An Equivalent Problem To The Twin Prime Conjecture
Francesca Balestrieri∗
July 1, 2011
Abstract
In this short paper we will show, via elementary arguments, the equivalence of
the Twin Prime Conjecture to a problem which might be simpler to prove. Some
conclusions are drawn, and it is shown that proving the Twin Prime Conjecture is
equivalent to proving that there cannot be an infinite string of consecutive natural
numbers satisfying some specified equations.
1 Main Theorem
The main theorem of this paper is the following.
Theorem 1.1 (Main Theorem). The Twin Prime Conjecture is true if, and only if, there
exist infinitely many n ∈ N such that n 6= 6xy+x−y and n 6= 6xy+x+y and n 6= 6xy−x−y,
for all x, y ∈ N.
In other words, the Conjecture is true iff ∄N ∈ N such that ∀n ≥ N, n is of one of the forms
n = 6xy + x − y or n = 6xy + x + y or n = 6xy − x − y, for some x, y ∈ N.
In order to prove the Main Theorem, we will need to prove some preliminary results; two
thirds of this paper are devoted to this aim.
2 Preliminary Results
Consider the two sequences:
an = 6n + 1 (1)
bn = 6n − 1 (2)
A simple argument shows that these two sequences generate all the prime numbers (and
some other non-prime numbers). The following Lemma is a useful criterion which tells us
for which n the terms an and bn are non-prime, and hence, by complement, for which n the
terms an and bn are prime.
∗
fb340@cam.ac.uk
1Lemma 2.1. Let an and bn be the two sequences specified before. Then
an non − prime ⇐⇒ n = 6xy + x − y
bn non − prime ⇐⇒ n = 6xy + x + y or n = 6xy + x + y
for all x, y ∈ N.
To prove this Lemma, we need to prove some minor lemmata first.
2.1 Some proofs
Let A = {an : n ∈ N} and let B = {bn : n ∈ N}, where an and bn are the sequences (1) and
(2) respectively.
Lemma 2.2. It is not possible to express any term ak of the sequence an as the product ax ·ay
for any (ax, ay) ∈ A×A, nor to express it as the product bx · by for any (bx, by) ∈ B ×B. It is
possible to express a term ak of the sequence an as the product at
· br of a couple of numbers
(at
, br) ∈ A × B if, and only if, k = 6tr + t − r. In other words, ak = at
· br if, and only if,
k = 6tr + t − r.
Proof. Let ak = 6k − 1, at = 6t − 1, ax = 6x − 1, ay = 6y − 1 and br = 6r + 1.
The last part of the theorem is almost trivial. In fact,
at
· br
= (6t − 1) · (6r + 1)
= 36tr − 6r + 6t − 1
= 6(6tr − r + t) − 1
and hence ak = at
· br if, and only if, k = 6tr + t − r .
For the first part, consider
ax · ay
= (6x − 1) · (6y − 1)
= 36xy − 6x − 6y + 1
= 36xy − 6x − 6y + 2 − 1
= 2(18xy − 3x − 3y + 1) − 1.
Hence, we must show that 18xy − 3x − 3y + 1 is not divisible by 3. This is easy, since
18xy − 3x − 3y + 1 = 3(9xy − x − y) + 1 ≡ 1 (mod 3).
Therefore, there are no x, y ∈ N such that ax · ay = ak .
Similarly, consider
bx · by
= (6x + 1) · (6y + 1)
2= 36xy + 6x + 6y + 1
= 36xy + 6x + 6y + 2 − 1
= 2(18xy + 3x + 3y + 1) − 1.
Hence, we must show that 18xy+3x+3y+1 is not divisible by 3. Again, this is straightforward
since
18xy + 3x + 3y + 1 = 3(9xy + x + y) + 1 ≡ 1 (mod 3).
Therefore, there are no x, y ∈ N such that bx · by = ak .
A similar lemma can be proved for the terms of the sequence bn.
Lemma 2.3. It is not possible to express any term bk of the sequence bn as the product ax ·dy
for any (ax, dy) ∈ A×B. It is possible to express a term bk of the sequence bn as the product
at
· ar of a couple of numbers (at
, ar) ∈ A × A if, and only if, k = 6tr − t − r, and as the
product bt
· br of a couple of numbers (bt
, br) ∈ B ×B if, and only if, k = 6tr +t+r. In other
words, bk = at
·ar if, and only if, k = 6tr−t−r and bk = bt
· br if, and only if, k = 6tr+t+r.
