A Geometric Proof of Riemann Hypothesis

Kaida Shi

Department of Mathematics, Zhejiang Ocean University,

Zhoushan City, Zip.316004, Zhejiang Province, China

Abstract Beginning from the resolution of Riemann Zeta function ζ (s ) , using the inner product
formula of infinite-dimensional vectors in the complex space, the author proved the world’s problem
Riemann hypothesis raised by German mathematician Riemannn in 1859.
Keywords: complex space, solid of rotation, axis-cross section, bary-centric coordinate, inner
product between two infinite-dimensional vectors in complex space.
MR(2000): 11M

1 Introduction

In 1859, German mathematician B. Riemann raised six very important hypothesizes about ζ (s ) in

[1]
his thesis entitled Ueber die Anzahl der Primzahlen unter einer gegebenen Große . Since his thesis

was published for 30 years, five of the six hypothesizes have been solved satisfactorily. The remained

one is the Super baffling problemRiemann hypothesis. That is the non-trivial zeroes of the Riemann

1
Zeta function ζ (s ) are all lie on the straight line Re( s ) = within the complex plane s. The well-
2

known Riemann Zeta function ζ ( s) raised by Swiss mathematician Leonard Euler on 1730 to 1750.

In 1749, he had proved the Riemann Zeta function ζ ( s) satisfied another function equation. For

Re( s) > 1 , the series

1 1 1 1
s
+ s + s +L+ s +L
1 2 3 n

is convergent, it can be defined as ζ ( s) . Although the definitional domain of the Riemann Zeta

∞
1
function ζ ( s) = ∑ s
( s = σ + it ) is Re( s) > 1 , but German mathematician Hecke used the
n =1 n

1
analytic continuation method to continue this definitional domain to whole complex plane (except

Re( s ) = 1 ).

In 1914, English mathematician G. H. Hardy declared that he had proved Riemann Zeta function

1
ζ ( s) equation has infinite non-trivial zeroes lie on the straight line Re( s) = .
2

In fact, about the Riemann Zeta functional equality

s 1− s
− s − 1− s
π 2 Γ ( )ζ ( s) = π 2
Γ( )ζ (1 − s), (*)
2 2

because

1
1− 12
−2 1
1 − 1 − 12 1
π 2
Γ( )ζ ( ) = π
2 2
Γ( )ζ (1 − ),
2 2 2 2

1 1
therefore, we have discovered that the solution of above equality is s = , namely Re( s ) = .
2 2

1
That is to say no matter how the value t takes, σ always equals to . This explains that the
2

definitional domain of the Riemann Zeta function ζ (s ) can be continued to at least

1
Re( s ) = .
2

2 The formal resolution of Riemann Zeta function ζ ( s)

Considering the geometric meaning of the Riemann Zeta function

1 1 1 1
ζ ( s) = s
+ s + s +L+ s +L,
1 2 3 n

we will have the following figure:

2
 w

C(1) A1
1
1s
A2
B1 1
2s
B2 A3
1
3s
B3

O D1(1) D2(2) D3(3) … z

Fig. 1
In above figure, the areas of the rectangles A1 D1 OC , A2 D2 D1 B1 , A3 D3 D2 B2 , L are respectively:

1 1 1
, , , L,
1s 2 s 3 s

therefore, the geometric meaning of the Riemann Zeta function ζ ( s) is the sum of the areas of a series of

rectangles within the complex space s.

Using the inner product formula between two infinite-dimensional vectors, the Riemann Zeta

function ζ ( s) equation

1 1 1 1
ζ ( s) = s
+ s + s +L+ s +L = 0 ( s = σ + it )
1 2 3 n

can be resolved as

1 1 1 1 1 1 1 1
( σ , σ , σ , L , σ , L) ⋅ ( it , it , it , L, it , L) = 0.
1 2 3 n 1 2 3 n

So, we should have (Please refer to the Appendix in the end)

1 1 1 1 1 1 1 1 r r
+ + +L+ +L⋅ ⋅ + + +L +L cos( N 1 , N 2 ) = 0, (1)
12σ 2 2σ 32σ n 2σ 12it 2 2it 32it n 2it

but in the complex space, if the inner product between two vectors equals to zero, then these two vectors

3
 r r π
are perpendicular, namely, ( N 1 , N 2 ) = , therefore
2
r r
cos( N 1 , N 2 ) = 0.

