International Journal of Pure and Applied Mathematics

Volume 84 No. 1 2013, 29-63

ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
doi: http://dx.doi.org/10.12732/ijpam.v84i1.3
AP
ijpam.eu

AN EXPLORATION ON GOLDBACH’S CONJECTURE

E. Markakis1 , C. Provatidis2 § , N. Markakis3

1 Vassilissis
Olgas 129B
54643, Thessaloniki, GREECE
2 National Technical University of Athens

9, Heroes of Polytechnion Ave., Zografou Campus

157 80, Athens, GREECE
3 Cram School “Methodiko”

Vouliagmenis and Kyprou 2,

16452 Argyroupolis, GREECE

Abstract: This paper divides the set of natural numbers in six equivalence
classes and determines two of them as candidate to include all prime numbers.
Concerning the even numbers themselves, these were divided into three subsets
using a basic cell (6n − 2, 6n and 6n + 2). Based on the aforementioned tools,
this paper proposes a deterministic process of finding all pairs (p, q) of odd
numbers (composites and primes) of natural numbers ≥ 3 whose sum (p + q) is
equal to a given even natural number 2n ≥6. Based on this procedure and also
relying on the distribution of primes in the set N of natural numbers, a closed
analytical formula is proposed for the estimation of the number of primes that
satisfy Goldbach’s conjecture for positive integers ≥6.

AMS Subject Classification: 11-XX

Key Words: Goldbach’s conjecture, prime numbers, statistics

c 2013 Academic Publications, Ltd.

Received: September 27, 2012 url: www.acadpubl.eu
§ Correspondence author
30 E. Markakis, C. Provatidis, N. Markakis

1. Introduction

As is known, on June 7, 1742, Christian Goldbach in a letter to Leonhard Euler

(see [1]) argued that, every even natural number greater than 4 can be written
as a sum of two primes, namely:

2n = p + q, where n > 2, and p, q are prime numbers. (1)

To save space, we do not use the original wording of that old time, but we
focus our attention on the modified formula (1) which is consistent with the
newest definition of the set of prime numbers P = {2,3,5,7,11,. . .}, which does
not include the unit.
On August 8, 1900, David Hilbert gave a famous speech during the second
International Congress of Mathematics in Paris, in which he proposed 23 prob-
lems for mathematicians of the 20th century, including Goldbach’s conjecture
(see [2]). Later, in 1912, Landau sorted four main problems for the prime num-
bers including Goldbach’s conjecture (see [3],[4]). The first scientific work on
Goldbach’s conjecture was made in the 1920’s. Note that in 1921, Hardy said
that Goldbach’s conjecture is not only the most famous and difficult problem
in number theory, but the whole of mathematics.
It is known that the most difficult so-called strong Goldbach conjecture was
preceded by important work in the so-called weak Goldbach conjecture. The
weak conjecture, which is known as the odd Goldbach conjecture or ternary
Goldbach problem or 3-primes problem, stated that: any number greater than 7
can be expressed as a sum of three primes (one prime number can be used more
than once in the same sum). The above assumption is called “weak” because
if the strong Goldbach’s conjecture (which concerns sums of two primes) is
proved, then the weak will be true.
The weak formulation of the conjecture has not been yet proven, but there
have been some useful although somewhat failed attempts. The first of these
works was in 1923 when, using the “circle method” and assuming the validity
of the hypothesis of a generalized Riemann, Hardy and Littlewood (see [5])
proved that every sufficiently large odd integer is sum of three odd primes and
almost all the even number is the sum of two primes. In 1919, Brun (see [6]),
using the method of his sieve proved that every large even number is the sum
of two numbers each of whom has at least nine factors of primes. Then in 1930,
using the Brun’s method along with his own idea of “density” of a sequence of
integers, Schnirelman (see [7]) proved that every sufficiently large integer is the
sum of maximum c primes for a given number c. Then in 1937, Vinogradov
(see [8]), using the circle method and his own method to estimate the expo-
AN EXPLORATION ON GOLDBACH’S CONJECTURE 31

nential sum in a variable prime number, was able to overcome the dependence
of the great Riemann hypothesis and thus provide the evidence of the findings
of Hardy and Littlewood now without conditions. In other words, he directly
proved (Vinogradov’s theorem) that all “sufficiently large” odd number can
be expressed as the sum of three primes. The original proof of Vinogradov,
based on inefficient theorem of Siegel-Walfisz, did not put a limit for the term
“sufficiently large” while his student Borozdkin (see [9]) showed in 1956 that
15
n0 =33 =314348907 is sufficiently large (has 6,846,169 digits). Later, after im-
provements in the method of Brun, in 1966 Jing-run Chen (see [10]) managed
to prove that every large integer is the sum of a prime and a product of at most
two primes. In 2002, Liu and Wang (see [11]) lowered the threshold around
n > e3100 ≈ 2 × 101346 . The exponent is too large to allow control of all smaller
numbers with the assistance of a digital computer. According to Internet re-
ports [12,13], the computer assisted search arrived for the strong Goldbach
conjecture up to order 1018 (http://www.ieeta.pt/∼tos/goldbach.html) and,
for the weak Goldbach conjecture not much more. In 1997, Deshouillers et al.
(see [14]) showed that the generalized Riemann hypothesis implies the weak
Goldbach’s conjecture for all numbers. Also, Kaniecki (see [15]) showed that
every odd number is the sum of at most five primes, provided the validity of
Riemann Hypothesis.
Most of these classic works have been included in a collective volume by
Wang (see [16]). Specifically, in that volume Section 1 includes the represen-
tation of an odd number as a sum of three primes in six papers (Hardy and
Littlewood; Vinogradov; Linnik; Pan; Vaughan; Deshouillers et al.), Section 2
includes the representation of an even number as a sum of two nearly primes
in six other works of (Brun; Buchstab; Kuhn; Selberg; Wang; Selberg) and
finally Section 3 includes the representation of an even number as a sum of
a prime and an almost prime in nine works (Renyi; Wang; Pan; Barban Til;
Buchstab; Vinogradov; Bombieri; Chen; Pan). Finally, apart from the indi-
vidual reports of certain articles, the collective volume includes 234 additional
citations arranged by author, referring to the period 1901-2001.
The strong formulation of Goldbach conjecture, which is the subject of this
paper, is much more difficult than the above weak one. Using the above method
of Vinogradov (see [8]), in separate works Chudakov (see [17]), van der Corput
(see [18]) and Estermann (see [19]) showed that almost all even number can be
written as a sum of two primes (in the sense that the fraction of even number
tends to the unit). As mentioned above, in 1930, Schnirelman (see [7]) showed
that every even number n ≥ 4 can be written as a sum of at most 20 primes.
This result in turn enriched by other authors; the most well-known result due to
32 E. Markakis, C. Provatidis, N. Markakis

Ramaré (see [20]) who in 1995 showed that every even number n ≥ 4 is indeed
a maximum sum of 6 primes. Indeed, resolving the weak Goldbach conjecture
will come through that every even number n ≥ 4 is the sum of at most 4 primes
(see [21]). In 1973, using sieve theory methods Jing-run Chen (see [22]) showed
that every sufficiently large even number can be written as a sum either of two
primes or of one prime and one semiprime (i.e. a product of two primes), e.g.
100 = 23 + 7·11. In 1975, Montgomery and Vaughan (see [23]) showed that
“most” even number is a sum of two primes. In fact, they showed that there
were positive constants c and C such that for all sufficiently large numbers N ,
every even number less than N is the sum of two primes with CN 1−c exceptions
at the most. In particular, all the even integers that are not sum of two primes
have zero density. Linnik (see [24]) proved, in 1951, the existence of a constant
K such that every sufficiently large even number is the sum of two primes and a
maximum of K powers of 2. Heath-Brown and Puchta (see [25]) in 2002 found
that the value K = 13 works well. The latter improved to K = 8 by Pintz and
Ruzsa (see [26]) in 2003.
It is noteworthy that, in 2000, the relation (1) was verified using computers
for even numbers up to 4 × 1016 (see [27]), and the attempt was repeated by T.
Oliveira e Silva with the help of distributed computing network to n ≤ 1.609 ×
1018 and in selected areas up to 4 × 1018 (see [13]). However, mathematically
these checks do not constitute conclusive evidence of validity of (1), and the
effort continues today [28].
In addition to the above papers, the interested reader can consult internet
sources [29-32]. Finally, Goldbach’s conjecture has been the subject of statisti-
cal approach [33], education [34], as well as narrative storytelling and popular
books [35-37].
In this paper we present a theoretical framework that provides an estimate
of the number of prime numbers satisfying Eq(1).

2. A Deterministic Procedure for the Representation of an Even

Number as Sum of Two Odds

By the definition of prime numbers, P = {2,3,5,7,11,. . .}, since they are divisible
only by themselves and the unit, it follows that “the only even prime number
is 2”. If for a moment we assume that p = 2, then the only case that fulfils
Eq(1) is when q = 2, otherwise the resulting sum would be equal to an odd
number. Because the resulting even number 4 (i.e. n = 2) is outside the scope
of interest (we care only for n > 2), it is obvious that Eq(1) makes sense only
AN EXPLORATION ON GOLDBACH’S CONJECTURE 33

for the odd primes, i.e. for all primes greater than 2, which obviously comprise
the set P − {2}.
In Subsections 2.1 and 2.2 we quote useful tools (for later use) and then in
2.3 we present a series of useful theorems.

