Skip to main content

Alternating double t-values and T-values

Abstract

Recently, Hoffman (Commun. Number Theory Phys. 13:529–567, 2019), Kaneko and Tsumura (Tsukuba J. Math. (in press), 2020) introduced and systematically studied two variants of multiple zeta values of level two, i.e., multiple t-values and multiple T-values, respectively. In this paper, by the contour integration and residue theorem, we establish two general identities, which further reduce to the expressions of the alternating double t-values and T-values. Some examples are also provided.

Introduction and notations

For positive integers n and p, let \(H^{(p)}_{n}\) and \({\overline{H}}^{(p)}_{n}\) stand for the nth generalized harmonic number and the nth generalized alternating harmonic number defined by

$$\begin{aligned} H_{n}^{(p)}: = \sum_{k=1}^{n} {\frac{1}{{{k^{p}}}}} \quad \text{and} \quad {\bar{H}}_{n}^{(p)}: = \sum_{k=1}^{n} { \frac{(-1)^{k-1}}{{{k^{p}}}}}, \end{aligned}$$

respectively. If \(p>1\) (or resp. \(p>0\)), the generalized harmonic number \(H^{(p)}_{n}\) (or resp. \({\bar{H}}_{n}\)) converges to the (Riemann) zeta value \(\zeta (p)\) (or resp. alternating zeta value \({\bar{\zeta }}(p)\)):

$$ \lim_{n\rightarrow \infty }H^{(p)}_{n}=\zeta (p)\quad \Bigl( \text{or resp. }\lim_{n\rightarrow \infty }{\bar{H}}^{(p)}_{n}={ \bar{\zeta }}(p)\Bigr). $$

When \(k=1\), \(H_{n}^{(1)}\equiv H_{n}\) (resp. \(\overline{H}_{n}^{(1)}\equiv \overline{H}_{n}\)) is the classical harmonic number (resp. the classical alternating harmonic number). The empty sums \(H^{(p)}_{0}\) and \({\bar{H}}^{(p)}_{0}\) are conventionally understood to be zero.

For positive integers \(p_{1},\ldots ,p_{k}\) with \(p_{1}>1\), the multiple zeta value (MZV for short) is defined by

$$\begin{aligned} \zeta (p_{1},p_{2},\ldots ,p_{k}):=\sum_{n_{1}>\cdots >n_{k}\geq 1} \frac{1}{n_{1}^{p_{1}}n_{2}^{p_{2}}\cdots n_{k}^{p_{k}}}. \end{aligned}$$
(1.1)

The study of multiple zeta values began in the early 1990s with the works of Hoffman [4] and Zagier [16]. The study of multiple zeta values have attracted numerous research interests in the area in the last two decades. For detailed history and applications, please see the book of Zhao [17].

Let \(h_{n}^{(p)}\) be the nth odd harmonic number, which is defined for \(n\in \mathbb{N}_{0}\) and \(p\in \mathbb{N}\) by

$$\begin{aligned} h_{n}^{(p)}:=\sum_{k=1}^{n} \frac{1}{(k-1/2)^{p}},\qquad h_{0}^{(p)}:=0, \qquad h_{n}:=h_{n}^{(1)}. \end{aligned}$$
(1.2)

If \(p>1\), the generalized harmonic number \(h^{(p)}_{n}\) converges to the -value:

$$\begin{aligned} \lim_{n\rightarrow \infty }h^{(p)}_{n}= \widetilde{t}(p)=\sum_{k=1}^{\infty } \frac{1}{(k-1/2)^{p}}. \end{aligned}$$
(1.3)

A twin sibling of the odd harmonic number is called alternating odd harmonic number, defined by

$$\begin{aligned} { \bar{h}}_{n}^{(p)}: = \sum _{k=1}^{n} { \frac{(-1)^{k-1}}{{{(k-1/2)^{p}}}}},\qquad { \bar{h}}_{0}^{(p)}:=0, \qquad {\bar{h}}_{n}:={ \bar{h}}^{(1)}_{n}, \end{aligned}$$
(1.4)

which was introduced in [14]. When taking the limit \(n\rightarrow \infty \) in above, we get the so-called alternating -value

$$\begin{aligned} {\bar{t}}(p): = \sum_{k = 1}^{\infty }{ \frac{{{{ ( { - 1} )}^{k - 1}}}}{{{(k-1/2)^{p}}}}}\quad (p\geq 1). \end{aligned}$$
(1.5)

Note that from [3], for nonnegative integer k, we have the generating function of \({\bar{t}}(2k+1)\)

$$\begin{aligned} \frac{\pi }{\cos (\pi s)}=2\sum_{k=0}^{\infty } { \bar{t}}(2k+1)s^{2k}= \sum_{k=0}^{\infty } \frac{(-1)^{k}E_{2k}\pi ^{2k+1}}{(2k)!}s^{2k}, \end{aligned}$$

where \(E_{2k}\) is the Euler number. Thus, we compute

$$\begin{aligned} {\bar{t}}(2k+1)=\frac{(-1)^{k}E_{2k}\pi ^{2k+1}}{2(2k)!}\quad (k\geq 0). \end{aligned}$$

In a recent paper [5], Hoffman introduced and studied the more general multiple t-values

$$\begin{aligned} t(p_{1},p_{2},\ldots ,p_{k})&:= \sum_{{n_{1}>\cdots >n_{k}\geq 1\atop n_{i} \text{ odd}}} \frac{1}{n_{1}^{p_{1}}n_{2}^{p_{2}}\cdots n_{k}^{p_{k}}} \\ &=\sum_{n_{1}>\cdots >n_{k}\geq 1} \frac{1}{(2n_{1}-1)^{p_{1}}(2n_{2}-1)^{p_{2}}\cdots (2n_{k}-1)^{p_{k}}}. \end{aligned}$$
(1.6)

As the normalized version,

$$\begin{aligned} \widetilde{t}(p_{1},p_{2},\ldots ,p_{k}):=2^{p_{1}+\cdots +p_{k}}t(p_{1},p_{2}, \ldots ,p_{k}), \end{aligned}$$
(1.7)

we call them multiple -values.

