Proofs Regarding Primorial Patterns

 

by

 

Dennis R. Martin

DP Technology Corp., Camarillo, CA

dennis.martin@dptechnology.com

BSME, Michigan Technological University, Houghton, MI

Department of Mathematics, University of California, Santa Barbara

drmartin@umail.ucsb.edu

 

Copyright © 2006 by Dennis R. Martin

 

ALL RIGHTS RESERVED

 

No part of this document may be reproduced, retransmitted, or redistributed by any means, without the expressed written consent of Dennis R. Martin.

 

This document is available in other formats. See www.primenace.com.

 

 Abstract

 

Observations are made regarding the pattern of the composite numbers that have a particular prime factor for their lowest prime factor. It is subsequently proven that this pattern repeats over intervals equal to the primorial of that lowest prime factor such that the number and distribution of such composites is constant. The value of that constant composite to primorial ratio is proven to be related to the previous prime numbers and its constant composite to primorial ratio.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 Introduction

 

It seems well known that the pattern of composite numbers repeats with a period equal to the primorial of each prime factor. Dickson [1] refers to remarks by H.J.S Smith in 1857 and papers by J. DeChamps published in 1907 regarding this property, and Weisstein [2] makes note of it. However, actual proofs of these properties do not seem to be readily available. This paper attempts to rectify that situation by starting from the ground up and deriving some general relationships and developing proofs based on them.

 

2 Singular Identities

 

A prime number is a number greater than 1 that has no positive integer divisors other than 1 and itself.[3] By the fundamental theorem of arithmetic, every positive integer greater than 1 can be uniquely represented by its prime divisors in what is called a prime factorization.[4, 5] Since a prime pN has no other positive integer divisors besides 1 and itself, the prime factorization of pN is simply pN. A positive integer greater than 1 which is not prime is a composite number.[6]

 

By convention the number 1 is considered neither prime nor composite. Every positive integer greater than 1 is either a prime number or a composite number. For every positive integer greater than 1, then, one of the following statements must be true:

            1. The prime pN is a factor in its prime factorization.

            2. The prime pN is not a factor in its prime factorization.

 

If a number n has a prime factorization where pN is a single factor, that is pNi is a factor and i = 1, and if pN1 is the only factor other than 1, then n = pN and it is prime. Otherwise it is a composite number which we can label C.

 

Lemma 1: For every composite number C having pN as a factor, one of the following statements must be true:

            1. The lowest prime factor of C is smaller than pN. That is, pN-J is the lowest prime factor and pN-J < pN.

            2. The lowest prime factor is pN and it is the lone prime factor, i.e. pNi = C where i ≥  2.

            3. pN is the lowest prime factor while the other factor is a larger prime pN+J > pN or a product of one or more larger primes such as (pN+J)k or (pN+J) • (pN+L) or combinations of their higher powers.

 

Proof: If pN is a factor by itself, then it is of the form pNi, but i cannot be 1 because then the number would be prime, so i must be greater than 1 and thus satisfy condition 2. If pN is not a factor by itself, then it must be combined with at least one other prime. If any of the other primes in the prime factorization is less than pN then condition 1 is satisfied, otherwise all of them must be greater than pN, in which is the case covered by condition 3.

                                                                                                                                    Q.E.D.

 

 


3 Primorial Soup

 

The primorial is analogous to a factorial applied to the sequence of prime numbers.[7] The primorial for the prime pN is the product of all primes up to and including pN, and it is denoted as  pN#. By the definition of factorial, p1# = 2, and then for every prime greater than 2:

 

            pN# = pN • pN–1#                                                                                               (1)

 

What happens, though, if instead of just multiplying pN–1# by pN to get the next primorial value, we multiply all of the positive integers in the interval up to and including pN–1# by pN? What can we say about these numbers, and what about the other integers that we might need to fill in the new interval up to and including pN#? What would happen if we were to add some multiple of pN# onto all of them? What can we say about the lowest prime factors of those numbers?

