Mathematics Letters
2019; 5(2): 13-22
http://www.sciencepublishinggroup.com/j/ml
doi: 10.11648/j.ml.20190502.11
ISSN: 2575-503X (Print); ISSN: 2575-5056 (Online)
The Inconsistency Problem of Riemann Zeta Function
Equation
Mei Xiaochun
Department of Theoretic Physics and Pure Mathematics, Institute of Innovative Physics in Fuzhou, Fuzhou, China
Email address:
To cite this article:
Mei Xiaochun. The Inconsistency Problem of Riemann Zeta Function Equation. Mathematics Letters. Vol. 5, No. 2, 2019, pp. 13-22.
doi: 10.11648/j.ml.20190502.11
Received: July 8, 2019; Accepted: July 31, 2019; Published: August 13, 2019
Abstract: Four basic problems in Riemann’s original paper are found. The Riemann hypothesis becomes meaningless. 1. It
is proved that on the real axis of complex plane, the Riemann Zeta function equation holds only at point Re(s)=1/2 (s = a+ib).
However, at this point, the Zeta function is infinite, rather than zero. At other points of real axis, the two sides of Zeta function
equation are contradictory. When one side is finite, another side may be infinite. 2. An integral item around the original point of
coordinate system was neglected when Riemann deduced the integral form of Zeta function. The item was convergent when
Re(s) > 1 but divergent when Re(s) < 1. The integral form of Zeta function does not change the divergence of its series form.
Two reasons to cause inconsistency and infinite are analyzed. 3. When the integral form of Zeta function was deduced, a
summation formula was used. The applicable condition of this formula is x > 0. At point x = 0, the formula is meaningless.
However, the lower limit of Zeta function integral is x = 0, so the formula can not be used. 4. A formula of Jacobi function was
used to prove the symmetry of Zeta function equation. The applicable condition of this formula was also x > 0. However, the
lower limit of integral in the deduction was x=0. So this formula can not be used too. The zero calculation of Riemann Zeta
function is discussed at last. It is pointed out that because approximate methods are used, they are not the real zeros of strict
Riemann Zeta function.
Keywords: Riemann Hypothesis, Riemann Zeta Function, Zeta Function Equation, Jacobi’s Function, Residue Theorem,
Cauchy-Riemann Equation
1. Introduction
Riemann Zeta function is an important one in modern
mathematics. Riemann proposed the Riemann hypothesis
about the zeros of Zeta function in 1859, but it can not be
proved up to now. In this is paper, the Riemann’s original
deduction is examined carefully and four basic mistakes are
revealed. Due to theses mistakes, the integral form of
Riemann Zeta function and its function equation can not hold.
Riemann hypothesis becomes meaningless.
Riemann Zeta function has two forms. One is the form of
series summation and another is the form of integral. The
series summation form is original and can be written as
∞
ζ ( s) = ∑ n − s = 1 +
n =1
1 1 1
+ + + ⋅⋅⋅
2s 3s 4s
(1)
Here s = a + ib ∈ C is a complex number, a ∈ R and
b ∈ R are real numbers. If s is a real number with b = 0
and a > 1 , the series is convergent, while b = 0 and a ≤ 1 ,
the series is divergent. For example
1 1 1
+ + + ⋅⋅⋅ → ∞
2 3 4
1
1 1
π2
ζ ( 2) = 1 + 2 + 2 + 3 + ⋅ ⋅ ⋅ =
2
3
4
6
ζ (1) = 1 +
ζ (−2) = 1 + 22 + 32 + 4 2 + ⋅ ⋅ ⋅ → ∞
(2)
Here ζ (1) is the harmonic series and ζ (2) is the Euler
formula. If s is a complex number, according to general
understanding, when Re( s ) > 1 , (1) is convergent. While
Re( s ) ≤ 1 , (1) is divergent and meaningless.
In order to make (1) meaningful for the situation Re( s ) < 1 ,
Riemann transformed it into the integral form by means of the
Gama function and obtained [1, 2]
14
Mei Xiaochun:
∞
x s −1
∞ −s
dx
∑ n Γ ( s ) = ∫ x
e −1
n =1
0
The Inconsistency Problem of Riemann Zeta Function Equation
Re( s) > 1
(3)
Here x ∈ R is still a real number. In order to calculate (3),
Riemann extended it to the integral of complex plane with
( z ∈ C ).
∞
∑n
n =1
−s
=
Γ(1 − s) (− z ) s -1
dz
2π i K∫ e z − 1
Re( s ) ≠ 1
(4)
By means of the method of residues and based on (4),
Riemann obtained following Zeta function equation
∞
∑n
n =1
−s
sπ ∞
= 2(2π ) s −1 Γ (1 − s ) sin ∑ n − (1− s )
2 n =1
(5)
By using the definition of (1), (5) was written as in general
sπ
ζ (1 − s )
2
ζ ( s ) = 2(2π ) s −1 Γ(1 − s ) sin
(6)
According to common understanding, after the complex
continuation, (6) was tenable on whole complex plane except
at the point Re( s )=1 [2]. Then Riemann introduced
following transformation [1, 3]
1
2
s
2
ξ ( s) = s( s − 1)π − s / 2 Γ ζ ( s)
(7)
and proved the existence of symmetry
ξ ( s ) = ξ (1 − s )
(8)
Because ζ ( s ) and ξ (s ) were considered to have the
same zeros, the zero’s calculation of Zeta function equation
was based on (7) actually.
