This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

On Primorial Numbers

Jonatan Gomez
jgomezpe@unal.edu.co
Universidad Nacional de Colombia
Abstract.

Prime numbers have attracted the attention of mathematicians and enthusiasts for millenniums due to their simple definition and remarkable properties. In this paper, we study primorial numbers (the product of the first prime numbers) to define primorial sets, primorial intervals, primorial tables, and primorial totative numbers. We establish relationships between prime numbers and primorial totative numbers and between admissible kk-tuples of prime numbers and admissible kk-tuples of primorial totative. Finally, we study the Goldbach conjecture and derive four Goldbach conjectures using primorial intervals, twin, cousin, and sexy prime numbers.

1. Introduction

1.1. Prime Numbers

A prime number is a natural number having exactly two factors, 11 and itself [Nar00]. The natural number 11 is not considered a prime number since it has only one factor (11). From now on, pnp_{n} will denote the nnth prime number, here n1n\geq 1. The first eleven prime numbers are 2,3,5,7,11,13,17,19,23,29,312,3,5,7,11,13,17,19,23,29,31. Although different classes of prime numbers have been defined [Wik], in this paper, we especially concentrate on some prime number classes derived from the notions of admissible kk-tuple and kk-constellation [For99, Wikb].

1.2. Admissible tuples

A kk-tuple a=(a1,a2,,ak)a=(a_{1},a_{2},\ldots,a_{k}) of natural numbers is admissible if and only if a1=0a_{1}=0, ai1<aia_{i-1}<a_{i} for all i=2,3,,ki=2,3,\ldots,k, and there is a prime number p>3p>3 such that (p,p+a2,,p+ak)(p,p+a_{2},\ldots,p+a_{k}) is a kk-tuple of prime numbers, i.e. p+aip+a_{i} is a prime number for all i=1,2,,ki=1,2,\ldots,k. We say that a prime number p>3p>3 satisfies an admissible kk-tuple aa (we denote this situation as [p|a][p|a]) if and only if (p,p+a2,,p+ak)(p,p+a_{2},\ldots,p+a_{k}) is a kk-tuple of prime numbers. The diameter of an admissible kk-tuple aa is dia(a)=akdia(a)=a_{k} while its gap is the maximum difference between two of its consecutive elements, i.e. gap(a)=max{aiai1|i=2,3,,k}gap(a)=max\left\{a_{i}-a_{i-1}\,\middle|\,i=2,3,\ldots,k\right\}. For example,

  1. (1)

    a=(0,2)a=(0,2) is an admissible 22-tuple. If p=5p=5, then 5+0=55+0=5 and 5+2=75+2=7, so (5,7)(5,7) is a 22-tuple of prime numbers. Clearly, dia(a)=a2=2dia(a)=a_{2}=2 and gap(a)=max{20}=2gap(a)=max\{2-0\}=2.

  2. (2)

    a=(0,2,6)a=(0,2,6) is an admissible 33-tuple. If p=5p=5 then 5+0=55+0=5, 5+2=75+2=7, and 5+6=115+6=11, so (5,7,11)(5,7,11) is a 33-tuple of prime numbers. Clearly, dia(a)=a2=6dia(a)=a_{2}=6 and gap(a)=max{20,62}=4gap(a)=max\{2-0,6-2\}=4.

  3. (3)

    (0,1)(0,1) is not admissible: since p>3p>3 is a prime number then p=2x+1p=2x+1 (odd number, i.e., not a multiple of 22). Then p+1=2x+1+1=2x+2=2(x+1)p+1=2x+1+1=2x+2=2(x+1) is an even number, therefore it is not a prime number (contradiction).

  4. (4)

    (0,2,4)(0,2,4) is not admissible 33-tuple: since p>3p>3 then p=3x+yp=3x+y with y=1,2y=1,2 (pp is prime so it is not a multiple of 33). If y=1y=1 then p+2=3x+1+2=3x+3)3(x+1)p+2=3x+1+2=3x+3)3(x+1) so p+2p+2 is a multiple of 33 then it is not a prime number (contradiction). If y=2y=2 then p+4=3x+2+4=3x+6=3(x+2)p+4=3x+2+4=3x+6=3(x+2). So p+4p+4 is a multiple of 33 then it is not a prime number (contradiction).

1.3. Constellations

An admissible kk-tuple a=(a1,a2,,ak)a=(a_{1},a_{2},\ldots,a_{k}) is kk-constellation if and only if for any admissible kk-tuple b=(b1,b2,,bk)b=(b_{1},b_{2},\ldots,b_{k}) we have that dia(a)dia(b)dia(a)\leq dia(b). For example, (0,2)(0,2), (0,4)(0,4), and (0,6)(0,6) are admissible 22-tuples but only (0,2)(0,2) is a 22-constellation (2462\leq 4\leq 6).

1.4. Prime Number Classes

The following is a short list of prime number classes defined in terms of admissible kk-tuples and kk-constellations:

  1. (1)

    Twin primes. Two prime numbers pp and qq are twin primes if and only if q=p+2q=p+2. For example, primes 55 and 77 are twin primes since they define the prime 22-tuple (5,5+2)(5,5+2) which satisfies the 22-prime constellation (0,2)(0,2).

  2. (2)

    Cousin primes. Two prime numbers pp and qq are called cousin primes if and only if q=p+4q=p+4. For example, primes 77 and 1111 are cousin primes since they define the prime 22-tuple (7,7+4)(7,7+4) which satisfies the admissible 22-tuple (0,4)(0,4).

  3. (3)

    Sexy primes. Two prime numbers pp and qq are called sexy primes if and only if q=p+6q=p+6. For example, primes 55 and 1111 are sexy primes since they define the prime 22-tuple (5,5+6)(5,5+6) which satisfies the admissible 22-tuple (0,6)(0,6).

