Examination of the Proof of Riemann’s hypothesis and the conditions for it to

hold for Generalized -functions. It holds for Riemann’s -function and the
resulting distribution of primes

N.Mantzakouras
e-mail: [email protected]
Athens - Greece

Abstract

Riemann’s zeta function is one of the most important and fascinating functions of Euler’s math-
ematics. By analyzing the material of Riemann’s hypothesis, we divide our analysis into the
function ζ(z) and the proof of the conjecture, which has very important consequences in terms
of the distribution of prime numbers. The proof of Riemann’s hypothesis follows from the simple
logic,that when two properties are related, i.e. these equations are zero i.e. ζ(z) = ζ(1 − z) = 0
while they have the proven 1 − 1 property of Riemann’s function ζ(z), then the Hypothesis will be
true, because it has been proved from before simultaneously that Riemann’s functional equations
hold. Therefore, there will be apply ζ(z) = 1/2 {because ζ(z) = ζ(1 − z) = 0 and also because
ζ(z) as well as ζ(1 − z) are 1-1}. This, being true, will give the direction of all non-trivial roots
to all be on the critical line, with a value of 1/2 on the real axis. Furthermore, to strengthen
the proof we consider ”5 important requirements” that clearly prove that 100% of the roots of
ζ(z) = 0 are only on the critical line. There is of course the case ζ(q ∗ z) = 0 with q > 1 in R,,
which has different critical lines, but all those who deal with the Hypothesis of Riemann, do not
mention it , because they claim that this form was not defined by Riemann from the begining.
But if we generalize the hypothesis to ζ(q ∗ z) = 0, q > 1 in R then we have infinite critical
lines that originate from and are related to the baseline critical line of Re(z) = 1/2, but are
not identical to it and lie within the interval (0, 1). In summary and by analytically examining
and other similar functions to ζ(z), we get zeros with Re(z) in the interval (0, 1) for the cases
ζ(q ∗ z) = 0 where q > 1 in R that mentioned, for Generalized case G − ζ(z, q) = 0 with 3 general
cases out of 5 which is the generalized sum including ζ(z) and end for the Davenport Heilbronn
case, which has zeros of a periodic Dirichlet series with different critical lines, within the interval
(0, 1). We are likely to have other constructed cases with relevance of ζ(z). Therefore, whether
or not the Riemann Hypothesis, holds depends on whether or not we accept the fact that the
Riemann Hypothesis is or is not fully mentioned, to functional equation ζ(z) = 0.

Part I. The function ζ(z) is 1-1 on the critical strip

1.Theorem

The Hypothesis of Riemann focuses on the point where we must prove that if s = Re(s)+
+Im(s)∗ I,

I) The functions ζ(s) and ζ(1 − s) are 1-1 on the critical strip.
II) The common roots of the equations ζ(s)− ζ(1 − s) = 0 they have Re(s) = Re(1 − s) = 1/2
within the interval (0,1) and determine unique position, which is called critical line.

I.1. Proof:
The functions ζ(s) and ζ(1 − s) are 1-1 on the critical strip. For this we need to analyze when and where
the exponential function nz is 1-1 when n ∈ Z + .
The exponential function nz :

1
I) is 1-1 in each of these strips defined by the intervals 2πk/ ln(n) and 2π(k + 1)/ ln(n) “where k ∈ Z.
For this we must prove two cases. If A ⊂ C ∧ f : A → C then if f (z1 ) = f (z2 ) then z1 = z2 that is..
x1 = x2 ∧ y1 = y2 . Indeed, if z1 = x1 + i · y1 , z2 = x2 + i · y2 are two points within into a such strip such
that ez1 = ez2 then,
nx1 +y1 ·i = nx1 +y1 ·i ⇒ |nz1 | = |nz2 | ⇒ nx1 ny1 ·i  = nx2 ny2 ·i  ⇒ x1 = x2
   

and since x1 = x2 , the relation ez1 ·ln(n) = ez2 ·ln(n) gives ei·y1 ·ln(n) = ei·y2 ·ln(n) , so y1 − y2 = 2kπ/ ln(n), is an
integer multiple of 2π. But z1 and z2 belong to the strip, so |y1 − y2 | < 2π/ ln(n). That is, the difference
y1 − y2 is at the same time a multiple of 2π/ ln(n) and at an absolute value of less than 2π/ ln(n). The
only case that this is true is when y1 = y2 . We finally conclude that z1 = z2 , so n2 is 1-1 in the strip
{z : 2kπ/ ln(n) ≤ Im(z) < 2(k + 1)π/ ln(n)}. ”lines down closed – open up”. We also notice that nz is on
C − {0}. Because if w 6= 0 and we put z = ln |w|/ ln(n) + iargw/ ln(n) then,
nz = nln |w|+iargw/ ln(n) = nln |w| niargw/ ln(n) = |w| · niargw/ ln(n) = w

Figure 1: The z1 and z2 belong to the lane width 2π/ ln(n)

Formation of the strips 1-1 for ζ(s) and ζ(1 − s).

a1.) If we accept the non-trivial zeroes on critical strip of the Riemann Zeta Function ζ(s) as s1 = σ1 + it1
and s2 = σ2 + it2 with |s2 | > |s1 |, and if we suppose that the real coordinates σ1 , σ2 of each non-trivial
zero of the Riemann Zeta, [1, 2, 3, 4] function ζ(s) correspond to two imaginary coordinates t1 and t2 , then,
we have the following equations group:
1 1 1 1
ζ(σ1 + i · t1 ) = + σ1 +i·t1 + σ1 +i·t1 + ... + σ1 +i·t1 + ... = 0
1σ1 +i·t1 2 3 n
1 1 1 1
ζ(σ2 + i · t2 ) = σ2 +i·t2 + σ2 +i·t2 + σ2 +i·t2 + ... + σ2 +i·t2 + ... = 0
1 2 3 n

Taking the first equation and deducting the second, we obtain:

∞  
X 1 1
ζ(σ1 + i · t1 ) − ζ(σ2 + i · t2 ) = − =
n=1
nσ1 +i·t1 nσ2 +i·t2
∞ ∞
X nσ2 +i·t2 − nσ1 +i·t1 X nσ2 ei·t2 ln(n) − nσ1 ei·t1 ln(n)
= =0
n=1
nσ2 +i·t2 · nσ1 +i·t1 n=1
nσ2 +i·t2 · nσ1 +i·t1

From the previous relation, we conclude that if

2kπ
nσ2 +i·t2 · nσ1 +i·t1 6= 0, n > 1 then σ2 = σ1 ∧ t2 = t1 ± , (k = 1, 2, ...).
ln(n)

2
That is to say t1 and t2 can take any value, but according to the previous relation. So from (Fig.1), that
means ζ(s) is 1-1 on the lane of critical Strip in Intervals such as defined, therefore and on the critical Line.
So n−z is 1-1 in the strip and {z : 2kπ/ ln(n) ≤ Im(z) < 2(k + 1)π/ ln(n)}.”lines down closed – open up”.

a2.) If we do the same work with |s2 | > |s1 | for the case ζ(1 − s) will we have:
∞  
X 1 1
ζ(1 − σ1 − i · t1 ) − ζ(1 − σ2 − i · t2 ) = − =
n=1
n1−σ1 −i·t1 n1−σ2 −i·t2

∞ ∞
X n1−σ2 −i·t2 − n1−σ1 −i·t1 X n1−σ2 e−i·t2 ln(n) − n1−σ1 e−i·t1 ln(n)
= =0
n=1
n1−σ2 −i·t2 · n1−σ1 −i·t1 n=1
n1−σ2 −i·t2 · n1−σ1 −i·t1
From the previous relation, we see that if

n1−σ2 −i·t2 · n1−σ1 −i·t1 6= 0, n > 1

2kπ
&({1 − σ2 = 1 − σ1 } ∨ σ2 = σ1 ) ∧ t2 = t1 ± , k = (1, 2, ...)
ln(n)
So, from (Fig.1), it follows that ζ(1 − s) is 1-1 on the critical Strip, So n−(1−z) is 1-1 in the strip and
{z : 2kπ/ ln(n) ≤ Im(z) ≤ 2(k + 1)π/ ln(n)}.”lines down closed – open up”

As we have seen before, the complex exponential form nz but and n−z is also periodic with a period
of 2π/ ln(n). For example nz+2·π·i = nz · n2·π·i/ log(n) = nz is repeated in all the horizontal strips with
2π/ ln(n) on (Fig.2) below.

Figure 2: All the ”bottom - closed upper “horizontal strips with 2π/ ln(n) for n−z , n ≥ 1, n ∈ Z.

Pn 1
If we are asked to find the strips that are 1-1 of the function ζ(z) = z
it will be the union of the
i=1 n
strips formed with period 2πk/ ln(n) where k, n ∈ Z. This results from the analysis we made for the cases
a1, a2. Therefore, if we analytically assume that n = 2, 3, ... then the exponential complex function will be
1-1 in each of the strips, i.e. in each one set of the form le

Sn = {z : 2kπ/ ln(n) ≤ Im(z) < 2(k + 1)π/ ln(n)}, k ∈ Z

∞
S
and therefore in each one subset of the unified expanded set S = Si . The largest strip obviously contains
i=2
all the rest it is wide, and has width 2π/ ln 2 i.e. with n = 2.

3
I.2. Overlapping contiguous Strips.
If we start from the upper bound of the strip with n = 2 and k = 1, Fig.3 the upper ones will descend to
zero by increasing n with a width difference the interval δ = 2π/ ln(n − 1) − 2π/ ln(n). In addition, we will
prove that the strip i = n ≥ 2 primary (k=1) that generated in ascending order, the upper limit of that
created of its double width enters the zone of the previous of (i = n − 1, k = 1) also primary. With this
logic we can find for which n we have overlapping strips in relation to the increase of n on the imaginary
axis. Below we find for which n this dimension applies which is useful to distinguish them but also to make
them visible schematically.

Figure 3: The upper ones will go down to zero by increasing n

√
3+ 5
Apply 2π/ ln(n − 1) < 2 · 2 · π/ ln n ⇒ n < (n − 1)2 ⇒ n2 − 3 · n + 1 > 0 ⇒ n > and because we
2
talked about integers then n ≥ 3. The same phenomenon is also created at higher level strips.

n
X 1
I.3. The ζ(z) = z
it is directly 1-1 from the comparison of 2 points z1 , z2 .
i=1
n

Proof:
From the equality
n n
X 1 X 1
ζ(z1 ) = z
= z2
= ζ(z2 ) ⇒ 1 + 2−(x1 +y1 ·i) + 3−(x1 +y1 ·i) + ... + n−(x1 +y1 ·i) = ζ(z2 ) =
i=1
n 1
i=1
n

= 1 + 2−(x2 +y2 ·i) + 3−(x2 +y2 ·i) + ... + n−(x2 +y2 ·i) ⇒
with comparison of similar terms the result arises.

2·π·k
x1 = x2 ∧ y1 = y2 − ,k ∈ Z
ln(n)

After to the analysis of cases before, will be arises 1-1 as shown in sections {I1 .a1 , I1 .a2 } pages 1& 2. Also
according to Lagrange’s generalized theory, each of the roots according to the obvious relation ζ(z) = ζ(1−z)
will result from the generalized theorem of Lagrange. From the equation that results from 3 Functional
Equations, with condition of common roots. We take ζ(z) = ζ(1 − z) and for first approach we then follow
the 3 first terms on each ζ-equation we will have in the Regular form ζ(z), ζ(1 − z) with the analysis below.

4
 1 1
1−z + 2−z + 3−z = 1−(1−z) + 2−(1−z) + 3−(1−z) ⇒ 2−z + 3−z = 2z + 3z ⇒
 z 2 3
3 2
2−z + 3−z − 2−(1−z) − 3−(1−z) = 0 ⇒ 6z + 4z − 3 − 3 =0
2 3
With the replacement
Log(y) + 2 · k · π · i
6z = y ⇒ z · Log(6) = Log(y) + 2 · k · π · i ⇒ z =
Log(6)

Log(y) + 2 · k · π · i
zin =
Log(6)
we consider it as initial value for the solution of Transcendental equation with the method of Lagrange
inversion, then:
  w 
∞ w w−1 
  Log(y) + 2 · k · π · i   Log(y) + 2 · k · π · i
Log(y) + 2 · kπ · i X (−1) d 1  3 Log(6) 2 Log(6)  
z= + 
w−1  y · Log(6)
·
2 · 4 −3·   (IL)
Log(6) w=1
Γ(w + 1) dy 3  
y→3

Some solutions for k ∈ Z (Table 1)

1st 0.4985184869351 -/+ 3.6072014300530 I, k=1

2nd 0.4997077392995 -/+ 10.329023684711 I, k=3
3rd 0.5003426560276 -/+ 13.990255078243 I, k=4
4th 0.4993087825728 -/+ 17.585326315022 I, k=5
5th 0.5026747758005 -/+ 21.304192633721 I, k=6

Table 1: 6 Roots of equation ζ(z) = ζ(1 − z)

And so on for the sequence of infinity. Now we see that we approach the real roots and its complexities.
With the additional method Newton or the Bisection method we approach more rapidly to them. And these
roots will be contained in corresponding strips that we have previously defined and in closer relation with
their position on the critical line.
By this logic for the previous approximation equation in relation to its solutions of, ζ(z) = 0 the strips
which determine the imaginary roots of the equation, will be between the intervals 2 · π · k/ ln 6, k ∈ Z as
defined by the analysis of the generalized Theorem of Lagrange. The strips of the imaginary part of the
roots, as they appear for 6 consecutive intervals of the approximate equation with method of Lagrange that
we have solved above and we will have the results on the (Table 2).

Strips – Imaginary Roots:

1st (7.013424977055179 3.506712488527589)
2nd (10.52013746558276 7.013424977055179
3rd (14.02684995411035 10.52013746558276)
4th (17.53356244263794 14.02684995411035)
5th (21.04027493116553 17.53356244263794)
6th (24.54698741969312 21.04027493116553)

Table 2: The strips of the imaginary part of the roots

It is clear the first root 3, 607..i is located in 1st of strip (7.01 − 3.506), the second 10.3290..i in the 2st strip
(7.013 − 10.52) etc. Now the root 13.99..i which approximates the first root of ζ(z) = ζ(1 − z) = 0 i.e. the
14.1347..i, is in the 4st strip (14.02 − 17.533) with a lower limit the 14.026..i a value that is very close to
the one required for the approximate and exclusive (1st root) when ζ(z) = 0 of the Riemann Hypothesis.

5
Corollary 1.
The only roots of the Zeta function not included in the set {z ∈ C : 0 ≤ Re(z) ≤ 1} are the points
−2, −4, −6.. [11. p47]

Proof
On the functional equation:
ζ(1 − z) = 2 · ζ(z) · Γ(z) · cos(π/2 · z) · (2π)−z we know that for Re(z) > 1 the functions ζ(z) and Γ(z) do
not equal zero (Proof-Th.3.p6). We also find that for Re(z) > 1 the 1 − Re(z) < 0 and putting u = 1 − z
we will find all the roots of ζ(u) for Re(u) < 0. Therefore we will have that ζ(u) will be zero, where it is
zeroed the i.e. z = 3, 5, 7. Then the roots of ζ(u) for Re(u) < 0 will be the points −2, −4, −6... and will be
all the roots of the function ζ out of the strip {z ∈ C : 0 ≤ Re(z) ≤ 1}.

Theorem-Helpful 1
The Riemann’s Z-function has no roots on the lines Re(z) = 1 and Re(z) = 0.
The proof is detailed in the book [11. p50-51].

II. The common roots of the equations ζ(s) − ζ(1 − s) = 0 they have Re(s) = Re(1 − s) = 1/2
within the interval (0, 1) if moreover apply ζ(s) = 0. (Refer to Theorem 3, p.8)

Proof: Let us assume z to be such that for complex z = x0 + iy0 , 0 < Re(z) < 1 ∧ Im(z) 6= 0 and
ζ(z) = ζ(1 − z) = 0. According to the two equations, they must apply to both, because they are equal to
zero that: ζ(z) = ζ(z) = 0 ∧ ζ(1 − z) = ζ(1 − z) = 0. But from [Theorem 1, II, a1, a2] the ζ(z) and ζ(1 − z)
are 1-1 on the critical Strip. If suppose generally that ζ(x0 ± y0 · i) = ζ(1 − x00 ± y00 · i) = 0 then:

y0 = y00 ∧ x0 = 1 − x00 ,

but because are the 1-1 then we will apply two cases for complex z i.e.
1
x00 = 1 − x0 , x0 6= x00 , y0 = y00 ∨ x00 = 1 − x0 , x0 = x00 = , y0 = y00
2
{Corollary 1, Theorem Helpful 1}, because we suppose ζ(z) = ζ1 − z = 0.] We conclude that 0 < x0 < 1.
Therefore we have three cases:

1
II.1. If x0 = x00 = we apply the obvious i.e. ζ (1/2 ± y0 · i) = ζ (1 − 1/2 ± y00 · i) = ζ (1/2 ± y00 · i) = 0∧
2
∧y0 = y00 . Which fully meets the requirements of hypothesis!

1
II.2. If 0 < x0 6= x00 6= < 1 in this case if x0 < x00 ∧ x0 + x00 = 1 or x0 > x00 ∧ x0 + x00 = 1 then for the
2
Functional equation ζ(s) − ζ(1 − s) = 0 simultaneously apply:

ζ (x0 + y0 · i) = ζ (x0 − y0 · i) = 0 ∧ ζ (x00 + y00 · i) = ζ (x00 − y00 i) = 0.