According to (6), when s = ±2 k ( k = 0, 1, 2, 3,⋅ ⋅ ⋅ ),
ζ ( s ) = 0 . For the situations with s = −2k , the zeros were called
as trivial ones [3, 4]. The situations with s = +2k were not
discussed in general.
Because the distribution of prime numbers is related to the
zeros of Zeta function, the judgment of zeros becomes an
impotent problem. Riemann hypothesis declared that all zeros
were distributed on the straight line Re( s )=1 / 2 of complex
plane. But the hypothesis can not be proved up to now [5, 6].
The original paper of Riemann is analyzed in this paper. Four
basic mistakes are found which makes Riemann hypothesis
meaningless. They are:
1. There exists inconsistency in the Riemann Zeta function
equation. After complex continuation, (5) and (6) were
considered to be tenable on whole complex plane except the
point Re( s )=1 . However, this is not true.
So-called continuation of function indicates to re-define the
function in a new domain where the original function has no
definition so that the function becomes meaningful. Therefore,
there is a basic principle for the continuation of function. At
each point of original domain, re-defined function should have
the same value with original function, though re-defined
function may have different form in new domain. Otherwise, it
can not be regarded as the continuation of original function.
According to this principle, on the domain of Re( s ) > 1 ,
the left side of (6) should be equal to (1). The right sides of
(5) and (6) have definition except at point Re( s ) = 1 , i.e.,
a ≠ 1 but b can be arbitrary. The domain of left sides of (5)
and (6) is Re( s ) > 1 , i.e., a > 1 but b can be arbitrary.
Therefore, to take a = 3.5 and b = 0 , (5) and (6) should
be effective. However, the result of practical calculation is
that the left side of (5) is a finite value, but the right side of (5)
is infinite. Therefore, two sides of Zeta function equation are
incompatible. (5) and (6) do not hold.
In fact, it is proved that on the real axis of complex plane,
(5) only hold at point a = 1 / 2 . But in this case, two sides
of Zeta function equation are equal to infinite, rather than
zero. That is to say, on the real axis, the point a = 1 / 2 is
the infinite point of Riemann Zeta function equation, rather
than zero.
When a is a non-integer with a > 1 , if the left side of
(5) is convergent, its right side is infinite, vice versa. When
0 < a < 1 , the two sides of (5) are infinite and become
meaningless. So the Riemann Zeta function equation is
contradictory.
2. An integral item around the original point of coordinate
system was neglected in Riemann’s original paper. The item
was convergent when Re( s ) > 1 , but infinite when Re( s ) ≤ 1 .
That is to say, the complex continuation of Riemann Zeta
function has not changed its divergence of series summation
form.
Two reasons to cause the inconsistency and divergence are
analyzed. They are
3. A summation formula was used in Riemann’s deduction.
The applicable condition of this formula is x > 0 . At point
x = 0 , the formula is meaningless. However, the lower limit
of Zeta function integral is x = 0 . So this formula can not be
used.
4. The formula θ (x ) = xθ (1 / x) of Jacobi function was used
to prove the symmetry of Zeta function. The applicable
condition of this formula is also x > 0 [7]. However, the
lower limit of integral involved in the deduction is x = 0 .
Therefore, the formula can not be used too, the symmetry
ξ ( s) = ξ (1 − s) does not hold.
The zero calculations of Riemann Zeta function was discussed
at last. It was pointed out that because approximate method was
used, the Cauchy-Riemann equation was not satisfied. Thought
great numbers of zeros were founded on the straight line
Re( s ) = 1 / 2 of complex plane, they were not real zeros of strict
Zeta function.
2. The Deduction of Riemann Zeta
Function Equation
2.1. The Deduction of Integral Form of Riemann Zeta
Function
Based on the form of series summation and by using Gama
Mathematics Letters 2019; 5(2): 13-22
function, Riemann deduced the integral form of ζ (s)
function [1, 2]. The definition of Gama function is
∞
Re( s ) > 0
Γ ( s) = ∫ e − x x s −1dx
(9)
15
According to Figure 1, (14) contained three items
δ
∞
( − x) s−1
( − z ) s −1
( − x) s−1
dx
dz + ∫ x
dx + ∫ z
x
e −1
e −1
∞
Ω
δ e −1
I ( s) = ∫
(15)
0
Here x ∈ R is a real number and s = a + ib ∈ C is a
complex number with Re( s ) > 0 . Let x → nx , (9) becomes
∞
∫
∫
0
0
s
− nx s −1
Γ ( s ) = e −( nx ) ( nx )s −1 d( nx ) = n e x dx
(10)
∞ ∞
∞ −s
or
n
Γ
(
s
)
=
e − nx x s −1dx
(11)
n
n
1
1
=
=
0
Then, Riemann used the following summation formula of
series (The original paper of Riemann had not provided this
formula but used it actually.) [8].
∑
∑
x s −1
dx
ex − 1
0
I ( s ) =(eiπs − e −iπs)∫
∞
∞
∞
Riemann’s paper provided following result directly without
concrete calculation [1, 2]
∫∑
It indicated that Riemann assumed that the medium item on
the right side of (15) was zero with
(− z ) s −1
dz = 0
ez −1
Ω
∫
eiπs − e −iπs
= sin(πs )
2i
n =1
(
)
∞
I ( s ) = i 2 sin(πs)Γ( s ) ∑ n − s
n =1
(12)
Re( s ) > 1
Γ ( s )Γ (1 − s) =
(13)
In order to calculate (13), Riemann extended the integral of
real number x into complex number z = x + iy ∈ C by
defining the function I (z ) with
(− z ) s −1
dz
ez − 1
L
Re( s) ≠ 1
(14)
The definition domain of function was extended to whole
complex plane except the point s = 1 . As shown in Figure 1,
the path of integral started from x → ∞ to x = δ along the
straight line B under x axis. Here δ was a small quantity.
