×^∞₀𝑡₁^2−𝑠F₂−𝑠23+1;2𝑠,2−2;𝑡²𝜋²4×cosh(𝜋𝑡)sinh⁻³(𝜋𝑡)𝑑𝑡,𝜎<0, (76) where the hypergeometric function arises because of [34, Equation 6.2.11(6)]. Write the hypergeometric function as a (uniformly convergent) series, thereby permitting the sum and integral to be interchanged, apply a second integration by parts, and find 𝜁(𝑠)=𝜋2^1−𝑠−1sin𝜋𝑠2^∞_𝑘=0𝜋^2𝑘Γ()×2𝑘+2_∞0𝑡^{1−𝑠+2𝑘}sinh²(𝜋𝑡)𝑑𝑡,𝜎<0. (77) Identify the integral in (77) using (18) giving 𝜁(𝑠)=2^𝑠𝜋^𝑠−1sin(𝜋𝑠/2)2^1−𝑠×−1_∞𝑘=02^−2𝑘𝜁(−𝑠+1+2𝑘)Γ(−𝑠+2+2𝑘)Γ(),2𝑘+2 (78) which is valid for all values of 𝜎 by the ratio test and the principle of analytic continuation. This result, which appears to be new, can be rewritten by extracting the 𝑘=0 term, and applying (13) to yield 𝜁(𝑠)=2^𝑠𝜋^𝑠−1sin(𝜋𝑠/2)2^1−𝑠×−2+𝑠_∞𝑘=12^−2𝑘𝜁(−𝑠+1+2𝑘)Γ(−𝑠+2+2𝑘)Γ(),2k+2 (79) a representation, valid for all values of 𝑠, that is reminiscent of the following similar sum recently obtained by Tyagi and Holm ([31, Equation (3.5)]): 𝜁(𝑠)=𝜋^𝑠−11−2^1−𝑠sin𝜋𝑠2×^∞_𝑘=12−2^𝑠−2𝑘𝜁(−𝑠+1+2𝑘)Γ(−𝑠+1+2𝑘)Γ(),2𝑘+2𝜎>0 (80) (convergent by Gauss’ test). Equation (79) can also be rewritten in the form 𝜁(𝑠)=1Γ(𝑠)(2^𝑠)×−1−𝑠_∞𝑘=12^−2𝑘𝜁(𝑠+2𝑘)Γ(1+𝑠+2𝑘)Γ(),2𝑘+2 (81) by replacing 𝑠→1−𝑠 and applying (13).

With reference to (7), express the numerator 𝑐𝑜𝑠𝑖𝑛𝑒 factor as a difference of exponentials each of which in turn is written as a convergent power series in 𝜋(𝑐+𝑖𝑡), and recognize that the resulting series can be interchanged with the integral and each term identified using (9) to find 𝜁(𝑠)=−√𝜋^∞_𝑘=0−𝜋²/4^𝑘𝜂(𝑠−2𝑘)Γ(𝑘+1/2)Γ(𝑘+1),𝜎>1. (82) A more interesting variation of (82) can be obtained by applying (3) and explicitly identifying the 𝑘=0 term of the sum giving 𝜁(𝑠)1=−2∑^∞_𝑘=1−𝜋^2𝑘1−2^{1−𝑠+2𝑘}/Γ(2𝑘+1)𝜁(𝑠−2𝑘)1−2^−𝑠, (83) a convergent representation if 𝜎>1 (by Gauss' test). An immediate consequence is the identification 𝜁(2)𝜋=−²3𝜁(0), (84) because only the 𝑘=1 term in the sum contributes when 𝑠=2. Additionally, when 𝑠=2𝑛, (83) terminates at 𝑛 terms, becoming a recursion for 𝜁(2𝑛) in terms of 𝜁(2𝑛−2𝑘),𝑘=1,…,𝑛—see Section 8 for further discussion. If the functional equation (13) is used to extend its region of convergence, (83) becomes 𝜁(𝑠)=12𝜋^𝑠∞_𝑘=1𝜁(1−𝑠+2𝑘)2^{𝑠−2𝑘−1}−1Γ()Γ()×22𝑘+1𝑠−2𝑘_−𝑠−1cos𝜋𝑠2⁻¹,𝜎>1, (85) which augments similar series appearing in [10, Sections 4.2 and 4.3], and corresponds to a combination of entries in [28, Equations (54.6.3) and (54.6.4)].

5.2. By Splitting

5.2.1. The Lower Range

After applying (13), the integral in (42) is frequently (e.g., [11]) and conveniently split in two, defining 𝜁₋(𝑠)=^𝜔₀𝑡^𝑠sinh²(𝑡)𝜁𝑑𝑡,𝜎>1,₊(𝑠)=^∞_𝜔𝑡^𝑠sinh²(𝑡)𝑑𝑡 (86)(87) with 𝜁(𝑠)Γ(𝑠+1)2^𝑠−1=𝜁₋(𝑠)+𝜁₊(𝑠) (88) for arbitrary 𝜔. Similar to the derivation of (53), after application of (13) we have 𝜁₋(𝑠)=−2√𝜋^∞_𝑘=0𝐵_2𝑘𝜔^{2𝑘+𝑠−1}Γ()Γ(),𝑘−1/2𝑘+1)(2𝑘−1+𝑠 (89) which can be rewritten using (51) as 𝜁₋(𝑠)=4𝜔^∞𝑠−1_𝑘=0−𝜔²/𝜋^2𝑘Γ(𝑘+1/2)𝜁(2𝑘)Γ(𝑘−1/2)(2𝑘−1+𝑠),𝜔≤𝜋. (90) Substitute (1) in (90), interchange the order of summation, and identify the resulting series to obtain 𝜁₋(𝑠)=2𝑠𝜔^𝑠+1𝜋²()×𝑠+1_∞𝑚=11𝑚²²𝐹₁11,2+𝑠2;32+𝑠2𝜔;−²𝜋²𝑚²−𝜔^𝑠coth(𝜔)+𝑠𝜔^𝑠−1,𝑠−1 (91) a convergent representation that is valid for all 𝑠.