2.1. Equivalence Classes

As known, an equivalence relation on the set A, such as e.g. K(α) = {x/x ∈

A ∧ xRα} divides the set A into subsets, named equivalence classes, which are
disjoint to each other and their union gives A.
Suppose Z is the set of integer numbers, Z2 its Cartesian (tensor) product
and R is the binary relation on the set Z2 , which is defined as follows:

R = {(α, β)/(α, β) ∈ Z2 and (α − β) is divisible by n ∈ N} (2)

that is (α − β) = λn or α = λn + β (the identity of division) and N is the

set of natural numbers.
The binary relation R is an equivalence relation because, as can be proven,
is reflexive, symmetric and transitive. The natural number n determines the
number of equivalence classes.
If we take the set of natural numbers N and define the divisor n = 6, then
N is divided into six equivalence classes, those of elements 0, 1, 2, 3, 4 and 5,
namely:

K(0) = {x/x = 6λ + 0, λ ∈ N} (3a)

K(1) = {x/x = 6λ + 1, λ ∈ N} (3b)
K(2) = {x/x = 6λ + 2, λ ∈ N} (3c)
K(3) = {x/x = 6λ + 3, λ ∈ N} (3d)
K(4) = {x/x = 6λ + 4, λ ∈ N} (3e)
K(5) = {x/x = 6λ + 5, λ ∈ N} (3f)

The above classes are disjoint each other and their union gives the set N, that
is:
K(0) ∪ K(1) ∪ K(2) ∪ K(3) ∪ K(4) ∪ K(5) = N (4)
Obviously, the set of natural numbers corresponding to the equivalence
classes K(2) and K(4) are divisible by 2; K(3) is divisible by 3, and K(0) is
divisible by both 2 and 3. Therefore the set of natural numbers contained in
the equivalence classes K(0), K(2), K(3) and K(4) are composite numbers as
being multiples of 2 and 3.
34 E. Markakis, C. Provatidis, N. Markakis

Conclusion-1: The remaining equivalence classes, K(1) and K(5), contain all
primes (except of 2 and 3) as well as the multiples of primes being > 3; these
can be combined in the formula:

6λ ± 1 = Primes + Multiples of Primes > 3 (5a)

In the representation of odd numbers according to Eq(5a), we achieved to con-
dense the feasible space of prime numbers in N by a factor of 3 (i.e. it is a
subset of two classes out of six).
For the sake of uniformity, the multiples of 3 will be denoted, as appropriate,
either as 6λ − 3 for λ = 1, 2, 3, . . . N , or like 6λ + 3 for λ = 0, 1, 2, . . . N , that is:
6λ ± 3 = Multiples of 3 (5b)

2.2. Even Numbers

From Number Theory we know that in any three consecutive even numbers one
of them is a multiple of 3, so this even number can be written as 3 · 2n =
6n. The triple of numbers (6n − 2, 6n, 6n + 2) are consecutive even numbers,
since they differ by 2 units. This implies that the subsets of even numbers
{x/x 6n − 2, 6n, 6n + 2, n ∈ N} are disjoint and their union is the set N2 of all
even numbers ≥ 4.
Conclusion-2: The set of even numbers ≥ 4 can be denoted with:

N2 = {x/x 6n − 2, 6n, 6n + 2} where n ≥ 1, (6)

and the above triple of even numbers is the basic cell to create the even numbers,
for n ≥ 1, in the sequence of our work.
If we replace n in Eq(6) with two natural numbers, i.e. λi and λj , where
(λi , λj ) ∈ N2 , such as:
n = λi + λj , (7)
the triple of the successive even numbers in the basic cell is written as:

6(λi + λj ) − 2 (8a)
6(λi + λj ) (8b)
6(λi + λj ) + 2. (8c)

Equations (8) may be further transformed as follows:

6(λi + λj ) − 2 = (6λi − 1) + (6λj − 1), (9a)

AN EXPLORATION ON GOLDBACH’S CONJECTURE 35

6(λi + λj ) = (6λi − 1) + (6λj + 1), (9b)

6(λi + λj ) = (6λi + 1) + (6λj − 1), (9c)
6(λi + λj ) + 2 = (6λi + 1) + (6λj + 1). (9d)

Conclusion-3: Thus symbolizing the even numbers, we managed to transform

all the even numbers into a sum of two odd numbers of the form 6λi ± 1, which,
as proved, either are primes or multiples of primes ≥ 5.
Of course the even numbers can be created also as a sum of two odd numbers
of which either one or both be multiples of 3. In this case however the even
numbers of the form 6n − 2 are generated solely as a sum according to Eq(10a),
the even numbers of the form 6n can be created only as a sum of two odd
multiples of 3 according to Eq(10b), whereas the even numbers of the form
6n + 2 are generated solely as a sum according to Eq(10c):

6n − 2 = (6λi + 1) + (6λj − 3) = 6(λi + λj ) − 2 (10a)

6n = (6λi + 3) + (6λj − 3) = 6(λi + λj ) (10b)
6n + 2 = (6λi − 1) + (6λj + 3) = 6(λi + λj ) + 2. (10c)

Equations (10) imply that, for each even number the pairs of odd numbers,
which are created on the basis of these equations, can give either zero or at
most only one pair prime-to-prime.
More specifically, given that the set of odd numbers being multiples of 3,
(6λ ± 3), from which the even numbers of the form 6n are formed, only the
number 3 is prime, this implies that under (10b), the only pair which verifies
Goldbach’s conjecture, is the “3+3=6”.
Based on Eq(10a), the even numbers of the form 6n − 2 verify Goldbach’s
conjecture only if (6λj − 3) = 3 while (6λi + 1) = prime.
Finally, the even numbers of the form 6n + 2 verify Goldbach’s conjecture,
by virtue of (10c), only if (6λj − 3) = 3 while (6λi − 1) = prime.

2.3. Three Basic Theorems

In the following we describe a deterministic procedure in which every even

natural number can be decomposed into all possible sums of odd integers, i.e.
primes or composites.
Theorem 1. Every even natural number 2n (independently of its specific
form 6n − 2, 6n, 6n + 2) can be decomposed into a sum of two odd natural
numbers (primes or composites) in so many different ways, ns , as the integer
part (floor) of the rational number (n − 1)/2, that is ⌊(n − 1)/2⌋. The index
36 E. Markakis, C. Provatidis, N. Markakis

“s” stands for the word “sample”, thus referring to a sample of ns odd numbers
from which will be later choose the prime numbers.

Proof. We distinguish two cases.

1) When n is odd, we form the sets:
A = {3, 5, . . . , n} and B = {2n − 3, 2n − 5, . . . , n}. Since the order of
items is not important in the sets, in order to maintain the desired sequence (in
→
−
the form of rows or columns) we form the vectors − →a = [3, 5, . . . , n] and b =
→
−
[2n − 3, 2n − 5, . . . , n]. It is obvious that all elements of the vector − →
c =− →
a+ b
are strictly defined and are equal to 2n as opposed to probabilistic pairs that
can be derived from the sets A and B. Also, it is evident that any enhancement
→
−
of the vector −→
a will give terms contained in the vector b , displayed from right
to left, so it makes no sense. Finally, it is obvious that the cardinality of two
sets is the same, i.e. cardA = cardB = (n − 1)/2.
2) When n is even, we form the sets:
A = {3, 5, . . . , n − 1} and B = {2n − 3, 2n − 5, . . . , n + 1}. As previously, we
→
−
consider the new vectors − →a ′ = [3, 5, . . . , n−1] and b ′ = [2n−3, 2n−5, . . . , n+1].
→
−
It is obvious that all elements of the vector − →c′ =−→a ′ + b ′ are again equal to
2n. As previously, any enhancement of the vector − →a ′ will give terms included
→′
−
into the vector b , displayed from right to left. Finally, it is obvious that the
cardinality of two sets is the same, i.e. cardA = cardB = n/2 − 1.
Summarizing the results of the two above cases, it is easily concluded that:

cardA = cardB = ⌊(n − 1)/2⌋ (11)

Conclusion-4: The higher an even number 2n is, the higher the number
of pairs of odd numbers ns .
Theorem 2. If the ns pairs of odd numbers (ps , qs ) involved in Theorem-
1, whose sum is equal to the even number 2n, are plotted in orthocanonical
system of axes ps qs , they will belong to a straight line which forms 45 degrees
to both axes ps and qs .

Proof. From Analytical Geometry we know that in a x-y system, every

straight line intersecting the x axis at the point A(α, 0) and the y axis at the
point B(0, β) satisfies the equation: x/α + y/β = 1. In our specific case, if we
select the x axis to represent the term ps ∈ N0 , while y axis to represent the term
qs ∈ N0 , then obviously it holds α = β = 2n (see Figure 1). It is also obvious
that the (2n + 1) integers of the interval [0, α] and the corresponding (2n + 1)
AN EXPLORATION ON GOLDBACH’S CONJECTURE 37

integers of the interval [0, β], of which sum equals to 2n, correspond to (2n + 1)
discrete points along the straight segment AB. Given that we are interested in
only the odd numbers ps , qs > 3 which satisfy Eq(1), without necessarily being
prime numbers, we must leave out the three pairs being closest to the x axis:
(2n, 0), (2n − 1, 1), (2n − 2, 2), as well as the three pairs closest to the y axis:
(0, 2n), (1, 2n−1), (2, 2n−2). These six points are denoted into Figure 1 by the
symbol (×). Therefore, the number of candidate points for further examination
is ns = ⌊(n − 1)/2⌋, two of which coincide with the ends C and D (note that
the set of all discrete points/pairs that correspond to even and odd integers is
2n − 5).

qs

β = 2n Β

D
P
Μ
P'

3 C
2
1 Α ps
O α = 2n
1 2 3

Figure 1: Diagram showing the interval CD on which lie the arranged

pairs (ps , qs ) which satisfy the relationship ps + qs = 2n, n > 2. To
avoid repeated pairs, we work only on the part DM, where M denotes
the common mid-point of the straight line segments AB and CD.

