Tải bản đầy đủ (.pdf) (48 trang)

Posthumous works - Chứng minh bài toán Ferma lớn

Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (527.08 KB, 48 trang )



N.V. TỪ-YÊN
*



ŒUVRES POSTHUMES

Posthumous works


*







HÀ-NỘI , VIỆT-NAM

MAI 2013

*


Ecce homo

Paroles de Pilate aux Juifs ,
Saint- Jean, XIX ,5.



Homo sapiens – demens
From Anthropology .
















TABLE DES MATIERES

1 - An elementary proof of the Fermat’s last theorem . ….
2 - Une preuve élémentaire de l’infinité des nombres
premiers jumeaux (p,p+2) et des paires voisines de
nombres premiers jumeaux (p,p+2; p+6,p+8) . ….
3 - Essai d’une preuve élémentaire de la conjecture de
Goldbach .

*
Le détail du déroulement de ces recherches

peut être trouvé dans mon autobiographie
“ Méditation posthume ”, 2007 .









à la mémoire de mes parents Hân-Đoàn
à la mémoire de ma femme Thọ

à ma fille Thu-Uyên et ma petite fille Hạnh-Duyên
*

PREFACE



Voici le bilan des recherches non-professionnelles d’un amateur durant
presque tout le temps libre de sa vie. Ce bilan a été récapitulé et écrit dans la
79-ième année de cet homo sapiens-demens pour le soumettre aux
professionnels. Non-professionnel et amateur, sapiens et demens – c’est le
trait général de l’auteur que le lecteur doive le méditer.

Que ces recherches tombent aux mains d’un examinateur de haute
responsabilité, qui n’est point allergique sur des preuves dites élémentaires, et
surtout , sur des chercheurs au nom des amateurs.


Que ces mémoires rencontrent les lecteurs curieux et méditatifs.

Que j’aime exprimer ma profonde reconnaissance à cet examinateur et à
ces lecteurs .

*

This is the researche’s recapitulation of an amateur in almost the spare
time of his life. The review is recapituled and written at the age of 79 by this
homo sapiens- demens with a view to submit to the professionals .
Unprofessional and amateur, sapiens and demens this is the trait general
of the author, on which the reader necessitate to meditate.

Let these researches fall into the hand of one reviewer of high responsibility,
who is not allergic to all elementary proofs for the difficult mathematic
problems, and more particularly , to all researchers named unprofessional .

Let these works are coming with the curious and meditative readers.

Let me express my heartfelt gratitude to this reviewer and to these readers.

*



















































MÉMOIRE 1
*
- soumis le 27 Juillet 1990
- mon père est décédé le 13 Octobre 1989
à mon âge de 54 .

* * *


1

AN ELEMENTARY PROOF OF
THE FERMAT’S LAST THEOREM


N V. TU -YEN
( Vietnam )


ABSTRACT- To examine the structural properties of the solutions ,
if existent , of the Diophantine equation : ( 0 ) a
pair of the two procedures ( I )
where x , ; x odd, even and
nnn
zyx =+
i
x
in
n
i
i
n
nn
x
exCxex

=

+=+
1
)(
+
∈ Zne
x
,
x
e
( II )
where y , ; y even, odd is proposed.

i
y
in
n
i
i
n
nn
y
eyCyey

=

+=+
1
)(
+
∈ Zne
y
,
y
e
The fives structural properties of these procedures reveal a
secret that if the eq.( 0 ) has solution, then the existence of one
consecutive solution as a first and smallest one is unavoidable .
From this the Fermat’s last theorem is proved simply by the
well- known fact in the number theory that when any
consecutive solution does not exist for the equation ( 0 ) .
3≥n


INTRODUCTION
Although the Fermat’s last theorem had been proved in 1994 by
the English mathematician A.Wiles, the question: “Is it possible to
prove this theorem in an elementary way using only the
mathematical knowledge in Fermat’s times ? ” posed during the
whole last 371 years had been left unsolved. In honour of
Fermat and of so many people, known and unknown after him,
who had essayed to prove this famous theorem, I introduce here
an elementary answer to this question.

I . TWO PROCEDURES PROPOSED FOR THE PROOF

1 -Two procedures to find the triplet- solution (x, y, z) of eq.( 0 ).
Assuming that there is a solution ( x, y, z ) of the eq.
( 0 ). Then we have : x, y, z and from this,
. Noting that the n- dimensional cubes
are always represented correspondingly by the 3-
dimensional right prisms with squares as their bases and
as their altitudes, we conclude the identical parity
of x, y and z for all n . Now let x be odd, y be even then z and
will be odd and will be even. For solving ( 0 ) we
consider constantly its two of three unknowns, say x and z as
positive intergers to find the third interger y, or y and z as
positive intergers to find the third interger x of the triplet-
solution. With this aim, two following procedures are proposed to
nnn
zyx =+
+
∈ Z
++

∈−=∈−= ZyzeZxze
yx
,
nnn
zandyx ,
222
, zandyx
222
,
−−− nnn
zandyx
2≥
y
e
x
e

2
find the solution of eq. ( 0 ) : ( I ) with
x odd , even ;

=

+=+
n
i
i
x
ini
n

nn
x
exCxex
1
)(
+
∈ Zex
x
,
x
e
i
y
in
n
i
i
n
nn
y
eyCyey

=

+=+
1
)(
( II ) with , y even , odd ;
+
∈ Zey

y
,
y
e
Clearly, in the procedure ( I ) because then if there
is an y =
+
∈ Zex
x
,
+
=









ZexC
n
n
i
i
x
ini
n
1

1
, or in the procedure ( II ) because
then if there is an x =
+
∈ Zey
y
,
+
=









ZeyC
n
n
i
i
y
ni
n
1
1
1
we

have certainly a solution ( x, y, z ) of the equation ( 0 ).