So long as the values of two radical expressions in the left side of (1) are finite (real or imaginary), then

we think the expression (1) is tenable.

Because Riemann had proved that when Re( s ) > 1 , the equation ζ ( s) = 0 has no solutions;

when Re( s ) < 0 , the Riemann Zeta function ζ ( s) has only the trivial zeroes such as

s = −2, −4, −6, L , therefore, the non-trivial zeroes of the Riemann Zeta function ζ ( s) will lie

on the banded domain 0 ≤ σ < 1 . This explains that German mathematician Hecke used the

analytic continuation method to continue the definitional domain of the Riemann Zeta function

ζ ( s) is tenable.

3 The relationship between the volume of the rotation solid and

the area of its axis-cross section within the complex space

1
From the expression (1), we can know that when Re( s ) = σ = , the series (harmonic series)
2
r r
within the first radical expression of left side is divergent. Because cos( N 1 , N 2 ) = 0, therefore the

situation of the complex series within the second radical expression of left side will change unable to

research. Hence, we must transform the Riemann Zeta function ζ ( s) equation. For this aim, let’s derive

the relationship between the volume of the rotation solid and the area of its axis-cross section within the

complex space.

We call the cross section which pass through the axis z and intersects the rotation solid as axis-

cross section. According to the barycentric formula of the complex plane lamina:

4
  b

ξ = ∫ z[ f ( z) − g ( z)]dz ,
a
 b

 ∫ [ f ( z) − g( z)]dz
a
 b
 ∫ [ f ( z) − g ( z)]dz .
1 2 2
2 a
η = b

 ∫ [ f ( z) − g( z)]dz
a

we have

b
π ∫ [ f 2 ( z) − g 2 ( z)]dz
2πη = a
b
. (**)
∫ a
[ f ( z ) − g ( z )]dz

The numerator of the fraction of right side of the formula (**) is the volume of the rotation solid,

and the denominator is the area of the axis-cross section. The η of left side of the formula (**) is the

longitudinal coordinate of the barycenter of axis-cross section. Its geometric explanation is: taking the

axis-cross section around the axis z to rotate the angle of 2π , we will obtain the volume of the rotation

solid.

4 The proof of divergence of two concerned serieses

∞ ∞
Now, let’s prove the divergence of the series ∑ cos(4t ln n) and the series ∑ sin(4t ln n) .
n =1 n =1

y
y = cos(4t ln x ) (t = 2)
1

0 1 2 3 4 5 6 7 8 x
-1

Fig. 2
∞
Obviously, from Fig.2, we can find that the area of c ⋅ ∑ cos(4t ln n)
n =1
is contained by the area

5
 +∞
of ∫ 1
cos(4t ln x )dx .

Suppose that c is a very small positive number, then the sum of the areas of a series of rectangles

(which take cos( 4t ln n)( n = 1,2,3, L ) as high and take c as width) is:

∞
c ⋅ ∑ cos(4t ln n) = c ⋅ cos(4t ln 1) + c ⋅ cos( 4t ln 2) + c ⋅ cos( 4t ln 3) + L .
n =1

Because

+∞ 1
∫ 1
cos( 4t ln x)dx = lim
w→ +∞ 1 + 16t 2
x[cos(4t ln x ) + 4t sin(4t ln x)]1w
t →∞

1
= lim [w(cos(4t ln w) + 4t sin(4t ln w)) − 1]
w→ +∞ 1 + 16t 2
t →∞

w  1 4t 
= lim  cos( 4t ln w) + sin( 4t ln w)  − lim 1 2
w→ +∞ 2   t →∞ 1 + 16t
t →∞ 1 + 16t  1 + 16t 2
1 + 16t 2


1 4t
Suppose that = sin δ and = cos δ , then above equals to
1 + 16t 2
1 + 16t 2

w 1
lim sin(δ + 4t ln w) − lim
w→ +∞
t →∞ 1 + 16t 2 t →∞ 1 + 16t 2
d
= lim sin(δ + 4t ln w)
w→ +∞
1
t →∞
+ 16
t2
d
= lim sin(δ + 4t ln w).
4 tw→→∞+∞

Because the relationship between w and t is linear, therefore, we denote w = dt .