So let’s assume that:

1 1
1st . x0 < x00 ∧ x0 + x00 = 1 ⇒ x0 = − a ∧ x00 = + b, a 6= b. But apply x0 + x00 = 1 ⇒ 1 + b − a = 1 ⇒
2 2
1
b = a. That is, a, b is symmetrical about of . But the ζ(z) and ζ(1 − z) are 1 - 1 on the critical Strip
2
then apply the relation:
   
1 1 1 1
ζ − a ± y0 · i = ζ + a ± y0 · i ∧ y0 = y00 ⇒ must − a = + a ⇒ 2a = 0 ⇒ a = 0
0
2 2 2 2
therefore x0 = x00

6
 1 1
2nd . x0 > x00 ∧ x0 + x00 = 1 ⇒ x0 = + a ∧ x00 = − a, as before. Then because the ζ(z) and ζ(1 − z) are
2 2
1-1 on the critical Strip will apply:
   
1 1
ζ + a ± y0 · i = ζ − a ± y0 · i ∧ y0 = y00
0
2 2

1 1
that + a = − a ⇒ 2a = 0 ⇒ a = 0 therefore x0 = x00 . We see that in all acceptable cases is true that
2 2
”The non-trivial zeros of ζ(z) − ζ(1 − z) = 0 have real part which is equal to 1/2 within the interval (0, 1).
In the end, we see still one partial case that appears in the roots of the equation ζ(z) −ζ(1 − z) = 0.

II.3. If 1 < x0 ∧ x00 < 0 furthermore, in this case with that is, it is symmetrical about x0 + x00 = 1 then
for the functional equations we apply:

ζ (x0 − y0 · i) = ζ (1 − x0 − y0 · i) ∧ ζ (x0 + y0 · i) = ζ (1 − x0 + y0 i) ∧ y0 = y00

and for the three cases it is valid y0 = 0 ∨ y0 6= 0. This case applies only when ζ(z) = ζ(1 − z) 6= 0 ∧ x0 =
= 1 + a, x00 = −a, and not for the equality of ζ function with zero. But in order to have common roots of
the two functions ζ(z) and ζ(1 − z), because they are equal to zero and equal to each other, then first of
all, it must be i.e x0 = x00 = 1 − x0 and also because the ζ(z) is 1-1 and for each case applies (y0 = 0 and
1
y00 = 0) or (y0 6= 0 and y00 6= 0), therefore we will apply x0 = x00 = ∧ y0 = y00 and with regard to the three
2
cases {II.1, II.2, II.3} the cases {II.2, II.3} cannot happen, because they will have to be within (0, 1) and
will therefore be rejected.

Therefore if ζ(z) = ζ(1 − z) = 0 then because ”these two equations of zeta function are 1-1 on the lane of
critical strip” as shown in theorem 1, is ”sufficient condition” that all non - trivial zeros are on the critical
line Re(z) = 1/2. That means that the real part of z of ζ(z) = 0, equals to 1/2”. So Theorem 1 & Corollary
1, they have been proved and also helps Theorem 3 below.

2. Theorem 2
For the non- trivial zeroes of the Riemann Zeta Function ζ(s) apply

There exists an upper-lower

 bound of Re(s) of the Riemann Zeta Function ζ(s) and more specifically
ln 2 ln π
in the closed interval , . The non-trivial zeroes of the Riemann Zeta function ζ(s) of the upper-
ln 2π ln 2π
lower bound are distributed symmetrically on the straight line Re(s) = 1/2. The average value of the upper
lower bound of Re(s) = 1/2.

Proof:
Here, we formulate two of the functional equations from E - q. Set

ζ(1 − s)/ζ(s) = 2(2π)−s cos(πs/2)Γ(s), Re(s) > 0

s−1
ζ(s)/ζ(1 − s) = 2(2π) sin(πs/2)Γ(1 − s), Re(s) < 1

We look at each one equation individually in order to identify the set of values that we want each time.

2.1. For the first equation and for real values with Re(s) > 0 and by taking the logarithm of two sides of
the equation [5], we have:
ζ(1 − s)/ζ(s) = 2(2π)−s cos(π/2)Γ(s) ⇒

7
 log[ζ(1 − s)/ζ(s)] = log[2] − s log[2π] + log[cos(πs/2)Γ(s)] + 2kπi

but solving for s and if f (s) = cos(πs/2)Γ(s)

 and from Lemma 2 [6, p.6]:
log[2] log[f (s)] + 2kπi
if lim ζ(1 − s)/ζ(s) = 1 or log lim ζ(1 − s)/ζ(s) = 0 we get s = + with
s→s0 s→s0 log[2π] log[2π]
 
log[f (s)] + 2kπi log[2]
Re ≥ 0, finally because we need real s we will have Re(s) ≥ = 0.3771 . This is
log[2π] log[2π]
the lower bound, which gives us the first ζ(s) of Riemann’s Zeta Function.

2.2. For the second equation, for real values if Re(s) < 1 and by taking the logarithm of the two parts of
the equation, we will have:

ζ(s)/ζ(1 − s) = 2(2π)s−1 sin(πs/2)Γ(1 − s) ⇒

log[ζ(s)/ζ(1 − s)] = log[2] + (s − 1) log[2π] + log[sin(π · s/2)Γ(1 − s)] + 2kπi

but solving for s and if f (s) = sin(πs/2)Γ(1 − s) and from Lemma 2[6, p.6] if lim [ζ(s)/ζ(1 − s)] = 1 or
   s→s0 
log[π] log[f (s)] + 2kπi log[f (s)] + 2kπi
log lim ζ(s)/ζ(1 − s) = 0 we get s = − with Re >= 0.
s→s0 log[2π] log[2π] log[2π]
log[π]
In the following, because we need real s we will take. Re(s) ≤ = 0.6228 . This is the upper bound,
log[2π]
which gives us the second of Riemann’s Zeta Function of ζ(s). So, we see that the lower and the upper
bound exist for Re(s) and they are well defined.

log[2] log[π]
2.3. Assuming that sk = σlow + itk and s0s = σupper + itk with σlow = and σupper = we
log[2π] log[2π]
apply σ1 = σlow and σ2 = σupper , (Fig.4). If we evaluate the difference and

Re ∆0κ = Re s0κ − 1/2 = σupper − 1/2 = 0.1228

Re ∆sκ = 1/2 − Re sκ = 1/2 − σlow = 0.1228.

This suggests for our absolute symmetry of

Figure 4: Arrangement of low and upper of real part of zeros from 2 functional equations of Riemann.

8
2.4. The average value of the upper lower bound is Re(s) = 1/2 because from (Figure 4): that is ap-
ply
log[2] + log[π]
Re(s) = 1/2 · = 1/2
log[2π]

3. Theorem 3
The Riemann Hypothesis states that all the non-trivial zeros of ζ(z) have real part equal to 1/2.

Proof: In any case, we Assume that: The Constant Hypothesis ζ(z) = 0, z ∈ C. In this case, we use
the two equations of the Riemann zeta function, so if they apply what they represent the ζ(z) and ζ(1 − z)
to equality. Before developing the method, we make the three following assumptions:

3.1. Analysis of specific parts of transcendental equations, which are detailed

3.1.1. For z ∈ C, {az = 0 ⇒ a = 0 ∧ Rez > 0} which refers to the inherent function similar to two Rie-
˜
mann zeta functions as (2π)−z = 0 or (2π)−1 = 0 and it seems that they do not have roots in C − Z, because
(2π) 6= 0.

3.1.2. This forms Γ(z) = 0 or Γ(1 − z) = 0 do not have roots in C − Z.

3.1.3. Solution of sin(π/2z) = 0 or cos(π/2z) = 0. More specifically if z = x + yi, then.

1. sin(π/2z) = 0 ⇒ cosh[(π · y)/2] sin[(π · x)/2] + I · cos[(π · x)/2] sinh[(π · y)/2] = 0

If k, m ∈ N (Integers) then all the solutions can be found with a program by the language of Mathematica.

i) {x = (2 + 4 · k), y = 2 · i · (1 + 2 · m}
ii) {x = 4 · k, y = 2 · i · (1 + 2 · m}
iii) {x = (−1 + 4 · k), y = i · (−1 + 4 · m}
iv) {x = (−1 + 4 · k), y = i · (1 + 4 · m}
v) {x = (1 + 4 · k), y = i · (−1 + 4 · m}
vi) {x = (1 + 4 · k), y = i · (1 + 4 · m}
vii) {x = (2 + 4 · k), y = i · (4 · m}
viii) {x = (2 + 4 · k), y = 2 · i · (1 + 2 · m}
From the generalized solution, it seems that in the pairs (x, y) will always arise integers which make impos-
sible the case 0 < x < 1, therefore there are no roots of the equation sin(πz/2) = 0 in C − Z.
Also, because as we see, all the roots given by the union of the sets x ≥ 1 ∨ x = −1 ∨ x = 0 all are Integers.
So it cannot be true that for our case:

cos(π/2z) = 0 ⇒ cosh[(πy)/2] cos[(πx)/2] − I sin[(π · x)/2] sinh[(πy)/2] = 0

If k, m ∈ N (Integers) and the solutions are:

i) {x = 4 · k, y = i · (−1 + 4 · m)}
ii) {x = 4 · k, y = i · (1 + 4 · m}
iii) {x = (−1 + 4 · k), y = i · (4 · m}
iv) {x = (−1 + 4 · k), y = 2 · i · (1 + 2 · m}
v) {x = (1 + 4 · k), y = 4 · i · m}
vi) {x = (1 + 4 · k), y = i · (1 + 4 · m}
vii) {x = (2 + 4 · k), y = i · (−1 + 4 · m}
viii) {x = (2 + 4 · k), y = i · (1 + 4 · m}

9
As we can see, again from the generalized solution, it seems that in the pairs (x, y) will always arise inte-
gers, which make impossible the case is 0 < x < 1, therefore there are not roots receivers of the equation
cos(πz/2) = 0. Since all the roots are given by the union of the sets x ≥ 1 ∨ x = −1 ∨ x = 0 it follows that
they are Integers.

Great Result
 With this three - cases analysis, we have proved that the real part of x, of the complex z = x + yi cannot
be in the interval 0 < x < 1, and in particular in cases {3.1.1, 3.1.2, 3.1.3}, when they are zeroed. Therefore
they cannot represent roots in the critical line.

3.2. Therefore, now we will analyze the two equations of the Riemann zeta function and we will try
to find any common solutions.

3.2.1. For the first equation and for real values with Re(z) > 0 we apply:
Where f (z) = 2(2π)−2 cos(π · z/2)Γ(z) ⇒ ζ(1 − z) = f (z) · ζ(z) but this means that the following two cases
occur:

3.2.1.1. If ζ(1 − z) = ζ(z) , where z is complex number.

This assumption implies that ζ(z) · (1 − f (z)) = 0 ⇒ ζ(z) = 0. In theorem 1.(I , II), (pages 1 − 8),
we showed that the functions ζ(z) and ζ(1 − z) are 1-1 and therefore if ζ (x0 + y0 i) = ζ (x00 + y00 i) then,
but we also apply that ζ (x0 + y0 i) = ζ (x0 − y0 i) = 0, because we apply it for complex roots. The form
ζ(1 − z) = ζ(z) means that if z = x0 + y0 i we apply 1 − x0 = x0 namely x0 = 1/2 because in this case
it will be verified that ζ (1 − x0 − y00 i) = ζ (1/2 − y00 i) = ζ (1/2 + y00 i) = ζ (1 − x0 + y00 i) = 0 and because
y0 = y00 ⇒ ζ (x0 − y0 i) = ζ (1/2 − y0 i) = ζ (1/2 + y0 i) = ζ (x0 + y0 i) = 0 and these are the two forms of
the ζ equation, they have common roots and therefore it can be verified by the definition of any complex
equation, when it is equal to zero. Therefore if z = x0 + y0 i then x0 = 1/2 which verifies the equation
ζ(z) = 0.

3.2.1.2. If ζ(1 − z) 6= ζ(z) , when z is a complex number.

To verify this case must be:
ζ(z) = ζ(1 − z) · f (z) ⇒ ζ(z) = 0 ∧ f (z) = 0.
But this case is not possible, because as we have shown in Section 3.1 (3.1.1, 3.1.2, 3.1.3), the individual
functions of f (z) cannot be zero when z is a complex number.

3.2.2. For the second equation and for any real values with Re(z) < 1 we apply:
ζ(z)/ζ(1−z) = 2(2π)z−1 sin(π·z/2)Γ(1−z) ⇒ ζ(z) = f (z)·ζ(1−z), where f (z) = 2(2π)z−1 sin(π·z/2)Γ(1−z)
but this also means that two cases occur:

3.2.2.1. If ζ(1 − z) = ζ(z) , when z is a complex number.

This case is equivalent to 3.2.1.1, and therefore if z = x0 + y0 i then x0 = 1/2 in order to verify the equation
ζ(z) = 0 we follow exactly the same process algebraically.

3.2.2.2. If ζ(1 − z) 6= ζ(z) , where z is a complex number. Similarly, the above case is equivalent
to 3.2.1.2 and therefore it cannot be happening, as it has been proved.

4. Forms of the Riemann ζ Functional Equations

The Riemann ζ function has three types of zeros: [7]

ζ(1 − s) = 2 · (2 · π)−S · cos(π · s/2) · Γ(s) · ζ(s),


Res > 0
Eq. Set (I, II)
ζ(s) = 2 · (2 · π)s−1 · sin(π · s/2) · Γ(1 − s) · ζ(1 − s), Res < 1

10
And

1−s
 
Γ
2
ζ(s) = π s−1/2 ·   ζ(1 − s) (III)
s
Γ
2

Usually referred to as the trivial zeros, and non-trivial complex zeros. Therefore has been proved that any
non-trivial zero lies in the open strip {s ∈ C : 0 < Re(s) < 1} that is called the critical Riemann
Strip: And all complex zeros of the function ζ lie in the line {s ∈ C : Re(s) = 1/2} which is called the
critical line.

Solving of the Riemann ζ Functional-Equations.

To find the imaginary part we must solve the functional equations (Eq. set), and for the cases which
the real part of the roots lies on the critical line Cases 4.1, 4.2 [6].

4.1. 1st type roots of the Riemann zeta functions (1st equation from the Eq. set)
For the first category roots and by taking the logarithm of two sides of the equations, and thus we get [5]:

ζ(1 − z)/ζ(z) = 2(2π)−x cos(π · z/2)Γ(z) ⇒

⇒ log[ζ(1 − z)/ζ(z)] = log[2] − z log[2π] + log[cos(π · z/2)Γ(z)] + 2kπi
and the total form from the theory of Lagrange inversion theorem, [5] for the root is p1 (z) = s which means
that f (s) = p−1
1 (s) = z, but with an initial value for s which is

log[2] 2·k·π·i
sin → + ,
log[2π] log(2π)

by setting values for k we can simply calculate the roots of the (1st equation from the Eq. set, (Table.3)).

1st 0.377145562795 -/+

2nd 0.377145562795 -/+ 3.41871903296 I
3rd 0.3771455627955 -/+ 6.83743806592 I
4th 0.3771455627955 -/+ 10.2561570988 I
5th 0.3771455627955 -/+ 13.6748761318 I
6th 0.3771455627955 -/+ 17.0935951646 I

Table 3: Solution of 1st equation from the Eq. set, The firsts 6 Roots.

A simple program in mathematica is:

k ≥ 4; t = (log[2] + 2kπi)/ log[2π];

∞
((−1/ log[2π])w /Gamma[w + 1]) + D [(s0 ) · log[cos[π · s/2]]+
P
z =s+
w=1
+ log[Gamma[s]] − log[Zetta[1 − s]/Zeta[s]])w , {s, w − 1}];
N [z/.s → t, 10]

With k = 0, ±1, ±2, . . .

11
But because the infinite sum approaching zero, theoretically s gets the initial value

log[2] 2·k·π·i
z→ +
log[2π] log(2π)
So we have in this case, in part, the consecutive intervals with k = n and k = n + 1 for any n ≥ 4 and for
the imaginary roots.

4.2. 2nd type of roots of the Riemann zeta functional equations (2nd equation from the Eq. set)

Same as in the first category roots by taking the logarithm of two sides of the equations, and thus we
get:
ζ(z)/ζ(1 − z) = 2(2π)z−1 sin(π · z/2)Γ(1 − z) ⇒
⇒ log[ζ(z)/ζ(1 − z)] = log[2] + (z − 1) log[2π] + log[sin(π · z/2)Γ(1 − z)] + 2 · k · π · i

Now, we will have for total roots of 3 groups fields (but I have interest for the first group), and therefore
for our case we will get p1 (z) = s which means that:

f (s) = p−1
1 (s) = z,

but with an initial value for s that is

log[π] 2·k·π·i
sin → +
log[2π] log(2π)

the overall form from the Lagrange inverse theory succeeds after replacing the above initial value and
therefore for the first six roots after calculating them we quote the following (Table.4).