2 - The use of procedures ( I ) and ( II ).
Using ( I ) : For all odd x successively in their increasing order
taking by turns each even beginning with , we
calculate ( I ). If for some odd x and some even , we have
y =
x
e
2
min
=
x
e
x
e
+
=









ZexC
n
n

i
i
x
ini
n
1
1
then the triplet obtained ( x, y, z = x+ )
is one solution of ( 0 ).
x
e
Using ( II ) : For all even y successively in their increasing
order taking by turns each odd beginning with = 1 , we
calculate ( II ). If for some even y and some odd , we have
x =
y
e
miny
e
y
e
+
=










ZeyC
n
n
i
i
y
ini
n
1
1
then the triplet obtained ( x, y, z = y+ )
is one solution of ( 0 ).
y
e

3 - Solution equivalence between equations ( I ), ( II ) and the
equation ( 0 ).
Let ( x, y, z ) be a solution of ( 0 ). Then there are an even
number and an odd number xze
x
−=
yze
y

=
. Intuitively
the solution equivalence of ( I ), ( II ) and ( 0 ) is evident from
the fact that, by using ( I ) only if we have then ( I ) and

( 0 ) will become identity, and by using ( II ) only if we have
then ( II ) and ( 0 ) will become identity. More rigorously
the proof of this solution -equivalence can be represented as
follows : Putting two integers : x , z = x+ of this solution into
( I ), we have with the same third integer y in
the solution of ( 0 ). Now let ( ) be a solution of ( I ).
From this and
+
∈ Zy
+
∈Zx
x
e
ni
x
in
n
i
i
n
yexC =

=

1
111
,, zyx
111
xze
x

−=
+
=








=

ZexCy
n
n
i
i
x
ini
n
1
1
111
, i. e. we

3
have with . In other words,
( ) is also a solution of the equation ( 0 ).
nnn

zyx
111
=+
+
∈+= Zexz
x111
111
,, zyx
Similarly, let ( ) be a solution of ( II ). From this
and
222
,, zyx
222
yze
y
−=
+
=








=

ZeyCx
n

n
i
i
y
ini
n
1
1
222
, i. e. we have
with . In other words, ( ) is
also a solution of the equation ( 0 ).
nnn
zyx
222
=+
+
∈+= Zeyz
y222 222
,, zyx

4 - Definitions connected with the procedures.
Suppose that ( 0 ) or ( I ) and ( II ) have some solution (x,y,z).
Then we have x, y, z, and n and, evidently,
and .
+
∈ Z
+
∈−= Zxze
x

+
∈−= Zyze
y
Definition1. Solution in the form (x,y,z) is called explicit solution.
Definition 2. Solution in the form ( ) is called implicit solution.
yx
ee ,
Definition 3. Solution is called irreducible if its greatest common
divisor equals 1, i. e. gcd (x,y,z) = gcd ( ) = 1.
yx
ee ,
Definition 4. Solution is called k- proportional if its greatest
common divisor equals k, i. e. gcd (x,y,z) = gcd ( ) = k.
yx
ee ,
Definition 5. Solution is called generating if it is found by using
2
min
=
=
xx
ee
in ( I ) and by using
1
min
=
=
yy
ee
in ( II ). Their

implicit forms are ( ) and ( ) respectively.
yx
ee ,
min min
,
yx
ee
Definition 6. Solution is called combining if its implicit form is
( ) where there are
yx
ee ,
minxx
ee

and
minyy
ee

.
Definition 7. Solution is called the first or the smallest if its
explicit form is ( ), and its implicit form is
( ). Here “ first ” is clear from the use of procedures ( I )
and ( II ) stated above, and “ smallest ” will be evidenced by the
property 2 presented below.
minminmin
,, zyx
minmin
,
yx
ee

Definition 8. Solution ( x, y, z ) is called consecutive if its explicit
form is (x, x+1, x+2) and its implicit form is ( ). This is
evident from the procedures ( I ) and ( II ) that : x+2 = y+1 = z,
i. e. there is ( x, y, z ) = ( x, x+1, x+2 ).
minmin
,
yx
ee
Furthermore :
+
Z
denotes the set of all positive integers;
R denotes the set of all real numbers.
(
)
)!(!
!
ini
n
C
n
i
i
n

== is the
i- th binomial coefficient.

II - STRUCTURAL PROPERTIES OF THE TRIPLET
NUMBERS ( x, y, z ) ON THE REAL NUMBER LINE R.

In the procedure ( I ) with , for each odd x and each
even , there is always an
+
∈ Zex
x
,
x
e
.
1
1
RexCy
n
n
i
i
x
ini
n







=

=


In the

4
particular case when then the triplet numbers (x, y, z=x+ )
obtained is a solution of ( 0 ). Similarly in the procedure ( II )
with for each even y and each odd , there is always
an
+
∈ Zy
x
e
+
∈Zey
y
,
y
e
.
1
1
ReyCx
n
n
i
i
y
ini
n








=

=

In the particular case when
then the triplet numbers (x, y, z=y+ ) obtained is a solution of (0)
+
∈Zx
y
e
1. Two ways to determine each solution.

Property 1 - If the equation ( 0 ) has some solution then this
solution can be found by two ways : Either by using the
procedure ( I ) or by using the procedure ( II ).

Proof - Suppose the equation ( 0 ) has a solution (x, y, z) in
explicit form and ( ) in implicit form. We will prove that if
this solution is found by the procedure ( I ) then it will be found
by the procedure ( II ) and vice versa. In fact if by using ( I ) :
we obtain a solution (x, y, z) but by using
( II ) we obtain an other one, say (x’, y, z), satisfying
.Then this leads to a contradiction that
Hence x’ = x. Similarly if by using ( II ) : +
but by using( I )we obtain an other one, say (x,

y’, z ), satisfying .Then this leads to a
contradiction that . Hence y’ = y.
yx
ee ,
nnn
x
yxex +=+ )(
n
z=
nnnn
y
zyxey =+=+ ')(
.'
nnnn
yxyx +=+
y(
nnnn
y
zxye =+=)
nnnn
x
zyxex =+=+ ')(
nnnn
yxyx '+=+

2. Proportionality of the triplet numbers (x, y, z) to the values of
and .
x
e
y

e

Property 2 - The values of each real triplet (x, y, z) of the
equation ( 0 ) is directly proportional to
2
min
=
x
e
in ( I ) and to
in ( II ).
1
min
=
y
e

Proof - Note that any even in ( I ) can be written by
where and
x
e
minxxx
eke =
+
∈ Zk
x
2
min
=
x

e
, and any odd in ( II )
can be written by
y
e
minyyy
eke
=
where and is odd;

+
∈ Zk
y
y
k
.1
min
=
y
e
1 ) In the procedure ( I ) when using , for each odd x, we
obtain a triplet of real numbers (x, y, z = x + , where x, z
; (1a) and in general
minx
e
)
minx
e
+
∈Z

xze
x
−=
min
.
1
1
min
RexCy
n
n
i
i
x
ini
n







=

=


(2a). Now when using
minxxx

eke
=
we will obtain a new triplet

5
on the real number line, say (X, Y, Z = X+ ) where
Z-X (3a). Evidently, X , Z and in general,
x
e
=
x
e
+
∈Z
ReXCY
n
n
i
i
x
ini
n