Kaneko and Tsumura [6, 7] introduced and studied a new kind of multiple zeta values of level two:

$$\begin{aligned} T(p_{1},p_{2},\ldots ,p_{k})&:=2^{k} \sum_{{m_{1}>m_{2}>\cdots >m_{k}>0 \atop m_{i}\equiv k-i+1\ {\mathrm{mod}}\ 2}} \frac{1}{m_{1}^{p_{1}}m_{2}^{p_{2}}\cdots m_{k}^{p_{k}}} \\ &=2^{k}\sum_{n_{1}>n_{2}>\cdots >n_{k}>0} \frac{1}{(2n_{1}-k)^{p_{1}}(2n_{2}-k+1)^{p_{2}}\cdots (2n_{k}-1)^{p_{k}}}, \end{aligned}$$
(1.8)

which were called multiple T-values (MTVs). As the normalized version,

$$\begin{aligned} \widetilde{T}(p_{1},p_{2},\ldots ,p_{k}):=2^{p_{1}+\cdots +p_{k}-k}T(p_{1},p_{2}, \ldots ,p_{k}), \end{aligned}$$
(1.9)

we call them multiple -values.

In (1.1) and (1.6)–(1.9), we put a bar on the top of \(p_{j}\) (\(j=1,\ldots k\)) if there is a sign \((-1)^{n_{j}}\) appearing in the denominator on the right. These with one or more \(p_{j}\) barred are called the alternating MZVs, alternating multiple t-values, alternating multiple -values, alternating multiple T-values, and alternating multiple -values, respectively. For example,

$$\begin{aligned} &\zeta (p_{1},{\bar{p}}_{2},p_{3},{ \bar{p}}_{4})=\sum_{n_{1}>n_{2}>n_{3}>n_{4}>0} \frac{(-1)^{n_{2}+n_{4}}}{n^{p_{1}}_{1}n^{p_{2}}_{2}n^{p_{3}}_{3}n^{p_{4}}_{4}}, \\ &t({\bar{p}}_{1},{\bar{p}}_{2},p_{3},p_{4})= \sum_{n_{1}>n_{2}>n_{3}>n_{4}>0} \frac{(-1)^{n_{1}+n_{2}}}{(2n_{1}-1)^{p_{1}}(2n_{2}-1)^{p_{2}}(2n_{3}-1)^{p_{3}}(2n_{4}-1)^{p_{4}}}, \\ &T({\bar{p}}_{1},p_{2},{\bar{p}}_{3},p_{4})=2^{4} \sum_{n_{1}>n_{2}>n_{3}>n_{4}>0} \frac{(-1)^{n_{1}+n_{3}}}{(2n_{1}-4)^{p_{1}}(2n_{2}-3)^{p_{2}}(2n_{3}-2)^{p_{3}}(2n_{4}-1)^{p_{4}}}. \end{aligned}$$

In all of these definitions, we call k the “depth” and \(p_{1}+\cdots +p_{k}\) the “weight”.

The motivation for this paper arises from the results of Flajolet and Salvy’s paper [2] and Wang and Xu’s papers [12, 14]. In [2], Flajolet and Salvy used the methods of contour integration and residue theorem to determine the reducibility of some classical Euler sums. Similarly, in [12, 14], Wang and Xu used the contour integration and residue theorem to evaluate (alternating) Euler sums and Euler T-sums. There have been numerous contributions on the theory of Euler sums in the last two decades, for example, see [1, 8, 9, 11, 13, 15] and the references therein.

The main purpose of this paper is to study the four (alternating) double t-values

$$ \widetilde{t}(q,p), \widetilde{t}(q,\bar{p}), \widetilde{t}( \bar{q},\bar{p}), \widetilde{t}(\bar{q},p) $$

and the four (alternating) double T-values

$$ \widetilde{T}(q,p), \widetilde{T}(q,\bar{p}), \widetilde{T}(\bar{q}, \bar{p}), \widetilde{T}(\bar{q},p) $$

by using the methods of contour integration and residue theorem.

Double t-values and T-values

In this section, we give explicit evaluations for some (alternating) double t-values and T-values. We will prove these results in Sect. 4.

