ebook img

Explicit evaluation of harmonic sums PDF

0.16 MB·
by  Ce Xu
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Explicit evaluation of harmonic sums

Explicit evaluation of harmonic sums Ce Xu∗ School of Mathematical Sciences, Xiamen University 7 Xiamen 361005,P.R. China 1 0 2 Abstract In this paper, we obtain some formulae for harmonic sums, alternating harmonic n a sums and Stirling number sums by using the method of integral representations of series. As J applications of these formulae, we give explicit formula of several quadratic and cubic Euler 2 sums through zeta values and linear sums. Furthermore, some relationships between harmonic ] numbers and Stirling numbers of the first kind are established. T N Keywords Harmonic number; Euler sum; Riemann zeta function; Stirling number. . h AMS Subject Classifications (2010): 11M06; 11M32; 33B15 t a m 1 Introduction [ 1 In this paper, the generalized harmonic numbers and alternating harmonic numbers of order k v are defined respectively as 8 8 n 1 n (−1)j−1 3 ζ (k) := , L (k) := , 1 ≤ n,k ∈ Z, (1.1) 0 n jk n jk 0 Xj=1 Xj=1 . 1 n 1 0 where H ≡ ζ (1) = denotes the classical harmonic number. n n 7 j j=1 1 X From [12,27], we know that the classical Euler sums are the infinite sums whose general term is : v product of 1 or (−1)n−1, harmonic numbers and alternating harmonic numbers of index n and i X a power of n−1. Therefore, the Euler sums of index r a π := k ,··· ,k ,··· ,k ,··· ,k ,π := h ,··· ,h ,··· ,h ,··· ,h ,p 1  1 1 m1 m1 2  1 1 m2 m2  q1 qm1   l1 lm2      are defined by | {z } | {z } | {z } | {z } m1 ζqi (k ) m2 Llj (h ) ∞ n i n j i=1 j=1 S ≡ S := , π1,π2,p mQ1kqi,mQ2hlj,p Q npQ i=1 i j=1 j nX=1 m1 ζqi (k ) m2 Llj (h ) ∞ n i n j S¯ ≡ S¯ := i=1 j=1 (−1)n−1, π1,π2,p mQ1kqi,mQ2hlj,p Q npQ i=1 i j=1 j nX=1 ∗Corresponding author. Email: [email protected] 1 m1 m2 where the quantity ω = π +π +p = (k q )+ (h l )+p being called the weight and the 1 2 i i j j i=1 j=1 X X quantities m ,m being thedegree. Here p(p > 1),m ,m ,q ,k ,h ,l are non-negative integers, 1 2 1 2 i i j j 1 ≤ k < k < ··· < k ∈Z, 1≤ h < h <··· < h ∈Z. For example, 1 2 m1 1 2 m2 ∞ H3ζ5(2)ζ2(3)L2 (1)L3 (2) ∞ L (p )L (p ) S132532,1223,4 = n n n n4 n n ,S¯0,p1p2,p = n 1np n 2 (−1)n−1 n=1 n=1 X X ∞ H2ζ4(2)ζ (3)L2 (1)L3 (2) ∞ ζ (p )ζ (p ) S¯12243,1223,5 = n n n n5 n n (−1)n−1,Sp1p2,0,p = n 1npn 2 . n=1 n=1 X X From the definition of Euler sums, there are altogether four types of linear sums: ∞ ∞ ∞ ∞ ζ (p) ζ (p) L (p) L (p) S = n ,S¯ = n (−1)n−1,S = n ,S¯ = n (−1)n−1. p,0,q nq p,0,q