Then path was along the circle Ω with radius
x2 + y2 = δ
around the original point of the coordinate system. At last, the
path was from x = δ to x → ∞ along the straight line A
above x axis.
π
sin(πs )
(20)
and considering (14), (19) can be written as
∞
∑ n− s =
n =1
I (s) = ∫
(19)
By using the complementary formula of Gama function
By substituting (12) in (11), Riemann obtained
∞
(18)
and substituting (16) and (18) in (13), the result was
e− x
1
=
= x
−x
1− e
e −1
x s −1
∞ −s
dx
∑ n Γ( s ) = ∫ x
e −1
n =1
0
(17)
By using the Euler’s formula
e − nx = e − x + e −2 x + e −3 x + ⋅⋅⋅
= e − x 1 + e− x + e−2 x + e−3 x + ⋅⋅⋅
(16)
I (s)
Γ(1 − s) (− z ) s −1
dz
=
i 2 sin(πs)Γ( s )
2πi ∫L e z − 1
(21)
By using the definition (1), (21) is generally written as
ζ ( s) =
I (s)
Γ(1 − s ) (− z ) s −1
dz
=
2πi ∫L e z − 1
i 2 sin(πs )Γ( s )
(22)
In this way, Riemann deduced the integral form of Zeta
function. However, according to (21) we should note that the
Zeta function ζ (s ) shown in (22) is in the form of (1). This
is very impotent for us to discuss the consistency problem of
Zeta function equation.
2.2. The Deduction of Riemann Zeta Function Equation
In order to calculate the right side of (22), Riemann used the
Residue theorem by considering the contour shown in Figure
2. The integrand function has infinite poles on the points
z = ±i 2nπ ( n = 1, 2, 3,⋅ ⋅ ⋅ ), but only three are drawn. The
path of integral in Figure 1 is enlarged into a closed curve C ,
composed of a big circle C0 and the paths shown in Figure 1.
Figure 1. The integral path of Riemann Zeta function.
16
Mei Xiaochun:
The Inconsistency Problem of Riemann Zeta Function Equation
According to the Residue theorem, when the radius of C
tends to infinite, the integral along big circle C0 is zero.
The calculation formula of Residue integral is [9]
∫
N
f ( z ) dz = i 2π ∑ a−( k1)
(23)
k =1
C
(− z ) s −1
(− z ) s −1
a−( k1) = z
= (±i 2nπ ) s −1 (24)
=
z
′
(
e
1
)
e
−
z = ± i 2 nπ
z = ± i 2 nπ
(29) is just the Riemann Zeta function equation, or called as
the algebraic relation of Zeta function. Based on it, Riemann
proposed the Riemann hypothesis that all zeros of Zeta
functions were located on the straight line of s = 1 / 2 + ib on
the complex plane.
At present, the common understanding is that through the
complex continuation, (29) becomes the new definition of
Riemann Zeta function. The function ζ (s ) on the left hand
side of (29) is defined by the right side, which is not the
original form defined in (1) again.
However, this is not the case. When we deduce (29), (1) and
(28) must be used finally. The prototype of (29) is (27). (29) is
only a simplified notation of (27). So, we should discuss the
consistency problem of the Riemann Zeta function equation
based on (27).
3. The Proof That Zeta Function
Equation Does not Hold on the Real
Axis of Complex Plane
Figure 2. The residual integral of Riemann Zeta function.
By considering
(i )s −1 + (−i) s −1 = 1 (eiπs / 2 − e-iπs / 2 ) = 2 sin πs
i
2
(25)
the calculation result of (23) is
N
∑a
k =1
(k )
−1
1
=
i 2π
∞
∑n
∞
(− z ) s−1
s −1
s −1
(i 2πn ) + (− i 2πn )
∫C e z − 1 dz = ∑
n =1
= 2(2π )
s −1
sin
πs
2
∞
∑n
− (1− s )
∑n
−s
= 2(2π )
s −1
(26)
n =1
Γ (1 − s ) sin
n =1
−a
πs
2
∑n
− (1− s )
∑n
(27)
aπ
2
∞
∑n
a −1
(30)
n =1
However, it can be proved that the two sides of (30) are
contradictory, so (30) can not hold.
I) When a is an integer, for example, to take a = −2 , the
left side of (30) is infinite with
∞
∞
=(
2 2π)a −1Γ (1 − a ) sin
n =1
Substituting (26) in (21), we have
∞
According to the theory of complex variable function, the
complex continuation indicates that a real function f (x) is
extended to the field of complex number with
f ( x) → G ( z ) = G ( x + iy ) . Thought the definition domain of
function is changed, on the real axis with y = 0 , the complex
continuation function should be the same as the original
function with G ( x) = f ( x) . If they are not the same on the real
axis, the complex continuation of function can not be
considered correct.
