By making use of the multisection series method, four classes of alternating infinite series are evaluated, in closed form, by the Riemann zeta function and the Dirichlet beta function.
Citation: Zhiling Fan, Wenchang Chu. Alternating series in terms of Riemann zeta function and Dirichlet beta function[J]. Electronic Research Archive, 2024, 32(2): 1227-1238. doi: 10.3934/era.2024058
[1] | Mehmet Ali Özarslan, Ceren Ustaoğlu . Extended incomplete Riemann-Liouville fractional integral operators and related special functions. Electronic Research Archive, 2022, 30(5): 1723-1747. doi: 10.3934/era.2022087 |
[2] | Nelson Vieira, M. Manuela Rodrigues, Milton Ferreira . Time-fractional telegraph equation of distributed order in higher dimensions with Hilfer fractional derivatives. Electronic Research Archive, 2022, 30(10): 3595-3631. doi: 10.3934/era.2022184 |
[3] | Xinguang Zhang, Yongsheng Jiang, Lishuang Li, Yonghong Wu, Benchawan Wiwatanapataphee . Multiple positive solutions for a singular tempered fractional equation with lower order tempered fractional derivative. Electronic Research Archive, 2024, 32(3): 1998-2015. doi: 10.3934/era.2024091 |
[4] | Aying Wan, Feng Qi . Power series expansion, decreasing property, and concavity related to logarithm of normalized tail of power series expansion of cosine. Electronic Research Archive, 2024, 32(5): 3130-3144. doi: 10.3934/era.2024143 |
[5] | Guo-Niu Han, Huan Xiong . Skew doubled shifted plane partitions: Calculus and asymptotics. Electronic Research Archive, 2021, 29(1): 1841-1857. doi: 10.3934/era.2020094 |
[6] | Mufit San, Seyma Ramazan . A study for a higher order Riemann-Liouville fractional differential equation with weakly singularity. Electronic Research Archive, 2024, 32(5): 3092-3112. doi: 10.3934/era.2024141 |
[7] | Mohra Zayed, Gamal Hassan . Kronecker product bases and their applications in approximation theory. Electronic Research Archive, 2025, 33(2): 1070-1092. doi: 10.3934/era.2025048 |
[8] | Yutong Chen, Jiabao Su . Resonant problems for non-local elliptic operators with unbounded nonlinearites. Electronic Research Archive, 2023, 31(9): 5716-5731. doi: 10.3934/era.2023290 |
[9] | Noosheza Rani, Arran Fernandez . An operational calculus formulation of fractional calculus with general analytic kernels. Electronic Research Archive, 2022, 30(12): 4238-4255. doi: 10.3934/era.2022216 |
[10] | Zhiliang Li, Lijun Pei, Guangcai Duan, Shuaiyin Chen . A non-autonomous time-delayed SIR model for COVID-19 epidemics prediction in China during the transmission of Omicron variant. Electronic Research Archive, 2024, 32(3): 2203-2228. doi: 10.3934/era.2024100 |
By making use of the multisection series method, four classes of alternating infinite series are evaluated, in closed form, by the Riemann zeta function and the Dirichlet beta function.
The Riemann zeta function and the Dirichlet beta function are important functions in number theory that are defined by (cf. [1, Chapter 25])
ζ(λ)=∞∑k=01(k+1)λandβ(λ)=∞∑k=0(−1)k(2k+1)λ, |
provided that ℜ(λ)>1 and ℜ(λ)>0 so that ζ(λ) and β(λ) are convergent. There are numerous useful formulae for these functions in the literature (cf. [2,3,4,5]). For example, we shall frequently utilize the following relation:
∞∑k=0(−1)k(k+1)λ=(1−21−λ)ζ(λ),ℜ(λ)>0. |
In particular, there exist closed form expressions in π
ζ(2n)=(−1)n−1(2π)2n2(2n)!B2nandβ(2n+1)=(−1)nπ2n+14n+1(2n)!E2n, |
where Bn and En are Bernoulli and Euler numbers (cf. [6, §6.5]), whose exponential generating functions are given by
yey−1=∞∑n=0Bnn!yn,|y|<2π;2ey1+e2y=∞∑n=0Enn!yn,|y|<π/2. |
For a real number x, denote by ⌊x⌋ and ⌈x⌉ the usual floor and ceiling functions. The aim this paper is to examine, for two natural numbers m,n∈N, the following four variants of the Riemann zeta function:
Um(n)=∞∑k=0(−1)⌊k+m2m⌋(k+1)2n,Um(n)=∞∑k=0(−1)⌊k+m2m⌋(2k+1)2n;Vm(n)=∞∑k=0(−1)⌊km⌋(k+1)2n−1,Vm(n)=∞∑k=0(−1)⌊km⌋(2k+1)2n−1. |
Several closed formulae will be established that express these alternating infinite series in terms of the Riemann zeta function and the Dirichlet beta function.