Theorem 2.1

For positive integers p and \(q>1\),

$$\begin{aligned} &\begin{aligned}[b] &\bigl(1-(-1)^{p+q} \bigr) \sum_{n=1}^{\infty }\frac{h^{(p)}_{n-1}}{(n-1/2)^{q}}\\ &\quad = \bigl(1-(-1)^{p+q}\bigr) \widetilde{t}(q,p) \\ &\quad =(-1)^{p+q}\widetilde{t}(p+q)-(-1)^{p}\bigl(1+(-1)^{q} \bigr)\widetilde{t}(p) \widetilde{t}(q) \\ &\qquad{}-(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}-1\bigr)\binom{p+q-k-2}{q-1} \widetilde{t}(k+1)\zeta (p+q-k-1) \\ &\qquad{}+2(-1)^{p} \sum_{{2k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \binom{k_{2}+p-2}{p-1}\widetilde{t}(2k_{1})\zeta (k_{2}+p-1), \end{aligned} \end{aligned}$$
(2.1)
$$\begin{aligned} &\begin{aligned}[b] &\bigl(1+(-1)^{p+q} \bigr) \sum_{n=1}^{\infty }\frac{{\bar{h}}^{(p)}_{n-1}}{(n-1/2)^{q}}\\ &\quad =- \bigl(1+(-1)^{p+q}\bigr) \widetilde{t}(q,{\bar{p}}) \\ &\quad =-(-1)^{p+q}{\bar{t}}(p+q)+(-1)^{p}\bigl(1+(-1)^{q} \bigr){\bar{t}}(p) \widetilde{t}(q) \\ &\qquad{}-(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}+1\bigr)\binom{p+q-k-2}{q-1}{\bar{t}}(k+1) \zeta (p+q-k-1) \\ &\qquad{}-2(-1)^{p}\sum_{{2k_{1}+k_{2}=q+2,\atop k_{1},k_{2}\geq 1}} \binom{k_{2}+p-2}{p-1}{\bar{t}}(2k_{1}-1){\bar{\zeta }}(k_{2}+p-1), \end{aligned} \end{aligned}$$
(2.2)
$$\begin{aligned} &\begin{aligned}[b] &\bigl(1-(-1)^{p+q} \bigr) \sum_{n=1}^{\infty }\frac{{\bar{h}}^{(p)}_{n-1}}{(n-1/2)^{q}}(-1)^{n-1}\\ &\quad = \bigl(1-(-1)^{p+q}\bigr) \widetilde{t}({\bar{q}},{\bar{p}}) \\ &\quad=(-1)^{p+q}\widetilde{t}(p+q)+(-1)^{p}\bigl(1-(-1)^{q} \bigr){\bar{t}}(p){ \bar{t}}(q) \\ &\qquad{}+(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}-1\bigr)\binom{p+q-k-2}{q-1} \widetilde{t}(k+1){\bar{\zeta }}(p+q-k-1) \\ &\qquad{}-2(-1)^{p} \sum_{{2k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \binom{k_{2}+p-2}{p-1}\widetilde{t}(2k_{1}){\bar{\zeta }}(k_{2}+p-1), & \end{aligned} \end{aligned}$$
(2.3)
$$\begin{aligned} &\begin{aligned}[b] &\bigl(1+(-1)^{p+q} \bigr) \sum_{n=1}^{\infty }\frac{{h}^{(p)}_{n-1}}{(n-1/2)^{q}}(-1)^{n-1}\\ &\quad =- \bigl(1+(-1)^{p+q}\bigr) \widetilde{t}({\bar{q}},p) \\ &\quad =-(-1)^{p+q}{\bar{t}}(p+q)-(-1)^{p}\bigl(1-(-1)^{q} \bigr)\widetilde{t}(p){ \bar{t}}(q) \\ &\qquad{}+(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}+1\bigr)\binom{p+q-k-2}{q-1}{\bar{t}}(k+1){ \bar{\zeta }}({p+q-k-1}) \\ &\qquad{}+2(-1)^{p}\sum_{{2k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \binom{p+k_{2}-2}{p-1}{\bar{t}}(2k_{1}-1)\zeta (p+k_{2}-1), \end{aligned} \end{aligned}$$
(2.4)

where \(\zeta (1):=-2\log (2)\)and \(\widetilde{t}(1):=0\).

Remark 2.2

Note that formulas (2.1) and (2.4) can also be found in Xu and Wang [14].

Example 2.1

We have

$$\begin{aligned} &\widetilde{t}(3,\bar{1})=\frac{1}{2}{\bar{t}}(4)-\frac{7}{8}\pi \zeta (3)- \frac{1}{4}\pi ^{3}\log (2), \\ &\widetilde{t}(2,\bar{2})=\frac{1}{2}{\bar{t}}(4)-3\zeta (2){\bar{t}}(2)+ \frac{7}{4}\pi \zeta (3), \\ &\widetilde{t}(\bar{2},\bar{1})=3\zeta (2)\log (2)-\frac{7}{2}\zeta (3), \\ &\widetilde{t}(\bar{3},1)=\frac{1}{2}{\bar{t}}(4)-\frac{1}{2}\log (2) \pi ^{3}+\frac{7}{8}\pi \zeta (3), \\ &\widetilde{t}(\bar{3},\bar{2})=-\frac{31}{2}\zeta (5)+\frac{1}{4} \pi ^{3}{ \bar{t}}(2)-\frac{27}{4}\zeta (2)\zeta (3). \end{aligned}$$

Theorem 2.3

For positive integers p and \(q>1\),

$$\begin{aligned} &\begin{aligned}[b] &\bigl(1-(-1)^{p+q} \bigr)\sum_{n=1}^{\infty }\frac{h^{(p)}_{n}}{n^{q}}\\ &\quad = \bigl(1-(-1)^{p+q}\bigr) \widetilde{T}(q,p) \\ &\quad =-(-1)^{p}\bigl(1+(-1)^{q}\bigr)\widetilde{t}(p)\zeta (q)-(-1)^{p} \binom{p+q-1}{p-1}\widetilde{t}(p+q) \\ &\qquad{}-(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}-1\bigr)\binom{p+q-k-2}{q-1} \widetilde{t}(k+1) \widetilde{t}(p+q-k-1) \\ &\qquad{}+(-1)^{p} \sum_{{k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \bigl(1+(-1)^{k_{1}}\bigr) \binom{k_{2}+p-2}{p-1}\zeta (k_{1}) \widetilde{t}(k_{2}+p-1)), \end{aligned} \end{aligned}$$
(2.5)
$$\begin{aligned} &\begin{aligned}[b] &\bigl(1+(-1)^{p+q} \bigr)\sum_{n=1}^{\infty }\frac{{\bar{h}}^{(p)}_{n}}{n^{q}}\\ &\quad =- \bigl(1+(-1)^{p+q}\bigr) \widetilde{T}(q,{\bar{p}}) \\ &\quad =(-1)^{p}\bigl(1+(-1)^{q}\bigr){\bar{t}}(p)\zeta (q)+(-1)^{p}\binom{p+q-1}{p-1}{ \bar{t}}(p+q) \\ &\qquad{}-(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}+1\bigr)\binom{p+q-k-2}{q-1}{\bar{t}}(k+1) \widetilde{t}(p+q-k-1) \\ &\qquad{}+(-1)^{p} \sum_{{k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \bigl(1+(-1)^{k_{1}}\bigr) \binom{k_{2}+p-2}{p-1}{\bar{\zeta }}(k_{1}){\bar{t}}(k_{2}+p-1)), \end{aligned} \end{aligned}$$
(2.6)
$$\begin{aligned} &\bigl(1+(-1)^{p+q} \bigr)\sum_{n=1}^{\infty }\frac{{\bar{h}}^{(p)}_{n}}{n^{q}}(-1)^{n-1} \\ &\quad =- \bigl(1+(-1)^{p+q}\bigr) \widetilde{T}({\bar{q}},{\bar{p}}) \\ &\quad =(-1)^{p}\bigl(1+(-1)^{q}\bigr){\bar{t}}(p){\bar{\zeta }}(q)-(-1)^{p} \binom{p+q-1}{p-1}{\bar{t}}(p+q) \\ &\qquad{}-(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}-1\bigr)\binom{p+q-k-2}{q-1}{ \widetilde{t}}(k+1){ \bar{t}}(p+q-k-1) \\ &\qquad{}+(-1)^{p} \sum_{{k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \bigl(1+(-1)^{k_{1}}\bigr) \binom{k_{2}+p-2}{p-1}{\zeta }(k_{1}){ \bar{t}}(k_{2}+p-1)), \end{aligned}$$
(2.7)
$$\begin{aligned} &\begin{aligned}[b] &\bigl(1-(-1)^{p+q} \bigr)\sum_{n=1}^{\infty }\frac{h^{(p)}_{n}}{n^{q}}(-1)^{n-1}\\ &\quad = \bigl(1-(-1)^{p+q}\bigr) \widetilde{T}({\bar{q}},p) \\ &\quad =-(-1)^{p}\bigl(1+(-1)^{q}\bigr)\widetilde{t}(p){\bar{ \zeta }}(q)+(-1)^{p} \binom{p+q-1}{p-1}\widetilde{t}(p+q) \\ &\qquad{}-(-1)^{p}\sum_{k=0}^{p-1} \bigl((-1)^{k}+1\bigr)\binom{p+q-k-2}{q-1}{\bar{t}}(k+1){ \bar{t}}(p+q-k-1) \\ &\qquad{}+(-1)^{p} \sum_{{k_{1}+k_{2}=q+1,\atop k_{1},k_{2}\geq 1}} \bigl(1+(-1)^{k_{1}}\bigr) \binom{k_{2}+p-2}{p-1}{\bar{\zeta }}(k_{1})\widetilde{t}(k_{2}+p-1)), \end{aligned} \end{aligned}$$
(2.8)