Therefore, (27) should hold on the real axis of complex
plane with b = 0 and a ≠ 1 . Because the definitions of Zeta
function ζ (s ) and ζ (1 − s ) shown in (1) and (28) are certain,
on the real axis of complex plane, (27) becomes
2
= 1 + 2 2 + 32 + 4 2 + ⋅ ⋅ ⋅ → ∞
(31)
n =1
n =1
But the right side of (30) is zero with
In light of the definition of (1), we have
∞
∑n
− (1− s )
∞
= ζ (1 − s )
(28)
(
2 2π)− 3Γ(3) sin( −π )∑ n − 3
n =1
n =1
1 1 1
= 0 × 1 + 3 + 3 + 3 + ⋅⋅⋅ = 0
3 4
2
So (27) can be written as
ζ ( s ) = 2(2π ) s −1 Γ(1 − s ) sin
πs
2
ζ (1 − s )
(29)
(32)
To take a = 4 , the left side of (30) is a limited and certain
value with
Mathematics Letters 2019; 5(2): 13-22
∞
∑n
−4
= 1+
n =1
1
1
1
+ 4 + 4 + ⋅⋅⋅ > 1
4
2
3
4
(33)
17
2 2π)2.5 Γ ( −2.5 )sin
=(
∞
∑n
n =1
(
)
(34)
The Zeta functions on the two sides of (30) are
incompatible.
II) When a is a non-integer, to take a = 1 / 2 , called
as the non-trivial zero of Riemann hypothesis, the left side of
(30) is
∞
∑n
−1 / 2
= 1+
n =1
1
1
1
+
+
+ ⋅⋅⋅
2
3
4
1 1 1
> 1 + + + + ⋅⋅⋅ → ∞
2 3 4
∞
∑n
(35)
∑n
n =1
(38)
∑n
−a
= 1 + 23.5 + 33.5 + 43.5 + ⋅⋅⋅ → ∞
But the right side of (30) is convergent. Due to
Γ(4.5) = 105 π / 16 and sin( −3.5π / 2) = 0.7091 , we have
aπ
2
∞
∑n
a −1
n =1
∞
−3.5π
-4.5
Γ ( 4.5 )sin
2 2π)
n −4.5
=(
2 n =1
∑
-4.5
=(2π)
π × 9.3069
n =1
(36)
(39)
n =1
1
1
1
× 1 + 4.5 + 4.5 + 4.5 + ⋅⋅⋅
3
4
2
The two sides of (30) are completely identical, but they are
infinite, rather than zero. So a = 1 / 2 is not the zero of
Riemann Zeta function equation on the real axis of complex
plane.
III) If a > 1 is a non-integer, for example, to take a = 3.5 ,
the left side of (30) is convergent with
∞
∞
−1/ 2
1
1
1
= 1 × 1 +
+
+
+ ⋅⋅⋅ → ∞
2
3
4
n =1
IV) If a < 0 is a non-integer, for example, to take
a = −3.5 , the situation is conversed. The left side of (30) is
divergent with
a −1
Γ ( 1 − a ) sin
(
2 2π)
Due to Γ(1 / 2) = π and sinπ / 4 = 2 / 2 , the right side
of (30) is
2
π
Γ ( 1 / 2 ) sin
4
2π
2.5
× 1 + 2 2.5 + 32.5 + 4 2.5 + ⋅⋅⋅)→ ∞
(
3
= 0 × 1 + 23 + 33 + 43 + ⋅⋅⋅ = 0 × ∞
∞
∑n
2.5
=(2π)
π × 0.7564
But the right side of (30) is uncertain with
3
(
2 2π)
Γ ( −3 ) sin( 2π )
3.5π
2
(40)
V) If 0 < a < 1 ,for example, to take a = 0.2 ,the left side
of (30) is divergent with
∞
∑n
n =1
−a
= 1+
1
1
1
+
+
+ ⋅⋅⋅
20.2 30.2 40.2
1 1 1
> 1 + + + + ⋅⋅⋅ → ∞
2 3 4
(41)
The right side of (30) is also divergent with
−a
= 1+
1
1
1
+
+
+ ⋅⋅⋅
23.5 33.5 43.5
1 1 1
< 1 + + + + ⋅⋅⋅
2 3 4
a −1
(
2 2π)
Γ ( 1 − a ) sin
(37)
But the right side of (30) is divergent. According the formula
of negative continuation of Gama function with
Γ( −2.5) = −8 π / 15 , as well as sin3.5π / 2 = −0.7091 , we
have
aπ
a −1
Γ ( 1 − a ) sin
(
2 2π)
2
∞
∑n
n =1
a −1
aπ
2
∞
∑n
a −1
n =1
=(
2 2π)−0.8 Γ ( 0.8 )sin( 0.1π ) ×
∞
∑n
−0.8
n =1
-4.5
=(2π)
× 0.7188
1
1
1
× 1 + 0.8 + 0.8 + 0.8 + ⋅⋅⋅ → ∞
3
4
2
(42)
Here Γ(0.8) = Γ(1.8) / 0.8 = 1.1643 , sin(0.1π ) = 0.3087 . In
this case, (30) is meaningless. Different from (35) and (36),
we can not ensure that the two sides of equal sign are certainly
equal.
18
Mei Xiaochun:
The Inconsistency Problem of Riemann Zeta Function Equation
In short, when a is an non-integer with a > 1 , the
Riemann Zeta function equation is inconsistent on the real
axis of complex plane. If one side of equation is infinite,
another side is limited. When 0 < a < 1 , the two sides of
equation are infinite. We can not ensure that two sides are
certainly equal. Only at the point a = 1 / 2 , the two sides of
equation are equal strictly. But they are infinite, rather than
zeros. That is to say, s = a = 1 / 2 is not the zero of Riemann
Zeta function equation, it is the point of infinite. This result is
different from the Riemann hypothesis.
=
=
δ a −1e −π b eibln δ eiπ ( a −1 )
( a − 1 )2 + b2
δ a −1e −π b
(
a − 1 )cos b ln δ + π ( a − 1 )
( a −1) + b
2
a − 1 ) sin b ln δ + π ( a − 1 )
+i (
(
)
− b cos ( b ln δ + π ( a − 1 ) )
4.1. Formula (17) Is Not Equal to Zero When Re( s ) < 1
the small circle Ω , we have dz = iδeiθ dθ , dz / z = idθ .