Throughout the paper, we shall make use of the following polynomial expression for higher derivatives of cotangent function (cf. [7,8])
dncotxdxn=Pn(cotx):Pn(x)=(2i)n(x−i)n∑k=0k!2k{nk}(ix−1)k, |
where {nk} are the Stirling numbers of the second kind (cf. [6, §6.1]). The first five polynomials read explicitly as follows:
P1(x)=−x2−1,P2(x)=2x3+2x,P3(x)=−6x4−8x2−2,P4(x)=24x5+40x3+16x,P5(x)=−120x6−240x4−136x2−16. |
By putting these in conjunction with the partial fraction decomposition (see, for example, Stromberg [9, Theorem 5.43] and Higgins [10])
cot(πx)=limn→∞1πn∑k=−n1k+x=1πx+2xπ∞∑k=11x2−k2, |
we deduce for m∈N, by computing mth derivative with respect to x, the following compact expression for the bilateral series
∞∑k=−∞1(k+x)m+1=(−1)mπm+1m!Pm(cot(πx)). | (1) |
In this section, we shall examine the series Um(n) and Um(n) by dividing them into multisection series. Two general summation theorems will be proved and several explicit formulae will be presented as consequences.
Since the series Um(n) is absolutely convergent, in order to evaluate it explicitly, we can regroup its terms. For example U1(n) and U2(n) can be expressed as follows:
U1(n)=∞∑k=0(−1)⌊k+12⌋(k+1)2n=1−122n−132n+142n+152n−162n−172n+182n+⋯={1−132n+152n−172n+⋯}−{122n−142n+162n−182n+⋯},U2(n)=∞∑k=0(−1)⌊k+24⌋(k+1)2n=1+122n−132n−142n−152n−162n+172n+182n+192n+1102n−1112n−1122n−1132n−1142n+1152n+1162n+⋯={1−152n+192n−1132n+⋯}+{122n−162n+1102n−1142n+⋯}−{132n−172n+1112n−1152n+⋯}−{142n−182n+1122n−1162n+⋯}. |
In general, by dividing the summation index k according to the residue classes modulo 4m, i.e., making the replacements k→4km+j−1 with 1≤j≤4m, we can express Um(n) in terms of multisection series
Um(n)=∞∑k=0(−1)⌊k+m2m⌋(k+1)2n=4m∑j=1∞∑k=0(−1)⌊4km+m+j−12m⌋(4km+j)2n=m∑j=1∞∑k=0{(−1)⌊m+j−12m⌋(4km+j)2n+(−1)⌊2m+j−12m⌋(4km+m+j)2n}+m∑j=1∞∑k=0{(−1)⌊3m+j−12m⌋(4km+2m+j)2n+(−1)⌊4m+j−12m⌋(4km+3m+j)2n}. |
For 1≤j≤m, observing that
(−1)⌊m+j−12m⌋=(−1)⌊4m+j−12m⌋=1,(−1)⌊2m+j−12m⌋=(−1)⌊3m+j−12m⌋=−1; |
and then putting the summation terms corresponding to j=m aside, we can reformulate further
Um(n)=∞∑k=0{(−1)k(2km+m)2n−(−1)k(2km+2m)2n}+m−1∑j=1∞∑k=0{1(4km+j)2n−1(4km+m+j)2n}+m−1∑j=1∞∑k=0{1(4km+3m+j)2n−1(4km+2m+j)2n}. |
Writing the first line directly in β(2n) and ζ(2n), then inverting simultaneously the summation indices by j→m−j and k→−k−1 in the last line, we derive the following bilateral series expression
Um(n)=β(2n)m2n−ζ(2n)(2m)2n{1−24n}+m−1∑j=1∞∑k=−∞{1(4km+j)2n−1(4km+m+j)2n}. |
By applying Eq (1), we find the simplified formula as in the following theorem.