Obviously,

d d d
− ≤ lim sin(δ + 4t ln w) ≤ ,
4 4 t →∞
w → +∞ 4

therefore

6
 ∞
d d
− ≤ c ⋅ ∑ cos( 4t ln n) ≤ ,
4 n =1 4

namely,

∞
d d
− ≤ ∑ cos( 4t ln n) ≤ .
4c n=1 4c

∞
Thus, although the series ∑ cos(4t ln n)
n =1
is divergent, but it is oscillatory and bounded.

∞
Similarly, although the series ∑ sin(4t ln n) is divergent, but it is also oscillatory and bounded.
n =1

∞
In a word, when t = 0 , the series ∑ cos(4t ln n)
n =1
is divergent (goes to infinite); when t ≠ 0 ,

it is oscillatory and bounded.

∞
Thus, the divergence (oscillatory and bounded) of ∑ cos(4t ln n)
n =1
has been proved.

Similarly, we can prove the divergence (oscillatory and bounded) of the series

∑ sin(4t ln n)
n =1
(t ≠ 0) .

5 Two theorems and their corollaries

Suppose that

1− s
1 − 1− s
ξ ( s) = s( s − 1)π 2
Γ( )ζ (1 − s),
2 2

then, we have

ξ ( s) = ζ ( s) .

According to the following:

7
 Theorem 1[2] The function ξ ( s) is the one class integer function, it has infinite zero points ρ n ,
∞ ∞

∑ ρn ε > 0 ，the series ∑ ρ n

−1 −1− ε
0 ≤ Re( ρ n ) ≤ 1 , the series is divergent. But for ∀ is
n =1 n =1

convergent. The zeroes of the function ξ ( s) are non-trivial zeroes of the Riemann Zeta function.

We can obtain:

Corollary 1[2] The formula

s
∞
 s  ρ
ξ ( s) = e A + Bs
∏ 1 −  e n
n =1  ρn 

is tenable.

Corollary 2[2] The non-trivial zeroes of the Riemann Zeta function ζ ( s) distribute

1
symmetrically on the straight line Re s = and Im s = 0 .
2

it σ 0 + it2

σ 1 + it0 • • • σ 2 + it0

o Re s = 1 2 σ
σ 0 + it1 •

Fig. 3
1
According to the symmetry (about Re( s ) = ) of distribution of non-trivial zeroes of the
2

Riemann Zeta function ζ ( s) stated by the Corollary 2, we can consider the uniqueness of non-trivial

zeroes of the Riemann Zeta function ζ ( s) for both the imaginative coordinate t 0 and the real

coordinate σ 0 , we therefore suggest:

Theorem 2 (The theorem about uniqueness and continuation of the non-trivial zero of the

8
Riemann Zeta function ζ ( s) ) The imaginative coordinate t 0 of each non-trivial zero of the

Riemann Zeta function ζ ( s) doesn’t correspond with two or over two real coordinates

σ k (k = 1, 2, 3, L) ; but the real coordinate σ 0 of each non-trivial zero of the Riemann Zeta

function ζ ( s) corresponds with infinite imaginative coordinates t k ( k = 1, 2, 3, L) .

