In press at Algebras, Groups and Geometreis, Vol.

21, 2004

DISPROOFS OF RIEMANNS HYPOTHESIS

Chun-Xuan, Jiang P.O.Box 3924, Beijing 100854, China and Institute for Basic Research P.O.Box 1577, Palm Harbor, FL 34682, U.S.A. [email protected]

Abstract As it is well known, the Riemann hypothesis on the zeros of the (s) function has been assumed to be true in various basic developments of the 20-th century mathematics, although it has never been proved to be correct. The need for a resolution of this open historical problem has been voiced by several distinguished mathematicians. By using preceding works, in this paper we present comprehensive disproofs of the Riemann hypothesis. Moreover, in 1994 the author discovered the arithmetic function Jn () that can replace Riemanns (s) function in view of its proved features: if Jn () = 0, then the function has innitely many prime solutions; and if Jn () = 0, then the function has nitely many prime solutions. By using the Jiang J2 () function we prove the twin prime theorem, Goldbachs theorem and the prime theorem of the form x2 + 1. Due to the importance of resolving the historical open nature of the Riemann hypothesis, comments by interested colleagues are here solicited.

AMS mathematics subject classication: Primary 11M26.

1. Introduction
In 1859 Riemann dened the zeta function[1] (s) =
p

(1 ps )1 =

1 , s n=1 n

(1)

where s = + ti, i = 1, and t are real, p ranges over all primes. (s) satises the functional equation [2]
(1s) s s 1s 2 ( )(s) = 2 ( )(1 s). 2 2

(2)

From (2) we have (ti) = 0. (3) Riemann conjectured that (s) has innitely many zeros in 0 1, called the critical strip. Riemann further made the remarkable conjecture that the zeros of (s) in the critical strip all lie on the central line = 1/2, a conjecture called the famous Riemann hypothesis (RH). It was stated by Hardy in 1914 that innitely many zeros lie on the line; A. Selberg stated in 1942 that a positive proportion at least of all the zeros lie on the line; Levinson stated in 1974 that more than one third of the zeros lie on the line; Conrey stated in 1989 that at least two fths of the zeros lie on the line. The use of the RH then lead to many mathematical problems, such as the generalized Riemann conjecture, Artins conjecture, Weils conjecture, Langlands program, quantum chaos, the hypothetical Riemann ow [3, 4], the zeta functions and L-functions of an algebraic variety and other studies. Similarly, it is possible to prove many theorems by using the RH. However, the RH remains a basically unproved conjecture to this day. In fact, Hilbert properly stated in 199 that the problem of proving or disproving the RH is one of the most important problems confronting 20th century mathematicians. In 2000 Griths and Graham pointed out that the RH is the rst challenging problem for the 21st century. The proof of the RH then become the millennium prize problem. In 1997 we studied the tables of the Riemann zeta function [5] and reached preliminary results indicating that the RH is false [6, 7, 8]. In 2

this paper we present a comprehensive disproof of the RH and show that the computation of all zeros of the (1/2 + ti) function done during the past 100 years is in error, as preliminarily indicated in Refs. [9, 10, 11]. Since the RH is false, all theorems and conjectures based to the same are also false.

2. Disproofs of Riemanns Hypothesis

Theorem 1. (s) has no zeros in the critical strip, that is (s) = 0, where 0 1. Proof 1. From (1) we have 1 = (s) where R=
p

(1
p

1 ) = Rei , ps 2 cos(t log p) 1 + 2 , p p p sin(t log p) . cos(t log p)

(4)

Rp , p ,
p

Rp =

(5) (6)

p = tan1

If = 0, from (5) we have Rp = 2 1 cos(t log p). If cos(t log p) = 1, we have Rp = 0 then R = 0. If > 0 from (5) we have Rp = 0. (s) = 0 if and only if Re (s) = 0 and Im (s) = 0, that is R = . From (5) we have that if cos(t log p) 0 then Rp > 1 and if cos(t log p) > 0 then Rp < 1. cos(t log p) is independent of the real part , but may well depend on primes p and imaginary part t. We write m+ (t) for the number of primes p satisfying cos(t log p) > 0, m (t) for the number of primes p satisfying cos(t log p) 0. For cos(t log p) > 0, we have 1 > Rp (1 + ti) > Rp (0.5 + ti). (7)

If m+ (t1 ) is much greater than m (t1 ) such that R(0.5 + t1 i) = min. From (5), (6) and (7) we have for given t1 minR(1 + t1 i) > minR(1 + t1 i) > minR(0.5 + t1 i) > minR(2 + t1 i) 0, 3

