A Proof of Goldbach Conjecture

Jinzhu Han1 Zaizhu Han2

1. Baotou North Hospital, Inner Mongolia, China
2. State Key Laboratory of Cognitive Neuroscience and Learning,
Beijing Normal University, China

Abstract
In this paper, the Goldbach Conjecture﹛1, 1﹜ is proved by the complex variable
integration. To prove the conjecture, a new function is introduced into Dirichlet series.
And then, by using the Perron Formula of Dirichlet Series and the Residue Theorem, we
conclude that any larger even integer can be expressed as the sum of two primes.

Notation：
p prime number
N positive integer

pN p exactly divides N

( p, N ) = 1 coprime between p and N

﹛1, 1﹜ the even integer as the sum of two primes

﹛a, b﹜ the even integer as the sum of product of at most a primes and product of at
most b primes

φ ( n) the Euler function

s = σ + it complex variable
∞
ζ (s) ζ ( s ) = ∑ n − s , the Riemann zeta function
n =1

ζ′ ζ′ ζ ′( s)
(s) (s) =
ζ ζ ζ (s)

⎧ log p, if n = p m
Λ (n) Λ ( n) = ⎨ , the Mangoldt function
⎩0, otherwise
π ( x) π ( x ) = ∑1
p≤ x

θ ( x) θ ( x ) = ∑ log p
p≤ x

ψ ( x) ψ ( x) = ∑ Λ ( n)
n≤ x

-1-
 ⎛ 1 ⎞ p −1
C(N) C ( N ) = ∏ ⎜1 − 2 ⎟∏
p>2 ⎝ ( p − 1) ⎠ p N p − 2
p>2

σc the convergence abscissa of Dirichlet series

σa the absolute convergent abscissa of Dirichlet series

resf ( s) the residue of function f ( s )

N ( f (s) ) the number of zero point of function f ( s )

B = O( A), B A there exists a calculable positive constant c , such that B ≤ cA

o (1) the constant tending to zero

γ the Euler constant, γ = 0.577 ⋅⋅⋅

ε the sufficiently small positive constant
c calculable positive constant

1. Introduction
The Goldbach Conjecture ﹛1, 1﹜ is one of the most famous still unresolved problems in
mathematics. The conjecture states that every even integer beyond 2 may be the sum of two prime
numbers (e.g., 12 = 5 + 7, 20 = 3 + 17). In over past two hundreds years, several relevant proofs
on this issue have been conducted. Using the sieve method, some mathematicians verified the
results including﹛9, 9﹜(Brun, 1920) [1], ﹛1, c﹜(R´enyi, 1947) [2], ﹛1, 5﹜(Pan, 1962) [3],
﹛1, 4﹜(Wang, 1962) [4], ﹛1, 3﹜(Richert, 1969) [5], ﹛1, 2﹜(Chen, 1973) [6], etc. In 1975,
Montgomery and Vaughan made the progress on the exceptional set in Goldbach’s problem by the
circle method [7]. However,﹛1, 1﹜ has not been proven up to now. In fact, one can enumerate
that ﹛1, 1﹜ is established for small values, but there is not enough evidence to demonstrate that
﹛1, 1﹜ is also established for larger values. In this case, once we can prove that any larger even
integer can be written as the sum of two primes, ﹛1, 1﹜ is entirely verified. In current paper, by
applying the complex variable integration, we are to prove that any larger even integer can
decompose into the sum of two primes.
First, we define a new function
⎧⎪0, if n ≡ N mod p
λ ( n) = ⎨ p≤ N
, (1.1)
⎪⎩ 1, otherwise

where ( p, N ) = 1 , p = {3,5, 7,…} is the prime sequence. Hence, this function is

-2-
provided with properties of sieve function.

Based on the series: ζ ( s ) , π ( x ) , θ ( x ) and ψ ( x ) , we obtain the new following

series:
∞
ζ ( s, λ ) = ∑ λ ( n ) n − s , (1.2)
n =1

π ( x, λ ) = ∑ λ ( p ) , (1.3)
p≤ x

θ ( x, λ ) = ∑ λ ( p ) log p , (1.4)
p≤ x

ψ ( x, λ ) = ∑ λ ( n ) Λ ( n ) , (1.5)
n≤ x

and
∞
Ψ ( s, λ ) = ∑ λ ( n ) Λ ( n ) n − s . (1.6)
n =1

In this case, if x = N , such that π ( N , λ ) > 0 , then ﹛1, 1﹜ is true.