 

For example, start with the number 1, and multiply it by the first prime p1 = 2. We now have an interval the width of the primorial p1# = 2 containing the numbers 1 and 2. Add any multiple of 2 onto these two numbers. That is, let n be a non-negative integer where n = 0, 1, 2, 3, 4, 5… The result of the addition is an interval of width p1# = 2 containing the numbers 2n + 1 and 2n + 2. For those numbers, the statements that follow, where º indicates congruence, are always true.[8]

 

            (2n + 1) º 1 (mod 2)

            (2n + 2) º 0 (mod 2)

 

Therefore within every interval of width p1# = 2 we have one number that has p1 = 2 for its lowest prime factor and another number that does not have 2 for a factor. Except for the case where n = 0 and that other number is 1, that other number must either be prime or it must be a composite that has a higher prime pN+J > p1 for its lowest prime factor. Essentially we have just found that every even number is evenly divisible by 2 and that every odd number is not.

 

Now take that first primorial and multiply it by the second prime, p2 = 3 and fill in the spaces in between. The result is an interval of width p2# = 6 containing the numbers 1, 2, 3, 4, 5, 6. Add any multiple of 6 on to those numbers. The result will be an interval the width of the primorial p2# = 6, still, containing the following numbers:

 

            (6n + 1), (6n + 2), (6n + 3), (6n + 4), (6n + 5), (6n + 6)

 

Let us factor these values. While we could factor a 3 from out of two of those numbers, though, let us instead only factor the lowest prime factor possible out of each. This factoring produces:

 

            (6n + 1) º 1 (mod 6)

            2 • (3n + 1) º 2 (mod 6) º 0 (mod 2)

            3 • (2n + 1) º 3 (mod 6) º 0 (mod 3)

            2 • (3n + 2) º 4 (mod 6) º 0 (mod 2)

            (6n + 5) º 5 (mod 6)

            2 • (3n + 3) º 0 (mod 6) º 0 (mod 2)

 

Performing a lowest prime factorization like this allows us to more easily count the numbers within the primorial interval in terms of their lowest prime factor.

 

Out of every p2# primorial interval, we can see that there are three numbers that have 2 for their lowest prime factor, one number that has 3 for its lowest prime factor (the 3 • (2n + 1) term), and two other numbers which do not have 2 or 3 as a factor at all. That there are three numbers that have 2 for their lowest prime factor makes sense because we multiplied the previous primorial by 3 and that primorial interval had one number that had 2 for its lowest prime factor. But notice also that the term that does have 3 for its lowest prime factor has the same form as the term that did not have 2 as a factor in that previous p1# primorial interval. That is:

 

            (2n + 1) º 1 (mod 2)    ®        3 • (2n + 1) º 3 (mod 6) º 0 (mod 3)

 

To summarize what can be concluded so far:

a.       1 out of every 2 and 3 out of every 6 numbers have 2 for their lowest prime factor.

b.      1 out of every interval of 6 numbers has 3 for its lowest prime factor.

c.       4 numbers total out of every 6 have either 2 or 3 for their lowest prime factor.

d.      2 out of every 6 numbers do not have 2 or 3 as a factor. Ignoring the trivial case where one of those is 1, those two numbers individually are either prime or have a prime that is higher than 3 for their lowest prime factor.

 

The count in (d) can be calculated as the primorial value minus the count from (a) that have 2 for their lowest prime factor minus the count from (b) that has 3 for its lowest prime factor. But the count from (a) corresponds back to the count in the p1# = 2 interval that had 2 for a lowest prime factor and the count from (b) matches the count from the previous p1# = 2 interval which did not have 2 as a factor. Thus the count in (d) is directly related to counts in the previous primorial.