=


=

1
1
(4a). But from (1a), there is
xkzkxzkeke
xxxxxx

=
−== )(
min
(5a). From (3a) and (5a) we
have Z = (6a) and X = (7a). Noting zk
x
xk
x
=
x
e and
putting (7a) into (4a) we get Y =
minxx
ek
n
n
i
i
x
ini
nx

exCk
1
1
min







=

R∈
(8a).
In comparison (8a) with (2a) we get Y = y
x
k
R

(9a).
Therefore if by procedure ( I ) using we get a triplet of
real numbers (x,y,z) positioned on the number line R, then when
using
minx
e
minxxx
eke
=
this triplet will be shifted forward and get a

new position at (X, Y, Z)= (x, y, z) = ( ). This
conclusion is evidently true also in the case where , i. e.
when (x, y, z) is a solution of the equation ( 0 ).
x
k zkykxk
xxx
,,
+
∈Zy
2 ) In the procedure ( II ) when using for each even y we
obtain a triplet of real numbers (x’, y’, z’= y’+ ) where =
z’-y’ (1b), y’, z’
miny
e
miny
e
miny
e
+

Z
and, in general, x’ =
ReyC
n
n
i
i
y
ini
n









=

1
1
min
'

(2b). Now when using
minyyy
eke
=
we will obtain a new triplet
on the real number line R, say (X’, Y’, Z’= Y+ ) , where
y
e
=Z’-Y’ (3b), Y’,Z’
y
e
+

Z
and in general

X’=
n
n
i
i
y
ini
n
eYC
1
1
'







=

R
∈ (4b). But from (1b) there is
'')''(
min
ykzkyzkeke
yyyyyy

=
−==

(5b). From (3b) and (5b) we
have Z’ = z’ (6b) and Y’ = y’ (7b). Noting
y
k
y
k
minyyy
eke
=
and
putting (7b) into (4b) we get X’ =
n
n
i
i
y
ini
ny
eyCk
1
1
min
'








=

R∈
(8b).
In comparison (8b) with (2b) we get X’ = x’ (9b).
y
k
Therefore if by procedure ( II ) using we get a triplet of
real numbers (x’, y’, z’) positioned on the number line R, then
when using this triplet will be shifted forward and get
a new position at (X’, Y’, Z’)= (x’,y’,z’) = ( ).
This conclusion is evidently true also in the case where x’
miny
e
minyyy
eke =
y
k ,'xk
y
',' zkyk
yy
+

Z
,
i. e. when (x’, y’, z’) is a solution of the equation ( 0 ).

3- Decisive role of and in the existence or the non-
existence of solution of the equation ( 0 ).
minx

e
miny
e

6

Property 3 - The equation ( 0 ) has no solution when either using
in the procedure ( I ) we have no any or using
in the procedure ( II ) we have no any .
minx
e
+
∈ Zy
miny
e
+
∈ Zx

Proof - As evidenced by the argumentation in the proof of
property 2, we see that :
1 ) By the use of procedure ( I ) to find the solution ( x, y, z ) of
eq. ( 0 ) with , because the two variables x and z are always
chosen as positive integers, the triplet obtained ( x, y, z ) will be a
solution of ( 0 ) if and only if the third variable calculated as
minx
e
n
n
i
i

x
ini
n
exCy
1
1
min






=

=

(2a) is a positive integer. Now when using
all other even greater than , say
x
e
minx
e
minxxx
eke
=
where
and > 1, a - proportional triplet (X,Y,Z)=
( ) will be obtained where X, Z
+

∈ Zk
x
x
k
x
k
zkykxk
xxx
,,
+

Z
and
Y =
n
n
i
i
x
ini
nx
exCk
1
1
min








=

(8a). Comparing (2a) and (8a) it can be
seen that for any even > the variable calculated Y in (8a)
will be positive integer if and only if the variable y in (2a)
calculated by using is a positive integer.
x
e
minx
e
minx
e
Therefore if using in the procedure ( I ) we find no any
then there will be no any to find the
solutions of the equation ( 0 ) for the procedure ( II ). Thus,
without using the procedure ( II ) it can be concluded in this case
that the equation ( 0 ) has no solution.
minx
e
+
∈ Zy
+
∈−= Zyze
y

2 ) By the use of procedure ( II ) to find the solution ( x’, y’, z’)
of eq. ( 0 ) with , because the two variables y’ and z’ are
always chosen as positive integers, the triplet obtained ( x’, y’, z’ )

will be a solution of ( 0 ) if and only if the third variable
calculated as x’ =
miny
e
n
n
i
i
y
ini
n
eyC
1
1
min
'







=

(2b) is a positive integer.
Now when using all other odd greater than , say
where is odd integer greater than 1, a -
proportional triplet (X’, Y’, Z’) = (
'

) can be obtained
where Y’ and Z’ are positive integers and X’ =
y
e
miny
e
minyyy
eke =
y
k
y
k
,',' zkykxk
yyy
n
n
i
i
y
ini
ny
eyCk
1
1
min
'








=

(8b). Comparing (2b) and (8b) it can be
seen that for any the variable calculated in (8b) will
minyy
ee >

7
be positive integer if and only if the variable x’ in (2b) calculated
by using is a positive integer.
miny
e
Therefore if using in the procedure ( II ) we find no any
x’
miny
e
+

Z
then there will be no any to find the
solutions of the equation ( 0 ) for the procedure ( I ). Thus without
using the procedure ( I ) it can be concluded in this case that the
equation ( 0 ) has no solution.
+
∈−= Zxze
x
''


3. Possibility of procedures ( I ) and ( II ) in finding an
irreducible solution of the equation ( 0 ) by any even and
by any odd .
x
e
y
e
Property 4 - If the equation ( 0 ) has some irreducible solution
then this solution will be found either by the procedure ( I ) at
any even , say or by the procedure ( II ) at any
odd , say as - proportional one or -
proportional one.
x
e
,
minxxx
eke =
y
e ,
minyyy
eke =
x
k
y
k

