The Riemann and Hurwitz zeta functions, Apery s constant and new rational series representations involving ζ(2k)

Transcription

1 The Riemann and Hurwitz zeta functions, Apery s constant and new rational series representations involving ζ(k) Cezar Lupu 1 1 Department of Mathematics University of Pittsburgh Pittsburgh, PA, USA Algebra, Combinatorics and Geometry Graduate Student Research Seminar, February, 017, Pittsburgh, PA

2 A quick overview of the Riemann zeta function. The Riemann zeta function is defined by ζ(s) = 1, Re s > 1. ns Originally, Riemann zeta function was defined for real arguments. Also, Euler found another formula which relates the Riemann zeta function with prime numbrs, namely ζ(s) = p 1 (1 1 p s ), where p runs through all primes p =, 3, 5,....

3 A quick overview of the Riemann zeta function. Moreover, Riemann proved that the following ζ(s) satisfies the following integral representation formula: where Γ(s) = ζ(s) = 1 Γ(s) 0 0 u s 1 e u du, Re s > 1, 1 t s 1 e t dt, Re s > 0 is the Euler gamma function. Also, another important fact is that one can extend ζ(s) from Re s > 1 to Re s > 0. By an easy computation one has and therefore we have (1 1 s )ζ(s) = ( 1) n 1 1 n s,

4 A quick overview of the Riemann function. ζ(s) = s ( 1) n 1 1, Re s > 0, s 1. ns It is well-known that ζ is analytic and it has an analytic continuation at s = 1. At s = 1 it has a simple pole with residue 1. We have lim (s 1) ζ(s) = 1. s 1 Let us remark that the alternating zeta function is called the Dirichlet eta function and it is defined as η(s) = ( 1) n 1 1, Re s > 0, s 1. ns

5 A quick overview of the Hurwitz zeta function. Another important function is the Hurwitz (generalized) zeta function defined by ζ(s, a) = n=0 1, Re s > 1, a 0, 1,,.... (n + a) s As the Riemann zeta function, Hurwitz zeta function is analytic over the whole complex plane except s = 1 where it has a simple pole. Also, from the two definitions, one has ζ(s) = ζ(s, 1) = 1 s 1 ζ ( s, 1 ) = 1 + ζ(s, ).

6 A quick overview of the Hurwitz zeta function. It can also be extended by analytic continuation to a meromorphic function defined for all complex numbers s 1. At s = 1 it has a simple pole with residue 1. The constant term is given by lim s 1 ( ζ(s, a) 1 s 1 ) = Γ (a) Γ(a) = Ψ(a), where Ψ is the digamma function. Also, the Hurwitz zeta function is related to the polygamma function, Ψ m (z) = ( 1) m+1 m!ζ(m + 1, z).

7 A quick overview of the Dirichlet beta function. Last but not least, we define the Dirichlet beta function as β(s) = n=0 ( 1) n, Re s > 0. (n + 1) s Alternatively, one can express the beta function by the following formula: β(s) = 1 (ζ(s, 1/4) ζ(s, 3/4)). 4s Equivalently, β(s) has the following integral representation: ζ(s) = 1 t s 1 e t dt. Γ(s) e t

8 A quick overview of the Dirichlet beta function. Note that β() = G (Catalan s constant), β(3) = π3 3, and β(n + 1) = ( 1)n E n π n+1 4 n+1 (n)! where E n are the Euler numbers in the Taylor series Other special values include e t + e t = n=0 E n n! tn. β(0) = 1, β(1) = π 4, β( k) = E k.

9 What is known about the values of ζ(s) at integers? ζ( n) = 0 for n = 1,,... (trivial zeros) ζ( n) = ( 1) n B n+1 n+1 ; with ζ( 1) = 1 1 The values ζ(n), for n = 1,,... have been found by Euler in 1740 The values ζ( n + 1), for n = 1,,... can be evaluated in terms of ζ(n). In fact, we have ζ( n + 1) = (π) n ( 1) n (n 1)!ζ(n). There is a mystery about ζ(n + 1) values ζ(0) = 1 ζ(1) does not exist, but one has the following ( lim ζ(s) 1 ) = γ. s 1 s 1

10 Another quick look at ζ(n) and ζ(n + 1) In 1734, Euler produced a sensation when he discovered that 1 n = π 6. Later, the same Euler generalized the above formula, ζ(n) = ( 1) n+1 Bn n 1 π n, (n)! where the coefficients B n are the so-called Bernoulli numbers and they satisfy z e z 1 = n=0 B n n! zn, z < π.

