AIMS Mathematics, 8(11): 25772–25803.

DOI:10.3934/math.20231315
Received: 17 May 2023
Revised: 10 July 2023
Accepted: 18 July 2023
http://www.aimspress.com/journal/Math Published: 07 September 2023

Research article

The Generalized Riemann Hypothesis on elliptic complex fields

Xian Hemingway*

Department of Mathematics, Hunan University, Changsha 410000, China

* Correspondence: Email: amicc-wx@outlook.com.

Abstract: In this paper, we will introduce a new algebraic system called the elliptic complex, and
consider the distribution of zeros of the function L(s, χ) in the corresponding complex plane. The key to
this article is to discover the limiting case of the Generalized Riemann Hypothesis on elliptic complex
fields and, taking a series of elliptic complex fields as variables, to study the ordinary properties of
their distributions about the non-trivial zeros of L(s, χ). It is on the basis of these considerations that
we will draw the following conclusions. First, the zeros of the function L(s, χ) on any two elliptic
complex planes correspond one-to-one. Then, all non-trivial zeros of the L(s, χ) function on each
elliptic complex plane are distributed on the critical line ℜ(s) = 12 due to the critical case of the
Generalized Riemann Hypothesis. Ultimately we proved the Generalized Riemann Hypothesis.

Keywords: the elliptic complex; the normal ellipse; the function L̂(s, χ) and ξ(s, χ); the critical case
of GRH; the Generalized Riemann Hypothesis
Mathematics Subject Classification: 11R42, 11R54

1. Introduction

It is well known that Riemann studied the distribution of nontrivial zeros of the function ζ(s) in the
complex plane and obtained the conclusion that the nontrivial zeros of ζ(s) were necessarily distributed
in the critical region 0 < ℜ(s) < 1 and symmetric about the point s = 1/2. He ventured to guess that
the nontrivial zeros of the function ζ(s) were distributed on the critical line ℜ(s) = 12 . This hypothesis
has not been proved so far.
In order to solve the problem of the distribution of primes in arithmetic series, Dirichlet introduced
the Dirichlet function L(s, χ). After analysis, mathematicians also gave the conjecture that the zeros of
L(s, χ) are all distributed on the line ℜ(s) = 1/2.
In the present paper, we will construct the binary algebraic system shown below as a basis for the
study of the L-function, which is also the key to solving the Generalized Riemann Hypothesis.
 25773

Definition 1.1. [1] Denote by ∀x, y ∈ R, the numbers of the form z = x + iy as the elliptic complexes,
where i satisfies i2 = λ, λ ∈ R− . The set of all elliptic complexes is denoted Cλ .

To avoid confusion we will refer to the elliptic complex with λ = −1 as the circular complex, i.e.,
the complex C = C−1 invented by the mathematical predecessors.
It is easy to show that all elliptic complexes satisfying the definition are divisible algebraic numbers
that satisfy both the multiplicative exchange law and the multiplicative combined law, and are the
number fields.
Let z∗ = x−iy be the conjugate complex of the complex z with z = x+iy ∈ Cλ , similarly, while N(z) =
zz∗ be the norm of the complex z which obviously satisfies √ N(z1 )N(z2 ) = N(z1 z2 ). Correspondingly, the
distance from the origin to the complex number z is |z| = N(z), which is also known as the modulus
of the complex z. Typically, the distance between any two complex numbers z1 and z2 is |z1 − z2 |, which
can also be derived accordingly.
The calculus is subsequently first introduced into the study of the elliptic complexes to reach some
basic conclusions on the theory of the elliptic complex functions where the results not relevant to this
paper are not given. These conclusions are used to study the basic properties and theory of the function
ζ(s) on the elliptic complex fields.

2. Preliminaries

Suppose that p = −λ = q2 would be made throughout the rest for convenience, and according to the
basic operations of elliptic complexes, it is easy to prove [1, 2]
!
x+ qi y i
e = e cos y + sin y ,
x
(2.1)
q

where e is the base of the natural logarithm, and sin θ and cos θ are the sine and cosine functions of
θ, respectively. This is Euler’s formula on elliptic complex fields whose proof, including some of the
details as follows, would be detailed in Appendix A.
For a point A(x, y) on the elliptic complex plane Cλ , let the origin of the coordinates be O, then, the
−−→ −−→ −−→
corresponding vector OA = x + iy. Now, consider the vector OB = iOA = −py + ix, which coordinates
of point B is B(−py, x). Thus it leads to the fact that

1 1
kOA · kOB = (−1) × = , (2.2)
p λ

−−→
where, with the case of y = 0 in particular, easy to get clearly that the vector OA is on the x-axis and
−−→
the vector OB on the y-axis in the complex plane. It can be seen that when λ , −1, the coordinate
system corresponding to the elliptic complex plane is no longer a right-angle coordinate system, but a
oblique coordinates system.
It is easy to get that φ = arcsin
q
1
is the angle between the positive y-axis and the positive x-axis in
the xOy coordinate system, so that a positive or negative q characterizes the chirality of the coordinate
system where in particular the negative q corresponds to the left-handed coordinate system.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25774

2.1. The normal ellipse

In order to clarify the representation of an ellipse in the elliptic complex plane, we need to redefine
the ellipse.
2 2
Definition 2.1. An ellipse of the form C : ax2 + by2 = 1, a > b > 0 is defined to have the (real) eccentricity
q √   √ 
of e = 1 − ba2 and two real foci of F1 = a2 − b2 , 0 and F2 = − a2 − b2 , 0 , while the imaginary
2

q  √   √ 
eccentricity of e = 1 − ab2 and two imaginary foci of F1 = 0, a2 − b2 and F2 = 0, − a2 − b2 .
′ 2 ′ ′

By the definition of the norm and modulus of the elliptic complex Cλ ,

x2 y2 x2 y2
+ = 1 ⇔ 2 +  2 = 1, (2.3)
N(z) N(z)
2
|z| |z|
q q

whose geometric meaning is an ellipse centred on the origin of the coordinates in the complex plane.
In the case where 0 < |q| < 1, the long semi-axis of the ellipse is m = |q| |z|
and the short semi-axis is
n = |z|, and the direction of the short axis at this point is said to be the direction of the major axis of the
ellipse. When |q| ≥ 1, the opposite is true and the direction of the long axis is said to be the direction
of the major axis of the ellipse. The length of the long axis (or short axis) in the direction of the major
axis is called the major axis length of the ellipse and half of the major axis length is called the principal
diameter. An ellipse with the principal semidiameter of 1 is called a unit ellipse.

Definition 2.2. When 0 < |q| < 1, an ellipse

q in the complex plane Cλ is said to be the normal ellipse
with its (real) eccentricity satisfing e = 1 + λ1 and the direction of its principal axis being parallel
to the x-axis. When |q| > 1, an ellipse qin the complex plane Cλ is said to be the normal ellipse with its
imaginary eccentricity satisfing e = 1+
′ 1
and the direction of its principal axis being parallel to
λ
q
the x-axis, at which time the eccentricity of the normal ellipse is also e = 1 + λ1 . In a special way,
the normal ellipse of the circular complex plane C−1 is a circle.

It is easy to know that the distance from any point on a normal ellipse to the center of the ellipse is
equal and all equal to its principal semidiameter. Further, the normal ellipse with z0 as the centre and r
as the principal semidiameter can be expressed as |z − z0 | = r.
The normal ellipse is the most fundamental geometric element in the elliptic complex plane,
and its use in many proofs in analysis is as important as that of the circle in the circular complex
plane. We consider that any geometric object in the complex plane is articulated by a series of
normal ellipses Γ1 , Γ2 , · · · , Γn , · · · , whose corresponding principal diameters r1 , r2 , · · · , rn , · · · satisfy
max(r1 , r2 , · · · , rn , · · · ) → 0.
Where not otherwise specified, all references to ellipses below are to the normal ellipses.

2.2. Basic results for the elliptic complex functions

Using the mathematician’s method of introducing calculus to circular complexes, we would arrive
at the following result.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25775

Theorem 2.3. Let f (z) = u(x, y) + iv(x, y) be a function of the complex variable defined on a region D
of the complex plane Cλ , then, a sufficient condition for f (z) f (z) at z = x + iy ∈ D is that the functions
u(x, y), v(x, y) are differentiable at (x, y) and satisfy [2, 3]

∂u ∂v ∂u ∂v
= , = −q2 , (2.4)
∂x ∂y ∂y ∂x
where q meets i2 = λ = −q2 (q ∈ R∗ ).
The proofs of Theorems 2.3, together with the other propositions and theorems below, are detailed
in Appendix B.
The basic elementary functions can be defined on elliptic complex fields next, and it is easy to know
that they are all analytic functions.  
1) The exponential function: ez = e x cos y + qi sin y , where z = x + qi y;
2) The trigonometric functions: the cosine and sine functions are defined as [2]
1  i qz z q  i qz z
cos z = e + e−i q , sin z = e − e−i q , (2.5)
2 2i
where it is easy to yield that the zero of the function sin z is kπ (k ∈ Z) and of the function cos z is
(k + 12 )π (k ∈ Z) on the complex plane Cλ . Further, there are the following conclusions apparently.
Proposition 2.4. Let Cλ1 and Cλ2 be two arbitrary complex planes, then, the function ez has the same
value on Cλ1 and Cλ2 if and only if z is a real number. Similarly, the functions cos z and sin z have the
same properties.
h i
Proof. Suppose z = a + ib, then, ez = ea cos(qb) + qi sin(qb) . As q is a non-zero constant, the function
ez is equal on any two complex planes Cλ1 and Cλ2 only if b = 0, i.e., z is a real number, which leads to
ez = ea .
The same conclusion can be likewise drawn due to the fact that [2, 4]
q  −qb 1
e − eqb sin a + e−qb + eqb cos a,
 
cos z =
2i 2
which leads to cos z = cos a if b = 0, and
q  −qb 1
e − eqb cos a + e−qb + eqb sin a,
 
sin z =
2i 2
which leads to sin z = sin a if b = 0. □
and cosh z = e +e
z −z z −z
Now, define sinh z = e −e
2 2
as the hyperbolic sine function and hyperbolic cosine
function of z, respectively. The definitions of the other trigonometric functions are the same as those
on the circular complex plane and are of no further interest here.
3) The definition of a argument function on an elliptic complex domain is the same as on a circular
complex domain. Define the logarithmic function as
i
w = Log z = log |z| + Arg z. (2.6)
q
As for the power and radical functions, they are also derived on the basis of logarithmic functions,
which would not be listed here.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25776

Proposition 2.5. Let z ∈ Cλ , then, z = z∗ holds when λ → 0− .