Proof - From the property 2, the existence of some solution (x,
y, z) of ( 0 ) is uniquely depended on its existence either by the
procedure ( I ) at or by the procedure ( II ) at . Now at

any , say
minx
e
miny
e
minxx
ee >
minxxx
eke
=
and at any , say
this solution will be shifted forward on the positive
integer number line
minyy
ee >
minyyy
eke =
+
Z
and positioned either at ( ) in
the procedure ( I ) or at ( ) in the procedure ( II ).
Hence after eliminating the common factors of proportionality
or we get always an irreducible solution (x, y, z) of the
equation ( 0 ).
zkykxk
xxx
,,
zkykxk
yyy
,,

x
k
y
k

*5. Generating characteristic of the set { } found by using
in the procedure ( II ) and the set { } found by using
in the procedure ( I )
ix
e
miny
e
jy
e
minx
e
About the last fifth property I would like to note that this
property does not play any role for the proof of the Fermat’s last
theorem. It is introduced here to understand the creation of the
combining solutions, found by each pair of the Cartesian product
{ } x { }. In other words for an elementary proof presented in
this paper only four properties given above are necessary and
sufficient.
ix
e
jy
e
* Property 5 - If the equation ( 0 ) has many solutions, say
( ) where i , j =1, 2, 3, 4, , n then :
jyix

ee ,
1 ) The solutions ( ) for all j and the solutions
( ) for all i are irreducible.
jyx
ee ,
min
min
,
yix
ee

8
2 ) Each pair ( ) of the Cartesian product { } x { }
where i, j = 2, 3, , n , corresponds with one solution of the
equation ( 0 ), which is either irreducible if gcd ( ) = 1 or
k – proportional if gcd ( ) = k.
jyix
ee ,
ix
e
jy
e
jyix
ee ,
jyix
ee ,

Proof - 1 ) The affirmation is evident by the fact that there are
always gcd ( ) = 1 for all j and gcd ( ) = 1 for all i.
jyx

ee ,
min min
,
yix
ee
2 ) Suppose that the implicit pair ( ) at some i,
j = 2,3, , n does not correspond with a solution, irreducible or
proportional , of the equation ( 0 ). Then when using the
procedure ( I ) with we have no any .
But this leads to , which contradicts the given
. As the same, when using the procedure ( II ) with
we have no any . But this leads to
, which contradicts the given . Hence
each combining pair ( ) at any i and j must be
corresponding with one triplet of positive integers ( ),
which satisfies the equality ( 0 ).
jyix
ee ,
+
∈ Zzex
jiixji
,,
+
∈ Zy
ji
+
∉−= Zyze
jijijy
+
∈ Ze

jy
+
∈ Zzey
jijyji
,,
+
∈ Zx
ji
+
∉−= Zxze
jijiix
+
∈ Ze
ix
jyix
ee ,
jijiji
zyx ,,
For resuming, we present here a distribution’s picture of the
considered solutions in the form of a ( n x n ) matrix as follows :
Table 1 . Representation of the solutions in the procedures (I), (II)

\ :
jy
e 1
min
=
y
e
2y

e
jy
e
ny
e
\ :
ix
e
:
:
2
min
=
x
e
minmin
,
yx
ee
2min
,
yx
ee
jyx
ee ,
min nyx
ee ,
min
:
2x

e
min2
,
yx
ee
22
,
yx
ee
jyx
ee ,
2 nyx
ee ,
2
: : : : : :
:
ix
e
min
,
yix
ee
2
,
yix
ee
jyix
ee ,
nyix
ee ,

: : : : : :
:
nx
e
min
,
ynx
ee
2
,
ynx
ee
jynx
ee ,
nynx
ee ,


III . APPLICATION OF THE PROPOSED PROCEDURES
TO FIND THE PYTHAGOREAN TRIPLETS
Before presenting a proof of the Fermat’s last theorem I would
like to illustrate an application of the two proposed procedures (I )
and ( II ) for finding all triplet – solutions of the Pythagorean case
n = 2, which are represented in form of a square matrix with
infinitely many rows and columns ( see Table 1 ).

Table 2 . Pythagorean triplets found by ( I ) and ( II )

9



yj
e
xi
e


2
1



2
3


2
5


2
7


2
9


2.1.1
3, 4, 5 15, 8, 17 35, 12, 37 63, 16, 65 99, 20,101


2.2.2
5, 12, 13 21, 20, 29 45, 28, 53 77,36,85 117, 44, 125

2.3.3
7, 24, 25 27, 36, 45 55, 48, 73 91, 60, 109 135, 72, 153

2.4.4
9, 40, 41 33, 56, 65 65, 72, 97 105, 88, 137 153,104,185

2.5.5
11, 60, 61 39, 80, 89 75, 100, 125 119,120,169 171,140,221

2.6.6
13,84,85 45,108,117 85,132,157 133,156,205 189,180,261

2.7.7
15,112,113 51,140,149 95,168,193 147,196,245 207,224,305

2.8.8
17,144,145 57,176,185 105,208,233 161,240,289 225,272,353

2.9.9
19,180,181 63,216,225 115,252,277 175,288,337 243,324,405



Let i , j be the i- th row and the j- th column respectively,
then the Pythagorean triplets ( ) and the sets { } ,
{ } are firstly represented here by the following formulas,

which are easily verified by the identity :
jijiji
zyx ,,
ix
e
jy
e
222
ijijij
zyx =+

]1)(2)[12(

+
−= jijx
ji


)12(2

+= ijiy
ji

=
2
2]1)(2)[12( ijijz
ji
+−+−=
2
)12()12(2 −+−+ jiji

{ } = { } and { } = { } with i , j = 1, 2, 3, ,
ix
e
2
2i
jy
e
2
)12( −j ∞
From the Table 2 it is easy to verify all the five properties
presented above. In particular, the property 5 shows us the
capacity for finding the existent solutions of the eq. ( 0 ) by the
generating sets
{
}
ix
e
given by using
1
min
=
y
e
in the procedure (II)
and
{
}
jy
e
given by using

2
min
=
x
e
in the procedure ( I ), with
considering each and each of them as an indivisible unit
such as . This important fact is evidenced from
the irreducibility of all the pairs
ix
e
jy
e
minmin yx
eande
).,(),(
minmin yixjyx
eeandee

IV - PROOF OF THE FERMAT’S LAST THEOREM

1 - Lemma . The Diophantine equation ( 0 ) for x,
y, z, n
nnn
zyx =+
+

Z
has solution if and only if it has a consecutive
solution as the smallest one.