1st 0.622854437204 -/+

2nd 0.622854437204 -/+ 3.41871903296 I
3rd 0.622854437204 -/+ 6.83743806592 I
4th 0.622854437204 -/+ 10.2561570988 I
5th 0.622854437204 -/+ 13.6748761318 I
6th 0.622854437204 -/+ 17.0935951648 I

Table 4: The firsts 6 Roots of 2nd equation from the Eq. set

A simple program in mathematica is:

k ≥ 4; t = (log[π] + 2kπi)/ log[2π];

∞
((−1/ log[2π])w /Gamma[w + 1]) + D [(s0 ) · log[sin[π · s/2]]+
P
z =s+
w=1
+ log[Gamma[1 − s]] − log[Zetta[s]/Zeta[1 − s]])w , {s, w − 1}];
N [z/.s → t, 10]

Therefore, with k = 0, ±1, ±2, . . .

But because the infinite sum approaches zero, theoretically s gets initial value

log[π] 2·k·π·i
z→ +
log[2π] log(2π)

12
So we have in this case, in part, the consecutive intervals with k = n and k = n + 1 for any n ≥ 4 and for
the imaginary roots. And for the cases we have for Im(s) the relationship
2·π·k
Im(z) = ,k ∈ Z
log[2π]
For the 3rd functional equation, one has been previously put on the (as in the other two in 4, page 10) and
we take as imaginary part
1 2·π·k
Re(z) = ∧ Im(z) = ,k ∈ Z
2 log[π]
Following to complete the roots of the two sets of the functional equations Eq. Set, we solve the functions
as cosine or sin according to its Generalized theorem of Lagrange [5].

4.3. Transcendental equations for zeros of the function (Explicit formula)

The main new results presented in the next few sections are transcendental equations satisfied by indi-
vidual zeros of some L-functions. For simplicity, we first consider the Riemann-function, which is the
simplest Dirichlet L-function.

*Asymptotic equation satisfied by the n-th zero on the critical line. [9, 10, 11]

As above, let us define the function x(z) ≡ π −z/2 Γ(z/2)ζ(z)√which satisfies the functional


equation x(z) =
x(1 − z). Now consider the Stirling’s approximation Γ(z) = 2πz z−1/2 e−z 1 + O z −1 where z = x + iy,
which is valid for large y. Under this condition, we also have
  πx  πy

z z = exp i y log y + + x log y − + x + O y −1 .
2 2
Therefore, using the polar representation ζ = |ζ|ei arg ζ and the above expansions, we can write as

√  y (x−1)/2
2ππ −x/2 e−πy/4 |ζ(x + iy)| 1 + O z −1



A(x, y) = ,
2
y  y  π
+ (x − 1) + arg ζ(x + iy) + O y −1 .


θ(x, y) = log
2 2πe 4

The final transactions we end up with

 
y  y  5 1 1
n= log − + lim arg ζ + δ + iy .
2π 2πe 8 δ→0+ π 2
Establishing the convention that zeros are labeled by positive integers, zn = 1/2+i·yn where n = 1, 2, 3, 4, . . .,
we must replace n → n−2. Therefore, the imaginary parts of these zeros satisfy the transcendental equation
   
tn tn 1 1 11
Eq. a log + lim+ arg ζ + δ + itn = n − .
2π 2πe δ→0 π 2 8

Let us recall the definition used in, namely:

    
1 1 1 1
S(y) = lim arg ζ + δ + iy = lim = log ζ + δ + iy .
δ→0+ π 2 δ→0+ π 2
These points are easy to find, since they do not depend on the uctuating S(y).

We have:    
1 1
ζ − iy =ζ + iy G(y), G(y) = e2iθ(y)
2 2

13
The Riemann-Siegel # function is defined by
√
 
1 i
ϑ ≡ arg Γ + y − y log π.
4 2
(+) (−)
Since the real and imaginary parts are not both zero, at yn then G = 1, whereas at yn then G = −1.
Thus
  
1 (+)

(+)

= ζ + iyn = 0 for ϑ yn = (n − 1)π,
2
  
1 (−)

(−)

< ζ + iyn = 0 for ϑ yn = (n − 1)π.
2
they can be written in the form of Eq.b.

2π(n − 7/8) 2π(n − 3/8)

Eq. b yn(+) = , yn(−) =
W [e−1 (n − 7/8)] W [e−1 (n − 3/8)]

(+)
where above n = 1, 2, . . . and the Labert - Function W denotes the principal branch W (0). The yn
are actually the Gram points. From the previous relation, we can see that these points (Fig. 5) are ordered
in a regular manner [8, 12]
(+) (−) (+) (−) (+) (−)
y1 < y1 < y2 < y2 < y3 < y3 < ···

Figure 5: Gram points

With this method, we are able to create intervals so as to approximate the correct values of imaginary
part of non-trivial zeros. We did something similar in the cases 4.2. What it has left, is to find a method
that approximates the values of imaginary part accurately.

14
 Figure 6: The black dots represent the zeros of ζ(s) function including
possible zeros which do not lie on the critical line.

The zeros of the Riemann Zeta function.

“PROGRAMMING”

5.1. The M Function – Bisection Method

2·π·k
Knowing the time of the successive steps (k, k + 1) of the relationship of imaginary parts Im(z) = ,
log[2π]
2·π·k
Im(z) = (Eq I, II, III, page 10) with k ∈ N , we can calculate the roots by solving the equation
log[π]
ζ(1/2 + i · y) = 0 using the Bisection Method.

Bisection is the division of a given curve, figure, or interval into two equal parts (halves). A simple
bisection procedure for iteratively converging on a solution, which is known to lie inside some interval [a, b]
a+b
proceeds by evaluating the function in question at the midpoint of the original interval y = and
2
testing to see in which of the subintervals [a, (a + b)/2] or [(a + b)/2], b] the solution lies. The procedure is
then repeated with the new interval as often as needed to locate the solution to the desired accuracy. Let
an , bn be the endpoints at the nth iteration (with a1 = a and b1 = b ) and let rn be the nth approximate
solution. Then, the number of iterations required to obtain an error smaller than ε is found by noting
b−a 1
that... bn − an = n−1 and that rn is defined by rn = (an + bn ). In order for the error to be smaller
2 2
1
than ε, then... |rn − r| ≤ (bn − an ) = 2−n (b − a) < ε. Taking the natural logarithm of both sides gives
2
−n ln 2 < ln ε − ln(b − a). Therefore, we have for steps:

ln(b − a) − ln ε
n> .
ln 2

5.2. M-function of the Bisection Method..

We define the functions Md+ , Md− on an interval (a, b) according to the scheme:

15
I. Md+ , with Md−1 the Nearest larger of Md−m where
Md−1 + Md−m
, d ≥ 2, d > m ≥ d − 2,
2
a+b
M1 = ,k = 1
2
II. M−
d , with Md−1 the Nearest smaller of Md−m where

Md−1 + Md−m
, d ≥ 2, d > m ≥ d − 2,
2
a+b
M1 = ,k = 1
2
For calculating the roots on solving the equation ζ(1/2 + i · y) = 0 take the limit rS according to the scheme:

rs = lim Mn+ = lim Mn− , s ∈ N with Md+ and Md− ,

n−∞ n−∞

 
2·π·k 2 · π · (k + 1)
belong in the interval (a, b), d ∈ N and also a= , b= , k ∈ N, k ≥ 3 for the x of
log[2π] log[2π]
ζ (1/2 + i · rs ) = 0.

5.3 Program in Mathematica for the Bisection method of ζ(1/2 + i · y) = 0.

 
2·π·k 2 · π · (k + 1)
Using the Intervals a = , b= and successive steps, we can compute all the roots
log[2π] log[2π]
of ζ(1/2 + i · y) = 0. We can of course use three types such intervals more specifically in generall than

2·π·k 2 · π · (k + 1)
1.a = ,b = or
log[2π] log[2π]
2·π·k 2 · π · (k + 1)
2.a = ,b = or
log[π] log[π]

3.a = yn− , b = yn+ .

2π(n − 7/8) 2π(n − 3/8)

yn(−) = −1
, yn(−) =
W [e (n − 7/8)] W [e−1 (n − 3/8)]
We always prefer an interval that is shorter, in order to locate fewer non trivial zeros. The most important
is to calculate all the roots in each successive
 interval and therefore
 only then we will have the program
8·π 10 · π
for data example: Integer k = 4, and a = , b= and Error approximate tol = 10−6 and
log[2π] log[2π]
Trials n = 22..

“Programm for Bisection method”

A program relevant by dividing intervals..
This program gives very good values as an approach to the roots we ask if we know the interval. Selecting
 
2·π·k 2 · π · (k + 1)
the interval for the case a = , b= , k ∈ N, k ≥ 3 the results are given below in
log[2π] log[2π]
(Table.5)..

16
 1
Calculation 1st Zetazero z1 = + 14.1347216..
2

Table 5: The first non-trivial Root, Imaginary Part.

With final value 14.1347216500, error near of 10−7 . This value is the approximate root of the by the nearest
error < 10−7 .
In the event that we have two or more uncommon roots within the interval, we divide similar successive
intervals in the order of finding of the first root either above or below. In such a case we have the k = 13,
and at the interval (47.8620664, 51.280785) the two roots are 48.0051088 and 49.7738324.

5.4 Explicit formula and the Zeros of ζ(1/2 + i · y) = 0.

Consider its leading order approximation, or equivalently

 its average since < arg ζ(1/2 + i · y) >≥ 0 [11, 12].

tn tn 11
Then we have the transcendental equation log = n− . Through the transformation tn =
2π 2πe 8

17
    
11 11
2π n − x−1
n this equation can be written a xn e xn
=e −1
n− . Comparing the previous results,
8 8
we obtain 
11
2π n −
8
tn =    where n = 1, 2, 3 . . .
11
W e−1 n −
8

5.5 Programm by Newton’s method, which finds the Zeros of ζ(1/2 + i · y) = 0

Using Newton’s method we can reach the roots of the equation ζ(1/2 + i · y) = 0 at a very good ini-
tial value from y → tn by the explicit formula.

It follows a mathematica program for the first 50 roots by the Newton’s method (Table.6). This method
determines and detects the roots at the same time in order to verify the relation ζ(z) = 0 and always
according to the relation...
 
11
2π n −
8
tn =    where n = 1, 2, 3 . . . and where W is the W-function.
11
W e−1 n −
8

Following is the program in mathematica and the Table.6 of complex roots. . .

Table 6: The firsts 50 complex Roots(with W-function)

A very fast method that ends in the root very quickly and very close to the roots of ζ(s) = 0 as shown.

5.6 Directly (from Explicit form) with the solution of this equation [8]

We know that the equation

   
tn tn 1 1 11
log + lim+ arg ζ + δ + itn =n− .
2π 2πe δ→0 π 2 8

18
Using the initial value the relation
 
11
2π n −
8
tn =    the explicit formula..
−1
11
W e n−
8

As already discussed, the function arg(ζ(1/2 + δ + i · y)) oscillates around zero. At a zero it can be well-
defined by the limit, which is generally not zero. For example, for the first Riemann zero y1 = 14.1347 the
limit δ → 0+ has value as..
lim arg ζ(1/2 + δ + i · y) = 0.157873.
The arg ζ term plays an important role and indeed improves the estimate of the nth zero. We can calculate
by Newton’s method, and we locate the first 30 imaginary part roots, of the equation ζ(1/2 + i · y) = 0, on
the bottom of the (Table.7), and where x = Im(1/2 + yi). The process we use here does not cease to be
approximate, although it gives us very good results quickly. The real solution to the problem, that is, the
correct solution of the before transcendental equation, is dealt with most effectively in Chapter 5.7.

Table 7: The firsts 50 roots of ZetaZero.(Arg() Method)

5.7 From Explicit form with the solution of method of Periodic Radicals..

Using the relation Explicit form..

   
tn tn 1 1 11
log + lim+ arg ζ + δ + itn =n− .
2π 2πe δ→0 π 2 8

We divide all the factors of the equation by 1/e and we take the more specialized form, according to the
replace x = tn /(2πe) therefore we have the form

x · log(x) + 1/(π · e) · arg(z(1/2 + I · (2 · π · e · x)) = 1/e · (n − 11/8)

There are 2 ways to solve this equation or find the inverse of the expression x · log(x) or of the expression
arg(z(1/2 + I · (2 · π · e · x))). Respectively the inversions are for each performance f1 (u) = eProductlog(u)
and f2 (u) = Zeta−1 arg−1 (u) . The preferred procedure for finding a solution is the first because it is

simpler, because the second it’s more tricky program. Following is a program in mathematica.

19
This program is based on the reversal of x log x term, the reverse of which is associated with W - Function,
a process that gives us a multiplicity to the Improved programming, which we take advantage in order to
find a plurality of roots. Those with the creation of this small Program we find it quickly and simply, any
set of complex roots we need. As we observe with this process we achieve a better approach and the roots

Table 8: The firsts 50 roots of ZetaZero.(xlog(x) Method)

are not removed from their correct values Perhaps it is the only case of finding complex roots in a large
area and number that we could achieve computationally. Unlike case # 5.6 which is widely used based on
the approximate type,
 
11
2π n −
8
tn =   
−1
11
W e n−
8
from explicit formula we may have discrepancies. That is, what is clearly seen in the calculation of complex
roots of ζ(z) = 0 is the fact that they must treat the equation as a whole and not in parts. Finally, as
we saw in this chapter, we fully tackled the transcendental equation as a whole and formulated without
gaps-free algorithm with a simple process that works perfectly and efficiently.

20
6. 100% of the zeros of ζ(z) = 0 are on the critical line

The elementary in our proof will be broken down into 5 parts.

6.1.1 Using direct method of inverse of ζ() on functional equations.

From the proofs of Theorems 1, 2, 3 we find it necessary to find the inverse of the function φ(z) =
ζ(z) − ζ(1 − z) by the lagrange method. First of all we have to mention why we got this equation now
and we want to examine its inverse. we saw in Theorem 3 since we proved from Th.1 that it is 1 − 1 and
therefore the function φ(z) is also inverted. We saw that the condition for the functional equations to hold
is necessary in the form . . . , φ(z) = 0. But from the fact that φ(z) is 1 − 1 since the partial functions that
make it up are also 1−1. So ζ(z) = ζ(1−z) => z = 1−z ⇒ Re(z) = 1/2. Also a very simple example is the
fact that the series 1−z + 2−z + 3−z + . . . + n−z = 1z−1 + 2z−1 + 3z−1 + . . . + nz−1 it is very simple to prove
that take as solution the unique real z = 1/2 comparing the terms one by one. Therefore, according to
Riemann’s functional equations, we have one critical critical line on which the non-significant
zeros will lie.

6.1.2 Solution of the functional Equation ζ(s), s Complex number

For real values with Re(s) > 0 and with form

 
−x
s
− 1−ν 1−s
π 2 Γ ζ(s) = π Γ
2 ζ(1 − s) ⇒
2 2
1−g


1 Γ
ζ(s) = π s− 2 2

ζ(1 − s)
Γ g2

if we take the logarithm but or z (we are making a change in the variable s), of the two parts of equation
[#4. Eq-set(III)] then we have
   
z− 21 2−1 1−z z
ζ(z)/ζ(1 − z) = π 2(2π) Γ /Γ =>
2 2
      
1 1−z z
log[ζ(z)/ζ(1 − z)] = − z log[π] + log Γ /Γ + 2kπi
2 2 2
The resolution analysis will be performed by the Lagrange inversion theorem, which distinguished the best
approach for transcendental equations. Using the correlation theory and after relations with 1 from 3 groups
fields and therefore for our case we will get:

p1 (z) = s which means that f (s) = p−1

1 (s) = z, but with an initial value

1 2·k·π·i
sin = +
2 log(π)
and total form from theory Lagrange for the root is:

Program
 in mathematica
Needs[”NumericalCalculus”]
k := 1; q := 25;

z = N [sin ]+
q
P w
((−1/ log[π]) / Gamma[w + 1]) ∗
w=1

21
N D[(s0 ) ∗ log[Gamma[1 − s]/2]/ Gamma[s/2]] − log[Zeta[s]/ Zeta[1 − s]])w , {s, w − 1},

t, WorkingPrecision → 250, Scale → .00001]


FQ = N[z/.Sin → t, 25]; N [πFQ − 1/2∗ Gamma [(1−FQ)/2]/ Gamma [FQ/2] - Zeta [FQ]/ Zeta [1 − FQ]]

The results for k = 1 and q = 25 limit indicator of series is 0.5 + 5.48879293259.. and with approxi-
mation 2.22044 ∗ 10−15 + 2.664535 ∗ 10−15 I of equation ζ. Note : The program is in language mathematica
dw−1 0 0
and where ND is ds w−1 and s (is first derivative of s) = 1, also k ∈ N .

Because of its limit summation tens to zero for a large n, we can roots of the functional equation to
write it simply as:
1 2·k·π·i
z→ + ,k ∈ Z
2 log(π)

We see, therefore, that the roots of the functional equation of the ζ, are on the critical line, with Re(s) = 1/2,
a very serious result for its complex roots. Below we see a list with k = 0, 1, 2, ..10 first values which are the
roots with a very good approach of the functional equation of the ζ() and if z a zeros we have solutions:

If we want to see the roots of ζ(z) = 0 ie analytically for the first 12 values:

we can compare them to the previous values that we found of the functional equation of the ζ. Below
is given the graphical representation of ζ- function in C for x ∈ (0, 1) and y ∈ (−31, 31),

22
 Figure 7: Plot of ζ(z) function

 
Zeta(1/2+I·y)
we notice in the diagram Fig.7 that for the blue lines apply. Im Zeta(1/2−I·y) = 0 and for the Red lines
 
Zeta(1/2+I·y)
Re Zeta(1/2−I·y) = 0.