nq 0,p,q nq 0,p,q nq n=1 n=1 n=1 n=1 X X X X The evaluation of linear sums in terms of values of Riemann zeta function and polylogarithm function at positive integers is known when (p,q) = (1,3),(2,2), or p + q is odd [6,13]. For instance, we have ∞ H 1 11 1 1 7 S¯ = n(−1)n−1 = −2Li + ζ(4)+ ζ(2)ln22− ln42− ζ(3)ln2, 1,0,3 n3 4 2 4 2 12 4 n=1 (cid:18) (cid:19) X ∞ L (1) 3 1 1 1 S¯ = n (−1)n−1 = ζ(4)+ ζ(2)ln22− ln42−2Li . 0,1,3 n3 2 2 12 4 2 n=1 (cid:18) (cid:19) X In [12], Philippe Flajolet and Bruno Salvy gave explicit reductions to zeta values for all linear sums S ,S¯ ,S ,S¯ when p+q is an odd weight. The relationship between the values p,0,q p,0,q 0,p,q 0,p,q oftheRiemannzetafunctionandEulersumshasbeenstudiedbymany authors,forexamplesee [2-3,6-14,16-27]. The Riemann zeta function and alternating Riemann zeta function are defined respectively by [1,4,5] ∞ 1 ζ(s):= ,ℜ(s) > 1, ns n=1 X and ∞ (−1)n−1 ζ¯(s):= , ℜ(s) ≥ 1. ns n=1 X The general multiple zeta functions is defined as 1 ζ(s ,s ,··· ,s ):= , 1 2 m ns1ns2···nsm n1>n2>X···>nm>0 1 2 m where s +···+s is called the weight and m is the multiplicity. In this paper, we show that 1 m the Euler-type sums with harmonic numbers: ∞ ∞ ζ (m) L (m) n , n ,1 ≤ m,k ∈ Z n(n+k) n(n+k) n=1 n=1 X X 2 can be expressed in terms of series of Riemann zeta values and harmonic numbers. We also provide an explicit evaluation of ∞ S(n+1,p) (p−1)! ,2 ≤ p ∈ Z,1 ≤ k ∈ Z n!n(n+k) n=p−1 X in a closed form in terms of zeta values and the complete exponential Bell polynomial Y (n), k S(n,k) stands for the Stirling number of the first kind. Specifically, we investigate closed-form representations for sums of the following form: ∞ H2 ∞ H L (1) ∞ H L (1) S12,0,p = npn,S1,1,p = nnnp ,S¯1,1,p = nnnp (−1)n−1, n=1 n=1 n=1 X X X ∞ ∞ ∞ H2 L2 (1) L2 (1) S¯12,0,p = npn(−1)n−1,S0,12,p = nnp ,S¯0,12,p = nnp (−1)n−1. n=1 n=1 n=1 X X X Furthermore, we evaluate several other series involving harmonic numbers. For example H3+3H ζ (2)+2ζ (3) 3ζ(3)+ k k k k ∞ H2 1 3 nX=1n(n+n k) = k −Hk2+kζk(2) −k−1 Hi2i +ζ(2)Hk−1. (1.2) i=1 X         2 Main Theorems and Proof In this section, we usecertain integral of polylogarithm function representations to evaluate sev- eral series with alternating (or non-alternating) harmonic numbers. The polylogarithm function defined as follows ∞ xn Li (x) := ,ℜ(p) > 1, |x| ≤ 1. p np n=1 X when x takes 1 and −1, then the function Li (x) are reducible to Riemann zeta function and p alternating Riemann zeta function, respectively. Theorem 2.1 For positive integers m and k, then ∞ m−1 k−1 ζ (m) 1 H n = ζ(m+1)+ (−1)j−1ζ(m+1−j)ζ (j)+(−1)m−1 i , (2.1) n(n+k) k  k−1 im nX=1  Xj=1 Xi=1   m−1  ζ¯(m+1)+ (−1)j−1ζ¯(m+1−j)ζ (j) ∞ k−1 nX=1nL(nn(+mk)) = k1 +(−1)m−1lnXj2=(1ζk−1(m)+Lk−1(m))+(−1)mk−1 (−i1m)i−1Li(1). (2.2) i=1  X    Proof. By the definition of polylogarithm function and Cauchy product formula, we can verify that ∞ Li (x) m = ζ (m)xn, x ∈ (−1,1), (2.3) n 1−x n=1 X 3 ∞ Li (−x) − m = L (m)xn, x ∈ (−1,1). (2.4) n 1−x n=1 X Multiplying (2.3) and (2.4) by xr−1−xk−1 and integrating over (0,1), we obtain 1 ∞ Li (x) ζ (m) xr−1−xk−1 m dx = (k−r) n (0 ≤ r < k, r,k ∈ Z), (2.5) 1−x (n+r)(n+k) Z0 (cid:16) (cid:17) nX=1 1 ∞ Li (−x) L (m) xk−1−xr−1 m dx = (k−r) n (0 ≤ r < k, r,k ∈ Z). (2.6) 1−x (n+r)(n+k) Z0 (cid:16) (cid:17) Xn=1 We now evaluate the integral on the left side of (2.5) and (2.6). Noting that 1 k−r 1 Li (x) xr−1−xk−1 m dx = xr+i−2Li (x)dx, (2.7) m 1−x Z0 (cid:16) (cid:17) Xi=1Z0 1 k−r −1 Li (−x) xk−1−xr−1 m dx = (−1)r+i xr+i−2Li (x)dx. (2.8) m 1−x Z0 (cid:16) (cid:17) Xi=1 Z0 Using integration by parts we have x q−1 xn (−1)q (−1)q n xk tn−1Li (t)dt = (−1)i−1 Li (x)+ ln(1−x)(xn −1)− . q ni q+1−i nq nq k ! Z i=1 k=1 0 X X (2.9) Taking r = 0 in (2.5)-(2.8), substituting (2.9) into (2.5)-(2.8), we can obtain (2.1) and (2.2). (cid:3) If taking r ≥ 1 in (2.5)-(2.8), using (2.9), we have m−1 (−1)j−1ζ(m+1−j)(ζ (j)−ζ (j)) ∞ k−1 r−1 (n+ζrn)((mn)+k) = k−1 r Xj=1 k−1 H r−1 H , (2.10) nX=1 +(−1)m−1 i − i  im im ! i=1 i=1  X X      m−1  (−1)j−1ζ¯(m+1−j)(ζ (j)−ζ (j)) k−1 r−1   ∞ Ln(m) = 1 +Xj=(1−1)m−1ln2(ζ (m)−ζ (m)+L (m)−L (m)). nX=1(n+r)(n+k) k−r +(−1)mk−1 (−1)ki−−11L (1) r−1 k−1 r−1  im i    Xi=r (2.11)     Letting m =1 in (2.1) and (2.2), we conclude that  ∞ H 1 1 1 H n = H2+ ζ (2)+ζ(2)− k , (2.12) n(n+k) k 2 k 2 k k n=1 (cid:18) (cid:19) X 4 1+(−1)k−1 ∞ Ln(1) = 1 ζ¯(2)+ln2(Hk +Lk(1))−ln2 k . (2.13) Xn=1n(n+k) k −12 L2k(1)+ζk(2) + Lkk(1)(−1)k−1   (cid:0) (cid:1)    Theorem 2.2 For integers n ≥1,k ≥ 0, we have 1 Y (n) tn−1lnk(1−t)dt = (−1)k k , (2.14) n Z 0 where Y (n)= Y (ζ (1),1!ζ (2),2!ζ (3),··· ,(r−1)!ζ (r),···), Y (x ,x ,···) stands for the k k n n n n k 1 2 complete exponential Bell polynomial is defined by (see [15]) tm tk exp x = 1+ Y (x ,x ,···) . (2.15) m k 1 2  m! k! m≥1 k≥1 X X   Proof. Using the definition of the complete exponential Bell polynomial, it is easily shown that tk 1 1+ Y (n) = k k! t t k≥1 (1−t) 1− ··· 1− X 2 n (cid:18) (cid:19) (cid:18) (cid:19) and n 1 t = 1+ t t t t (1−t) 1− ··· 1− m=1m(1−t) 1− ··· 1− 2 n X 2 m (cid:18) (cid:19) (cid:18) (cid:19) (cid:18) (cid:19) (cid:18) (cid:19) Therefore