Proof - First of all, remember some facts given above. From
the definition 8, a consecutive solution must be either of the
implicit form ( ) or of the explicit form ( x, x+1, x+2 ).
Then from the use of the procedures ( I ) and ( II ); and the
property 2 , this consecutive solution must be either the first and
the smallest one found by the procedure ( I ) using , or the
minmin
,
yx
ee
minx
e

10
first and the smallest one found by the procedure ( II ) using
. Following the property 1, this solution is also the first and
the smallest one found by both procedures ( I ) and ( II )
simultaneously.
miny
e
Now we will prove the respective contrapositions of the
lemma.
Suppose that the equation ( 0 ) has no consecutive solution
( ). From the property 3, this equation has no any
solution found by the procedure ( I ) using and by procedure
( II ) using . Therefore the two generating sets { } and
{ } are empty, i. e. the set {( )} of the solutions of eq.
( 0 ) is empty. This means that there does not exist any solution
of the equation ( 0 ).

minmin
,
yx
ee
minx
e
miny
e
ix
e
jy
e
jyix
ee ,
Suppose that the equation ( 0 ) has no solution. Then this
evidently leads to the non- existence of any solution, including the
first smallest ( ) that is consecutive one : ( )
or ( x, x+1, x+2 ).
111111
,, zyx
minmin
,
yx
ee

2 - The Fermat’s last theorem . The Diophantine equation
( 0 ) has no non- zero positive integer solution for x,
y, z when and n >2.
nnn
zyx =+

+
∈ Zn

Proof - The lemma established the equivalence between the
statement of this theorem and the non- existence of the
consecutive solution for the equation ( 0 ). But the non- existence
of the consecutive solution for all n > 2 of the equation ( 0 ) had
been easily and beautifully proved long ago in the number theory
in form of an exercise, for example in [ 1 ] ( Chapter 2, section
18, exercise 5 ). For convenience let us copy out this proof.

“ Exercise 5 - If n is a natural number greater than 2, then
the equation : has no solution in natural
number.
nnn
xxx )2()1( +=++
Solution -1 ) n is an odd number greater than 2 . Putting
y=x+1, we have , whence :
. This proves that is
a divisor of the number 2, which by y = x+1 > 1, is impossible.
nnn
yyy )1()1( −−+=
2222
223311
=−−−−
−−−
yCyCyCy
n
n
n

n
n
n
n
L
2
y
2 ) n is an even number greater than 2 . Putting
y=x+1, we obtain ,
whence : . The first equality
shows that whence y > 2n . The second equality shows
that y is a divisor of 2n. This is a contradiction . ”
0222
13311
=−−−−
−−−
yCyCyCy
n
n
n
n
n
n
n
L
0222
43211
=−−−−
−−−
nyCyCy

n
n
n
n
n
L
1
2

>
nn
nyy

3 . Corollary –

11
1 ) For positive integers x, and n, the n- th root
x
e
n
n
i
i
x
ini
n
exC
1
1








=

is not an integer when n > 2.
2 ) The Diophantine equation in three unknowns x, y and :
x
e
ni
x
in
n
i
i
n
yexC =

=

1
( III ) has no non- zero positive integer solution
when n
+

Z
and n > 2.

3 ) The Diophantine equation in five unknowns x, y, z, and :
x
e
y
e
ni
y
in
n
i
i
n
i
x
in
n
i
i
n
zeyCexC =+

=

=
∑∑
11
( IV ) has no non- zero positive
integer solution when n
+


Z
and n > 2.

Proof - 1) and 2) : Evidently, if
+
=

∈=







ZyexC
n
n
i
i
x
ini
n
1
1

then with when n > 2.
This contradicts the Fermat’s last theorem.
nnnn
x

zyxex =+=+ )(
+
∈ Zzandyx,
3 ) : This assertion is directly resulted from both
assertions 1 ) and 2 ) above.

* * *
Acknowlegment - I would like to express my deep gratitude
to all mathematicians - referees for their important contribution to
evaluate this proof and to bring this paper to the readers.
To them I would like to express my heartfelt thankfulness.



References
[ 1 ] W. Sierpin’ski , Elementary theory of numbers , Polski
Academic Nauk, Warzawa , 1964





Author’s Address

N.V. TU -YEN E- mail :
NHA E 1 , P. 406 VAN CHUONG
DONG DA - HANOI- VIETNAM






























MÉMOIRE 2
*
- soumis en 2007
à mon âge de 72 .

* * *


UNE PREUVE ÉLÉMENTAIRE DE L’INFINITÉ
DES NOMBRES PREMIERS JUMEAUX ( p, p+2 )
ET DES PAIRES VOISINES DE NOMBRES
PREMIERS JUMEAUX ( p, p+2 ; p+6, p+8 )
N. V. TU- YEN
( Vietnam )
*
ABSTRACT
- In imitation of the Eratosthenes sieve for sifting the primes (p),
two sieves for sifting the twin- primes (p, p+2) and the neighbouring pairs of
twin- primes (p,p+2 ; p+6,p+8) are proposed. From the structure of these sieves,
after the Legendre’s manner of deducing the density- limit of the primes
, the formulas describing the density- limits of the twin- primes
and of the neighbouring pairs of twin- primes
)(pD

)2,( +

ppD
)8,6;2,(
+
++

ppppD
are deduced and expressed in the same
form
)1(

3
1
3


=

−=
i
i
p
m
D
as , where is an i-th prime ; m =1 for
m = 2 for
)(pD

i
p
),(pD

)2,(
+

ppD
and m = 4 for
).8,6;2,(
+
++


ppppD
Finally, the last two density- limits obtained just
had been used to prove the infinity of the twin- primes and the neighbouring
pairs of twin- primes. If the proof proposed here is an accepted one, it will
be reasonable to place both these proved here conjectures into the Arithmetic’s
edifice already constructed as one Corollary of the Euclid’s theorem on the
infinitude of the primes.