11 Another quick look at ζ(n) and ζ(n + 1) An elementary but sleek proof of Euler s result was recently given in E. De Amo, M. Diaz Carrillo, J. Hernandez-Sanchez, Another proof of Euler s formula for ζ(k), Proc. Amer. Math. Soc. 139 (011), The authors proved Euler s formula using the Taylor series for the tangent function and Fubini s theorem. Unlike ζ(n), the values ζ(n + 1) are still mysterious! One of the most important results was produced by Roger Apery in 1979, when he proved that ζ(3) is irrational by using the "fast converging" series representation ζ(3) = 5 ( 1) n n 3( ). n n

12 Another quick look at ζ(n) and ζ(n + 1) Amazingly, there exist similar formulas for ζ() and ζ(4), namely ζ() = 3 1 n ( ), ζ(4) = 36 n 17 n 1 n 4( ). n n Recently, other substantial results were obtained. In 00, K. Ball and T. Rivoal proved the following K. Ball, T. Rivoal, Irrationalite d une infinite de la fonction zetaaux entiers impairs, Invent. Math. 146 (001), Theorem There are infinitely many irrational values of the Riemann zeta function at odd positive integers. Moreover, if N(n) = #{irrational numbers among ζ(3), ζ(5),..., ζ(n + 1)}, 1 then N(n) log n for large n. (1 + log )

13 Another quick look at ζ(n) and ζ(n + 1) Other remarkable results in this direction are given by Rivoal (001) and Zudilin (001), Theorem At least four numbers ζ(5), ζ(7),..., ζ(1) are irrational. and W. Zudilin, One of the numbers ζ(5), ζ(7), ζ(9), ζ(11) is irrational, Russ. Math. Surv. 56 (001), Theorem At least one of the four numbers ζ(5), ζ(7), ζ(9), ζ(11) is irrational.

14 Some Taylor series representations For the sake of completeness we display the Taylor series for the tangent, cotangent, secant and cosecant functions: tan x = csc x = cot x = ( 1) n 1 n ( n 1)B n x n 1, x < π (n)! n=0 sec x = n=0 (1) ( 1) n n B n x n 1, x < π () (n)! n=0 ( 1) n E n x n, x < π (n)! (3) ( 1) n+1 ( n 1 1)B n x n 1, x < π (4) (n)!

15 Clausen integral The Clausen function (Clausen integral), introduced by Thomas Clausen in 183, is a transcendental special function of single variable and it is defined by Cl (θ) := k=1 sin kθ k θ ( ( x )) = log sin 0 It is intimately connected with the polylogarithm, inverse tangent integral, polygamma function, Riemann zeta function, Dirichlet eta function, and Dirichlet beta function. Some well-known properties of the Clausen function include periodicity in the following sense: Cl (kπ ± θ) = Cl (±θ) = ± Cl (θ). dx.

16 Clausen integral Moreover, it is quite clear from the definition that Cl (kπ) = 0 for k integer. For example, for k = 1 we deduce π 0 ( ( x )) log sin ( π By periodicity we have Cl the Catalan constant defined by G := π dx = 0, log(sin x) dx = π log. 0 ) ( ) 3π = Cl = G, where G is n=0 ( 1) n (n + 1)

17 Clausen integral. Evaluation of some elementary integrals More generally, one can express the above integral as the following: θ 0 θ 0 log(sin x) dx = 1 Cl (θ) θ log, log(cos x) dx = 1 Cl (π θ) θ log, and θ 0 log(1 + cos x) dx = Cl (π θ) θ log, θ 0 log(1 + sin x) dx = G Cl ( π + θ ) θ log.

18 Clausen acceleration formula Theorem We have the following representation for the Clausen function Cl (θ), Cl (θ) θ = 1 log θ + ζ(n) (π) n n(n + 1) θn, θ < π. (5) Sketch of the proof. Integrating by parts the function xy cot(xy) and using the product formulas for the sine function, we have π 0 xy cot(xy)dx = π ( πy ) log π k=1 ( y 4k ) n n + π log +

19 Clausen acceleration formula + 1 y Cl (πy). On the other hand, by the Taylor series for the cotangent function, we obtain π 0 Therefore, we obtain xy cot(xy) dx = π ζ(n) ( ) π n+1 π n (n + 1) y n. π ζ(n) ( ) π n+1 π n (n + 1) y n = π log(πy) π Equate the coefficients of y n and we obtain our formula. ζ(n) n4 n y n + 1 y Cl (πy).