 

4 A Preliminary Proof by Induction

 

By induction then, if we multiply the first interval of 6 by p3 = 5 and then add any multiple of p3# = 5 • 6 = 30, we should expect the three numbers that had 2 as their lowest prime factor in the previous p2# = 6 primorial to lead to 3 • 5 = 15 = 30 / 2 which have 2 as their lowest prime factor in each p3# = 30 primorial interval. We should also expect the one that had 3 as its lowest prime factor in the previous to lead to 1 • 5 = 5 = 30 / 6 that have 3 as their lowest prime factor in this interval, and the 2 that had neither 2 nor 3 as a factor should now relate to two composites that have 5 for their lowest prime factor. That will leave 30 – 15 – 5 – 2 = 8 that have do not have 2, 3 or 5 as a factor. If not equal to 1, then each of those eight must either be prime themselves or must have a prime higher than 5 as their lowest prime factor.

 

Let us set about proving what has been implied by induction. Take the first 30 positive integers and add any multiple of p3# = 30 onto them. The result is a set having the values {(30n + 1), (30n + 2), (30n + 3), (30n + 4), (30n + 5), (30n + 6), …(30n + 29), (30n + 30)}. Now factor the lowest prime factor possible out of each of them. The results of this lowest prime factorization including various relevant residues are shown in Table 1.

 

 

Table 1: Lowest Prime Factorization of Any Multiple of the p3# = 5# = 30 Primorial Interval

 

Primorial

Interval

Member

Lowest Prime

Factorization

Residues

Modulo p3#

Modulo p2#

Modulo

Lowest pN

30n + 1

(30n + 1)

1 (mod 30)

1 (mod 6)

 

30n + 2

2 • (15n + 1)

2 (mod 30)

2 (mod 6)

0 (mod 2)

30n + 3

3 • (10n + 1)

3 (mod 30)

3 (mod 6)

0 (mod 3)

30n + 4

2 • (15n + 2)

4 (mod 30)

4 (mod 6)

0 (mod 2)

30n + 5

5 • (6n + 1)

5 (mod 30)

5 (mod 6)

0 (mod 5)

30n + 6

2 • (15n + 3)

6 (mod 30)

0 (mod 6)

0 (mod 2)

30n + 7

(30n + 7)

7 (mod 30)

1 (mod 6)

 

30n + 8

2 • (15n + 4)

8 (mod 30)

2 (mod 6)

0 (mod 2)

30n + 9

3 • (10n + 3)

9 (mod 30)

3 (mod 6)

0 (mod 3)

30n + 10

2 • (15n + 5)

10 (mod 30)

4 (mod 6)

0 (mod 2)

30n + 11

(30n + 11)

11 (mod 30)

5 (mod 6)

 

30n + 12

2 • (15n + 6)

12 (mod 30)

0 (mod 6)

0 (mod 2)

30n + 13

(30n + 13)

13 (mod 30)

1 (mod 6)

 

30n + 14

2 • (15n + 7)

14 (mod 30)

2 (mod 6)

0 (mod 2)

30n + 15

3 • (10n + 5)

15 (mod 30)

3 (mod 6)

0 (mod 3)

30n + 16

2 • (15n + 8)

16 (mod 30)

4 (mod 6)

0 (mod 2)

30n + 17

(30n + 17)

17 (mod 30)

5 (mod 6)

 

30n + 18

2 • (15n + 9)

18 (mod 30)

0 (mod 6)

0 (mod 2)

30n + 19

(30n + 19)

19 (mod 30)

1 (mod 6)

 

30n + 20

2 • (15n + 10)

20 (mod 30)

2 (mod 6)

0 (mod 2)

30n + 21

3 • (10n + 7)

21 (mod 30)

3 (mod 6)

0 (mod 3)

30n + 22

2 • (15n + 11)

22 (mod 30)

4 (mod 6)

0 (mod 2)

30n + 23

(30n + 23)

23 (mod 30)

5 (mod 6)

 

30n + 24

2 • (15n + 12)

24 (mod 30)

0 (mod 6)

0 (mod 2)

30n + 25

5 • (6n + 5)

25 (mod 30)

1 (mod 6)

0 (mod 5)

30n + 26

2 • (15n + 13)

26 (mod 30)

2 (mod 6)

0 (mod 2)

30n + 27

3 • (10n + 9)

27 (mod 30)

3 (mod 6)

0 (mod 3)

30n + 28

2 • (15n + 14)

28 (mod 30)

4 (mod 6)

0 (mod 2)

30n + 29

(30n + 29)

29 (mod 30)

5 (mod 6)

 

30n + 30

2 • (15n + 15)

0 (mod 30)

0 (mod 6)

0 (mod 2)

 

As expected there are 15 numbers within each interval of 30 that have 2 as their lowest prime factor. All of them relate to a value within a primorial interval of the previous prime. That is, all of them are congruent to either 2 or 4 or 0 (mod 6).