where \(\zeta (1):=-2\log (2)\)and \(\widetilde{t}(1):=0\).

Remark 2.4

Note that the explicit evaluation of \(T(q,p)\) with odd weight was also proved by Kanenko and Tsumura [6, 7] by another method.

Example 2.2

We have

$$\begin{aligned} &\widetilde{T}(4,1)=\frac{31}{2}\zeta (5)-7\zeta (2)\zeta (3), \\ &\widetilde{T}(3,2)=-62\zeta (5)+35\zeta (2)\zeta (3), \\ &\widetilde{T}(3,\bar{1})=\frac{1}{2}{\bar{t}}(4)+\frac{1}{2}\zeta (2){ \bar{t}}(2)-\frac{7}{2}\pi \zeta (3), \\ &\widetilde{T}(2,\bar{2})=-\frac{3}{2}{\bar{t}}(4)-\frac{3}{2} \zeta (2){ \bar{t}}(2)+7\pi \zeta (3), \\ &\widetilde{T}(\bar{3},\bar{1})=-\frac{1}{2}{\bar{t}}(4)+\zeta (2){ \bar{t}}(2), \\ &\widetilde{T}(\bar{2},\bar{2})=\frac{3}{2}{\bar{t}}(4)- \frac{9}{2} \zeta (2){\bar{t}}(2), \\ &\widetilde{T}(\bar{4},1)=\frac{1}{2}\pi {\bar{t}}(4)-\frac{31}{2} \zeta (5)-\frac{7}{2}\zeta (2)\zeta (3), \\ &\widetilde{T}(\bar{3},2)=-\frac{3}{2}\pi {\bar{t}}(4)+62\zeta (5)+7 \zeta (2)\zeta (3). \end{aligned}$$

Notations and related expansions

In this section, we give some basic notations, definitions, and lemmas. Let \(A:=\{a_{k}\}\), \(-\infty < k < \infty \) be a sequence of complex numbers with \({a_{k}} = o ( {{k^{\alpha }}} )\) (\(\alpha < 1\)) if \(k\rightarrow \pm \infty \). For convenience, let \(A_{1}\) and \(A_{2}\) denote the constant sequence \(\{(1)^{k}\}\) and the alternating sequence \(\{(-1)^{k}\}\), respectively.

Notations and definitions

Now, we give three definitions.

Definition 3.1

With A defined above, we define the parametric digamma function \(\varPhi ( { - s;A} )\) by

$$\begin{aligned} \varPhi ( { - s;A} ):= \frac{{{a_{0}}}}{s} + \sum_{k = 1}^{\infty }{ \biggl( {\frac{{{a_{k}}}}{k-1/2} - \frac{{{a_{k}}}}{{k - s}}} \biggr)}. \end{aligned}$$
(3.1)