Other representations are easily obtained by applying any of the standard hypergeometric linear transforms as in (118)—see Section 6. One such leads to 𝜁₋(𝑠)=𝑠𝜔^𝑠−1Γ12+𝑠2×^∞_𝑘=0Γ(𝑘+1)Γ(𝑘+3/2+𝑠/2)^∞_𝑚=11𝜋²𝑚²/𝜔²+1^𝑘+1−𝜔^𝑠coth(𝜔)+𝑠𝜔^𝑠−1,𝑠−1 (92) a rapidly converging representation for small values of 𝜔, valid for all 𝑠. The inner sum of (92) can be written in several ways. By expanding the denominator as a series in 1/(𝑡²𝑚²) with 𝑟<2𝑝−1 and interchanging the two series, we find the general form 𝜅𝑟,𝑝,𝑡≡^∞_𝑚=1𝑚^𝑟1+𝑡²𝑚^2𝑝=^∞_𝑚=0Γ𝑝+𝑚(−1)^𝑚𝑡^{−2𝑚−2𝑝}𝜁2𝑚+2𝑝−𝑟Γ𝑝Γ(),𝑚+1 (93) where the right-hand side continues the left if |𝑡|>1 and 𝑟−2𝑝≥−1. Alternatively, [28, Equation (6.1.32)] identifies ^∞_𝑚=01𝑚²𝑡²+𝑦²=12𝑦²+𝜋2𝑡𝑦coth𝜋𝑦𝑡, (94) so for 𝑝=𝑘+1 and 𝑟=0 in (93), each term of the inner series (92) can be obtained analytically by taking the 𝑘th derivative of the right-hand side with respect to the variable 𝑦² at 𝑦=1, leading to a rational polynomial in coth(𝜔) and sinh(𝜔). The first few terms are given in Table 1 for two choices of 𝜔.

Table 1 

The first few values of 𝜅(0,𝜅+1,𝜋²/𝜔²) in (93).

Application of (93) to (92) leads to a less rapidly converging representation that can be expressed in terms of more familiar functions: 𝜁₋(𝑠)=𝑠𝜔^𝑠−1Γ12+𝑠2×^∞∞_𝑘=0𝑚=0(−1)^𝑚Γ(𝑚+𝑘+1)𝜁(2𝑚+2𝑘+2)Γ()Γ()×𝜔𝑘+3/2+𝑠/2𝑚+1₂𝜋^{2(𝑚+𝑘+1)}−𝜔^𝑠coth(𝜔)+𝑠𝜔^𝑠−1.𝑠−1 (95) A second transform of the intermediate hypergeometric function yields another useful result: 𝜁₋(𝑠)=2𝑠𝜔^𝑠−1Γ(1/2+𝑠/2)^∞_𝑘=0𝜔²𝜋^2(𝑘+1)×^∞_𝑚=0𝜔²𝜋^2𝑚(−1)^𝑚Γ1𝑚+𝑘+2+𝑠2()×(Γ()Γ(×𝜁2𝑚+2𝑘+2𝑚+1𝑘+1)(2𝑘+1+𝑠))⁻¹−𝜔^𝑠coth(𝜔)+𝑠𝜔^𝑠−1.𝑠−1 (96) Both (95) and (96) can be reordered along a diagonal of the double sum (∑^∞_𝑘=0∑^∞_𝑚=0𝑓(𝑚,𝑘)=∑^∞_𝑘=0∑^𝑘_𝑚=0𝑓(𝑚,𝑘−𝑚)), allowing one of the resulting (terminating) series to be evaluated in closed form. In either case, both eventually give 𝜁₋(𝑠)=−2𝑠𝜔^∞𝑠−1_𝑘=0−𝜔²𝜋^2𝑘𝜁(2𝑘)(2𝑘−1+𝑠)−𝜔^𝑠coth(𝜔), (97) reducing to (49) in the case 𝜔=𝜋 taking (13) into consideration. Series of the form (97) have been studied extensively [10, 11] when 𝑠=𝑛 for special values of 𝜔. These representations will find application in Section 8.