Theorem 1 (m,n∈N).
Um(n)=∞∑k=0(−1)⌊k+m2m⌋(k+1)2n=β(2n)m2n+ζ(2n)(2m)2n{24n−1}+(−1)n4nζ(2n)43nm2nB2nm−1∑j=1{P2n−1(cot(jπ4m))−P2n−1(cot(m+j4mπ))}. |
For small m,n∈N, this theorem can be utilized to compute exact values of Um(n). The following interesting formulae are exhibited as examples.
Example 1.
U1(1)=∞∑k=0(−1)⌊k+12⌋(k+1)2=β(2)−ζ(2)8,U1(2)=∞∑k=0(−1)⌊k+12⌋(k+1)4=β(4)−7ζ(4)128,U1(3)=∞∑k=0(−1)⌊k+12⌋(k+1)6=β(6)−31ζ(6)2048,U1(4)=∞∑k=0(−1)⌊k+12⌋(k+1)8=β(8)−127ζ(8)32768,U1(5)=∞∑k=0(−1)⌊k+12⌋(k+1)10=β(10)−511ζ(10)524288. |
Example 2.
U2(1)=∞∑k=0(−1)⌊k+24⌋(k+1)2=β(2)4+3ζ(2)4√2−ζ(2)32,U2(2)=∞∑k=0(−1)⌊k+24⌋(k+1)4=β(4)16+165ζ(4)128√2−7ζ(4)2048,U2(3)=∞∑k=0(−1)⌊k+24⌋(k+1)6=β(6)64+22743ζ(6)16384√2−31ζ(6)131072,U2(4)=∞∑k=0(−1)⌊k+24⌋(k+1)6=β(8)256+369165ζ(8)262144√2−127ζ(8)8388608,U2(5)=∞∑k=0(−1)⌊k+24⌋(k+1)10=β(10)1024+94810353ζ(10)67108864√2−511ζ(10)536870912. |
Example 3.
U3(1)=∞∑k=0(−1)⌊k+36⌋(k+1)2=β(2)9+ζ(2)√3+7ζ(2)72,U3(2)=∞∑k=0(−1)⌊k+36⌋(k+1)4=β(4)81+115ζ(4)72√3+553ζ(4)10368,U3(3)=∞∑k=0(−1)⌊k+36⌋(k+1)6=β(6)729+11767ζ(6)6912√3+22537ζ(6)1492992,U3(4)=∞∑k=0(−1)⌊k+36⌋(k+1)8=β(8)6561+1287715ζ(8)746496√3+832993ζ(8)214990848,U3(5)=∞∑k=0(−1)⌊k+36⌋(k+1)10=β(10)59049+744010091ζ(10)429981696√3+30173017ζ(10)30958682112. |
Performing the replacement k→4km+j−1 with 1≤j≤4m, we can express Um(n) as the following multisection series
Um(n)=∞∑k=0(−1)⌊k+m2m⌋(2k+1)2n=∞∑k=04m∑j=1(−1)⌊m+j−12m⌋(8km+2j−1)2n=∞∑k=0{m∑j=1(−1)⌊m+j−12m⌋(8km+2j−1)2n+m∑j=1(−1)⌊2m+j−12m⌋(8km+2m+2j−1)2n+m∑j=1(−1)⌊3m+j−12m⌋(8km+4m+2j−1)2n+m∑j=1(−1)⌊4m+j−12m⌋(8km+6m+2j−1)2n}. |
The above expression can be further reduced as follows:
Um(n)=∞∑k=0m∑j=1{1(8km+2j−1)2n−1(8km+2m+2j−1)2n}+∞∑k=0m∑j=1{1(8km+6m+2j−1)2n−1(8km+4m+2j−1)2n}. |
For the two sums on the ultimate line, making the changes on summation indices by j→1+m−j and k→−k−1
−1∑k=−∞m∑j=1{1(8km+2j−1)2n−1(8km+2m+2j−1)2n} |
and then unifying this resulting expression to the first line, we find the bilateral series expression
Um(n)=m∑j=1∞∑k=−∞{1(8km+2j−1)2n−1(8km+2m+2j−1)2n}=m∑j=11(8m)2n∞∑k=−∞{1(k+2j−18m)2n−1(k+2m+2j−18m)2n}. |
By invoking Eq (1), we establish the following summation formula.