Proof. Let bk = 6k + 1, at = 6t − 1, ar = 6r − 1, ax = 6x − 1, bt = 6t + 1, br = 6r + 1 and
by = 6y + 1.
The last part of the theorem is almost trivial. In fact,
at
· ar
= (6t − 1) · (6r − 1)
= 36tr − 6r − 6t + 1
= 6(6tr − r − t) + 1
and hence bk = at
· ar if, and only if, k = 6tr − t − r .
Also,
bt
· br
= (6t + 1) · (6r + 1)
= 36tr + 6r + 6t + 1
= 6(6tr + r + t) + 1
and hence bk = bt
· br if, and only if, k = 6tr + t + r .
For the first part, consider
ax · by
= (6x − 1) · (6y + 1)
= 36xy + 6x − 6y − 1
= 36xy + 6x − 6y − 2 + 1
= 2(18xy + 3x − 3y − 1) + 1.
Hence, we must show that 18xy + 3x − 3y − 1 is not divisible by 3. This is easy, since
18xy + 3x − 3y − 1 = 3(9xy + x − y) − 1 ≡ −1 (mod 3).
Therefore, there are no x, y ∈ N such that ax · by = bk .
3The following is an obvious result.
Lemma 2.4. Given any term ak of the sequence an, all the primes smaller than ak have
already been generated by the sequences an and bn for n < k . Similarly, given any term bk of
the sequence bn, all the primes smaller than bk have already been generated by the sequences
an and bn for n ≤ k.
Proof. Consider the sequence pn = (a1 , b1 , a2 , b2, a3 , b3 , . . . , ak , bk, . . .). Evidently, pn
is a strictly increasing sequence. Furthermore, pn contains all the prime numbers. Suppose
that one prime number p smaller than ak is not generated before the term ak; then, since all
the prime numbers are generated by the sequence pn, p must be generated after ak. But the
sequence pn is strictly increasing, and therefore p is greater than ak. This is a contradiction,
and hence all the primes smaller than ak are generated before ak.
The second half of the Lemma can be proved in a similar way.
2.2 More proofs
A few remarks and observations are now necessary.
2.2.1
Given a term ak, we have shown that
ak = at
· br ⇐⇒ k = 6tr + t − r;
otherwise, ak cannot be expressed as the product of any other pair of terms in A×A, B ×B
or A × B.
This can be generalised.
Note: An improper but self-evident use of notation will be made in the next paragraph.
Let ak be a non-prime term of the sequence an. Using Lemma 2.4, all the factors of ak
will be terms ai
, bj
for some i, j < k. Let say that α factors of ak belong to the set A, and
β factors of ak belong to the set B. Since ak is non-prime, α + β > 1. In short form,
ak = A
α
· B
β
(with α + β > 1),
where A (improperly) denotes an element of A and B (improperly) denotes an element of
B.
There are a few cases to consider, depending on the values of α and β. Notice that in
each case we will repetively use Lemma 2.2 and Lemma 2.3.
41. If α = 0, then for any natural value of β (> 1), we will have that ak = Bβ
. But Bβ
is
an element of the set B. Hence, we have an equation with an element of the set A in
the LHS and an element of the set B in the RHS, and A ∩ B = ∅. This is nonsense,
and hence it cannot be that α = 0.
2. If α is even, then for any β ∈ N ∪ {0}, we can write ak as
ak = A
α
· B
β = (A · A) · (A · A) · . . . · (A · A) · B
β = B
α/2
· B
β = B
(α/2)+β
and we have a situation analogous to the one in case 1. Hence, α cannot be even.
3. If α is odd, then α − 1 is positive even or zero) and, for any β ∈ N ∪ {0}, we can write
ak as
ak = A
α
· B
β = A · A
α−1
· B
β = A · B
(α−1)/2
· B
β = A · B
(α−1)/2+β
and this is consistent with what we have already proved, since A denotes an element
at of the set A and B(α−1)/2+β
is an element br of B, that is, A · B(α−1)/2+β = at
· br for
some t, r ∈ N.
So, if ak is non-prime, it is necessarely of the form ak = Aα
· Bβ
, where α is odd, β is
any number in N ∪ {0}, and (α + β) > 1. Of course, the converse is also true.