5.2.2. The Upper Range

Rewrite (87) in exponential form and expand the factor (1−𝑒^−2𝑡)⁻² in a convergent power series, interchange the sum and integral, and recognize that the resulting integral is a generalized exponential integral 𝐸_𝑠(𝑧) usually defined in terms of incomplete Gamma functions ([1, 35], Section 8.19) as follows: 𝐸_𝑠(𝑧)=𝑧^𝑠−1Γ(1−𝑠,𝑧), (98) giving a rapidly converging series representation: 𝜁₊(𝑠)=4𝜔^∞𝑠+1_𝑘=1𝑘𝐸_−𝑠(2𝜔𝑘). (99) A useful variant of (99) is obtained by manipulating the well-known recurrence relation ([1, Equation 8.19.12]) for 𝐸_𝑠(𝑧) to obtain 𝐸_−𝑠(𝑧)=1𝑧²e^−𝑧(𝑠+𝑧)+𝑠(𝑠−1)𝐸_2−𝑠(𝑧). (100) Apply (100) repeatedly 𝑁 times to (99) and find 𝜁₊(𝑠)=−𝑠𝜔^−1+𝑠log𝑒^2𝜔−1+2𝑠𝜔^𝑠−2𝜔^𝑠1−𝑒^2𝜔+2𝜔^𝑠Γ(𝑠+1)^𝑁+1_𝑘=2Li_𝑘𝑒^−2𝜔Γ(𝑠−𝑘+1)(2𝜔)^𝑘+2𝜔^{𝑠−𝑁−1}Γ(𝑠+1)Γ(𝑠−𝑁−1)2^∞𝑁+1_𝑘=1𝐸_{𝑁+2−𝑠}(2𝑘𝜔)𝑘^𝑁+1 (101) in terms of logarithms and polylogarithms (Li_𝑘(𝑧)), a result that loses numerical accuracy as 𝑁 increases, but which also, usefully, terminates if 𝑠=𝑛, 𝑁≥𝑛≥0. This is discussed in more detail in Section 8.

Another useful representation emerges from (99) by writing the function 𝐸_𝑠(2𝑘𝜔) in terms of confluent hypergeometric functions. From [35, Equation (2.6b)] and [36, Equation (9.10.3)], we find 𝜁₊(𝑠)=4𝜔^∞𝑠+1_𝑘=1𝑘𝐸_−𝑠(2𝑘𝜔)=2^1−𝑠Γ(𝑠+1)𝜁(𝑠)−4𝜔^𝑠+1×𝑠+1_∞𝑘=1𝑘𝑒₁^−2𝑘𝜔𝐹₁(1;2+𝑠;2𝑘𝜔), (102) where the series on the right-hand side converges absolutely if 𝜎>1 independent of 𝜔. Comparison of (99), (88), and (102) identifies 𝜁₋(𝑠)=4𝜔^𝑠+1𝑠+1^∞_𝑘=1𝑘𝑒₁^−2𝑘𝜔𝐹₁(1;2+𝑠;2𝑘𝜔) (103) which can be interpreted as a transformation of sums, or an analytic continuation, for the various guises of 𝜁₋(𝑠) given in Section 5.2.1, if both, or either, of the two sides converge for some choice of the variables 𝑠 and/or 𝜔, respectively. The fact that a term containing 𝜁(𝑠) naturally appears in the expression (102) for 𝜁₊(𝑠) suggests that there will be a severe cancellation of digits between the two terms 𝜁₋(𝑠) and 𝜁₊(𝑠) if (88) is used to attempt a numerical evaluation of 𝜁(𝑠), as has been reported elsewhere [37]. Essentially, (88) is numerically useless (with common choices of arithmetic precision) for any choices of 𝜌>10, even if 𝜔 is carefully chosen in an attempt to balance the cancellation of digits. However, with 20 digits of precision, using 20 terms in the series, it is possible to find the first zero of 𝜁(1/2+𝑖𝜌) correct to 10 digits, deteriorating rapidly thereafter as 𝜌 increases. For any choice of 𝜔, the convergence properties of the two sums representing 𝜁₋(𝑠) and 𝜁₊(𝑠) anti-correlate, as can be seen from Table 1. This can also be seen by rewriting (102) to yield the combined representation: 2^1−𝑠Γ(𝑠+1)𝜁(𝑠)=4𝜔^𝑠+1∞_𝑘=1𝑘𝐸_−𝑠(2𝑘𝜔)+¹𝐹₁(𝑠+1;2+𝑠;−2𝑘𝜔).𝑠+1 (104) In (104), the first inner term (𝐸_−𝑠(2𝑘𝜔)) vanishes like exp(−2𝑘𝜔)/(2𝑘𝜔) whereas the second term (₁𝐹₁/(𝑠+1)) vanishes like 1/𝑘^(1+𝑠)−exp(−2𝑘𝜔)/(2𝑘𝜔), so for large values of 𝑘 cancellation of digits is bound to occur. (In fact, the coefficients of both asymptotic series prefaced by the exponential terms cancel exactly term by term.)