Theorem 2 (m,n∈N).
Um(n)=∞∑k=0(−1)⌊k+m2m⌋(2k+1)2n=(−1)n4nζ(2n)44nm2nB2n×m∑j=1{P2n−1(cot(2j−18mπ))−P2n−1(cot(2m+2j−18mπ))}. |
For m=1,2,3, we record the following closed formulae as applications.
Example 4 (m=1).
U1(1)=∞∑k=0(−1)⌊k+12⌋(2k+1)2=3ζ(2)4√2,U1(2)=∞∑k=0(−1)⌊k+12⌋(2k+1)4=165ζ(4)128√2,U1(3)=∞∑k=0(−1)⌊k+12⌋(2k+1)6=22743ζ(6)16384√2,U1(4)=∞∑k=0(−1)⌊k+12⌋(2k+1)8=369165ζ(8)262144√2,U1(5)=∞∑k=0(−1)⌊k+12⌋(2k+1)10=94810353ζ(10)67108864√2. |
Example 5 (m=2).
U2(1)=∞∑k=0(−1)⌊k+24⌋(2k+1)2=3ζ(2)16√10+√2,U2(2)=∞∑k=0(−1)⌊k+24⌋(2k+1)4=15ζ(4)2048√15898+241√2,U2(3)=∞∑k=0(−1)⌊k+24⌋(2k+1)6=63ζ(6)1048576√267655210+494881√2,U2(4)=∞∑k=0(−1)⌊k+24⌋(2k+1)8=15ζ(8)67108864√19853759178298+4207474321√2,U2(5)=∞∑k=0(−1)⌊k+24⌋(2k+1)10=33ζ(10)68719476736√4327811705145000010+103023433654081√2. |
Example 6 (m=3).
U3(1)=∞∑k=0(−1)⌊k+36⌋(2k+1)2=ζ(2)24√218+24√3,U3(2)=∞∑k=0(−1)⌊k+36⌋(2k+1)4=ζ(4)6912√1635146+22968√3,U3(3)=∞∑k=0(−1)⌊k+36⌋(2k+1)6=ζ(6)2654208√138927209978+219727704√3,U3(4)=∞∑k=0(−1)⌊k+36⌋(2k+1)8=ζ(8)1146617856√52163307935186666+9179017200888√3,U3(5)=∞∑k=0(−1)⌊k+36⌋(2k+1)10=ζ(10)2641807540224√57564351813037136458778+1125649024023711384√3. |
This section will be devoted to the two remaining series Vm(n) and Vm(n). We shall establish two general summation theorems. Several explicit formulae in terms of the Riemann zeta function and the Dirichlet beta function will be deduced as particular cases.
Making the replacement k→km+j−1 with 1≤j≤m, we can rewrite Vm(n) as
Vm(n)=∞∑k=0(−1)⌊km⌋(k+1)2n−1=∞∑k=0m∑j=1(−1)⌊km+j−1m⌋(km+j)2n−1=∞∑k=0(−1)⌊km+m−1m⌋(km+m)2n−1+∞∑k=0m−1∑j=1(−1)⌊km+j−1m⌋(km+j)2n−1. |
We remark that the above equalities are valid for n≥1. When n=1, the corresponding formula in Theorem 3 can be shown analogously by examining the truncated sum below
Vm(1)=limn→∞n∑k=0(−1)⌊km⌋k+1. |
The first series in the last line can be evaluated directly as follows:
∞∑k=0(−1)⌊km+m−1m⌋(km+m)2n−1=1m2n−1∞∑k=0(−1)k(k+1)2n−1={ln2m,n=1;22n−1−2(2m)2n−1ζ(2n−1),n>1. |
According to the parity of k, the double series can be split into two
∞∑k=0m−1∑j=1(−1)⌊km+j−1m⌋(km+j)2n−1=∞∑k=0m−1∑j=1(−1)k(km+j)2n−1=∞∑k=0m−1∑j=11(2km+j)2n−1−∞∑k=0m−1∑j=11(2km+m+j)2n−1. |
Now making the replacements on summation indices by j→m−j and k→−k−1 in the latter double sum and then unifying it to the former, we can simplify the resulting expression into the following bilateral series:
∞∑k=0m−1∑j=1(−1)⌊km+j−1m⌋(km+j)2n−1=m−1∑j=11(2m)2n−1∞∑k=−∞1(k+j2m)2n−1. |
Finally, by making use of (1), we can express the rightmost bilateral series in terms of Dirichlet beta function:
∞∑k=−∞1(k+j2m)2n−1=π2n−1(2n−2)!P2n−2(cot(jπ2m))=(−1)n−14nβ(2n−1)E2n−2P2n−2(cot(jπ2m)). |
Summing up, we have proved the following theorem.