Proof First, suppose that t 0 corresponds with two real coordinates σ 1 and σ 2 ，then, we

have following equation group:

 1 1 1 1
ζ (σ 1 + it 0 ) = 1σ 1 +it0 + 2 σ 1 +it 0 + 3σ 1 +it0 +L+ n σ 1 +it0 +L = 0,

ζ (σ 2 + it 0 ) = 1 + 1 + 1 +L+ 1 +L = 0.
 1σ 2 +it0 2 σ 2 +it 0 3σ 2 +it0 n σ 2 +it0

Taking the first equation minus the second equation, we will obtain

∞
1 1
ζ (σ 1 + it 0 ) − ζ (σ 2 + it 0 ) = ∑ ( σ 1 + it0
− σ 2 + it0
)
n =1 n n
∞
nσ 2 − nσ 1
=∑ ( σ 1 + σ 2 ) + it 0
,
n =1 n

because σ 1 ≠ σ 2 ，therefore, n σ 2 − n σ 1 ≠ 0 . On the other hand, because

n (σ 1 +σ 2 )+it 0 = n σ1 +σ 2 ⋅ n it0 = n σ 1 +σ 2 ⋅ e it 0 ln n
= n σ 1 +σ 2 (cos(t 0 ln n) + i sin(t 0 ln n)) ≠ 0,

therefore we have

ζ (σ 1 + it 0 ) − ζ (σ 2 + it 0 ) ≠ 0

namely

ζ (σ 1 + it 0 ) ≠ ζ (σ 2 + it 0 ) .

This is contradictory with the fact that the first equation minus the second equation must equal to zero.

Hence, the uniqueness of the real coordinate σ of each non-trivial zero of the Riemann Zeta

function ζ ( s) has been proved.

9
 Second, suppose that the real coordinate σ 0 of each non-trivial zero of the Riemann Zeta

function ζ ( s) corresponds with two imaginative coordinates t1 and t 2 , then, we have following

equation group:

 1 1 1 1
ζ (σ 0 + it1 ) = 1σ 0 + it1 + 2 σ 0 + it1 + 3σ 0 + it1 + L + n σ 0 +it1 + L = 0,

ζ (σ 0 + it 2 ) = 1 + 1 + 1 + L + 1 + L = 0.
 1σ 0 +it 2 2σ 0 +it 2 3σ 0 +it 2 n σ 0 + it2

Taking the first equation minus the second equation, we obtain

∞
1 1
ζ (σ 0 + it 1 ) − ζ (σ 0 + it 2 ) = ∑ ( σ 0 +it1
− σ 0 + it2
)
n =1 n n

∞
n it 2 − n it1 ∞
e it 2 ln n − e it1 ln n
= ∑ σ 0 +i ( t1 +t 2 ) = ∑
n =1 n n =1 n σ 0 +i ( t1 + t2 )

∞
(cos(t 2 ln n) − cos(t1 ln n)) + i(sin(t 2 ln n) − sin(t1 ln n))
=∑ ,
n =1 n σ 0 +i (t1 +t2 )

where

n σ 0 + i (t1 + t2 ) = n σ 0 ⋅ n i (t1 +t 2 ) = n σ 0 ⋅ e i (t1 +t2 ) ln n

= n σ 0 ⋅ (cos(t1 ln n) + i sin(t1 ln n)) ⋅ (cos(t 2 ln n) + i sin(t 2 ln n))
≠ 0.

Enable above expression equals to zero, we must have

cos(t 2 ln n) = cos(t1 ln n),

 (n = 1, 2, 3, L)
sin(t 2 ln n) = sin(t1 ln n)

2kπ
so, we obtain t1 = t2 + (k = 1, 2, 3,L) .That is to say t1 and t 2 can take any value, but
ln n

2kπ
t1 − t 2 = (k = 1, 2, 3,L) .
ln n

Thus, theorem 2 has been proved.

Corollary 3 According to the solution of the equality (*) and the symmetry of Corollary 2 and

10
the uniqueness and the continuation of Theorem 2, we can say certainly that if t ∈ ( −∞, +∞) , then

1
all non-trivial zeroes of the Riemann Zeta function ζ ( s) lie on straight line Re( s) = (except a
2

certain point).