It should be noted that (102) is representative of a family of related transformations. For example, although setting 𝑠=0 in (102) does not lead to a new result, operating first with 𝜕/𝜕𝑠 followed by setting 𝑠=0 leads to a transformation of a related sum: ^∞_𝑘=1𝐸₁()√2𝑘𝜔=−𝜔arctanh−𝜔²/𝜋𝜋√−𝜔²−^∞_𝑘=2𝜔arctanh√−𝜔²/𝜋𝑘𝜋𝑘√−𝜔²+12𝜔log𝜋+1+𝛾2𝜔2, (105) where the first term corresponding to 𝑘=1 has been isolated, since it contains the singularity that occurs due to divergence of the left-hand sum if 𝜔=𝑖𝜋. (In (86) and (87), 𝜔∈ℭ is permitted by a fundamental theorem of integral calculus.) Similarly, operating with 𝜕/𝜕𝑠 on (102) at 𝑠=1 leads to the identification ^∞_𝑘=11log1+𝜋²𝑘²=−1−log(2)+log𝑒²−1 (106) (which can alternatively be obtained by writing the infinite sum as a product). A second application of 𝜕/𝜕𝑠 on (102), again at 𝑠=1 using (106), leads to the following transformation of sums: 12^∞_𝑘=1𝜁(2𝑘)(−1)^𝑘𝜋^2𝑘𝑘²=12ln(2)²−𝛾²2+𝜋²(1)−12−1−𝛾^∞_𝑘=1𝐸₁(2𝑘)𝑘. (107) This transformation can be further reduced by applying an integral representation given in [1, Section (6.2)], leading to ^∞_𝑘=1𝐸₁(2𝑘)𝑘=−^∞₁log1−𝑒^−2𝑡𝑡𝑑𝑡. (108) Correcting a misprinted expression given in ([38, Equation (4.8)], [10, 11], Equation 3.3(43)) which should read ^∞_𝑘=1𝑡^𝑘𝑘²𝜁(2𝑘)=^𝑡₀log𝜋√𝑣csc𝜋√𝑣𝑑𝑣𝑣, (109) followed by a change of variables, eventually yields the representation ^∞_𝑘=1𝜁(2𝑘)(−1)^𝑘𝜋^2𝑘𝑘²=−2¹₀1𝑡1log2𝑒^𝑡−12𝑒^−𝑡−log(𝑡)𝑑𝑡. (110) Substituting (108) and (110) into (107) followed by some simplification finally gives ¹₀log1−𝑒^−2/𝑣𝑒^2𝑣−12𝑣𝑑𝑣𝑣𝜋=−²+𝛾122²(1)−1+𝛾2log²(2)+2, (111) a result that appears to be new.

5.3. Another Interpretation

These expressions for the split representation can be interpreted differently. For the moment, let 𝜎<−3, and insert the contour integral representation [35, Equation (2.6a)], for 𝐸_−𝑠(𝑧) into the converging series representation (99), noting that contributions from infinity vanish, giving 𝜁₊(𝑠)=−4𝜔^∞𝑠+1_𝑘=112𝜋𝑖^𝑐+𝑖∞_{𝑐−𝑖∞}Γ(−𝑡)(2𝑘𝜔)^𝑡𝑠+1+𝑡𝑑𝑡, (112) where the contour is defined by the real parameter 𝑐<−2. With this caveat, the series in (112) converges, allowing the integration and summation to be interchanged, leading to the identification 𝜁₊(𝑠)=−4𝜔^𝑠+12𝜋𝑖^𝑐+𝑖∞_{𝑐−𝑖∞}Γ(−𝑡)(2𝜔)^𝑡𝜁(−1−𝑡)𝑠+1+𝑡𝑑𝑡. (113) The contour may now be deformed to enclose the singularity at 𝑡=−𝑠−1, translated to the right to pick up residues of Γ(−𝑡) at 𝑡=𝑛,𝑛=0,1,…  and the residue of 𝜁(−1−𝑡) at 𝑡=−2. After evaluation, each of the residues so obtained cancels one of the terms in the series expression (90) for 𝜁₋(𝑠) leaving (after a change of integration variables) Γ(𝑠+1)𝜁(𝑠)2^𝑠−1𝜔^𝑠+11=−Γ()𝜁(𝑡)2𝜋𝑖𝑡+1(𝑠−𝑡)2^𝑡−1𝜔^𝑡+1𝑑𝑡, (114) the standard Cauchy representation of the left-hand side, valid for all 𝑠, where the contour encloses the simple pole at 𝑡=𝑠.

6. Integration by Parts

Integration by parts applied to previous results yields other representations.

From (17), we obtain (known to the Maple computer code, but only in the form of a Mellin transform) 𝜁(𝑠)1=−Γ(𝑠−1)^∞₀𝑡^𝑠−2log1−exp(−𝑡)𝑑𝑡,𝜎>1, (115) and from (64), we find 𝜁(𝑠)=12−𝑠𝜋×^∞₀log(sinh(𝜋𝑡))cos((s+1)arctan(𝑡))1+𝑡^2(1+𝑠)/2𝑑𝑡,𝜎>1 (116) as well as the following intriguing result (discovered by the Maple computer code): 𝜁(𝑠)=12+12√𝜋sin(𝜋𝑠/2)Γ(𝑠/2−1/2)Γ(𝑠/2)+𝜋^∞₀𝑡sin(𝑠arctan(𝑡))sinh²()×𝜋𝑡₂𝐹₁12,𝑠2;32;−𝑡²𝑑𝑡−𝑠^∞₀𝑡cosh(𝜋𝑡)cos(𝑠arctan(𝑡))sinh(𝜋𝑡)1+𝑡²×₂𝐹₁12,𝑠2;32;−𝑡²𝑑𝑡,𝜎>0. (117) This result has the interesting property that it is the only integral representation of which I am aware in which the independent variable 𝑠 does not appear as an exponent. Many variations of (117) can be obtained through the use of any of the transformations of the hypergeometric function, for example, ₂𝐹₁12,𝑠2;32;−𝑡²=1+𝑡^2−𝑠/2×₂𝐹₁𝑠1,2;32;𝑡²1+𝑡², (118) which choice happens to restore the variable 𝑠 to its usual role as an exponent.