Theorem 3 (m,n∈N).
∞∑k=0(−1)⌊km⌋(k+1)2n−1=(−1)n−12β(2n−1)m2n−1E2n−2m−1∑j=1P2n−2(cot(jπ2m))+{ln2m,n=1;22n−1−2(2m)2n−1ζ(2n−1),n≥2. |
When m=1, it is trivial to check that
V1(n)=∞∑k=0(−1)k(k+1)2n−1={ln2,n=1;(1−41−n)ζ(2n−1),n≥2. |
For m=2,3 and 1≤n≤5, the closed formulae are highlighted as examples.
Example 7 (m=2).
V2(1)=∞∑k=0(−1)⌊k2⌋k+1=β(1)+ln22,V2(2)=∞∑k=0(−1)⌊k2⌋(k+1)3=β(3)+332ζ(3),V2(3)=∞∑k=0(−1)⌊k2⌋(k+1)5=β(5)+15512ζ(5),V2(4)=∞∑k=0(−1)⌊k2⌋(k+1)7=β(7)+638192ζ(7),V2(5)=∞∑k=0(−1)⌊k2⌋(k+1)9=β(9)+255131072ζ(7). |
Example 8 (m=3).
V3(1)=∞∑k=0(−1)⌊k3⌋k+1=8β(1)3√3+ln23,V3(2)=∞∑k=0(−1)⌊k3⌋(k+1)3=160β(3)81√3+ζ(3)36,V3(3)=∞∑k=0(−1)⌊k3⌋(k+1)5=2176β(5)1215√3+5ζ(5)1296,V3(4)=∞∑k=0(−1)⌊k3⌋(k+1)7=232960β(7)133407√3+7ζ(7)15552,V3(5)=∞∑k=0(−1)⌊k3⌋(k+1)9=425805824β(9)245348595√3+85ζ(9)1679616. |
Under the replacements k→2km+j−1 with 1≤j≤2m, for n>1, we can rewrite Vm(n) as the following multisection series
Vm(n)=∞∑k=0(−1)⌊km⌋(2k+1)2n−1=∞∑k=02m∑j=1(−1)⌊j−1m⌋(4km+2j−1)2n−1=∞∑k=0{m∑j=1(−1)⌊j−1m⌋(4km+2j−1)2n−1+m∑j=1(−1)⌊m+j−1m⌋(4km+2m+2j−1)2n−1}=m∑j=1∞∑k=01(4km+2j−1)2n−1−m∑j=1∞∑k=01(4km+2m+2j−1)2n−1. |
When n=1, the corresponding formula in Theorem 4 can be confirmed by examining the following truncated sum
Vm(1)=limℓ→∞ℓ∑k=0(−1)⌊km⌋2k+1. |
Now, making the replacements j→m+1−j and k→−k−1 simultaneously for the latter double sums and then unifying it to the former one, we obtain the bilateral series expression below
Vm(n)=∞∑k=0m∑j=11(4km+2j−1)2n−1+−1∑k=−∞m∑j=11(4km+2j−1)2n−1=m∑j=1∞∑k=−∞1(4km+2j−1)2n−1=m∑j=11(4m)2n−1∞∑k=−∞1(k+2j−14m)2n−1. |
Keeping in mind (1) and writing the rightmost sum in terms of the Dirichlet beta function
∞∑k=−∞1(k+2j−14m)2n−1=π2n−1(2n−2)!P2n−2(cot(2j−14mπ))=(−1)n−14nβ(2n−2)E2n−2P2n−2(cot(2j−14mπ)), |
we arrive at the compact expression as in the following theorem.
