Some Infinite Series Involving the Riemann Zeta Function

Some Infinite Series Involving the Riemann Zeta Function

International Journal of Mathematics and M Computer Science, 7(2012), no. 1, 11–83 CS Some infinite series involving the Riemann zeta function Donal F. Connon Elmhurst Dundle Road Matfield, Kent TN12 7HD, United Kingdom email: [email protected] (Received May 16, 2010, Accepted July 12, 2012) Abstract This paper considers some infinite series involving the Riemann zeta function. Some examples are set out below ∞ n n − kxk 1 ( 1) s x, s ,y − x x, s, y n k k y s = Φ( +1 ) log Φ( ) n=0 +1k=0 ( + ) ∞ n n k − n 2 1 − 1−s ζ s ( 1) k k s = 1 2 ( ) n=0 k=0 ( +1) 2 ∞ n n − k − k 1 ( 1) ( 1) π πy n+1 k k y + k − y = cosec( ) n=0 2 k=0 + +1 ∞ 2n 2n u n −1 2 sin 2 u2 u u u u − ζ n3 n =4 log(2 sin )+2Cl3(2 )+4 Cl2(2 ) 2 (3) n=1 ∞ 4 7 1 24n [n!] ζ(3) − πG = n 3 n 2 4 2 n=0 (2 +1) [(2 )!] where Φ(x, s, y) is the Hurwitz-Lerch zeta function and Cln(t) are the Clausen functions. Key words and phrases: Infinite series, Riemann zeta Function. AMS (MOS) Subject Classifications: 40A05, 11M32. 12 Donal F. Connon 1 Some Hasse-type series It was shown in [21] that ∞ n k k y 1 n (−1) y log y = Lis(y) − Lis−1(y)(1) − k s−1 − s 1 n=0 n +1 k=0 (k +1) s 1 where Lis(y) is the polylogarithm function [39] ∞ yn Lis(y)= s n=1 n and, with y = 1, this becomes the formula originally discovered by Hasse [32] in 1930 ∞ n k 1 1 n (−1) = Lis(1) = ζ(s)(2) − k s−1 s 1 n=0 s +1 k=0 (k +1) ∞ s−1 −t The gamma function is defined as Γ(s)= 0 t e dt, s>0. Using the substitution t =(k + y)u,weobtain ∞ 1 1 s−1 −(k+y)u s = u e du (3) (k + y) Γ(s) 0 We now consider the finite sum set out below n k n x Sn(x, y)= (4) k s k=0 (k + y) and combine (3) and (4) to obtain n k n ∞ n x n k 1 s−1 −(k+y)u Sn(x, y)= = x u e du k s k k=0 (k + y) k=0 Γ(s) 0 ∞ n 1 s−1 n k −(k+y)u = u x e du k Γ(s) 0 k=0 Some infinite series involving the Riemann zeta function 13 ∞ n 1 s−1 −yu n −u k = u e xe du k Γ(s) 0 k=0 Using the binomial theorem this becomes ∞ 1 s−1 −yu −u n Sn(x, y)= u e 1+xe du (5) Γ(s) 0 Making the summation, we see that ∞ n k ∞ ∞ n n x 1 n s−2 −yu −u n t = t u e 1+xe du k s n=0 k=0 (k + y) Γ(s) n=0 0 The geometric series gives us for |t (1 + xe−u)|< 1 ∞ n 1 t xe−u 1+ = − −u n=0 1 t (1 + xe ) and we then have ∞ n k ∞ s−1 −yu n n x 1 u e t = du (6) k s − −u n=0 k=0 (k + y) Γ(s) 0 1 t (1 + xe ) We now integrate (6) with respect to t ∞ n+1 n k w ∞ s−1 −yu w n x 1 u e = dt du k s − −u n=0 n +1 k=0 (k + y) Γ(s) 0 0 1 t(1 + xe ) ∞ s−1 −yu − −u − 1 u e log [1 w(1 + xe )] = −u du Γ(s) 0 1+xe and obtain 14 Donal F. Connon ∞ n+1 n k ∞ s−1 −(y−1)u −u w n x 1 u e log [1 − w(1 + xe )] = − du k s u n=0 n +1 k=0 (k + y) Γ(s) 0 e + x (7) When w =1andx →−x we get ∞ n k k ∞ s−1 −(y−1)u −u 1 n (−1) x 1 u e log [xe ] = − du (8) k s u − n=0 n +1 k=0 (k + y) Γ(s) 0 e x ∞ s −(y−1)u ∞ s−1 −(y−1)u 1 u e − log x u e = u du u du Γ(s) 0 e − x Γ(s) 0 e − x We see from (3) that k k ∞ x x s−1 −(k+y)u s = u e du (k + y) Γ(s) 0 and we have the summation ∞ k ∞ ∞ x 1 s−1 −yu k −ku s = u e x e du k=0 (k + y) Γ(s) 0 k=0 1 ∞ us−1e−yu = −u du Γ(s) 0 1 − xe We therefore obtain the well-known formula [43, p. 121] for the Hurwitz- Lerch zeta function Φ(x, s, y) ∞ xk 1 ∞ us−1e−(y−1)u x, s, y du Φ( )= s = u − (9) k=0 (k + y) Γ(s) 0 e x and with y =1weobtain ∞ k ∞ k x 1 x Lis(x) Φ(x, s, 1) = s = s = k=0 (k +1) x k=1 k x Some infinite series involving the Riemann zeta function 15 giving us [46, p. 280] x ∞ us−1 Lis(x)= u du (10) Γ(s) 