7. Conversion to Integral Representations

From (79), identify the product 𝜁(⋯)Γ(⋯) in the form of an integral representation from one of the primary definitions of 𝜁(𝑠) ([13, Equation (9.511)]) and interchange the series and sum. The sums can now be explicitly evaluated, leaving an integral representation valid over an important range of 𝑠: 𝜁(𝑠)=𝜋^𝑠−14^𝑠sin(𝜋𝑠/2)2−2^1+𝑠+𝑠2^𝑠×^∞₀𝑡^−𝑠(𝑠−(2𝑠sinh(𝑡/2)/𝑡)−1+cosh(𝑡/2))𝑒^𝑡−1𝑑𝑡,𝜎<2. (119) Each of the terms in (119) corresponds to known integrals ([13, Equations (3.552.1) and (8.312.2)]), and if the identification is taken to completion, (119) reduces to the identity 𝜁(𝑠)=𝜁(𝑠) with the help of (13). Other useful representations are immediately available by evaluating the integrals in (119) corresponding to pairs of terms in the numerator, giving 𝜁(𝑠)=𝜋^𝑠2^𝑠−2(1−2^𝑠)()×()2cos𝜋𝑠/2sin𝜋𝑠^𝑠𝜋^∞₀𝑡^−𝑠(−1+cosh(𝑡/2))𝑒^𝑡+1−1𝑑𝑡Γ(𝑠)𝜁(𝑠),𝜎<2,=𝜋^𝑠−1sin𝜋𝑠2×^∞₀𝑣^−𝑠−1𝑣(𝑣)𝑒sinh−1^−𝑣−𝜋𝑑𝑣sin(𝜋𝑠)Γ(1+𝑠),𝜎<2. (120)(121)

8. The Case 𝑠=𝑛

Several of the results obtained in previous sections reduce to helpful representations when 𝑠=𝑛.

8.1. Finite Series Representations

From (79), we find, for even values of the argument, 𝜁()12𝑛=−2^𝑛_𝑘=1(−1)^𝑘𝜋^2𝑘(2𝑛−2𝑘−1)𝜁(2𝑛−2𝑘)Γ()×22𝑘+2_−2𝑛+𝑛−1⁻¹, (122) from (83), we find 𝜁()=12𝑛22^−2𝑛×−1_𝑛𝑘=1−𝜋^2𝑘1−2^{1−2𝑛+2𝑘}𝜁(2𝑛−2𝑘)Γ(),2𝑘+1 (123) and, based on (80), we find [39] a similar, but independent, result: 𝜁()22𝑛=2^−2𝑛1−2^1−2𝑛×^𝑛_𝑘=11−2^{2𝑛−2𝑘−1}(2𝜋)^2𝑘𝜁(2𝑛−2𝑘)(−1)^𝑘Γ(),2𝑘+2 (124) three of an infinite number of such recursive relationships [40]. The latter two may be identified with known results by reversing the sums, giving 𝜁(2𝑛)=−−𝜋^2𝑛2^1−2𝑛×+2𝑛−2_{𝑛−1𝑘=0}(2𝑘−1)𝜁(2𝑘)−𝜋^2𝑘Γ(),2𝑛−2𝑘+2 (125) thought to be new, 𝜁(2𝑛)=−𝜋^2𝑛2^1−2𝑛−2^𝑛−1_𝑘=01−2^1−2𝑘𝜁(2𝑘)−𝜋^2𝑘Γ(2𝑛−2𝑘+1) (126) equivalent to [11, Equation 4.4(11)], and 𝜁()=22𝑛−𝜋^2𝑛1−2^{1−2𝑛𝑛−1}_𝑘=01−2^2𝑘−1𝜁(2𝑘)(−1)^𝑘(2𝜋)^2𝑘Γ(),2𝑛−2𝑘+2 (127) equivalent to [11, Equation 4.4(9)].

8.2. Novel Series Representations

By adding and subtracting the first 𝑛−1 terms in the series representation of the hypergeometric function in (91), we find ([36, Chapter 9]), for the case 𝑠=2𝑛+2  (𝑛≥0), ₂𝐹₁1,𝑛+1/2;𝑛+3/2,−𝜔²/𝜋²𝑚²𝑚²=1𝑛+2−𝜔²𝜋²𝑚^2−𝑛×2𝜋𝜔𝑚𝜔arctan−1𝜋𝑚𝑚^2𝑛−1_𝑘=0−𝜔²/𝜋²𝑚^2𝑘,1/2+𝑘 (128) and for the case 𝑠=2𝑛+3₂𝐹₁1,𝑛+1;𝑛+2;−𝜔²/𝜋²𝑚²𝑚²=(𝑛+1)−𝜔²/𝜋²𝑚^2−𝑛×𝜋²𝜔²𝜔ln1+²𝜋²𝑚²−1𝑚^2𝑛−1_𝑘=0−𝜔²/𝜋²𝑚^2𝑘.𝑘+1 (129) Substituting (128) or (129) in (88), (91), and (101) gives series representations for 𝜁(2𝑛) and 𝜁(2𝑛+1), respectively, for which I find no record in the literature, with particular reference to works involving Barnes double Gamma function (e.g., [30]). For example, employing (128) in the case 𝑛=0, we obtain what appears to be a new representation (or the evaluation of a new sum): 𝜁(2)=−4𝜔^∞_𝑚=1𝜋𝑚𝜔𝜔arctan𝜋𝑚−1+Li₂𝑒^−2𝜔−2𝜔log−1+𝑒^2𝜔+3𝜔²+2𝜔. (130) The series (130) (compare with ([28, Equation (42.1.1)])) is absolutely convergent by Gauss' test for arbitrary values of 𝜔 and reduces to an identity in the limiting case 𝜔→0.