Definition 3.2

For nonnegative integers \(j\geq 1\) and n, we define

$$\begin{aligned} &D^{(A)}(j):=\sum_{k=1}^{\infty } \frac{a_{k}}{k^{j}},\qquad E^{(A)}_{n}(j):= \sum _{k=1}^{n}\frac{a_{n-k}}{k^{j}},\qquad E^{(A)}_{0}(j):=0, \\ &F^{(A)}_{n}(j)= \textstyle\begin{cases} \sum_{k=1}^{\infty }\frac{a_{k+n}-a_{k}}{k} ,& j=1, \\ \sum_{k=1}^{\infty }\frac{a_{k+n}}{k^{j}},& j>1, \end{cases}\displaystyle \qquad \overrightarrow{F_{n}}^{(A)}(j)= \textstyle\begin{cases} \sum_{k=1}^{\infty } (\frac{a_{k+n}}{k}-\frac{a_{k}}{k-1/2} ) ,& j=1, \\ F^{(A)}_{n}(j),& j>1, \end{cases}\displaystyle \\ &{\bar{E}}^{(A)}_{n}(j):=\sum_{k=1}^{n} \frac{a_{k-n-1}}{k^{j}},\qquad { \bar{E}}^{(A)}_{0}(j):=0,\qquad \widehat{E}^{(A)}_{n}(j):=\sum_{k=1}^{n} \frac{a_{n-k}}{(k-1/2)^{j}},\qquad \widehat{E}^{(A)}_{0}(j):=0, \\ &{\bar{F}}^{(A)}_{n}(j)= \textstyle\begin{cases} \sum_{k=1}^{\infty }\frac{a_{k-n}-a_{k}}{k} ,& j=1, \\ \sum_{k=1}^{\infty }\frac{a_{k-n}}{k^{j}},& j>1, \end{cases}\displaystyle \qquad \widehat{F}^{(A)}_{n}(j)= \textstyle\begin{cases} \sum_{k=1}^{\infty }\frac{a_{k+n}-a_{k}}{k-1/2} ,& j=1, \\ \sum_{k=1}^{\infty }\frac{a_{k+n}}{(k-1/2)^{j}},& j>1, \end{cases}\displaystyle \\ &\widetilde{E}^{(A)}_{n}(j):=\sum _{k=1}^{n} \frac{a_{k-n-1}}{(k-1/2)^{j}},\qquad \widetilde{E}^{(A)}_{0}(j):=0, \qquad \widetilde{F}^{(A)}_{n}(j)= \textstyle\begin{cases} \sum_{k=1}^{\infty }\frac{a_{k-n}-a_{k}}{k-1/2} ,& j=1, \\ \sum_{k=1}^{\infty }\frac{a_{k-n}}{(k-1/2)^{j}},& j>1, \end{cases}\displaystyle \\ &G^{(A)}_{n}(j):=E^{(A)}_{n}(j)-{ \bar{E}}^{(A)}_{n-1}(j)- \frac{a_{0}}{n^{j}},\qquad G^{(A)}_{0}(j):=0,\qquad L^{(A)}_{n}(j):=F^{(A)}_{n}(j)+(-1)^{j}{ \bar{F}}^{(A)}_{n}(j), \\ &M^{(A)}_{n}(j):=E^{(A)}_{n}(j)+(-1)^{j} \overrightarrow{F_{n}}^{(A)}(j), \qquad N^{(A)}_{n}(j):= \widehat{E}^{(A)}_{n}(j)+(-1)^{j} \widehat{F}^{(A)}_{n-1}(j), \\ &{\bar{N}}^{(A)}_{n}(j):=\widetilde{F}^{(A)}_{n}(j)- \widetilde{E}^{(A)}_{n-1}(j), \qquad R^{(A)}_{n}(j):=G^{(A)}_{n}(j)+(-1)^{j} L^{(A)}_{n}(j), \end{aligned}$$

with \(D^{(A)}(1):=\overrightarrow{{F}_{0}}^{(A)}(1)\). Clearly, \(D^{(A_{1})}(1):=-2\log (2)\) and \(D^{(A_{2})}(1):=-\log (2)+\frac{\pi }{2}\).

Remark 3.1

It should be emphasized that many notations in Definition 3.2 were introduced in the reference [12].

Obviously,

$$\begin{aligned} \overrightarrow{F_{n}}^{(A_{1})}(1)=-2\log (2),\qquad \overrightarrow{F_{n}}^{(A_{2})}(1)=(-1)^{n-1}\log (2)+ \frac{\pi }{2}. \end{aligned}$$

Clearly, if we let \(A=A_{1}\) or \(A_{2}\) in Definition 3.2, elementary calculations yield

$$\begin{aligned} &M^{(A_{1})}_{n}(j)={H}^{(j)}_{n}+(-1)^{j} \textstyle\begin{cases} -2\log (2) ,& j=1, \\ \zeta (j),& j>1, \end{cases}\displaystyle \\ &M^{(A_{2})}_{n}(j)=(-1)^{n-1}{\bar{H}}^{(j)}_{n}+(-1)^{j} \textstyle\begin{cases} (-1)^{n-1}\log (2)+\frac{\pi }{2} ,& j=1, \\ (-1)^{n-1}{\bar{\zeta }}(j),& j>1, \end{cases}\displaystyle \\ &N^{(A_{1})}_{n}(j)=h^{(j)}_{n}+(-1)^{j} \widetilde{t}(j),\quad \widetilde{t}(1):=0, \\ &N^{(A_{2})}_{n}(j)=(-1)^{n-1}{\bar{h}}^{(j)}_{n}+(-1)^{j} \textstyle\begin{cases} ((-1)^{n}+1){\bar{t}}(1) ,& j=1, \\ (-1)^{n}{\bar{t}}(j),& j>1, \end{cases}\displaystyle \\ &{\bar{N}}^{(A_{1})}_{n}(j)=\widetilde{t}(j)-h^{(j)}_{n-1}, \quad \widetilde{t}(1):=0, \\ &{\bar{N}}^{(A_{2})}_{n}(j)=(-1)^{n}{ \bar{h}}^{(j)}_{n-1}+ \textstyle\begin{cases} ((-1)^{n-1}+1){\bar{t}}(1) ,& j=1, \\ (-1)^{n-1}{\bar{t}}(j),& j>1, \end{cases}\displaystyle \\ &R^{(A_{1})}_{n}(j)=\bigl(1+(-1)^{j}\bigr)\zeta (j), \qquad R^{(A_{2})}_{n}(j)=(-1)^{n-1}\bigl(1+(-1)^{j} \bigr){ \bar{\zeta }}(j). \end{aligned}$$

Definition 3.3

([12, Def. 1.2])

Define the cotangent function with sequence A by

$$\begin{aligned} \pi \cot ( {\pi s;A} )= \frac{{{a_{0}}}}{s} - 2s\sum _{k = 1}^{\infty }{\frac{{{a_{k}}}}{{{k^{2}} - {s^{2}}}}}. \end{aligned}$$
(3.2)

It is clear that if we let \(A=A_{1}\) and \(A_{2}\) in (3.2), respectively, then it becomes

$$\begin{aligned} &\cot ( {\pi s;A_{1}} ) = \cot ( {\pi s} ) \quad \text{and}\quad \cot ( {\pi s;A_{2}} ) = \csc ( {\pi s} ). \end{aligned}$$