Hence, ak is non-prime ⇐⇒ ak = A
α
· Bβ where α is odd, β is any number in N ∪ {0},
and (α + β) > 1 ⇐⇒ ak = at
· br for some t, r ∈ N ⇐⇒ k = 6tr + t − r.
By Lemma 2.4, in all the other cases ak is prime.
2.2.2
Given a term bk, we have shown that
bk = at
· ar ⇐⇒ k = 6tr − t − r;
bk = bt
· br ⇐⇒ k = 6tr + t + r;
otherwise, bk cannot be expressed as the product of any other pair of terms in A × A, B × B
or A × B.
This, again, can be generalised.
Note: An improper but self-evident use of notation will be made in the next paragraph.
Let bk be a non-prime term of the sequence bn. Using Lemma 2.4, all the factors of bk
will be terms ai
, bj
for some i ≤ k and some j < k. Let say that α factors of bk belong to
the set A, and β factors of bk belong to the set B. Since bk is non-prime, α+ β > 1. In short
form,
bk = A
α
· B
β
(with α + β > 1),
where A (improperly) denotes an element of A and B (improperly) denotes an element of
B.
5There are a few cases to consider, depending on the values of α and β. Notice that in
each case we will repetively use Lemma 2.2 and Lemma 2.3.
1. If α = 0, then for any natural value of β (> 1), we will have that bk = Bβ
, and this is
consistent with what we have already proved, since B denotes an element bt of the set
B and Bβ−1
is an element br of B, that is, B · Bβ−1 = bt
· br for some t, r ∈ N.
2. If α is even, then for any β ∈ N ∪ {0}, we can write bk as
bk = A
α
· B
β = (A · A) · (A · A) · . . . · (A · A) · B
β = B
α/2
· B
β = B
(α/2)+β
and we have a situation analogous to the one in case 1, which is consistent.
Furthermore, if α is even, then α−2 is positive even or zero. Then for any β ∈ N∪{0},
we can write bk as
bk = A
α
· B
β = A · A · A
α−2
· B
β = A · A · B
(α−2)/2
· B
β = (A · B
(α−2)/2
) · (A · B
β
)
and this is consistent with what we have already proved, since (A · B(α−2)/2
) is an
element at of the set A and (A · Bβ
) is an element ar of the set A , that is, (A ·
B(α−2)/2
) · (A · Bβ
) = at
· ar for some t, r ∈ N.
3. If α is odd, then α − 1 is positive even or zero and, for any β ∈ N ∪ {0}, we can write
bk as
bk = A
α
· B
β = A · A
α−1
· B
β = A · B
(α−1)/2
· B
β = A · B
(α−1)/2+β
.
But A · B(α−1)/2+β
is an element of the set A, and hence we have an equation with
an element of the set B in the LHS and an element of the set A in the RHS. Since,
A ∩ B = ∅, this is nonsense, and therefore it cannot be that α is odd.
So, if bk is non-prime, it is necessarely of the form bk = Aα
· Bβ
, where α is pos
An Equivalent Problem To The Twin Prime ConjectureFrancesca Balestrieri∗July 1, 2011AbstractIn this short paper we will show, via elementary arguments, the equivalence ofthe Twin Prime Conjecture to a problem which might be simpler to prove. Someconclusions are drawn, and it is shown that proving the Twin Prime Conjecture isequivalent to proving that there cannot be an infinite string of consecutive naturalnumbers satisfying some specified equations.1 Main TheoremThe main theorem of this paper is the following.Theorem 1.1 (Main Theorem). The Twin Prime Conjecture is true if, and only if, thereexist infinitely many n ∈ N such that n 6= 6xy+x−y and n 6= 6xy+x+y and n 6= 6xy−x−y,for all x, y ∈ N.In other words, the Conjecture is true iff ∄N ∈ N such that ∀n ≥ N, n is of one of the formsn = 6xy + x − y or n = 6xy + x + y or n = 6xy − x − y, for some x, y ∈ N.In order to prove the Main Theorem, we will need to prove some preliminary results; twothirds of this paper are devoted to this aim.2 Preliminary ResultsConsider the two sequences:an = 6n + 1 (1)bn = 6n − 1 (2)A simple argument shows that these two sequences generate all the prime numbers (andsome other non-prime numbers). The following Lemma is a useful criterion which tells usfor which n the terms an and bn are non-prime, and hence, by complement, for which n theterms an and bn are prime.