  4. (4)

    Sexy primes triplet. Three prime numbers pp, qq, and rr are called a sexy prime triplet if and only if q=p+6q=p+6 and r=p+12r=p+12. For example, primes 55, 1111, and 1717 are a sexy prime triplet since they define the prime 33-tuple (5,5+6,5+12)(5,5+6,5+12) which satisfies the admissible 33-tuple (0,6,12)(0,6,12).

  5. (5)

    Sexy primes quadruplet. Four prime numbers p,q,r,p,q,r, and ss are called a sexy prime quadruple if and only if q=p+6q=p+6, r=p+12r=p+12, and s=p+18s=p+18. For example, primes 55, 1111, 1717, and 2323 are a sexy prime quadruplet since they define the prime 44-tuple (5,5+6,5+12,5+18)(5,5+6,5+12,5+18) which satisfies the admissible 44-tuple (0,6,12,18)(0,6,12,18).

1.5. Primorial

Many interesting operations can be defined over prime numbers, for example, we can define the primorial of the nnth prime number as the product of the first n+n\in\mathbb{N}^{+} prime numbers [Dub87], i.e., #(n)=pn#=i=1npk\#(n)=p_{n}\#=\prod_{i=1}^{n}p_{k}. This definition is similar to the definition of the factorial function, so it is possible to define the primorial as a recursive function #(0)=1\#(0)=1 and #(n)=pn#(n1)\#(n)=p_{n}\#(n-1) for n1n\geq 1. The first six primorials are 1,2,6,30,210,23101,2,6,30,210,2310.

1.6. Coprime or Relative-prime Numbers

Two natural numbers a,ba,b\in\mathbb{N} are coprime or relative-prime numbers if and only if their greatest common divisor is 11 (gcd(a,b)=1gcd(a,b)=1). According to this definition, i) 11 is relative-prime with any other positive natural number, ii) 0 is not relative-prime with any natural number, and iii) any prime number is relative-prime with any other prime number. Notice that we can define prime numbers in terms of coprime numbers: A natural number p>1p>1 is a prime number if and only if pp is relative-prime to qq for all natural numbers 1q<p1\leq q<p.

1.7. Natural numbers less than nn (n\mathbb{Z}_{n})

Let nn\in\mathbb{N}, the set of natural numbers less than nn, is n={0,1,,n1}\mathbb{Z}_{n}=\{0,1,\ldots,n-1\}. For example, 2={0,1}\mathbb{Z}_{2}=\{0,1\} and #(2)=23=6={0,1,3,4,5}\mathbb{Z}_{\#(2)}=\mathbb{Z}_{2*3}=\mathbb{Z}_{6}=\{0,1,3,4,5\}. Amount the properties hold by n\mathbb{Z}_{n}, we are especially interested in the structure of the subset of natural numbers that are relative-prime to #(n)\#(n) (sometimes called as totative numbers of #(n)\#(n)). For instance, consider #(2)=6\mathbb{Z}_{\#(2)}=\mathbb{Z}_{6}, the subset of 6\mathbb{Z}_{6} defined by the natural numbers that are relative-prime to 66 is {1,5}\{1,5\}. It is clear that any prime number qq such pn<q<#(n)p_{n}<q<\#(n) will be a relative-prime number to #(n)\#(n).

1.8. Euler’s totient function (φ(n)\varphi(n))

Euler’s totient function [Eul63] counts the natural numbers, in n\mathbb{Z}_{n}, which are relative-prime to nn. Take for example #(3)=30\mathbb{Z}_{\#(3)}=\mathbb{Z}_{30}, the subset of natural numbers relative-prime to 3030 is {1,7,11,13,17,19,23,29}\{1,7,11,13,17,19,23,29\}, therefore φ(30)=8\varphi(30)=8. Euler’s totient function is a multiplicative function, i.e., φ(ab)=φ(a)φ(b)\varphi(ab)=\varphi(a)\varphi(b) for two coprime numbers aa and bb. Moreover, for a prime number pp, we have that p{0}\mathbb{Z}_{p}-\{0\} is the set of totative numbers of pp, therefore φ(p)=p1\varphi(p)=p-1. Using these two properties of Euler’s totient function, we can easily compute it on primorial numbers: φ(#(n))=i=1n(pk1)=i=1nφ(pk)\varphi(\#(n))=\prod_{i=1}^{n}(p_{k}-1)=\prod_{i=1}^{n}\varphi(p_{k}).

2. Primorial Sets, Primorial Intervals, and Primorial Totatives

2.1. Primorial sets

We define a nn-primorial set by ’replacing’ numbers 0 and 11 in #(n)\mathbb{Z}_{\#(n)}, with numbers #(n)\#(n) and #(n)+1\#(n)+1, respectively. We do this since the number 11 is always coprime to another positive natural number. Let nn\in\mathbb{N}, the nn-primorial set is defined as n#={2,3,,#(n),#(n)+1}\mathbb{Z}^{\#}_{n}=\{2,3,\ldots,\#(n),\#(n)+1\}. The following is the list of the first five nn-primorial sets:

  1. (1)

    1#={2,3}\mathbb{Z}^{\#}_{1}=\{2,3\}

  2. (2)

    2#={2,3,4,5,6,7}\mathbb{Z}^{\#}_{2}=\{2,3,4,5,6,7\}

  3. (3)

    3#={2,3,,30,31}\mathbb{Z}^{\#}_{3}=\{2,3,\ldots,30,31\}

  4. (4)

    4#={2,3,,210,211}\mathbb{Z}^{\#}_{4}=\{2,3,\ldots,210,211\}

  5. (5)