Theorem 4 (m,n∈N).
∞∑k=0(−1)⌊km⌋(2k+1)2n−1=β(2n−1)m2n−1E2n−2(−14)n−1m∑j=1P2n−2(cot(2j−14mπ)). |
When m=1, it is obvious to see that
V1(n)=∞∑k=0(−1)k(2k+1)2n−1=β(2n−1). |
For m=2,3, the corresponding closed formulae are displayed as follows.
Example 9 (m=2).
V2(1)=∞∑k=0(−1)⌊k2⌋2k+1=√2β(1),V2(2)=∞∑k=0(−1)⌊k2⌋(2k+1)3=3β(3)2√2,V2(3)=∞∑k=0(−1)⌊k2⌋(2k+1)5=57β(5)40√2,V2(4)=∞∑k=0(−1)⌊k2⌋(2k+1)7=2763β(7)1952√2,V2(5)=∞∑k=0(−1)⌊k2⌋(2k+1)9=250737β(9)177280√2. |
Example 10 (m=3).
V3(1)=∞∑k=0(−1)⌊k3⌋2k+1=5β(1)3,V3(2)=∞∑k=0(−1)⌊k3⌋(2k+1)3=29β(3)27,V3(3)=∞∑k=0(−1)⌊k3⌋(2k+1)5=245β(5)243,V3(4)=∞∑k=0(−1)⌊k3⌋(2k+1)7=2189β(7)2187,V3(5)=∞∑k=0(−1)⌊k3⌋(2k+1)9=19685β(9)19683. |
The authors declare that they have not used Artificial Intelligence (AI) tools in the creation of this article.
The authors declare that there are no conflicts of interest regarding the publication of this paper.
[1] | T. M. Apostol, Zeta and related functions, in NIST Handbook of Mathematical Functions, Cambridge University Press, Cambridge, (2010), 601–616. |
[2] |
G. Bretti, C. Cesarano, P. E. Ricci, Laguerre-type exponentials and generalized Appell polynomials, Comput. Math. Appl., 48 (2004), 833–839. https://doi.org/10.1016/j.camwa.2003.09.031 doi: 10.1016/j.camwa.2003.09.031
![]() |
[3] |
E. Guariglia, Fractional calculus, zeta functions and Shannon entropy, Open Math. 19 (2021), 87–100. https://doi.org/10.1515/math-2021-0010 doi: 10.1515/math-2021-0010
![]() |
[4] |
S. D. Lin, H. M. Srivastava, Some families of the Hurwitz-Lerch Zeta functions and associated fractional derivative and other integral representations, Appl. Math. Comput., 154 (2004), 725–733 https://doi.org/10.1016/S0096-3003(03)00746-X doi: 10.1016/S0096-3003(03)00746-X
![]() |
[5] |
, W. Ramírez, D. Bedoya, A. Urieles, C. cesarano, M. Ortega, New U-Bernoulli, U-Euler and U-Genocchi polynomials and their matrices, Carpathian Math. Publ., 15 (2023), 449–467 https://doi.org/10.15330/cmp.15.2.449-467 doi: 10.15330/cmp.15.2.449-467
![]() |
[6] | R. L. Graham, D. E. Knuth, O. Patashnik, Concrete mathematics, Addison-Wesley Professional, Massachusetts, 1989. |
[7] | K. N. Boyadzhiev, Derivative polynomials for tanh, tan, sech and sec in explicit form, Fibonacci Quart., 45 (2007), 291–303. |
[8] |
W. Chu, C. Wang, Convolution formulae for Bernoulli numbers, Integr. Transf. Spec. Funct., 21 (2010), 437–457. https://doi.org/10.1080/10652460903360861 doi: 10.1080/10652460903360861
![]() |
[9] | K. Stromberg, An Introduction to Classical Real Analysis, American Mathematical Society, California, 1981. |
[10] |
J. R. Higgins, Two basic formulae of Euler and their equivalence to Tschakalov's sampling theorem, Sampl. Theory Signal Image Process., 2 (2003), 259–270. https://doi.org/10.1007/BF03549398 doi: 10.1007/BF03549398
![]() |