By the Prime Number Theorem [8], we have

θ ( x, λ ) = π ( x, λ ) log x − ∫ π ( u , λ )u −1du ,
x
(1.7)
2

π ( x, λ ) = θ ( x, λ )( log x ) + ∫ θ ( x, λ ) ( u log 2 u ) du ,
−1 x −1
(1.8)
2

and

ψ ( x, λ ) = θ ( x, λ ) + O x ( ).
1
2
(1.9)

So we can estimate the value of function π ( x, λ ) by function ψ ( x, λ ) .

In present paper, by using both the Perron Formula of Dirichlet Series and the Residue
Theorem, we will verify the following theorems.
Theorem 1. If N is any larger even integer, then

ψ ( N , λ ) = 4e−γ (1 + o (1) ) C ( N )
N
log N
+ O N e− c ( log N
). (1.10)

Theorem 2. If N is any larger even integer, then

π ( N , λ ) = 4e−γ (1 + o (1) ) C ( N )
N
2
log N
+ O Ne − c ( log N
). (1.11)

Remark: The propositions (1.10) and (1.11) are equivalent.

By Theorem 2 we have π ( N , λ ) > 0 , the Goldbach Conjecture (i.e., {1, 1}) is established.
To prove Theorem 1 and Theorem 2, we need the following lemmas.

-3-
2. Lemmas
Lemma 1. The Perron Formula of Dirichlet Series.
For the Dirichlet series
∞
A ( s ) = ∑ a ( n )n − s , σ a < +∞ , (2.1)
n =1

if there exist increasing functions H ( u ) and B ( u ) , such that

a ( n ) ≤ H ( n ) , n = 1, 2,3… , (2.2)

and
∞

∑ a ( n ) n σ ≤ B (σ ) ,
n =1
−
σ > σa , (2.3)

for any s0 = σ 0 + it0 , let b0 ≥ σ 0 + b > σ a , T ≥ 1 , x ≥ 1 (when x is a positive integer), then

1 1 b +iT xs
∑ a (n) n
n≤ x
− s0
+ a ( x) x =
2
− s0

2π i ∫b −iT
A ( s0 + s ) ds
s

⎛ xb B ( b + σ 0 ) ⎞ ⎛ 1−σ ⎛ log x ⎞ ⎞
+O ⎜ ⎟ +O ⎜ x 0 H ( 2 x ) min ⎜1, ⎟⎟ , (2.4)
⎝ T ⎠ ⎝ ⎝ T ⎠⎠

the constant implied in O depends on σ a and b0 .

Let s0 = 0 , then

1 1 b +iT xs ⎛ xb B ( b ) ⎞
∑ a ( )
n +
2
a ( )
x =
2π i ∫b −iT
A ( )
s
s
ds + O ⎜ ⎟
n≤ x ⎝ T ⎠

⎛ ⎛ log x ⎞ ⎞
+O ⎜ xH ( 2 x ) min ⎜1, ⎟⎟ (2.5)
⎝ ⎝ T ⎠⎠
(for the proof of Lemma 1, see [8]).

Lemma 2. There exists a positive constant c1 , such that ζ (σ + it , λ ) has not zero point in the
range:

σ ≥ 1 − c1 log −1 ( t + 2 ) (2.6)

Proof. Let ζ ( s, λ ) be defined by

∞
ζ ( s, λ ) = ∑ (1 − λ ( n ) )n − s . (2.7)
n =1

When σ ≥ 1 − c1 log
−1
( t + 2 ) , we obtain
-4-
 ζ ( s , λ ) ≤ ζ ( s, λ ) , (2.8)

and

ζ ( s, λ ) + ζ ( s , λ ) = ζ ( s ) . (2.9)

By the Rouché Theorem, we deduce

N ( ζ ( s, λ ) ) = N (ζ ( s ) ) . (2.10)

Since ζ (σ + it ) has not zero point in the range σ ≥ 1 − c1 log −1 ( t + 2 ) , then ζ (σ + it , λ )

has not zero point in the same range σ ≥ 1 − c1 log −1 ( t + 2 ) too.