    5#={2,3,,2310,2311}\mathbb{Z}^{\#}_{5}=\{2,3,\ldots,2310,2311\}

2.2. Primorial intervals

Let nn\in\mathbb{N}, the nn-primorial interval is defined as n={m|#(n1)+1x#(n)+1}\mathcal{I}_{n}=\left\{m\,\middle|\,\#(n-1)+1\leq x\leq\#(n)+1\right\}. The following is the list of the first five nn-primorial intervals:

  1. (1)

    1={2,3}\mathcal{I}_{1}=\{2,3\}

  2. (2)

    2={3,4,5,6,7}\mathcal{I}_{2}=\{3,4,5,6,7\}

  3. (3)

    3={7,8,,30,31}\mathcal{I}_{3}=\{7,8,\ldots,30,31\}

  4. (4)

    4={31,32,,210,211}\mathcal{I}_{4}=\{31,32,\ldots,210,211\}

  5. (5)

    5={211,212,,2310,2311}\mathcal{I}_{5}=\{211,212,\ldots,2310,2311\}

Notice that primorial intervals are not disjoint sets: the last element of n\mathcal{I}_{n} is the first element of n+1\mathcal{I}_{n+1}.

2.3. Primorial nn-totatives

We can extend the notion of totatives of #(n)\#(n) to set n#\mathbb{Z}^{\#}_{n} by considering the subset of natural numbers in n#\mathbb{Z}^{\#}_{n} being relative-prime to #(n)\#(n). The following is the list of the first three nn-totative sets:

  1. (1)

    The 11-totative set is tot(1)={3}tot(1)=\{3\}

  2. (2)

    The 22-totative set is tot(2)={5,7}tot(2)=\{5,7\}

  3. (3)

    The 33-totative set is tot(3)={7,11,13,17,19,23,29,31}tot(3)=\{7,11,13,17,19,23,29,31\}

Notice that any prime number pn#p\in\mathbb{Z}^{\#}_{n} must be either pip_{i} for some i=1,2,,ni=1,2,\ldots,n or a nn-totative number, i.e., we can express the Prime Number Theorem in terms of primorials and nn-totative numbers. We will explore this relationship in a future paper. In this section, we just extend definitions in sections 1.2-1.4 to nn-totative numbers.

2.4. nn-admissible

A kk-tuple a=(a1,a2,,ak)a=(a_{1},a_{2},\ldots,a_{k}) of natural numbers is nn-admissible if and only if a1=0a_{1}=0, ai1<aia_{i-1}<a_{i} for all i=2,3,,ki=2,3,\ldots,k, and for all mnm\geq n there is a mm-totative number tmt_{m} such that (tm,tm+a2,,tm+ak)(t_{m},t_{m}+a_{2},\ldots,t_{m}+a_{k}) is a mm-totative kk-tuple. Notice that a nn-admissible kk-tuple is also a mm-admissible kk-tuple for all m>nm>n. A nn-totative number tt satisfies an nn-admissible kk-tuple aa if and only if a[t]=(t,t+a2,,t+ak)a[t]=(t,t+a_{2},\ldots,t+a_{k}) is a kk-tuple of nn-totative numbers. The gap of a kk tuple aa is the maximum difference between two consecutive elements of the kk-tuple, i.e., gap(a)=max{aiai1|i=2,3,,k}gap(a)=max\left\{a_{i}-a_{i-1}\,\middle|\,i=2,3,\ldots,k\right\}.

2.5. nn-strong

A nn-admissible kk-tuple a=(a1,a2,,ak)a=(a_{1},a_{2},\ldots,a_{k}) is nn-strong if and only if for all mnm\geq n we have that mod(ai,pm)mod(aj,pm)mod(a_{i},p_{m})\neq mod(a_{j},p_{m}) with i=1,,k1i=1,\ldots,k-1 and j=i+1,,kj=i+1,\ldots,k. Notice that being nn-strong implies that kpnk\leq p_{n}.

2.6. nn-constellation

A nn-admissible kk-tuple a=(a1,a2,,ak)a=(a_{1},a_{2},\ldots,a_{k}) is nn-totative kk-constellation if and only if for akbka_{k}\leq b_{k} for any nn-admissible kk-tuple b=(b1,b2,,bk)b=(b_{1},b_{2},\ldots,b_{k}).

2.7. Classes of primorial nn-totative numbers

The following is a short list of nn-totative numbers classes:

  1. (1)

    Isolated. Let a=(0,a2,,ak)a=(0,a_{2},\ldots,a_{k}) a nn-admissible kk-tuple, and tt a nn-totative number such that a[t]a[t] is a nn-totative kk-tuple. a[t]a[t] is isolated if and only if both a[ta2]a[t-a_{2}] and a=[t2]a=[t_{2}] are not nn-totative kk-tuples.

  2. (2)

    nn-Twin. Two nn-totative numbers pp and qq define a nn-twin couple if and only if q=p+2q=p+2. Notice that a nn-twin couple satisfies the nn-strong nn-totative 22-constellation (0,2)(0,2). Moreover, it is an isolated 22-tuple when n>2n>2 since the 33-tuple (0,2,4)(0,2,4) is not a nn-admissible 33-tuple: tt, t+2t+2, and t+4t+4 all must be coprime to 33, therefore t=3s+1t=3s+1 or t=3s+2t=3s+2 for some natural number ss. If t=3s+1t=3s+1 then t+2=3s+1+2=3(s+1)t+2=3s+1+2=3(s+1), so t+2t+2 is not coprime to 33. Now if t=3s+2t=3s+2 then t+4=3s+2+4=3(s+2)t+4=3s+2+4=3(s+2), so t+4t+4 is not coprime to 33.