A better representation, especially for various small numbers is detailed in

Table 1. In full agreement with the immediately above, we observe that the
middle M appears to be among the candidate pairs only when the number n is
odd, i.e. for the even numbers: 6, 10, 14, ..., 50, and so on. But if we consider
the middle M of segment CD, each pair P(ps , qs ) has a corresponding equivalent
pair being represented by the symmetric point P′ (qs , ps ) of P with respect to
M. If we want to exclude the repetition of a pair (ps , qs ), then we can restrict
our attention only to half of CD, e.g. the segment DM. Concerning the point
M, it satisfies the relation ps + qs = 2n only when the number n is odd.

Theorem 3. Suppose the point P in Figure 1 starts from point D and

38 E. Markakis, C. Provatidis, N. Markakis

2ν+1 ν
25 12 28 30 32 34 36 38 40 42 44 46 48 50
23 11 26 28 30 32 34 36 38 40 42 44 46 48
21 10 24 26 28 30 32 34 36 38 40 42 44 46
19 9 22 24 26 28 30 32 34 36 38 40 42 44
17 8 20 22 24 26 28 30 32 34 36 38 40 42
15 7 18 20 22 24 26 28 30 32 34 36 38 40
13 6 16 18 20 22 24 26 28 30 32 34 36 38
11 5 14 16 18 20 22 24 26 28 30 32 34 36
9 4 12 14 16 18 20 22 24 26 28 30 32 34
7 3 10 12 14 16 18 20 22 24 26 28 30 32
5 2 8 10 12 14 16 18 20 22 24 26 28 30
3 1 6 8 10 12 14 16 18 20 22 24 26 28
0 1 2 3 4 5 6 7 8 9 10 11 12 ν
3 5 7 9 11 13 15 17 19 21 23 25 2ν+1

Table 1: Representative way of decomposing even numbers in a sum of

two odd ones. Every even integer has been put in the form 2n ≡ 2(ν+2),
and is produced by the sum of red odd numbers (2ν+1) of the horizontal
axis and the corresponding red values in the vertical axis. The colors
displayed in green, blue and magenta correspond to the even numbers
of the form 6n, 6n + 2 and 6n − 2, respectively (n ∈ N).

moves towards M traversing all distinct pairs (ps , qs ) on the straight segment
DM with ps ≤ qs . In this movement, two sets A and B of odd natural numbers
are created, of which the first (A) is formed by the values of ps while the second
(B) by the values of qs . Both sets have the same cardinality: ns = ⌊(n − 1)/2⌋,
which equals the number of distinct pairs that satisfy the relationship ps + qs =
2n. Assuming that the number of prime numbers in sets A and B are α1 and
β1 , respectively, and that the distribution of primes along the arithmetic line is
completely random, the number of pairs of primes (p, q) that satisfy Eq(1), is
approximated by:
np = (α1 β1 )/ns (12)

Proof. According to the theory of probability, in each pair (ps , qs ) selected

from the set of ns = ⌊(n − 1)/2⌋ elements, there are four possible events as
shown in Table 2.
If we select a random pair (ps , qs ), where ps ∈ A ∧ qs ∈ B, the probability
ps be prime is PA = α1 /ns , while the probability qs be prime is PB = β1 /ns .
Since the first event (ps =prime) is entirely independent on the second event
(qs =prime), the probability Pprime of the intersection of these two independent
AN EXPLORATION ON GOLDBACH’S CONJECTURE 39

Column A Column B Number of pairs

Prime Prime npp = (α1 β1 )/ns (13a)
Prime Composite npc = α1 (ns − β1 )/ns (13b)
Composite Prime ncp = (ns − α1 )β1 /ns (13c)
Composite Composite ncc = (ns − α1 )(ns − β1 )/ns (13d)
Sum of pairs ns

Table 2: All possible combinations for the formation of a pair (ps , qs )

from the columns A and B.

events equals to the product of their probabilities, that is:

Pprime = PA · PB = (α1 β1 )/n2s (14)

Since the probability Pprime on the sample of ns pairs is known, the number
np of pairs of prime numbers (p, q) will be also known and will be approximated
by the relationship:
np = Pprime · ns = (α1 β1 )/ns (15)
The relationship (15) completes the proof of Theorem-3.

It is noted that, generally, probabilistic analyzes are conducted under condi-

tions of uncertainty. When talking about probability, we refer to the realization
of one event in relation to other possible events. Obviously, the possibility to
verify Goldbach’s conjecture on the number of np pairs, belongs to the first
category (13a).
Remark. The distribution of prime numbers on the numerical line is not
accidental (stochastic) but strictly predetermined. It is therefore a purely de-
terministic phenomenon. The prime numbers are in a predetermined position,
waiting to be discovered.
Today, we know the number and distribution of prime numbers for a very
large number of natural numbers. Therefore, we know the number of primes,
both in column A (α1 ) and in column B (β1 ), arranged in ns pairs in which the
even numbers are decomposed on the basis of equations (9) and (10). So, we
are able to rigorously examine whether the predetermined distribution of prime
numbers is such as to ensure the validity of probabilistic relationship (13a),
“prime-to-prime” in a representative sample where the number of pairs ns is
small.
40 E. Markakis, C. Provatidis, N. Markakis

As an example, Table 3 represents the way in which we decompose a triple

(triad) of even numbers (6n − 2, 6n, 6n + 2), where n = 15. To facilitate dis-
crimination of pairs “prime-to-prime”, we encode the composite numbers with
gray color, the odd (prime numbers) of the form 6λ − 1 with turquoise color, the
odd (prime numbers) of the form 6λ + 1 with tile color, and finally the prime
number 3 with yellow color. Since even numbers of the form 6n are produced
on the basis of two equations, i.e. (9b) and (9c), while the even numbers of
either the form 6n − 2 or 6n + 2 are produced by only one equation [(9a) or
(9d), respectively], an asymmetry appears in the number of ordered pairs that
verify (13a), as shown in Table 3.
Figure 2 shows, with a remarkable correlation, the expected ‘coincidence’
of the curve which represents the number of cases that verify Goldbach’s con-
jecture, only for the even numbers of the form 6n (on one hand) and the sum of
cases for the even numbers in the form 6n − 2 and 6n + 2 (on the other hand).
Figure 3 shows, separately for each form of the triad of even numbers
(6n, 6n − 2, 6n + 2), the number of pairs (dyads) of prime numbers that verify
Goldbach’s conjecture. As previously observed, even in this case the anticipated
‘coincidence’ by equations (9a) and (9d) occurs.

To remove the above asymmetry, due to the four relations (9), which in
turn reflect the three types of even numbers, we transform (13a) by introducing
three weights whose sum is equal to 3 (the weight of 6n is twice the others),
and therefore adapt in more detail as follows:

For 6n − 2 : npp = 3/4(α1 β1 /ns ) (16a)

For 6n : npp = 3/2(α1 β1 /ns ) (16b)
For 6n + 2 : npp = 3/4(α1 β1 /ns ) (16c)

Therefore, if we assume that Goldbach’s conjecture is not verified for some

even numbers, these cases should be essentially searched within the even num-
bers of the form 6n−2 and 6n+2, which have the smallest values of verification,
those of Eq(16a) and Eq(16c).

3. The Distribution of Prime Numbers

It is known, by virtue of the Prime Number Theorem (PNT) [see Appendix]

that, as we move to larger numbers N , the density of primes 1/logN gradually
decreases. In the following, the symbol logx [or log(x)] is equivalent to the
AN EXPLORATION ON GOLDBACH’S CONJECTURE 41

6n-2 = 88 6n = 90 6n+2 = 92

COLUMN COLUMN COLUMN

Α Β Α Β Α Β
1 3 85 1 3 87 1 3 89
2 5 83 2 5 85 2 5 87
3 7 81 3 7 83 3 7 85
4 9 79 4 9 81 4 9 83
5 11 77 5 11 79 5 11 81
6 13 75 6 13 77 6 13 79
7 15 73 7 15 75 7 15 77
8 17 71 8 17 73 8 17 75
9 19 69 9 19 71 9 19 73
10 21 67 10 21 69 10 21 71
11 23 65 11 23 67 11 23 69
12 25 63 12 25 65 12 25 67
13 27 61 13 27 63 13 27 65
14 29 59 14 29 61 14 29 63
15 31 57 15 31 59 15 31 61
16 33 55 16 33 57 16 33 59
17 35 53 17 35 55 17 35 57
18 37 51 18 37 53 18 37 55
19 39 49 19 39 51 19 39 53
20 41 47 20 41 49 20 41 51
21 43 45 21 43 47 21 43 49
22 45 45 22 45 47

Sum of pairs that Sum of pairs that Sum of pairs that

fulfill Goldbach fulfill Goldbach fulfill Goldbach
conjecture: 4 conjecture: 9 conjecture: 4

Sum of Pairs: 21 Sum of Pairs: 22 Sum of Pairs: 22

[Equation (9a)] [Equations (9b) and [Equation (9d)]
(9c)]
Composite number
Odd prime in the form (6λ-1)
Odd prime in the form (6λ+1)
Prime number 3

(Sum of pairs fulfilling Golbach’s conjecture for the entire triad: 4+9+4 = 17)

Table 3: Example for the decomposition of a triad of numbers (6n −

2, 6n, 6n + 2), for n = 15, in sum of two odd numbers.