(8) (1 + t1 i) = (1 + t1 i) = (0.5 + t1 i) = (2 + t1 i) = const where 1 > 1 and 0 2 < 0.5. 1 Since | (s) |= R from (8) we have max | (1 + t1 i) |< max | (1 + t1 i) | < max | (0.5 + t1 i) |< max | (2 + t1 i) | . For cos(t log p) < 0 we have 1 < Rp (0.5 + ti) < Rp (0.4 + ti) < Rp (0.3 + ti). (11) (10) (9)

If m (t1 ) is much greater than m+ (t1 ) such that R(0.5 + t1 i) = max. From (5), (6) and (11) we have for given t1 maxR(1 + t1 i) < maxR(0.5 + t1 i) < maxR(0.4 + t1 i), < maxR(0.3 + t1 i) < maxR(2 + t1 i) = , (1 + t1 i) = (0.5 + t1 i) = (0.4 + t1 i) = (0.3 + t1 i) = (2 + t1 i) = const, (13) where 1 > 0.5 and 0 2 < 0.3. 1 Since | (s) |= R from (12) we have min | (1 + t1 i) |> min | (0.5 + t1 i) |> min | (0.4 + t1 i) |> min | (0.3 + t1 i) |> min | (2 + t1 i) |= 0. Proof 2. We dene the beta function (s) =
p

(12)

(14)

(1 + ps )1 =

(n) , s n=1 n

(15)

where (1) = 1, (n) = (1)a1 ++ak if n = pa1 pak , t = 0. 1 k

From (15) we have 1 = (s) where R=

p

(1 +
p

1 ) = Rei , s p 2 cos(t log p) 1 + 2 , p p

(16)

Rp , Rp =

1+

(17) (18)

=
p

sin(t log p) p , p = tan1 . p + cos(t log p)

For cos(t log p) < 0, we have 1 > Rp (1 + ti) > Rp (0.5 + ti). (19)

If m (t1 ) is much greater than m+ (t1 ) such that R(0.5 + t1 i) = min. From (17), (18) and (19) we have for given t1 minR(1 + t1 i) > minR(1 + t1 i) > minR(0.5 + t1 i) > minR(2 + t1 i) 0, (1 + t1 i) = (1 + t1 i) = (0.5 + t1 i) = (2 + t1 i) = const, where 1 > and 0 2 < 0.5. 1 Since | (s) |= R from (20) we have max | (1 + t1 i) |< max | (1 + t1 i) | < max | (0.5 + t1 i) |< max | (2 + t1 i) | . For cos(t log p) > 0 we have 1 < Rp (0.5 + ti) < Rp (0.4 + ti) < Rp (0.3 + ti). (23) (22) (20) (21)

If m+ (t1 ) is much greater than m (t1 ) such that that R(0.5 +t1 i) = max. From (17), (18) and (23) we have for given t1 maxR(1 + t1 i) < maxR(0.5 + t1 i) < maxR(0.4 + t1 i) < maxR(0.3 + t1 i) < maxR(2 + t1 i) = , 5 (24)

 (1 + t1 i) = (0.5 + t1 i) = (0.4 + t1 i) = (0.3 + t1 i) = (2 + t1 i) = const, (25) where 1 > 0.5 and 0 2 < 0.3. 1 Since | (s) |= R from (24) we have min | (1 + t1 i) |> min | (0.5 + t1 i) | > min | (0.4 + t1 i) |> min | (0.3 + t1 i) |> min | (2 + t1 i) |= 0. From (1) and (15) we have (2s) = (s)(s). (27) (26)

In 1896 J. Hadamard and de la Vallee Poussin proved independently | (1 + ti) |= 0. From (27) we have 1 1 | (1 + 2ti) |=| ( + ti) || ( + ti) |= 0. 2 2 From (28) we have 1 | ( + ti) |= 0 2 and 1 | ( + ti) |= 0. 2 (29) (28)

(s) and (s) are the dual functions. From (22) we have 1 | ( + ti) |= . 2 Therefore we have In the same way we have 1 1 1 | ( + 2ti) |=| ( + ti) || ( + ti) |= 0. 2 4 4 (32) 1 | ( + ti) |= 0. 2 (30)

(31)

From (32) we have 1 | ( + ti) |= 0 4 In the same way we have | ( As n we have | (ti) |= 0. Proof 3. For > 1 we have

and

1 | ( + ti) |= 0. 4

(33)

1 + ti) |= 0. 2n

(34)

(35)

log (s) =
p m=1

m1 pm exp(itm log p).