20 Clausen acceleration formula. Some remarks. In particular case of θ = π, using the fact that Cl ( π ) = G, we obtain ζ(n) n(n + 1)16 n = G ( π ) π 1 + log. (6) Recently, Wu, Zhang and Liu developed the following representation for the Clausen function, J. Wu, X. Zhang, D. Liu, An efficient calculation of the Clausen functions, BIT Numer. Math. 50 (010), ( Cl (θ) = θ θ log sin θ ) ζ(n) (n + 1)(π) n θn+1, θ < π. (7)

21 Clausen acceleration formula. Some well-known representation for ζ(3). It is interesting to see that integrating the above formula from 0 to π/ we have the following representation for ζ(3) due to Choi, Srivastava and Adamchik, H. M. Srivastava, M. L. Glasser, V. S. Adamchik, Some definite integrals associated with the Riemann zeta function, J. Z. Anal. Anwendungen. 19 (000), ζ(3) = 4π 35 ( 1 + G ) π ζ(n) (n + 1)(n + 1)16 n. (8)

22 Other representations for ζ(3) Also, Strivastava, Glasser and Adamchik derive series representations for ζ(n + 1) by evaluating the integral π/ω 0 t s 1 cot tdt, s, ω integers in two different ways. One of the ways involves the generalized Clausen functions. When they are evaulated in terms of ζ(n + 1) one obtains the following formula for ζ(3), ζ(3) = π 9 ( log + n=0 ) ζ(n) (n + 3)4 n. (9)

23 Other representations for ζ(3) Other representations for Apery s constant are given by Cvijovic and Klinowski, D. Cvijovic, J. Klinowski, New rapidly convergent series representations for ζ(n + 1), Proc. Amer. Math. Soc. 15 (1997), and ζ(3) = π 3 n=0 (n + 5)ζ(n). (10) (n + 1)(n + )(n + 3)n ζ(3) = 4π 7 n=0 ζ(n). (11) (n + 1)(n + )n

24 New series representations for Apery s constant ζ(3) Theorem ζ(3) = 4π 35 ( 3 ( π ) log + ( ζ(3) = 64 4 ( π ) 3π β(4)+8π 9 3 log and ) ζ(n) n(n + 1)(n + 1)16 n, (1) + 3 ) ζ(n) n(n + 1)(n + 3)16 n, (13) ( ζ(3) = 64 16π 1 β(4)+ 3π 7 + 3G ) π 3 ζ(n) (n + 1)(n + 3)16 n, (14)

25 Ideas of the proof The main ingredients in the proof of the above theorem are the following: Fubini s theorem, Clausen acceleration formulas and π/4 u log(sin u)du = ζ(3) πg 8 π log to find 3 0 π/ 0 Cl (y) = 35 3 ζ(3). It can be proven using Fubini s theorem a formula for π /4 0 Cl ( y)dy which combined with the polygamma formula related to the Hurwitz zeta function give us π /4 0 Cl ( y)dy = 3π ζ(3) + β(4). 3

26 Rational series representation involving ζ(n) We shall call rational ζ-series of a real number x, the following representation: x = q n ζ(n, m), n= where q n is a rational number and ζ(n, m) is the Hurwitz zeta function. For m > 1 integer, one has m 1 x = ζ(n) j n. n= q n j=1 J. M. Borwein, D. M. Bradley, R. E. Crandall, Computational strategies for the Riemann zeta function, J. Comp. Appl. Math. 11 (000),

27 Rational series representation involving ζ(n). Examples In the particular case m =, one has the following series representations: 1 = 1 γ = log = (ζ(n) 1) n= n= n= 1 (ζ(n) 1) n 1 (ζ(n) 1), n where γ is the Euler-Mascheroni constant.

28 New rational series representation involving ζ(n) Theorem The following representation is true ζ(n) n4 n ( ) n = m 1 m 1 m ( ( ζ(m) 1 1 ) ) m 1 m odd, m even. (15)

29 Some corollaries In particular cases we obtain the following Corollary We have and ζ(n) = log π 1. (16) n(n + 1)4n

30 Some corollaries Corollary We have the following series representations ζ(n) 4 n = 1, (17) ζ(n)(n 1)(n ) 4 n = 1, (18) ζ(n)(n 1) 4 n = π 8 1, (19) ζ(n)n 4 n = π 16, (0)

31 New rational series representation involving ζ(n) Theorem We have the following series representation ζ(n) n16 n ( ) n = m 1 (1 β(m)) m m 1 m ( ( ζ(m) 1 1 ) ) m 1 odd, m even, ()

32 Some corollaries again From the previous theorem we recover some well-known rational series representations for π Corollary and ζ(n) n16 n ( ) π = log, (3) ζ(n) 16 n = 4 π. (4) 8

33 Some corollaries again and again Corollary We have the following series ζ(n) n4 n ( 1 1 ) ( ) n 4 n = ζ(k) (1 14 ) k k k, (5) and ζ(n) n4 n ( 1 1 ) ( ) n β(k + 1) 4 n = k + 1 k + 1. (6)

34 And one more... Corollary We have ζ(n)(n 1) 16 n = π 16 1, (7) ζ(n)(n 1)(n ) 16 n = 1 π3 96, (8) ζ(n)n ζ(n)n 16 n = π ( π ) 16 1, (9) 16 n = π 3 ( ) 3π π 4 1. (30)

35 Thank you for your attention!!!