 

There are 5 numbers within each interval of 30 that have 3 as their lowest prime factor. Those 5 relate to a specific value within a primorial interval of p2# = 6 in that all of them have a residue of 3 (mod 6).

 

Then there are 2 numbers (as highlighted in bold in Table 1) within each interval of 30 that have 5 as their lowest prime factor. Those two are directly related to the two factors that did not have 2 or 3 as a factor within the primorial of the previous prime. They are the values with residues of 1 (mod 6) and 5 (mod 6):

 

            (6n + 1) º 1 (mod 6)    ®        5 • (6n + 1) º 1 (mod 6) º 0 (mod 5)

            (6n + 5) º 5 (mod 6)    ®        5 • (6n + 5) º 5 (mod 6) º 0 (mod 5)

 

Finally that leaves 8 numbers within each interval of 30 that do not have 2, 3, or 5 as a factor. The results for the primorial p3# = 30 match what we predicted by induction from p2#. Let us now generalize this as a theorem and prove it for any primorial.

 

5 Deriving the General Theorems

 

Theorem 1: Over any interval equal to the primorial pN# of a particular prime pN, the count of the numbers having pN as their lowest prime factor is constant, as is the count of the numbers having any prime pN-J  that is less than pN as their lowest prime factor a constant as well.

 

Proof: Let the sequence of numbers a0, a1, a2, … am-2, am-1 represent all of the integers within any interval of the primorial pN# such that ai Î {a} and (n • pN# + A) < ai ((n + 1) • pN# + A), where n is a non-negative integer and A is an integer offset. Since we are only concerned with prime factors, and 1 has no prime factors, let us also stipulate that 1 ≤  A pN# so that all ai > 1.

 

Divide every integer in the interval by the primorial value. Since every integer is unique, each one will have a unique residue with respect to the primorial. That is, each will produce an integer residue going from 0 up to the primorial value minus 1 with the primorial as the modulus.

 

            pN# | {a}          ®        [ º {0, 1, 2, 3 … (pN# – 2), (pN# –1)} (mod pN#) ]

 

The entity on the right is a residue system. Because each integer value from zero up to the modulus minus 1 is represented as a residue, this is a complete residue system.[9]

 

Label this residue system {r}N. Since the primorial is a factorial product of the previous lower primes, let us divide those residues by the primorial value of the next lower prime. Since each residue is unique within the range from 0 up to the original primorial, each new residue with respect to the primorial of the previous prime will be unique within an interval of that primorial, and the number of subintervals of that primorial will be equal to the value of the original prime pN that we started with. Thus each lower residue system with the lower primorial value as the modulus will be repeated a number of times equal to the value of that original, next higher prime.

 

            pN-1# | {a}        ®        [ º {0, 1, 2, 3 … (pN-1# – 2), (pN-1# –1)} (mod pN-1#) ] x pN

 

Label the new residue system in brackets as {r}N-1. There are pN of these {r}N-1 residue systems within each pN# primorial, and each of them is a complete residue system. We can continue by dividing all of these residue systems by pN-2# to produce the {r}N-2 system in brackets below:

 

            pN-2# | {a}        ®        [ º {0, 1, 2, 3 … (pN-2# – 2), (pN-2# –1)} (mod pN-2#) ] x (pN •  pN-1)

 

There are (pN •  pN-1) of these {r}N-2 residue systems within the original pN# primorial. This process can continue all the way down to p1# = 2, where the complete residue system labeled {r}1 is [ º {0, 1} (mod 2) ]. There would be (pN • pN-1 • pN-2 •... • p2) = (pN# / 2) of these {r}1 residue systems within an interval of pN#.