we obtain n Y (m) Y (n)= k k−1 , Y (n)= 1. (2.16) k 0 m m=1 X It is easily shown using integration by parts that x 1 n xj tn−1ln(1−t)dt = xnln(1−x)− −ln(1−x) ,−1≤ x < 1. n j  Z0  Xj=1  Letting x → 1−, we have   x H Y (n) lim tn−1ln(1−t)dt = − n = − 1 . (2.17) x→1−Z0 n n Using integration by parts and (2.17) , we can find that x 1 2 n 1 k xj tn−1ln2(1−t)dt = (xn −1)ln2(1−x)− xkln(1−x)− −ln(1−x) n n k  j  Z0 Xk=1  Xj=1  (2.18)   5 Letting x tend to 1− in (2.18), we deduce that x 2 n 1 k 1 Y (n) lim tn−1ln2(1−t)dt = = 2 . (2.19) x→1−Z0 n k=1 k j=1 j n X X Further, it is easily verify that x (−1)m n 1 k1 1 km−1 1 Y (n) lim tn−1lnm(1−t)dt = m! ··· = (−1)m m ,1 ≤ m ∈ Z. x→1−Z0 n kX1=1 k1 kX2=1k2 kXm=1km n (2.20) We complete the proof of (2.14). (cid:3) From the definition of the complete exponential Bell polynomial, we have Y (n)= H ,Y (n)= H2+ζ (2),Y (n)= H3+3H ζ (2)+2ζ (3), 1 n 2 n n 3 n n n n Y (n)= H4+8H ζ (3)+6H2ζ (2)+3ζ2(2)+6ζ (4). 4 n n n n n n n Theorem 2.3 For integer p ≥ 2,k ≥ 1, we have ∞ S(n+1,p) 1 Y (k) Y (k) p p−1 (p−1)! = (p−1)!ζ(p)+ − . (2.21) n!n(n+k) k p k n=p−1 (cid:26) (cid:27) X where S(n,k) are the Stirling number of the first kind defined by n x x n!(1+x) 1+ ··· 1+ = S(n+1,k+1)xk. 2 n (cid:16) (cid:17) (cid:16) (cid:17) Xk=0 From the definition S(n,k), we can rewrite it as n n x S(n+1,k+1)xk = n!exp ln 1+  j  Xk=0 Xj=1 (cid:18) (cid:19)  n ∞ xk = n!exp (−1)k−1  kjk Xj=1Xk=1  ∞ ζ (k)xk = n!exp (−1)k−1 n  . k ( ) k=1 X Therefore, we know that S(n,k) is a rational linear combination of products of harmonic num- bers. The following identities is easily derived (n−1)! S(n,1)= (n−1)!,S(n,2)= (n−1)!H ,S(n,3) = H2 −ζ (2) , n−1 2 n−1 n−1 (n−1)! (cid:2) (cid:3) S(n,4)= H3 −3H ζ (2)+2ζ (3) , 6 n−1 n−1 n−1 n−1 (n−1)!(cid:2) (cid:3) S(n,5)= H4 −6ζ (4)−6H2 ζ (2)+3ζ2 (2)+8H ζ (3) . 24 n−1 n−1 n−1 n−1 n−1 n−1 n−1 (cid:2) (cid:3) 6 Proof. From [15], we have ∞ xn lnp(1−x)= (−1)pp! S(n,p) ,1 ≤ p ∈ Z,−1≤ x < 1. (2.22) n! n=p X Differentiating this equality, we obtain lnp−1(1−x) ∞ xn = (−1)p−1(p−1)! S(n+1,p) ,p ≥ 2. (2.23) 1−x n! n=p−1 X Let k,r be integers with k > r ≥ 0, then multiplying (2.23) by xr−1−xk−1 and integrating over (0,1), we get 1 lnp−1(1−x) ∞ (k−r)S(n+1,p) xr−1−xk−1 dx = (−1)p−1(p−1)! ,p ≥ 2. (2.24) 1−x n!(n+r)(n+k) Z0 (cid:16) (cid:17) n=Xp−1 Noting that 1 lnp−1(1−x) k−r 1 xr−1−xk−1 dx = xr+i−2lnp−1(1−x)dx (2.25) 1−x Z0 (cid:16) (cid:17) Xi=1Z0 and 1 lnp−1(1−x) dx = (−1)p−1(p−1)!