Several identities among Φ-functions

Proposition 3.2

Let \(p\geq 1\)and n be nonnegative integers, if \(|s-n+1/2|<1\), then

$$\begin{aligned} \frac{\varPhi ^{(p-1)}(-s;A)}{(p-1)!}=\sum_{j=1}^{\infty }(-1)^{j-1} \binom{j+p-2}{p-1} N^{(A)}_{n}(j+p-1) (s-n+1/2)^{j-1}. \end{aligned}$$
(3.3)

Proposition 3.3

Let \(p\geq 1\)and n be nonnegative integers, if \(|s-n|<1\)with \(s\neq n\), then

$$\begin{aligned} \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;A} )}}{{ ( {p - 1} )!}}= \frac{1}{{{{ ( {s - n} )}^{p}}}} \Biggl\{ a_{n}- \sum_{j=1}^{\infty }(-1)^{j} \binom{j+p-2}{p-1} M^{(A)}_{n}(j+p-1) (s-n)^{j+p-1} \Biggr\} . \end{aligned}$$
(3.4)

If we set \(n=0\), then for any \(|s|<1\) with \(s\neq 0\),

$$\begin{aligned} \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;A} )}}{{ ( {p - 1} )!}}= \frac{{{a_{0}}}}{{{s^{p}}}} + { ( { - 1} )^{p}}\sum_{j = 1}^{\infty }{ \binom{j+p-2}{j-1} D^{(A)}(j+p-1){s^{j - 1}}}. \end{aligned}$$
(3.5)

Proposition 3.4

Let p and n be positive integers, if \(|s+n-1/2|<1\), then

$$\begin{aligned} \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;A} )}}{{ ( {p - 1} )!}}=(-1)^{p} \sum _{j=1}^{\infty }\binom{j+p-2}{p-1} {\bar{N}}^{(A)}_{n}(j+p-1) (s+n-1/2)^{j-1}. \end{aligned}$$
(3.6)

The method of the proofs of identities (3.3)–(3.6) is completely similar to that in [12, Theorems 2.1–2.3]. Thus, we omit it.

Lemmas

We define \(\tan (s;A):=\cot (\pi /2-s;A)\). It is clear that \(\tan (s;A_{1})=\tan (s)\) and \(\tan (s;A_{2})=\sec (s)\).

Lemma 3.5

([12, Thm. 2.3])

With \(\cot (\pi s;A)\)defined above, if \(|s-n|<1\)with \(s\neq n\) (\(n\in \mathbb{Z}\)), then

$$\begin{aligned} \pi \cot (\pi s;A)=\frac{a_{|n|}}{s-n}-\sum_{j=1}^{\infty }(- \sigma _{n})^{j} R^{(A)}_{|n|}(j) (s-n)^{j-1}, \end{aligned}$$
(3.7)

where \(\sigma _{n}\)is defined by the symbol of n, namely

$$\begin{aligned} \sigma _{n}:= \textstyle\begin{cases} 1 ,& n\geq 0, \\ -1,& n< 0. \end{cases}\displaystyle \end{aligned}$$

Hence, an elementary calculation yields

$$\begin{aligned} \pi \tan (\pi s;A)=-\frac{a_{|n-1|}}{s-n+1/2}+\sum_{j=1}^{\infty } \sigma _{n-1}^{j} R^{(A)}_{|n-1|}(j) (s-n+1/2)^{j-1} \end{aligned}$$
(3.8)

for \(|s-n+1/2|<1\) with \(s\neq n-1/2\) (\(n\in \mathbb{Z}\)),

Let \(B:=\{b_{k}\}\), \(-\infty < k < \infty \) be a sequence of complex numbers with \({b_{k}} = o ( {{k^{\beta }}} )\) (\(\beta < 1\)) if \(k\rightarrow \pm \infty \). Define a kernel function \(\xi ( s )\) by the two requirements: 1. \(\xi ( s )\) is meromorphic in the whole complex plane. 2. \(\xi ( s )\) satisfies \(\xi ( s )=o(s)\) over an infinite collection of circles \(\vert s \vert = {\rho _{k}}\) with \({\rho _{k}} \to \infty \). Applying these two conditions of kernel function \(\xi ( s )\), Flajolet and Salvy showed the following residue theorem.

Lemma 3.6

([2])

Let \(\xi ( s )\)be a kernel function, and let \(r(s)\)be a rational function which is \(O(s^{-2})\)at infinity. Then

$$\begin{aligned} \sum_{\alpha \in O} \operatorname{Res} {{ \bigl[ {r ( s ) \xi ( s )},s = \alpha \bigr]}} + \sum_{\beta \in S} \operatorname{Res} {{ \bigl[ {r ( s )\xi ( s )},s = \beta \bigr]}} = 0, \end{aligned}$$
(3.9)

where S is the set of poles of \(r(s)\)and O is the set of poles of \(\xi ( s )\)that are not poles of \(r(s)\). Here \(\operatorname{Res} { [ {r ( s )},s = \alpha ]} \)denotes the residue of \(r(s)\)at \(s= \alpha \).

Two general theorems

In this section, we prove two general theorems which will be used to obtain the explicit evaluations of (alternating) double t-values and (alternating) double T-values.

Theorem 4.1

For positive integers p and \(q>1\),

$$\begin{aligned} &{-}\sum_{n=1}^{\infty }\frac{N^{(B)}_{n}(p)}{(n-1/2)^{q}}a_{n-1}-(-1)^{p+q} \sum_{n=1}^{\infty }\frac{{\bar{N}}^{(B)}_{n}(p)}{(n-1/2)^{q}}a_{n} \\ &\quad{}-(-1)^{p} \sum_{k=0}^{p-1} \frac{(-1)^{k}}{k!}\binom{p+q-k-2}{q-1} \sum_{n=1}^{\infty } \frac{b_{n}}{n^{p+q-k-1}}\lim_{s\rightarrow n} \frac{d^{k}}{ds^{k}}\bigl(\pi \tan ( \pi s;A)\bigr) \\ &\quad{}+\operatorname{Res}\bigl[f_{1}(s;A,B),s=0\bigr]=0, \end{aligned}$$
(4.1)