*
RÉSUMÉ - À l’imitation du crible d’Ératosthène pour cribler les nombres
premiers ( p ), deux cribles pour cribler les nombres premiers jumeaux ( p,p+2 )
et les paires voisines de nombres premiers jumeaux ( p,p+2 ; p+6,p+8 ) sont
proposés. De la structure de ces cribles et suivant la déduction connue de
Legendre pour la densité limite des nombres premiers , deux formules
décrivant les densités limites de nombres premiers jumeaux et
des paires voisines de nombres premiers jumeaux
)(pD

)2,( +

ppD
)8,6;2,( ++
+

ppppD

respectives sont établies en même forme comme celle de nombres premiers :
)1(
3
1

3
i
i
p
m
D −=


=

où est le i- ième nombre premier , entier positif
égale 1 pour des nombres premiers, égale 2 pour des nombres premiers
jumeaux et égale 4 pour des paires voisines de nombres premiers jumeaux.
Enfin ces deux dernières formules ont utilisées effectivement pour démontrer
l’infinité des nombres premiers jumeaux et l’infinité des paires voisines de
nombres premiers jumeaux. Si la preuve proposée ici est acceptable, il serait
raisonnable de placer ces deux anciennes conjectures comme un corollaire du
théorème d’Euclide de l’infinitude des nombres premiers dans l’édifice d’
Arithmétique déjà construit.
i
p
m
*


MOTS CLEFS - Crible d’Ératosthène ; Suite binaire ; Bande criblée ; Bande
criblante ; Bande résultante ; Nombre premier jumeau ; Paire de nombres
premiers jumeaux ; Densité et Densité limite .

*


1








PRÉLIMINAIRE


Dans cet article le crible d’Ératosthène sera concrétisé par un
système de 3 composants : D’abord une bande criblée comportant
une suite infinie de bits 1 qui représentent les nombres naturels
que l’on devrait les faire trier, puis une infinité des bandes
criblantes donnant pour chacune une suite binaire de période
avec étant le i-ième nombre premier, où les bits 0 codant
les multiples de et les bits 1 codant ses non- multiples, et
enfin, une bande résultante qui est la suite binaire restée de la
bande criblée après le triage. Cette approche est en fait une
concrétisation des concepts comme fonction crible, fonction
Ératosthène , rappelés dans les ouvrages d’Informatique, par
exemple [2].
i
p
i
p
i

p
i
p
Alors après avoir criblé la suite infinie des nombres naturels
par la première bande criblante
2
1
=
p
, qui est 010101 tous les
nombres pairs sauf 2 sont annulés et les bits 1 représentant les
nombres naturels restés sur la bande criblée ou trouvés sur la
bande résultante notent des nombres impairs. En codant chaque
période 2 : ‘01’, qui représente un nombre impair par un bit 1,
on a une suite infinie des bits 1, donnant une bande criblée pour
trier tous les nombres premiers à partir de 9. Les bandes
criblantes sont placées en lignes horizontales de l’ordre
croissant de haut en bas avec le premier bit 0, ou le premier
trou, situé à l’extrémité gauche de valeur , mais plus
simplement pour écrire le schéma du crible, de valeur suivant
la propriété 1 ci- après :
i
p
2
i
p
i
p3

1 - Propriété 1

- À chaque différence de 2 unités entre deux
nombres premiers successifs correspond un pas avancé vers
droite d’une longueur de 3 bits binaires ‘011’ du départ ou du
premier bit 0 de la bande criblante par rapport à celui de
la bande criblante .
1+i
p
i
p

Preuve - Car tous les nombres considérés sont impairs, en
supposant et 12 += kp
i
(
)
12
1
+
+
=
+
mkp
i
, avec entiers positifs k
et m, la différence entre ces deux premiers bits 0, ou trous, est

2
égale à mpp
ii
633

1
=

+
unités ou 3m bits binaires utilisés.
Notons que chaque unité de m correspond à un écart entre deux
nombres impairs consécutifs, c’est à dire à 2 unités de nombres
naturels, la propriété 1 est évidente.

( Symbole

utilisé dans
cet article pour noter la fin d’une preuve ).
Utilisant les composants du crible et la propriété 1 décrits au-
dessus, un schéma du crible d’Ératosthène pour déterminer les
nombres premiers à partir de 9 est présenté ci- après.

2 - Schéma 1 : Crible 1 d’Ératosthène pour nombres premiers.

9 11 13 15 17 19 21 23 25 27 29 31 33 35 37
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

0 1 1 . 0 1 1 . 0 1 1 . 0 1 1 . 0 1 1.
0 1 1 1 1 . 0 1 1 1 1 . 1 1

0 1 1 0 1 1 0 1 0 0 1 1 0 0 1

Remarques - 1) Car 37 < , il est nécessaire et suffisant
d’utiliser seulement 3 bandes criblantes dans le crible,
parmi lesquelles celle de

22
4
7=p
321
, petpp
2
1
=
p
a été utilisée pour donner la
bande criblée codée ( ligne ) donc restent dans cet exemple
deux bandes criblantes
è
2
53
32
=
=
petp ( lignes
respectivement ). La dernière ligne est la bande résultante. Sur la
bande résultante de cet exemple on trouve 8 bits 1 correspondant
8 nombres premiers notés à la 1
èè
et 43
ère
ligne dans leurs mêmes
colonnes respectivement. Ce sont 11 ;13 ;17 ; 19 ; 23 ; 29 ; 31 et 37.
2) Les points dans chaque bande criblante notent ses périodes .
i
p

3) L’usage seul des bandes criblantes ‘ premières ’ pour trier les
nombres premiers (p), premiers jumeaux (p,p+2) et les paires
voisines de nombres premiers jumeaux (p,p+2 ; p+6,p+8) eux-
mêmes n’est pas un cercle vicieux car un système de n bandes
criblantes ‘premières’ , i = 1, 2, , n est capable toujours de
trouver un bon nombre d’autres nombres premiers plus grands
que , qui sont prêts à continuer sans cesse le criblage.
i
p
i
p
n
p

I - Schémas du crible d’ Ératosthène utilisés

1 - Cas des nombres premiers jumeaux ( p,p+2 )

En utilisant la bande criblante
3
2
=
p
, ‘011011011 ’ pour la
bande criblée, qui se compose seulement des nombres impairs

3
jumeaux ‘011’, on voit bien qu’à partir de 9, il serait possible
d’en trouver un nombre premier jumeau si tous les deux bits 1
d’une période 3, ‘011’, restaient conservés sans être annulés par

des bandes criblantes du crible. En codant chaque période 3,
‘011’ de la bande criblante
i
p
3
2
=
p
par un bit 1, c’est à dire, en
posant 011 = 1, on obtient une suite infinie des bits 1 utilisée
comme une bande criblée pour trier les nombres premiers
jumeaux.
Dans ce cas, parce que la longueur de chaque période de la
bande criblée était diminuée 3 fois, toutes les bandes criblantes
à partir de
i
p 5
3
=
p devraient se rétrécir en une même échelle
1/3 d’après la propriété 2 suivante :

1 ) Propriété 2 - Pour cribler les nombres premiers jumeaux
(p,p+2), la i- ième bande criblante à partir de est une
suite binaire périodique infinie de période bits binaires
commencée par
i
p 5
3
=p

i
p






3
i
p
bits 1, puis continue par
)}2
3
2({ +







i
i
p
p
bits 1 placés entre deux bits 0, et finie par







3
i
p
bits 1. Ici
[
]
x
est la partie entière de x.