  3. (3)

    nn-Cousin. Two nn-totative numbers pp and qq are a nn-cousin couple if and only if q=p+4q=p+4. Notice that a nn-cousin couple satisfies the nn-strong nn-admissible 22-tuple (0,4)(0,4). Moreover, it is an isolated 22-tuple when n>3n>3 ((0,4,8)(0,4,8) is not a nn-admissible 33-tuple, we left the proof to the reader).

  4. (4)

    nn-Sexy. Two nn-totative numbers pp and qq are a nn-sexy couple if and only if q=p+6q=p+6. Notice that a nn-sexy couple (p,p+6)(p,p+6) satisfies the nn-admissible 22-tuple (0,6)(0,6). Moreover, it is isolated if and only if p6p-6 and p+12p+12 are not nn-totative.

  5. (5)

    nn-Sexy triplet. Three nn-totative numbers pp, qq, and rr are a nn-sexy triplet iff q=p+6q=p+6 and r=p+12r=p+12. Notice that a nn-sexy triplet (p,p+6,p+12)(p,p+6,p+12) satisfies the strong nn-admissible 33-tuple (0,6,12)(0,6,12). Moreover, it is isolated if and only if p6p-6 and p+18p+18 are not nn-totative.

  6. (6)

    nn-Sexy primes quadruplet. Four nn-totative numbers pp, qq, rr, and ss are a nn-sexy quadruplet iff q=p+6q=p+6, r=p+12r=p+12, and s=p+18s=p+18. Notice that any nn-sexy quadruplet satisfies the nn-strong nn-admissible 44-tuple (0,6,12,18)(0,6,12,18). Moreover, it is an isolated 44-tuple ((0,6,12,18,24)(0,6,12,18,24) is a non nn-admissible 55-tuple, we left the proof to the reader).

3. Primorial Tables

Primorial sets can be arranged as tables, see Tables 1, 2, and 3. Notice that i) Multiples of prime numbers p<pnp<p_{n} are located on columns, of the nn-primorial table, columns that are completely marked as red, i.e., a column with elements that are not relative-primes to #(n1)\#(n-1); ii) Prime pnp_{n} will mark just one element of each column (multiple of pnp_{n}); and iii) There are columns maintaining almost exclusively the nn-totative numbers, except made for the one in each column that is a multiple of pnp_{n}. We call these columns nn-totative columns. Table 4 shows just the 44-totative columns.

2
3
Table 1. 11-primorial table (#(1)=2\#(1)=2). Multiples of p1=2p_{1}=2 are shown in blue. 11-totatives are shown in white.
2 3
4 5
6 7
Table 2. 22-primorial table (#(2)=23=6\#(2)=2*3=6). Multiples of p2=3p_{2}=3 are shown in blue, multiples of p1=2p_{1}=2 are shown in red, and multiples of p1=2p_{1}=2 and p2=3p_{2}=3 are shown in yellow. 22-totatives are shown in white.
2 3 4 5 6 7
8 9 10 11 12 13
14 15 16 17 18 19
20 21 22 23 24 25
26 27 28 29 30 31
Table 3. 33-primorial table (#(3)=235=30\#(3)=2*3*5=30). Multiples of p3=5p_{3}=5 are shown in blue, multiples of p1=2p_{1}=2 or p2=3p_{2}=3 are shown in red, and multiples of p1=2p_{1}=2 or p2=3p_{2}=3, and p3=5p_{3}=5 are shown in yellow. 33-totatives are shown in white.
7 11 13 17 19 23 29 31
37 41 43 47 49 53 59 61
67 71 73 77 79 83 89 91
97 101 103 107 109 113 119 121
127 131 133 137 139 143 149 151
157 161 163 167 169 173 179 181
187 191 193 197 199 203 209 211
Table 4. 44-totative columns (#(4)=2357=210\#(4)=2*3*5*7=210). Multiples of p4=7p_{4}=7 are shown in blue.

We define rows and columns in a nn-primorial table as follow: the kkth row of a nn-primorial table is row(n,k)={mm=k#(n1)+jrow(n,k)=\{m\mid m=k*\#(n-1)+j for 2j#(n1)+1}2\leq j\leq\#(n-1)+1\} with 0k<pn0\leq k<p_{n}. The nn-totative column defined by the (n1)(n-1)-totative number tt is col(n,t)={mm=t+k#(n1)col(n,t)=\{m\mid m=t+k*\#(n-1) for 0k<pn}0\leq k<p_{n}\}. For example, the 44-totative column col(4,13)={mm=13+k30col(4,13)=\{m\mid m=13+k*30 for 0k<7}0\leq k<7\}, i.e., col(4,13)={13,43,73,103,133,163,193}col(4,13)=\{13,43,73,103,133,163,193\}, is defined by the 33-totative number 1313.

Lemma 1.

There is one and just one element of col(n,t)col(n,t) that is a multiple of pnp_{n}.

Proof.

We need to prove that each element of col(n,t)col(n,t) produces a different residue module pnp_{n}. Suppose that there are two elements of col(n,t)col(n,t) producing the same residue cc module pnp_{n}, i.e., two elements t1=t+k1#(n1)=apn+ct_{1}=t+k_{1}*\#(n-1)=a*p_{n}+c and t2=t+k2#(n1)=bpn+ct_{2}=t+k_{2}*\#(n-1)=b*p_{n}+c such that 0k1<k2<pn0\leq k_{1}<k_{2}<p_{n} and 0c<pn0\leq c<p_{n}. If we subtract the first number from the second we have that (k2k1)#(n1)=(ba)pn(k_{2}-k_{1})*\#(n-1)=(b-a)*p_{n} then #(n1)\#(n-1) must be a multiple of pnp_{n} since 0<k2k1<pn0<k_{2}-k_{1}<p_{n}, but it is impossible by definition of #(n1)\#(n-1). Now, we have pnp_{n} different elements in col(n,t)col(n,t) so we need pnp_{n} different residues module pnp_{n}, therefore, there is one residue that must be 0, i.e., there is only one element of col(n,t)col(n,t) that is a multiple of pnp_{n}. ∎

Corollary 2.