Proof. Since z can be expressed as z = Reiθ , the equation z = z∗ clearly holds when R = 0. When
R , 0, if λ = −q2 → 0− , then, based on Eq (A.1) in Appendix A.
" #
z i
= lim e = lim cos(2qθ) + sin(2qθ) = 1,
2iθ
z∗ λ→0− q→0 q
which means that the equation z = z∗ holds constantly [2, 3, 5]. □
In fact, Proposition 2.5 tells us that the geometric plane corresponding to the complex plane Cλ as
λ → 0− is a plane where the angle between the x-axis and the y-axis tends to zero.
Consider the integral of the function f (z) = u(x, y) + iv(x, y) on the curve C in the elliptic complex
plane, which is easily accessible by Eq (2.4).
Z Z Z
f (z)dz = u(x, y)dx − q v(x, y)dy + i u(x, y)dx + v(x, y)dy.
2
(2.7)
C C C

With regard to the integral operations, the following results are available.
Proposition 2.6. Let C be the elliptical circumference |z − α| = ρ, taking the positive direction of the
complex plane, then, we will see that [2]

2πi
, n = 1;
Z (
1
dz = q
(2.8)
C (z − α)n 0, n , 1, n ∈ Z,
which is essential for the derivation of the elliptic complex analysis.
Proof. It is clear that C : |z − α| = ρ, i.e., C : z = α + ρei q (0 ≤ t ≤ 2π), thus dz = i ρq ei q dt. When n = 1,
t t

the integral sought is

Z Z 2π i ρ ei qt dt
1 q 2πi
dz = = .
C z−α
t
0 ρe qi q
As n , 1, the integral is

i ρq ei q dt
t
Z Z 2π Z 2π
1 i i
dz = = n−1 e− q (n−1)t dt = 0.
C (z − α)n 0 ρn e i ntq qρ 0

The proposition is thus proved. □

Similarly, there is a corresponding Cauchy integral theorem for curves in the elliptic complex plane
as follows.
Theorem 2.7. Let C be any simple closed curve in the region D, then,
Z
f (z)dz = 0. (2.9)
C
As a further step, the integration equation can be accessed by combining Eq (2.8) with Cauchy’s
integral theorem.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25777

Theorem 2.8. Let D be a bounded region bounded by a finite number of simple closed curves C, and
the function f (z) be analytic over the closed region D consisting of D and C, then, with ∀z ∈ D there
is the formula
I
q f (ζ)
f (z) = dζ, (2.10)
2πi C ζ−z
and the higher order derivative formula
I
q · n! f (ζ)
f (z) =
(n)
dζ (n = 1, 2, . . .). (2.11)
2πi C (ζ − z)n+1
With reference to the basic theory of series, the results on the elliptic complex fields are almost
identical to those on the circular complex fields and will not be repeated. As a matter of course, the
following conclusions are drawn with regard to the residues.
Theorem 2.9. Supposing D being a bounded region in the complex plane whose boundary is a (or
a finite composition of) simple closed curves, if the function f (z) is analytic in D except for a finite
number of isolated singularities z1 , z2 , · · · , zn , and also analytic on the boundary C = ∂D, then, there
is [2, 5]

I n
2πi X
f (z)dz = Res ( f, zk ), (2.12)
C q k=1
where the integral along C is drawn in the positive direction* about region D in the complex plane.

2.3. Fourier and Merlin transforms on the elliptic complex plane

The formula for the Fourier integral
Z ∞ "Z ∞ #
1 −i ws wt
f (t) = f (s)e q ds ei q dw, (2.13)
2π −∞ −∞

over the elliptic complex field is easily derived from the Fourier series, thus there is the concept of the
Fourier transform.
Definition 2.10. † If the function f (t) satisfies the conditions of the Fourier integral theorem on
(−∞, +∞), the function [5, 6] Z ∞
wt
F(w) = f (t)e−i q dt (2.14)
−∞
is said to be the Fourier transform of f (t). While the function
Z ∞
1 wt
F (F(w)) := f (t) =
−1
F(w)ei q dw (2.15)
2π −∞
is the inverse Fourier transform of F(w).
* Our thumb passes from the inner side of the complex plane to the outer side, and the other four fingers bend from the positive x-axis

to the positive y-axis, so that the direction of bending of the four fingers is the positive direction of the complex plane. It is easily shown
that the positive direction of the complex plane is counterclockwise with the case of q > 0 corresponding to a right-handed coordinate
system, and counterclockwise with the case of q < 0 corresponding to a right-handed coordinate system.
†
Correspondingly, the Fourier transform on the elliptic complex field has two forms, the other of which would be derived in Appendix
C.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25778

Suppose s = c − qi ω, then, ω = qi (s − c) and dω = qi ds which leads to that

Z ∞ Z ∞
F[i(s − c)] = (s−c)t
e f (t)dt = e st · [e−ct f (t)]dt, (2.16)
−∞ −∞

and
c−i∞ c+i∞
q · ect
Z Z
i
f (t) = (c−s)t
e F[i(s − c)]ds = e−st F[i(s − c)]ds. (2.17)
2πq c+i∞ 2πi c−i∞

Next, suppose again that t = ln x, dt = x−1 dx, yielding

Z ∞
F[i(s − c)] = x s−1 [x−c f (ln x)]dx, (2.18)
0

along with
c+i∞
q · xc
Z
f (ln x) = x−s F[i(s − c)]ds, (2.19)
2πi c−i∞
where with the case of assuming g(x) = x−c f (ln x) and G(s) = F[i(s − c)], we would arrive at
Z ∞
G(s) = {Mg}(s) = x s−1 g(x)dx, (2.20)
0

which is said to be the Merlin transform of g(s), and

Z c+i∞
q
g(x) = {M G}(t) =
−1
x−sG(s)ds, (2.21)
2πi c−i∞

which is the inverse Merlin transform of G(s). The presence of a real number c which could be
appropriately selected in the inverse conversion formula could avoid possible poles in the integration
path.

3. Dirichlet functions on the elliptic complex fields

3.1. Dirichlet characters and Dirichlet functions

In a similar way to the study of the Riemann hypothesis, the definition of the arithmetic series is
here first extended to the elliptic complex fields.
Definition 3.1. Let χ : Zm ∗ → Cλ ∗ be a group homomorphism satisfying the multiplicative
operation [5, 7]
χ(ab) = χ(a)χ(b), ∀a, b ∈ Zm ∗ .
Thus call χ the Dirichlet character (mod m) on the elliptic complex field Cλ . For convenience, χ
can also be written as χ(n) = χ(n; m). Further, define
∞
X χ(n)
L(s, χ) =
n=1
ns

to be the Dirichlet function with respect to the character χ on Cλ .

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25779

It is easy to conclude that χ(1) = 1, and χ(n) φ(m) = 1, which means that the identity χ is the φ(m)-
 
th root of unity for (n, m) = 1 and further that χ(−1) = ±1. If χ(−1) = 1, then χ is said to be an even
character. If χ(−1) = −1, then χ is said to be an odd character. Since χ is a unit root, the inverse of the
identity χ satisfies χ−1 = χ.

Definition 3.2. If χ(n) = 1 is constant when (n, m) = 1, call it the trivial character, denoted χ0 (n). The
rest of the characters are termed the non-trivial characters. If the values of them take only real values,
they are referred to as the real characters, otherwise they are referred to as complex characters.

Apparently, the real characters on any complex field Cλ1 are the same as those on Cλ2 , but the
difference is the complex characters. Let χ(n)λ be the character on the complex field Cλ . Then, based
on the properties of χ and the Euler formula shown in Eq (A.6), it follows that ℜ χ(n)λ1 = ℜ χ(n)λ2
   
and [1, 5, 8]
q1 · ℑ χ(n)λ1 = q2 · ℑ χ(n)λ2 ,
   
(3.1)

where λ1 = −q1 2 and λ2 = −q2 2 .

3.2. Gauss sums on the elliptic complex fields

It is easy to know that the n-th unit root on Cλ is
!
i 2kπ 2kπ i 2kπ
wk = e q n = cos + sin , k = 0, 1, · · · , n − 1, (3.2)
n q n

where w0 = 1, when k = 0, and the conjugate

! corresponding to wk is wk ∗ = wn−k .
2πi
Now, note that e(x) =: exp x is a function on Cλ . From the property of the unit root [5, 9],
q
! 
X kn h,
 h|n;
=

e (3.3)
k∈Zh
h 0,
 otherwise.