This proves Lemma 2.
∞
Lemma 3. For Dirichlet series Ψ ( s, λ ) = ∑ λ (n) Λ (n) n
n =1
−s
, we have

σ c = σ a = 1. (2.11)

Proof. By the Prime Number Theorem for Arithmetic Progressions, when x → ∞ and

( q, N ) = 1 , we have
x
lim
x →∞
∑
n ≡ N mod q
Λ (n) =
φ (q)
(2.12)
n≤ x

⎛ 1 ⎞
lim ∑ λ ( n ) Λ ( n ) = x ∏ ⎜⎜1 − ⎟⎟
⎝ φ ( p) ⎠
x →∞
n≤ x p≤ N
( p , N ) =1

⎛ 1 ⎞
=x ∏ ⎜1 − ⎟
p −1 ⎠
p≤ N ⎝
( p , N ) =1

⎛ ⎛ 1 ⎞⎞ x
= 4e − γ C ( N ) ⎜ 1 + O ⎜ ⎟⎟ , (2.13)
⎝ ⎝ log N ⎠ ⎠ log N
Hence
x
σ c = lim log −1 x log ∑ λ ( n ) Λ ( n ) = 1 . (2.14)
x →∞
n =1

Since λ ( n) ≥ 0 , σ c = σ a = 1.
This proves Lemma 3.
∞
Lemma 4. At s = 1 , Ψ ( s, λ ) = ∑ λ (n) Λ (n) n
n =1
−s
exists a pole, and the residue of function

-5-
Ψ ( s, λ ) is that

⎛ ⎛ 1 ⎞⎞ 1
res Ψ ( s, λ ) = 4e −γ C ( N ) ⎜1 + O ⎜ ⎟⎟ . (2.15)
s =1
⎝ ⎝ log N ⎠ ⎠ log N
Proof. Because
∞
lim (1 − s ) ∑ Λ ( n ) n −1 = 1 , (2.16)
s →1
n =1

and
1 ∞
lim
x →∞
∑n≤ x
Λ ( n ) n −1 = ∑
φ ( q ) n =1
Λ ( n ) n −1 , (2.17)
n ≡ N mod q

consequently

⎛ 1 ⎞∞
lim ∑ λ ( n )Λ ( n ) n −1 = ∏ ⎜⎜ 1 − ⎟⎟∑ Λ ( n ) n
−1

⎝ φ ( p ) ⎠ n =1
x →∞
n≤ x p≤ N
( p , N ) =1

⎛ 1 ⎞∞
∏ ⎟∑ Λ ( n ) n
−1
= ⎜ 1 −
p≤ N ⎝ p − 1 ⎠ n =1
( p , N ) =1

⎛ ⎛ 1 ⎞⎞ 1 ∞
= 4e − γ C ( N ) ⎜ 1 + O ⎜ ⎟⎟ ∑ Λ (n) n −1
. (2.18)
⎝ ⎝ log N ⎠ ⎠ log N n =1

Therefore
∞
res Ψ ( s, λ ) = lim (1 − s ) ∑ λ ( n )Λ ( n ) n −1
s =1 s →1
n =1

⎛ ⎛ 1 ⎞⎞ 1 ∞
= 4e − λ C ( N ) ⎜ 1 + O ⎜ ⎟⎟ lim
s →1
(1 − s ) ∑ Λ ( n ) n −1
⎝ ⎝ log N ⎠⎠ log N n =1

⎛ ⎛ 1 ⎞⎞ 1
= 4e − γ C ( N ) ⎜ 1 + O ⎜ ⎟⎟ . (2.19)
⎝ ⎝ log N ⎠ ⎠ log N
Lemma 4 is proved.

3. Proof of Theorem 1
For Dirichlet series (1.6)
∞
Ψ ( s, λ ) = ∑ λ ( n ) Λ ( n ) n − s ,
n =1

(T + 2 ) , b = 1 + log −1 x , log T = ( log x ) 1+α ( 0 < α < 1) , we have

1
−1
let a = 1 − c1 log

-6-
 H ( u ) ≤ log u , (3.1)

and

B ( u ) ≤ c2 log x , (3.2)

where c2 is a positive constant.

By (2.5) in Lemma 1, we have

ψ ( x, λ ) = ∑ λ ( n ) Λ ( n )
n≤ x

1 b +iT xs
( ) ds + R1 ,
2π i ∫b −iT
= Ψ s , λ (3.3)
s
where

x log 2 x
R1 ≤ xτ log 2 x, 0 < τ < 1 . (3.4)
T
By letting a ± iT , b ± iT to be the fixed points for closed contour Γ , we then have

1 xs
Ψ ( s, λ ) ds + R1 + R2 ,
2π i ∫Γ
ψ ( x, λ ) = (3.5)
s
where

R2
1
2π i (∫ a − iT

b − iT
+∫
a + iT

a − iT
+∫
b + iT

a + iT ) Ψ ( s, λ )
xs
s
ds

(∫ ) ∑ Λ ( n) n
∞
1 a −iT a + iT b + iT xs
≤ +∫ +∫ −s
ds
2π i b −iT a −iT a + iT
n =1 s

=
1
2π i (∫ a − iT

b − iT
+∫
a + iT

a − iT
+∫
b + iT

a + iT ) −
ζ′
ζ
s
xs
( ) ds
s

1
2π i (∫ a − iT

b −iT
+∫
a + iT

a −iT
+∫
b + iT

a + iT ) log 2 t
xs
s
ds

xe− c log x
. (3.6)

∞
ζ′
Remark: When σ > 1 − c1 log 2 t , ∑ Λ ( n) n
n =1
−s
=−
ζ
(σ + it ) log 2 t (see [8]).