ζ(p), 2 ≤ p ∈ Z. (2.26) x Z 0 Taking r = 0,k ≥ 1 in (2.24), using (2.14)(2.16)(2.25) and (2.26), we can obtain (2.21). (cid:3) If taking r ≥ 1 in (2.24), then we have ∞ S(n+1,p) 1 (p−1)!(k−r) = {Y (k−1)−Y (r−1)}. (2.27) p p n!(n+r)(n+k) p n=p−1 X Letting p = 3,4 in (2.8), we can give the following Corollary. Corollary 2.4 If 1≤ k ∈ Z, then we have ∞ H2−ζ (2) 1 H3+3H ζ (2)+2ζ (3) H2+ζ (2) n n = 2ζ(3)+ k k k k − k k , (2.28) n(n+k) k 3 k n=1 (cid:26) (cid:27) X H4+8H ζ (3)+6H2ζ (2)+3ζ2(2)+6ζ (4) ∞ H3−3H ζ (2)+2ζ (3) 1 k k k k k k k n n n n = 4 . nX=1 n(n+k) k −Hk3+3Hkζkk(2)+2ζk(3) +6ζ(4)  (2.29)     From (2.1), taking m = 2, we have ∞ k−1 ζ (2) 1 H n i = ζ(3)+ζ(2)H − . (2.30) n(n+k) k k−1 i2 ( ) n=1 i=1 X X Substituting (2.30) into (2.28), we obtain (1.2). In the same manner, we obtain the following Theorem: 7 Theorem 2.5 For integer m,k > 0, then ∞ ζ (m) n (−1)n−1 n+k n=1 X ∞ ζ (m) = (−1)k n (−1)n−1−ζ¯(m+1) +(−1)k+m+1ln2(ζ (m)+L (m)) n k−1 k−1 ! n=1 X m−1 k−1 L (1) +(−1)k (−1)j−1ζ¯(m+1−j)L (j)+(−1)m+k i , (2.31) k−1 im j=1 i=1 X X ∞ L (m) n (−1)n−1 n+k n=1 X ∞ k−1 L (m) H = (−1)k−1 ζ(m+1)− n (−1)n−1+(−1)m i(−1)i−1 n im ! n=1 i=1 X X m−1 +(−1)k (−1)j−1ζ(m+1−j)L (j) (2.32) k−1 j=1 X From (2.31) and (2.32), we obtain ∞ k−1 H 1 L (1) n (−1)n−1 = (−1)k−1 ln22−ln2(H +L (1))+ i , (2.33) n+k 2 k−1 k−1 i ! n=1 i=1 X X ∞ L (1) ζ(2)−ln22 k−1 H n (−1)n−1 = (−1)k−1 − i(−1)i−1 . (2.34) n+k 2 i ! n=1 i=1 X X Theorem 2.6 For 1≤ l ,l ,m ∈ Z and x,y,z ∈ [−1,1), we have the following relation 1 2 ∞ ∞ ∞ ζ (l ,x)ζ (l ,y) ζ (l ,x)ζ (m,z) ζ (l ,y)ζ (m,z) n 1 n 2 zn+ n 1 n yn+ n 2 n xn nm nl2 nl1 n=1 n=1 n=1 X X X ∞ ∞ ∞ ζ (m,z) ζ (l ,x) ζ (l ,y) = n (xy)n+ n 1 (yz)n+ n 2 (xz)n nl1+l2 nm+l2 nl1+m n=1 n=1 n=1 X X X +Li (z)Li (x)Li (y)−Li (xyz) (2.35) m l1 l2 l1+l2+m n xk where the partial sum ζ (l,x) of polylogarithm function is defined by ζ (l,x) := . n n kl k=1 X ∞ Proof. We construct the function F (x,y,z) := {ζ (l ,x)ζ (l ,y)−ζ (l +l ,xy)}zn−1, n 1 n 2 n 1 2 n=1 X z ∈(−1,1). By the definition of ζ (l,x), we have n ∞ ζ (l ,x) ζ (l ,y) F (x,y,z)= zF (x,y,z)+ n 1 yn+1+ n 2 xn+1 zn. (2.36) ((n+1)l2 (n+1)l1 ) n=1 X 8 Moving zF(x,y,z) from right to left and then multiplying (1−z)−1 to the equation (2.36) and integrating over the interval (0,z), we obtain ∞ ζ (l ,x)ζ (l ,y)−ζ (l +l ,xy) n 1 n 2 n 1 2 zn n n=1 X ∞ ζ (l ,x) ζ (l ,y) = n 1 yn+1+ n 2 xn+1 {Li (z)−ζ (1,z)}. (2.37) ((n+1)l2 (n+1)l1 ) 1 n n=1 X Furthermore, using integration and the following formula ∞ ζ (l ,x) ζ (l ,y) n 1 yn+1+ n 2 xn+1 = Li (x)Li (y)−Li (xy), ((n+1)l2 (n+1)l1 ) l1 l2 l1+l2 n=1 X we deduce (2.35) to complete the proof of Theorem 