Denote the Gauss sum G(n; χ) of the character χ(mod h) by

!
X kn
G(n; χ) = χ(k)e , (3.4)
k∈Zh
h

which means that, multiplying the Gauss sum by χ(n),

! X
X kn m
χ(n)G(n; χ) = χ(nk)e = χ(m)e ,
k∈Zh
h m∈nZ
h
h

with the case of (n, h) = 1, which leads to that nZh = Zh . Thus,

G(n; χ) = χ(n)G(1; χ). (3.5)

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25780

Furthermore, combining Eqs (3.3) and (3.5) yields that

!
X −k
|G(1; χ)| = G(1; χ)G(1; χ) =
2
χ(k)G(1; χ)e
k∈Zh
h
! ! !
X −k X km −k
= G(k; χ)e = χ(m)e e
k∈Zh
h k,m∈Zh
h h
X X k(m − 1) ! X
= χ(m) e = χ(1) 1 = h,
m∈Z k∈Z
h k∈Z
h h h

√
which leads to |G(1; χ)| = h.
The definitions of the induced modulus and the primitive character over Cλ are not repeated here.
It is straightforward to show that the function L(s, χ) has no zeros in the region ℜ(s) ⩾ 1 and is
convergent in the region ℜ(s) > 0 when χ , χ0 . These results are all consistent with those in the
circular complex field.

4. The functional equations for Dirichlet functions

Suppose that χ is the character (mod h) on Cλ , and [5, 10]


0, χ(−1) = 1;

ε(χ) = 

(4.1)
1, χ(−1) = −1.


Define the Dirichlet L-function of the completion, supposing ΓR (s) = π−s/2 Γ(s/2), as
s
L̂(s, χ) = h 2 ΓR (s + ε(χ))L(s, χ), (4.2)
which could eliminate the effect of the trivial zeros of L(s, χ).
Lemma 4.1. Supposing the function

∞
X
e−(n+a) πx , x > 0,
2
θ(x, a) = (4.3)
n=−∞
it follows that ! ∞
1 √ X
e−πxn + q na .
2 2πi
θ ,a = x (4.4)
x n=−∞
!
1 √
In particular, θ = xθ(x) when a = 0, where θ(x) = θ(x, 0).
x
Proof. Suppose the function f (u) = exp − πx (u + a)2 while u + a = xy. According to Eq (C.11), the
 

Fourier transform of f (u) is [5, 11]

Z ∞  π  2πi
g(v) = exp − (u + a)2 e− q vu du
−∞ x
Z ∞
−πxv2 + 2πi i 2
= xe q va
e−πx(y+ q v) dy.
−∞

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25781

Then, applying the integral theorem as shown in Theorem 2.7,

Z ∞ Z ∞
−πx(y+iv)2 2 1
e dy = e−πxy dy = √ , (4.5)
−∞ −∞ x
√
which resulted in g(v) = xe−πxv + q va .
2 2πi

Following the Poisson summation formula as shown in Eq (C.19),

∞ ∞ ∞
1 X 2π
X √ X
e−πxn + q na .
2 2πi
θ( , a) = e−(n+a) x = g(n) = x (4.6)
x n=−∞ n=−∞ n=−∞

Thus the proposition is proved. □

Theorem 4.2. Suppose that χ is the character (mod h) on Cλ with λ = −q2 , and the function [5, 12]

∞
X
χ(m)e−m πx/h ,
2
ψ(x, χ) =: (4.7)
m=−∞
when χ(−1) = 1, along with the function
∞
X
mχ(m)e−m πx/h ,
2
ϕ(x, χ) =: (4.8)
m=−∞

when χ(−1) = −1. Then, it follows that

1
!  x  12
ψ , χ = τ(χ) ψ(x, χ), (4.9)
x h
and !  x  12
1 i
ϕ , χ = − τ(χ)x ϕ(x, χ), (4.10)
x q h
where τ(χ) = G(1; χ).
Proof. In accordance with Lemma 4.1,
∞
m X
e−(nh+m) πx/h .
2
θ(hx, ) =
h n=−∞

Thus, combining Eq (3.5) gives

! X h ! h h
1 h m X X πxn2 2πi nm
e− h + q h
p
ψ ,χ = χ(m) · θ , = x/h χ(m)
x m=1
x h m=1 m=1
∞
 x  12 X πxn 2
!
= G(n; χ) exp −
h n=−∞ h
∞
 x  12 πxn2
X !
= G(1; χ) χ(n) exp −
h n=−∞
h
 x  12
= G(1; χ)ψ (x, χ) ,
h

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25782

which yields Eq (4.9).

To differentiate with respect to a on both sides of Eq (4.4) gives
∞ ∞
π(n + a)2
!
X i 3 X −πxn2 + 2πiq na
(n + a) exp − = − x2 ne . (4.11)
n=−∞
x q n=−∞

Consequently,
! h ∞  !
1 X X m h  m 2
ϕ ,χ = h χ(m) l+ exp −π l +
x m=1 l=−∞
h x h
∞
πxl2
1 !
i  x2 X
=− x G(l; χ)l exp −
q h l=−∞ h
i  x  21
= − G(1; χ)x ϕ(x, χ),
q h
which yields Eq (4.10). □
Due to the fact that χ(0) = 0 and, for any positive integer m, that χ(−m) = χ(−1)χ(m) = χ(m) when
χ(−1) = 1, and (−m)χ(−m) = mχ(m) when χ(−1) = −1, it follows that
∞
X 1
χ(m)e−m πx/h = ψ(x, χ),
2
ψ1 (x, χ) = (4.12)
m=1
2

and ∞
X 1
mχ(m)e−m πx/h = ϕ(x, χ).
2
ϕ1 (x, χ) = (4.13)
m=1
2
Next use these conclusions to prove the following proposition.
Theorem 4.3. The Dirichlet L-function can be analytically extended to the entire complex plane and
satisfies the functional equation

L̂(s, χ) = W(χ)L̂(1 − s, χ), (4.14)

where
G(1, χ)
W(χ) =  ε(χ) √ (4.15)
i
q
h
is the unit root on the complex field Cλ and χ is the primitive character (mod h) over Cλ with λ = −q2 .
Proof. When χ(−1) = 1, according to the Laplace transform formula on the elliptic complex fields
which is the same as one on the circular complex field,
Γ(z)
Z ∞
xz−1 e−λx dx = z . (4.16)
0 λ
Now supposing z = s/2, λ = πn2 /h,
∞
Γ(s/2)
Z
2 x/h
x s/2−1 e−πn dx = ,
0 (πn2 /h) s/2

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25783

which leads to, following Eq (4.12), that

Z ∞ ∞
X Z ∞
e−n πx/h x s/2−1 dx
2
ψ1 (x, χ)x s/2−1
dx = χ(n)
0 n=1 0
∞
X χ(n) Γ(s/2)
= = L(s, χ)Γ(s/2)π−s/2 h s/2 .
n=1
n s (π/h) s/2

Thus, by the definition of the function L̂(s, χ),

Z ∞
1
L̂(s, χ) = ψ(x, χ)x s/2−1 dx, (4.17)
2 0

which result in, in combination with Eq (4.9), that

1 1 1 ∞
Z Z
s s
L(s, χ) =
b ψ(x, χ)x dx +
2 −1
ψ(x, χ)x 2 −1 dx
2 0 2 1
Z ∞
1 1 1
! Z
1 1 − 2s −1 s
= ψ ,χ x dx + ψ( , χ)x− 2 −1 dx
2 1 x 2 0 x
Z ∞
1 1−s
= G(1; χ)h−1/2 ψ (x, χ) x 2 −1 dx,
2 0

thereby providing that

G(1, χ)
L(s, χ) =  0 √ b
b L(1 − s, χ). (4.18)
i
q
h

When χ(−1) = −1, supposing z = (s + 1)/2, λ = πn2 /h on Eq (4.16),

∞ Γ( s+1 )
Z
2 x/h
x(s+1)/2−1 e−πn dx = 2
,
0 (πn /h)
2 (s+1)/2

Therefore,
Z ∞ ∞
X Z ∞
e−n πx/h x(s+1)/2−1 dx
2
ϕ1 (x, χ)x (s+1)/2−1
dx = nχ(n)
0 n=1 0

χ(n) Γ( 2 )
X∞ s+1
=
n=1
n s (π/h)(s+1)/2
s + 1 −(s+1)/2 (s+1)/2
= h1/2 L(s, χ)Γ( )π h
2
= h1/2 L̂(s, χ),
which leads to, following Eq (4.13), that
Z ∞
1
L̂(s, χ) = h−1/2 ϕ(x, χ)x(s+1)/2−1 dx. (4.19)
2 0

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25784

Hence, combining Eq (4.10),

1 −1/2 1 1 −1/2 ∞
Z Z
L̂(s, χ) = h ϕ(x, χ)x (s+1)/2−1
dx + h ϕ(x, χ)x(s+1)/2−1 dx
2 0 2 1
Z ∞
1 −1/2 1
Z
1 −1/2
= h ϕ(x , χ)x
−1 −(s+1)/2−1
dx + h ϕ(x−1 , χ)x−(s+1)/2−1 dx
2 1 2 0
Z ∞
1 −1/2 i
=− h G(1; χ)h−1/2 ϕ(x, χ)x−s/2 dx,
2 q 0

that is,
G(1, χ)
L̂(s, χ) = i 1/2
L̂(1 − s, χ). (4.20)
q
h
In summary, the proposition is proved. □

Under Theorem 4.3, the function L̂(s, χ) is invariant under the substitution s → 1 − s, χ → χ, which
leads to the following conclusion.