6 The proof of Riemann hypothesis

Because we have the volume formula of the rotation solid formed by rotating the rectangle

An Dn Dn −1 Bn−1 around the axis z:

n π π π
Vn = ∫ 2s
dz = 2s
z n
n −1 = ,
n −1 n n n2s

therefore, the sum of the volumes of a series of the cylinders formed by rotating a series of the

rectangles A1 D1 OC , A2 D2 D1 B1 , A3 D3 D2 B2 ,L around the axis z is:

∞
1 1 1 1
V = ∑ Vn = π ( 2s
+ 2 s + 2 s +L+ 2 s +L ) .
n =1 1 2 3 n

But the sum of the areas of a series of the rectangles A1 D1 OC , A2 D2 D1 B1 , A3 D3 D2 B2 ,L is:

1 1 1 1
s
+ s + s +L+ s +L .
1 2 3 n

By the formula (**) in Section 2, we have

1 1 1 1
π( 2s
+ 2 s + 2 s +L+ 2 s +L )
1 2 3 n

1 1 1 1
= 2πη ( s
+ s + s +L+ s +L ). (2)
1 2 3 n

Substituting the Riemann Zeta function ζ ( s) equation

1 1 1 1
ζ ( s) = s
+ s + s +L+ s +L = 0
1 2 3 n

11
into (2), we can obtain the transformed equation:

1 1 1 1
2s
+ 2 s + 2 s +L+ 2 s +L = 0 .
1 2 3 n

This can be explained geometrically as: if the area of the axis-cross section equals to zero, then it

corresponds to the volume of the rotation solid also equals to zero.

Because s = σ + it , therefore

1 1 1 1
2 (σ + it )
+ 2 (σ + it )
+ 2 (σ + it )
+L+ 2 ( σ + it )
+L = 0 .
1 2 3 n

According to the inner product formula between two infinite-dimensional vectors, above expression can

be written as:

1 1 1 1 1 1 1 1
( 2σ , 2σ , 2σ , L , 2σ , L) ⋅ ( 2it , 2it , 2it , K, 2it , L) = 0.
1 2 3 n 1 2 3 n
r r
Denoting above two infinite-dimensional vectors as N 3 and N 4 respectively, we have

1 1 1 1 1 1 1 1 r r
+ + + L + + L ⋅ + + + L + L ⋅ cos( N 3 , N 4 ) = 0 （3）
14σ 2 4σ 3 4σ n 4σ 14it 2 4it 3 4it n 4it

Now, let’s prove Riemann hypothesis. If t will take all values of a certain infinite interval, then we

think that Riemann hypothesis has been proved.

From the Riemann Zeta functional equality

s 1− s
− s − 1− s
π 2 Γ ( )ζ ( s) = π 2
Γ( )ζ (1 − s), (*)
2 2

1
we have known its solution is Re( s) = . Substituting it into the expression (3), we will have:
2

1 1 1 1 1 1 1 1 r r
+ + + L + + L ⋅ + + + L + + L ⋅ cos( N 3, N4) = 0
12 2 2 3 2 n2 14it 2 4it 3 4it n 4it

namely,

12
 π 1 1 1 1 r r
4it
+ 4it + 4it + L + 4it + L ⋅ cos( N 3 , N 4 ) = 0 .
6 1 2 3 n

r r π
When ( N 3 , N 4 ) = , we have
2

π 1 1 1 1
4 it
+ 4it + 4it +L+ 4it +L ⋅ 0 = 0 .
6 1 2 3 n

In above expression, when t = 0 , although the series

∞ ∞

∑ sin(4t ln n) = ∑ sin(4 ⋅ 0 ⋅ ln n)
n =1 n =1

= 0 + 0 + 0 + L + 0 + L = 0,

but because the series

∞ ∞

∑ cos(4t ln n) = ∑ cos(4 ⋅ 0 ⋅ ln n)
n =1 n =1

= 1 + 1 + 1 + L + 1 + L = ∞,

therefore, the complex series

1 1 1 1
4 it
+ 4it + 4it + L + 4it + L
1 2 3 n
∞
= ∑ [cos(4t ln n) − i sin( 4t ln n)]
n =1

= (1 + 1 + 1 + L + 1 + L) −
− i ⋅ (0 + 0 + 0 + L + 0 + L) = ∞.

Because “infinite” multiplied by zero doesn’t equal to zero, therefore

π 1 1 1 1
4it
+ 4it + 4it + L + 4it + L ⋅ 0
6 1 2 3 n
π ∞
=
6
⋅ ∑ [cos(4t ln n) − i sin(4t ln n)] ⋅ 0
n =1

π
= ⋅ ∞ ⋅ 0 ≠ 0.
6

13
 1
So, Re( s ) = (t = 0) is not a non-trivial zero of the Riemann Zeta function ζ (s ) , it must be
2

removed.