If n>1n>1 and tt is nn-totative number then there is a (n1)(n-1)-totative number tt^{\prime} such that tt belongs to the nn-totative column defined by tt^{\prime}, i.e., t=t+k#(n1)t=t^{\prime}+k*\#(n-1) for some 0k<pn0\leq k<p_{n}.

Proof.

For any tt in the nn-primorial set, there is a tt^{\prime} in the (n1)(n-1) primorial set and 0k<pn0\leq k<p_{n} such that t=t+k#(n1)t=t^{\prime}+k*\#(n-1) (definition of nn-table rows). Suppose that tt^{\prime} is not a (n1)(n-1)-totative number, then gcd(t,#(n1))>1gcd(t^{\prime},\#(n-1))>1. Therefore, there is a prime number p<pnp<p_{n} such that t=pst^{\prime}=p*s for some s1s\geq 1. Using definition of tt we have that t=ps+p#(n1)p=p(s+#(n1)p)t=p*s+p*\frac{\#(n-1)}{p}=p*\left(s+\frac{\#(n-1)}{p}\right). Since pp also divides #(n)\#(n) then gcd(t,#(n))pgcd(t,\#(n))\geq p but this is impossible since tt is nn-totative, i.e., gcd(#(n),t)=1gcd(\#(n),t)=1. Therefore tt^{\prime} is a #(n1)\#(n-1)-totative number. ∎

4. Counting Primorial nn-Totative Numbers

By using primorial tables, we can count nn-totative numbers and their different classes. Numeric results are obtained with the C++ program (onprimorials.cpp) freely available at the prime numbers github repository of professor Jonatan Gómez [Gom].

Theorem 3.

The number of nn-totatives can be defined recursively as tot(1)=1tot(1)=1 and tot(n)=(pn1)tot(n1)tot(n)=(p_{n}-1)*tot(n-1) for n>1n>1.

Proof.

Clearly tot(1)=1tot(1)=1, see Table 1. Since there is only one multiple of pnp_{n} in each nn-totient column such an element will be marked (it is non nn-totative), then there are (pn1)tot(n1)(p_{n}-1)*tot(n-1) elements that are nn-totative. ∎

Result in Theorem 3 is expected since #(n)\#(n) and (#(n)+1)(\#(n)+1) are coprime numbers, therefore the number of nn-totatives is equal to the totatives of #(n)\#(n), i.e, the number of nn-totatives is φ(#(n))=i=1nφ(pk)=i=1n(pk1)=tot(n)\varphi(\#(n))=\prod_{i=1}^{n}\varphi(p_{k})=\prod_{i=1}^{n}(p_{k}-1)=tot(n). However, we will consider the idea of ’marking’ just one element of the nn-totative columns for counting nn-totative kk-tuples, and kk-constellations. Notice that any prime number in the nn-primorial set must be a nn-totative number or pip_{i} for i=1,2,,ni=1,2,\ldots,n. Table 5 shows this relation between the count of nn-totative numbers (tot(n)tot(n)) and the total number of prime numbers in the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (π(#(n)+1)\pi(\#(n)+1)).

nn #(n)\#(n) tot(n)tot(n) π(#(n)+1)\pi(\#(n)+1) tot(n)π(#(n)+1)\frac{tot(n)}{\pi(\#(n)+1)}
3 30 8 11 0.727273
4 210 48 47 1.02128
5 2310 480 344 1.39535
6 30030 5760 3248 1.7734
7 510510 92160 42331 2.17713
8 9699690 1658880 646029 2.56781
9 223092870 36495360 12283531 2.97108
10 6469693230 1021870080 300369796 3.40204
Table 5. Relation between the count of nn-totative numbers (tot(n)tot(n)) and the total number of prime numbers in the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (π(#(n)+1)\pi(\#(n)+1)).

4.1. Admissible Tuples of Totatives

For determining the number of different nn-totative number classes, we need some technical Theorems about nn-admissible kk-truples.

Theorem 4.

Let a=(0,a2,,ak)a=(0,a_{2},\ldots,a_{k}) be a (n1)(n-1)-admissible kk-tuple such that gap(a)<pn1gap(a)<p_{n}-1 and tt a nn-totative number. If tt satisfies aa then there is a (n1)(n-1)-totative number tt^{\prime} such that t=t+#(n1)jt=t^{\prime}+\#(n-1)*j for some j=0,1,,pn1j=0,1,\ldots,p_{n}-1 and tt^{\prime} satisfies aa.

Proof.

The first nn-totative column is generated by prime pnp_{n} (any natural number greater than 11 and lower than pnp_{n} must be a multiple of some pip_{i} with i=1,2,,n1i=1,2,\ldots,n-1) and the last nn-totative column is generated by #(n1)+1\#(n-1)+1. Therefore, the gap between the element in the last nn-totative column of row rr and the first element in the nn-totative column of row r+1r+1 is (#(n1)r+pn)(#(n1)r+1)=#(n1)1=pn1(\#(n-1)*r+p_{n})-(\#(n-1)*r+1)=\#(n-1)-1=p_{n}-1. Since gap(a)<pn1gap(a)<p_{n}-1 it is impossible that the element in the last nn-totative column of row rr and the element in the first column of row r+1r+1 satisfy the kk-tuple aa. Therefore, it is impossible that a nn-totative number tt defines a nn-admissible kk-tuple (with gap(a)<pn1gap(a)<p_{n}-1) distributed in two or more different rows, i.e, t+ai#(n1)(j+1)+1t+a_{i}\leq\#(n-1)*(j+1)+1 for all i=1,2,,ki=1,2,\ldots,k. According to Corollary 2, t=t+#(n1)jt=t^{\prime}+\#(n-1)j for some j=0,1,,pn1j=0,1,\ldots,p_{n}-1, then t+ai=t+ai+#(n1)j#(n1)(j+1)+1t+a_{i}=t^{\prime}+a_{i}+\#(n-1)*j\leq\#(n-1)*(j+1)+1, so t+ai#(n1)+1t^{\prime}+a_{i}\leq\#(n-1)+1, i.e., t+ait^{\prime}+a_{i} is (n1)(n-1)-totative for all i=1,2,,ki=1,2,\ldots,k. ∎

Theorem 5.