When δ << 1 , we have
2π
1
1
=
e − 1 1 + z + z 2 / 2!+ z 3 / 3!+ ⋅⋅⋅ − 1
k1 ∫ ( − z )s −1 dz = k1 ∫ ( −δ eiθ )s −1δ deiθ
Ω
1
1
=
= (1 + k1 z + k2 z 2 + ⋅⋅⋅)
2
z(1 + z / 2!+ z / 3!+ ⋅⋅⋅) z
∫
Ω
( −z )
dz =
ez −1
∫
Ω
( −z )
z
( 1 + k1 z + k2 z 2 + ⋅⋅⋅ )dz
(44)
Let’s only calculate the first item with
∫
Ω
( − z )s −1
dz =
z
= ( −δ )s −1
2π
∫
2π
∫
= k1 ( eiπ )s −1 δ s ∫ eisθ d( iθ ) =
0
0
=
( −δ )s −1 i 2π s −i 2π
−1
e e
s −1
(
)
=
2π
0
( −δ )s −1 i 2π s
− 1) (45)
(e
s −1
Substituting s = a + ib in (45) and considering
e i 2πs = e −2πb ei 2πa = e −2πb (cos 2πa + i sin 2πa )
and
k1eiπ ( s-1 )δ s i 2π s
(e − 1)
( a − ib )~k1′δ a
a 2 + b2
2π
0
(47)
a > 0 , when δ → 0 , δ a → 0 , (47) is convergent. If
a < 0 , when δ → 0 , δ a → ∞ , (47) is divergent.
If
( − z ) s −1
∫ e z − 1 dz = Q(δ )
Ω
0
( −δ )s −1 i( s −1 )θ
e
s −1
=
k1eiπ ( s-1 )δ s isθ
e
s
So, the last result of (44) should be
( −δ eiθ )s −1 idθ
ei( s −1 )θ d( iθ ) =
0
2π
(43)
Here ki are the developing parameters. Substituting (43)
in the left side of (17), we obtain
s −1
(46)
When δ → 0 we have ln δ → −∞ , sin(b ln δ ) and
cos(b ln δ ) are uncertain. If a > 1 , when δ → 0 , δ a −1 → 0 ,
so (46) is convergent. If a < 1 , when δ → 0 , δ a −1 → ∞ . In
this case, (46) is infinite and (44) is divergent too.
The calculation result of second item in (44) is
z
s −1
2
+ b sin ( b ln δ + π ( a − 1 ) )
4. The Non-Negligible Item in the
Integral Form of Zeta Function
We now prove that the integral (17) is divergent rather than
zero when Re( s ) < 1 . Let z = δeiθ describe the path around
( a − 1 − ib )
δ ib = eib ln δ , we get
( −δ )s −1 ( δ eiπ )a −1+ib δ a −1δ ib e −π b eiπ ( a −1 )
=
=
s −1
a − 1 + ib
a − 1 + ib
(48)
If a < 1 , when δ → 0 , we have Q(δ ) → ∞ . Similar to the
summation form of series shown in (1), the integral form of
Riemann Zeta function is still divergent when Re( s ) = a < 1 .
Therefore, it is meaningless to discuss the zeros of integral
form of Riemann Zeta function at point a = 1 / 2 < 1 .
Some authors also estimated and make approximate
calculations on the situation a > 1 and came to the same
conclusion that (17) could be ignored, but unfortunately no
further calculations were carried out [2,10]. In fact, if they also
did in-depth calculations for the situation a < 1 , they would
have reached the same conclusion that the item can not be
neglected.
4.2. The Numerical Calculation of Divergent Item
In order to have more clear understanding of divergence,
let’s take numerical calculation. We have from (15)
Mathematics Letters 2019; 5(2): 13-22
δ
19
∞
(− x) s −1
(− z ) s −1
(− x) s −1
dx + ∫ z
dz + ∫ x
dx
x
e −1
e −1
e −1
∞
Ω
δ
I ( s) = ∫
∞
=(eiπs − e − iπs)∫
δ
( − z )s −1
× ∫ z
dz + 2 2 × 1010
e
−
1
L
x s −1
dx + Q(δ )
ex − 1
The additional item has the magnitude of 1010 . In fact, for
Re( s ) < 1 ,as long as δ is small enough, − Q(δ ) can be
great enough with
∞
x s −1
= 2i sin(πs) ∫ x
dx + Q(δ )
e −1
δ
(49)
( − z ) s −1
dz
ez −1
L
− Q(δ ) >> ∫
By considering (14) and (49), we get
(− z ) s −1
x s −1
1
=
dx
∫δ e x − 1 2i sin(πs) ∫L e z − 1 dz − Q(δ )
(50)
∞
x s −1
1
( − z )s −1
dx ≠
dz
x
e −1
2i sin( π s ) ∫L e z − 1
0
ζ ( s )Γ ( s ) = ∫
If δ ≠ 0 , we have
∞
0
∞
x s −1
dx
ex −1
0
ζ ( s )Γ ( s ) = ∫
s −1
(51)
=
On the other hand, on the real axis of complex plane with