symbol lnx [or ln(x)] corresponding to the Neperian (natural) logarithm of

base e.
Here we should clarify that, despite the gradual decrease in the density of
the primes (see percentage in the fifth column of Table 4), the odd numbers
42 E. Markakis, C. Provatidis, N. Markakis

250
Dyads (6n)
Dyads (6n−2)+(6n+2)

200
Exact number of dyads of primes

150

100

50

0
0 1000 2000 3000 4000 5000 6000
6n

Figure 2: Number of pairs (dyads) of prime numbers that fulfill Gold-

bach’s conjecture for the lowest 6000 numbers, in two characteristic cat-
egories [magenta line: dyads 6n, red line: dyads (6n − 2) and (6n + 2)].

starting from the number 3 (from which we start the column A) until the even
number of 6n + 2 = 92 (see right column of Table 3) is 44. Within these 44
odd numbers, 23 of primes are contained. Thus the percentage of primes in
the ordered pairs of odd number in which the even number 92 is analyzed will
be 23 × 44/100 = 52.3%. This implies that in any distribution of prime and
composite numbers, in the ordered in pairs in which the even numbers between
6 and 92 are analyzed, the value of any prime-to-prime event will be larger
than 1. This means that Goldbach’s conjecture is a priori verified for the even
numbers with n ≤ 15.
Based on the default deterministic creation of ordered pairs, to ensure the
sum of all dyads be 6n − 2, and given the distribution of prime numbers in
between 3 and 6 × 15 − 2 = 88, the number npp of verifications is illustrated
in Table 4, where we present the data α1 , β1 and ns , involved in (16a) to
AN EXPLORATION ON GOLDBACH’S CONJECTURE 43

250
Dyads (6n)
Dyads (6n−2)
Dyads (6n+2)
200
Exact number of dyads of primes

150

100

50

0
0 1000 2000 3000 4000 5000 6000
6n

Figure 3: Number of dyads of prime numbers that fulfill Goldbach’s

conjecture for the smallest 6000 numbers, in three characteristic cate-
gories according to the cell (6n − 2, 6n, 6n + 2).

calculate npp, for even numbers of the form 6n − 2 with n varying from 2 to
15. A first observation arising from the data of Table 4 is that the density and
distribution of primes in that interval is such as to ensure in a great approach,
the coincidence of the values derived from (16a) with actual checks.
The density and distribution of prime numbers also ensure that α1 and β1
increase in such a way that the product (α1 β1 ) increases at a faster rate than
the rate at which the number ns of ordered pairs increases. Direct result of the
distribution of prime numbers is that as we move to larger even numbers, the
verification of Goldbach’s conjecture constantly increases, and is increasingly
removed far away from the value ‘1’ required by the conjecture.
Therefore, besides the percentage (%) which, in the test sample, ensures a
priori verification of Goldbach conjecture, it is also the distribution of prime
numbers, and more especially the distribution of the primes, which ensures the
44 E. Markakis, C. Provatidis, N. Markakis

n 6n − 2 α1 β1 Percentage ns Number True npp

(%) of verifica-
tions npp
according
to Eq(16a)
2 10 2 2 100 2 1.5 2
3 16 3 2 80 3 1.5 2
4 22 4 4 80 5 2.4 3
5 28 5 3 67 6 1.9 2
6 34 6 5 69 8 2.8 4
7 40 7 4 61 9 2.3 3
8 46 8 6 64 11 3.3 4
9 52 8 6 58 12 3.0 3
10 58 9 7 57 14 3.4 4
11 64 10 7 57 15 3.5 5
12 70 10 8 53 17 3.5 5
13 76 11 9 55 18 4.1 5
14 82 12 10 55 20 4.5 5
15 88 13 9 52 21 4.2 4

Table 4: Estimation of the number of verifications npp concerning Gold-

bach’s conjecture based on Eq(16a) compared with true values.

appropriate values in α1 and β1 so that the results derived from (16) to be in

a much closed agreement with the actual verification. This fact urges us to
investigate whether the distribution of prime numbers, in a larger and more
representative sample of even numbers, where the percentage (%) of the primes
with its continuous reduction, goes down below 50%, shall ensure, by itself, the
values for α1 and β1 which are also in agreement with the actual verifications.
Due to the reduced space, in Table 5 the following quantities are randomly
recorded:
-the values of α1 and β1 ,
-the number of columns ns
-and the verifications npp based on equations (16),
for several values of the even numbers in the form 6n − 2, 6n and 6n + 2 for
n = 20 until n = 2000, which correspond to the even numbers (118, 120, 122)
until (11998, 12000, 12002).
The even number 12002 includes 6001 odd numbers and 1437 primes. There-
fore, the percentage (in %) of the primes included in the 6001 odd numbers by
AN EXPLORATION ON GOLDBACH’S CONJECTURE 45

which we create the ordered pairs, will be 1437/6001 × 100 = 23.9%. Obvi-
ously, this percentage does not ensure the a-priori verification of Goldbach’s
conjecture.
In the sequence, Table 6 includes both the actual verifications of Goldbach’s
conjecture, in red, and the verifications based on formula (16); the latter are
due to the distribution of prime numbers (α1 , β1 ), in columns A and B, in
which each even number is decomposed according to Eq(9) and Eq(10), in blue
(already included in Table 5). From the results in Table 6, one can notice
that the distribution of prime numbers is enough to ensure the values of the
verification under Eq(16) to be in close accordance with the reality, as clearly
shown in Figure 4 (for these particular results only).

600
Exact
Eq(16)
500
Number of dyads of primes

400

300

200

100

0
0 2000 4000 6000 8000 10000 12000
6n

Figure 4: Number of pairs (dyads) of primes that fulfill Goldbach Con-

jecture for the smallest 12000 numbers (the red line represents the real
number of dyads of primes that fulfill Goldbach Conjecture, while the
blue line corresponds to the Eq(16)). The graph is based on Table 6
only.
46 E. Markakis, C. Provatidis, N. Markakis

For n = 20 α1 β1 ns npp [Eqs (16)]

6n-2=118 16 14 29 3/4α 1β1/ns = 3/4*16*14/29 = 5.8 : Eq 16(a)
6n=120 16 13 29 3/2α 1β1/ns = 3/4*16*13/29 = 10.7 : Eq 16(b)
6n+2=122 17 13 30 3/4α 1β1/ns = 3/4*17*13/30 = 5.5 : Eq 16(c)
Sum of triad = 21.9
For n = 40 α1 β1 ns
6n-2=238 29 21 59 3/4α 1β1/ns = 3/4*29*21/59 = 7.7
6n=240 29 21 59 3/2α 1β1/ns = 3/2*29*21/59 = 15.5
6n+2=242 29 21 60 3/4α 1β1/ns = 3/4*29*21/60 = 7.6
Sum of triad = 30.8
For n = 60 α1 β1 ns
6n-2=358 40 31 89 3/4α 1β1/ns = 3/4*40*31/89 = 10.4
6n=360 40 30 89 3/2α 1β1/ns = 3/2*40*30/89 = 20.2
6n+2=362 41 30 90 3/4α 1β1/ns = 3/4*41*30/90 = 10.2
Sum of triad = 40.8
For n = 80 α1 β1 ns
6n-2=478 51 40 119 3/4α 1β1/ns = 3/4*51*40/119 = 12.9
6n=480 51 39 119 3/2α 1β1/ns = 3/2*51*39/119 = 25.1
6n+2=482 52 39 120 3/4α 1β1/ns = 3/4*54*41/120 = 12.7
Sum of triad = 50.7
For n = 100 α1 β1 ns
6n-2=598 61 46 149 3/4α 1β1/ns =3/4*61*46/149=14.1
6n=600 61 46 149 3/2α 1β1/ns =3/2*61*46/149=28.2
6n+2=602 61 47 150 3/4α 1β1/ns =3/4*61*47/150=14.3
Sum of triad = 56.6
For n = 200 α1 β1 ns
6n-2=1198 108 88 299 3/4α 1β1/ns = 3/4*108*88/299 = 23.8
6n=1200 108 87 299 3/2α1β1/ns = 3/2*108*87/299 = 47.1
6n+2=1202 109 87 300 3/4α 1β1/ns = 3/4*109*87/300 = 23.7
Sum of triad = 94.6
For n = 300 α1 β1 ns
6n-2=1798 153 124 449 3/4α 1β1/ns =3/4*153*124/449=31.7
6n=1800 153 124 449 3/2α 1β1/ns =3/2*153*124/449=63.4
6n+2=1802 153 124 450 3/4α 1β1/ns =3/4*153*124/450=31.6
Sum of triad = 126.7
For n = 400 α1 β1 ns
6n-2=2398 195 160 599 3/4α 1β1/ns = 3/4*195*160/599 = 39.0
6n=2400 195 160 599 3/2α 1β1/ns = 3/2*195*160/599 = 78.1
6n+2=2402 196 161 600 3/4α 1β1/ns = 3/4*196*161/600 = 39.4
Sum of triad = 156.5
For n = 500 α1 β1 ns
6n-2=2998 238 191 749 3/4α 1β1/ns = 3/4*238*191/749 = 45.5
6n=3000 238 190 749 3/2α 1β1/ns = 3/2*238*191/749 = 91.0
6n+2=3002 238 191 750 3/4α 1β1/ns = 3/4*238*191/450 = 45.4
Sum of triad = 181.9
For n = 750 α1 β1 ns
6n-2=4498 333 276 1124 3/4α 1β1/ns = 3/4*333*276/1124 = 61.3
6n=4500 333 276 1124 3/2α 1β1/ns = 3/2*333*276/1124 = 126.6
6n+2=4502 334 276 1125 3/4α 1β1/ns = 3/4*334*276/1122 = 61.3
Sum of triad = 249.2
For n = 1000 α1 β1 ns
6n-2=5998 429 354 1499 3/4α 1β1/ns = 3/4*429*354/1499 = 76
6n=6000 429 354 1499 3/2α 1β1/ns = 3/2*429*354/1499 = 152
6n+2=6002 430 354 1500 3/4α 1β1/ns = 3/4*430*354/1500 = 76
Sum of triad = 304
For n = 1500 α1 β1 ns
6n-2=8998 609 506 2249 3/4α 1β1/ns = 3/4*609*506/2249 = 102.8
6n=9000 609 506 2249 3/2α 1β1/ns = 3/2*609*506/2249 = 205.5
6n+2=9002 609 507 2250 3/4α 1β1/ns = 3/4*609*507/2250 = 102.7
Sum of triad = 411
For n = 2000 α1 β1 ns
6n-2=11998 782 655 2999 3/4α 1β1/ns = 3/4*782*655/2999 = 128.1
6n=12000 782 655 2999 3/2α 1β1/ns = 3/2*782*655/2999 = 256.2
6n+2=12002 782 655 3000 3/4α 1β1/ns = 3/4*782*655/3000 = 128.0
Sum of triad = 512.3