 

We can use {r}NL to represent the count of complete residue systems {r}L that have for their modulus the primorial pL# of a lower prime pL < pN within an interval of the primorial pN#.

 

            {r}NL = pN# / pL#                                                                                             (2)

 

There is always just one complete residue system {r}N within an interval of pN#, thus {r}NN = 1, and there would be (pN# / p1#) = (pN# / 2) = {r}N1 instances of the {r}1 residue system within pN#. All of the values from the original pN# primorial interval that are congruent to 0 (mod 2) have p1 = 2 as their lowest prime factor. There would be (pN# / 2) = {r}N1 such values. Only those values that are congruent to 1 (mod 2) can have a higher prime number pH > p1 as their lowest prime factor.

 

Let rN represent the ratio of numbers within the primorial for pN that have pN as their lowest prime factor. There is only one value that is congruent to 0 (mod 2) in each interval of p1# = 2, therefore r1 = 1. That means that there is always one number in each interval of 2 that is not a multiple of 2. That is the value that is congruent to 1 (mod 2) and its count can be calculated as p1# – r1 = 2 – 1 = 1. Theorem 1 definitely applies to the first prime and its primorial.

 

For an interval of p2# = 6, the residue system {r}2 contains {0, 1, 2, 3, 4, 5} (mod 6). There must be r1 • {r}21 = r1 • (p2# / p1#) = 3 numbers within that interval that have 2 as their lowest prime factor. Those three are congruent to {0, 2, 4} (mod 6). Another way to think of this is that, since initially multiplying the primorial of 2 by p2 = 3 generates the primorial of 6, there must be three multiples of 2 within that 6, because there is always one multiple of 2 within each interval of 2.

 

When the value congruent to 0 (mod 2) in an interval of 2 is multiplied by 3 in generating an interval of 6, the result is congruent to 0 (mod 6) and since 6 is divisible by 2, it is still congruent to 0 (mod 2). An even number times any positive integer is always an even number. So while there are two multiples of 3 in each interval of 6, one of them must have 2 as its lowest prime factor. The other one corresponds back to a value that was congruent to 1 (mod 2) in the interval of 2, the count of which was p1# – r1. When it is multiplied by 3 the result is congruent to 3 (mod 6) and hence to 0 (mod 3). That there is one such value means r2 = 1 = p1# – r1.

 

The count of what is left, then, is p2# – r1 • (p2# / p1#) – r2 = 6 – 3 – 1 = 2. There are two numbers in each multiple of 6 that do not have 2 or 3 as their lowest prime factor. They are the numbers congruent to 1 or 5 (mod 6). Those numbers are each either a composite that has a prime greater than p2 as their lowest prime factor or they themselves are prime. When an interval of 6 is multiplied by p3 = 5 to initially generate an interval of p3# = 30, it is these two values and only these two values that can produce composites that have 5 as their lowest prime factor.

 

The other four numbers in the interval of 6 will, when multiplied by 5, result in composites that still have 2 or 3 as a prime factor. This implies that r3 = 2 = p2# – r1 • (p2# / p1#) – r2.

 

Since p2# = p2 •  p1#, we can factor p2 from the first two terms on the right side of that expression:

 

            r3 = p2 • (p1# – r1 • (p1# / p1#)) – r2 = p2 • (p1# – r1) – r2

 

But r2 = (p1# – r1), so substituting yields:

 

            r3 = p2 • r2 – r2 = r2 • (p2 – 1) = 1 • (3 – 1) = 2

 