6.1.3 Solutions of the Equation ζ(s)/ζ(1 − s) = 1, s is Complex number.

A functional equation that hides all secrets of the ζ(s) = 0. Based on the analytical extension of the
zeta function we conclude that ζ(1 − s) is analytical at C − {0} and zero will be a simple pole. So we will
have:
lim ζ(1 − z) = ∞
z→0
H (z)
lim − (2π)z = ∞
z→0 2i sin π2 z
ie equality applies to z0 , where

H(z) = lim H (z) = −2i sin(πz)Γ(z)ζ(z).

→0

As for the regard 4 categories roots of the equation F(s) = ζ(s) − ζ(1 − s) = 0 and s = x0 + y0 i we have a
brief summary of the form:
 
1
 x 0 = 2 , ±y0 , if (y 0 6
= 0) 
 * 
 x0 , ±y0 i 1


if x0 6= 2 , y0 6= 0 
 

1 − x0 , ±y0 i
 
 
s=  *


 x0 , 1



 if x0 6= 2 , y0 = 0  

 1 − x0 , 

 
1
x0 = 1 − x0 = 2 , if (y0 = 0)

The equation F(s) is related to the Hypothesis Riemann and helps in the final proof of Theorem 3(pages
8-10) seen before. It is a general form that also contains the non-trivial roots of ζ(s) = 0 and ζ(1 − s) = 0.

23
Below we see a list with 15 first values, which are the roots with a very good approach of the equa-
tion of the F (s) = 0 with method Newton in language Mathematica.

Reduce[Zeta[z]==Zeta[1-z]&& Abs[z]<15,z]

z = {−12.16278 − 2.58046 I}||z = {−12.1627878 + 2.5804649 I}||z = {−7.9909145 − 4.5105941 I}||z =

{−7.99091453+4.51059414 I}||z = {0.5−14.13472514 I}||z = {0.5−9.666908 I}||z = {0.5−3.4362182 I}||z =
{1/2 I}||z = {0.5 + 3.4362182 I}||z = {0.5 + 9.66690805 I}||z = {0.5 + 14.1347251 I}||z = {8.990914533 −
4.510594140 I}||z = {8.990914533 + 4.51059414069 I}||z = {13.1627870 − 2.580464 I}||z = {13.16278786 +
2.58046497 I}

Further, is given the graphical representation of F (s) = ζ(s) − ζ(1 − s) functional equation.

Figure 8: Plot of ζ(z) − ζ(1 − z) function

6.2. With general inverse random function theory.

According to the theory of inversion of functions that help us to solve them as equations, it completely
guides us to a single solution i.e. Re(z) = 1/2 for ζ(z) = 0. In this article, we seek to develop a formula
for an inverse Riemann zeta function w = ζ(s) we have an inverse function s = ζ −1 (w) and whics implies
that ζ −1 (ζ(s)) = s ∧ ζ −1 (ζ(w)) = w for real and complex Domain w and s. The presented method can
also recursively compute these multiple solutions on other branches, but as we will find, the computational
requirements become very high and start exceeding the limitations of the test computer, so we will primarily
focus on the principal solution. We develop a recursive formula for an inverse Riemann zeta function as:
" 1
!#− 2m
2m
 n
1 d X 1
s = ζ −1 (w) = lim m −

log [(ζ(s) − w)(s − 1)] −
m→∞ (2m − 1) ds2m s→0 s2m
k
k=1

Where Sn is the multi-value for which w = ζ(s).

24
6.2.1. Theorem 4. To prove that for the formula [26]

1 dn
− log |f (z)|z→0 = Z(m) − P (m)
(m − 1)! dz m

that valid for a positive integer variable m > 1. This formula generates Z(m) − P(m) over all zeros and
poles in the whole complex plane, as opposed to being enclosed in some contour.

Proof. If we model an analytic function f(z) having simple zeros and poles by admitting a factorization of
the form
N2   YNp  −1
Y z z
f (z) = g(z) · 1− · 1−
n=1
zn n=1
pn

Where g(z) is a component not any zeros or poles, then

Nz   XNp  
X z z
log[f (z)] = log[g(z)] + log 1 − − log 1 −
n=1
zn n=1
pn

From theorem Residue if i take

f 0 (z) 1
Z
1
PΩ dz = Z(s) − P(s)
2iπ f (z) z s

And with use Taylor Theorem and |z| < 1 we take in final
∞ ∞
X zk X zk
log[f(z)] = log[g(z)] − Z(k) + P(k)
k k
k=1 k=1

From this form, we can now extract Z(m) − P(m) by the nth order differentiation as

1 dm
− log |f(z)|z→0 = Z(m) − P(m)
( m − 1)! dz m

6.2.2. Theorem 5. If Zn is a set of positive real numbers ordered such that 0 < z1 < z2 < . . . < zn , and
so on, then the recurrence relation for the nth+1 term is

n
!−1/s
X 1
Zn+1 = lim Z(s) −
s→∞ Zsk
k=1

Proof.

The generalized zeta series over zeros Zn of a function

Nz
X 1 1 1 1
Z(s) = s
= s + s + s + ...
z
n=1 n
z1 z2 z3

and let us also assume that the zeros are positive, real, and ordered from smallest to largest such that
0 < z1 < z2 < . . . < zn then the asymptotic relation holds
1 1 −s
⇒ O zn−s > O zn+1



s
> s , s→∞
zn zn+1

25
Final we take  −1/s
−1/s 1
z1 = lim (Z(s)) , z2 = lim Z(s) − s ,...,
s→∞ s→∞ z1
n
!−1/s
X 1
zn+1 = lim Z(s) − ,
s→∞ zsk
k=1

thus all zeros up to the nth order must be known in order to generate the nth+1 zero. Exist the Voros’s
closed-form formula for depends on (RH), and the formula which we can express the zeta and beta terms
in terms of a Hurwitz zeta function, and then substituting the the Voros’s closed-form formula we obtain
another formula for non-trivial zeros a for in terms of the Zeta .If the limit converges to 1/2, it would imply
(RH) is correct. The formula of Voros’s is especially for real part (of ρn.nt the non-trivial zeros) of
ζ(s) = 0 is
" − 1s
1 2 1
< (ρ1,nt ) =σ1 = lim Z (m) − Znt (2m) +
m→∞ 2 nt 2
− 1s # 12
(−1)m
    
2m 1 (2m) 1 1 5 1
− 2 − log(|ζ|) − 2m ζ 2m, = .
2 (2m − 1)! 2 2 4 2
V.roots

m R(ρ1, nt)
15 0.4730925331369
20 0.4898729067547
25 0.4993065936936
50 0.5000000028549
100 0.4999999999999
150 0.5000000000000 Table 9: Table Voro’s. Approximation of
200 0.4999999999999 Re(z) of z() function

The computation of the real part of the first non-trivial zero for m = 200 places (13 digits) using this
relationship Vroots and with 200 iterations we achieve a very satisfactory result. It is of course obvious that
it approximates by the direct inverse of ζ() , the real part of the non-trivial roots to 1/2.

6.3. Statistical evidence

Here, we formulate two of the functional equations from E-q.Set (Theorem 2, page 7)

ζ(1 − s)/ζ(s) = 2(2π)−s Cos(πs/2)Γ(s), s > 0

ζ(s)/ζ(1 − s) = 2(2π)z−1 Sin(πs/2)Γ(1 − s), Re < 1

We look at each one equation individually in order to identify the set of values that we want each time.
For the first equation and for real values with Re(s) > 0 and by taking the logarithm of two sides of the
equation [5], we have...

log[2]
real s we will have Re(s) ≥ = 0.3771 . This is the lower bound, which gives us the first ζ(s)
log[2π]
of Riemann’s Zeta Function.

For the second equation, for real values if Re(s) < 1 and by taking the logarithm of the two parts of

26
the equation, we will have:

log[π]
In the following, because we need real s we will take Re(s) ≤ = 0.6228
log[2π]

average value of the upper lower bound is Re(s) = 1/2 because from that is apply

log[2] + log[π]
Re(s) = 1/2 · = 1/2
log[2π]

6.4. The critical line also follows from the definition of λ(N )

Will define the function λ(N ) as follows:

λ (N) = 1 + 2−S + 3−S + . . . + N−S

The next step is to perform a mathematical analysis on the functions of λ after allocating N value by
equalizing it with Pn+1 , because the values of λ function in mainly depends on the prime numbers. In
this way, the equations that represent the imaginary part of zeros which we are going to have will be as a
function of the prime numbers. But the equations include all the natural numbers, so we’ll generalize the
solutions which depend on the prime numbers after obtaining all natural numbers with the same values
which approach to those prime numbers.

Now we get the equation...

s(1 − s)p−s
    
n 1 1 1 1
= 1− 1− ··· 1 −
p1−s
n+1 2 3 pn−1 pn

With replacing 1 − s by s and vice versa in the same equation, we get the

s(1 − s)ps−1
    
n 1 1 1 1
= 1 − 1 − · · · 1 −
psn+1 2 3 pn−1 pn

By dividing equations the previous two, we get rid of the two factors s and (1 − s) which cause the
emergence of the classical zeros of eta function

psn+1 p−s
n p2s−1
n+1
= 1 ⇒ 2s−1 − 1 = 0
ps−1
n p1−s
n+1 pn

s−1/2
( s− 12  −(s− 12 ) )
pn+1 pn+1 pn+1
⇒ s−1/2
− =0
pn+1 pn pn

27
And therefore
 P −i P −i 
1 i(s−1/2) ln Pn+1
−i(s−1/2) ln n+1
e n −e Pn
=0
2i
(  −i )
Pn+1
⇒ sin (s − 1/2) ln =0
Pn
 
Pn+1
⇒ −i(s − 1/2) ln = ±kπ, k = 1, 2, 3, . . .
Pn
1 kπ
⇒s= ±i   , k = 1, 2, 3, . . .
2 ln PPn+1
n

(6.4.I)

this last equation (6.4.I) proves the truth of the Hypothesis Riemann.

6.4.1. A very important inequality arises from the 3rd functional equation for Number Theory

From the 3rd Solve of the functional equations equation of ζ(s)(page 10, III) we have for real values
with Re(s) > 0 where s is Complex number and with form
 
x
s 1−x 1−s
π− 2 Γ ζ(s) = π − 2 Γ ζ(1 − s) ⇒
2 2
1−s


s− 21 Γ 2

ζ(s) = π ζ(1 − s)
Γ g2
if we take the logarithm but or z (we are making a change in the variable s), of the two parts of equation
then we have
   
1 1−z z
ζ(z)/ζ(1 − z) = π z− 2 2(2π)z−1 Γ /Γ ⇒
2 2
       
1 1−z z
log(ζ(z)/ζ(1 − z)) = − z log(π) + log Γ /Γ + 2kπi
2 2 2

The resolution analysis will be performed by the Lagrange inversion theorem, which distinguished the best
approach for transcendental equations. Using the correlation theory and after relations with 1 from 3 groups
fields and therefore for our case we will get

p1 (z) = s which means that f(s) = p−1

1 ( s) = z but with an initial value

1 2·k·π·i
sn = +
2 log(π)
And total from Lagrange inversion theorem for the root Z is if q → ∞
q
X
z = sin + ((−1/ log[π])w /Gamma[w + 1])
w=1
dw−1

∗ (log[(Gamma[(1 − s)/2]) ∗ (Gamma[x/2])] − log[(Zeta[x])(Zeta[1 − x])])w
dsw−1
Because of its limit summation tens to zero for a large n, we can roots of the functional equation to write
it simply as:
1 2·k·π·i
z→ + ,k ∈ Z
2 log(π)

28
We see, therefore, that the roots of the functional equation of the ζ, are on the critical line, with Re(s) = 1/2,
a very serious result for its complex roots. Below we see a list with k = 0, 1, 2, ..10 first values, which are
the roots with a very good approach of the functional equation of the ζ and if z we have solutions:
{z == 0.5, z = = 0.5 + 5.4887929325942295t, z == 0.5 + 10.977585865188459i, z =
= 0.5 + 16.46637879778269i, z == 0.5 + 21.955171730376918i, z =
= 0.5 + 27.443964662971148i, z == 0.5 + 32.93275759556538i, z =
= 0.5 + 38.42155052815961i, z == 0.5 + 43.910343460753836i, z =
= 0.5 + 49.399136393348066i, z == 0.5 + 54.887929325942295t}

we can compare them to the previous values that we found of the functional equation of the ζ(z) equation.
But from the second proof the non-trivial zeros z has a lower value than the first one. This means
that that is: z < s(from 6.4.1) that is
√
 
2κπi κπi pn+1 1 pn+1
<   ⇒ log < log(π) ⇒ < π
log(π) p
log n+1 pn 2 pn
pn

Sure it is obvious
pn+1 pn+1 √
>1⇒1< < π (6.4.2.II)
pn pn
√
The ratio of the largest to the smallest consecutive prime has an upper bound of the real number π. A
relation that enters into many prime number conjectures.

6.5. From the inverse of the function ζ()

First we need to formulate the Riemann functional equation. It is necessary to define the function ξ(s)
   
1 −s/2 1 −s/2 1
ξ(s) := s(s − 1)π Γ s ζ(s) = (s − 1)π Γ (s + 1) ζ(s)
2 2 2

Where we have used the identity 21 sΓ 12 s = Γ 21 s + 1 .

We use the following theorem,

Theorem 6. The function ξ has an analytic continuation to C. For continuation we have ξ(1 − s) = ξ(s)
for s ∈ C.

6.5.2. Corollary 2. The function ζ an analytic continuation to C\{1} with a simple pole with Residue 1
at s = 1.

For this continuation we have

 
1
ζ(1 − s) = 21−s π −s cos πs Γ(s)ζ(s) for s ∈ C\{0, 1}
2

Proof. We define the analytic continuation of ζ by

ξ(s)π s/2 Γ 1 1s+1

(2 )
ζ(s) =
s−1

1/Γ is analytic on C and the other function in the numerator are the analytic on C.Hence ζ is analytic on

29
C\{1}. The analytical continuation stated here coincides with that stated by the well-known theory {s in
C : Re s > 0}\{1} since analytic continuations to connected sets are uniquely determined. Hence ζ(s) has
a simple pole with residue 1 at s = 1. We derive the functional equation. By Theorem 6.5.1 we have, for s
in C\{1},

ξ(1 − s) ξ(s)
ζ(1 − s) = 1 1

= 1 1


2 (1 − s)(−s)π −(1−s)/2 Γ 2 (1 − s) 2 s(s − 1)π −(1−s)/2 Γ 2 (1 − s)
− 1)π −s/2 Γ 12 s
1


2 s(s
· ζ(s) = F (s)ζ(s)
= 1 −(1−s)/2 Γ 1 (1 − s)
2 s(s − 1)π 2

Now we have
1 1 1



(1/2)−s Γ 2 s Γ
2 s + 2
F (s) = π · 1 1 1 1


Γ 2 − 2s Γ 2 + 2s
1−s
√
2 πΓ(s)
= π (1/2)−s
π/ sin π 12 − 12 s



 
1
= π −s 21−s cos πs Γ(s).
2

This implies this Corollary.

6.5.3. Direct solution of the equation ζ(q ∗ z) = 0, q ∈ R

Keeping the basic logic of the corollary 6.5.2., declare the definition s 7→ q · z, q ∈ Z who is analytic
on C, for every fixed q ∈ R, q 6= 0. Then is obvious. According Corollary 6.5.2. we have
 
1
ζ(1 − q · z) = 21−q·z π −q·z cos π · q · z Γ(q · z)ζ(q · z) for q · z ∈ C\{0, 1}.
2

But according to theorems 1 & 3 the function ζ() is 1-1 in the critical strip, which means that:

ζ(1 − q · z) = ζ(q · z) and therefore

1
z= , q 6= 0, q ∈ R (a)
2q

This relationship always determines the position of the critical line, and of course atom is directly related
to the ζ() function as we proved in Theorem 3. Let us now define a function ZetaZero() which gives us the
zeroes of ζ(s) = 0 as defined in sections {5.4, 5.5, 5.6, 5.7}.