b = 0 ,according to (46), we have approximately
Q(δ ) =
δ a −1
a −1
cos(π (a − 1) )
(52)
If a > 1 , when δ → 0 , we have Q(δ ) → 0 . This is just the
result of Riemann’s original paper
∞
x s −1
1
( − z )s −1
ζ ( s )Γ ( s ) = ∫ x dx =
dz
e −1
2i sin( π s ) ∫L e z − 1
0
(53)
If a < 1 , for example, to take a = 3 / 4 , we have from (52)
Q (δ ) = −
δ −1 / 4
1/ 4
2 2
π
cos − = − 1 / 4
δ
4
(54)
When δ ≠ 0 , taking δ = 10 −4 , we have Q(δ ) = −2 2 × 10 ,
(50) becomes
x s −1
1
∫−4 e x − 1 dx = 2i sin( π s )
10
( − z )s −1
1
dz + ∞
∫ z
2i sin( π s ) L e − 1
4.3. The Applicable Condition of Formula (12)
The reasons to cause the inconsistency and divergence of
Zeta function equation are discussed below. According to the
definition of Γ(s ) , when Re( s ) > 0 , the integral of (9) is
limited at its lower limit x = 0 . (12) is used when Riemann
deduced (13). It noted that the applicable condition of (12)
is 0 < x ≤ ∞ . This condition can be found in any
mathematics handbook [8] (Let
x′ > 1 ). At point
x = 0 , two sides of (12) become infinite and meaningless
∞
∑e
− nx
= e − x + e −2 x + e −3 x + ⋅⋅⋅ =
1
→∞
ex −1
1
→∞
(60)
0
Besides, contradictory result would be caused by (12) when
x = 0 . For example, when 0 < x ≤ ∞ , according to (12),
we have
1 + 1 + 1 + 1 + ⋅⋅⋅ =
or
There is an additional item 2 2 ×10 in (55). Taking
δ = 10 −40 , we have Q(δ ) = −2 2 × 1010 ,(50)becomes
x′ = e − x ,
with
n =1
(55)
∞
∑e
− nx
= e − x + e − 2 x + e − 3 x + ⋅ ⋅ ⋅ + 1000 >
n =1
∞
s −1
x
1
dx =
x
e −1
2i sin( π s )
10−40
∫
(59)
In this case, the integral form of Riemann Zeta function has
not changed the divergence of its summation form and is
actually meaningless.
∞
( − z )s −1
× ∫ z
dz + 2 2 × 10
L e −1
(58)
If taking δ → 0 ,when Re( s ) < 1 , we have Q (δ ) → ∞
and
∞
x s −1
x s −1
dx
≠
∫ x dx
ex − 1
δ ≠0 e − 1
( −z )
1
dz − Q( δ ≠ 0 )
=
∫ z
2i sin( π s ) L e − 1
(57)
so that we have
∞
ζ ( s )Γ ( s ) = ∫
(56)
∞
∑e
n =1
− nx
= e − x + e − 2 x + e − 3 x + ⋅ ⋅ ⋅ − 1000 <
1
(61)
e −1
x
1
e −1
x
(62)
20
Mei Xiaochun:
If let
The Inconsistency Problem of Riemann Zeta Function Equation
x = 0 , (61) and (62) becomes
The Jacobi’s function satisfied following equation [7, 11]
∞ + 1000 = ∞ > ∞ and ∞ − 1000 = ∞ < ∞
1
θ ( x ) = xθ
x
(63)
These results are absurd, so (12) can not be used at point
x = 0 and (13) can not hold. In Riemann’s original paper,
(12) and its applicable condition had not been mentioned.
Besides (12) is improper at x = 0 , we seems unable to find
other reason to cause the inconsistency in (30) and the infinite
in (59). Meanwhile, just by using (12) and introducing the
Residue theorem, the summation form of ζ (1 − s) was
obtained which caused contradiction with ζ (s ) on the two
sides of Zeta function equation.
x>0
(70)
So (69) can be written as
1 + 2ψ ( x ) = θ ( x ) =
=
1
x
1 1
θ
x x
1
2ψ x + 1
(71)
i.e.
5. The Formula of Jacobi Function is
Improper and Unusable
1 1 1
2ψ + 1 − 1
2 x x
ψ( x ) =
Riemann proved that the Zeta function equation had the
symmetry ξ ( s ) = ξ (1 − s ) in his original paper. The proof
was not clear enough. We give the detail below and prove that
the symmetry does not exist due to the incorrect using of the
formula of Jacobi function. The following Gama function was
used in the deduction which was different from (9) [1, 2].