An interval of p3# = 30 would contain {r}31 = (p3# / p1#) = 15 instances of the {r}1 residue system, {r}32 = (p3# / p2#) = 5 instances of the {r}2 residue system, and of course {r}33 = (p3# / p3#) = 1 instance of the {r}3 residue system. Thus there would be r1 • {r}31 = 15 numbers that have 2 as their lowest prime factor, r2 • {r}32 = 5 numbers that have 3 as their lowest prime factor, and r3 • {r}33 = 2 numbers that have 5 as their lowest prime factor. The numbers that are left do not have 2, 3, or 5 as a factor. When this interval of p3# = 30 is multiplied by p4 = 7 to generate an interval of p4# = 210, it is these numbers that will have p4 = 7 as their lowest prime factor:

 

            r4 = p3# – r1 • {r}31 – r2 • {r}32 – r3 • {r}33 = p3# – S (rL • {r}3L)                     {for L = 1 to 3

 

In this case, r4 = 30 – 15 – 5 – 2 = 8. The factorization of those 8 numbers is shown in Table 2.

 

Table 2: Factorization of Any p4# = 210 Primorial Interval for the Lowest Prime Factor p4 = 7

 

Primorial

Interval Member

Lowest Prime

Factorization

Residues

Modulo p4#

Modulo p3#

Modulo p2#

210n + 7

7 • (30n + 1)

7 (mod 210)

1 (mod 30)

1 (mod 6)

210n + 49

7 • (30n + 7)

49 (mod 210)

7 (mod 30)

1 (mod 6)

210n + 77

7 • (30n + 11)

77 (mod 210)

11 (mod 30)

5 (mod 6)

210n + 91

7 • (30n + 13)

91 (mod 210)

13 (mod 30)

1 (mod 6)

210n + 119

7 • (30n + 17)

119 (mod 210)

17 (mod 30)

5 (mod 6)

210n + 133

7 • (30n + 19)

133 (mod 210)

19 (mod 30)

1 (mod 6)

210n + 161

7 • (30n + 23)

161 (mod 210)

23 (mod 30)

5 (mod 6)

210n + 203

7 • (30n + 29)

203 (mod 210)

29 (mod 30)

5 (mod 6)

 

The count of the numbers that do not have a prime pN or lower as their lowest prime factor within an interval of pN# represents the count of the numbers that will have pN+1 as their lowest prime factor in an interval of pN# • pN+1 = pN+1#. The previous expression can be generalized as:

 

            rN+1 = pN# – S (rL • {r}NL)       {for L = 1 to N                                     (3)         

 

From equation (2) we have {r}NL = pN# / pL#, and from equation (1) we have pN# = pN • pN-1#, so pN can be factored from all of the terms on the right of equation (3) similar to what was done with r3, which leads to rN+1 = rN • (pN – 1), or alternatively, with N > 1:

 

            rN = rN-1 • (pN-1 – 1)                                                                                        (4)

 

Thus rN is a constant for all pN# intervals. Solving equation (4) for r4 gives r4 = r3 • (p3 – 1) = 2 • (5 – 1) = 8, which is in agreement with what we arrived at previously. Not only have we proven that rN is a constant, but we have proven that it is related to previous values of r.

                                                                                                                                    Q.E.D.

 

This relationship also corresponds to the local minima of Euler’s totient (phi) function and appears in the On-Line Encyclopedia of Integer Sequences as A005867.[10]

 

Obviously there are no numbers less than pN that can have pN as a factor, and since pN itself is prime, it is not necessary to consider any primorial intervals that start at pN or lower. It is the intervals that start at pN + 1 and above that are important. In those intervals, the numbers represented by rN all must be composite; therefore we can refer to rN as the composite to primorial ratio. The interval that starts at pN + 1 would complete its first full primorial interval at pN + pN#. Because in some ways we can think of composite numbers as molecules composed of their constituent prime factor atoms and arranged in a lattice across their primorial interval, this value shall be called the first atomic boundary and labeled aN. If we make a function TN(x) to count the number of composites that have pN as their lowest prime factor, then we can say that rN = TN(aN). Table 3 lists these values for the first twelve primes.