Corollary 4.4. When χ is a real character, the zeros of the function L̂(s, χ) are symmetric with respect
1 1
to s = and the function L̂( + it, χ) is a real even function.
2 2
By Proposition 2.5, the character χ on the complex field Cλ satisfies χ = χ when λ → 0− , so there
is the following conclusion.
1
Corollary 4.5. When λ → 0− , the zeros of the function L̂(s, χ) are symmetric with respect to s = if χ
2
is a complex character on the field Cλ .

When χ(−1) = −1, suppose ξ(s, χ) = h s/2 L̂(s, χ) for convenience of the next operation.

5. An equivalent proposition for the Generalized Riemann Hypothesis

Without the special remarks, in the following we assume that χ is the primitive character (mod h)
i
with h ⩾ 3 over the complex field Cλ and that s = σ + t ∈ Cλ . The Generalized Riemann Hypothesis
q
is noted as GRH later for convenience [5].

5.1. The case when χ is a real character and satisfies χ(−1) = 1

When χ(−1) = 1, depending on Eq (4.17),

1 ∞
Z Z ∞
h π Γ(s/2)L(s, χ) =
s/2 −s/2
ψ(x, χ)x s/2−1
dx = ψ(x2 , χ)x s−1 dx. (5.1)
2 0 0

According to Eq (2.20), Eq (5.1) is exactly of the form of a Merlin transformation, i.e.,

h s/2 π−s/2 Γ(s/2)L(s, χ) = {Mψ(x2 , χ)}(s). (5.2)

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25785

Therefore, using the inverse formula of the Merlin transform,

2 +i∞
Z 1
q
ψ(x , χ) =
2
h s/2 π−s/2 Γ(s/2)L(s, χ)x−s ds, (5.3)
2πi 1
2 −i∞

due to the fact that the integrand has no poles at σ > 0. Replacing x → eu gives

2 +i∞
Z 1
q
ψ(e , χ) =
2u
h s/2 π−s/2 Γ(s/2)L(s, χ)e−us ds. (5.4)
2πi 1
2 −i∞

1 i
And now let s = + t, then Eq (5.4) can be reduced to
2 q
Z ∞ !
1 − qi ut 1 i
ψ(e , χ)e =
2u u/2
e L̂ + t, χ ds, (5.5)
| {z } 2π −∞ 2 q
Ψ1 ( u2 ,χ)

where, according to Corollary 4.4, the function

u  ∞
X
χ(n)e−πn e /h+u/2
2 2u
Ψ1 , χ = ψ(e2u , χ)eu/2 =
2 n=−∞

is also an even function, so the combination of the Fourier inversion formula as seen in Definition 2.10
shows that ! Z ∞  Z ∞ 
1 i u  qi ut u 
L̂ + t, χ = Ψ1 , χ e du = 2 Ψ1 , χ cos(ut)du
2 q −∞ 2 0 2
Z ∞
=4 Ψ1 (u, χ) cos(2xt)du.
0
Z ∞
eϖu Ψ1 (u, χ) cos(zu)du,
2
Supposing 10 H(ϖ, z) =
0

!
1 1 iz
1
0 H(0, z) = L̂ + , χ , (5.6)
4 2 2q

which the following conclusion can be drawn by combining the properties of L̂(s, χ) by.

Proposition 5.1. Let χ be a real primitive character ( mod h) with h ⩾ 3 over Cλ and satisfy χ(−1) = 1.
The GRH with respect to χ holds when and only when all zeros of the function 10 H(0, z) are real.

Based on the definition of the function 10 H(0, z), the following conclusion holds in conjunction with
Proposition 2.4.

Proposition 5.2. Let Cλ1 and Cλ2 be any two complex planes. The function 10 H(0, z) has equal values
in Cλ1 and Cλ2 provided that z is a definite real number.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25786

5.2. The case when χ is a real character and satisfies χ(−1) = −1

When χ(−1) = −1, following Eq (4.19),
s+1 1 ∞
Z Z ∞
hπ Γ(
s −(s+1)/2
)L(s, χ) = ϕ(x, χ)x (s+1)/2−1
dx = ϕ(x2 , χ)x s dx. (5.7)
2 2 0 0
Similarly, Eq (5.7) is exactly of the form of a Merlin transformation, that is
h s−1 π−s/2 Γ(s/2)L(s − 1, χ) = {Mϕ(x2 , χ)}(s), (5.8)
which leads to, using the inverse formula of the Merlin transform, that
Z 21 +i∞
q
ϕ(x , χ) =
2
h s−1 π−s/2 Γ(s/2)L(s − 1, χ)x−s ds. (5.9)
2πi 12 −i∞
Replacing Eq (5.9) by x → eu yields [5]
2 +i∞
Z 1
q s−1
ϕ(e , χ) =
2u
h 2 L̂(s − 1, χ)e−us ds. (5.10)
2πi 1
2 −i∞

1 i
Hereby let s − 1 = + t, then, Eq (5.10) can be turned into
2 q
Z ∞ !
1 − qi ut 1 i
ϕ(e , χ)e
2u 3u/2
= e ξ + t, χ ds, (5.11)
| {z } 2π −∞ 2 q
Ψ2 ( u2 ,χ)

where the function ∞

u  X
nχ(n)e−πn e /h+3u/2
2 2u
Ψ2 , χ = ϕ(e , χ)e
2u 3u/2
=
2 n=−∞
is equally an even function, resulting in that
! Z ∞  Z ∞ 
1 i u  qi ut u 
ξ + t, χ = Ψ2 , χ e du = 2 Ψ2 , χ cos(ut)du
2 q −∞ 2 0 2
Z ∞
=4 Ψ2 (u, χ) cos(2xt)du,
0
Z ∞
eϖu Ψ2 (u, χ) cos(zu)du,
2
Given the function 20 H(ϖ, z) =
0
!
1 1 iz
2
0 H(0, z) = ξ + ,χ , (5.12)
4 2 2q
which likewise brings us to the following conclusion.
Proposition 5.3. Let χ be a real primitive character (mod h) with h ⩾ 3 over Cλ and satisfy χ(−1) =
−1. The GRH with respect to χ holds when and only when all zeros of the function 20 H(0, z) are real.
Proposition 5.4. Let Cλ1 and Cλ2 be any two complex planes. The function 20 H(0, z) has equal values
in Cλ1 and Cλ2 provided that z is a definite real number.
The zeros of the function L̂(s, χ) corresponding to the complex characters are not necessarily
symmetric about the point s = 1/2, so its equivalence condition about the GRH will be discussed
later.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25787

6. Correspondence of zeros on the different complex planes

Since the function L̂(s, χ) has no poles in the region ℜ(s) ∈ (0, 1), allowing ∆ ∈ (0, 1) to be a given
constant, Eqs (5.4) and (5.10) can be easily transformed into that
Z ∆+i∞
q
ψ(e , χ) =
2u
L̂(s, χ)e−us ds, (6.1)
2πi ∆−i∞
which corresponds to the case of even character, along with that [5, 6, 13]
Z ∆+i∞
q s−1
ϕ(e , χ) =
2u
h 2 L̂(s − 1, χ)e−us ds, (6.2)
2πi ∆−i∞
i
corresponding to the case of the odd character. To let s = ∆ + t on Eq (6.1), it follows that
q
Z ∞ !
1 − qi ut i
ψ(e , χ)e =
2u u∆
e L̂ ∆ + t, χ ds. (6.3)
| {z } 2π −∞ q
Ψ∆1 ( 2u ,χ)

i
And suppose s − 1 = ∆ + t on Eq (6.2), to find that
q
Z ∞ !
1 − qi ut i
ϕ(e , χ)e
2u u(∆+1)
= e ξ ∆ + t, χ ds. (6.4)
| {z } 2π −∞ q
Ψ∆2 ( u2 ,χ)

From Eq (6.3) combined with the inverse Fourier transform formula,

! Z ∞ Z ∞
i ∆ u
  i
Ψ∆1 (u, χ) e q u(2t) du.
i
L̂ ∆ + t, χ = Ψ1 , χ e q du = 2
ut
(6.5)
q −∞ 2 −∞

Now suppose that Z ∞

∆
eϖu Ψ∆1 (u, χ) cos(zu)du,
2
1
H (ϖ, z) =
−∞
and that Z ∞
∆
eϖu Ψ∆1 (u, χ) sin(zu)du.
2
1
F (ϖ, z) =
−∞
Depending on the definition of the trigonometric function as seen in Eq (2.5),
! ! " !#
1 ∆ 1 iz 1 iz iz
H (0, z) = L̂ ∆ + , χ + L̂ ∆ − , χ = ℜ L̂ ∆ + , χ , (6.6)
2 2q 2 2q 2q
and ! ! " !#
∆ 1 iz 1 iz iz
1
F (0, z) = L̂ ∆ + , χ − L̂ ∆ − , χ = qℑ L̂ ∆ + , χ , (6.7)
2 2q 2 2q 2q
which could lead to that !
i i
L̂ ∆ + z, χ = 1 H ∆ (0, z) + · 1 F ∆ (0, z), (6.8)
2q q