∗fb340@cam.ac.uk1Lemma 2.1. Let an and bn be the two sequences specified before. Thenan non − prime ⇐⇒ n = 6xy + x − ybn non − prime ⇐⇒ n = 6xy + x + y or n = 6xy + x + yfor all x, y ∈ N.To prove this Lemma, we need to prove some minor lemmata first.2.1 Some proofsLet A = {an : n ∈ N} and let B = {bn : n ∈ N}, where an and bn are the sequences (1) and(2) respectively.Lemma 2.2. It is not possible to express any term ak of the sequence an as the product ax ·ayfor any (ax, ay) ∈ A×A, nor to express it as the product bx · by for any (bx, by) ∈ B ×B. It ispossible to express a term ak of the sequence an as the product at· br of a couple of numbers(at, br) ∈ A × B if, and only if, k = 6tr + t − r. In other words, ak = at· br if, and only if,k = 6tr + t − r.Proof. Let ak = 6k − 1, at = 6t − 1, ax = 6x − 1, ay = 6y − 1 and br = 6r + 1.The last part of the theorem is almost trivial. In fact,at· br= (6t − 1) · (6r + 1)= 36tr − 6r + 6t − 1= 6(6tr − r + t) − 1and hence ak = at· br if, and only if, k = 6tr + t − r .For the first part, considerax · ay= (6x − 1) · (6y − 1)= 36xy − 6x − 6y + 1= 36xy − 6x − 6y + 2 − 1= 2(18xy − 3x − 3y + 1) − 1.Hence, we must show that 18xy − 3x − 3y + 1 is not divisible by 3. This is easy, since18xy − 3x − 3y + 1 = 3(9xy − x − y) + 1 ≡ 1 (mod 3).Therefore, there are no x, y ∈ N such that ax · ay = ak .Similarly, considerbx · by= (6x + 1) · (6y + 1)2= 36xy + 6x + 6y + 1= 36xy + 6x + 6y + 2 − 1= 2(18xy + 3x + 3y + 1) − 1.Hence, we must show that 18xy+3x+3y+1 is not divisible by 3. Again, this is straightforwardsince18xy + 3x + 3y + 1 = 3(9xy + x + y) + 1 ≡ 1 (mod 3).Therefore, there are no x, y ∈ N such that bx · by = ak .A similar lemma can be proved for the terms of the sequence bn.Lemma 2.3. It is not possible to express any term bk of the sequence bn as the product ax ·dyfor any (ax, dy) ∈ A×B. It is possible to express a term bk of the sequence bn as the productat· ar of a couple of numbers (at, ar) ∈ A × A if, and only if, k = 6tr − t − r, and as theproduct bt· br of a couple of numbers (bt, br) ∈ B ×B if, and only if, k = 6tr +t+r. In otherwords, bk = at·ar if, and only if, k = 6tr−t−r and bk = bt· br if, and only if, k = 6tr+t+r.Proof. Let bk = 6k + 1, at = 6t − 1, ar = 6r − 1, ax = 6x − 1, bt = 6t + 1, br = 6r + 1 andby = 6y + 1.The last part of the theorem is almost trivial. In fact,at· ar= (6t − 1) · (6r − 1)= 36tr − 6r − 6t + 1= 6(6tr − r − t) + 1and hence bk = at· ar if, and only if, k = 6tr − t − r .Also,bt· br= (6t + 1) · (6r + 1)= 36tr + 6r + 6t + 1= 6(6tr + r + t) + 1and hence bk = bt· br if, and only if, k = 6tr + t + r .For the first part, considerax · by= (6x − 1) · (6y + 1)= 36xy + 6x − 6y − 1= 36xy + 6x − 6y − 2 + 1= 2(18xy + 3x − 3y − 1) + 1.Hence, we must show that 18xy + 3x − 3y − 1 is not divisible by 3. This is easy, since18xy + 3x − 3y − 1 = 3(9xy + x − y) − 1 ≡ −1 (mod 3).Therefore, there are no x, y ∈ N such that ax · by = bk .3The following is an obvious result.Lemma 2.4. Given any term ak of the sequence an, all the primes smaller than ak havealready been generated by the sequences an and bn for n < k . Similarly, given any term bk ofthe sequence bn, all the primes smaller than bk have already been generated by the sequencesan and bn for n ≤ k.Proof. Consider the sequence pn = (a1 , b1 , a2 , b2, a3 , b3 , . . . , ak , bk, . . .). Evidently, pnis a strictly increasing sequence. Furthermore, pn contains all the prime numbers. Supposethat one prime number p smaller than ak is not generated before the term ak; then, since allthe prime numbers are generated by the sequence