We therefore have the relation s = ZetaZero(±k), k ∈ N, k ≥ 1. Therefore

s = Re(ZetaZero(k)) ± Im(ZetaZero)(k) · i, i is imaginary unit, k ∈ N+ (b)

As we know that for the simple function ζ(s) = 0 the

1
Re(s) = Re(ZetaZero(k)) = , ∀k ∈ N+
2

30
From relations (a,b) we have the relation for non trivial zeroes,

1 1
zzeroes = · Re(ZetaZero(k)) ± · Im(Zeta Zero(k)) · i ⇔
q q
1 1
zzeroes = ± · Im(ZetaZero (k)) · i,
2·q q
i is imaginary unit, k ∈ N+ , q ∈ R, where q 6= 0. (c)

from the last relation (c) it follows that the Critical line will be given by the relation

1
zc.line = · Re(ZetaZero(k)) ⇔
q
1
zc.line = , ∀k ∈ N+ , q ∈ R, where q 6= 0. (d)
2q

Finaly for the case of trivial zeroes we have the relation,

2·k
zzeroes = − , k ∈ N+ , q ∈ R, where q 6= 0. (e)
q

With this analysis we have essentially achieved the general solution of the transcendental equation ζ(q · z) =
0, q ∈ R, and we observe that for positive values of q, the critical lines, infinite in number, run through the
interval (0, 1) for three cases (1&2&3),

1 1 1
1. 0 < zc.line < then 0 < < therefore q > 1
2 2q 2
1 1 1 1
2. < zc.line < 1 then < < 1 therefore < q < 1
2 2 2q 2
1 1
3. zc.line > 1 then > 1 therefore 0 < q <
2q 2

Similarly for the interval (−1, 0) we will have the relations (4&5&6),

1 1 1
4. < zc.line < 0 then − < < 0 therefore q < −1
2 2 2q
1 1 1 1
5. − 1 < zc.line < − then − 1 < < − therefore − 1 < q < −
2 2q 2 2
1 1
6. zc.line < −1 then < −1 therefore − < q < 0
2q 2

6.5.4. Two Characteristic examples

For the Riemann hypothesis the most important thing we are interested in is the interval (0, 1). The
transcendental equation z (q ∗ z) = 0 as we have shown has infinite critical lines in this interval, which
depend exclusively on ato q.

i) I take for begin the mean value of interval (0, 1/2) and we have:

31
1. zc.line = 41 = 2q 1
⇔q=2
 
 non trivials zeroes 
1
 
zeroes = q ( ZetaZero (±m)), m = 1 ÷ 5
 z 
2.
z = {0.25 + 7.06736I, 0.25 + 10.511I, 0.25 + 12.5054I,
 zeroes

 

0.25 + 15.2124I, 0.25 + 16.4675I}


 
 trivials zeroes 
3. zzeroes = −2 × mq ,m = 1 ÷ 5
zzeroes = {−1., −2., −3., −4., −5.}
 

ii) Also the mean value of interval (1/2, 1) is 3/4. Therefore we get the relations for this case:

1. zc.line = 34 = 2q1
⇔ q = 23
 
 non trivials zeroes 
1
 
zeroes = q ( ZetaZero (±m)), m = 1 ÷ 5
 z 
2.
z = {0.75 + 21.2021I, 0.75 + 31.533I, 0.75 + 37.5163I
 zeroes

 

0.75 + 45.6373I, 0.75 + 49.4016I}


 
 trivials zeroes 
3. zzeroes = −2 × mq , m = 1 ÷ 5
zzeroes = {−3., −6., −9., −12., −15.}
 

6.6. Deniers of the Riemann Hypothesis.

The most characteristic case that can be considered in relation to ζ(z) = 0 is the Davenport-Heilbronn
function (introduced by Titchmarsh), is a linear combination of the two L -functions with a complex char-
acter mod 5, with a functional equation of L -function type but for which the analogue of the Riemann
hypothesis fails. In this lecture, we study the Moebius inversion for functions of this type and show how its
behavior is related to the distribution of zeros in the half-plane of absolute convergence. In general, however,
these cases do not refer to ζ(z) = 0 but to the L -function. The most important element in the negators
of the Riemann hypothesis to consider is whether it is consistent with the Riemann equation ζ(z) = 0.
Therefore, we will not consider cases that do not fall within our hypothesis, as defined by Riemann.

6.6.1. Introduction of the denial conjecture through control one case of ζ(6, z) = 0.

However, there is one interesting case announced by the Enrico Bombieri .In a video by Enrico Bombieri[23]
it is clear that there is a root that belongs to the

ζ s, m2 + 5 · n2 = 0

The root is s = 0.93296997 + 15.668249531∗ I and is the prime for m = n = 1. Thus the equation becomes
ζ(s, 6) = 0 as he mentions. We have to prove that it is not the root of the equation.

True that ζ(s, 6) = −0.101884 − 0.00598394I approximately. In contrast, the ζ(6, z) case gives a value
to zero.

32
In more detail approximately,

ζ(6, z) = 2.93166 · 10−8 + 2.10446 · 10−7 · I

But if we do a simulation with the equation ζ (q ∗ z) = 0 then we need q = 1/(2(Re(z))) = 1/ (2∗ 0.93296997)
i.e q = 0.535923 so we will have another root z = 0.93297 + 26.3745I approximately, for which

ζ(q · z) = −2.46975 · 10−15 + 5.81423 · 10−15 · I

a very good approximation and correct this time. To verify the fact that this root is a false root we will use 2
methods. One is with the Newton method locally and the other is to solve also locally the transcendental
equation ζ(6, z) i.e. near the real root of the complex Re(s) = 0.93296997. If in this zone we do not find
such a value this root will no longer exist.

6.6.2. Approximate proof of the Newton method for checking ζ(6, z).

So assuming that the root is actually s1 = 0.93296997 − 15.668249531∗ I, we set it as the initial root in
FindRoot[Zeta [6, z], {z, s1}, WorkingPrecision → 50] with a dynamic iteration 50 times. We imme-
diately see that it locally diverges to infinity and gives us a value

z → 3.5898 · 107 − 1.29617 · 109 I

This means that it failed near there to approach any root that would zero the equation ζ(6, z). In contrast,
by the same method we observe that the local root s1 = 0.93297 + 26.3745I for the equation ζ (q∗ z) very
quickly locates root i.e. using the command FindRoot[Zeta [q ∗ z] , {z, s1}, WorkingPrecision − > 50]
with q = 0.535923 then z = 0.93296997 + 26.374547818I with approximation

ζ(q · z) ≈ 2.46975 · 10−15 − 1.29617 · 10−15 .

Therefore the root s1=0.93296997-I*15.668249531, failed the first test to confirm that it is a root.

6.6.3. Direct proof of ζ(6, z) = 0.

In mathematics, the polygamma function of order m is a meromorphic function on the complex

numbers C, defined as the (m + 1) th derivative of the logarithm of the gamma function.

It satisfies the Reflection relation:

∂m
(−1)m ψ(m, 1 − z) − ψ(m, z) = π · cot(π · z) (6.6.3.1)
∂zm

The polygamma function has the series representation

∞
X 1
ψ(m, z) = (−1)m+1 m! (6.6.3.2)
(z + k)m+1
k=0

which holds for integer values of m > 0 and any complex z not equal to a negative integer. This represen-
tation can be written more compactly in terms of the Hurwitz zeta function as

ψ(m, z) = (−1)m+1 m!ζ(m + 1, z) (6.6.3.3)

33
Alternately, the Hurwitz zeta can be understood to generalize the polygamma to arbitrary, non-integer
order. If we generally define ζ(m + 1, z) = f(z) we can with improved programming solve this equation based
on the 2 previous equations (6.6.3.1) & (6.6.3.3). In this case we ask to find the roots when ζ(m + 1, z) = 0
therefore f(z) = 0 with method Periodic Radicals.

In this case we are interested in the fact that ζ(6, z) = 0 and therefore ψ(5, z) = 0 from relation (6.6.3.3).
From (6.6.3.1) we have

∂5
(−1)5 ψ(5, 1 − z) − ψ(5, z) = π · cot(π · z) (6.6.3.4)
∂z5

But because Ψ(5, z) = 0 the resulting final relation will be

∂5
−ψ(5, 1 − z) = π · cot(π · z) (6.6.3.5)
∂z 5

Performing the process of finding roots according to the Generalized Root Existence Theorem (of an equa-
tion) [24] we find the complementary function after solving the form

ψ(5, 1 − z) + π 6 · −16 · (cot(πz))4 · (csc(πz))2 − 88 · (cot(πz))2 · (csc(πz))4 − 16 · (csc(πz))6 = 0

(6.6.3.6)

in final form

ψ(5, 1 − z)/π 6 + −16 · (cot(πz))4 · (csc(πz))2 − 88 · (cot(πz))2 · (csc(πz))4 − 16 · (csc(πz))6 = 0

(6.6.3.7)

To solve such a transcendental equation, we will use Periodic Radicals, which means we will split the equa-
tion into 2 parts. We solve the 2 nd part because it’s a trigonometric equation and we have,

−16 · (cot(πz))4 · (csc(πz))2 − 88 · (cot(πz))2 · (csc(πz))4 − 16 · (csc(πz))6 = 0

whose has in general 12 solutions j := ±1; m := 1 ÷ 6; as below,

1 hp i 
z := j ? 2 ArcTan Root [15 + 30#1 + 17#12 + (4 + 8u)#13 + 17#14 + 30#15 + 15#16 &, m] + 2πk
π

A radical relation which gives us the inverse of the 2nd part of equation 6.6.3.7.The whole secret is based
on the parameter k of the trigonometric equation, which in turn will give us multiple different roots, which
solve our basic equation ζ(6, z) = 0. The total cases we consider are 2X6=12, and of these we keep the
values where 0 < Re(z) < 1. With these zeros we make the following table of the program below.

34
Then the matrix program in mathematica witn infinity Periodic Radicals is,

The final table of results is given below,

Table 9: Zeros of equation ζ(6, z) = 0

As we observe Re(z) < 0 and k < 0, for all roots. If we solve the equation for k > 0 we obtain exponential
roots, i.e. large roots, were simply omitted. Therefore there is no solution for Re(z) in the interval (0, 1) as
a critical line.

6.6.4. Direct proof of Epstein zeta equations ζ(Q(m, n), s) = 0 and the contradiction of the
deniers of R-H.

In 1935 Potter and Titchmarsh for equation ζ(Q, z) proved has infinite zeros on the critical line and they
also reported that a numerical calculation had shown that in the case, the Epstein function zeta has a zero
on the critical strip but not on the critical line. This put an end to the speculation that the functional
equation was the reason for the prevalence of a Riemann hypothesis. The deniers state ”It is widely accepted
that the generalized Riemann hypothesis is valid, with strong numerical proofs supporting it, but the proof
remains long elusive.” The Epstein zeta functions have the form of a double sum

35
 n
X 1  2 2 2

ζ(λ, s) = s , λ (p1 , p2 ) = a · p1 + b · p1 p2 + c · p2 , D = b − 4 · a · c , {p1 , p2 , a, b, c ∈ Z+}
λ (p1 , p2 )
p1 ,p2 6=0

To complete the functions related to the ζ-function we have to deal with the ζ(4, z), ζ(5, z) which the
deniers mention. We consider the sum:

n
X 1
ζ(λ, s) = s
p1 ,p2
(p21 + p22 · λ2 )

where the sum over the integers p1 and p2 runs over all integer pairs, apart from (0, 0), as indicated by the
superscript prime. The quantity Q corresponds to the period ratio of the rectangular lattice, and s is an
arbitrary complex number. For Q2 an integer, this is an Epstein zeta function, but for Q2 non-integer we
will refer to it as a lattice sum over the rectangular lattice.
∞
!
s−1/2+n
X p
K(n, m; s; λ) = π n 2
s−1/2−n
Ks−1/2+m (2πp1 p2 λ) .
p1 ,p2 =1 p1

For λ >= 1 and the (possibly complex) numbers small in magnitude, such sums converge rapidly, facili-
tating numerical evaluations. (The sum gives accurate answers as soon as the argument of the MacDonald
function exceeds the modulus of its order by a factor of 1.3 or so.) The double sums satisfy the following
symmetry relation, obtained by interchanging p1 and p2 in the definition:

K(n, −m; s; λ) = K(n, m; 1 − s; λ).

Finally we get the relation for [25]

n
X 1 2 2 2
S0 (λ, s) = ζ(Q, s) = s , Q = p1 + p2 · λ
p1 ,p2
(p21 + p22 × λ2 )
√ Γ(s − 1/2)ζ(2s − 1) 8π s
 
2ζ(2s) 1
S0 (λ, s) = +2 π + K 0, 0; s; .
λ2s Γ(s)λ Γ(s)λs+1/2 λ

From the general relations below we will derive the final relations of interest for the function z(Q, s) :
√
S0 (1, s) = 4ζ(s)L−4 (s), S0 ( 2, s) = 2ζ(s)L−8 (s)
√ √
S0 ( 3, s) = 2 1 − 21−2s ζ(s)L−3 (s), S0 ( 4, s) = 2 1 − 2−s + 21−2s ζ(s)L−4 (s)



√ √
S0 ( 5, s) = ζ(s)L−20 (s) + L−4 (s)L+5 (s), S0 ( 6, s) = ζ(s)L−24 (s) + L−3 (s)L+8 (s)
√
S0 ( 7, s) = 2 1 − 21−s + 21−2s ζ(s)L−7 (s).

From the√zeroing√of the first

√ part of the transcendental representations for the cases where the values for λ
are λ = 3, λ = 4, λ = 7 and will we take:
√
 
1 (2n + 1)πi
S0 ( 3, s) : s = 1+
2 ln 2
√
√ 1 i arctan 7 2nπi
S0 ( 4, s) : s = ± +
2 ln 2 ln 2
√ 1 iπ 2nπi
S0 ( 7, s) : s = + +
2 4 ln 2 ln 2

36
We note that the critical line 1/2 is technically constructed by solving the representations for these cases
and, of course, a number of infinite imaginary roots on the critical line. And in the case of the Epstein
Function we have points on the critical line. The only way to refute any negation for the z(Q, z) functions is
to solve completely the 3 transcendental equations ζ(4, z) = 0, ζ(5, z) = 0, ζ(8, z) = 0 and check the results
if any cases belong on the critical line.

Solving the other cases of Epstein functional equation approximately with the Periodic Radicals method.

If we assume that

Q(m, n) = am2 + bmn + cn2

= a(m + nτ )(m + nτ̄ )
= a|m + nτ |2
∞
X 1
Z(s) = s |m + nτ |2s
, σ > 1.
m,n=−∞
a

in addition, the following relationships apply

√
b+ b2 − 4ac
τ=
2a
√ √
b D b+i D
x= 2a , y= 2a , τ = x + iy = 2a

b
τ + τ̄ = a

c
τ τ̄ = a

separating the term with n = 0 and we have

∞ ∞ ∞
2 X 1 2 X X 1
Z(s) = s 2s
+ ,σ > 1 (6.6.4.I)
a m=1 m a n=1 m=−∞ |m + nτ |2s
s

which is done
∞
√ √ ∞
X 1 Γ s − 12 π 4 π X 1

2s
= 2s−1 y 2s−1 Γ(s)
+ 2s−1 y 2s−1 Γ(s)
(mnπy)s− 2 cos(2mnπx)Ks− 12 (2mnπy)
m=−∞
|m + nτ | n n m=1

with use the Bessel function

 ν  Z ∞
1 2 1 cos yt 1
Kν (y) = √ Γ ν+ ν+ 1
dt, y > 0, <(v) > − ,
π y 2 0 (1 + t2 ) 2 2

and finally we get the equation for the Epstein function in the form

√ 1 ∞ ∞
−s −s 1−2s Γ s − 12 π 8a−s y 2 −s π s X 1−2s X 1
Z(s) = 2a ζ(2s) + 2a y ζ(2s − 1) + n (mn)s− 2 cos(2mnπx)Ks− 21 (2mnπy)
Γ(s) Γ(s) n=1 m=1

(6.6.4.II)

37
It also turns out that is valid

√ !s √ !1−s
D D
Γ(s)Z(s) = Γ(1 − s)Z(1 − s) (6.6.4.III)
2π 2π

which is the functional equation of the Epstein function. But as we proved [Theorem 1, 2, 3 pages 1-10] with
the function ζ( ) that it is 1 − 1, the same is proved here by (6.6.4.3), which follows because Z(s) = Z(1 − s)
i.e. s = 1−s => s = 1/2 for the Real part of the variable s. But we can also prove it from equation (6.6.4.II)
if we consider the third part of the sum with a Bessel function as a small loss in the final result and because it
is always imaginary, keeping the first 2 parts and obtain equation (6.6.4.IV) and work on it to find the roots.

√
−s −s 1−2s Γ s − 21 π
Z(s) = 2a ζ(2s) + 2a y ζ(2s − 1) (6.6.4.IV)
Γ(s)

Thus we secure the real part which will show us the critical line and indeed uniquely. With the method
we follow [24] (Generalized Theorem -Method Periodic Radicals) we find the generator (inverse function)
of a part of the equation and solve the equation as a whole. For this equation a good generator is the term
y1−2s . The analysis that will result after logarithm will give us multiple different roots, which is exactly
what we have in such a case. Below we give the relationship of the generator,

y(1−2x) = u ⇒ y = log(u) + 2 · k · i · π
  (6.6.4.V)
1 log(u)+2·k·i·π
x= 2 − 2·log(y)

Then the matrix program in mathematica witn infinity Periodic Radicals is,

By running the program, all the roots with real part equal 1/2 are obtained. Here is a table with partial
roots for λ2 = 2, 3, 4, 5, 6, 7, 8, 9.

Let’s look in detail at the actual zeros per case in Table 17.

38
 Table 10: Zeros of equation Z(z) = 0 for λ2 = 2, 3, 4, 5, 6, 7, 8, 9

6.6.5. Direct proof of ζ(4, z) = 0.

The whole procedure follows from chapter 6.3 and equation 6.3.3.1 It satisfies the Reflection relation:

∂n
(−1)n ψ(n, 1 − z) − ψ(n, z) = π · cot(π · z) (6.7.I)
∂z n

this equation can be written making the first part positive as

∂n
(−1)n+1 ψ(n, 1 − z) + ψ(n, z) = −π · cot(π · z) (6.7.II)
∂zn

But since we associate the polygamma function and Hurwitz zeta function arising

ψ(n, z) = (−1)n+1 n!ζ(n + 1, z) (6.7.III)

As a complementary relation we take the n -th derivative of cot(πz) which will be

∂n
cot(π · z) = π 4 4 · cot(πz)2 csc(πz)2 + 2 csc(πz)4


−π · n
(6.7.IV)
∂z

We come therefore to the final system

(−1)n+1 ψ(n, 1 − z) + ψ(n, z) = π 4 4 · cot(πz)2 csc(πz)2 + 2 csc(πz)4
ψ(n, z) = (−1)n+1 n!ζ(n + 1, z) (6.7.V)
ζ(n + 1, z) = 0

39
Which takes the final form


(−1)n+1 ψ(n, 1 − z) = π 4 4 · cot(πz)2 csc(πz)2 + 2 csc(πz)4
ψ(n, z) = (−1)n+1 n!ζ(n + 1, z) = 0 (6.7.VI)
ζ(n + 1, z) = 0

The whole philosophy is therefore that if we admit that we zero ζ(n + 1, z) then automatically ψ(n, z)
will also be zero. If we know a method that can solve 6.7.II we can also solve ζ(n + 1, z) = 0. The method
we follow here is the Infinite Periodic Radicals, which is considered here the fastest and safest, because
we have high polynomial powers resulting from the multiple derivation of cot(πz) of the equation (6.7.II).
In our case n = 3 to solve ζ(4, z) = 0.