(36)

If (s) had a zero at 1 + ti, then log | ( + ti) | would tend to as 2 tends to 1 from the right. From (36) we have 2

log | (s) |=
p m=1

m1 pm cos(tm log p),

(37)

with t replaced by 0, t, 2t, , Ht, it gives

H1 j=0

2H j

log | ( + (H j)ti) | +

1 2

2H H (38)

log () =
p m=1

m1 pm A 0,

where A =
H1 j=0

2H j

cos ((H j)tm log p) +

1 2

2H H

(39)

= 2H1 [1 + cos(tm log p)]H 0, H is an even number.

From (38) we have

1 2

(())

2H H

H1

|( + (H j)ti)|
j=0

2H j

1.

(40)

Since | ( 1 + eti) |= [5], where e = 1, 2, , H, from (40) we have 2 1 | ( 2 + eti) |= 0 for suciently large even number H. Min | ( 1 + ti) 0 but = 0. The computation of all zeros of ( 1 + ti) 2 2 is error, which satises the the error RH. From (39) we have cos 2 + 4 cos + 3 = 2(1 + cos )2 , cos 4 + 8 cos 3 + 28 cos 2 + 56 cos + 35 = 8(1 + cos )4 , cos 5 + 12 cos 5 + 66 cos 4 + 220 cos 3 +495 cos 2 + 792 cos + 462 = 32(1 + cos )6 .

3. The Arithmetic Function Jn () Replacing Riemanns Hypothesis

In view of the preceding results, the RH has no value for the study of prime distributions. In 1994 the author discover the arithmetic function Jn () [12, 13] that is able to take the place of Riemanns zeta-functions and L-functions because of the following properties: Jn () = 0, then the function has innitely many prime solutions; and if Jn () = 0, then the function has nitely many prime solutions. By using Jiangs Jn () function we have proved numerous theorems including the twin prime theorem, Goldbachs theorem, the prime ktuples theorem, Santillis theory for a prime table, the theorem of nite Fermat primes, the theorem of nite Mersenne primes, the theorem of nite repunit primes, there are innitely many triples of n, n+1 and n+2 that each is the product of k distinct primes, there are innitely many Carmichael numbers which are product of exactly ve primes, there there are nitely many Carmichael numbers which are product of exactly six primes in the prime distributions [14]. We give some theorems below Theorem 2. Twin prime theorem: p1 = p + 2. 8

We have the arithmetic function J2 () =

3ppi

(p 2) = 0.

Since J2 () = 0, there are innitely many primes p such that p1 is a prime. Theorem 3. Goldbach theorem: N = p1 + p2 . We have the arithmetic function J2 () =
3ppi

(p 2)
p|N

p1 = 0. p2

Since J2 () = 0, every even number N greater than 4 is the sum of two odd primes. Theorem 4. p1 = (p + 1)2 + 1. We have the arithmetic function J2 () =
3ppi

(p 2 (1)

p1 2

) = 0.

Since J2 () = 0, there are innitely many primes p such that p1 is a prime. Theorem 5. p1 = p2 2. We have 2 J2 () = (p 2 ( )) = 0. p 3ppi Since J2 () = 0, there are innitely many primes p such that p1 is a prime. Theorem 6. p1 = p + 4 and p2 = 4p + 1. We have J2 () = 3 (p 3) = 0.
7ppi

Since J2 () = 0, there are innitely many primes p such that p1 and p2 are primes. 9

Theorem 7. p1 = (p + 1)2 + 1 and p2 = (p + 1)2 + 3. We have p1 3 J2 () = (p 3 (1) 2 ( )) = 0. p 5ppi Since J2 () = 0, there are innitely many primes p such that p1 and p2 are primes. Theorem 8. The prime 13-tuples theorem: p+b: b = 0, 4, 6, 10, 16, 18, 24, 28, 34, 40, 46, 48, 90. Since J2 (13) = 0, there are no prime 13-tuples if p = 13. Theorem 9. The prime 14-tuples theorem: p+b: b = 0, 2, 6, 8, 12, 18, 20, 26, 30, 32, 36, 42, 48, 50. We have J2 () = 300 (p 14) = 0.
29ppi

Since J2 () = 0, there are innitely many prime 14-tuples. Theorem 10. p1 = 6m + 1, p2 = 12m + 1, p3 = 18m + 1, p4 = 36m + 1, p5 = 72m + 1. We have J2 () = 12 (p 6) = 0.
13ppi