∞
By means of (72), we obtain [1, 11]
1
∫x
1
ψ ( x )dx = 1 ∫ x s / 2−1 1 2ψ 1 + 1 − 1 dx
s / 2 −1
(64)
=∫x
1
1
ψ dx +
2
x
s / 2 −3 / 2
0
Let x → n 2πx in (64), Riemann obtained
∫(x
− s / 2 −3 / 2
− x s / 2 −1 ) dx (73)
0
1
(65)
(74)
Let x = 1 / y in (73), we get
1
1
s / 2− 3 / 2 1
− s / 2+3/ 2
ψ ( y )( − y −2 )dy
∫0 x ψ x dx = ∞∫ y
∞
2
∞
∞ −s −s / 2 s
∑ n π Γ = ∫ x s / 2 −1 ∑ e − n πx dx
2 0
n =1
n =1
1
x
1
1
x s / 2 −3 / 2 − x s / 2 −1 ) dx =
(
∫
20
s( s − 1 )
∞
2
s
Γ = ∫ e − n πx n sπ s / 2 x s / 2 −1dx
2 0
x
20
0
1
s
Γ = ∫ e − x x s / 2 −1dx
2 0
(72)
(66)
∞
= ∫ y −( s +1 )/ 2ψ ( y )dy
Set
∞
ψ ( x ) = ∑ e − n πx
(67)
n =1
Let y → x on the right side of (75) and considering (74),
(73) becomes
By means of the definition (1), (66) was written as
ζ ( s )π − s / 2 Γ = ∫ x s / 2 −1ψ ( x )dx
2 0
1
=∫x
∞
ψ ( x )dx + ∫ x
s / 2 −1
0
∞
0
1
ψ ( x )dx
∑e
n =−∞
−n π x
2
∞
= 1 + 2∑ e − n π x = 1 + 2ψ ( x )
n =1
(76)
s
(68)
ζ ( s )π − s / 2 Γ
2
1
∞
1
s( s − 1 )
Thus, (68) can be written as
s / 2 −1
In order to calculate (68), the Jacobi’s function was
introduced with
θ( x ) =
1
s / 2 −1
−(s +1)/ 2
ψ ( x )dx +
∫ x ψ ( x )dx = ∫ x
∞
s
(75)
1
2
∞
= ∫ψ ( x ) ( x −(s +1)/ 2 + x s / 2 −1 ) dx +
1
2
(69)
1
s( s − 1 )
(77)
The right side of (77) is unchanged under the replace
s → 1 − s , so its left side should be invariable too under
Mathematics Letters 2019; 5(2): 13-22
s → 1 − s . Set [12, 13]
s
Φ( s) = π − s / 2Γ ζ ( s)
2
6. Discussion on the Zero’s Calculation of
Zeta Function Equation
(78)
(77) indicates the existence of symmetry
1− s
Φ ( s) = Φ(1 − s) = π −(1− s ) / 2Γ
ζ (1 − s) (79)
2
From (78) and (79), we obtain
1 − s
ζ (1 − s )
2
s
2
π − s / 2Γ ζ ( s ) = π −(1− s ) / 2Γ
sπ
s 1− s
ζ ( s) = π s − 3 / 2 sin Γ1 − Γ
ζ (1 − s ) (82)
2 2 2
On the other hand, based on (9), Riemann obtained (13). By
extending (13) to the field of complex plane and using the
residue theorem, Riemann deduced the function equation (5).
If these two methods are consistent, (82) and (5) should be
consistent. We should have
s 1 − s
s −1
= 2(2π ) Γ(1 − s)
2 2
f ( z ) = u ( x, y ) + iv( x, y )
(83)
∂u ∂v
=
∂x ∂y
ξ ( s ) = u (a, b) + iv(a, b)
On the other hand, we have the Legendre double formula of
Gama function
1
Γ s + Γ( s ) = 21− 2 s π 1 / 2Γ(2 s )
2
(85)
Let s → (1 − s ) / 2 in (85), we get (84) and prove (83), so
the results of two methods are the same. But if the neglected
item in (5) which is infinite when Re( s ) < 1 is considered,
both calculations are not consistent again.
Meanwhile, this deduction had a serious problem. The
applicable condition of Jacobi’s equation (70) is x > 0 . If
x = 0 , we have 1 / x → ∞ , (70) becomes meaningless.
Riemann did not mention this condition in his original paper.
Because the lower limits of integrals (73) and (74) are x = 0 ,
the Jacobi’s equation (70) can not be used, (77) can not hold.
The symmetry of (79) does not exist.
(88)
u (a, b) and v(a, b) should satisfy (87). ξ ( s ) = 0 means
that u (a, b) and v(a, b) are equal to zeros simultaneously.
However, in the zero’s calculation of Zeta function,
approximate method was used so that the Cauchy-Riemann
equation (87) was not satisfied.
2. For example, let s = 1 / 2 + ib , (7) was written as [14, 16]
(84)
(87)
Riemann Zeta function is an analytic one, so (7) can also be
written as;
s 1− s
s 1/ 2
Γ 1 − Γ
= 2 π Γ(1 − s )
2 2
∂u
∂v
=−
∂y
∂x
ξ ( 1 / 2 + ib ) = e Reln Γ ( s / 2 )π −1/ 4
or
(86)
If f (z ) is analytic one, its real part and imaginary part are
independent. They should satisfy the following
Cauchy-Riemann equation [9]
(81)
(80) can be written as
π s − 3 / 2 Γ 1 − Γ
By means of manual and numerical calculations, a great
number of zeros were founded up to now (about one thousand
billion) [14, 15]. All of them were regarded to locate on the
straight line of Re( s ) = 1 / 2 . However, they are not the real
zeros of strict Zeta function. Let’s discuss this problem.
1. For a common complex function f ( z ) = f ( x + iy ) , its
real part and imaginary part can always be separated and
written in following form
(80)
By means of the complemented formula of Gama function
π
s s
Γ Γ1 − =
2
2
sin(
s
π / 2)
21
−b 2 − 1 / 4
2
× ei Imln ln Γ ( s / 2 ) π −ib / 4 ζ ( 1 / 2 + ib )
(89)
The practical process to calculate zero was to calculate the
changes of signs in (89). If the sign of Zeta function changed
from positive to negative or from negative to positive, a zero
was considered to be found. Because the item in the first
bracket was always negative, it could be neglected. We only
need to calculate the item in the second bracket. Let
Z (b) = ei Im ln ln Γ ( s / 2 ) π − ib / 4 ζ (1 / 2 + ib)
(90)
By developing Z(b) into the progressive form, the
Riemann-Siegal formula was obtained [14]
Z(b) = 2
∑n
n < t / 2π
−1 / 2
cos(θ (b) − b ln n ) + R(b)
(91)
The zeros calculation was based on (91), in which θ (b) and
R(b) were very complex functions. By taking the different
22
Mei Xiaochun:
The Inconsistency Problem of Riemann Zeta Function Equation
orders of Z(b) , different zeros were found. For example, the
first item of (91) was Z(b) = 2 cosθ (t ) and the first zero was at
b ≈ 14.5 . Then, by considering the revised factor R(b) , the
value was revised into b ≈ 14.1345 , and so on.