We therefore have the following program,

Table 11: Zeros of equation ζ(4, z) = 0

40
As we can see the program has 2 variables j = +/ − 1 and m = 1 ÷ 4 and gives us 8 cases that have been
listed in the table 11, in detail for 5 consecutive roots. Moreover, as we can see, there is no critical line for
these roots (because Re(z) < 0), in the interval (0, 1) but not in the interval greater than 1. Also the cases
of transcendental equations ζ(2, z) = 0, ζ(3, z) = 0, ζ(7, z) = 0 do not have the real part of complex zeros in
the interval (0, 1) have been checked.

6.6.6. The zeros of equations ζ(8, z) = 0, ζ(5, z) = 0, ζ(3, z) = 0, ζ(2, z) = 0 and the tables
of these.

Based on the same theory and method of chapter 6.5 we calculate the zeros around the interval (0, 1)
below zero and above unity. If the values are not in this interval for the values of Re(z) then it means that
there are no zeros in the interval (0, 1) and therefore there are no critical lines other than 1/2. From the
given examination we see that there are all points outside the interval and per pair of points of conjugate
complexes. They have all in the interval (−1, 0) from 1 pair upwards as q grows in the form ζ(q, z) that we
consider. This means that as q increases in value, the probability of finding a critical line in the interval
(0, 1) is zero. The process of developing complex pairs is given on page 42 and they show a peculiarity
around −1/2 and −1.

I. Zeros of ζ(8, z) = 0.

Table 12: Zeros of equation ζ(8, z) = 0

II. Zeros of ζ(5, z) = 0.

Table 13: Zeros of equation ζ(5, z) = 0

41
III. Zeros of ζ(3, z) = 0.

Table 14: Zeros of equation ζ(3, z) = 0

IV. Zeros of ζ(2, z) = 0.

Table 15: Zeros of equation ζ(2, z) = 0

Preceded by the given tables with zeros near the interval (0, 1) laterally in order of magnitude. The
colored zeros are those closest to the interval (0, 1); notice immediately that they are divided into 3 cat-
egories. Those where Re(z) is around −1/2 and those where Re(z) is close to −1. Following the logic of
the matrices we construct the following Conclusion, which is general for the class of functional equations
z(q, z), with q > 1 in Z+.

6.6.7. Corollary 3.

The zeros of the Generalized Riemann Zeta function ζ(q, z) = 0, q > 1 in Z+, do not be-
long to the critical strip of the interval (0, 1), but are outside.

According to the Generalized root finding theorem [24] and the Infinite Periodic Radicals method, by
which we solved the whole series of equations with q = 2, 3, 4, 5, 6, 8 we observe that the roots, are divided
into 3 categories, 2 in those where is around −1/2 and which approach it as q increases, and another one
which moves away from −1/2 towards −1 as q increases. Of course we are looking at these roots which are
their real parts in the interval (0, 1).

We could also consider the case for q in R in general, and such a brief examination will be done in section
6.7 where we will consider the generalized equation G − ζ(z, q), i.e. we will also see some instances of ζ(q, z)
with positive and negative q. Thus we will have a complete picture of the functions ζ(q, z) and G − ζ(z, q)
which are closely related to Riemann’s ζ(z).

Let’s look in detail at the actual zeros per case in Table 16.

42
 Table 16: Zeros table for q = {2, 3, 4, 5, 6, 8}

Suppose we have the zeros

n o n o n o n o
zq11 , zq21 , zq21 , zq12 , zq21 , zq21 , . . . , zq1k , zq2k , zq3k , . . . , zq1in , zq2in , zq3in ,
i1 i2 i3 i1 i2 i3 i1 i2 i3 1 2 3

where {κ = 1 ÷ n, i1 , i2 , i3 ∈ N }

with ascending sequence of q, such that for the real parts of zeros apply the following

   
 
= − 21 − ε1qk
1
 
Re z

 
  1 
qik εqk → 0 Re zqk → − 12
1
   
i1

 1

   
 

i1
 i1 

   
 
 
 
 

 2 1 2
  ε2k → 0
   
Re zqk = − 2 + εqk
    
qi
 i2  i2 ⇔ 2
2 ∧ Re zq2k → − 21

 
 
 εq k →0 
 
  i2  

3 3
Re zqk = −1 + εqk  i3

 
 
 
  
  
κ >> 2  Re zq3k → −1
   
i3 i3

 
  

i3
 
κ, n ∈ N, κ = 1 ÷ n
 

Therefore all zeros they have a negative real part of the complex zero, which shows that no zero is in
the critical Strip of the interval (0, 1). So the Generalized Riemann Zeta equation ζ(q, z) = 0, q > 1
in Z+ , is outside of the critical line, but also has no critical line in the interval (0,1). For the Table 16,
are used the functions Hurwitzzeta[q, z] or Zeta(q, z) to check the zeros with the program mathematica 12,
clearly with more decimals to achieve approximations above 10−10 .

The whole procedure for values of q > 1, clearly shows that the function ζ(q, z) = 0 does not generate
zeros even just in the interval (0, 1) which is the only critical line for the function ζ(z) = 0. Of course this
does not exclude the possibility for other values of q, maybe it gives us some pairs or entire critical lines.
This research will be completed below.

43
6.6.8. The zeros of equations ζ(−3, z) = 0, ζ(−2, z) = 0, ζ(−2/3, z) = 0, ζ(−1/2, z) = 0,
ζ(−1/4, z) = 0, and of equations ζ(1/4, z) = 0, ζ(1/2, z) = 0, ζ(2/3, z) = 0 and the tables of
these.

To be able to close the important issue of the zeros of ζ(q, z) = 0 and the positions of the values of
Re(z) within the intervals (0, 1), (−1, 0)&(−∞, −1) we need to solve several forms of transcendental equa-
tions and derive some characteristic matrices. The only method that has this possibility is the Periodic
Radicals method, the same method that we solved the previous forms. the zeros we find will show the
selective positions that are found and to which of these three intervals the Re(z) of each equation belongs.

I. Control if Re(z) of zeros in interval (0, 1)

I.1. Bernoulli polynomials

Here there are 2 ways for integer values. Both the Infinity Periodic Radicals method and the Bernoulli
polynomials. In this case we will prefer the Bernoulli polynomials. The Bernoulli polynomials are a gen-
eralization of the Bernoulli numbers. They have a variety of interesting properties, and will feature in our
proof of the Euler-Maclaurin Summation Formula.
∞
xexy X Bk (y)xk
=
ex − 1 k!
k=0

The generating function for the Bernoulli polynomials is the generating function for the Bernoulli numbers
is multiplied by a term of exy . Our first observation of the Bernoulli polynomials is that the constant term
of Bk (y) is in fact B. If we set y = 0, then Bk (0) = Bk .

xexy x
x
= x
e −1 e −1
Proposition. The Bernoulli polynomials, Bk (y) satisfy the recurrence relation
k  
X k
Bk (y) = Bn y k−n
n
n=0

∞
X Bk (y)xk xexy
=
k! ex − 1
k=0
x
= x · exy
e −1
∞ ∞
X Bk xk X (xy)k
= ·
k! k!
k=0 k=0

We find the Cauchy product by the same process as earlier to obtain

∞ ∞ Xk
X Bk (y)xk X (xy)k−n Bn xn
= ·
k! n=0
(k − n)! n!
k=0 k=0
∞ X k
X y k−n Bk k
= x
(k − n)!n!
k=0 n=0
∞ X k 
xk

X k
= y k−n Bn
n k!
k=0 n=0

44
Now, if we compare terms of x in the right and left-hand summations, we see the following:
k  
X k
Bk (y) = Bn y k−n
n
n=0

as desired Using this recurrence, we can calculate the rst few Bernoulli polynomials:

B0 (y) = 1
1
B1 (y) = y −
2
1
B2 (y) = y 2 − y +
6
3 1
B3 (y) = y 3 − y 2 + y
2 2
1
B4 (y) = y 4 − 2y 3 + y 2 −
30
5 5 1
B5 (y) = y 5 − y 4 + y 3 − y
2 3 6
5 1 1
B6 (y) = y 6 − 3y 5 + y 4 − y 2 +
2 2 42

Notice our earlier result that the constant term of each polynomial is a Bernoulli number. Let us consider
one of these polynomials, say B5 (y), more closely. What if we differentiate it, or integrate it over 0 to 1?

d 1
B5 (y) = 5y 4 − 10y 3 + 5y 2 −
dy 6
 
1
= 5 y 4 − 2y 3 + y 2 −
30
= 5B4 (y)

so we observe that in this case,

Bk0 (y) = kBk−1 (y)

We also see 1
Z 1
1 6 1 5 5 1 2 
B5 (y) = y − y + y4 − y 
0 6 2 12 12 0
=0

These two observations are, remarkably, true in general. In fact, they give us the following inductive defi-
nition for the Bernoulli numbers.

More generally about Bernoulli polynomials apply:

1. B0 (y) = 1
2. Bk0 (y) = kBk−1 (y);
R1
3. 0 Bk (y)dy = 0 for k ≥ 1

45
I.2. Zeros of ζ(−3, z) = 0 and ζ(−2, z) = 0.

In special functions, the following applies to the Zeta function

B(q + 1, z)
ζ(−q, z) = −
q+1

Therefore in these 2 cases we will have:

B(3, z)
ζ(−2, z) = − =0⇒
4
⇒ 1/3 (−(z/2) + (3z ∧ 2) /2 − z ∧ 3) = 0
1
the zeros are : z = 0kz = kz = 1
2

And for the case ζ(−3, z) = 0 we have:

B(4, z)
ζ(−3, z) = − =0⇒
4
⇒ 1/4 1/30 − z 2 + 2z 3 − z 4 = 0

the zeros are:

 s r  s r
1 2 1 2
z= 1− 1−2 kz = 1+ 1−2 k
2 15 2 15
 s r  s r
1 2 1 2
z= 1− 1+2 kz = 1+ 1+2 .
2 15 2 15

ie
z = 0.240335kz = 0.759665k
z = −0.157704kz = 1.1577

Solving other cases we come to the conclusion that we have 2 zeroes only real in the interval (0, 1). Therefore
we have no critical lines in the interval (0, 1).

I.3. Zeros of ζ(−2/3, z) = 0, ζ(−1/2, z) = 0, ζ(−1/4, z) = 0

This case can only be solved with Infinity Periodic Radicals with use Generalized Theorem[24]. The matrix
of the program that generates all solutions is the following:

46
 The matrix program for ζ(−2/3, z) = 0, ζ(−1/2, z) = 0, ζ(−1/4, z) = 0 in mathematica with infinity
Periodic Radicals is:

1. The zeros of ζ(−2/3, z) = 0 are:

z1 = 0.703226180818014
z2 = 0.112695242215194

2. The zeros of ζ(−1/2, z) = 0 is:

z1 = 0.066489124132138

3. The zeros of ζ(−1/4, z) = 0 is:

z1 = 0.582245789118785

As we observe in all three cases we have only real roots although they are located in the interval (0, 1) they
obviously neither create nor belong to any critical line.

I.4. Zeros of ζ(1/4, z) = 0, ζ(1/2, z) = 0, ζ(2/3, z) = 0

Using the same program matrix we find all 3 zeros respectively. The matrix of the program that gen-
erates all solutions is the following:

47
1. The zeros of ζ(1/4, z) = 0 are:

z1 = 0.408064210108089

2. The zeros of ζ(1/2, z) = 0 is:

z1 = 0.302721828598366

3. The zeros of ζ(2/3, z) = 0 is:

z1 = 0.221547787091544

Therefore for all cases with q < 0 and q in (0, 1) while there are zeros in the set of real numbers in the
interval (0, 1) there are no critical lines in this interval.

6.6.9. Dirichlet L-Series and their relationship to the critical line.

I. Dirichlet L-series is a series of the form [27]

∞
X
Lk (s, χ) ≡ χκ (n) · n−s
n=0

where the number theoretic character χk (n) is an integer function with period k, are called Dirichlet L-series.
These series are very important in additive number theory (they were used, for instance, to prove Dirichlet’s
theorem), and have a close connection with modular forms. Dirichlet L-series can be written as sums of

48
Lerch transcendents with z a power of e2πi/k . Dirichlet L-series is implemented in the Wolfram Language as
DirichletL [k, j, s] for the Dirichlet character χ(n) with modulus k and index j. The generalized Riemann
hypothesis conjectures that neither the Riemann zeta function nor any Dirichlet L-series has a zero with
real part larger than 1/2.

The Dirichlet lambda function also is,

∞
X 1
= 1 − 2−s ζ(s)


λ(s) ≡ s
n=0
(2n + 1)

The function Dirichlet (k, j, s), gives the Dirichlet L-function L(χ, s) for the Dirichlet character χ(n) with
modulus k and index j.

Example(with mathematica):

FindRoot [Dirichlet [8, 3, t] == 0, {t, 14I}], Find Root[Dirichlet[8, 3, t] == 0, {t, 15I}]

Out : {t → 0.5 + 12.9881i}, {t → 0.5 + 16.3426i}

We therefore always have as a critical line 1/2.

II. The Dirichlet beta function is defined by the sum

∞
(−1)n
 
X
−s 1
β(s) ≡ = 2 Φ −1, s,
n=0
(2n + 1)s 2

where Φ(z, s, a) is the Lerch transcendent. The beta function can be written in terms of the Hurwitz zeta
function ζ(x, a) by
    
1 1 3
β(s) ≡ s ζ x, − ζ s,
4 4 4

The beta function can be defined over the whole complex plane using analytic continuation,
 z  
2 1
β(1 − z) = sin · π · z · Γ(z) · β(z)
π 2

where Γ(z) is the gamma function.

Example(with mathematica):

Find Root[DirichletLBeta[s] == 0, {t, 14I}]

Out : {t → 0.5 + 12.9881i}

We therefore always have as a critical line 1/2.

49
III. Dirichlet Eta Function

The Dirichlet eta function is the function η(s) defined [28] by

∞
X (−1)n−1
= 1 − 21−s · ζ(s)


η(s) ≡
n=1
(n)s

where ζ(s) is the Riemann zeta function. Note that Borwein and Borwein (1987, p. 289) use the nota-
tion α(s) instead of η(s). The function is also known as the alternating zeta function and denoted ζ ∗ (s),
η(0) = 1/2 is defined by setting s = 0 in the right-hand side of (2), while η(1) = ln 2 (sometimes called
the alternating harmonic series) is defined using the left-hand side. The function vanishes at each zero of
1 − 21−s except s = 1. The eta function is related to the Riemann zeta function and Dirichlet lambda
function by

ζ(v) λ(v) η(v)

v
= v = v
2 2 −1 2 −2

And ζ(v) + η(v) = 2λ(v). The eta function is also a special case of the polylogarithm function,

η(x) = − Lix (−1)

The derivative of the eta function is given by

η 0 (x) = 21−x (ln 2)ζ(x) + 1 − 21−x ζ 0 (x)

Therefore, the natural logarithm of 2 is

∞
X (−1)n−1
η(1) ≡ = ln(2)
n=1
(n)1

Example(with mathematica):

FindRoot [DirichletLEeta [s] = =0,{t,14I}],

Out : {t → 0.5 + 14.1347i}

We therefore always have as a critical line 1/2.

IV. Ramanujan z-Diriehlet series

In this paper we consider the zeros of the Dirichlet series

∞
X τ (n)
L(s) = (σ = Re s > 13/2)
n=1
ns

formed with Ramanujan’s tau-function which may be defined by

∞
X ∞
Y 24
τ (n)z n = z (1 − z n ) (|z| < 1)
n=1 n=1

50
It is known that L(s) has all of its non-real zeros in the strip 11/2 < σ < 13/2 and it is conjectured that all
these zeros are on σ = 6 and are simple. It has been shown (by Hafner [29]) that a positive proportion of
the zeros of L(s) are of odd multiplicity and are on σ = 6, but it is unknown whether any of the zeros are
simple. Some other Dirichlet series are known to have infinitely many simple zeros. Heath-Brown [30] and
Selberg independently observed that the work [31] of Levinson in 1974 implies that a positive proportion of
the zeros of the Riemann zeta function are simple. Of course, the same result works for a Dirichlet L-function.

Example(with mathematica):

FindRoot[RamanujanTauL[s]==0,{t,5+9I}],
Out : {t → 6 + 9.22238i}

We therefore always have as a critical line in 6.