2.6. (cid:3) Letting (x,y,z) = (−1,−1,1), (l ,l ,m) = (1,1,m) and (x,y,z) = (−1,−1,−1), (l ,l ,m) = 1 2 1 2 (1,1,m) in (2.35) gives the following Corollary 2.7 For integer m > 1 we have that ∞ ∞ L2 (1) L (1)ζ (m) n +2 n n (−1)n−1 nm n n=1 n=1 X X ∞ ∞ ζ (m) L (1) = n +2 n (−1)n−1+ln22ζ(m)−ζ(m+2), (m > 1) (2.38) n2 nm+1 n=1 n=1 X X ∞ L2 (1) ∞ L (1)L (m) n (−1)n−1+2 n n (−1)n−1 nm n n=1 n=1 X X ∞ ∞ L (m) L (1) = n +2 n +ln22ζ¯(m)−ζ¯(m+2), (m > 0). (2.39) n2 nm+1 n=1 n=1 X X In fact, multiplying (2.35) by (1−z)−1 and integrating over (0,z) (z ∈ (−1,1)), we can obtain the following Corollary. Corollary 2.8 For positive integers l > 0,l > 0,m > 1 and x,y,z ∈ [−1,1), then we have 1 2 n n ζ (1,z) ζ (1,z) k k ζ (l ,x) ζ (l ,y) ∞ ∞ n 1 km ∞ n 2 km ζ (l ,x)ζ (l ,x)ζ (1,z) ! ! n 1 n 2 n + Xk=1 yn+ Xk=1 xn nm nl2 nl1 n=1 n=1 n=1 X X X n ζ (1,z) k ∞ km ∞ ∞ ! ζ (l ,x)ζ (1,z) ζ (l ,y)ζ (1,z) = Xk=1 xnyn+ n 1 n yn+ n 2 n xn nl1+l2 nm+l2 nm+l1 n=1 n=1 n=1 X X X ∞ ∞ ζ (1,z) ζ (1,z) +Li (x)Li (y) n − n xnyn . (2.40) l1 l2 nm nm+l1+l2 ! ! n=1 n=1 X X 9 Letting x,y,z → 1 in (2.40), we arrive at the conclusion that, for integers m,l ,l > 1 1 2 n n H H k k ζ (l ) ζ (l ,y) ∞ ∞ n 1 km ∞ n 2 km H ζ (l )ζ (l ) ! ! n n 1 n 2 k=1 k=1 + X + X nm nl2 nl1 n=1 n=1 n=1 X X X n H k ∞ km ∞ ∞ ∞ ∞ ! H ζ (l ) H ζ (l ) H H = Xk=1 + n n 1 + n n 2 +ζ(l )ζ(l ) n − n . nl1+l2 nm+l2 nm+l1 1 2 nm nm+l1+l2 ! ! n=1 n=1 n=1 n=1 n=1 X X X X X (2.41) From [13], we derive the following identity ∞ n ∞ ∞ ∞ 1 H H H H ζ (p+1) k n n n n = +ζ(p+1) − . (2.42) np+1 km nm+p+1 nm nm ! n=1 k=1 n=1 n=1 n=1 X X X X X Therefore, we may rewrite (2.41) as n n H H k k ζ (l ) ζ (l ,y) ∞ ∞ n 1 km ∞ n 2 km H ζ (l )ζ (l ) ! ! n n 1 n 2 k=1 k=1 + X + X nm nl2 nl1 n=1 n=1 n=1 X X X ∞ H ζ (l ) H ζ (l ) H ζ (l +l ) n n 1 n n 2 n n 1 2 = + − nm+l2 nm+l1 nm n=1(cid:26) (cid:27) X ∞ ∞ H H n n +ζ(l +l ) +ζ(l )ζ(l ) . (2.43) 1 2 nm 1 2 nm ! ! n=1 n=1 X X Taking l = l = m = 2l+1 (l is a positive integer) in (2.43), we conclude that 1 2 n H k ζ (2l+1) ∞ H ζ2(2l+1) ∞ n k2l+1! n n +2 Xk=1 n2l+1 n2l+1 n=1 n=1 X X ∞ ∞ H ζ (2l+1) H ζ (4l+2) H = 2 n n − n n + ζ(4l+2)+ζ2(2l+1) n . (2.44) n4l+2 n2l+1 n2l+1 ! n=1(cid:26) (cid:27) n=1 X (cid:0) (cid:1) X 3 Closed form of quadratic Euler sums In this section we evaluate some quadratic Euler sums involving harmonic numbers and alter- nating harmonic numbers. Theorem 3.1 For integers l > 0,m > 0,p > 1, we have ∞ n ∞ n ζ (l) H ζ (m) H (−1)m−1 n k −(−1)p+l n k np+1 km np+1 kl ! ! n=1 k=1 n=1 k=1 X X X X 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.