My Math Forum Twin prime conjecture proof

 Number Theory Number Theory Math Forum

January 1st, 2017, 06:09 PM   #11
Senior Member

Joined: May 2016
From: USA

Posts: 467
Thanks: 198

Quote:
 Originally Posted by MrAwojobi So why do you have a problem with the fact that I have made my proof accessible to college kids by stating 5 lines that you have compressed into 1 line. At least you didn't have a problem with the mathematics. Make comments about the mathematics instead of criticizing the presentation.
This is entirely wrong. Conciseness is the result of careful expression. That which is unnecessarily prolix invites confusion.

Perhaps I am incorrect that all you intended by your first five lines is

"All primes > 3 can be expressed in the form $6n \pm 1$" and

"Consider the sequence of all primes > 3 in ascending order."

If I am incorrect, what did you mean?

If I am correct, however, then there is no question that 2117 and 2119 are not in that sequence because they are not prime. So skipjack's example is not on point. But who is at fault for making obscure what could be clear? You are. And as skipjack says, your saying that something is clear does not make it so. It is the writer's obligation to present things as clearly as possible. You have failed to do so as demonstrated by the responses you have received.

January 1st, 2017, 10:33 PM   #12
Member

Joined: Sep 2016
From: USA

Posts: 98
Thanks: 36

Math Focus: Dynamical systems, analytic function theory, numerics
Quote:
 Originally Posted by MrAwojobi In the sieve described here, it should be clear that the number of multiples of 5 eliminated in the number sequence > the number of non-multiples of 5 eliminated for any N.
Things that are "clear" in math often have a bad habit of being wrong. In this case, what you are claiming is indeed wrong. The prime number theorem gives asymptotic estimates for the number of composite numbers less than $N$ explicitly. This is provably $N - \pi(N) \approx N - \frac{N}{\log N}$. You may notice that as a ratio of $N$, this approaches 1.

On the other hand, the number of multiples of 5 less than $N$ is approximately $\frac{N}{5}$ which you may notice as a ratio of $N$ is bounded away from 1.

In fact, the density of multiples of 5 to non-multiples of 5 is 0 which you might recognize contradicts your assertion that it is "clear" that the multiples of 5 is greater than non-multiples of 5.

Quote:
 Originally Posted by MrAwojobi Also the minimum difference between any 2 multiples of 5 is 10 and there will be a pair of numbers, 6n-1 and 6n+1, in between these 2 multiples of 5.
I hope it is clear why this statement is false given the flaw in the previous statement.

January 2nd, 2017, 05:35 AM   #13
Global Moderator

Joined: Dec 2006

Posts: 16,369
Thanks: 1172

Quote:
 Originally Posted by MrAwojobi . . . the minimum difference between any 2 multiples of 5 is 10 and there will be a pair of numbers, 6n-1 and 6n+1, in between these 2 multiples of 5.
Yes, but such a pair needn't be prime. If the odd multiples of 5 are 2115 and 2125, the only such pair lying between them is (2117, 2119), and 2117 and 2119 are both composite.

Twin primes greater than (3, 5) must have the form (6n - 1, 6n + 1). From all such number pairs, MrAwojobi considers sieving out those which aren't prime pairs by removing (composite) multiples of successive odd primes. As there are infinitely many primes, there are infinitely many pairs of the form (6n - 1, 6n + 1) for which at least one of these numbers is prime. It seems highly plausible that the other number will sometimes be prime as well, even if one uses only very large values of n, but this is currently unproved. The sieving method is a very simple way of finding primes, and can in principle be used to find all primes. Using this method to find twin primes will in principle find all such pairs, but examining this process doesn't seem to provide any proof that there are infinitely many of them. The attempted proof by MrAwojobi has flaws, as has been pointed out.

 January 2nd, 2017, 06:56 AM #14 Senior Member   Joined: Aug 2010 Posts: 158 Thanks: 4 I have noticed an error concerning a claim I made in the proof. I will reword that part of the proof and post an update.
 January 4th, 2017, 03:31 AM #15 Math Team     Joined: Jul 2011 From: North America, 42nd parallel Posts: 3,140 Thanks: 157 I think 2117 and 2119 are not in the final sequence because they have already been eliminated by 29 and 13 sieving. The way I understand his post is ... he list's the sequence of positive integers of the form $6n \pm 1$ ... then he applies the sieve (like Eratosthenes) on this list ... then he examines the numbers that remain and draws his conclusions on those remaining numbers. For example 35 gets eliminated by 5 so is NOT there to be eliminated by 7 and since 35 is gone by the time we get around to eliminating multiples of 7 , it does not add to the count of numbers eliminated by 7. So his claim ... "it should be clear that the number of multiples of 5 eliminated in the number sequence > the number of non-multiples of 5 eliminated for any N." seems true to me However , I don't think the final claim of his post follows logically IMHO
January 4th, 2017, 07:01 AM   #16
Senior Member

Joined: May 2016
From: USA

Posts: 467
Thanks: 198

Quote:
 Originally Posted by agentredlum For example 35 gets eliminated by 5 so is NOT there to be eliminated by 7 and since 35 is gone by the time we get around to eliminating multiples of 7 , it does not add to the count of numbers eliminated by 7. So his claim ... "it should be clear that the number of multiples of 5 eliminated in the number sequence > the number of non-multiples of 5 eliminated for any N." seems true to me
"Seems true to me" has a tad less force than QED.

Let's try to figure out what is being asserted. First note that N and n are different variables: n is any natural number whereas N is a natural number such that $N = 6k \pm 1,\ k \in \mathbb N^+.$

Anyway, the assertion is (assuming that I am reading this prolix stuff correctly) that in the finite sequence of numbers of the form $6n\pm 1$ from 5 to N, the number that are evenly divisible by 5 exceeds the number that are not evenly divisible by 5 but are evenly divisible by EVERY other odd prime < N.

I suspect it is fairly easy to prove formally that the number of such numbers divisible by 5 exceeds the number of such numbers evenly divisible by 7 but not 5 (though the OP has not bothered to do so). It does not, however, obviously follow from that that the number of numbers in the form of $6n \pm 1$ evenly divisible by 5 exceeds the number of such numbers evenly divisible by 7 but by no lower prime plus the number of such numbers evenly divisible by 11 but by no lower prime plus the number of such numbers evenly divisible by 13 but by no lower prime, etc. However intuitive it may or may not be (and it is not intuitive to me), an appeal to intuition is not a proof.

Last edited by JeffM1; January 4th, 2017 at 07:03 AM.

 Tags conjecture, prime, proof, twin

 Thread Tools Display Modes Linear Mode

 Similar Threads Thread Thread Starter Forum Replies Last Post Al7-8Ex5-3:Fe#!D%03 Number Theory 3 September 30th, 2013 04:52 PM Hedge Number Theory 41 August 22nd, 2012 09:54 AM Macky Number Theory 8 September 28th, 2010 11:39 AM MrAwojobi Number Theory 51 August 9th, 2010 11:09 AM ogajajames Number Theory 4 April 26th, 2010 05:51 AM

 Contact - Home - Forums - Top