Escolar Documentos
Profissional Documentos
Cultura Documentos
Introduction
In this paper we discuss the links between Legendre's conjecture and Bertrand's
postulate, the latter also known as Chebiscev theorem.
n 1, p : n p 2n
In other words, for every integer n greater than 1, there exists at least one
prime number between it 2n, namely that:
(2n) (n)
Legendre's conjecture says that there is always a prime between n2 and (n + 1)2,
namely:
n 1, p : n2 p (n 1)2
This conjecture is one of the problems of Landau and, to date, has not yet been
demonstrated.
In [3] we have seen evidence of the conjecture, a proposal so, the formulas for
the estimates of the number of primes in a quadratic interval.
An idea, therefore, may be to work at PBG, and use the combinatorial as in [1]
and [4].
1
Rosario Turco is an engineer, who works in a society Information Technology of the group
TelecomItalia (NA – Italy)
2
Maria Colonnese is a teacher of mathematics, who works at the High School “De Bottis” of Torre del
Greco (NA - Italy)
1
The work is divided, then, in PART ONE Bertrand's postulate, PART TWO
Generalized Bertrand's Postulate and Legendre's conjecture.
Definitions
We define with ( a, b)
a p b
p the product of primes in the interval a,b with the
primes greater than a, but not greater than b; con (a) p the product of
pa
primes not greater than a. We define the binomial coefficient with:
n n!
h h !( n h)!
n n
h n h .
n
n
(a b)n a hb n h .
h 0 h
Then we define with p|c that p is a divisor of c; while with the symbol x we
show the floor of x, the biggest integer less or equal than x. Next with pk show
the k-th prime number.
PART ONE
Betrand’s Postulate
For better readability, we will introduce a series of terms, which are valid for
theorems, following the paper [4].
2n
Lemma 1 – For each n>0 is always 4n .
n
Proof
2n
2 n 2n
4n 22 n (1 1)2 n
h 0 h n
2
n
Lemma 2 – If n is odd for each 0 kn is always 2n 1 .
k
Proof
N is odd then:
n
n ( n 1) / 2
n n n
2 (1 1) 2 2 2n 1
n n
h 0 h h 0 h k k
Lemma 3 – If p is a prime number with p>n then allora p isn‟t a divisor of n!.
Proof
The proof is trivial because a prime p cannot be written as a product of primes
smaller than p. So if p> n then even if we have that:
n! = n *(n-1)*(n-2)…3*2*1,
p cannot be its divisor.
Lemma 4 – If a is a real number with a2, then the product of the prime number
less or equal than a is less than 4a: (a) p 4a
pa
Proof
For 2 a 3 it‟s trivial. If it‟s true for each odd integer n 3 it will be
also true for each real number b 3, because for any b 3 always exists a odd
n b n+2, then as b aboves n+2 so the number of prime numbers less than b
are equal to those of n:
pp4
p b pn
n
4b
So we have to prove that the Lemma is true for any integer n 3, and we do by
induction.
For n = 3 is true because 6 <64, which represents the starting point of the
induction.
For an odd n 5, let us assume that the theorem is true for every odd k < n
and we choose a sign such that k = (n 1)/2 is odd. In this mode is also k
3.
n n!
If p is a divisor of n! it involves that p| . This is true for k < p
k k !(n k )!
n. So we can say:
3
n n
p| p
k pn k k pn k
If we use the Lemma 2 now it is:
k pn
p 2n 1
p p
pn pk k pn
p
As we have previously speculated k odd and less than n, the first part is less
than 4K, while the second product that we know is less than or equal to 2 n-1,
which is:
p4
pn
k
2n1 22 k 2n1 22 k n1
p2
pn
2n
4n
2n 4n
Lemma 5 – If n > 1 then
n 2 n
Proof
2n
Now we are interested in such primes p divide and with what esponent.
n
n
v j (a)
j 1 p
The sum in (a) is always finite, because when j grows, it‟s pj > n and the ratio
tends to zero.
For example with p=2, the largest power that divides 55! is 250, infact from (a):
4
= 27 + 13 + 6 + 3 + 1 = 50.