0 e − x Reference to (8) then shows that ∞ n k k 1 n (−1) x = s Φ(x, s +1,y) − log x Φ(x, s, y) (11) k s n=0 n +1 k=0 (k + y) With x = 1 we obtain Hasse’s formula (2) ∞ n k 1 n (−1) = s Φ(1,s+1,y)=sζ(s +1,y) k s n=0 n +1 k=0 (k + y) With y = 1 equation (11) becomes ∞ n k k 1 n (−1) x = s Φ(x, s +1, 1) − log x Φ(x, s, 1) k s n=0 n +1 k=0 (k +1) or equivalently ∞ n k k 1 n (−1) x s 1 = Lis+1(x) − log xLis(x) (12) k s n=0 n +1 k=0 (k +1) x x which corresponds with (1). A closed form expression may be obtained for example with x =1/2ands =2. We see from (6) that ∞ n k ∞ s−1 −yu n n x 1 u e t = du k s − − −u − n=0 k=0 (k + y) (1 t)Γ(s) 0 1 e tx/(1 t) 1 ∞ us−1e−(y−1)u = u du (1 − t)Γ(s) 0 e − tx/(1 − t) 16 Donal F. Connon and hence referring to (9) we have ∞ n k n n x 1 tx t = Φ ,s,y (13) k s − − n=0 k=0 (k + y) 1 t 1 t With t =1/2weobtain ∞ n k 1 n x =Φ(x, s, y) (14) n+1 k s n=0 2 k=0 (k + y) which is clearly an example of Euler’s transformation of series [33, p. 244]. tx Letting w = 1−t , (13) may be represented by ∞ n n k w n x w + x = Φ(w, s, y) (15) k s n=0 w + x k=0 (k + y) x and with x = −1 this becomes ∞ n n k −w n (−1) =(1− w)Φ (w, s, y) (16) − k s n=0 1 w k=0 (k + y) as previously reported by Guillera and Sondow [57]. Letting w =1andx = −1 in (15) results in another identity due to Hasse ∞ n k 1 n (−1) =Φ(−1,s,y)=ζa (s, y) (17) n+1 k s n=0 2 k=0 (k + y) where ζa (s, y) is the alternating Hurwitz-Lerch zeta function ∞ (−1)k ζa (s, y)= s k=0 (k + y) With y =1wehave ∞ n k 1 n (−1) = ζa (s, 1) = ζa (s) (18) n+1 k s n=0 2 k=0 (k +1) Some infinite series involving the Riemann zeta function 17 Letting w = x = 1 in (15) results in ∞ n 1 n 1 =Φ(1,s,y)=ζ (s, y) (19) n+1 k s n=0 2 k=0 (k + y) Letting s =1andy =1/2 in (17) results in ∞ n k 1 n (−1) = ζa (1, 1/2) n k n=0 2 k=0 2k +1 We have from [15, p. 523] π ζa (1, 1/2) = 2 and we therefore obtain ∞ n k 1 n (−1) π = n k n=0 2 k=0 2k +1 2 which was also determined in a different manner in equation (8.11e) in [22]. Since [15, p. 523] ζa (1,y)+ζa (1, 1 − y)=π cosec(πy) we have for 0 <y<1 ∞ n k k 1 n (−1) (−1) + = π cosec(πy) n+1 k − n=0 2 k=0 k + y k +1 y We now divide (7) by w and integrate to obtain ∞ n+1 n k v −u ∞ s−2 −(y−1)u v n x 1 log [1 − w(1 + xe )] u e = − dw du 2 k s u n=0 (n +1) k=0 (k + y) Γ(s) 0 w 0 e + x Since 18 Donal F. Connon log(1 − ax) dx = −Li2(ax) x we have ∞ n+1 n k ∞ s−1 −(y−1)u −u v n x 1 u e Li2 [v(1 + xe )] = du 2 k s u n=0 (n +1) k=0 (k + y) Γ(s) 0 e + x (20) Further operations of the same kind will result in ∞ n+1 n k ∞ s−1 −(y−1)u −u v n x 1 u e Lir [v(1 + xe )] = du r k s u n=0 (n +1) k=0 (k + y) Γ(s) 0 e + x (21) With t = −1 in (6) we obtain ∞ n k ∞ s−1 −yu n n x 1 u e (−1) = du k s −u n=0 k=0 (k + y) Γ(s) 0 2+xe and letting x →−x we get ∞ n k k ∞ s−1 −yu n n (−1) x 1 u e (−1) = du k s − −u n=0 k=0 (k + y) Γ(s) 0 2 xe With x =2wehave ∞ n k k ∞ s−1 −yu ∞ s−1 −(y−1)u n n (−1) 2 1 u e 1 u e (−1) = du = du k s − −u u − n=0 k=0 (k + y) 2Γ(s) 0 1 e 2Γ(s) 0 e 1 Hence we obtain using (9) ∞ n k k n n (−1) 2 1 (−1) = Φ(1,s,y) (22) k s n=0 k=0 (k + y) 2 Some infinite series involving the Riemann zeta function 19 and with s = n +1andy = 1 this becomes ∞ n k k n n (−1) 2 1 (−1) = ζ(n + 1) (23) k n+1 n=0 k=0 (k +1) 2 This identity, in the case where n is a positive integer, was recently reported by Alzer and Koumandos [6] where it was derived in a very different manner.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    73 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us