8.3. Connection with Multiple Gamma Function

From Section 5.2 with 𝜔=1, we have, after some simplification, 𝜁(2)=5−4^∞_𝑚=1(−1)^𝑚𝜋^(−2𝑚)𝜁(2𝑚)2𝑚+1+Li₂𝑒⁻²−2log𝑒²−1. (131) Comparison of the previous infinite series with [10, Equation 3.4(464)] using, 𝑧=𝑖/𝜋 identifies 𝐺()log1+𝑖/𝜋𝐺()=𝑖1−𝑖/𝜋()4𝜋6+4log2𝜋+2Li₂𝑒⁻²−4log𝑒²−𝜋−1²3=0.232662175652953𝑖, (132) where 𝐺 is Barnes double Gamma function. From [10, Section 1.3], unnumbered equation following 26, we also find 𝐺()log1+𝑖/𝜋𝐺()=()1−𝑖/𝜋𝑖log2𝜋𝜋−𝑖𝜋+^∞_𝑘=1−2𝑖𝜋+𝑘ln𝑘+𝑖/𝜋𝑘−𝑖/𝜋 (133) thereby recovering (130) after representing the arctan function in (130) as a logarithm (with 𝜔=1).

9. The Critical Strip

In (58) and (59) with 0≤𝜎≤1 let 𝑓(𝑠)𝜋=−^𝑠−1sin(𝜋𝑠/2),𝑔(𝑠)=2𝑠−1_𝑠−1Γ(),ℎ(𝑡)=√𝑠+1𝑡1sinh²𝑡−1𝑡². (134)(135)(136) Then, after adding and subtracting, we find a composition of symmetric and antisymmetric integrals about the critical line 𝜎=1/2: 𝜁(𝑠)=2𝑓(𝑠)𝑔(𝑠)𝑔(𝑠)−𝑓(𝑠)×^∞₀ℎ(𝑡)(𝑡)1cos𝜌lnsinh2(𝑡)+(𝑠)𝑔(𝑠)−𝜎ln𝑑𝑡2𝑖𝑓𝑓(𝑠)(𝑠)×−𝑔_∞0ℎ(𝑡)(𝑡)1sin𝜌lncosh2−𝜎ln(𝑡)𝑑𝑡, (137) and simultaneously 𝜁(𝑠)=2𝑓(𝑠)𝑔(𝑠)𝑓(𝑠)+𝑔(𝑠)×^∞₀ℎ(𝑡)(𝑡)1cos𝜌lncosh2(𝑡)−(𝑠)𝑔(𝑠)−𝜎ln𝑑𝑡2𝑖𝑓𝑓(𝑠)(𝑠)×+𝑔_∞0ℎ(𝑡)(𝑡)1sin𝜌lnsinh2−𝜎ln(𝑡)𝑑𝑡. (138) Let C𝜌=^∞₀ℎ(𝑡)(𝑡)𝑆𝜌=cos𝜌ln𝑑𝑡,_∞0ℎ(𝑡)sin𝜌ln(𝑡)𝑑𝑡. (139) For 𝜁(𝑠)=0 on the critical line 𝜎=1/2, both (137) and (138) can be simultaneously satisfied when both 𝐶(𝜌)≠0 and 𝑆(𝜌)≠0 only if 𝑓12+𝑖𝜌²=𝑔12+𝑖𝜌², (140) a true relationship that can be verified by direct computation. Set 𝜎=1/2 in (137) and (138) and find 𝜁12=𝑔+𝑖𝜌2𝑓1/2+𝑖𝜌1/2+𝑖𝜌𝑓𝐶𝜌=𝑔1/2+𝑖𝜌+𝑔1/2+𝑖𝜌2𝑖𝑓1/2+𝑖𝜌1/2+𝑖𝜌𝑓𝑆𝜌.1/2+𝑖𝜌−𝑔1/2+𝑖𝜌 (141) Equation (141) can be used to obtain a relationship between the real and imaginary components of 𝜁(1/2+𝑖𝜌). Let 𝜁12+𝑖𝜌=𝜁_𝑅12+𝑖𝜌+𝑖𝜁_𝐼12+𝑖𝜌. (142) Then, by breaking 𝑓(1/2+𝑖𝜌) and 𝑔(1/2+𝑖𝜌) into their real and imaginary parts, (141) reduces to 𝜁_𝑅1/2+𝑖𝜌𝜁_𝐼=𝐶1/2+𝑖𝜌_𝑚𝜌cos𝜌ln(2𝜋)−𝐶_𝑝𝜌sin𝜌ln(2𝜋)𝐶_𝑝𝜌cos𝜌ln(2𝜋)+𝐶_𝑚𝜌sin𝜌ln(2𝜋)−√𝜋, (143) where 𝐶_𝑝𝜌=Γ_𝑅12+𝑖𝜌cosh𝜋𝜌2+Γ_𝐼12+𝑖𝜌sinh𝜋𝜌2,𝐶_𝑚𝜌=Γ_𝐼12+𝑖𝜌cosh𝜋𝜌2−Γ_𝑅12+𝑖𝜌sinh𝜋𝜌2, (144)(145) in terms of the real and imaginary parts of Γ(1/2+𝑖𝜌). This is a known result [41] that depends only on the functional equation (13) (See the Appendix for further discussion on this point). In [41], it was shown that the numerator and denominator on the right-hand side of (11) can never vanish simultaneously, so this result defines the half-zeros (only one of 𝜁_𝑅(1/2+𝑖𝜌) or 𝜁_𝐼(1/2+𝑖𝜌) vanishes) (and usually brackets the full-zeros) of 𝜁(1/2+𝑖𝜌) according to whether the numerator or denominator vanishes. This was studied in [41]. The full-zeros (𝜁_𝑅(1/2+𝑖𝜌₀)=𝜁_𝐼(1/2+𝑖𝜌₀)=0) of 𝜁(1/2+𝑖𝜌) occur when the following condition 𝑆𝜌₀𝐶𝜌=0,₀=0 (146) is simultaneously satisfied for some value(s) of 𝜌=𝜌₀. In that case, the defining equation (143) becomes 𝜁_𝑅1/2+𝑖𝜌^′𝜁_𝐼1/2+𝑖𝜌^′𝜁=−^′1/2+𝑖𝜌_𝐼𝜁^′1/2+𝑖𝜌_𝑅=𝐶_𝑚𝜌cos𝜌ln(2𝜋)−𝐶_𝑝𝜌()×𝐶sin𝜌ln2𝜋_𝑝𝜌cos𝜌ln(2𝜋)+𝐶_𝑚𝜌sin𝜌ln(2𝜋)−√𝜋⁻¹, (147) where the derivatives are taken with respect to the variable 𝜌, and (147) is evaluated at 𝜌=𝜌₀. Many solutions to (146) are known [42], so there is no doubt that these two functions can simultaneously vanish at certain values of 𝜌. (A proof that this cannot occur in (137) and (138) when 𝜎≠1/2 is left as an exercise for the reader.)