With what we saw earlier, for example, we could indicate with vp the largest
v 2n
integer exponent such that p p | .
n
We also know that the prime factors of (2n)! are all less or equal than 2n; so:
2n
p
vp
n p2n
34
For example with PARI/GP for is:
17
? r = binomial(34,17)
%17 = 2333606220
? factor( r )
%18 =
[2 2]
[3 3]
[5 1]
[11 1]
[19 1]
[23 1]
[29 1]
[31 1]
The factorization on the right of the brackets shows the exponents v p we are
looking for, the binomial coefficient is decomposed into factors, then, that we
are finding; so the binomial coefficient is: 2^2*3^3*5*11*19*23*29*31 with v2=2,
v3=3, v5=1, v11=1, v19=23, v23=1, v29=1, v31=1.
2n n
Lemma 7 – For each n, vp j 2 j (c)
j 1 p p
2n
The (c) comes by (a), where instead of n there is , and by (b). In the (c)
n
the differences are 0 if the ratio is even, 1 otherwise.
p p 2n
v
Lemma 8 –
Proof
5
p t 2n
It means that for jt+1 pj > 2n, from t+1 onwards all the exponents are zero in
the (c). On the other hand, in the(c) the terms are 0 or 1 so vp t and
p p pt 2n . This proves the Lemma 8.
v
The Lemma 8 suggests a technique for the evaluation of exponents: not only must
be p 2n but also that pvp 2n.
For example if n=17 then we can establish that v3 cannot be greater than 3
because 2n=34 3^3 34; consequently for each 5 < p < 34 we have that vp
cannot be greater than 1.
With this technical we can work on the exponents and introduce the last Lemma.
Lemma 9 – We suppose n 3
1. If p 2n then vp 1
2. If 2n/3 < p n then vp = 0
3. If n < p 2n then vp = 1
Corollary
For each n1 pn1 2 pn
To demonstrate the Theorem we must use the Prime Numbers Theorem (PNT) and see
the upper and lower boundaries. On these types of reasoning Chebiscev was just a
true master.
2n
p p p
vp vp vp
(d)
n p2n p n n p 2n
We use Lemma 9, 3th point, the exponent values 1, so the (d) is:
2n
p p
vp
(e)
n pn n p 2 n
6
As 2nd point by Lemma 9, if 2n/3 < p n then vp = 0 so the (e) is:
2n
p p Qn Pn (†)
vp
n p2n / 3 n p 2n
At this point we want to find an upper boundary for Qn. We can break Qn in two
products to use the Lemma 9: the product of prime numbers for p 2n/3 where vp
= 1 and the product of prime numbers p 2n/3 where vp > 1.
p 2n p
vp
p
n p 2 n p 2n
Then:
Qn pp p pp p
v v v
p p (††)
p2n / 3 p2 n / 3 p 2 n / 3 p 2 n / 3 p 2n
1. If we use Lemma 4 is
p2n / 3
p 42 n / 3 .
prime numbers p 2n .
3. The prime numbers are odd and the number of prime numbers less than Y is
certainly less than Y/2 so we can write that: 2n / 2 and that:
p p (2n)
v 2n / 2
.
p 2n
Qn 42n / 3 (2n) 2n / 2
Qn 1 42 n / 3 (2n) 2n / 2
2n 2n 1 2n 2 n / 3
Qn Pn Pn Qn 4 (2n) 2n / 2
n n
n
so:
4n 2 n / 3
1
Pn 4 (2n) 2n / 2
4n / 3 2 n (2n) 2n / 2
2 n
If we say:
7
1
2n
(n)
4n / 3 2 n (2n) 2n / 2
so:
log 4n / 3 2 n (2n)
1
2n / 2
( n) (*)
log 2n
Pn
n p 2 n
p (+)
So we talk of the prime numbers between n and 2n, referenced from Betrand‟s
Postulate and we know by PNT that they are exactly (2n) (n) , so:
Pn 2n 2n
(2 n ) ( n )
Pn 2n 2n 2n Pn 2n
(2 n ) ( n ) (n) (2 n ) ( n )
then
(n) (2n) (n) (++)
If we go back to (*) we say that:
n
for n > 2190 ( n)
3log 2n
n
By (++) is (2n) (n) .
3log 2n
If we go back to (+) and if we use log, it is: log( Pn )
n p 2 n
log p .
But each p of the summation is p>n so log p > log n, then it is:
8
log( Pn ) log p (2n) (n) log n
n p 2 n
It is also:
2n
Pn and by Lemma 1:
n
2n
log Pn log log(4n ) n log(4) (2n) (n) log n n log(4)
n
So:
n log(4)
(2n) (n)
log n
7n
Now 7/5=1.4 and log(4)1.38629… so (2n) (n)
5log n
This is the demonstration of the Lemma 10, but it is also the demonstration of
Betrand‟s Postulate, because (2n) (n) 0 , in fact
n 7n
(2n) (n) .