Let a=(0,a2,,ak)a=(0,a_{2},\ldots,a_{k}) be a strong (n1)(n-1)-admissible kk-tuple such that gap(a)<pn1gap(a)<p_{n}-1. For any isolated (n1)(n-1)-totative kk-tuple satisfying aa there are pnkp_{n}-k isolated nn-totative kk-tuples satisfying aa.

Proof.

If tt is a n1n-1-totative number such that tt satisfies aa and a[t]a[t] is an isolated (n1)(n-1)-totative kk-tuple then there are pnp_{n} candidate isolated nn-totative kk-tuples (one per each row of the nn-primorial table on the same columns of the isolated (n1)(n-1)-totative kk-tuple. Since pnp_{n} marks only one element in a single nn-totative column defined by aa (Lemma 1) and pnp_{n} cannot mark two different columns t+ait+a_{i} and t+ajt+a_{j} iji\neq j (aa is strong), pnp_{n} marks kk different candidate nn-admissible kk-tuples generated by a[t]a[t]. Therefore, counta(m)=counta(m1)(pmk)count_{a}(m)=count_{a}(m-1)*(p_{m}-k). ∎

4.2. Twin Totative Numbers and Twin Primes

We can now count twin nn-totatives and compare them against twin primes.

Corollary 6.

The number of nn-twin couples (twin(n)twin(n)) can be defined recursively as twin(1)=0twin(1)=0, twin(2)=1twin(2)=1 and twin(n)=(pn2)twin(n1)twin(n)=(p_{n}-2)*twin(n-1) for n>2n>2.

Proof.

Clearly twin(1)=0twin(1)=0 and twin(2)=1twin(2)=1, see Tables 1 and 2 respectively. Notice that gap((0,2))=2gap\left((0,2)\right)=2 and p3=5p_{3}=5, so gap((0,2))<p31gap\left((0,2)\right)<p_{3}-1. Since (0,2)(0,2) is a strong 33-admissible 22-tuple and every nn-twin couple is isolated for n>2n>2, then by Theorem 5 we have that twin(n)=(pn2)twin(n1)twin(n)=(p_{n}-2)*twin(n-1) for all n>2n>2. ∎

Table 6 shows the relationship between the count of nn-totative twin couples (twin(n)twin(n)) and the total number of twin prime couples in the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (twin(#(n)+1)twin_{*}(\#(n)+1)).

nn #(n)\#(n) twin(n)twin(n) twin(#(n)+1)twin_{*}(\#(n)+1) twin(n)twin(#(n)+1)\frac{twin(n)}{twin_{*}(\#(n)+1)}
3 30 3 5 0.6
4 210 15 15 1
5 2310 135 70 1.92857
6 30030 1485 468 3.17308
7 510510 22275 4636 4.80479
8 9699690 378675 57453 6.59104
9 223092870 7952175 896062 8.87458
10 6469693230 214708725 18463713 11.6287
Table 6. Relation between the count of nn-totative twin couples (twin(n)twin(n)) and the total number of twin prime couples the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (twin(#(n)+1)twin_{*}(\#(n)+1)).

4.3. Cousin Totative Numbers and Cousin Primes

We can now count cousin nn-totatives and compare them against cousin primes.

Corollary 7.

The number of nn-cousin couples (cousin(n)cousin(n)) can be defined recursively as cousin(1)=cousin(2)=0cousin(1)=cousin(2)=0, cousin(3)=3cousin(3)=3, and cousin(n)=(pn2)cousin(n1)cousin(n)=(p_{n}-2)*cousin(n-1) for n>3n>3.

Proof.

Clearly cousin(1)=0cousin(1)=0, cousin(2)=0cousin(2)=0 and cousin(3)=3cousin(3)=3, see Tables 1 to 3 respectively. Notice that gap((0,4))=4gap\left((0,4)\right)=4 and p4=7p_{4}=7, so gap((0,4))<p41gap\left((0,4)\right)<p_{4}-1. Since (0,4)(0,4) is a strong 44-admissible 22-tuple and every nn-cousin couple is isolated for n>3n>3, then by Theorem 5 we have that cousin(n)=(pn2)cousin(n1)cousin(n)=(p_{n}-2)*cousin(n-1) for all n>3n>3. ∎

Corollary 8.

twin(n)=cousin(n)twin(n)=cousin(n) for all n3n\geq 3.

Table 7 shows the relationship between the count of nn-totative cousin couples (cousin(n)cousin(n)) and the total number of cousin prime couples in the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (cousin(#(n)+1)cousin_{*}(\#(n)+1)).

nn #(n)\#(n) cousin(n)cousin(n) cousin(#(n)+1)cousin_{*}(\#(n)+1) cousin(n)cousin(#(n)+1)\frac{cousin(n)}{cousin_{*}(\#(n)+1)}
3 30 3 4 0.75
4 210 15 14 1.07143
5 2310 135 71 1.90141
6 30030 1485 468 3.17308
7 510510 22275 4630 4.81102
8 9699690 378675 57065 6.63585
9 223092870 7952175 896737 8.8679
10 6469693230 214708725 18460319 11.6308
Table 7. Relation between the count of nn-totative cousin couples (cousin(n)cousin(n)) and the total number of cousin prime couples the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (cousin(#(n)+1)cousin_{*}(\#(n)+1)).