By Lemma 2 and Lemma 4, ζ ( s, λ ) has not zero point in closed contour Γ . So function

xs
Ψ ( s, λ ) only exists a pole at s = 1 , we obtain
s

-7-
 xs 1 xs
res Ψ ( s, λ ) ( ) ds
s 2π i ∫Γ
= Ψ s , λ
s =1 s

⎛ ⎛ 1 ⎞⎞ x
= 4e − γ C ( N ) ⎜ 1 + O ⎜ ⎟⎟ . (3.7)
⎝ ⎝ log N ⎠ ⎠ log N
By (3.5) and (3.7), we have

⎛ ⎛ 1 ⎞⎞ x
ψ ( x , λ ) = 4e − γ C ( N ) ⎜ 1 + O ⎜ ⎟⎟ + R1 + R2 . (3.8)
⎝ ⎝ log N ⎠ ⎠ log N
Thus, combining (3.8) with (3.4) and (3.6) we have

⎛ 1 ⎞⎞ x
ψ ( x , λ ) = 4e − γ C ( N ) ⎜ 1 + O ⎜
⎝
⎛
⎝ log N
⎟⎟
⎠⎠ log N
+ O xe − c ( log x
). (3.9)

Let x = N , and N is any larger number, we then obtain (1.10), namely,

ψ ( N , λ ) = 4e−γ (1 + o (1) ) C ( N )
N
log N
+ O N e− c ( log N
).
This completely proves Theorem 1.

4. Proof of Theorem 2
Combining (3.9) with (1.7), (1.8) and (1.9) we have

π ( x, λ ) = 4e−γ C ( N ) (1 + o (1) )
x
+ R3 (4.1)
log N log x
where

x ψ ( u, λ ) xe− c log x x ψ (u )
∫ + ≤ ∫ du + xe− c log x
R3 2
du 2
. (4.2)
2 u log u log x 2 u log u
By the Prime Number Theorem

(
ψ ( x ) = x + O xe− c log x
), (4.3)

we have

x ψ (u ) x u x ue
− c log u

∫ ∫ + ∫2 u log 2 u du xe− c log x

du du . (4.4)
2 u log 2 u 2 u log 2 u
By (4.1), (4.2) and (4.4), we have

π ( x, λ ) = 4e−γ C ( N ) (1 + o (1) )
x
log N log x
+ O xe− c ( log x
). (4.5)

Let x = N , and N is any larger number, we then obtain (1.11), i.e.,

π ( N , λ ) = 4e−γ (1 + o (1) ) C ( N )
N
2
log N
+ O Ne − c ( log N
).

-8-
 Theorem 2 is proved. Therefore, we have π ( N , λ ) > 0 . Thus, the Goldbach Conjecture
﹛1, 1﹜ is established.

References
[1] V. Brun, De cribe d'Eratosth6ne et le th6orhme de Goldbach, Videnskabs-selskabt: Kristiania
Skrifter I, Mate-Naturvidenskapelig Klasse. 3 (1920) 1–36.
[2] A. R´enyi. On the representation of an even number as the sum of a single prime and a single
almost-prime number. Dokl. Akad. Nauk SSSR. 56 (1947), 455–458.
[3] C.T. Pan, On the representation of an even number as the sum of a prime and of an almost
prime, Acta Mathematica Sinica. 12 (1962) 95–106.
[4] Y. Wang, On the representation of a large integer as the sum of a prime and of an almost prime,
Sci. Sin. 11 (1962) 1033–1054.
[5] H.E. Richert, Selberg’s sieve with weights, Mathemetika. 16 (1969) 1–22.
[6] J.R. Chen, On the representation of a large even integer as the sum of a prime and the product
of at most two primes, Sci. Sin. 17 (1973) 157–176.
[7] H.L. Montgomery, R.C. Vaughan, The exceptional set in Goldbach’s problem, Acta. Arith. 27
(1975) 353–370.
[8] C.D. Pan, C.B. Pan, Basic Analytic Number Theory, Science Press, Beijing, 1999.

-9-