Preuve - L’action d’annulation des bits 1 causée par la bande
criblante sur la bande criblée à l’échelle 1 : 1 comme
‘011011011 ’ est représentée par une suite binaire périodique
infinie de période bits, où il y a périodes 3, ‘011’, sur la
bande criblée et 3 périodes sur la bande criblante . Dans
chaque période commune de la bande criblante il y a 4
bits 0 pour la diviser en 3 périodes , parmi lesquels deux bits
0 à ses deux extrémités sont déjà coïncidés avec ceux de la
bande criblée qui n’annulent aucun bit 1 des périodes 3, ‘011’, et
seulement deux bits 0 à l’intérieur de chaque période jouent
le rôle des trous, dits ‘ trous actifs ’, pour faire annuler deux bits
1 situés à deux périodes 3, ‘011’ distinguées, c’est à dire ces deux
trous actifs annulent deux paires possibles de nombres premiers
jumeaux sur la bande criblée. On voit bien qu’à l’échelle rétrécie
1/3, la période commune bits de la bande criblante devient
bits, où chacune de ses trois périodes composantes possède
i
p

i
p3
i
p
i
p
i
p
i
p3
i
p
i
p
i
p3
i
p3
i
p
i
p
[]
3
i
p
bits binaires. En notant la symétrie centrale ou le palindrome
d’une telle période commune, on trouve facilement la répartition

4

des bits 1 codés, c’est à dire, 1
011

dans les trois périodes
composantes à l’échelle 1/3, qui est représentée par le schéma
suivant :
i
p
( 0 )
[]
13 bitsp
i
. 0 {
[
]
)232(
+

ii
pp } 0
[
]
13 bitsp
i
. ( 0 )
bits 1 ( I )
où ( 0 ) note le bit 0 coïncident avec celui de la bande criblée
que nous ne l’écririons pas dans la bande criblante considérée.
i
p

Il est important de noter que notre argumentation utilisée dans
cette preuve reste valable en cas des périodes communes de
bits binaires, où n est un entier positif arbitraire.
Dans ce cas, il y aura périodes 3 ‘011’ dans la
bande criblée et périodes dans la bande
criblante . * De ( I ) nous trouvons facilement, par exemple :
ni
pppp 3
43
ni
pppp
43
nii
ppppp 3
1143 +− i
p
i
p
Période : 10101 ; Période 5
3
=p 11
5
=
p : 11101110111 ; etc

2) Schéma 2 : Crible 2 d’Ératosthène pour les nombres
premiers jumeaux (p, p+2).

9 15 21 27 33 39 45 51 57 63 69 75 81
13 19 25 31 37 43 49 55 61 67 73 79 85

1 1 1 1 1 1 1 1 1 1 1 1 1

1 0 1 0 1 . 1 0 1 0 1 . 1 0
1 1 0 1 0 1 1 . 1 1 0 1

1 1 0 1 0 1 0 0 1 0 1 0 0

Remarques - 1) Car 85 < , les bandes criblantes
utilisées sont inférieures à
22
5
11=p
i
p
11
5
=
p parmi lesquelles celles de
sont déjà utilisées pour donner la bande criblée
( ) donc restent seulement deux bandes criblantes
( et . Notant que tous les nombres de
la première ligne divisent 3 et sont codés par des bits 0, leur
deux impairs consécutifs où le deuxième est écrit à la ligne
sont des non- multiples de
32
21
== petp
ligne
è
3 5

3
=p
ligne
è
4) )5(7
4
lignep
è
=
è
2
3
2
=
p
et codés par des bits 1. On voit
donc que tous les bandes considérées et leur écarts des bits binaires
aux extrémités gauches dans ce crible sont rétrécies à l’échelle
1/3. Dans cet exemple sur la bande résultante on trouve 6 bits 1
correspondant 6 nombres premiers jumeaux qui sont les deux
nombres impairs consécutifs après ceux notés à la 1
ère
ligne dans
leurs mêmes colonnes respectivement. Ce sont : 11,13 ; 17,19 ;
29,31 ; 41,43 ; 59,61 et 71,73.

5
2) et 3) sont identiques à celles du Crible 1 précédent.

2 - Cas des paires voisines de nombres premiers jumeaux

( p, p+2 ; p+6, p+8 )

1 ) Propriété 3 - Pour cribler les paires voisines de nombres
premiers jumeaux (p,p+2 ;p+6,p+8), la i- ième bande criblante
à partir de a une période de bits binaires avec 4
bits 0 actifs inclus, qui comporte :
i
p 5
3
=p
i
p
*
[
6
i
p
]
bits 1 : À chacun intervalle de ses deux bouts.
* 1 bit 1, si 93
<

i
p.
2k+1 bits 1, si )1(6363
+
+
<

+

kpk
i
: À l’intervalle
entre deux bits 0 au milieu, qui sont le 2
ème
et le 3
ème
bits 0
actifs notés de gauche à droite de la période .
i
p
*
[]
)}562({
2
1
+−
ii
pp , si 93
<

i
p

[]
)},5262({
2
1
++− kpp
ii

si
)}1(6363 +
+
<

+
kpk
i
:
À chacun intervalle entre deux bits 0 aux demi- périodes
gauche et droite respectivement. Ici
[
]
x
note la partie entière de x.