V. Zeros of the Davenport-Heilbronn counterexample

We compute zeros off the critical line of a Dirichlet series considered by H. Davenport and H. Heil-
bronn. This computation is accomplished by deforming a Dirichlet series with a set of known zeros into the
Davenport- Heilbronn series.

For s = σ + i t with σ > 1, let

ξ ξ 1 0
f1 (s) = 1 + − s − s + s + ··· (1)
2s 3 4 5

be a Dirichlet series with periodic coefficients of period 5. Then f1 (s) defines an entire function satisfying
the following functional equation
1
f (s) = T −s+ 2 χ2 (s)f (1 − s) (2)

with T = 5 and
 πs 
χ2 (s) = 2(2π)s−1 Γ(1 − s) cos (3)
2

In 1936, H. Davenport and H. Heilbronn (see [32]) showed that f1 (s), as defined in (1), has zeros off the
critical line σ = 1/2. In 1994, R. Spira (see [33]) computed the following zeros of the Davenport-Heilbronn
example:

0.808517 + 85.699348i, 0.650830 + 114.163343i

0.574356 + 166.479306i, 0.724258 + 176.702461i
In this note we present a scheme for computing additional zeros of the Davenport- Heilbronn Dirichlet
series. We certainly see isolated cases of pairs of complexes that individually define critical lines. We have
seen this before in other cases with ζ-functions and corresponding functional equations.

”It is not a rare phenomenon but here we have to clarify which equations the Riemann
hypothesis accepts as real cases as true cases that define the structure of the given hypothe-
sis”

51
6.6.10. Solve the generalized equation G − ζ(z, q) = 0 of Riemann.

To get a general picture of all forms of ζ() functions we have to deal with the general case which we
will call G − ζ(z, q). If we define a complex z = σ + τ i and D is the set of real σ, it will converge absolutely
for σ > 1. For the value s = 1 it will diverge. So it will have 2 poles s = 1 and where the denominator is
zeroed. The form of the sum will now be,

∞
X 1 α+β
G − ζ(z, q) = , where q = , {α, β ∈ R}, z ∈ C (G.1)
(α · k + β)z α
k=1

Therefore we have 2 poles, s = 1 and s = −β/α. There are 4 forms ( of functions that are equivalent
to the generalized function of G − ζ(z, g), which means, that they are not always equal. The last 2 provide
the solution of the generalized equation G − ζ (z, q) = 0. The Important point of solving G − ζ(z, q) = 0 is
to check if there are zeros with 0 < σ < 1, i.e. if Re(z) is within the open interval (0, 1).

Equivalent relations functions that hold with respect to the values a, b and q.

α+β
If where q = α , {α, β ∈ R}, z ∈ C, apply per case,

G − ζ(z, q) = α−z · HurwitzZeta(z, q) (1)

G − ζ(z, q) = α−z · Zeta(z, q) (2) (G.2)

 z 
G − ζ(z, q) = α−z · 1
− (q−1) + Zeta(z, q − 1) = T1 − Zeta(z, q) (3)
 z 
G − ζ(z, q) = α−z · 1
q + Zeta(z, q + 1) = T2 − Zeta(z, q) (4)

But for the process of solving with the Infinite Periodic Radicals method and the generalized G.Rt theorem
we need to analyze further the cases related to the values of α, β and for which functions are equal to the
G − ζ(z, q). The main thing we are mainly interested in are the functions T1 − Zeta (z, q) & T2 − Zeta(z, q)
because they are represented by exponential functions with respect to q and therefore give multiple different
zeros. In total we distinguish 5 cases related to the signs of α, β and q.

52
I. The case where α < 0, β > 0 and q > 0 & |q| < 1.

In this case the following equations apply:

 z 
1
G − ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z · + Zeta(z, q + 1) = T2 − Zeta(z, q) ∀z ∈ C (G.2)
q

Since the objective is G − ζ(z, q) = 0 with T2 − Zeta(z, q) = 0 implies:

 z 
G − ζ(z, q) = α−z · HurwitzZeta (z, q) = α−z · 1
q + Zeta(z, q + 1) = 0
 z
with Generator 1q = u ⇒
(G.3)
{k ∈ Z, u 6= 0, q 6= 0} and
2·i·π·k log(u)
z = f (u) = + , Generator function.
log(1/q) log(1/q)

Program(example) G.ζ.1. The matrix program in mathematica with infinity Periodic Radicals is:

This case is very basic like the next one because it produces zeros belonging to the interval (0, 1) that are
doubly conjugate and in particular 0 < Re(z) < 1/2. They indicate that the hypothesis is rejected if its
formulation refers to the most generalized form of the z-function. Of course, this fact if it holds depends on
the assumptions we make in the definition of the Hypothesis from the beginning.

Some results of the project are given in the table 18:

Example α = −3 & β = 2.

53
 Table 17: 5 zeros for Re(z) in interval (0, 1)

Therefore, as we observe, we have a number of conjugate zeros in the interval (0, 1). This means that
we will have scattered critical lines in pairs of complex zeros, which are infinite. The percentages cording
to the number of zeros requested for more than d = 10 ranges from 76% to 99.99...% as the number of zeros
requested increases.

II. The case where α < 0, β < 0 and q > 0 & |q| > 1.

In this case equations apply:

 z 
G − ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z · 1
q+ Zeta(z, q + 1) = T2 − Zeta(z, q)
  z 
G − ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z · − q−1
1
+ Zeta(z, q + 1) = T1 − Zeta(z, q) (G.4)

G − ζ(z, q) = α−z · Zeta(z, q) ∀z ∈ C

Since the objective is G − ζ(z, g) = 0 with T2 − Zeta (z2 q) = 0 implies:

 z 
G − ζ(z, q) = α−z · HurwitzZeta (z, q) = α−z · 1
q + Zeta(z, q + 1) = 0
 z
with Generator 1q = u ⇒
(G.5)
{k ∈ Z, u 6= 0, q 6= 0} and
2·i·π·k log(u)
z = f (u) = + , Generator function.
log(1/q) log(1/q)

Program(example) G.ζ.2. The matrix program in mathematica with infinity Periodic Radicals is:

Some results of the project are given in the table 19:

Example α = −3 & β = −1

54
 Table 18: 5 zeros for Re(z) in interval (0, 1)

So here too we have a number of conjugate zeros in the interval (0, 1).This means that we will have scattered
critical lines in pairs of complex zeros, which are infinite. The percentages from d = 20 range from 40% to
99.99 . . . % as the number of required zeros increases.

III. The case where α > 0, β < 0, |α| > |β| and q > 0 6 & |q| < 1 & The case where α > 0, β < 0,
|α| < |β| and q > 0 6 |q| < 1

They have no solution because the G − ζ (z, g) tends to infinity:

∞
X 1 α+β
G − ζ(z, q) = → ∞, where q = α , {α, β ∈ R}, z ∈ C
(α · k + β)z
k=1

We therefore do not examine these cases.

55
IV. The case where α > 0, β > 0 and q > 0 & |q| > 1

In this case equations apply:

 z 
G − ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z · 1
q + Zeta(z, q + 1) = T2 − Zeta(z, q)
  z 
G − ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z 1
· − q−1 + Zeta(z, q + 1) = T1 − Zeta(z, q) (G.7)

G − ζ(z, q) = α−z · Zeta(z, q) ∀z ∈ C

Since the objective is G − ζ(z, q) = 0 with T2 − Zeta(z, q) = 0 implies:

 z 
G − ζ(z, q) = α−z · HurwitzZeta (z, q) = α−z · 1
q + Zeta(z, q + 1) = 0
 z
with Generator 1q = u ⇒
(G.8)
{k ∈ Z, u 6= 0, q 6= 0} and

2·i·π·k log(u)
z = f (u) = log(1/q) + log(1/q) , Generator function.

Program(example) G.ζ.4. The matrix program in mathematica with infinity Periodic Radicals is:

Some results of the project are given in the table 19

Example α = 3 & β = 7

56
 Table 19: 5 zeros for Re(z) in interval (0, 1)

Note that the zeros all with Re(z) outside the interval (0, 1) per pair. We also notice that the values
of Re(z) are very close and there is also a net true root at 85.475..about.

V. The case where α < 0, β > 0 and q < 0 & |q| > 1

The equations that apply in this case are:

 z 
1
G − ζ(z, q) = α−2 · HurwitzZeta(z, q) = α−z · + Zeta(z, q + 1) = T2 − Zeta(z, q)
q
  z 
1 (G.9)
G − ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z · − + Zeta(z, q + 1) = T1 − Zeta(z, q)
q−1
G − ζ(z, q) = α−2 · Zeta(z, q) ∀z ∈ C

Since the objective is G − ζ(z, q) = 0 with T2 − Zeta(z, q) = 0 implies:

 z 
1
G - ζ(z, q) = α−z · HurwitzZeta(z, q) = α−z · + Zeta(z, q + 1) = 0
q
 z
1
with Generator =u⇒
q (G.10)
{k ∈ Z, u 6= 0, q 6= 0} and
2·i·π·k log(u)
z = f (u) = + , Generator function.
log(1/q) log(1/q)

This is a peculiar case we are considering, which has no double conjugates but only simple conjugates,
so simple points. It is not a rejection of Riemann’s Hypothesis because no critical lines are defined. Obvi-
ously we can look at many more examples and draw some conclusion about the region we are considering.
But for cases I&II we have drawn very useful conclusions that warn us that the Riemann Hypothesis bends
when it faces the generalized function G − ζ(z, q).

57
Program(example) G.ζ.5. The matrix program in mathematica with in nity Periodic Radicals is:

Some results of the project are given in the table 19

Example α = −3 & α = 10

Table 20: 5 zeros for Re(z) in interval (0, 1)

We see that the zeros of this class are all with Re(z) within the interval (0, 1) per root. We also note that
the values of Im(z) do not have opposite signs are what we call orphan complex roots it is now obvious that
for the generalized equation G − ζ(z, q) = 0 its zeros where R(z) is in the interval (0,1), correspond only to
cases I,II,V. But this equation is derived from the generalized z() function, so the Riemann hypothesis is
Rejected in these cases as well.

58
 Epilogue to Chapter 6

I. The zeros of ζ(z) = 0

The fact that the roots of ζ(z) = 0 all lie on the critical line, follows from 5 proofs, both basic and el-
ementary.

1.Using direct method of inverse of ζ() on functional equations, chapter 6.1.1.

2.Solutions of the Equation ζ(s)/ζ(1 − s) = 1, s is Complex number, chapter 6.1.3.

3.With general inverse random function theory, chapter 6.2.

4.Statistical evidence method chapter 6.3.

5.With the define the function λ(N), chapter 6.4.1.

Therefore, the proof that all zeros of ζ(z) = 0 belong on the critical line is complete, since the func-
tion ζ(z) = 0 as proved many times has zeros only on the critical line which means that Re(z) = 1/2 and
only this one.

II. Other variations of the function ζ(z) = 0

Finally, the cases ζ(z, q) = 0 or ζ(q, z) = 0 or G − ζ(z, q) = 0, q ∈ R, are considered in detail in these 4
intervals to clarify whether and where there are zeros and whether they correspond to critical lines. The
analytical examination yields zeros with Re(z) in the interval (0, 1) for the cases ζ (q ∗ z) = 0 where q > 1
in R, G − ζ(z, q) = 0 for 3 general cases out of 5 , and for case Davenport Heilbronn case, which has zeros of
a Dirichlet series periodic. Therefore, whether or not the Riemann Hypothesis holds depends on whether or
not we accept the fact if we consider only ζ(z) = 0. If in the definition of the Hypothesis the other forms or
part of them also hold i.e. ζ (q ∗ z) = 0 or ζ(q, z) = 0 or G- ζ(z, q) = 0 with q ∈ R in general or the Epstein
Function or the Dirichlet series and others, then the Riemann Hypothesis does not hold in general, but only
in special cases which are stated in detail. But we have to accept an indisputable fact that in order to have
a visual perception of which ζ-functions have zeros with Re(z) = 1/2 or have Re(z) different from 1/2 or
other critical lines outside the interval (0, 1), we have to solve transcendental equations [24] with accuracy.

III. In which cases the Riemann case is valid or rejected.

Cases I, II of the characteristic generalized equation G − ζ(z, q) = 0, q > 0, |q| < 1 or |q| > 1, q ∈ R
(pages 52-58) as well as the case ζ (q ∗ z) = 0 with q > 1 (pages 30-32) are subclass of the generalized one
mentioned above, and some cases of the Dirihlet periodic series Davenport-Heilbronn (page 51) reject the
Riemann Hypothesis. So if the hypothesis is formulated as follows: ”For any ζ-function, the zeros of the
resulting equation coincide on the critical line with Re(z) = 1/2, the formulation is complicated as a Latent
Hypothesis and does NOT hold”. But if it is formulated as ”The zeros only of ζ(z) = 0 coincide on the
critical line for Re(z) = 1/2 is correct and the Hypothesis is Valid”, as has been shown. Obviously this
happened because the equations of other related functions i.e other related equations arising from related
functions with the ζ-function were not completely or adequately or at all and solved and there was a large
knowledge gap related to where their zeros are located with respect mainly to the interval (0, 1). These
equations are forms of Transcendental equations quite difficult and to solve them a strong method and a
generalized Theorem for solving such equations is needed.

59
Part II. Distribution of primes numbers and Zeros
7. Introduction

One way to get a feeling for where the primes are within the natural numbers is to count the number
of primes π(n) less than or equal to a given number n. There is no real hope for finding a simple rule
for π(n), but the Prime Number Theorem1 (PNT) says that π(n) can be approximated by the logarithmic
integral function. The distribution of prime numbers is most simply expressed as the (discontinuous) step
function π(x), where π(x) (Fig.7), is the number of primes less than or equal to x.

Figure 9: The primes of function π(x)

The function π(x) in relation to the random int. x. It turns out that π(x) can be expressed exactly as the
limit of a sequence of smooth functions Rn (x). To define Rn (x) we first introduce the logarithmic integral
function Li(x),
Z x
du
Li(x) =
2 log u

which appears throughout the analytical theory of the prime distribution:

This is a smooth function which simply gives the area under the curve of the function 1/ log u in the
interval [2, x]. Don Zagier explains the reasoning behind the function Li in his excellent introductory ar-
ticle “The first 50 million prime numbers” [14], based on his inaugural lecture held at Bonn University,
May 5, 1975: “A good approximation to π(x)”, which was first given by Gauss is obtained by taking as
starting point the empirical fact that the frequency of prime numbers near a very large number x is almost
exactly 1/ log x. From this, the number of prime numbers up to x should be approximately given by the
logarithmic sum Li(x) = 1/ log 2 + 1/ log 3 + 1/ log x or, what is essentially the same, by the logarithmic
integral. Z x
du
Li(x) =
2 log u

Using Li(x) we then define another smooth function, R(x), first introduced by Riemann in his original eight
page paper, and given by
∞
X µ(n)
R(x) = Li(x1/n )
n=1
n

Riemann’s research on prime numbers suggests that the probability for a large number x to be prime should

60
be even closer to 1/ log x if one counted not only the prime numbers but also the powers of primes, count-
ing the square of a prime as half a prime, the cube of a prime as a third, etc. This leads to the approximation:
1 1
π(x) + π(x1/2 ) + π(x1/3 ) + ... ≈ Li(x)
2 3

or, equivalently [by means of the Möbius inversion formula]

1 1
π(x) ≈ Li(x) − Li(x1/2 ) − Li(x1/3 ) − ...
2 3

The function on the right side of this formula is denoted by R(x), in honour of Riemann. It represents
an amazingly good approximation to π(x). For those in the audience who know a little function theory,
perhaps I might add that R(x) is an entire function of log x, given by the rapidly converging power series:
∞
X (ln x)k
R(x) = 1 +
kk!ζ(k + 1)
k=1

where ζ(k + 1) is the Riemann zeta.

Figure 10: The non-trivial zeros in the Critical Line

Here we see the zeros of the Riemann zeta function in the complex plane. These fall into two categories,
trivial and nontrivialzeros (Fig.8). Here are some tables on nontrivial zeros compliled by Andrew Odlyzko[4].
The trivial zeros are simply the negative even integers. The nontrivial zeros are known to all lie in the
critical strip that is 0 < Re[s] < 1, and always come in complex conjugate pairs. All known nontrivial
zeros lie on the critical line Re[s] = 1/2. The Riemann Hypothesis states that they all lie on this line.
The difference between the prime counting function and its “amazingly good approximation” R(x), i.e. the
fluctuations in the distribution of primes, can be expressed in terms of the entire set of zeros of zeta, which
we shall represent by ρ, via the function R itself:
X
R(x) − π(x) = R (xρ )
ρ

Obviously some of the xρ are complex values, so here R is the analytic continuation of the real-valued
function R defined previously. This was mentioned above by Zagier, and is known as the Gram Series

61
expansion:
∞
X
[R (xρk ) + R (xρ−k )]
k=1

The contributions from the complex-conjugate pairs ρk and ρ−k = ρk cancel each others’ imaginary parts,
so
∞
X ∞

X
π(x) = R(x) − R x−2m + Tk (x)
m=1 k=1

Where Tk (x) = −R (xρk ) − R (xρ−k ) are real-valued. We can now define the sequence of functions Rn (x)
which approximate π(x) in limit:
∞
X n
X
−2m


Rn (x) = R(x) − R x + Tk (x) (1)
m=1 k=1

∞
! Z ∞
X 1
f (x) = −2Re Ei (ρk log(x)) + dt − log(2) (2)
x (t3 − 1) log(t)
k=1

where ρκ is the κth complex zero of the zeta function. In this formula, Ei(z) (Mathematica’s built-in
function ExpIntegral Ei(z) is the generalization of the logarithmic integral to complex numbers. These
equations come from references [13], [14], and [15]. First, let M be the smallest integer such that x1/M < 2.
We need to add only the first M-1 terms (that is, n = 1, 2, .., M ) in the
sum in equation (1). For each of
these values of N , we use equation (2) to compute the value of f x1/n . However, we will add only the first
N terms (that is, κ = 1, 2, .., N ) in the sum in equation (2). Because the purpose of this Demonstration is to
show how the jumps in the step function π(x) can be closely approximated by adding to R(x) a correction
term that involves zeta zeros, we ignore the integral and the log 2 in second equation; this speeds up the
computation and will not noticeably affect the graphs, especially for x more than about 5. The more zeros
we use, the closer we can approximate π(x). For larger x, the correction term must include more zeros in
order to accurately approximate π(x).