pn, p must be generated after ak. But thesequence pn is strictly increasing, and therefore p is greater than ak. This is a contradiction,and hence all the primes smaller than ak are generated before ak.The second half of the Lemma can be proved in a similar way.2.2 More proofsA few remarks and observations are now necessary.2.2.1Given a term ak, we have shown thatak = at· br ⇐⇒ k = 6tr + t − r;otherwise, ak cannot be expressed as the product of any other pair of terms in A×A, B ×Bor A × B.This can be generalised.Note: An improper but self-evident use of notation will be made in the next paragraph.Let ak be a non-prime term of the sequence an. Using Lemma 2.4, all the factors of akwill be terms ai, bjfor some i, j < k. Let say that α factors of ak belong to the set A, andβ factors of ak belong to the set B. Since ak is non-prime, α + β > 1. In short form,ak = Aα· Bβ(with α + β > 1),where A (improperly) denotes an element of A and B (improperly) denotes an element ofB.There are a few cases to consider, depending on the values of α and β. Notice that ineach case we will repetively use Lemma 2.2 and Lemma 2.3.41. If α = 0, then for any natural value of β (> 1), we will have that ak = Bβ. But Bβisan element of the set B. Hence, we have an equation with an element of the set A inthe LHS and an element of the set B in the RHS, and A ∩ B = ∅. This is nonsense,and hence it cannot be that α = 0.2. If α is even, then for any β ∈ N ∪ {0}, we can write ak asak = Aα· Bβ = (A · A) · (A · A) · . . . · (A · A) · Bβ = Bα/2· Bβ = B(α/2)+βand we have a situation analogous to the one in case 1. Hence, α cannot be even.3. If α is odd, then α − 1 is positive even or zero) and, for any β ∈ N ∪ {0}, we can writeak asak = Aα· Bβ = A · Aα−1· Bβ = A · B(α−1)/2· Bβ = A · B(α−1)/2+βand this is consistent with what we have already proved, since A denotes an elementat of the set A and B(α−1)/2+β
is an element br of B, that is, A · B(α−1)/2+β = at
· br for
some t, r ∈ N.
So, if ak is non-prime, it is necessarely of the form ak = Aα
· Bβ
, where α is odd, β is
any number in N ∪ {0}, and (α + β) > 1. Of course, the converse is also true.
Hence, ak is non-prime ⇐⇒ ak = A
α
· Bβ where α is odd, β is any number in N ∪ {0},
and (α + β) > 1 ⇐⇒ ak = at
· br for some t, r ∈ N ⇐⇒ k = 6tr + t − r.
By Lemma 2.4, in all the other cases ak is prime.
2.2.2
Given a term bk, we have shown that
bk = at
· ar ⇐⇒ k = 6tr − t − r;
bk = bt
· br ⇐⇒ k = 6tr + t + r;
otherwise, bk cannot be expressed as the product of any other pair of terms in A × A, B × B
or A × B.
This, again, can be generalised.
Note: An improper but self-evident use of notation will be made in the next paragraph.
Let bk be a non-prime term of the sequence bn. Using Lemma 2.4, all the factors of bk
will be terms ai
, bj
for some i ≤ k and some j < k. Let say that α factors of bk belong to
the set A, and β factors of bk belong to the set B. Since bk is non-prime, α+ β > 1. In short
form,
bk = A
α
· B
β
(with α + β > 1),
where A (improperly) denotes an element of A and B (improperly) denotes an element of
B.
5There are a few cases to consider, depending on the values of α and β. Notice that in
each case we will repetively use Lemma 2.2 and Lemma 2.3.
1. If α = 0, then for any natural value of β (> 1), we will have that bk = Bβ
, and this is
consistent with what we have already proved, since B denotes an element bt of the set
B and Bβ−1
is an element br of B, that is, B · Bβ−1 = bt
· br for some t, r ∈ N.
2. If α is even, then for any β ∈ N ∪ {0}, we can write bk as
bk = A
α
· B
β = (A · A) · (A · A) · . . . · (A · A) · B
β = B
α/2
· B
β = B
(α/2)+β
and we have a situation analogous to the one in case 1, which is consistent.