Table 5: Estimation of the number of verifications npp of Goldbach’s

conjecture based on relations (16) for several values of even numbers in
the form 6n − 2, 6n, and 6n + 2.
AN EXPLORATION ON GOLDBACH’S CONJECTURE 47

n n = 20
Type of even number 6n-2 6n 6n+2 Sum
Even number 118 120 122
Verifications of GC (Facts) 6 12 4 22
Verifications using Eq(16) 5.8 10.7 5.5 21.9
n n = 40
Type of even number 6n-2 6n 6n+2 Sum
Even number 238 240 242
Verifications of GC (Facts) 9 18 7 34
Verifications using Eq(16) 7.7 15.5 7.6 30.8
n n = 60
Type of even number 6n-2 6n 6n+2 Sum
Even number 358 360 362
Verifications of GC (Facts) 10 22 7 39
Verifications using Eq(16) 10.4 20.2 10.2 40.8
n n = 80
Type of even number 6n-2 6n 6n+2 Sum
Even number 478 480 482
Verifications of GC (Facts) 11 29 10 50
Verifications using Eq(16) 12.9 25.1 12.7 50.7
n n = 100
Type of even number 6n-2 6n 6n+2 Sum
Even number 598 600 602
Verifications of GC (Facts) 15 32 11 58
Verifications using Eq(16) 14.1 28.2 14.3 56.8
n n = 200
Type of even number 6n-2 6n 6n+2 Sum
Even number 1198 1200 1202
Verifications of GC (Facts) 24 54 19 97
Verifications using Eq(16) 23.8 47.1 23.7 94.6
n n = 300
Type of even number 6n-2 6n 6n+2 Sum
Even number 1798 1800 1802
Verifications of GC (Facts) 27 74 31 132
Verifications using Eq(16) 31.7 63.4 31.6 126.7
n n = 400
Type of even number 6n-2 6n 6n+2 Sum
Even number 2398 2400 2402
Verifications of GC (Facts) 37 90 37 164
Verifications using Eq(16) 39.0 78.1 39.4 156.5
n n = 500
Type of even number 6n-2 6n 6n+2 Sum
Even number 2998 3000 3002
Verifications of GC (Facts) 46 103 39 188
Verifications using Eq(16) 45.5 91.0 45.4 181.9
n n = 750
Type of even number 6n-2 6n 6n+2 Sum
Even number 4498 4500 4502
Verifications of GC (Facts) 64 138 52 254
Verifications using Eq(16) 61.3 126.6 61.3 249.2
n n = 1000
Type of even number 6n-2 6n 6n+2 Sum
Even number 5998 6000 6002
Verifications of GC (Facts) 72 179 62 313
Verifications using Eq(16) 76.0 152.0 76.0 304.0
n n = 1500
Type of even number 6n-2 6n 6n+2 Sum
Even number 8998 9000 9002
Verifications of GC (Facts) 101 243 110 454
Verifications using Eq(16) 102.8 205.5 102.7 411.0
n n = 2000
Type of even number 6n-2 6n 6n+2 Sum
Even number 11998 12000 12002
Verifications of GC (Facts) 144 303 115 562
Verifications using Eq(16) 128.1 256.2 128.0 512.3

Table 6: Real verifications of Goldbach’s Conjecture-GC (in red colour)

compared with the verifications based on Eq(16) (in blue colour).
48 E. Markakis, C. Provatidis, N. Markakis

Also in this second sample of 12,000 numbers (6×2000 = 12000) the growth
rate of α1 and β1 is greater than the growth rate of the ns ordered pairs, so the
number of verifications npp of Eq(16) increases continuously and removes from
the unity (“1”), as is the case with actual verifications.
Comment. Among a probabilistic experiment and the creation of ordered
pairs when decomposing an even number in the sum of two odd numbers, there
is a substantial difference. In a probabilistic experiment, the likelihood that one
out of the four possibilities of Table 2 occurs is a matter of coincidence, or luck.
The order in which the various possibilities they appear in the potential space
is completely random. There is no rule which specifies the type of possibility
in a particular test. Completely different is the case of an even number’s de-
composition in a sum of two odd numbers with the creation of ordered pairs,
where the potential for each of the pairs are arranged, is strictly predetermined.
Therefore the analysis of an even number in sum of two odd numbers with the
creation of ordered pairs is purely deterministic.

4. The Critical Question

4.1. General

Generalization is a very important process in mathematics. We take a problem

and examine its behavior in a limited area, and then we try to expand the
conclusions arising from the study of this sample to larger areas. We studied
the behavior of a sample of all the even numbers up to 12,000 and found that
the number of pairs that verify Goldbach’s conjecture, have a clear upward
trend, which has a close relationship with the probabilistic equation (16).
Equation (16) imposes no restriction on the size of the sample. Not even the
deterministic decomposition of an even number in odd pairs, based on equations
(9) and (10), has a similar problem.
The crucial question that arises here is, whether for sufficiently large even
numbers the distribution of primes in the arithmetic line continues to be such
as to ensure similar behavior in deterministic ordered pairs in which sufficiently
large even numbers are analyzed, to that even numbers of the relatively small
sample that we previously looked, so as to legitimize to generalize to all the
even numbers.
Our belief is strongly ’YES’, since, as we explain below, the creation of
composite numbers on the numeric (arithmetic) line, is determined by strictly
defined rules, uniform for the whole crowd of natural numbers. Hence the dis-
AN EXPLORATION ON GOLDBACH’S CONJECTURE 49

E3 = 3× 3 5 7 9 11 13 15 17 ... 2n + 1 ...
E5 = 5× 5 7 9 11 13 15 17 19 ... 2n + 1 ...
E7 = 7× 7 9 11 13 15 17 19 21 ... 2n + 1 ...
E9 = 9× 9 11 13 15 17 19 21 23 ... 2n + 1 ...
E11 = 11× 11 13 15 17 19 21 23 25 ... 2n + 1 ...
E13 = 13× 13 15 17 19 21 23 25 27 ... 2n + 1 ...

Table 7: Decomposition of Composite Numbers (CN) in a product of

two odd numbers.

tribution of prime numbers, which is formed by the relationship:

Prime numbers = Natural numbers - Composite numbers has to follow
the same inevitability and it is strictly prescribed. The view that the distribu-
tion of primes along the arithmetic line is random and chaotic is believed to be
wrong and misleading.

4.2. Basic Rule for the Creation of Composite Numbers

As the prime numbers (except of 2) are odd, we will seek the prime numbers
in the set of odd numbers N1 = {x/x = 2n + 1, n ∈ N}. This set, which can be
written as N1 = {α1 , α2 , α3 , . . . αn , . . .}, is a countable set equivalent to N.
The two factors in the product (β · υ) in which every odd composite is
analyzed, should be odd, because only the product of odd numbers results in an
odd number. Therefore, the composite numbers can be partitioned into subsets
of the form:

Ei = {x/x = (2n + 1) × [(2n + 1), (2n + 3), . . . (2n + 2k + 1) . . .]} (17)

Such composite odd numbers are those of Table 7, where we see that the
smallest of the infinite odd composite numbers generated in each row is the
product of the first number in the series, by itself, i.e. 3 × 3, 5 × 5, 7 × 7, 9 × 9,
11 × 11, . . .
In each of the above infinite series of composite odds, which are multiples
of 3, 5, 7, 9, etc., we have the first number to be a “square” (32 , 52 , 72 , . . .),
whereas everyone else is a “rectangle”. Obviously, the “squares” of prime num-
bers cannot be transformed into “rectangles” and will be called the “original
squares”. Unlike the squares of prime numbers, squares of composite numbers,
such as the 9 × 9, can be also transformed into rectangles 3 × 27, which we call
“abusive squares”.
50 E. Markakis, C. Provatidis, N. Markakis

Easily see that the elements of the set E3 = 3 × (3, 5, 7, . . . 2n + 1) in the

first row of Table 7, which are multiples of 3, is a periodic ‘phenomenon’ in the
set of integer numbers and may be derived from formula (18):

E3 : (CN )3 = x = 9 + 6µ = 32 + 2 × 3µ, µ = 0, 1, 2, . . . (18)