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25788
!
i
where L̂ ∆ + z, χ = 0 if and only if 1 H ∆ (0, z) = 0 and 1 F ∆ (0, z) = 0.
2q
By Proposition 2.4, the functions 1 H ∆ (0, z) or 1 F ∆ (0, z) have the same value on any two complex
planes Cλ1 and Cλ2 when z is a definite real number, leading to the following conclusion.
Proposition 6.1. The zeros of the function L̂(s, χ) on any two complex planes Cλ1 and Cλ2 are in one-
i
to-one correspondence. Let λ1 = −q1 2 and λ2 = −q2 2 . Precisely, if s = ∆ + z, where z is a real
q1
′ i
number, is the zero of the function L̂(s, χ) on the complex plane Cλ1 , then s = ∆ + z has to be the
q2
zero of the function L̂(s, χ) on the complex plane Cλ2 .
1
+ Re q1 θ and
i
If we set ∆ = 1
2
+ R cos θ and z = R sin θ, where R ⩾ 0, it follows that s =
2
′ 1 i
θ
s = + Re q2 . Based on the definition of the normal ellipse, there is the following conclusion.
2
Corollary 6.2. If the zero of the function L̂(s, χ) on the complex plane Cλ1 is on the normal ellipse
1
centred at the point s = and with R ⩾ 0 as its principal semi-diameter, then its zero on the complex
2
1
plane Cλ2 is also on the normal ellipse centred at the point s = and with R as its principal semi-
2
diameter correspondingly [6, 13].
Equivalently, when χ is an odd character, it follows from Eq (6.4) combined with the inverse Fourier
transform formula, it follows that
! Z ∞ Z ∞
i ∆ u
  i
Ψ∆2 (u, χ) e q u(2t) du.
i
ξ ∆ + t, χ = Ψ2 , χ e du = 2
q ut
(6.9)
q −∞ 2 −∞

Now assume that Z ∞

∆
eϖu Ψ∆2 (u, χ) cos(zu)du,
2
2
H (ϖ, z) =
−∞
and that Z ∞
∆
eϖu Ψ∆2 (u, χ) sin(zu)du,
2
2
F (ϖ, z) =
−∞
which could analogously lead to that
!
i i
ξ ∆ + z, χ = 2 H ∆ (0, z) + · 2 F ∆ (0, z), (6.10)
2q q
!
i
where ξ ∆ + z, χ = 0 if and only if 2 H ∆ (0, z) = 0 and 2 F ∆ (0, z) = 0.
2q
The following conclusions can be drawn in the same way.
Proposition 6.3. The zeros of the function ξ(s, χ) on any two complex planes Cλ1 and Cλ2 are in one-
i
to-one correspondence. Precisely, if s = ∆ + z, where z is a real number, is the zero of the function
q1
′ i
ξ(s, χ) on the complex plane Cλ1 , then s = ∆ + z has to be the zero of the function ξ(s, χ) on the
q2
complex plane Cλ2 .

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25789

Corollary 6.4. If the zero of the function ξ(s, χ) on the complex plane Cλ1 is on the normal ellipse
1
centred at the point s = and with R ⩾ 0 as its principal semi-diameter, then its zero on the complex
2
1
plane Cλ2 is also on the normal ellipse centred at the point s = and with R as its principal semi-
2
diameter correspondingly [6].
Obviously, Propositions 6.1 and 6.3, along with Corollaries 6.2 and 6.4 above, hold irrespective of
whether χ is a real or complex character.
From the above analysis, it is easy to conclude that the equivalent proposition of the GRH
corresponding to the complex character is as follows.
1
Proposition 6.5. Let ∆ = and χ be the complex primitive character (mod h) with h ⩾ 3 over
2
Cλ . If χ(−1) = 1, the corresponding GRH holds when and only when the zeros of the function
1 ∆ i
H (0, z) + · 1 F ∆ (0, z) are all real. Additionally, in the event that χ(−1) = −1, the corresponding
q
i
GRH holds when and only when the zeros of the function 2 H ∆ (0, z) + · 2 F ∆ (0, z) are all real.
q

7. Proof of the GRH

According to our previous definition of the elliptic complex Cλ , the equation corresponding to the
norm degenerates to the form x2 = N(z) when λ = 0, representing two straight lines symmetric about
the y-axis perpendicular to the x-axis in the complex plane. Considering that any geometric feature
in the complex plane is articulated by a corresponding normal ellipse, and that the normal ellipse on
the complex plane Cλ corresponding to λ = 0 is an ellipse whose principal axis, i.e., the x-axis, is
infinitely compressed, the figure articulated by such ellipses can only be a line perpendicular to the x-
axis. However, it is clearly inappropriate to study the complex number Cλ directly with λ = 0, because
such an algebraic system is not divisible.

7.1. The critical case of the GRH on elliptic complex fields

When λ → 0− , the complex number Cλ is clearly a divisible algebra, which of course satisfies
Theorem 4.3, and the norm of the complex number s = x + iy is

N(s) = lim− (x2 − λy2 ) = x2 .

λ→0

From the symmetry of the function L̂(s, χ) for the case of χ(−1) = 1 and ξ(s, χ) for the case of
χ(−1) = −1, it follows that the zeros of L̂(s, χ) as well as ξ(s, χ) are symmetric about the point s = 1/2.
Therefore, the equation L̂(s, χ) = 0, together with ξ(s, χ) = 0, holds if and only if N(s) = x2 = 1/4,
1
i.e., ℜ(s) = when λ → 0− .
2
Pursuant to Corollaries 4.4 and 4.5, this result holds for both real and complex characters.
The geometric significance of this result can be further obtained from Corollaries 6.2 and 6.4.
Since all the zeros of the function L̂(s, χ) or ξ(s, χ) must occur symmetrically on a series of normal
ellipses centred at the point s = 1/2, thus the geometry of this result is such that, as λ → 0− , all such

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25790

normal ellipses are compressed into the straight line segments passing through the point s = 1/2 and
symmetrical about the x-axis. Therefore, with λ → 0− , all the zeros of the function L̂(s, χ) or ξ(s, χ)
1
are distributed on the critical line ℜ(s) = perfectly naturally.
2
Now, let λ0 = −p0 → 0− . The distribution of zeros of the function L̂(s, χ) or ξ(s, χ) on the elliptic
complex field Cλ with p = −λ ∈ [p0 , 1] would be explored, according to the continuity approach in
analysis, while consider such a proposition as follows.

Proposition 7.1. The zeros of the function L̂(s, χ) or ξ(s, χ) on the complex plane Cλ1 , with the case of
1
p1 = −λ1 ∈ [p0 , 1), are all distributed on ℜ(s) = , then the zeros of L̂(s, χ) or ξ(s, χ) on the complex
2
1
plane Cλ2 , with the case of p2 = −λ2 ∈ (p1 , 1], are all distributed on ℜ(s) = .
2
If this proposition can be shown, combined with the critical case when λ → 0− , the GRH on elliptic
complex fields can also be proved.

7.2. The ultimate proof

In actual fact, there is a more powerful formulation for Proposition 7.1 as the following proposition.

Proposition 7.2. For any ε > 0 and one real number p = −λ ∈ [p0 , 1), if the zeros of the function
1
L̂(s, χ) or ξ(s, χ) on the complex plane Cλ are all distributed on ℜ(s) = , then the zeros of L̂(s, χ) or
2
1
ξ(s, χ) on the complex plane Cλ1 are all distributed on ℜ(s) = where λ1 meets p1 = p + ε = −λ1 .
2
Proof. Now let p = q2 and ε = δ2 where q and δ are real numbers that are both positive or negative at
the same time.
The case of the real characters is discussed at the outset.
As assumed in Proposition 7.2, in conjunction with Propositions 5.1 and 5.3, the zeros of the
function ! !
1 1 iz 1 1 iz
0 H(0, z) = L̂ + , χ or 0 H(0, z) = ξ + , χ
1 2
4 2 2q 4 2 2q
on the complex plane Cλ are all real. Based on Propositions 5.2 and 5.4, it follows that these
corresponding points are also the zeros of the function
! !
1 1 iz 1 1 iz
0 Hε (0, z) = L̂ + ,χ or 20 Hε (0, z) = ξ + ,χ
1
4 2 2(q + δ) 4 2 2(q + δ)

on the complex plane Cλ1 .

Combining this with Propositions 6.1 and 6.3 shows that all zeros of the function 10 Hε (0, z) or
0 Hε (0, z) are also real, so that all zeros of the function L̂(s, χ) or ξ(s, χ) on the complex plane Cλ1
2

1
are distributed on ℜ(s) = .
2
Similarly, Proposition 6.5, used in combination with Propositions 6.1 and 6.3, could prove the case
for the complex characters. □

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25791

In the above, the proof of the GRH on elliptic complex fields is complete. This conclusion holds for
all elliptic complex planes Cλ with p = −λ ∈ (0, +∞), including the circular complex plane C naturally.
As a matter of fact, for any λ1 = −q21 and λ2 = −q22 with q1 , q2 ∈ R∗ , if the imaginary part of the
n-th zero of the function L̂(s, χ) or ξ(s, χ) on the complex plane Cλ1 is βn , then correspondingly the
′ q1
imaginary part of the n-th zero on the complex plane Cλ2 is βn = βn , while their real part is 1/2.
q2

8. Concluding remarks

In this paper, we first construct the elliptic complexes Cλ and introduce them into the calculus to
obtain the corresponding theory of complex variables’ function. Further, the problem of the distribution
of the zeros of the function L(s, χ) on the corresponding elliptic complex fields is discussed, which
contributed to the analytic extended form and functional equation of L̂(s, χ).
One of the difficulties of this essay is to find the correspondence between the zeros of the function
L̂(s, χ) or ξ(s, χ) in the elliptic complex plane.
And the key to this paper is to find a critical case of the GRH bout the real and complex characters
on the elliptic complex planes and to discover the equivalent propositions of the GRH. Based on
the continuity method in analysis, we eventually proved the GRH on all complex planes due to
Proposition 2.5 which shows the basic relationship of functions between the elliptic complex fields
Cλ .