Furthermore, if α is even, then α−2 is positive even or zero. Then for any β ∈ N∪{0},
we can write bk as
bk = A
α
· B
β = A · A · A
α−2
· B
β = A · A · B
(α−2)/2
· B
β = (A · B
(α−2)/2
) · (A · B
β
)
and this is consistent with what we have already proved, since (A · B(α−2)/2
) is an
element at of the set A and (A · Bβ
) is an element ar of the set A , that is, (A ·
B(α−2)/2
) · (A · Bβ
) = at
· ar for some t, r ∈ N.
3. If α is odd, then α − 1 is positive even or zero and, for any β ∈ N ∪ {0}, we can write
bk as
bk = A
α
· B
β = A · A
α−1
· B
β = A · B
(α−1)/2
· B
β = A · B
(α−1)/2+β
.
But A · B(α−1)/2+β
is an element of the set A, and hence we have an equation with
an element of the set B in the LHS and an element of the set A in the RHS. Since,
A ∩ B = ∅, this is nonsense, and therefore it cannot be that α is odd.
So, if bk is non-prime, it is necessarely of the form bk = Aα
· Bβ
, where α is pos
การแปล กรุณารอสักครู่..

An Equivalent Problem To The Twin Prime Conjecture
Francesca Balestrieri∗
July 1, 2011
Abstract
In this short paper we will show, via elementary arguments, the equivalence of
the Twin Prime Conjecture to a problem which might be simpler to prove. Some
conclusions are drawn, and it is shown that proving the Twin Prime Conjecture is
equivalent to proving that there cannot be an infinite string of consecutive natural
numbers satisfying some specified equations.
1 Main Theorem
The main theorem of this paper is the following.
Theorem 1.1 (Main Theorem). The Twin Prime Conjecture is true if, and only if, there
exist infinitely many n ∈ N such that n 6= 6xy+x−y and n 6= 6xy+x+y and n 6= 6xy−x−y,
for all x, y ∈ N.
In other words, the Conjecture is true iff ∄N ∈ N such that ∀n ≥ N, n is of one of the forms
n = 6xy + x − y or n = 6xy + x + y or n = 6xy − x − y, for some x, y ∈ N.
In order to prove the Main Theorem, we will need to prove some preliminary results; two
thirds of this paper are devoted to this aim.
2 Preliminary Results
Consider the two sequences:
an = 6n + 1 (1)
bn = 6n − 1 (2)
A simple argument shows that these two sequences generate all the prime numbers (and
some other non-prime numbers). The following Lemma is a useful criterion which tells us
for which n the terms an and bn are non-prime, and hence, by complement, for which n the
terms an and bn are prime.
∗
fb340@cam.ac.uk
1Lemma 2.1. Let an and bn be the two sequences specified before. Then
an non − prime ⇐⇒ n = 6xy + x − y
bn non − prime ⇐⇒ n = 6xy + x + y or n = 6xy + x + y
for all x, y ∈ N.
To prove this Lemma, we need to prove some minor lemmata first.
2.1 Some proofs
Let A = {an : n ∈ N} and let B = {bn : n ∈ N}, where an and bn are the sequences (1) and
(2) respectively.
Lemma 2.2. It is not possible to express any term ak of the sequence an as the product ax ·ay
for any (ax, ay) ∈ A×A, nor to express it as the product bx · by for any (bx, by) ∈ B ×B. It is
possible to express a term ak of the sequence an as the product at
· br of a couple of numbers
(at
, br) ∈ A × B if, and only if, k = 6tr + t − r. In other words, ak = at
· br if, and only if,
k = 6tr + t − r.
Proof. Let ak = 6k − 1, at = 6t − 1, ax = 6x − 1, ay = 6y − 1 and br = 6r + 1.
The last part of the theorem is almost trivial. In fact,
at
· br
= (6t − 1) · (6r + 1)
= 36tr − 6r + 6t − 1
= 6(6tr − r + t) − 1
and hence ak = at
· br if, and only if, k = 6tr + t − r .
For the first part, consider
ax · ay
= (6x − 1) · (6y − 1)
= 36xy − 6x − 6y + 1
= 36xy − 6x − 6y + 2 − 1
= 2(18xy − 3x − 3y + 1) − 1.