For µ = 0 : x = 9 = 3 × 3
For µ = 1 : x = 15 = 3 × 5
For µ = 2 : x = 21 = 3 × 7
For µ = 3 : x = 27 = 3 × 9, e.t.c.
As a periodic phenomenon, it can also be represented as a transverse wave,
as follows. In a rectangular system of axes xOy, we identify the straight line
that represents the set of natural numbers, with the axis Ox. For the sake of
clarity of shape, we ignore the even numbers, and we indicate only the odd ones.
If one constructs a transverse wave that starts at the number 9, of wavelength
λ = 12 units in the set of natural numbers N or λ′ = 12/2 = 6 units in the
set of odd numbers N1 , the Composite Numbers (CN )3 described by equation
(18) coincide with the intersections of the transverse wave and the axis Ox
(zero point deviation from the axis Ox in the direction of y axis, as shown in
Figure 5b). The appearance of composite numbers that are multiples of 3 on
the arithmetic line is the most common of any other odd number; it divides
the sum of odd pairs N1 in infinite pairs of consecutive odd numbers, starting
with the square of 3 and reaching the utmost ends of the arithmetic line of
Figure 5a. These successive pairs, of which, as we have explained, their first
number is of the form (6λ − 1) while the second form (6λ + 1), will be either
primes or multiple of primes > 3. All these pairs are candidates to become the
Twin Primes 1 , as long as none of the two numbers of the pair is crossed by a
subsequent wave, thus remain to be primes, as shown in Figure 5b.
But also the elements of the set E5 = 5 × (5, 7, 9, 11, 13, . . . , 2κ + 1 . . .)
constitute a periodic ‘phenomenon’ in the set of odd numbers N1 , starting from
25, with a wavelength λ = 45 − 25 = 20 or λ′ = 20/2 = 10. Formula (19) gives
the multiples of 5 that are transverse sections of the wave with the axis Ox,
which apparently are composite numbers:

E5 : (CN )5 = x = 25 + 10µ = 52 + 2 × 5µ, µ = 0, 1, 2, . . . (19)

Quite similarly, the elements of E7 = 7 × (7, 9, 11, 13, 15, . . . , 2κ + 1 . . .) are also
periodic at a frequency 14µ, where µ ∈ N1 . The first term is 49 and the formula
1
A twin prime is a prime number that differs from another prime number by two, for
example the twin prime pair (3, 5).
 AN EXPLORATION ON GOLDBACH’S CONJECTURE 51

that gives the multiples of 7 is

E7 : (CN )7 = x = 49 + 14µ = 72 + 2 × 7µ, µ = 0, 1, 2, . . . (20)

1,5
Waves from primes

0,5

0 n
0 100 200 300 400 500 600 700 800 900 1000
-0,5

-1

-1,5 3^2

(a)

1,5
3^2
Waves from primes

0,5

0 n
3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91
-0,5

-1

-1,5
3^2

(b)

Figure 5: Formation of composite numbers starting with the square of

3 until (a) 1000 and (b) detail until 91.

Generalizing, we get a similar rationale in the calculation of the formula

which gives us all the multiples of the odd number (2n + 1):

E2n+1 : (CN )2n+1 = (2n + 1)2 + 2(2n + 1)µ, µ = 0, 1, 2, . . . (21)

From Eq(21) we conclude that, as we move to increasingly odd numbers, so

the composite numbers (CN) set up by their respective primes are ever fewer
because, firstly the start of a wave is shifted to the right of the numerical line,
and secondly the wavelength increases.
Through a set of thousands, millions, (even for infinite remote parts of
arithmetic line) waves with different wavelengths and the phase difference, some
numbers of the numerical line, which is located in the heart of the package, are
left untouched by all this multitude of waves. This means that there are some
NP natural numbers which remain primes, because apparently the following
relationship is satisfied:
(NP − Pi2 )/2Pi µi ∈
/Z (22)
52 E. Markakis, C. Provatidis, N. Markakis

√
where Pi consecutive primes ≤ N . That means all the following conditions
are fulfilled:

(NP − 25)/10µi , (NP − 49)/14µi , · · · , (NP − P i2 )/2Pi µi ∈

/ Z. (23)

These numbers, being infinite and endless, as Euclid proved using the “Re-
ductio ad absurdum” method, are the Prime Numbers (PN) that appear in Fig-
ure 6.
3
Waves from primes

0 n
9 109 209 309 409 509 609 709 809 909
-1

-2

-3
3^2 5^2 7^2 11^2 13^2 17^2 9^2 15^2

(a)
3
2,5
Waves from primes

2
1,5
1
0,5 n
0
-0,5 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91
-1
-1,5
-2
3^2 5^2 7^2 9^2

(b)

Figure 6: (a) Waves proceeding from the primes (3, 5, 7, 11, 13, 17) as
well as the composites (9,15). (b) Detail for the primes (3, 5, 7) and
the composite number 9.

As we can see from Figure 6, only the waves starting at the squares of
primes (original squares) create composite numbers. In contrast, the waves
starting from composite numbers, having a growing density as we move to
larger numbers, do not create new composite numbers. That’s why we call
them “sterile waves”.
If therefore from the set N1 of odd numbers we create subsets Ai , having as
first term of each successive subset the squares of all odd numbers and last term
the odd number which is the next smallest square of the next odd, namely:

Ai = {x/x : (2i + 1)2 ≤ x < (2i + 3)2 , x = odd}, (24)

such a subset will include (2i + 3)2 − (2i + 1)2 = 8(i + 1) terms that belong
to the set of natural numbers N. Therefore, the number of terms in the set N1
AN EXPLORATION ON GOLDBACH’S CONJECTURE 53

of odd ones will be:

N1 = 8(i + 1)/2 = 4(i + 1) where i ∈ N. (25)

Due to the way of their creation (from the square of the odd 2i + 1 up to
the previous odd number than the square of 2i + 3), the subsets Ai are disjoint,
and the union of all these subsets equals the set N1 of the odd numbers:

A0 ∪ A1 ∪ A2 ∪ A3 · · · = N1 (26)

where (card = cardinality):

cardA0 = 4, and cardAi+1 − cardAi = 4, i = 1, 2, . . . (27)

In Table 8 we present these subsets of the odd numbers Ai from 1 to 1087

(for n = 0 to n = 15), which correspond to the squares of the odd numbers that
√
are smaller or equal to 1087, that is 31. In these subsets for easier distinction,
the composites are encoded in ‘turquoise’ while the primes in ‘black’ colour.
By the term “wave number” mentioned in the sequence, we mean the num-
ber of waves which propagate through the particular subset. The first subset,
1 to 7 (n = 0), is not traversed by any wave. The second, 9 to 23, (n = 1), is
traversed by a wave that gives the multiples of 3. The third, 25 to 47 (n = 2),
by two waves that give the multiples of 3 and 5, and so on.
According to the above, the conversion of some natural numbers to com-
posites and the distribution of the arithmetic line is governed by specific rules
that are uniform for all natural numbers. The procedure is deterministic and
therefore independent of sample size.
From what we have presented here the validity of the statements below is
obvious.
Sentence-1: The relation between the first term x0 of every subset An and
its ascending number n is: x0 = (2n + 1)2 . At the same time, the symbol n
represents also the number of waves that transverse the concrete subset (An ).
Sentence-2: Every new subset Ai has 4 terms (odd numbers) more than
its previous subset, Ai−1 . See also Eq(27).
Sentence-3: In every new subset Ai only one new wave acts, starting from
(2n + 1)2 and step (half wave) 2(2n + 1). Since the term (2n + 1) is a prime
number, every new wave creates composite numbers using the formula (21),
that is, (CN )(2n+1) = (2n + 1)2 + 2(2n + 1)µ.
Sentence-4: Easily proved that for any n, in the concrete wave µ may
take only the values µ = 0, 1 and 2 (three intersections of the transverse wave
with the axis Ox for each new subset). Since the term (2n + 1) is a composite
54 E. Markakis, C. Provatidis, N. Markakis

For n = 0 => A0 : Sum of odd terms = 4(n +1) = 4 => Number of waves n = 0: {001 003 005 007}
For n = 1 => A1 : Sum of odd terms = 4(n +1) = 8 => Number of waves n = 1: {009 011 013 015 017 |
019 021 023}
For n = 2 => A2: Sum of odd terms = 4(n +1) =12 => Number of waves n = 2
{025 027 029 031 033 | 035 037 039 041 043 | 045 047}
For n = 3 => A3: Sum of odd terms = 4(n +1) =16 Number of waves n = 3
{049 051 053 055 057 | 059 061 063 065 067 | 069 071 073 075 077 | 079
For n = 4 => A4: Sum of odd terms = 4(n +1) =20 Number of waves n = 4
{081 083 085 087 089 | 091 093 095 097 099 | 101 103 105 107 109 | 111 113 115 117 119}
For n = 5 =>A5: Sum of odd terms = 4(n +1) =24 Number of waves n = 5
{121 123 125 127 129 | 131 133 135 137 139 | 141 143 145 147 149 | 151 153 155 157 159 |
161 163 165 167}
For n = 6 =>A6: Sum of odd terms = 4(n +1) =28 Number of waves n = 6
{169 171 173 175 177 | 179 181 183 185 187 | 189 191 193 195 197 | 199 201 203 205 207
209 211 213 215 217 | 219 221 223}

For n = 7 => A7: Sum of odd terms = 4(n +1) =32 Number of waves n = 7
{225 227 229 231 233 |235 237 239 241 243 | 245 247 249 251 253 | 255 257 259 261 263 |
265 267 269 271 273 | 275 277 279 281 283| 285 287 }
For n = 8 => A8: Sum of odd terms = 4(n +1) =36 Number of waves n = 8
{289 291 293 295 297| 299 301 303 305 307| 309 311 313 315 317|319 321 323 325 327
329 331 333 335 337 | 339 341 343 345 347 | 349 351 353 355 357 | 359}