Since J2 () = 0, there are innitely many integers m such that p1 , p2 , p3 , p4 and p5 are primes. n = p1 p2 p3 p4 p5 is the Carmichael numbers. Theorem 11. p3 = p1 + p2 + p1 p2 . We have J3 () = (p2 3p + 3) = 0.
3ppi

Since J3 () = 0, there are innitely many primes p1 and p2 such that p3 is a prime. Theorem 12. p3 = (p1 + 1)5 + p2 . We have (p2 3p + 3) = 0. J3 () =
3ppi

10

Since J3 () = 0, there are innitely many primes p1 and p2 such that p3 is a prime. Theorem 13. p4 = p1 (p2 + p3 ) + p2 p3 . We have (p 1)4 1 J4 () = + 1 = 0. p 3ppi Since J4 () = 0, there are innitely many primes p1 , p2 and p3 such that p4 is a prime. Theorem 14. Each of n, n + 1 and n + 2 is the product of k distinct primes. Suppose that each of m1 , m2 = m1 +1 and m3 = m1 +2 is the product of k 1 distinct primes. We dene p1 = 2m2 m3 x + 1, p2 = 2m1 m3 x + 1, p3 = 2m1 m2 x + 1. (41)

We have the arithmetic function J2 () =

3ppi

(p 4 (p)) = 0,

(42)

where (p) = 2 if p | m1 m2 m3 ; (p) = 0 otherwise. Since J2 () = 0, there exist innitely many integers x such that p1 , p2 and p3 are primes. From (41) we have n = m1 p1 = 2m1 m2 m3 x+ m1 , n +1 = m1 p1 +1 = 2m1 m2 m3 x + m1 + 1 = m2 (2m1 m3 x + 1) = m2 p2 , n + 2 = m1 p1 + 2 = 2m1 m2 m3 x + m1 + 2 = m3 (2m1 m2 x + 1) = m3 p3 . If p1 , p2 and p3 are primes, then each of n, n + 1 and n + 2 is the product of k distinct primes. For example, n = 1727913 = 3 11 52361, n + 1 = 1727914 = 2 17 50821, n + 2 = 1727915 = 5 7 49369. Jn () is a generalization of Eulers proof for the existence of innitely many primes. It has a wide application in various elds.

Acknowledgements
The Author would like to express his deepest appreciation to Professors R. M. Santilli, G. Weiss, D. Zuckerman, Ke-xi Liu, Mao-xian Zuo, Zhongdian Wang, Chang-pei Wang, and Xin-ping Tong for their helps and supports. 11

References
[1] B. Riemann, Uber die Anzahl der Primzahlen under einer gegebener Grsse, Monatsber. Akad. Berlin. 671-680 (1859). o [2] H. Davenport, Multiplicative Number Theory, Springer Verlag (1980). [3] N. Katz and P. Sarnak, Zeroes of zeta functions and symmetry, Bull. AMS, 36, 1-26 (1999). [4] A. Connes,Trace formula in noncommutative geometry and the zeros of the Riemann zeta function, Sel. Math., New Ser. 5, 29106(1999). [5] C. B. Haslgrove,Tables of the Riemann zeta function, Roy. Soc. Math. Tables, Vol.6, Cambridge Univ. Press, London and New York (1960). [6] Chun-xuan, Jiang, The study of the Riemann Zeta function, Unpublished Oct. (1997). [7] Chun-xuan,Jiang,Foundations of Santillis isonumber theory II, Algebras Groups and Geometries, 15, 509-544 (1998). [8] Chun-xuan, Jiang,Foundations of Santillis isonumber theory. In: Foundamental open problems in sciences at the end of the millennium, T. Gill, K. Liu and E. Trell (Eds) Hadronic Press, USA, 105139 (1999). [9] Richard P. Brent,On the zeros of the Riemann zeta function in the critical strip, Math. Comp., 33, 1361-1372(1979). [10] J. van de Lune, H. J. J. te Riele and D. T. Winter, On the zeros of the Riemann zeta function in the critical strip, Math. Comp. 46, 667-681 (1986). [11] A. M. Odlyzko, 1020 -th zero of the Riemann zeta function and its neighbors, Preprint (1989).

12

[12] Chun-xuan, Jiang, On the Yu-Goldbach prime theorem, Guangxi Sciences (in Chinese) 3. 9-12 (1996). [13] Chun-xuan, Jiang,Foundations of Santillis isonumber theory I, [14] C.X.Jiang. Foundation of Santillis isonumber theory with applications to new cryptograms, Fermats theorem and Goldbachs conjecture. International Academic Press, 2002 (also available in dpf format at http://www.i-b-r.org/docs/magneh.pdf

13