 

Table 3: Composite to Primorial Ratio and Ratio Summation for the First Twelve Primes

 

N

pN

pN#

aN = pN# + pN

rN = TN(aN)

Σ (Numerator)

1

2

2

4

1

1

2

3

6

9

1

4

3

5

30

35

2

22

4

7

210

217

8

162

5

11

2310

2321

48

1830

6

13

30030

30043

480

24270

7

17

510510

510527

5760

418350

8

19

9699690

9699709

92160

8040810

9

23

223092870

223092893

1658880

186597510

10

29

6469693230

6469693259

36495360

5447823150

11

31

200560490130

200560490161

1021870080

169904387730

12

37

7420738134810

7420738134847

30656102400

6317118448410

 

Now that we have a proof that the count of the numbers having pN as their lowest prime factor is constant over any interval of pN#, what can we say about the pattern of those numbers?

 

Theorem 2: The pattern of the numbers having pN as their lowest prime factor repeats over intervals of the primorial pN#.

 

Proof: Again let the sequence of numbers a0, a1, a2, … am-2, am-1 represent all the integers within any interval of the primorial pN# such that ai Î {a} and (n • pN# + A) < ai ((n + 1) • pN# + A), where n is a non-negative integer and A is an integer offset where 1 ≤  A pN# so that all ai > 1.

 

When the primorial pN# is divided into {a}, the resulting residue system {r}N is a complete residue system. This means that each residue ri in {r}N can be mapped to exactly one of the integers in {a}. Map the residue r0 = 0 to a0, and the residue r1 = 1 to a1, and so on, up to rm-1 = m – 1, which maps to am-1. The result of this mapping is that ai º i (mod pN#) for all i where 0 ≤  i ≤  m – 1 and where m represents the modulus pN# so that m – 1 = pN# – 1.

 

Let n = n + 1 so that a new sequence {a¢} is generated for the next primorial interval. That is, let a0¢, a1¢, a2¢, … am-2¢, am-1¢ represent all the integers within the next primorial interval such that ((n + 1) • pN# + A) < ai¢ ((n + 2) • pN# + A) for all ai¢.

 

This means that ai¢ = ai + pN# for all i where again 0 ≤  i ≤  m – 1 and m – 1 = pN# – 1. Since ai º i (mod pN#) for all i and pN# º 0 (mod pN#), this implies that ai¢ º i (mod pN#) for all i, meaning that the residues for the next primorial interval can be mapped to the integers within the primorial in the exact same order every time.

 

As shown in the proof of theorem 1, the numbers that have pN as their lowest prime factor have residue modules the primorial pN# that relates to a specific residue in the previous pN-1# primorial. For example, the numbers that are 0 (mod 5) within a primorial of 5# = 30 and thus have p3 = 5 for their lowest prime factor have residues of 1 or 5 (mod 6). Since those residues are mapped in the same order for every interval, the numbers that have pN as their lowest prime factor occur in the same order in every interval as well. This can easily be seen in Table 1 where the residues modulo p3# = 30 would repeat in the same order for the next n + 1 interval.

 

                                                                                                                                    Q.E.D.

 

Theorem 3: The pattern of the numbers having pN as their lowest prime factor is symmetrical within intervals of the primorial pN#.

 

Proof: Again let the sequence of numbers a0, a1, a2, … am-2, am-1 represent all the integers within any interval of the primorial pN# such that ai Î {a} and (n • pN# + A) < ai ((n + 1) • pN# + A), where n is a non-negative integer and A is an integer offset where 1 ≤  A pN# so that all ai > 1.

 

Then once again divide the primorial pN# into {a} and map the resulting residues r0 = 0 to a0, and r1 = 1 to a1, and so on, up to rm-1 = m – 1 to am-1, such that ai º i (mod pN#) for all i.

 

Now pair up the members of the interval based on the residues. Pair a0 with am-1, a1 with am-2, and so on, such that for each pair (aj, ak) the sum j + k = m – 1. Since aj º j (mod pN#) and ak º k (mod pN#), then by the properties of congruence, (aj + ak) º (j + k) (mod pN#) º (m – 1) (mod pN#).[11] Since m – 1 = pN# – 1, that is equivalent to saying (aj + ak) º (–1) (mod pN#).