Preuve - L’action d’annulation des bits 1 séparés sur la
bande criblée non- codée ‘011011011 ’ causée par la bande
criblante est représentée par une suite binaire périodique infinie
de période bits où il y a périodes 6 bits binaires
‘011011’ sur la bande criblée et 6 périodes bits binaires sur la
bande criblante . Dans chaque période commune il y a 7
bits 0 pour la diviser en 6 périodes où deux bits 0 situés à
ses deux extrémités et 1 bit 0 à son point médian sont toujours
coïncidés avec ceux de la bande criblée qui n’annulent pas aucun
bit 1 de la période 6 ‘011011’. Alors restent sur chaque période
de la bande criblante du crible, 4 bits 0 ‘actifs’ ou trous,
pour faire annuler 4 bits 1 distinguées de la bande criblée, et
bits 1.
i

p
i
p6
i
p
i
p
i
p
i
p6
i
p
i
p
)46( −
i
p
Évidemment à l’échelle 1/6, la longueur d’une période
bits de la bande criblante devient celle d’une période bits et
chacune de ses 6 périodes composantes rétrécies possède
i
p6
i
p
[]
6
i
p
bits binaires. Considérons la répartition de ces bits binaires

dans une période commune rétrécie à 1/6. Les 6 périodes
composantes, notées par des nombres ordinaux ci- après de dans
une période commune sont présentées comme :
i
p
i
p
i
p6

6
(0) 0 0 (0) 0 0 (0)
1
ère
2
ème
3
ème
4
ème
5
ème
6
ème
( II )
où 3 bits 0 coïncidents notés par (0) ne seront comptés pas dans
les bits, d’où une répartition de
i
p )4(


i
p bits 1 dans ces 6
intervalles doive être réalisée. En notant que cette période est un
palindrome ou une symétrie centrale et que le bit central (0) se
trouve à la médiane d’une période 6, ‘011011’, de la bande
criblée, tandis que deux autres bits (0) situés aux extrémités de
chacune période de la bande criblante sont toujours coïncidés
avec deux bits 0 de la bande criblée, nous déterminons les bits 1
restés pour ces 6 intervalles de ( II ) selon un ordre suivant :
i
p
i
p
*
[
6
i
p
]
bits 1 : pour le 1
ère
et le 6
ème
intervalle chacun.
* 1 bit 1 si 93
<

i
p
(2k + 1) bits 1 si

)1(6363
+
+
<

+
kpk
i

pour l’ensemble des intervalles 3
ème
et 4
ème
. Et enfin,
*
[]
)}4162({
2
1
++−
ii
pp bits 1 si
93 <

i
p


[]
)}4)12(62({

2
1
+++− kpp
ii
bits 1 si )1(6363 ++
<

+
kpk
i

pour le 2
ème
et le 5
ème
intervalle chacun. *
D’après ces formules chaque période de la bande criblante
prend une suite binaire concrète pour chaque , par exemple :
i
p
i
p
5
3
=p : 0 0 1 0 0
7
4
=p
: 1 0 0 1 0 0 1
11

5
=p : 1 0 1 0 1 1 1 0 1 0 1
13
6
=p : 1 1 0 1 0 1 1 1 0 1 0 1 1

2 ) Schéma 3 : Crible 3 d’Ératosthène pour les paires voisines
de nombres premiers jumeaux (p,p+2 ; p+6,p+8)

9 21 33 45 57 69 81 93 105 117 165 177 189
19 31 43 55 67 79 91 103 115 127 175 187 199
1 1 1 1 1 1 1 1 1 1 1 1 1
11 11 11 11 11 1 1 11 11 11 11 11 11 1 1

1 01 01 10 10 1.1 01 01 10 10 10 10 1.1
11 01 01 11 1 0 10 11 .11 01 10 11 . 1 1
11 10 11 1 0 11 11 11 01 . 11 10 1 1
1 11 10 1 1 10 11 11 11 01 11 1.1

1 0 0 0 0 0 0 1 0 0 0 1


7
Remarques - 1 ) Puisque 199 < les bandes criblantes
utilisées sont inférieures à
2
7
17=p
i
p

17
7
=
p , parmi lesquelles celles de
étaient déjà utilisées pour donner la bande
criblée ( 3
32
21
== petp
ème
et 4
ème
lignes ), il y a dans cet exemple 4 bandes
criblantes : et
( 8
)7(11),6(7),5(5
543
lignepligneplignep
èèè
===
13
6
=p
ème
ligne ). En tenant compte du fait indiqué par la
propriété 1 que l’écart minimal possible entre deux premiers bits
0 des bandes criblantes successives égale 3 bits
binaires, pour distinguer bien la position exacte des 4 bits 0 actifs
de chaque période , qui font annuler les demi- périodes des
périodes 6, ‘011011’, à l’échelle 1/6 utilisée, chaque bit 1 de la

bande criblée ( 3
1+ii
petp
i
p
ème
ligne ) et de toutes les bandes criblantes se
divise en deux bits 1 et chaque bit 0 des bandes criblantes se
divise en deux bits binaires différents ‘01’ ou ‘10’, dépendant de
la position d’un bit 1 annulé, qui se trouve à la demie gauche
ou droite d’une période 6 non- rétrécie, ‘011011’, respectivement.
Plus concrètement, chaque bit 0 de la bande criblante à
l’échelle 1/6 est décrit par ‘01’ pour representer une période 6
annulée non- rétrécie de forme ‘01011’ ou de forme ‘001011’, et
par ‘10’ pour représenter une période 6 annulée non- rétrécie de
forme ‘011010’ ou de forme ‘011001’. Ainsi donc, une paire
voisine de nombres premiers jumeaux ( p,p+2 ; p+6,p+8 ) sera
trouvée quand il y a dans la matrice binaire formée par des
bandes criblantes deux colonnes avoisinantes, qui comportent
seulement des bits 1. Pour cet exemple, dans l’intervalle de 9 à
199 considéré, la bande résultante ( dernière ligne ) nous donne
trois telles paires de colonnes avoisinantes, qui correspondent à
trois paires voisines de nombres premiers jumeaux, dites ( 11,13 ;
17,19 ), ( 101,103 ; 107,109 ) et ( 191,193 ; 197,199 ).
i
p
i
p
2 ) et 3 ) sont identiques à celles du Crible 1 précédent.


II - Densités limites des nombres premiers jumeaux

)2,(
+

ppD
et des paires voisines de nombres
premiers jumeaux
)8,6;2,(
+
+
+

ppppD
.

Après être filtré par n bandes criblantes, dans une longueur de
la période commune de la bande résultante, le rapport
du nombre des bits 1 restés sans être annulés sur la longueur de
cette période commune s’appelle densité de l’ensemble des
nombres considérés dans l’ensemble des nombres naturels de cette
période. Quand n tend vers l’infini, cette densité est appelée
densité limite des ces nombres considérés et notée par .
n
ppp
21
n
D

D


8

×