8. The GUE hypothesis.

While many attempts to prove the RH had been made, a few amount of work has been devoted to the
study of the distribution of zeros of the Zeta function. A major step has been done toward a detailed
study of the distribution of zeros of the Zeta function by Hugh Montgomery [18], with the Montgomery pair
correlation conjecture. Expressed in terms of the normalized spacing.

log (γn /(2π))

δn = (γn+1 − γn )
2π
this conjecture is that, for M → ∞
Z b Z b  2
1 sin πu
# {n : N + 1 ≤ n ≤ N + M, δn ∈ [α, β]} ∼ p(0, u)du ∼ 1− du
M a a πu
2 
sin πu
In other words, the density of normalized spacing between non-necessarily consecutive zeros is 1− .
πu
It was first noted by the Freeman Dyson, a quantum physicist, during a now-legendary short teatime ex-
change with Hugh Montgomery[18], that this is precisely the pair correlation function of eigenvalues of
random hermitian matrices with independent normal distribution in (Fig.9&10) of its coefficients. Such
random hermitian matrices are called the Gauss unitary ensemble (GUE). As referred by Odlyzko in [16] for
example, this motivates the GUE hypothesis which is the conjecture that the distribution of the normalized
spacing between zeros of the Zeta function is asymptotically equal to the distribution of the GUE eigen val-
ues. Where p(0, u) is a certain probability density function, quite complicated to obtain for an expression of

62
 π2 2 π4
it). As reported by Odlyzko in [16], we have the Taylor expansion around zero. p(0, u) = u −2 u4 +. . .
3 45
which under the GUE hypothesis entails that the proportion of δ n less than a given small value δ is asymp-
π2 3

totic to δ + O δ 5 . Thus very close pair of zeros are rare.
9

Figure 11: Probability density of the normalized spacing

between non-necessarily consecutive zeros and the GUE prediction[20]

Figure 12: Probability density of the normalized spacing δn and the GUE prediction[20].

9. Gaps between zeros

The table below lists the minimum and maximal values of normalized spacing between zeros δn and of
δn + δn+1 , and compares this with what is expected under the GUE hypothesis. It can be proved that
π2 2 π4
p(0, t) have the following Taylor expansion around 0, p(0, u) = u − 2 u4 + · · · so in particular, for
3 45
small delta
Z δ
π2 3
P rob (δn < δ) = p(0, u)du ∼ δ
0 9
so that the probability that the smallest δn are less than δ for M consecutive values of δn is about.
M
π2 3 π2
  
1− 1− δ ' 1 − exp − δ 3 M
9 9
This was the value used in the sixth column of the table. The result can be also obtained for the δn +δn+1 . . .

π6 8
P rob (δn + δn+1 < δ) ∼ δ
32400

63
from which we deduce the value of the last column (Table 21).

Table 21: The values of δn + δn+1

For very large spacing in the GUE, as reported by Odlyzko in [16], des Cloizeaux and Mehta [17] have
proved that
log p(0, t) ∼ −π 2 t2 /8 (t → ∞)
which suggests that
(8 log M )1/2
max δn ∼
N +1≤n≤N +M π

Statistics of False zetazeros, δ-intervals and count of primes.

10. General equations.

With this statistic we find the crowd of individual parts intervals defined by consecutive Zetazeros in a
fixed integer interval. Here we use δ = 1000. We have the general Equation δ+ Zetazero (kin) - Ze-
tazero (kf ) = 0 and after given initial value in kin, i calculate the kf usually by the Newton method.
From the initial and final value of kin, kf and by performing the process of successive intervals, i calcu-
late the number(count) of the primes ones that are within the intervals. In this way will have consecutive
intervals. Sk = { Zetazero (k + 1) < p < Zetazero (k), k ∈ (kin, kf )} from where resulting the number of
the primes in the given consecutive interval. The sum of the primes in the given interval δ = 1000 will be
obvious.
kf
[
Sδ = Si
i=kin

The False intervals will be in a normalized form Fδ = 1000 − Sδ over N >= 4000.

11. The statistics

The First Statistic looking for ways to show us which function is the most ideal to get closer the points of
interest, uses the NonlinearModelFit method of the function y = (a + b · x)/(log(d + c · x) and therefore
after determining the variables {a, b, c, d} we are able to make statistical and probable prediction at higher
levels of numbers . This function is directly related to the function π(x) = x/ log x which was reported in
the introduction that is, defining the number of primes numbers relative to x. We will do a double statistic
of the intervals {δ, False intervals } and {δ, number of primes } that we are ultimately interested in
the statistics mainly the count of the primes inside at δ -intervals.

64
11.1 The first statistic is about count of False intervals and the count of the primes that correspond-
ing in them. The table below gives it in aggregate until the count of primes equal pδ = 29 corresponding to
the count of False Fδ = 971 and the Total range Integers for a interval D = 1000 to 2 · 1014 + 1000, ( Table
22).

Table 22: The data.1

The diagram given below (Fig.11) shows the arrangement of the points at the level (x, y) according to the
data1[19].

Figure 13: The depiction of the intervals False and the primes ones (file:data1) contained in the intervals
N + δ − N = 1000.

Even more macroscopically (Fig.12), the points are shown by the line y = 988.709 − 0.988372 ∗ x, which,
as they appear, are stacked on its lower right,
It is obvious that for {x = 999, y = 1.32} and for {y = 1000, x = 0.33}, a value located below the unit and
means that it can in such a interval, and we are talking about this is for high order integers intervals so
there is not one prime within the interval δ = 1000, chosen at random.

Basic thinking usually in statistics is to calculate at constant shows the number of data we need respec-
tively, finding the frequencies that appear. This method is a real percentage of realistic Espigue isolates the
properties showing a group of numbers.

11.2 The second statistic refers to the count of the primes ones that located in the intervals [N, N + δ]
with N = 1000 up to 2·1014 , range δ = 1000 and the count of the primes within successive ZetaZeros.

65
 Figure 14: Graphic depiction of the line y = 988.709 − 0.988372 ∗ x with the archive(data1) points.

After we found the roots kf of general Equation δ+ Zetazero (kin) -Zetazero (kf ) = 0 of given initial
value kin, usually by the Newton method. The data 19 (Table 23), that we have met with the above method
are.

Table 23: The data

Using the NonlinearModelFit[19] method of the function y = (a + c ∗ x)/(log[d + b ∗ x]) and after specifying
the variables {a, b, c, d} the following result the function will be:
1108.254246288494 − 1.116325646187134 × 10−12 x
y=
log[−4503.90336177023 + 5.425635171403081x]
with very good approach and value performance in each pairing.The diagram (Fig.13) of the above equation
is shown in more detail below with gravity in the latest data.
y = (a + c ∗ x)/(log[d + b ∗ x]). Thus a test value for x = 2 · 1014 gives us y = 25.6, close to 29 we took with
the analysis. They follow statistics ANOVA and t-Statistic [19, 20] and we get the (Tables 24, 25).

Analysis of variance (ANOVA) is an analysis tool used in statistics that splits an observed aggregate vari-
ability found inside a data set into two parts: systematic factors and random factors. The systematic factors

66
 Figure 15: The graphic depiction of the function

have a statistical influence on the given data set, while the random factors do not.An ANOVA test is a way
to find out if survey or experiment results are significant. In other words, they help you to figure out if you
need to reject the null hypothesis or accept the alternate hypothesis.

Analysis-ANOVA

Table 24:

Analysis-T-statisrtic

Table 25:

As we can see from the results, we have an important Standard Error only for the variable a. The other
variables are observed to have a low statistical error and have good compatibility. By adapting the method
as we see, we associate two lists of results of the number of primes and long intervals, which although
disproportionately, work together with impeccable and good contact.

Ronald Fisher introduced the term variance and proposed its formal analysis in a 1918 article The Corre-
lation
Between Relatives on the Supposition of Mendelian Inheritance. His first application of the analysis of vari-
ance was published in 1921. Analysis of variance became widely known after being included in Fisher’s 1925
book Statistical Methods for Research Workers.

One of the attributes of ANOVA that ensured its early popularity was computational elegance. The structure
of the additive model allows solution for the additive coefficients by simple algebra rather than by matrix

67
calculations. In the era of mechanical calculators this simplicity was critical. The determination of statis-
tical significance also required access to tables of the F function which were supplied by early statistics texts.

12. Standard–Error and Confidence–Interval

The standard error (SE) of a statistic (usually an estimate of a parameter) is the standard deviation of
its sampling distribution or an estimate of that standard deviation. If the statistic is the sample mean, it
is called the standard error of the mean (SEM). The sampling distribution of a population mean is
generated by repeated sampling and recording of the means obtained. This forms a distribution of different
means, and this distribution has its own mean and variance. Mathematically, the variance of the sampling
distribution obtained is equal to the variance of the population divided by the sample size. This is because
as the sample size increases, sample means cluster more closely around the population mean. Therefore,
the relationship between the standard error of the mean and the standard deviation is such that, for a given
sample size, the standard error of the mean equals the standard deviation divided by the square root of the
sample size. In other words, the standard error of the mean is a measure of the dispersion of sample means
around the population mean. Where we observed, in addition to the initial measurements #11-#12, the
Standard - Error also the confidence interval stabilized at good and acceptable values with central average
values around the lateral expected values. The values are given in two consecutive tables that we add below
and refer to the whole values of range (Table 26)

Table 26:

68
13. Statistical comparison

With statistical comparison it appears that as long as we moving to a higher order of integer-size that
are within a given interval, we have chosen δ = 1000, the count of the primes diminishes to disappear or to
there are 1 − 2 primes at high order intervals of more than 1020 . This also agrees with of Gram’s law. In
particular, the problem of distribution of the differences tn+1 − tn (that is of difference of ZetaZeros)
is considered [21]. If we accept Gram’s law then the order of this difference does not exceed the quantity
2π
tn+1 − tn ≈ ln(n) → 0, n → ∞ for much larger integers [20, 21] then mean and their mean value is close to,

ln ln(n)
tn+1 − tn ≈ → 0, n → ∞
ln(n)

And as the above analysis of-False intervals- as we have shown, it is compatible with this result of
law Gram’s.

14. From above analysis they arise 3 big conclusions:

1st . The number(Count) of primes located within δ intervals is gradually decreasing with a higher or-
der size n of 10n .
2nd . The distribution of the number(Count) of the primes, within interval δ follows a Nonlinear correlation
by the function y = (a + b · x)/(log(d + c · x) similar to the number (Count) of the primes π(x) ≈ x/ log(x),
which is apply approximating for large numbers.
3rd . The measurement(Count) of false intervals and the count of the primes follows a linear correlation and
it increases if the order of magnitude size of the integer increases.

Epilogue

This analysis has proved the Riemann Hypothesis is correct, since the real part of the non-trivial zero-
equation functions is always constant and equals 1/2. This proven by 3 independent methods. Theorems
{1.Theorem I, II (pages 1 − 7), 2.Theorem (pages 7 − 8), 3.Theorem (pages 8 − 10)}. This analysis has also
demonstrated that the imaginary part of the non trivial zero zeta function accepts certain values according
to the intervals defined by the solution of the 3 based functional equations {Eq.set (I, II), III (page 10)}.

The Riemann Zeta function as defined by Riemann is ζ(s, 0) = ζ(s) and applies Riemann’s functional
equations. If the equations are solved, the only solution for the real part of s is 1/2. Many other scientists
refer to other forms such as ζ(s, a) or ζ(Q(x), a) or Dirichlet series and many other cases, which are not
related to the original definition and therefore pertain to the analysis of other cases.

The question is not simply whether Riemann’s Hypothesis is valid or not, but which ζ-functions do we

accept that the Hypothesis as defined by Riemann includes; should we accept the general view of existing
ζ-functions that the Hypothesis includes or not? So if we take as a general assumption of the Hypothesis all
parallel ζ-functions then it is not 100% true. Of course it is also true that the Riemann Hypothesis has mul-
tiple connections with other sciences and gives directions appropriate and grouped according to the function
that we accept to be valid in each particular application case. If we accept generalized ζ-functions, this is
taken as the only real answer to the Riemann Hypothesis conjecture that has occupied the mathematical
community for more than 200 years because it generalizes the hypothesis

69
References
[1] A Proof of the Riemann Hypothesis, Murad Ahmed Abu Amr (2016, arXiv.org)
[2] A Geometric Proof of Riemann Hypothesis, Kaida Shi p.20-45,(2003 arXiv.org)

[3] An Exploration of the Riemann Zeta Function and its Applications p.15-60, 2006.
[4] Themis Mitsis,(Complex analysis I) p.17-30,
http://fourier.math.uoc.gr/~mitsis/notes/undergrad_complex.pdf.
[5] Mantzakouras Nikos Solve Polynomials and Transcendental equations using Mathematica,
p.50-80,2012, also in internet a part
https://www.academia.edu/21964173/Solve_Equations_With_method_Lagrange_
[6] Proof of Hypothesis Riemann,Part I. Mantzakouras Nikos (2017). The zeros of the ζ(z) = 0.
https://www.academia.edu/26990409/Proof_of_the_Riemann_Hypothesis_part_I
[7] Nikolentzoc Poluqronic, R-H p.9-22, 2008.

[8] A theory for the zeros of Riemann and other 2014, L-functions Guilherme Frand Andre Le Clairy
p.38-50
[9] Titchmarsh E C. The zeros of the Riemann zeta-function. Proc Roy Soc Ser, 1935,151: 234-255.
[10] Titchmarsh E C. The zeros of the Riemann zeta-function. Proc Roy Soc Ser, 1936,157: 261-263.

[11] Van de Lune J, te Riele H J J, Winter D T. On the zeros of the Riemann zeta-functionin the critical
strip. Math. Com, 1986, 46: 667-681.
[12] Hardy G H, Littlewood J E. The zeros of Riemann’s zeta-function on the critical line. Math Z, 1921,
10: 283-317.

[13] H.M. Edwards, Riemann’s Zeta Function, Academic Press, New York, London, 1974.
[14] D. Zagier, The first 50 million prime numbers”, Mathematical Intelligencer (1977), p.7–19.
[15] The Mathematical Intelligencer (1977), p.7-19.
[16] Andrew Odlyzko. The 1021 -st zeros of the Riemann zeta function, 1988. Tables of Zeros.

[17] Cloizeaux and Mehta, The Widom–Dyson constant for the gap probability in random matrix theory,
2007, p.26-47.
[18] Hugh Montgomery, The pair correlation of the zeros function, 1970.
[19] The “data1 & data & NonlinearModelFit Statistic”, have been done with the program mathematica.

[20] ’The 1013 first zeros of the Riemann Zeta function, and zeros computation at very large heigh’,
Xavier Gourdon, version 2004.
[21] On the Gram’s Law in the Theory of Riemann Zeta Function. M.A. Korolev, 2010, p.4-6.
[22] Euler Gamma function page 123, http://www.math.leidenuniv.nl/~evertse/ant13-8.pdf

[23] Video of Enrico Bombieri, https://www.ias.edu/video/analysis/davenport-hellbronn

[24] Solve Polynomial and transcendental Equations with use Generalized Theorem (Method Lagrange),
https://arxiv.org/search/?query=mantzakouras+&searchtype=all&source=header,
Mantzakouras Nikos 2021

70
[25] Zeros of Lattice Sums: 1. Zeros of the Critical Line R.C. McPhedran,
https://arxiv.org/pdf/1601.01724.pdf
[26] The inverse Riemann zeta function, A Kawalec (p. 61-63).18 June 2021.

[27] E. Saias, A. Weingartner, Zeros of Dirichlet series with periodic coefficients, Acta Arith. 140 (2009)
[28] On the concavity of Dirichlet’s eta function and related functional inequalities.Horst Alzer,Man
KamKwong,2015
[29] Hafner, J.: Zeros on the critical line for Dirichlet series attached to certain cusp forms. Math. Ann.
264, 2t-37 (1983)
[30] Heath-Brown, D.R.: Simple zeros of the Riemann zeta-function on the critical line. Bull. London
Math. Soc. 11, 17-18 (1979)
[31] Levinson, N.: More than one-third of zeros of Riemann’s zeta-function are on t = 1/2. Adv. Math.
13, 383-436 (1974)

[32] Davenport, H.; Heilbronn, H., On the zeros of certain Dirichlet series I, II. J. London Math. Soc. 11
(1936), 181-185, 307-312.
[33] Spira, Robert., Some zeros of the Titchmarsh counterexample. Math. Comp. 63 (1994), no.208,
747-748. MR1254148 (95a:11081)

71