4.4. Sexy Totative Numbers and Sexy Primes

We can now count sexy nn-totatives and compare them against sexy primes.

Corollary 9.

The number of nn-sexy quadruplets (quad(n)quad(n)) can be defined recursively as quad(1)=quad(2)=0quad(1)=quad(2)=0, quad(3)=1quad(3)=1, quad(4)=6quad(4)=6, and quad(n)=(pn4)quad(n1)quad(n)=(p_{n}-4)*quad(n-1) for n>4n>4.

Proof.

Clearly quad(1)=quad(2)=0quad(1)=quad(2)=0, quad(3)=1quad(3)=1 and quad(4)=6quad(4)=6, see Tables 1 to 4 respectively. Notice that gap((0,6,12,18))=6gap\left((0,6,12,18)\right)=6 and p5=11p_{5}=11, so gap((0,6,12,18))<p51gap\left((0,6,12,18)\right)<p_{5}-1. Since (0,6,12,18)(0,6,12,18) is a strong 55-admissible 44-tuple and every nn-quadruple is isolated for n>4n>4, then by Theorem 5 we have that quad(n)=(pn4)quad(n1)quad(n)=(p_{n}-4)*quad(n-1) for all n>4n>4. ∎

Lemma 10.

The number of isolated nn-sexy triplets (itriple(n)itriple(n)), nn-sexy triplets (triple(n)triple(n)), isolated nn-sexy couples (isexy(n)isexy(n)), and nn-sexy couples (sexy(n)sexy(n)) can be defined as follow:

  1. (1)

    itriple(1)=itriple(2)=0itriple(1)=itriple(2)=0, itriple(3)=1itriple(3)=1, itriple(4)=4itriple(4)=4, and itriple(n)=(pn3)itriple(n1)+2quad(n1)itriple(n)=(p_{n}-3)*itriple(n-1)+2*quad(n-1) for n>4n>4.

  2. (2)

    triple(n)=itriple(n)+2quad(n)triple(n)=itriple(n)+2*quad(n) for all n>0n>0.

  3. (3)

    isexy(1)=isexy(2)=isexy(3)=0isexy(1)=isexy(2)=isexy(3)=0, isexy(4)=4isexy(4)=4, and isexy(n)=(pn2)isexy(n1)+2itriple(n1)+2quad(n1)isexy(n)=(p_{n}-2)*isexy(n-1)+2*itriple(n-1)+2*quad(n-1) for n>4n>4.

  4. (4)

    sexy(n)=isexy(n)+2itriple(n)+3quad(n)sexy(n)=isexy(n)+2*itriple(n)+3*quad(n).

Proof.

We left proofs to the reader (just consider which columns are marked by prime number pnp_{n} in each case). ∎

Corollary 11.

The number of nn-sexy couples (sexy(n)sexy(n)) can be defined recursively as sexy(1)=sexy(2)=0sexy(1)=sexy(2)=0, sexy(3)=5sexy(3)=5, sexy(4)=30sexy(4)=30, and sexy(n)=(pn2)sexy(n1)sexy(n)=(p_{n}-2)*sexy(n-1) for all n>4n>4.

Proof.

Clearly, sexy(1)=sexy(2)=0sexy(1)=sexy(2)=0, sexy(3)=5sexy(3)=5, and sexy(4)=30sexy(4)=30 by computing sexy(n)sexy(n) with definition 4) of Lemma 10. Now, by item 4) of Lemma 10 we have (pn2)sexy(n1)=(pn2)(isexy(n1)+2itriple(n1)+3quad(n1))(p_{n}-2)*sexy(n-1)=(p_{n}-2)*\left(isexy(n-1)+2*itriple(n-1)+3*quad(n-1)\right). By distributing the product on the right side, then expressing 2(pn2)2(p_{n}-2) as 2(pn3)+22(p_{n}-3)+2 and 3(pn2)3(p_{n}-2) as 3(pn4)+2+43(p_{n}-4)+2+4 and organizing terms we have (pn2)sexy(n1)=(pn2)isexy(n1)+2itriple(n1)+2quad(n1)+2(pn3)itriple(n1)+4quad(n1)+3(pn4)quad(n1)(p_{n}-2)*sexy(n-1)=(p_{n}-2)*isexy(n-1)+2*itriple(n-1)+2*quad(n-1)+2*(p_{n}-3)*itriple(n-1)+4*quad(n-1)+3*(p_{n}-4)*quad(n-1). Now, by grouping terms 131-3, 454-5, and using Lemma 10 we have (pn2)sexy(n1)=isexy(n)+2itriple(n)+3quad(n)=sexy(n)(p_{n}-2)*sexy(n-1)=isexy(n)+2*itriple(n)+3*quad(n)=sexy(n). ∎

Corollary 12.

sexy(n)=2twin(n)sexy(n)=2*twin(n) for all n>=4n>=4.

Table 8 shows the relationship between the count of nn-totative sexy couples (sexy(n)sexy(n)) and the total number of sexy prime couples in the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (sexy(#(n)+1)sexy_{*}(\#(n)+1)). Notice that the behavior of the ratio between the number of twin primes and the number of nn-twins is the same as the behavior of the ratio between the number of cousin primes and the number of nn-cousins, and the behavior of the ratio between the number of sexy primes and the number of nn-sexy numbers.

nn #(n)\#(n) sexy(n)sexy(n) sexy(#(n)+1)sexy_{*}(\#(n)+1) sexy(n)sexy(#(n)+1)\frac{sexy(n)}{sexy_{*}(\#(n)+1)}
3 30 5 6 0.833333
4 210 30 26 1.15385
5 2310 270 140 1.92857
6 30030 2970 951 3.12303
7 510510 44550 9331 4.77441
8 9699690 757350 114189 6.63243
9 223092870 15904350 1792173 8.87434
10 6469693230 429417450 36921295 11.6306
Table 8. Relation between the count of nn-totative sexy couples (sexy(n)sexy(n)) and the total number of sexy prime couples in the nn-primorial set, i.e., up to #(n)+1\#(n)+1 (sexy(#(n)+1)sexy_{*}(\#(n)+1)).