For n = 9 => A8: Sum of odd terms = 4(n +1) =40 Number of waves n = 9
{361 363 365 367 369 | 371 373 375 377 379 | 381 383 385 387 391 | 389 393 395 397 399 |
401 403 405 407 409 | 411 413 415 417 419 | 421 423 425 427 429 | 431 433 435 437 439}
For n = 10 =>A10: Sum of odd terms = 4(n +1) =44 Number of waves n = 10
{441 443 445 447 449 | 451 453 455 457 459 | 461 463 465 467 469 | 471 473 475 477 479 |
481 483 485 487 489| 491 493 495 497 499 | 501 503 505 507 509| 511 513 515 517 519 |
521 523 525 527}
For n = 11 =>A11: Sum of odd terms = 4(n +1) =48 Number of waves n = 11
{529 531 533 535 537| 539 541 543 545 547| 549 551 553 555 557|559 561 563 565 567 |
569 571 573 575 577| 579 581 583 585 587| 589 591 593 595 597| 599 601 603 605 607|
609 611 613 615 617* 619 621 623 }
For n = 12=>A12: Sum of odd terms = 4(n +1) =52 Number of waves n = 12
{625 627 629 631 633| 635 637 639 641 643| 645 647 649 651 653| 655 657 659 661 663|
665 667 669 671 673| 675 677 679 681 683|685 687 689 691 693 | 695 697 699 701 703|
705 707 709 711 713* 715 717 719 721 723* 725 727}
For n =13 => A13: Sum of odd terms = 4(n +1) =56 Number of waves n = 13
{729 731 733 735 737| 739 741 743 745 747| 749 751 753 755 757|759 761 763 765 767|
769 771 773 775 777| 779 781 783 785 787| 789 791 793 795 797| 799 801 803 805 807|
809 811 813 815 817| 819 821 823 825 827| 829 831 833 835 837| 839 }
For n =14 => A14: Sum of odd terms = 4(n +1) =60 Number of waves n = 14
{841 843 845 847 849| 851 853 855 857 859| 861 863 865 867 869|871 873 875 877 879|
881 883 885 887 889| 891 893 895 897 899| 901 903 905 907 909| 911 913 915 917 919|
921 923 925 927 929|931 933 935 937 939| 941 943 945 947 949| 951 953 955 957 959| }
For n =15 => A15: Sum of odd terms = 4(n +1) =64 Number of waves n = 15
{961 963 965 967 969| 971 973 975 977 979| 981 983 985 987 989| 991 993 995 997 999|
1001 003 005 007 009| 011 013 015 017 019| 021 023 025 027 029| 031 033 035 037 039|
041 043 045 047 049| 051 053 055 057 059| 061 063 065 067 069| 071 073 075 077 079|
081 083 085 087}

Table 8: Subsets of odd numbers Ai (from n = 0 to n = 15), the

composites encoded in turquoise colour while the primes in black. The
vertical bar (|) appears only for visual purposes, every five numbers.
AN EXPLORATION ON GOLDBACH’S CONJECTURE 55

number, it does not create new composites based on the formula: (CN )(2n+1) =
(2n + 1)2 + 2(2n + 1)µ.
Sentence-5: In this way the complex numbers are created, each new wave
will form 0, 1, or at most 2 new composite numbers, in the first subset it acts.
The value of µ = 2 coincides with the values that are multiples of 3.
Sentence-6: From the above analysis, it is showed that the creation of
composite numbers follows a uniform determinism, from the first up to the last
subset Ai we created, using the formula Ai = {x/x = (2n + 1)2 , x < (2n + 3)2 }
(24) . Therefore, both the crowd and the distribution of prime numbers will
also be deterministically defined and uniform throughout the set of natural
numbers.
Finding: Concerning the distribution of prime numbers, in the extended
sample that we examined (and which we can expand indefinitely) is such that
it follows very closely the probabilistic relationships (16) and this distribution
is strictly deterministically defined for all of natural numbers, we have the right
to extend the validity of (16) for the entire set of natural numbers.

5. Extension of Equation (13a) for the Entire set of Natural

Numbers

Theorem 4. If we call p(N ) the function that counts the number of

ordered pairs of primes that fulfill Goldbach’s conjecture for a natural number
N , this crowd is approximated by the formula:
p(N ) = N/[log(N/2)]2 (28)
Proof. Let us consider a sufficiently large even number N . According to
the Prime Number Theorem (PNT), the primes that exist in the set from 1 to
N , are approximated by:
π(N ) = N/log(N ) (29)
If we decompose this even number in ordered pairs of odd numbers according
to equations (9) and (10), the odd numbers that correspond to the column A,
and are in increasing order, will be all less than N/2. Therefore, the primes α1 ,
which exist in column A, will be given by:
α1 = π(N/2) = (N/2)/[log(N/2)]. (30)
As a result, the primes β1 that exist in the column B, will be the subtraction
of (30) from (29):
β1 = N/log(N ) − (N/2)/log(N/2). (31)
56 E. Markakis, C. Provatidis, N. Markakis

But N/log(N ) can be written as N/log[2 · (N/2)] or N/[log(N/2) + log2], and

for a sufficiently large N the latter is approximated by N/log(N/2), for which
Eq(31) becomes:
β1 ∼
= N/log(N/2) − (N/2)/log(N/2) = (N/2)/[log(N/2)]. (32)
Thus Eq(30) and Eq(32) suggest that, for very large even numbers N , the
primes α1 in column A and the primes β1 in column B, tend to become equal
(α1 ∼
= β1 ).
From what we have mentioned for the crowd ns of ordered pairs in which
the even number N is decomposed, if it is put in the form 6n − 2 and 6n it will
be (N/4) − 1, while if it is put in the form 6n + 2 it will be N/4. For very large
N , we can consider all three cases with ns = N/4, without causing any serious
mistake.
Substituting the above values of α1 , β1 and ns into Eq(13a), we obtain:
npp ∼
= (N/2)/[log(N/2)]2 /(N/4). (33)
If both parts of Eq(33) are mupltiplied by ns ∼
= N/4, then Eq(28) is ob-
tained. This completes the proof of Theorem 4.

The graphical representation of Eq(28) is illustrated by the green line in

Figure 7.
In Figure 7 we observe that, the graph of Eq(28) representing the number
of verifications of Goldbach’s conjecture is below the corresponding number us-
ing Eq(13a) and is located just below the average line of true verifications (red
line). Note that the red line of Figure 7 does not correspond to the red line in
Figure 2, but the union of sets of values corresponding together to the purple
and red line in Fig.2, which corresponds to the sum of true pairs. This is a
natural consequence of the fact that the number of prime numbers calculated
from (PNT), is below the true number of primes π(N ), as shown in Figure 8.

6. Monotonically Increasing Function

Theorem 5. The estimator of the verifications p(N ) of Goldbach’s con-

jecture is a monotonically increasing function of N .

Proof. We consider the continuous function p(x) = x/[log(x/2)]2 . The first

and second derivatives are given by:
p′ (x) = [log(x/2) − 2]/[log(x/2)]3 (34)
AN EXPLORATION ON GOLDBACH’S CONJECTURE 57

250
REAL
Eq(13a)
Eq(28)
200
Number of dyads of primes

150

100

50

0
0 1000 2000 3000 4000 5000 6000
6n

Figure 7: Comparison between the true number of pairs of prime num-

bers that fulfill Goldbach’s conjecture and the approximations using
eq(13a) and eq(28).

p′′ (x) = −2[log(x/2) − 3]/x[log(x/2)]4 (35)

Since for the first derivative it obviously holds that:

p′ (x) > 0, ∀x > 2e2 , (36)

while the second derivative never becomes zero :

p′′ (x) > 0, ∀x > 2e3 , (37)

it is finally concluded that the function p(x) is monotonically increasing and

convex.
58 E. Markakis, C. Provatidis, N. Markakis

180
Li(x)
160 True
x/log(x)
140

120
Number of primes

100

80

60

40

20

0
0 200 400 600 800 1000
x

Figure 8: Prime Number Theorem [red line: according to the function

Li(x), Blue line: true number, Green line: according to x/log(x)].

7. Discussion

This paper includes ten original points as follows:

1) The definition of six equivalence classes in N through Eq(3) is original.
The profit from their definition is that, in this way we achieved to condense the
feasible set of primes and their multiples in one-third, through the expression
6λ ± 1.
2) Concerning the even numbers themselves, we divided them into three
subsets, i.e. like a basic cell (6n − 2, 6n, 6n + 2). These subsets were further
put as a sum of two odd numbers of the form 6λ ± 1 (equations 9, 10).
3) For the three even numbers that appear in the above cell (6n−2, 6n, 6n+
2), we found that for odd numbers in the form of the centered numbers, i.e. 6n,
the probability to fulfill Goldbach’s conjecture is about twice as that for either
6n − 2 or 6n + 2.
AN EXPLORATION ON GOLDBACH’S CONJECTURE 59

4) The first three theorems (Theorem-1, Theorem-2 and Theorem-3), through

which we determine the number of couples of odd numbers (not necessarily
primes) in which the even numbers are decomposed, are new.
5) In contrast to the established opinion that the distribution of the primes
along the arithmetic line is random, we depicted that it is a severely determin-
istic ‘phenomenon’.
6) We found that the density and distribution of prime numbers ensure that
α1 and β1 increase in such a way that, the product (α1 β1 ) increases faster than
what the total number of couples ns increases, thus resulting in a continuously
increasing number of verifications of Goldbach’s conjecture.
7) We refined a rule (sieve) to create Composite Numbers, which is based
on waves of prescribed starting point and predetermined wave-length (equa-
tion(21)). We established that between all the aforementioned waves there
are some numbers that remain untouched. This means that there are natural
numbers NP that remain primes, obviously because the following relationship
√
is fulfilled: (NP − Pi2 )/2Pi µi , where Pi are successive primes ≤ N such as
5, 7, 11, etc. Also, using Eq(22) is helpful to make the decision whether any
arbitrary natural number N is prime or not.
8) The extension of Eq(13a) to the entire set of natural numbers is novel.
9) The closed form expression for the estimation of pairs that fulfill Gold-
bach’s conjecture.
10) We proved that the function p(N ) is monotonically increasing.