3log 2n 5log n
PART TWO
n 1, (n2 n) (n2 ) 0
Corollary
n 1, (n2 n) (n2 ) ((n 1)2 ) (n2 ) ((n 1)2 ) (n2 ) 0
So it will be true the Legendre‟s Conjecture.
p 4n 22 n
2 2
Lemma 11 –
p n2
Proof
As Lemma 4.
9
Lemma 12 – pvp n(n 1)
Proof
n(n+1)
2
n ( n1)
Lemma 13 – per n>1, 2
n
Proof
They are true also Lemma 3, Lemma 4 and Lemma 6, means that for
v (n 1)n
n2 < p n(n+1), p is a divisor of n(n+1)! Consequently p p | 2 .
n
So we can break the binomial coefficient in the product of prime numbers in this
mode:
(n 1)n
p p p (***)
vp vp vp
2
n p ( n 1) n p n2 n 2 p ( n 1) n
n(n 1)
2
n
Lemma 14 – If n > 1 then è 2
n
Proof
10
n 1
The Lemma 8 here also suggests a technique for the evaluation of exponents: not
only must be p n(n+1) but also that pvp n(n+1).
If n=3 yhen we can say that v3 cannot be greater than 2 because n(n+1)=12
3^2 12; consequently for each 9 < p 12 we have also that vp cannot be
greater then 2.
Lemma 16 – If n 3
1. If p n(n 1) then vp 1
2. If n2 < p n(n+1) then vp = 1
Example
n=5
n2=25
n(n+1)=30
radice = sqrt(n(n+1))= 5.477225575051661134569697828
p=7
n(n+1)/3=10
By (***) we have:
(n 1)n
p p p
vp vp vp
2
n p ( n 1) n p n2 n 2 p ( n 1) n
pp p Qn Pn
v
p n2 n 2 p ( n 1) n
Where:
Qn p
vp
p n2
Pn p (^)
n 2 p ( n 1) n
If:
p n(n 1) p
vp
p
p n2 p n ( n 1)
Some considerations are:
11
1. by Lemma 11
p n ( n 1)
p 22 n ( n 1)
Then:
Qn p p p p p p
v v vp
p
p n2 p n2 p n2 p n2 p n ( n 1)
Qn 22 n n(n 1)
2 n ( n 1) / 2
By (***) we have:
(n 1)n
p p Qn Pn
vp
2
n p n2 n 2 p ( n 1) n
(n 1)n 1 (n 1)n 2 n2
2 n(n 1)
n ( n 1) / 2
Pn 2 Qn 2
n n
n(n 1)
n ( n 1)
2n (1 n ) 2n (1n )
n(n 1)
n ( n 1) / 2
Pn 2 2n n2
n(n 1) n(n 1)
n ( n 1)
n ( n 1)
If we write:
n(n 1)
n ( n 1)
(n)
2n (1n )
(n(n 1)) (=)
n(n 1)
n ( n 1)
So we have
Proof
12
By (=) and if we use the log, we obtain:
n (1 n )
n(n 1)
n ( n 1)
2
(n) log n(n 1) log
n(n 1)
n ( n 1)
For n the square bracket goes to zero because the numerator goes to the
infinite lower.
Now it is:
(n) 0
otherwise the (=) is 1 for the rules of powers: a finite number x with exponent
zero is x^0 = 1. It is a absurd.
By (^) is:
log( Pn ) log p .
n p ( n 1) n
2
n(n 1)
Pn 2 so:
n
n(n 1)
log( Pn ) log 2 log(4
n ( n 1)
) n(n 1) log 4 (n(n 1)) (n2 ) log n(n 1) n(n 1) log 4
n
This proves Lemma 17. At this point it is automatically shown that generalized
Bertrand's postulate is true and his Corollary:
13
References:
[1] http://it.wikipedia.org/wiki/Dimostrazione_del_postulato_di_Bertrand
[2] M. El Bachraoui. Primes in the Interval [2n,3n]. International Journal of
Contemporary Mathematical Sciences, 1(3):617–621, 2006.
[3] La congettura di Legendre – R. Turco, M. Colonnese et al.
[4] Il postulato di Bertrand e i primi di Ramanujan - Umberto Cerruti
[5] Towards Proving Legendre‟s Conjecture – Shiva Kintali
CNR SOLAR
http://150.146.3.132/
14