Hence, we must show that 18xy − 3x − 3y + 1 is not divisible by 3. This is easy, since
18xy − 3x − 3y + 1 = 3(9xy − x − y) + 1 ≡ 1 (mod 3).
Therefore, there are no x, y ∈ N such that ax · ay = ak .
Similarly, consider
bx · by
= (6x + 1) · (6y + 1)
2= 36xy + 6x + 6y + 1
= 36xy + 6x + 6y + 2 − 1
= 2(18xy + 3x + 3y + 1) − 1.
Hence, we must show that 18xy+3x+3y+1 is not divisible by 3. Again, this is straightforward
since
18xy + 3x + 3y + 1 = 3(9xy + x + y) + 1 ≡ 1 (mod 3).
Therefore, there are no x, y ∈ N such that bx · by = ak .
A similar lemma can be proved for the terms of the sequence bn.
Lemma 2.3. It is not possible to express any term bk of the sequence bn as the product ax ·dy
for any (ax, dy) ∈ A×B. It is possible to express a term bk of the sequence bn as the product
at
· ar of a couple of numbers (at
, ar) ∈ A × A if, and only if, k = 6tr − t − r, and as the
product bt
· br of a couple of numbers (bt
, br) ∈ B ×B if, and only if, k = 6tr +t+r. In other
words, bk = at
·ar if, and only if, k = 6tr−t−r and bk = bt
· br if, and only if, k = 6tr+t+r.
Proof. Let bk = 6k + 1, at = 6t − 1, ar = 6r − 1, ax = 6x − 1, bt = 6t + 1, br = 6r + 1 and
by = 6y + 1.
The last part of the theorem is almost trivial. In fact,
at
· ar
= (6t − 1) · (6r − 1)
= 36tr − 6r − 6t + 1
= 6(6tr − r − t) + 1
and hence bk = at
· ar if, and only if, k = 6tr − t − r .
Also,
bt
· br
= (6t + 1) · (6r + 1)
= 36tr + 6r + 6t + 1
= 6(6tr + r + t) + 1
and hence bk = bt
· br if, and only if, k = 6tr + t + r .
For the first part, consider
ax · by
= (6x − 1) · (6y + 1)
= 36xy + 6x − 6y − 1
= 36xy + 6x − 6y − 2 + 1
= 2(18xy + 3x − 3y − 1) + 1.
Hence, we must show that 18xy + 3x − 3y − 1 is not divisible by 3. This is easy, since
18xy + 3x − 3y − 1 = 3(9xy + x − y) − 1 ≡ −1 (mod 3).
Therefore, there are no x, y ∈ N such that ax · by = bk .
3The following is an obvious result.
Lemma 2.4. Given any term ak of the sequence an, all the primes smaller than ak have
already been generated by the sequences an and bn for n < k . Similarly, given any term bk of
the sequence bn, all the primes smaller than bk have already been generated by the sequences
an and bn for n ≤ k.
Proof. Consider the sequence pn = (a1 , b1 , a2 , b2, a3 , b3 , . . . , ak , bk, . . .). Evidently, pn
is a strictly increasing sequence. Furthermore, pn contains all the prime numbers. Suppose
that one prime number p smaller than ak is not generated before the term ak; then, since all
the prime numbers are generated by the sequence pn, p must be generated after ak. But the
sequence pn is strictly increasing, and therefore p is greater than ak. This is a contradiction,
and hence all the primes smaller than ak are generated before ak.
The second half of the Lemma can be proved in a similar way.
2.2 More proofs
A few remarks and observations are now necessary.
2.2.1
Given a term ak, we have shown that
ak = at
· br ⇐⇒ k = 6tr + t − r;
otherwise, ak cannot be expressed as the product of any other pair of terms in A×A, B ×B
or A × B.
This can be generalised.
Note: An improper but self-evident use of notation will be made in the next paragraph.
Let ak be a non-prime term of the sequence an. Using Lemma 2.4, all the factors of ak
will be terms ai
, bj
for some i, j < k. Let say that α factors of ak belong to the set A, and
β factors of ak belong to the set B. Since ak is non-prime, α + β > 1. In short form,
ak = A
α
· B
β
(with α + β > 1),
where A (improperly) denotes an element of A and B (improperly) denotes an element of
B.
There are a few cases to consider, depending on the values of α and β. Notice that in
each case we will repetively use Lemma 2.2 and Lemma 2.3.