A further interesting representation can be obtained inside the critical strip by setting 𝑐=𝜎 in (9) and separating 𝜂(𝑠) into its real and imaginary components: 𝜂_𝑅=1𝜎+𝜌𝑖2𝜎^(1−𝜎)∫^∞_−∞1+𝑡^2−𝜎/2𝑒^{𝜌arctan(𝑡)}(cos(Ω(𝑡))sin(𝜋𝜎)cosh(𝜋𝜎𝑡)−sin(Ω(𝑡))cos(𝜋𝜎)sinh(𝜋𝜎𝑡))cosh²(𝜋𝜎𝑡)−cos²(𝜋𝜎)𝑑𝑡,𝜂_𝐼1𝜎+𝜌𝑖=−2𝜎^(1−𝜎)∫^∞_−∞1+𝑡^2−𝜎/2𝑒^{𝜌arctan(𝑡)}(sin(Ω(𝑡))sin(𝜋𝜎)cosh(𝜋𝜎𝑡)+cos(Ω(𝑡))cos(𝜋𝜎)sinh(𝜋𝜎𝑡))cosh²(𝜋𝜎𝑡)−cos²(𝜋𝜎)𝑑𝑡 (148)(149)where Ω(𝑡)=12𝜌log𝜎²1+𝑡²+𝜎arctan(𝑡). (150) In analogy to (137) and (138), the representations (148) and (149) include a term cos(𝜋𝜎) that vanishes on the critical line 𝜎=1/2, both reducing to the equivalent of (26) in that case. (I am speculating here that such forms contain information about the distribution and/or existence of zeros of 𝜂(𝑠) and hence 𝜁(𝑠) inside the critical strip.) The integrands in both representations vanish as exp(−𝜋𝜎𝑡) as  𝑡→∞ and are finite at 𝑡=0; most of the numerical contribution to the integrals comes from the region 0<𝑡≅𝜌. Numerical evaluation in the range 0≤𝑡 can be acclerated by writing e^{𝜌arctan(𝑡)}→e^{𝜌(arctan(𝑡)−𝜋/2)}, (151) and extracting a factor e^𝜌𝜋/2 outside the integral in that range. It remains to be shown how the remaining integral conspires to cancel this factor in the asymptotic range 𝜌→∞. Finally, (148) is invariant and (149) changes sign under the transformations 𝜌→−𝜌 and 𝑡→−𝑡, demonstrating that 𝜂(𝑠) is self-conjugate.

10. Comments

A number of related results that are not easily categorized and which either do not appear in the reference literature or appear incorrectly should be noted.