It is obvious that original complex Zeta function had
changed into real function in (91). Z(b) was neither real part
nor imaginary part of original Zeta function. The restriction of
the Cauchy-Riemann equation did not exist again. Because
(91) contains trigonometric function, it had infinite b to
satisfy Z(b) = 0 . However, they were not real zeros of the
strict Zeta function.
3. It assumed Re( s ) = 1 / 2 in advance in all computer
calculations about the zeros of Zeta function. Because this is a
precondition, we can not say that the calculations of computer
have proved that all zeros are located on the line of
Re(s) = 1 / 2 . What we can say is that there are great numbers
of zeros on the line of Re( s) = 1 / 2 .
4. In fact, according to (1) and (28), ζ (s ) and ζ (1 − s )
are already series. We can discuss their zeros directly. It is
unnecessary for us to develop them into series again. The
Riemann-Siegal formula (91) not only causes errors, but also
perplexes the problem.
Riemann hypothesis becomes meaningless. Mathematicians
should consider whether it makes sense to research the
distribution of prime numbers based on the Riemann Zeta
function equation again. In other words, we should consider
whether or not the study on prime number distribution should
return to the traditional model in the real number domain.
References
[1]
Riemann G. F. B., Uber die Anzabl der Primahlem unter einer
gegebenen Grosse, Monatsberichte der Berliner Akademine,
1859, 2, 671-680.
[2]
Bent E. Petersen, Riemann Zeta Function, https://pan.
baidu.com/s/1geQsZxL.
[3]
Neukirch, J., Algebraic Number Theory, 1999, Springer,
Berlin, Heidelberg (The original German edition was
published in 1992 under the title Algebraische Zahlentheorie).
[4]
Iwaniec H., Lectures on the Riemann Zeta Function,
American Mathematical Society, 2014, Providence.
[5]
Bender, C. M., Brody, D. C., M¨uller, M. P., Hamiltonian for
the zeros of the riemann zeta function, Physical Review
Letters, 2017, 118 (13), 130201.
[6]
Lagarias, J. C. An elementary problem equivalent to the
Riemann hypothesis, The American Mathematical Monthly,
2002, 109 (5), 534–543.
[7]
Jacobi, C. G. J., Fundamenta Nova Theoriae Functionum
Ellipticarum, Regiomonti, Borntraeger, Konigsberg, 1829,
(Reprinted by Cambridge University Press, 2012).
[8]
Mathematics Handbook, Scientific Publishing House, 1980,
p. 144.
[9]
Guo Dunreng, The Methods of Mathematics and Physics,
Education publishing House, 1965,p. 109.
7. Conclusion
The Riemann hypothesis was one of 23 mathematical
problems presented by Hilbert at the world mathematical
congress in 2000. It was not solved in the 20th century and
was chosen by Clay Institute of Mathematics as one of the
eight millennium mathematical problems.
The Riemann hypothesis is so famous, also due to the fact
that it has become the important basis of modern prime
number distribution theory. Because of Riemann's work, the
traditional prime distribution theory was extended from the
real domain to the complex domain. According to statistics,
nearly a thousand theorems had been proposed based on the
premise of assuming the Riemann hypothesis to be correct so
far. We can see the depth of its influence on modern
mathematics.
Since the Riemann hypothesis had not been proved for a
long time, some mathematicians believed that it was not true.
But most mathematicians believed it to be true, and they
wanted it to be true. However, this paper gives third result.
The Riemann hypothesis is proved to be meaningless, which is
unexpected by most people.
The problem lied in Riemann's original paper of 1859. In
the Riemann’s deduction, two formulas were improperly used
without considering their applicable conditions which caused
the inconsistency of the Zeta function equation. Meanwhile,
an integral term around the origin point of coordinate system
was omitted. This item was infinite when Re( s ) < 1 , so the
integral form of Zeta function had not changed the divergence
of its series summation form.
Because the integral form of Zeta function and its function
equation are not tenable generally when Re( s ) < 1 , the
[10] Felix Rubin, Riemann's First Proof of the Analytic
Continuation of Zeta function, http://www2.math. ethz.ch/edu
cation/bachelor/seminars/ws0607/modular-forms/Riemanns_fi
rst_proof. pdf.
[11] Gerald Tenenbaum, Michel Mendes France, Les Nombres,
Premiers, Presses Universitaires de France, 1997, 44-52.
[12] Gelbart, S., Miller, S., Riemann’s zeta function and beyond,
Bulletin of the American Mathematical Society, 2004, 41 (1),
59–112.
[13] Jessen, B., Wintner, A., Distribution functions and the
Riemann zeta function, Transactions of the American
Mathematical Society, 1935, 38 (1), 48–88.
[14] Ru Changhai, The Riemann Hypothesis, Qianghua University
Publishing Company, 2016, p. 61, 52, 192.
[15] Gourdon X., The 10^(13) first zeros of the Riemann Zeta
function, and zeros computation at very large height, 2004,
http://pdfs.semanticscholar.org/6eff/62ff5d98e8ad2ad8757c0f
af4bac87546f27. pdf.
[16] Katz N M, Sarnak P. Zeroes of zeta functions and symmetry [J].
AMS,1999,36 (1) 1-26.