Use of AI tools declaration

The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

Conflict of interest

The authors declare no conflicts of interest.

References

√
1. X. K. Zhang, Congruences mudulo 23 for class numbers of general quadratic fields Q( m) and
√
Q( m) , J. Number Theory, 32 (1989), 332–338. https://doi.org/10.1016/0022-314X(89)90089-9
2. N. A. Lebedev, Geometric theory of functions of a complex variable, Washington: American
Mathematical Society, 1969.
3. T. S. Fiske, Theory of functions of a complex variable, Science, 14 (1901), 59–65.
4. M. Morse, Topological methods in the theory of functions of a complex variable, Princeton:
Princeton University Press, 1949.
5. C. Pan, C. Pan, Basics analytic number theory, Harbin: Harbin University of Technology Press,
2016.
6. V. Chandee, K. Soundararajan, Bounding |ξ( 21 + it)| on the Riemann hypothesis, B. Lond. Math.
Soc., 43 (2011), 243–250. https://doi.org/10.1112/blms/bdq095

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25792

7. J. B. Conrey, A. Ghosh, D. Goldston, N. S. M. Gonek, D. R. Heath-Brown, On the

distribution of gaps between zeros of the zeta-function, Quart. J. Math., 36 (1985), 43–51.
https://doi.org/10.1093/qmath/36.1.43
8. P. Bourgade, J. Kuan, Strong Szeg asymptotics and zeros of the Zeta-function, Commun. Pur. Appl.
Math., 67 (2014), 1028–1044. https://doi.org/10.1002/cpa.21475
9. M. R. Murty, J. Esmonde, Problems in algebraic number theory, New York: Springer, 2005.
https://doi.org/10.1007/b138452
10. A. Selberg, Old and new conjectures and results about a class of Dirichlet series, Amalfi Conference
on Analytic Number Theory, Italy: Maiori, 1992, 367–385.
11. K. Broughan, Equivalents of the Riemann hypothesis, Cambridge: Cambridge University Press,
2017. https://doi.org/10.1017/9781108178228
12. H. Davenport, Multiplicative number theory, 2 Eds., New York: Springer, 1980.
https://doi.org/10.1007/978-1-4757-5927-3
13. A. M. Odlyzko, On the distribution of spacings between zeros of the zeta function, Math. Comput.,
48 (1987), 273–308. https://doi.org/10.2307/2007890

A. Appendix 1: The geometric significance of elliptic complexes and the proof of Euler’s
formula

There are various ways of proving Euler’s formula on elliptic complex fields, and two of the more
general methods are given merely here.

Proposition A.1. Suppose φ ∈ R and i2 = λ = −q2 , q ∈ R∗ , then,

1
eiφ = cos(qφ) + i sin(qφ), (A.1)
q
where e is the base of the natural logarithm, and sin θ, cos θ are the sine and cosine functions of θ
respectively.

Proof. From the expansion of the MacLaurin series it follows that

x x2 xn
ex = 1 + + + ... + + o (xn ) ;
1! 2! n!

x2 x4 x6 x2n  
cos x = 1 − + − + . . . + (−1)n + o x2n+1 ;
2! 4! 6! (2n)!

x x3 x5 x7 x2n+1  
sin x = − + − + . . . + (−1)n + o x2n+2 .
1! 3! 5! 7! (2n + 1)!

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25793

So there

iφ (iφ)2 (iφ)n
e =1+ +
iφ
+ ... + + o ((iφ)n )
1! 2! n!
(qφ)2 (qφ)4 (qφ)6 (qφ)2n  
=1− + − + . . . + (−1)n + o (qφ)2n+1
2! 4! 6! (2n)!
3 5 7 2n+1
" #
1 (qφ) (qφ) (qφ) (qφ) n (qφ)

+i − + − + . . . + (−1) + o (qφ)2n+2
q 1! 3! 5! 7! (2n + 1)!
1
= cos(qφ) + i sin(qφ).
q

Thus proving the proposition. □

For Eq (A.1), consider the case where the elliptic complex is a pure imaginary number, i.e.,
supposing cos(qφ) = 0 at which point sin(qφ) = 1, to obtain φ = arcsin q
1
. It can be seen that φ is
the angle between the positive y-axis and the positive x-axis in the xOy coordinate system.
An alternative approach to the proof requires further consideration of the geometric significance of
elliptic complexes. For a point A(x, y) on the elliptic complex plane Cλ , let the origin of the coordinates
−−→
be O. Then, the corresponding vector OA = x + iy, now consider the vector
−−→ −−→
OB = iOA = i(x + iy) = −py + ix.

So the coordinates of point B is B(−py, x), which gives that

1 1
kOA · kOB = (−1) × = , (A.2)
p λ
−−→
where, with the case of y = 0 in particular, easy to get clearly that the vector OA is on the x-axis and
−−→
the vector OB on the y-axis in the complex plane. It can be seen that when λ , −1, the coordinate
system corresponding to the elliptic complex plane is no longer a right-angle coordinate system, but a
oblique coordinates system. It leads to the following definitions.

Definition A.2. Two lines l1 , l2 in the elliptic complex plane Cλ are orthogonal if and only if

1
kl1 · kl2 = , (A.3)
λ
which is denoted l1 ⊥l2 . With the case of λ = −1 especially, the two lines are orthogonal while kl1 · kl2 =
−1 in the circular complex plane.

Definition A.3. One line l is said to be orthogonal to one normal ellipse Γ if it passes through the
centre O of the ellipse Γ on the elliptic complex plane Cλ .

It is easy to see that if the line l is orthogonal to the normal ellipse Γ, then the tangent lc at the
intersection of them satisfies kl · klc = λ1 which shows that, namely, the line l is orthogonal to the tangent
lc . A detailed proof of this result would be given now.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25794

Proof. Based on the translation invariance of one geometric figure in the complex plane, consider the
special regular ellipse
x2 y2
Γ: + = 1,
N(z) N(z)/p
where N(z) is the square of the principal diameter of this ellipse. Let p(x0 , y0 ) be the (non-endpoint)
tangent point on F, then the equation of its tangent line is
xx0 yy0
lc : + = 1.
N(z) N(z)/p

The slopes thus obtained are respectively

x0 1 y0
klc = − · , kl = .
y0 p x0

This leads to the required conclusion. □

Definition A.4. If two normal ellipses Γ1 and Γ2 , with the centres O1 and O2 respectively, intersect at
one point P on the elliptic complex plane Cλ , and the product of the slopes of the lines PO1 and PO2 is

1
kPO1 · kPO2 = ,
λ
then, the normal ellipse Γ1 is said to be orthogonal to the normal ellipse Γ2 .
Consider now the vector
−−→ i −−→ i
OC = OA = −qy + x.
q q
−−→ −−→ −−→ −−→
Obviously, there would be OC⊥OA, along with |OC| = |OA| = x2 + q2 y2 on the elliptic complex
p

plane Cλ . Thus, the vectors ⃗a and ⃗b on the elliptic complex plane Cλ are mutually orthogonal and equal
in length, equivalent to
i
⃗a + ⃗b = ⃗0, (A.4)
q
which is the geometric meaning of the relationship between a complex number (or vector) z and another
complex number (or vector) qi z on the elliptic complex plane Cλ .
Now begins the proof of another approach to Euler’s formula.

Proof. Set the initial value of the function y = f (x) to


y = q y;

 ′ i
(A.5)
y(0) = 1.



We find that the first equation of Eq (A.5) is a first-order ordinary differential equation in separable
variables, and hence,
Z Z
dy i dy i i
= y ⇔ = dx ⇔ ln |y| = x + C0 .
dx q y q q

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25795
i i
Thereby, |y| = e q x eC0 , i.e., y = ±eC0 e q x . Suppose C = ±eC0 , then y = Ceix which was taken into the
second equation while we can easily solve for C = 1. Consequently, according to the Picard-Lindelof
i
theorem, y = e q x is the only solution to this problem.
Now let y = cos(x) + qi sin(x), which yields

i i i i
y′ = − sin(x) + cos(x) = (cos(x) + sin(x)) = y.
q q q q

i
So it satisfies the first equation of (A.5) and obviously satisfies the second. It follows that y = e q x
and y = cos(x) + qi sin(x) are both solutions to the problem and, based on the Picard-Lindelof theorem,
we reach the conclusion that
i i
e q x = cos(x) + sin(x), (A.6)
q
which is equivalent to Eq (A.1) apparently. In general, Eq (A.6) is more commonly used. □

 2
In deed, the above result could be derived from i2 = −q2 , namely, qi = −1, which could be
substituted in the proof of Proposition A.1. Similar results for circular complex domains are easily
 2  4  
obtained with 1 + qi = 2iq , 1 + qi = −4 and qi 1 − qi = 1 + qi .
According to the above analysis about the angle of co-ordinate axes, when q = 1, the angle is θ = π2
between the forward y-axis and the forward x-axis, where the corresponding xOy coordinate system is
the so-called right-handed right-angle coordinate system. When q = −1, similarly, the angle is θ = − π2
which corresponds to the left-handed right-angle coordinate√system. In fact, the case q = −1 is the
system of complex numbers corresponding to the root i = − −1 of the equation i2 = −1, which is one
situation that the mathematicians have overlooked.
when q , 1 and q , 0, there would be |θ| = | arcsin q
1
| > π2 with the case of |q| ∈ (0, 1) and be
|θ| = | arcsin
q
1
| < π2 with the case of |q| ∈ (1, ∞). In this situation, the angle between the positive y-axis
and the positive x-axis in the corresponding xOy coordinate system is no longer a right angle.
Further for k > 0, the cases of q = k and q = −k correspond to two systems with exactly opposite
properties, similar to the positive and negative particles in quantum physics.
Of particular interest is the fact that above we have merely considered the case where q is a constant,
corresponding to a linear coordinate system. If q = q(t) is a function of a variable t, there would be
q (t) = dtd q(t), while introducing mathematical analysis, which represents the change in slope of the
′

y-axis in the positive direction.