∞ ∞
When t ≠ 0 , although the series ∑ cos(4t ln n) and series ∑ sin(4t ln n) are all divergent (see
n=1 n =1

Section 3), but they are all oscillatory and bounded, therefore we can know the complex series

1 1 1 1
4 it
+ 4it + 4it + L + 4it + L
1 2 3 n
∞
= ∑ [cos(4t ln n) − i sin( 4t ln n)]
n =1

is also oscillatory and bounded.

From following figure:

∞
d
∑ [cos( 4t ln n) − i sin( 4t ln n)]
n =1 4c
i

d o d x
−
4c 4c

d
− i
4c

Fig. 4

∞
we can know the change scope of the complex series ∑ [cos( 4t ln n) − i sin( 4t ln n)]
n =1
is the square.

Obviously, the finite complex numbers multiplied by zero equal to zero, therefore we have

14
 π 1 1 1 1
4it
+ 4it + 4it + L + 4it + L ⋅ 0
6 1 2 3 n
π ∞
=
6
⋅ ∑ [cos(4t ln n) − i sin(4t ln n)] ⋅ 0
n =1

= 0, ( t ≠ 0)

s
By the expression (1), considering the situation of Γ( ) in ζ ( s) , we have known that when
2

s = 0 , ζ ( s) ≠ 0 , therefore, s = 0 is not a non-trivial zero of the Riemann Zeta function ζ ( s) .

Because s = σ + it ，therefore t = 0 is also not a non-trivial zero of the Riemann Zeta function

ζ ( s) . This is identical with the situation of the sufficient and essential condition of divergence of the

complex series, therefore, we obtain

t ∈ (−∞, 0) U (0, +∞) .

1
This explains that t takes all values of the straight line Re( s) = except t = 0 .
2

Obviously, this result is identical with the corollaries of theorem 1 and theorem 2 within Section 4.

7 The progress in the process for proving Riemann hypothesis

Riemann has proved that Zeta function satisfied the function equation:

s
s −
ζ (1 − s) = ζ ( s)Γ ( )π 2
,
2

in which, when s is the positive number, we have

∞
Γ( s ) = ∫ x s −1e − x dx .
0

Especially, when s is the positive integral, we have

Γ ( s + 1) = s! ,

15
when s is arbitrary number (include negative number), we have

n ! n s −1
Γ ( s) = lim .
n →∞ s( s + 1)( s + 2)L ( s + n − 1)

In 1942, Norway mathematician A. Selberg proved

N 0 (T ) ≥ CT log T， (C > 0) .

In fact, Selberg proved

N 0 (T ) ≥ CN (T ) ， (C > 0，C ∈ R) .

but, Selberg obtained the constant C has only 0.01.

In 1974, mathematician N. Levinson at Massachusetts Institute of Technology of USA proved

successfully that for full great real number T,

1
N 0 (T ) ≥ N (T ) .
3

that is to say Riemann Zeta function ζ ( s) has at least a third non-trivial zeroes lie on the straight

1
line Re( s) = .
2

In 1980, Chinese mathematician Shituo Lou after improving slightly Levinson’s result and proved:

N 0 (T ) > 0.35N (T ) .

These are most new general survey about the research of Riemann hypothesis[3
～8]
.

In 1968, by using great computer, three mathematicians at University of Wisconsin of USA proved

1
first 3,000,000 non-trivial zeroes of Riemann Zeta function ζ ( s) lie on the straight line Re( s) = .
2

After that, even one man using great computer checked the first 1,500,000,000 non-trivial zeroes of

1 [8]
Riemann Zeta function ζ ( s) lie on the straight line Re( s) = . But the ability of the computer is
2

limited, it can’t achieve infinite, therefore, that can’t to say whether Riemann hypothesis about the non-

16
trivial zeroes of Zeta function ζ ( s) is correct or not, it only provides the basis for supporting the

hypothesis.