where

$$\begin{aligned}& \frac{d^{k}}{ds^{k}}\bigl(\pi \tan (\pi s;A)\bigr)= \textstyle\begin{cases} (1-(-1)^{k})k!\widetilde{t}(k+1) ,& A=A_{1}, \\ (-1)^{n}(1+(-1)^{k})k!{\bar{t}}(k+1),& A=A_{2}, \end{cases}\displaystyle \end{aligned}$$
(4.2)
$$\begin{aligned}& \begin{aligned}[b] &\operatorname{Res}\bigl[f_{1}(s;A_{1},B),s=0 \bigr]\\ &\quad =b_{0}\bigl(1+(-1)^{p+q}\bigr)\widetilde{t}(p+q) \\ &\qquad {}+2(-1)^{p} \sum_{{2k_{1}+k_{2}=q+1,\atop k_{1},k_{2} \geq 1}} \binom{k_{2}+p-2}{p-1} \widetilde{t}(2k_{1})D^{(B)}(k_{2}+p-1), \end{aligned} \end{aligned}$$
(4.3)
$$\begin{aligned}& \begin{aligned}[b] &\operatorname{Res}\bigl[f_{1}(s;A_{2},B),s=0 \bigr]\\ &\quad =b_{0}\bigl(1-(-1)^{p+q}\bigr){\bar{t}}(p+q) \\ &\qquad {}+2(-1)^{p} \sum_{{2k_{1}+k_{2}=q+2,\atop k_{1},k_{2} \geq 1}} \binom{k_{2}+p-2}{p-1} {\bar{t}}(2k_{1}-1)D^{(B)}(k_{2}+p-1). \end{aligned} \end{aligned}$$
(4.4)

Proof

Apply the kernel function

$$ \pi \tan ( {\pi s};A ) \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;B} )}}{{ ( {p - 1} )!}} $$

to the base function \(r(s)=s^{-q}\). Namely, we need to compute the residue of the function

$$ {f_{1}}( {s;A,B}): = \pi \tan ( {\pi s} ;A ) \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;B} )}}{{ ( {p - 1} )!{s^{q}}}}. $$

Clearly, \(f_{1}(s;A,B)\) only has poles at \(s=0,\pm (n-1/2)\) and n (n is a positive integer). With the help of identities (3.3)–(3.6), we deduce the following residues:

$$\begin{aligned} &\operatorname{Res}[f_{1},s=n-1/2]=-\frac{N^{(B)}_{n}(p)}{(n-1/2)^{q}}a_{n-1}, \\ &\operatorname{Res}[f_{1},s=1/2-n]=-(-1)^{p+q} \frac{{\bar{N}}^{(B)}_{n}(p)}{(n-1/2)^{q}}a_{n}, \\ &\operatorname{Res}[f_{1},s=n]=-(-1)^{p} \sum _{k=0}^{p-1} \frac{(-1)^{k}}{k!} \binom{p+q-k-2}{q-1} \frac{b_{n}}{n^{p+q-k-1}}\lim_{s\rightarrow n} \frac{d^{k}}{ds^{k}}\bigl(\pi \tan ( \pi s;A)\bigr) \end{aligned}$$

and (4.2)–(4.4). Applying Lemma 3.6 yields the desired result. □

Proof of Theorem 2.1

Setting \(A,B\in \{A_{1},A_{2}\}\) in Theorem 4.1 yields the four desired evaluations. □

Theorem 4.2

For positive integers p and \(q>1\),

$$\begin{aligned} &{-}\sum_{n=1}^{\infty } \frac{N^{(B)}_{n}(p)}{n^{q}}a_{n-1}-(-1)^{p+q} \sum _{n=1}^{\infty }\frac{{\bar{N}}^{(B)}_{n+1}(p)}{n^{q}}a_{n+1} \\ &\quad{}-(-1)^{p} \sum_{k=0}^{p-1} \frac{(-1)^{k}}{k!}\binom{p+q-k-2}{q-1} \sum_{n=0}^{\infty } \frac{b_{n}}{(n+1/2)^{p+q-k-1}}\lim_{s \rightarrow n} \frac{d^{k}}{ds^{k}}\bigl(\pi \tan ( \pi s;A)\bigr) \\ &\quad{}-(-1)^{p}a_{1}\binom{p+q-1}{p-1}W^{(B)}(p+q) \\ &\quad{}+(-1)^{p}\sum_{{k_{1}+k_{2}=q+1,\atop k_{1},k_{2} \geq 1}} \binom{k_{2}+p-2}{p-1}Z^{(A)}(k_{1})W^{(B)}(k_{2}+p-1)=0, \end{aligned}$$
(4.5)

where

$$\begin{aligned} W^{(B)}(j):=\widehat{F}^{(B)}_{-1}(j)= \textstyle\begin{cases} \sum_{k=1}^{\infty }\frac{a_{k-1}-a_{k}}{k-1/2} ,& j=1, \\ \sum_{k=1}^{\infty }\frac{a_{k-1}}{(k-1/2)^{j}},& j>1, \end{cases}\displaystyle \end{aligned}$$
(4.6)

and

$$\begin{aligned} Z^{(A)}(j):=(-1)^{j} R^{(A)}_{1}(j)=\sum _{k=1}^{\infty }\frac{a_{k+1}+(-1)^{j}a_{k-1}}{k^{j}}. \end{aligned}$$
(4.7)

Proof

Apply the kernel function

$$ \pi \tan ( {\pi s};A ) \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;B} )}}{{ ( {p - 1} )!}} $$

to the base function \(r(s)=(s+1/2)^{-q}\). Namely, we need to compute the residues of the function

$$ {f_{2}}( {s;A,B}): = \pi \tan ( {\pi s} ;A ) \frac{{{\varPhi ^{ ( {p - 1} )}} ( { - s;B} )}}{{ ( {p - 1} )!{(s+1/2)^{q}}}}. $$