′
It is easy to know that if q (t) is constantly greater than 0 and |q(t)| ∈ (1, ∞), such a coordinate system
′
corresponds to the geometric object described by Riemannian geometry. While q (t) is constantly less
than 0 and |q(t)| ∈ (0, 1), such a coordinate system corresponds to the geometric object described by
Lobachevskian geometry.
Obviously, there are more possible curvilinear coordinate systems to describe geometric objects
′
than the above two cases, such as the coordinate system corresponding to q (t) not constantly less than
0 (or not constantly greater than 0) and |q(t)| ∈ (s, t), where s ∈ (0, 1) and t ∈ (1, ∞), which would be
more complex and will not be discussed here.

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25796

B. Appendix 2: Proof of the theory of the elliptic complex functions

In fact, according to Subsection 2.1, the elliptic complex plane also has its own topology.
Definition B.1. The set of points in the interior of a normal ellipse with z0 as centre and δ > 0 as
principal semidiameter is called the δ-neighbourhood of z0 in the complex plane, denoted by U(z0 , δ),
namely,
U (z0 , δ) = {z| |z − z0 |< δ, z ∈ Cλ } . (B.1)
Apparently, it’s an open elliptical disc.

B.1. Proof of Cauchy-Riemann equation

Proof. For the necessity, let f (z) = a + ib and ∆z = ∆x + i∆y, then by the definition of the derivative,
′

f (z + ∆z) − f (z) = f ′ (z)∆z + o(|∆z|) while ∆z → 0 namely,

[u(x + ∆x, y + ∆y) + iv(x + ∆x, y + ∆y)] − [u(x, y) + iv(x, y)]
= (a + ib)(∆x + i∆y) + o(|∆z|)
 
= a∆x − q2 b∆y + i(b∆x + a∆y) + o(|∆z|).
Comparing the real and imaginary parts on both sides gives that, with ρ → 0,
u(x + ∆x, y + ∆y) − u(x, y) = a∆x − q2 b∆y + o(ρ)
v(x + ∆x, y + ∆y) − v(x, y) = b∆x + a∆y + o(ρ),
which leads to
∂u ∂u ∂v ∂v
a= , −q2 b = , b= , a= ,
∂x ∂y ∂x ∂y
that is,
∂u ∂v ∂u ∂v
= , = −q2 . (B.2)
∂x ∂y ∂y ∂x
Thus, the necessity is shown, next is the sufficiency.
Due to the differentiability of the functions u(x, y) and v(x, y) at the point (x, y), with ρ → 0,
∂u ∂u
u(x + ∆x, y + ∆y) − u(x, y) = ∆x + ∆y + o(ρ),
∂x ∂y
∂v ∂v
v(x + ∆x, y + ∆y) − v(x, y) = ∆x + ∆y + o(ρ).
∂x ∂y
Then, the combination of the two formulas yields
[u(x + ∆x, y + ∆y) + iv(x + ∆x, y + ∆y)] − [u(x, y) + iv(x, y)]
∂u ∂u ∂v ∂v
! !
= ∆x + ∆y + i ∆x + ∆y + o(|∆z|) (substitute Eq (B.2) into it)
∂x ∂y ∂x ∂y
∂u 2 ∂v ∂v ∂u
! !
= ∆x − q ∆y + i ∆x + ∆y + o(|∆z|)
∂x ∂x ∂x ∂x
∂u ∂v ∂u ∂v
! !
= +i (∆x + i∆y) + o(|∆z|) = +i ∆z + o(|∆z|).
∂x ∂x ∂x ∂x

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25797

Therefore, f (z) is differentiable at z = x + iy ∈ D and its derivative is

′ ∂u ∂v
f (z) = +i . (B.3)
∂x ∂x

The sufficiency of the proposition is thus proved. □

Since the in-region derivability is equivalent to the in-region resolution, Theorem 2.3 is proved
naturally.

B.2. Proof of the integral theorem i.e., Theorem 2.7

Proof. Let z = x + iy ∈ D and f (z) = u(x, y) + iv(x, y). As f (z) is analytic in the region D, the functions
u(x, y) and v(x, y) are differentiable in D and satisfy Eq (B.2) and the following Green’s formula
"
∂Q ∂P
I !
Pdx + Qdy = − dxdy, (B.4)
C D ∂x ∂y
"
∂u
2 ∂v
Z !
which leads to that u(x, y)dx − q v(x, y)dy =
2
−q − dxdy, namely,
C D ∂x ∂y
"
2 ∂v 2 ∂v
Z !
u(x, y)dx − q v(x, y)dy =
2
−q +q dxdy = 0, (B.5)
C D ∂x ∂x

and that "

∂u ∂v
Z !
v(x, y)dx + u(x, y)dy = − dxdy = 0. (B.6)
C D ∂x ∂y
Therefore, Z Z Z
f (z)dz = u(x, y)dx − q v(x, y)dy + i
2
u(x, y)dx + v(x, y)dy = 0, (B.7)
C C C

which means that Theorem 2.7 is shown. □

So, there are the following conclusions.

Corollary B.2. Let Z

C be a simple closed curve and the function f (z) analytic in the bounded region
bounded by C, then f (z)dz = 0 holds.
C

Theorem B.3. Let C0 and C1 , C2 , · · · , Cn enclose a multiply connected region D,ZD and its boundary
form a closed region D. If the function f (z) is analytic over the region D, then, f (z)dz = 0, where
C
C = C0 + C1− + C2− + · · · + Cn− are all the boundaries of the region D, which is also equivalent to
Z Z Z Z
f (z)dz = f (z)dz + f (z)dz + · · · + f (z)dz. (B.8)
C0 C1 C2 Cn

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25798

B.3. Proof of the integral formula

We know that |z − z0 | denotes the normal ellipse centered at z0 = ( (x0 , y0 ) , which can be expressed
T

(x − x0 )2
(y − y0 )2 x − x0 = a cos t,
as F : + = 1 , i.e., the parametric equation F :
y − y0 = b sin t,
t ∈ [0, 2π]. And
a2 b2
according to the symmetry of the ellipse, the circumference of this normal ellipse is
Z π2 q Z π2 p
L=4 (x (t)) + (y (t)) dt = 4
′ 2 ′ 2
a2 sin2 t + b2 cos2 tdt.
0 0

For a unit normal ellipse in the complex plane Cλ , let λ = −q2 , and the principal semidiameter of
a2
the ellipse be r, then q2 = .
b2
When |q| ≥ 1, r = a and the ellipse circumference is
Z π2 r 2
Z π2 s !
b 1
L = 4a 1 − cos2 t + 2 cos2 tdt = 4r 1 − 1 − 2 cos2 tdt =: 4rE(q).


0 a 0 q

And when 0 < |q| ≤ 1, r = b and the ellipse circumference is

Z π2 r 2 Z π2 q
a  
L = 4b 2
+ 2
= 2 sin2 tdt =: 4rE ′ (q),


sin t 1 − sin t dt 4r 1 − 1 − q
0 b2 0

where s
Z π ! Z π
1
2 2
q
E(q) = E (q) =
′
1 − 1 − q2 sin2 tdt


1 − 1 − 2 cos2 tdt and (B.9)
0 q 0

are called the perimeter coefficients of the normal ellipse on the complex plane Cλ when |q| ≥ 1 and
when 0 < |q| ≤ 1, respectively. Clearly, for the definited complex plane Cλ , the perimetric coefficient
π
is a constant. When λ = −1, the perimeter coefficient meets E(q) = E ′ (q) = corresponding to the
2
circular complex plane C.
Now begin the proof of Eq (2.10).
Proof. For any point z inside D, draw an elliptic circumference Cρ : |ζ − z| = ρ such that the closed
elliptic disk enclosed by Cρ all falls inside D.
Due to the fact that f (ζ) is continuous at ζ = z, it follows that ∃δ > 0(δ < ρ) for ∀ε > 0 such that
I ! I
1 f (ζ) 2πi 1 f (ζ) − f (z)
dζ − f (z) = dζ
4E(q) Cr ζ − z q 4E(q) Cr ζ−z
1 ε
< · · 4rE(q) = ε,
4E(q) r
with the case of |q| > 1, and
I ! I
1 f (ζ) 2πi 1 f (ζ) − f (z)
dζ − f (z) = dζ
4E ′ (q) Cr ζ−z q ′
4E (q) Cr ζ−z
1 ε ′
< ′ · · 4rE (q) = ε,
4E (q) r

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25799

with the case of 0 < |q| < 1.