8. Conclusion

We proved a stronger sentence than the original Goldbach’s conjecture, which

can be formulated as follows:
Final Sentence. The strictly predetermined distribution of primes along
the arithmetic line is such as to ensure that as we move towards larger and
larger even numbers N , the growth of the product (α1 β1 ) is greater than the
growth rate of the number of ordered pairs of odd numbers ns in which the
even number under question is analyzed using equations (9) and (10), where
α1 is the number of prime numbers in column A and β1 is the number of prime
numbers in column B of ordered pairs. The direct result of this property of
prime numbers is that:
Not only Goldbach’s conjecture is true, but the number of verifications has a
clearly increasing trend, as we move to larger even numbers, which is determined
by relation (28).
60 E. Markakis, C. Provatidis, N. Markakis

References

[1] Christian Goldbach, Letter to L. Euler, June 7 (1742).

[2] David Hilbert, Mathematical Problems, Bulletin of the American Math-

ematical Society, 8, No. 10 (1902), 437-479. Earlier publications (in the
original German) appeared in Göttinger Nachrichten (1900), 253-297, and
Archiv der Mathematik und Physik, 3dser., 1 (1901), 44-63 and 213-237.

[3] János Pintz, Landau’s problems on primes, Journal de théorie des nombres
de Bordeaux, 21, No. 2 (2009) 357-404.

[4] Richard K. Guy, Unsolved problems in number theory, Third Edition,

Springer-Verlag, Berlin (2004).

[5] G. H. Hardy and J. E. Littlewood, Some problems of ‘Partitio numerorum’;

III: On the expression of a number as a sum of primes, Acta Mathematica,
44, No. 1 (1923), 1-70.

[6] Viggo Brun, Le crible d’Eratosthene et le theoreme de Goldbach, C. R.

Acad. Sci., Paris, 168 (1919), 544-546.

[7] L. G. Schnirelman, On additive properties of numbers. Izv. Donsk. Po-

litehn. Inst. 14 (1930), 328 (In Russian); Also: L. G. Schnirelman, Über
additive Eigenschaften von Zahlen, Math. Ann., 107 (1933), 649-690.

[8] I. M. Vinogradov, Some theorems in analytic theory of numbers, Dokl.

Akad. Nauk SSSR, 4 (1934), 185-187. Also: I. M. Vinogradov, Elements of
Number Theory, Dover Publications, Mineola, NY (2003).

[9] K. G. Borozdkin, On a problem of Vinogradov’s constant, Trudy Mat. Soc,

SSSR, 1 (1956), 3.

[10] Jingrun Chen, On the representation of a large even integer as the sum
of a prime and the product of at most 2 primes, it Kexue Tongbao, 17
(1966), 385-386.

[11] M. C. Liu, T. Z. Wang, On the Vinogradov’s bound in the three primes

Goldbach conjecture, Acta Arithmetica, 105, No. 2 (2002), 133-175.

[12] N. Lygeros, F. Morain, O. Rozier, http://www.lix.polytechnique.fr/∼morain

/Primes/myprimes.html

[13] Tomás Oliveira e Silva, http://www.ieeta.pt/∼tos/goldbach.html

AN EXPLORATION ON GOLDBACH’S CONJECTURE 61

[14] J.-M. Deshouillers, G. Effinger, H. Te Riele and D. Zinoviev, A complete

Vinogradov 3-primes theorem under the Riemann hypothesis. Electronic
Research Announcements of the American Mathematical Society, 3, No. 15
(1997), 99-104.

[15] Leszek Kaniecki, On Šnirelman’s constant under the Riemann hypothesis,

Acta Arithmetica, 4 (1995), 361-374.

[16] Yuan Wang, Goldbach Conjecture, 2nd edn., World Scientific Publishing,
New Jersey, Singapore (2000).

[17] Nikolai G. Chudakov, On the Goldbach problem, Doklady Akademii Nauk

SSSR, 17 (1937), 335-338.

[18] J. G. Van der Corput, Sur l’hypothèse de Goldbach, Proc. Akad. Wet.
Amsterdam, 41 (1938), 76-80.

[19] T. Estermann, On Goldbach’s problem: proof that almost all even positive
integers are sums of two primes, Proc. London Math. Soc., Sér. 2, 44
(1938), 307-314.

[20] Ramar, Olivier, On Šnirel’man’s constant, Annali della Scuola Normale

Superiore di Pisa - Classe di Scienze, Sér. 4, 22, No. 4 (1995), 645-706.

[21] Matti K. Sinisalo, Checking the Goldbach Conjecture up to 4 · 1011 , Math-

ematics of Computation, 61, No. 204 (1993), 931-934.

[22] J. R. Chen, On the representation of a larger even integer as the sum of

a prime and the product of at most two primes, Sci. Sinica, 16 (1973),
157-176.

[23] H.L. Montgomery, R.C. Vaughan, The exceptional set in Goldbach’s prob-
lem, Acta Arith., 27 (1975), 353-370.

[24] Yu. V. Linnik, Prime numbers and powers of two, Collection of articles. To
the sixtieth birthday of academician Ivan Matveevich Vinogradov, Trudy
Mat. Inst. Steklov., 38, Acad. Sci. USSR, Moscow (1951), 152-169.

[25] D. R. Heath-Brown, J. C. Puchta, Integers represented as a sum of primes

and powers of two, Asian Journal of Mathematics, 6, No. 3 (2002), 535-565.

[26] J.Pintz, I. Z. Ruzsa, On Linnik’s approximation to Goldbach’s problem, I,

Acta Arithmetica, 109, No. 2 (2003), 169-194.
62 E. Markakis, C. Provatidis, N. Markakis

[27] Jörg Richstein, Verifying the Goldbach conjecture up to 4 · 1014 , Mathe-

matics of Computation, 70, No. 236 (2000), 1745-1749.

[28] I. Mittas, Generalization of Goldbach’s conjecture and some special cases,

Italian Journal of Pure and Applied Mathematics, 27 (2010), 241-254.

[29] http://en.wikipedia.org/wiki/Goldbach’s weak conjecture

[30] Eric W. Weisstein, Goldbach Conjecture. From MathWorld - A Wolfram

Web Resource, http://mathworld.wolfram.com/GoldbachConjecture.html

[31] http://jtnb.cedram.org/item?id=JTNB 2009 21 2 357 0

[32] http://jtnb.cedram.org/cedram-bin/article/JTNB 2009 21 2 357 0.pdf

[33] Neil Sheldon, A statistician’s approach to Goldbachs Conjecture, Teaching

Statistics 25, No. 1 (2003), 12-13.

[34] Song Y. Yan, A simple verification method for the Goldbach conjecture,
International Journal of Mathematical Education in Science and Technol-
ogy, 25, No. 5 (1994), 681-688.

[35] John Derbyshire, Prime Obsession: Bernhard Riemann and the Greatest
Unsolved Problem in Mathematics, Joseph Henry Press, Washington, D.C.
(2003).

[36] Apostolos Doxiadis, Uncle Petros and Goldbach’s Conjecture, Bloomsbury,

New York (2001).

[37] Marcus du Sautoy, The Music of the Primes: Searching to Solve the Great-
est Mystery in Mathematics, HarperCollins Publishers, New York (2003).

[38] John Harrison, Formalizing an Analytic Proof of the Prime Number The-
orem, J Autom Reasoning, 43 (2009), 243-261.

[39] D. J. Newman, Simple Analytic Proof of the Prime Number Theorem, The
American Mathematical Monthly, 87, No. 9 (1980), 693-696.
AN EXPLORATION ON GOLDBACH’S CONJECTURE 63

Appendix A

Prime Number Theorem (PNT). Let π(x) be the prime-counting func-

tion that gives the number of primes less than or equal to x, for any real number
x. The prime number theorem then states that the limit of the quotient of the
two functions π(x) and x/ln(x) ≡ x/log(x) as x approaches infinity is 1, which
is expressed by the formula:

π(x)
lim =1 (38)
x→∞ x/logx

known as “the asymptotic law of distribution of prime numbers”. Using asymp-

totic notation this result can be restated as:

π(x) ≃ x/log(x) (39)

This theorem does not say anything about the limit of the difference of the
two functions as x approaches infinity. Instead, the theorem states that x/ln(x)
approximates π(x) in the sense that the relative error of this approximation
approaches 0 as x approaches infinity.
The prime number theorem is equivalent to the statement that the nth
prime number pn is approximately equal to nln(n), again with the relative
error of this approximation approaching 0 as n approaches infinity.
For analytical proofs we refer to [38,39] among others.
Corrolary. The average density of primes is given by: π(N ) ∼ 1/logN .
Obviously, it decreases by increasing N . As a result, the density of primes in
the column A is greater than that of column B and at the same time the density
of primes in the two columns is different.
64