Clearly, \({f_{2}}( {s;A,B})\) only has poles at 0, n and \(\pm (n-1/2)\) (n is a positive integer). With the help of identities (3.3)–(3.6), these residues are

$$\begin{aligned} &\operatorname{Res}[f_{2},s=n-1/2]=-\frac{N^{(B)}_{n}(p)}{n^{q}}a_{n-1} \quad (n \geq 1), \\ &\operatorname{Res}[f_{2},s=1/2-n]=-(-1)^{p+q} \frac{{\bar{N}}^{(B)}_{n}(p)}{(n-1)^{q}}a_{n}\quad (n\geq 2), \\ &\operatorname{Res}[f_{2},s=n] \\ &\quad =-(-1)^{p} \sum_{k=0}^{p-1} \frac{(-1)^{k}}{k!}\binom{p+q-k-2}{q-1} \frac{b_{n}}{(n+1/2)^{p+q-k-1}}\lim_{s\rightarrow n} \frac{d^{k}}{ds^{k}}\bigl(\pi \tan (\pi s;A)\bigr)\quad (n\geq 0), \\ &\operatorname{Res}[f_{2},s=-1/2]\\ &\quad =-(-1)^{p}a_{1} \binom{p+q-1}{p-1}W^{(B)}(p+q) \\ & \qquad {}+(-1)^{p} \sum_{{k_{1}+k_{2}=q+1,\atop k_{1},k_{2} \geq 1}} \binom{k_{2}+p-2}{p-1}Z^{(A)}(k_{1})W^{(B)}(k_{2}+p-1)\\ &\quad =0. \end{aligned}$$

Then summing these four contributions and using Lemma 3.6, we may easily deduce the desired evaluation. □

Proof of Theorem 2.3

Setting \(A,B\in \{A_{1},A_{2}\}\) in Theorem 4.2 yields the four desired evaluations. □

It is possible that closed form representations of some other similar infinite series can be proved using techniques of the present paper.

Remark 4.3

It should be emphasized that Xu [12] defined another parametric digamma function \(\varPsi (-s;A)\). Very recently, Wang and Xu [10] used the parametric digamma function \(\varPsi (-s;A)\) to define several new kernel functions. Then they used the methods of contour integration and residue theorem to prove two general theorems (using the two theorems, they obtained Theorems 2.1 and 2.3), which are similar to Theorems 4.1 and 4.2. Moreover, they also showed many other types of results.

References

  1. 1.

    Bailey, D.H., Borwein, J.M., Girgensohn, R.: Experimental evaluation of Euler sums. Exp. Math. 3(1), 17–30 (1994)

    MathSciNet  Article  Google Scholar 

  2. 2.

    Flajolet, P., Salvy, B.: Euler sums and contour integral representations. Exp. Math. 7(1), 15–35 (1998)

    MathSciNet  Article  Google Scholar 

  3. 3.

    Guo, D.R., Tu, F.-Y., Wang, Z.X.: Special Functions. World Scientific, Singapore (1989), pp. 498–574

    Google Scholar 

  4. 4.

    Hoffman, M.E.: Multiple harmonic series. Pac. J. Math. 152, 275–290 (1992)

    MathSciNet  Article  Google Scholar 

  5. 5.

    Hoffman, M.E.: An odd variant of multiple zeta values. Commun. Number Theory Phys. 13, 529–567 (2019)

    MathSciNet  MATH  Google Scholar 

  6. 6.

    Kaneko, M., Tsumura, H.: Zeta functions connecting multiple zeta values and poly-Bernoulli numbers. Adv. Stud. Pure Math. (2020, in press). arXiv:1811.07736v1

  7. 7.

    Kaneko, M., Tsumura, H.: On a variant of multiple zeta values of level two. Tsukuba J. Math. (2020, in press). arXiv:1903.03747v2

  8. 8.

    Mezö, I.: Nonlinear Euler sums. Pac. J. Math. 272, 201–226 (2014)

    MathSciNet  Article  Google Scholar 

  9. 9.

    Wang, W., Lyu, Y.: Euler sums and Stirling sums. J. Number Theory 185, 160–193 (2018)

    MathSciNet  Article  Google Scholar 

  10. 10.

    Wang, W., Xu, C.: Alternating Euler T-sums and Euler -sums. arXiv:2004.04556v3

  11. 11.

    Xu, C.: Multiple zeta values and Euler sums. J. Number Theory 177, 443–478 (2017)

    MathSciNet  Article  Google Scholar 

  12. 12.

    Xu, C.: Explicit formulas for general Euler type sums. arXiv:2002.12107v2

  13. 13.

    Xu, C., Li, Z.: Tornheim type series and nonlinear Euler sums. J. Number Theory 174, 40–67 (2017)

    MathSciNet  Article  Google Scholar 

  14. 14.

    Xu, C., Wang, W.: Two variants of Euler sums. arXiv:1906.07654v3

  15. 15.

    Xu, C., Yan, Y., Shi, Z.: Euler sums and integrals of polylogarithm functions. J. Number Theory 165, 84–108 (2016)

    MathSciNet  Article  Google Scholar 

  16. 16.

    Zagier, D.: Values of zeta functions and their applications. In: First European Congress of Mathematics, Volume II, vol. 120, pp. 497–512. Birkhäuser, Boston (1994)

    Google Scholar 

  17. 17.

    Zhao, J.: Multiple Zeta Functions, Multiple Polylogarithms and Their Special Values. Series on Number Theory and Its Applications, vol. 12. World Scientific, Hackensack (2016)

    Google Scholar 

Download references

Acknowledgements

We thank the anonymous referee for suggestions which led to improvements in the exposition.

Availability of data and materials

Data sharing not applicable to this article as no data sets were generated or analysed during the current study.

Funding

The research is not funded.

Author information

Affiliations

Authors

Contributions

All authors declare that they have reviewed and approved the final manuscript for publication.

Corresponding author

Correspondence to Junjie Quan.

Ethics declarations

Competing interests

The authors declare that they have no competing interests.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Quan, J. Alternating double t-values and T-values. Adv Differ Equ 2020, 450 (2020). https://doi.org/10.1186/s13662-020-02917-1

Download citation

MSC

  • 11M32
  • 11B65

Keywords

  • Multiple zeta values
  • Multiple t-values
  • Multiple T-values
  • Odd harmonic numbers
  • Contour integration