5. Goldbach Conjecture

There are a lot of different conjectures about prime numbers [Wika]. In particular, Goldbach’s conjecture stays that any even natural number greater than six (66) can be expressed as the sum of two prime numbers [Ras17, Est38]. In this section, we derive a version of Goldbach’s conjecture but for primorial intervals and combine Goldbach’s conjecture with twin, cousin, and sexy prime numbers. From now on, 𝒫\mathcal{P} denotes the set of prime numbers, \mathcal{E} denotes the set of even natural numbers, 𝒫n\mathcal{P}_{n} denotes the set of primes in the nn-primorial interval, i.e., 𝒫n=𝒫n\mathcal{P}_{n}=\mathcal{P}\bigcap\mathcal{I}_{n}, and n\mathcal{E}_{n} denotes the set of even number in the nn-primorial interval, i.e., n=n\mathcal{E}_{n}=\mathcal{E}\bigcap\mathcal{I}_{n}.

Conjecture 13.

(Goldbach) For any positive even number natural m4m\geq 4 there exist at least two prime numbers p,qPp,q\in P such that m=p+qm=p+q.

Conjecture 14.

(Goldbach-Intervals) For any positive natural number nn and any mn+1m\in\mathcal{E}_{n+1} there are at least two prime numbers p𝒫np\in\mathcal{P}_{n} and q𝒫n𝒫n+1q\in\mathcal{P}_{n}\bigcup\mathcal{P}_{n+1} such that m=p+qm=p+q.

The pair of prime numbers satisfying the Goldbach-Intervals conjecture for the first two primorial intervals are shown in Table 9. Goldbach-Intervals conjecture for the first 1010-primorial intervals can be validated with the C++ program (onprimorials.cpp) freely available at the prime numbers github repository of professor Jonatan Gómez [Gom].

nn 𝒫n\mathcal{P}_{n} 𝒫n𝒫n+1\mathcal{P}_{n}\bigcup\mathcal{P}_{n+1} mEn+1m\in E_{n+1} pp qq
11 {2,3}\left\{2,3\right\} {2,3,5,7}\left\{2,3,5,7\right\} 44 22 22
66 33 33
22 {3,5,7}\left\{3,5,7\right\} {3,5,7,11,13,17,19,23,29,31}\left\{3,5,7,11,13,17,19,23,29,31\right\} 88 33 55
1010 33 77
1212 55 77
1414 33 1111
1616 33 1313
1818 55 1313
2020 33 1717
2222 33 1919
2424 55 1919
2626 33 2323
2828 55 2323
3030 77 2323
Table 9. Validation of the Goldbach-Intervals conjecture for the first two primorial intervals.
Theorem 15.

If the Goldbach-Intervals conjecture 18 is true then the Goldbach conjecture 13 is true.

Conjecture 16.

(Goldbach-Twin) For any positive even natural number nn there are at least two prime numbers p𝒫p\in\mathcal{P} and q𝒫q\in\mathcal{P} such that m=p+qm=p+q and pp or qq is a twin prime.

Conjecture 17.

(Goldbach-Cousin) For any positive even natural number nn there are at least two prime numbers p𝒫p\in\mathcal{P} and q𝒫q\in\mathcal{P} such that m=p+qm=p+q and pp or qq is a cousin prime.

Conjecture 18.

(Goldbach-Sexy) For any positive even natural number nn there are at least two prime numbers p𝒫p\in\mathcal{P} and q𝒫q\in\mathcal{P} such that m=p+qm=p+q and pp or qq is a sexy prime.

Goldbach-Twin, Goldbach-Cousin, and Goldbach-Sexy conjectures for the first 1010-primorial sets can be validated with the C++ program (onprimorials.cpp) freely available at the prime numbers github repository of professor Jonatan Gómez [Gom].

6. Conclusions and Future Work

We have shown that primorial numbers may be useful for understanding some properties of prime numbers and prime numbers classes. For instance, we defined the concept of nn-primorial set and established a relation between prime numbers in the interval [pn,#(n)+1][p_{n},\#(n)+1] and nn-totative numbers. We used primorial intervals and primorial tables to define functions that count nn-admissible kk-tuples and to establish a relationship between them and their prime admissible kk-tuples counterparts. In particular, we showed that the behavior of the ratio between the number of twin primes and the number of nn-twins is the same as the behavior of the ratio between the number of cousin primes and the number of nn-cousins, and the behavior of the ratio between the number of sexy primes and the number of nn-sexy numbers. Finally, we stated variation on Goldbach’s conjecture one in terms of primorial intervals and three in terms of twin, cousin, and sexy prime numbers. We computationally validate such conjectures for even numbers up to the 10th primorial number,i.e., up to #(10)=6469693230\#(10)=6469693230, see C++ program onprimorial.cpp at professor Jonatan Gomez github repository [Gom]. Our future work will concentrate on expressing the Prime Number Theorem [Nar00] in terms of both primorial numbers and nn-totative numbers. Also, we will study the relationship between the asymptotic behavior of nn-totative admissible kk-tuples as nn-twin, nn-cousin, and nn-sexy couples and their corresponding twin, cousin, and sexy prime counterparts. Finally, we will try to find proof of the stated conjectures.

References