Comparison of (49) and (54) suggests the identification 𝑣=𝑖. Such sums and integrals have been studied [30, 43] for the case 𝑠=−𝑛 in terms of Barnes double Gamma function 𝐺(𝑧) and Clausen's function defined by Cl₂(𝑧)=^∞_𝑘=1sin(𝑧𝑘)𝑘², (152) with the (tacit) restriction that 𝑧∈ℜ. That is not the case here. Thus straightforward application of that analysis will sometimes lead to incorrect results. For example [30, Equation (4.3)], gives (with 𝑛=1) 𝜋^𝑡₀𝑣^𝑛()()𝐺()cot𝜋𝑣𝑑𝑣=𝑡log2𝜋+log1−𝑡𝐺(),1+𝑡 (153) further ([43, Equation (13)]) identifying 𝐺()log1−𝑡𝐺()()1+𝑡=𝑡logsin𝜋𝑡𝜋+12𝜋Cl₂(2𝜋𝑡),0<𝑡<1. (154) To allow for the possibility that 𝑡∈ℭ, this should instead read 𝐺()log1−𝑡𝐺()()1+𝑡=𝑡logsin𝜋𝑡𝜋+𝑖𝜋2𝐵₂(𝑡)+e^{−𝑖𝜋/2}2𝜋Li₂e^{2𝜋𝑖𝑡}, (155) which is valid for all 𝑡 in terms of Bernoulli polynomials 𝐵_𝑛(𝑡) and polylogarithm functions. However, similar results are given in the equivalent, general form in [20, Equation (5.12)]. Interestingly, both the Maple and Mathematica computer codes somehow (neither [44] nor references contained therein, both of which study such integrals, list the computer results, which are likely based on [20, Equation (5.12)]) manage to evaluate the integral in (153) in terms similar to (155) for a variety of values of 𝑛 provided that the user restricts |𝑡|<1, although the right-hand side provides the analytic continuation of the left when that restriction is lifted. This has application in the evaluation of 𝜁(3) according to the method of Section 8.3, where, using the methods of [10, Chapter 3], we identify ^∞_𝑘=1𝜁(2𝑘)𝑡^(2𝑘)=1𝑘+12−𝜋𝑡^2𝑡₀𝑣²cot(𝜋𝑣)𝑑𝑣, (156) a generalization of integrals studied in [44]. From the computer evaluation of the integral, (156) eventually yields ^∞_𝑘=1𝜁(2𝑘)𝑡^2𝑘=1𝑘+131𝑖𝜋𝑡+2−log1−𝑒^{2𝑖𝜋𝑡}+𝑖Li₂𝑒^{2𝑖𝜋𝑡}−𝜋𝑡Li₃𝑒^{2𝑖𝜋𝑡}−𝜁(3)2𝜋²𝑡², (157) as a new addition to the lists found in [10, Section (3.4)], and [11, Sections (3.3) and (3.4)], and a generalization of (97) when 𝑠=3.

Appendix

In [41], an attempt was made to show that the functional equation (13) and a minimal number of analytic properties of 𝜁(𝑠) might suffice to show that 𝜁(𝑠)≠0 when 𝜎≠1/2. This was motivated by a comment attributed to H. Hamburger ([8, page 50]) to the effect that the functional equation for 𝜁(𝑠) perhaps characterizes it completely. (An unequivocal equivalent claim can also be found in [15]. As demonstrated by the counterexample given here, and in [45], this claim is false.) The purported demonstration contradicts a result of Gauthier and Zeron [45], who show that infinitely many functions exist, arbitrarily close to 𝜁(𝑠), that satisfy (13) and the same analytic properties that are invoked in [41], but which vanish when 𝜎≠1/2. (To quote Prof. Gauthier: “We prove the existence of functions approximating the Riemann zeta-function, satisfying the Riemann functional equation and violating the analogue of the Riemann hypothesis.” (private communication).)

In fact, an example of such a function is easily found. Let 𝜈(𝑠)=𝐶sin𝜋𝑠−𝑠₀sin𝜋𝑠+𝑠₀×sin𝜋𝑠−𝑠^∗₀sin𝜋𝑠+𝑠^∗₀, (A.1) where 𝑠^∗₀ is the complex conjugate of 𝑠₀, 𝑠₀=𝜎₀+𝑖𝜌₀ is arbitrary complex, and 𝐶, given by 1𝐶=cosh²𝜋𝜌₀−cos²𝜋𝜎₀², (A.2) is a constant chosen such that 𝜈(1)=1. Then 𝜁_𝜈(𝑠)≡𝜈(𝑠)𝜁(𝑠) (A.3) satisfies analyticity conditions (1−3) of [45] (which include the functional equation (13)). Additionally, 𝜁_𝜈(𝑠) has zeros at the zeros of 𝜁(𝑠) as well as complex zeros at 𝑠=±(𝑛+𝑠₀) and 𝑠=±(𝑛+𝑠^∗₀) and a pole at 𝑠=1 with residue equal to unity, where 𝑛=0,±1,±2,….

Thus the last few paragraphs of Section 2 of [41] beginning with the words “There are two cases where (2.9) fails…” cannot be valid. However, the remainder of that paper, particularly those sections dealing with the properties of 𝜁(1/2+𝑖𝜌) as referenced here in Section 9, are valid and correct. I am grateful to Professor Gauthier for discussions on this topic.

Acknowledgments

The author wishes to thank Vinicius Anghel and Larry Glasser for commenting on a draft copy of this paper. Leslie M. Saunders obituary may be found in the journal Physics Today, issue of Dec. 1972, page 63, http://www.physicstoday.org/resource/1/phtoad/v25/i12/p63_s2?bypassSSO=1.