 

Now suppose we pair up the members of the interval again, but this time pair a0 with a1, then a2 with am-1, and a3 with am-2, and so on, such that for each pair (aj, ak) either the sum j + k = 1 or the sum j + k = m + 1. Since m represents the modulus pN#, that is equivalent to saying (aj + ak) º 1 (mod pN#).

 

Likewise it is possible to produce pairs (aj, ak) for every residue ri in the original complete residue system so that (aj + ak) º ri (mod pN#). We can say that the residue system {r} is invariant under this pair-wise transformation.

 

We can also invert the residue system by subtracting each ai from ((n + 1) • pN# + A) + 1 to generate a new sequence {a¢}. Dividing the primorial pN# into {a¢} will result in the complete residue system {r¢}, where ri¢ = m – ri. For example, we previously noted that an interval for the primorial p2# = 6 contains the values (6n + 1), (6n + 2), (6n + 3), (6n + 4), (6n + 5), and (6n + 6) which produces the residue system {1, 2, 3, 4, 5, 0} (mod 6). Subtract these values from (6n + 7) and the results are the values 6, 5, 4, 3, 2, 1 which produces the residue system {0, 5, 4, 3, 2, 1} (mod 6), which is the inverse of the original residue system. So we can also say that the residue system {r} is invariant under inversion.

 

That {r} is invariant under these transformations allows us to conclude that the pattern is symmetrical. That symmetry can readily be seen in Table 1.

                                                                                                                                    Q.E.D.

 

Conclusions

 

The technique of characterizing composites by their lowest prime factor over primorial intervals potentially has several useful applications. It is hoped that these proofs can help in the development of such applications.

 

Acknowledgments

 

I am grateful to the following individuals for their assistance and advice: Dr. Peter Gibson, The University of Alabama in Huntsville; Dr. Charles Ryavec and Dr. Jeffrey Stopple, University of California, Santa Barbara; Kaustuv M. Das and Joseph Paunovich of DWT. And finally a special thanks to DP Technology Corp. and its management for their support and encouragement.

 

References:

 

1.      Dickson, Leonard E. “History of the Theory of Numbers Volume I: Divisibility and Primality”. Mineola, NY: Dover Publications, Inc., 2005, p. 439.

 

2.      Weisstein, Eric W. "Twin Peaks." From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/TwinPeaks.html

 

3.      Eric W. Weisstein. “Prime Number.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/PrimeNumber.html

 

4.      Eric W. Weisstein. “Fundamental Theorem of Arithmetic.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/FundamentalTheoremofArithmetic.html

 

5.      Eric W. Weisstein. “Prime Factorization.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/PrimeFactorization.html

 

6.      Eric W. Weisstein. “Composite Number.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/CompositeNumber.html

 

7.      Eric W. Weisstein. “Primorial.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/Primorial.html

 

8.      Eric W. Weisstein. “Congruence.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/Congruence.html

 

9.      Eric W. Weisstein. “Complete Residue System.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/CompleteResidueSystem.html

 

10.  Sloane, N. J. A. Sequence A005867 in "The On-Line Encyclopedia of Integer Sequences."

 

11.  Eric W. Weisstein. “Congruence.” From MathWorld--A Wolfram Web Resource. http://mathworld.wolfram.com/Congruence.html

 

Bibliography:

 

Caldwell, Chris K. “The Prime Pages”, http://primes.utm.edu/.

 

Crandall, R. and Pomerance, C. Prime Numbers: A Computational Perspective, 2nd ed. New York: Springer-Verlag, 2005.

 

Dickson, Leonard E. “History of the Theory of Numbers Volume I: Divisibility and Primality”. Mineola, NY: Dover Publications, Inc., 2005.

 

Guy, Richard K. “Unsolved Problems in Number Theory, Third Edition”. New York: Springer-Verlag, 2005.

 

Riesel, H. Prime Numbers and Computer Methods for Factorization, 2nd ed. Boston, MA: Birkhäuser, 1994.