′
In summary, as E(q) and E (q) being non-zero constants for a determined q , 0,
I I
f (ζ) 2πi q f (ζ)
lim dζ = f (z) ⇔ f (z) = lim dζ.
r→0 C ζ − z q 2πi r→0 Cr ζ − z
r

According to Theorem B.3,

I I I
f (ζ) f (ζ) f (ζ)
dζ = dζ = dζ(0 < r < ρ).
C ζ −z Cρ ζ − z Cr ζ − z

Eventually to conclude with I

q f (ζ)
f (z) = dζ. (B.10)
2πi C ζ−z
□

B.4. Proof of the higher order derivative formula

Z
q f (ζ)
Proof. Consider first the case n = 1 where we need to prove that f (z) = ′
dζ. For any
2πi C (ζ − z)2
z ∈ D, ∃d > 0 such that U(z, 2d) ⊂ D. Let M = maxζ∈C {| f (ζ)|} and L be the arc length of C. When
0 < |h| < d,
|ζ − z| > 2d > d,
(

|ζ − z − h| ≥ |ζ − z| − |h| > d,
which leads to
f (z + h) − f (z)
I
q f (ζ)
− dζ
h 2πi C (ζ − z)2
" I I # I
1 q f (ζ) q f (ζ) q f (ζ)
= dζ − dζ − dζ
h 2πi C ζ − z − h 2πi C ζ − z 2πi C (ζ − z)2
I
q·h f (ζ) |q| · |h| M
= 2
dζ ≤ · 3 · L → 0 (h → 0).
2πi C (ζ − z − h)(ζ − z) 2π d
Thus,
f (z + h) − f (z)
I
′ q f (ζ)
f (z) = lim = dζ.
h→0 h 2πi C (ζ − z)2
Now, prove the general case using mathematical induction. Let the conclusion hold for n = k,
namely, I
q · k! f (ζ)
f (z) =
(k)
dζ (k = 1, 2, . . .). (B.11)
2πi C (ζ − z)k+1
Then, when n = k + 1,
f (k) (z + h) − f (k) (z) q · (k + 1)!
I
f (ζ)
− k+2
dζ
h 2πi C (ζ − z)
q · (k + 1)!
" I I # I
1 q · k! f (ζ) q · k! f (ζ) f (ζ)
= k+1
dζ − k+1
dζ − k+2
dζ
h 2πi C (ζ − z − h) 2πi C (ζ − z) 2πi C (ζ − z)
q · (k + 1)!
I " #
1 1
= f (ζ) − dζ + hO(1) → 0 (h → 0),
2πi C (ζ − z − h)k+1 (ζ − z) (ζ − z)k+2

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25800

which results in
f (k) (z + h) − f (k) (z) q · (k + 1)!
I
f (ζ)
f (k+1)
(z) = lim = dζ.
h→0 h 2πi C (ζ − z)k+2
The proposition is thus proved. □
Corollary B.4. Let the function f (z) be analytic in the region D, then, f (z) has derivatives with any
order in D.

C. Appendix 3: Fourier transform and Poisson summation formula

Any periodic function is capable of becoming the union of different sine functions, i.e.,
∞ !
X 2πnx
f (x) = A0 + An sin +φ , (C.1)
n=1
T

where T is the period and φ is the offset of the sine function, also known as the initial phase.

C.1. The Fourier series

Use of the formula
sin(α + β) = sin α cos β + sin β cos α
could simplify C.1 to obtain
∞ " ! !#
X 2πnx 2πnx
f (x) = A0 + An sin φ cos + An cos φ sin .
n=1
T T

an = An sin φ,


Then, apply the substitution: 
bn = An cos φ,

which results in the Fourier series
∞ " ! !#
X 2πnx 2πnx
f (x) = A0 + an cos + bn sin . (C.2)
n=1
T T

By means of Euler’s formula, we can solve for

1  i qz z q  i qz z
cos z = e + e−i q , sin z = e − e−i q . (C.3)
2 2i
And substituting it into Eq (C.2) yields
∞ " #
X an  qi 2πnx − qi 2πnx
 ibn  i 2πnx − qi 2πnx
f (x) = A0 + e T +e T − e q T −e T

n=1
2 2q
∞
 an − qi bn i 2πnx an + qi bn − i 2πnx 
 
X
= A0 +  · eq T + · e q T  .
n=1
2 2

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25801
a − i b
, n > 0;
 n q n

2



Now, apply the substitution : cn =  A0 , n = 0;



+ i

 a b
 n q n , n < 0,


2
which leads to the exponential form of the Fourier series
∞ 
X ∞
X
i 2πnx i 2πnx  i 2πnx
f (x) = c0 + cn e q T + c−n e q T =
−
cn e q T . (C.4)
n=1 n=−∞
 ∞ 
i 2πkx  X i 2πnx  i 2πkx
Thus, f (x)e− q =  cn e q T  · e− q T , which is then integrated over one of the periods of f (x).
T

n=−∞
And the integration and summation signs are interchangeable, according to the dominant convergence
theorem, so that Z x0 +T Z x0 +T X
∞
i 2π(n−k)x
− qi 2πkx
f (x)e T dx = cn e q T dx
x0 x0 n=−∞
∞ Z x0 +T
i 2π(n−k)x
X
= cn eq T dx.
n=−∞ x0
Z x0 +T
Provided that this integral converges absolutely, i.e., | f (x)| dx < ∞, we are left with that the
x0
integral can be simplified to
Z x0 +T Z x0 +T
i 2π(k−k)x
ck e q T dx = ck dx = ck T,
x0 x0

with the case of n = k, and to

Z x0 +T x0 +T
i 2π(n−k)x qT cn i 2π(n−k)x
cn eq T = eq T = 0,
x0 2πi(n − k) x0

with the case of n , k. Therefore,

Z x0 +T
1 i 2πkx
ck = f (x)e− q T dx, k ∈ Z. (C.5)
T x0

C.2. The Fourier transform

∞
X i 2πnt
Now, further on, study the Fourier series f (t) = cn e q T
, whose period is taken from -T/2 to
n=−∞
Z T
1 2 i 2πkt
T/2, where ck = f (t)e− q T dt.
T − T2
Z T !
2
− qi 2πξt 1 k
Suppose g(ξ) = f (t)e T dt, then, ck = g , and it follows that
− T2 T T
∞
X n 1
e2π q ( T )t ·
i n
f (t) = g . (C.6)
n=−∞
T T

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25802

Following this, we then replace n

T
with ξn and 1
T
with ∆ξ to get that

∞
X
g(ξn )e2π q ξn t ∆ξ.
i
f (t) = (C.7)
n=−∞

In order for the Fourier series to be used to represent functions without periods, we can find the
limit where ∆ξ tends to 0, i.e., T → ∞, which means
∞
X
g(ξn )e2π q ξn t ∆ξ.
i
f (t) = lim+ (C.8)
∆ξ→0
n=−∞

According to the theory of calculus, the right-hand side of the equation is found to be a Riemann
sum, and the series becomes an integral
Z ∞
g(ξ)e2π q ξt dξ.
i
f (t) = (C.9)
−∞

Following the same approach, we arrive at

Z ∞
f (t)e−2π q ξt dt.
i
g(ξ) = (C.10)
−∞

Hence, it is said that g(ξ) is the Fourier transform of f (t), that is,
Z ∞
f (t)e−2π q ξt dt.
i
g(ξ) = F { f (t)} = (C.11)
−∞

Sometimes, for simplicity, F { f (t)} is denoted directly by fˆ(ξ). And f (t) is the inverse Fourier
transform of g(ξ), namely,
Z ∞
g(ξ)e2π q ξt dξ,
i
f (t) = F {g(ξ)} =
−1
(C.12)
−∞

which, along with Eq (C.11), is called another form of the Fourier transform.

C.3. Poisson’s summation formula

A Fourier expansion of f (t) with period T yields
∞ ∞
(t+nT )
X i
X i
f (t + nT ) = ck e2π q k T = ck e2π q kt/T , (C.13)
k=−∞ k=−∞

where
Z t+nT + 12 T
1 i
ck = f (x)e−2π q kx/T dx. (C.14)
T t+nT − 12 T

AIMS Mathematics Volume 8, Issue 11, 25772–25803.

 25803
Z ∞
Let f (t) satisfy | f (x)|2 dx < ∞, then we can sum over f (t + nT ) to attain
−∞

∞
X ∞ X
X ∞
i
f (t + nT ) = ck e2π q kt/T
n=−∞ n=−∞ k=−∞
∞ X ∞ Z t+nT + 12 T
1 X
2π qi kt/T i
= e f (x)e−2π q kx/T dx
T n=−∞ k=−∞ t+nT − 21 T
∞ ∞ Z t+nT + 12 T
1 X
2π qi kt/T
X i
= e f (x)e−2π q kx/T dx,
T k=−∞ n=−∞ t+nT − 21 T
| {z }
S

where the part S of it can be further simplified. Due to the fact that for all n ∈ Z, the intervals
 T T
t + nT − , t + nT + will eventually merge without overlap into the integration interval from
2 2
−∞ to +∞, Z ∞ Z ∞ !
−2π qi kx/T −2π qi (k/T )x k
S = f (x)e dx = f (x)e dx = fˆ , (C.15)
−∞ −∞ T
which could be substituted back into the above summation formula to yields
∞ ∞
1 X 2π qi kt/T ∞
Z !
−2π qi kx/T 1 X ˆ k 2π qi kt/T
e f (x)e dx = f e , (C.16)
T k=−∞ −∞ T k=−∞ T

namely,
∞ ∞ !
X X 1 ˆ k 2π qi kt/T
f (t + nT ) = f e , (C.17)
n=−∞ k=−∞
T T
which is called Poisson’s summation formula on elliptic complex fields. In special, given that T = 1,
it follows that X X i
f (t + n) = fˆ(k)e q (2πkt) . (C.18)
n∈Z k∈Z

If let T = 1 and t = 0, then X X

f (n) = fˆ(k). (C.19)
n∈Z k∈Z

Many summation problems can be simplified by Poisson’s summation formula, especially for
exponential functions which can be transformed into problems of integration.
© 2023 the Author(s), licensee AIMS Press. This
is an open access article distributed under the
terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/4.0)

AIMS Mathematics Volume 8, Issue 11, 25772–25803.