41. If α = 0, then for any natural value of β (> 1), we will have that ak = Bβ
. But Bβ
is
an element of the set B. Hence, we have an equation with an element of the set A in
the LHS and an element of the set B in the RHS, and A ∩ B = ∅. This is nonsense,
and hence it cannot be that α = 0.
2. If α is even, then for any β ∈ N ∪ {0}, we can write ak as
ak = A
α
· B
β = (A · A) · (A · A) · . . . · (A · A) · B
β = B
α/2
· B
β = B
(α/2)+β
and we have a situation analogous to the one in case 1. Hence, α cannot be even.
3. If α is odd, then α − 1 is positive even or zero) and, for any β ∈ N ∪ {0}, we can write
ak as
ak = A
α
· B
β = A · A
α−1
· B
β = A · B
(α−1)/2
· B
β = A · B
(α−1)/2+β
and this is consistent with what we have already proved, since A denotes an element
at of the set A and B(α−1)/2+β
is an element br of B, that is, A · B(α−1)/2+β = at
· br for
some t, r ∈ N.
So, if ak is non-prime, it is necessarely of the form ak = Aα
· Bβ
, where α is odd, β is
any number in N ∪ {0}, and (α + β) > 1. Of course, the converse is also true.
Hence, ak is non-prime ⇐⇒ ak = A
α
· Bβ where α is odd, β is any number in N ∪ {0},
and (α + β) > 1 ⇐⇒ ak = at
· br for some t, r ∈ N ⇐⇒ k = 6tr + t − r.
By Lemma 2.4, in all the other cases ak is prime.
2.2.2
Given a term bk, we have shown that
bk = at
· ar ⇐⇒ k = 6tr − t − r;
bk = bt
· br ⇐⇒ k = 6tr + t + r;
otherwise, bk cannot be expressed as the product of any other pair of terms in A × A, B × B
or A × B.
This, again, can be generalised.
Note: An improper but self-evident use of notation will be made in the next paragraph.
Let bk be a non-prime term of the sequence bn. Using Lemma 2.4, all the factors of bk
will be terms ai
, bj
for some i ≤ k and some j < k. Let say that α factors of bk belong to
the set A, and β factors of bk belong to the set B. Since bk is non-prime, α+ β > 1. In short
form,
bk = A
α
· B
β
(with α + β > 1),
where A (improperly) denotes an element of A and B (improperly) denotes an element of
B.
5There are a few cases to consider, depending on the values of α and β. Notice that in
each case we will repetively use Lemma 2.2 and Lemma 2.3.
1. If α = 0, then for any natural value of β (> 1), we will have that bk = Bβ
, and this is
consistent with what we have already proved, since B denotes an element bt of the set
B and Bβ−1
is an element br of B, that is, B · Bβ−1 = bt
· br for some t, r ∈ N.
2. If α is even, then for any β ∈ N ∪ {0}, we can write bk as
bk = A
α
· B
β = (A · A) · (A · A) · . . . · (A · A) · B
β = B
α/2
· B
β = B
(α/2)+β
and we have a situation analogous to the one in case 1, which is consistent.
Furthermore, if α is even, then α−2 is positive even or zero. Then for any β ∈ N∪{0},
we can write bk as
bk = A
α
· B
β = A · A · A
α−2
· B
β = A · A · B
(α−2)/2
· B
β = (A · B
(α−2)/2
) · (A · B
β
)
and this is consistent with what we have already proved, since (A · B(α−2)/2
) is an
element at of the set A and (A · Bβ
) is an element ar of the set A , that is, (A ·
B(α−2)/2
) · (A · Bβ
) = at
· ar for some t, r ∈ N.
3. If α is odd, then α − 1 is positive even or zero and, for any β ∈ N ∪ {0}, we can write
bk as
bk = A
α
· B
β = A · A
α−1
· B
β = A · B
(α−1)/2
· B
β = A · B
(α−1)/2+β
.
But A · B(α−1)/2+β
is an element of the set A, and hence we have an equation with
an element of the set B in the LHS and an element of the set A in the RHS. Since,
A ∩ B = ∅, this is nonsense, and therefore it cannot be that α is odd.
So, if bk is non-prime, it is necessarely of the form bk = Aα
· Bβ
, where α is pos
การแปล กรุณารอสักครู่..