References

1. Lou S. T., Wu D. H., Riemann hypothesis, Shenyang: Liaoning Education Press, 1987. pp.152-154.

2. Chen J. R., On the representation of a large even integer as the sum of a prime and the product of at most two

primes, Science in China, 16 (1973), No.2, pp. 111-128..

3. Pan C. D., Pan C. B., Goldbach hypothesis, Beijing: Science Press, 1981. pp.1-18; pp.119-147.

4. Hua L. G., A Guide to the Number Theory, Beijing: Science Press, 1957. pp.234-263.

5. Chen J. R., Shao P. C., Goldbach hypothesis, Shenyang: Liaoning Education Press, 1987. pp.77-122; pp.171-205.

6. Chen J. R., The Selected Papers of Chen Jingrun, Nanchang: Jiangxi Education Press, 1998. pp.145-172.

7. Lehman R. S., On the difference π(x)-lix, Acta Arith., 11(1966). pp.397～410.

8. Hardy, G. H., Littlewood, J. E., Some problems of “patitio numerorum” III: On the expression of a number as a

sum of primes, Acta. Math., 44 (1923). pp.1-70.

9. Hardy, G. H., Ramanujan, S., Asymptotic formula in combinatory analysis, Proc. London Math. Soc., (2) 17 (1918).

pp. 75-115.

10. Riemann, B., Ueber die Anzahl der Primzahlen unter einer gegebenen Große, Ges. Math. Werke und

Wissenschaftlicher Nachlaß , 2, Aufl, 1859, pp 145-155.

11. E. C. Titchmarsh, The Theory of the Riemann Zeta Function, Oxford University Press, New York, 1951.

12. Morris Kline, Mathematical Thought from Anoient to Modern Times，Oxford University Press, New York, 1972.

13. A. Selberg, The Zeta and the Riemann Hypothesis, Skandinaviske Mathematiker Kongres, 10 (1946).

Appendix

The inner product formula between two vectors in real

space can be extended formally to complex space

17
 Dear Mr. Referee,

Thank you for your review.

First, please observe following examples.

Example 1 A{1 + i, 3}, B{−i, 2i}, C{1, −i}.

AB = {−1 − 2i, −3 + 2i}, AC = {−i, −3 − i}, BC = {1 + i, −3i}.

AB = (−1 − 2i ) 2 + (−3 + 2i) 2 = 2 − 8i ,

AC = (−i ) 2 + (−3 − i ) 2 = 7 + 6i ,

BC = (1 + i ) 2 + (−3i ) 2 = − 9 + 2i .
According to Cosine theorem

2 2 2
AB + AC − BC = 2 AB AC cos( AB, AC ) ,
we have

2 2 2
AB + AC − BC (2 − 8i ) + (7 + 6i) − (−9 + 2i)
= = 9 − 2i.
2 2
On the other hand, we have

AB ⋅ AC = AB AC cos( AB, AC ) = (−1 − 2i )(−i) + (−3 + 2i )(−3 − i ) = 9 − 2i.

According to the Cosine theorem

2 2 2
AC + BC − AB = 2 AC BC cos( AC , BC )

2 2 2
AC + BC − AB (7 + 6i ) + (−9 + 2i ) − (2 − 8i)
= = −2 + 8i.
2 2

On the other hand, we have

AC ⋅ BC = AB AC cos( AB, AC ) = ( −i)(1 + i ) + ( −3 − i)(−3i) = −2 + 8i.

18
 1 1 − 15 − 8i 1
S ∆ABC = AB AC sin θ 1 = 2 − 8i ⋅ 7 + 6i ⋅ = − 15 − 8i ,
2 2 62 − 44i 2

1 1 − 15 − 8i 1
S ∆ACB = AC BC sin θ 21 = 7 + 6i ⋅ − 9 + 2i ⋅ = − 15 − 8i .
2 2 − 75 − 40i 2

Example 2 A{1 + i, 1 − i, 2i}, B{1 − i, 1 + i, −2i}, C{1, 0, i}.

AB = {−2i, 2i, −4i}, AC = {−i, −1 + i, −i}, BC = {i, −1 − i, 3i}.

AB = (−2i) 2 + (2i ) 2 + (−4i ) 2 = − 24 ,

AC = (−i ) 2 + (−1 + i ) 2 + (−i ) 2 = − 2 − 2i ,

BC = (i ) 2 + (−1 − i) 2 + (3i ) 2 = − 10 + 2i .
According to the Cosine theorem

2 2 2
AB + AC − BC = 2 AB AC cos( AB, AC ) ,
we have

2 2 2
AB + AC − BC (−24) + (−2 − 2i) − (−10 + 2i)
= = −8 − 2i.
2 2
On the other hand, we have

AB ⋅ AC = AB AC cos( AB, AC ) = ( −2i )(−i) + ( 2i)(−1 + i) + (−4i)(−i) = −8 − 2i.

According to the Cosine theorem

2 2 2
AC + BC − AC = 2 AC BC cos( AC , BC ) ,

we have

2 2 2
AC + BC − AB (−2 − 2i) + (−10 + 2i) − (−24)
= = 6.
2 2

On the other hand, we have

AC ⋅ BC = AC BC cos( AC , BC ) = (−i )(i) + (−1 + i )(−1 − i ) + (−i )(3i) = 6.

19
 1 1 − 12 + 16i
S ∆ABC = AB AC sin θ 1 = − 24 ⋅ − 2 − 2i ⋅ = − 3 + 4i ,
2 2 48 + 48i

1 1 − 12 + 16i
S ∆ACB = AC BC sin θ 21 = − 2 − 2i ⋅ − 10 + 2i ⋅ = − 3 + 4i .
2 2 24 + 16i

Example 3 A{8i, 14, 8 − i, 1}, B{6, 15i, 17, −8}, C{3 − i, 10 + 7i, 11, 3i}.

AB = {6 − 8i, −14 + 15i, 9 + i, −9}, AC = {3 − 9i, −4 + 7i, 3 + i, −1 + 3i},

BC = {−3 − i, 10 − 8i, −6, 8 + 3i}.
AB = (6 − 8i ) 2 + (−14 + 5i ) 2 + (9 + i ) 2 + (−9) 2 = 104 − 498i ,

AC = (3 − 9i ) 2 + (−4 + 7i ) 2 + (3 + i ) 2 + (−1 + 3i ) 2 = − 105 − 110i ,

BC = (−3 − i) 2 + (10 − 8i ) 2 + (−6) 2 + (8 + 3i ) 2 = 135 − 106i .

According to the Cosine theorem

2 2 2
AB + AC − BC = 2 AB AC cos( AB, AC ) ,
we have

2 2 2
AB + AC − BC (104 − 498i ) + (−105 − 110i ) − (−135 − 106i )
= = −68 − 251i.
2 2
On the other hand, we have

AB ⋅ AC = AB AC cos( AB, AC ) =
= (6 − 8i)(3 − 9i) + ( −14 + 15i)(−4 + 7i) + (9 + i)(3 + i ) + ( −9)(−1 + 3i) = −68 − 251i.

According to the Cosine theorem

2 2 2
AC + BC − AC = 2 AC BC cos( AC , BC ) ,

we have

2 2 2
AC + BC − AB (−105 − 110i ) + (135 − 106i ) − (104 − 498i)
= = −37 + 141i.
2 2

On the other hand, we hand

20
AC ⋅ BC = AC BC cos( AC , BC ) =
= (3 − 9i )(−3 − i ) + ( −4 + 7i )(10 − 8i ) + ( −6)(3 + i ) + (−1 + 3i )(8 + 3i ) = −37 + 141i.

1 1 − 7323 + 6714i
S ∆ABC = AB AC sin θ 1 = 104 − 498i ⋅ − 105 − 110i ⋅
2 2 − 65700 + 40850i
1
= − 7323 + 6714i ,
2

1 1 − 7323 + 6714i
S ∆ACB = AC BC sin θ 21 = − 105 − 110i ⋅ 135 − 106i ⋅
2 2 − 25835 − 3720i
1
= − 7323 + 6714i .
2

21