mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Twin Prime Search

Reply
 
Thread Tools
Old 2019-02-25, 22:32   #1
Pietro Maiorana
 
Feb 2019

3 Posts
Default I found a sieve to search all pairs of twin primes

Through the use of some sequences that make up a useful sieve to calculate all the pairs of first twins it proves that they are endless.

Sieve_for_Oeis_A077800_Twin_prime_conjec.pdf
Pietro Maiorana is offline   Reply With Quote
Old 2019-02-25, 23:25   #2
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

9,127 Posts
Default

Will the colors in this PDF make my eyes bleed?

Quote:
THE ANSWER WITHOUT THE SHADE OF DOUBT IS YES !!
Batalov is offline   Reply With Quote
Old 2019-03-09, 06:14   #3
bbb120
 
Feb 2019

3010 Posts
Default

Quote:
Originally Posted by Pietro Maiorana View Post
Through the use of some sequences that make up a useful sieve to calculate all the pairs of first twins it proves that they are endless.

Attachment 19954
please tell me the least twin primes which great than 10^1000
bbb120 is offline   Reply With Quote
Old 2019-03-09, 14:09   #4
Dr Sardonicus
 
Dr Sardonicus's Avatar
 
Feb 2017
Nowhere

1101101011002 Posts
Default

Quote:
Originally Posted by bbb120 View Post
please tell me the least twin primes which great than 10^1000
This may already be known, but a cursory search failed to turn it up. I did find a lot of larger pairs of twin primes, though, so a least pair greater than 10^1000 certainly exists.

Judging from, e.g. Twin Prime Statistics, they could be found as follows:

1) Sieve a "suitably" long interval (10^1000, 10^1000 + N) to eliminate all numbers with "small" factors.

2) Subject any remaining pairs differing by 2, starting with the smallest, to a pseudoprime test.

3) If you're really ambitious, subject any twin pseudoprimes you find to a rigorous primality test until you get a pair of proven primes. With numbers of this size, this might take a while.

Based on asymptotic estimates and my own sloppy estimates, you'd probably have to take N on the order of 10^7 or 10^8 to stand a decent chance of success.

I have no idea what a good sieving limit for "small" factors would be.

Last fiddled with by Dr Sardonicus on 2019-03-09 at 14:15 Reason: Clarification
Dr Sardonicus is offline   Reply With Quote
Old 2019-03-09, 14:40   #5
danaj
 
"Dana Jacobsen"
Feb 2011
Bangkok, TH

38A16 Posts
Default

Sigh, I now realize this was a test for the OP to show how well his method works. Well, ok um you can use the rest of this as a comparison I guess.


Quote:
Originally Posted by bbb120 View Post
please tell me the least twin primes which great than 10^1000
10^1000 + 9705091 is the smaller of the pair.

I'm sure it's online somewhere. It took about 2 minutes for my Macbook to find it. They are ES BPSW, and then I let Pari/GP prove them, a little over 1 minute each.

The software I used have chosen 80000 * log2(high) as the depth, purely because empirically that seemed to work well at a variety of sizes. Each candidate that passes this sieve is subjected to MR base 2 for both low and high, then ES Lucas for both low and high. It took 2.5 seconds for a width of 1e5, 13s for 1e6, 124s for 1e7 where it found one (I just ran it multiple times with a bigger window, redoing the previous part).

Last fiddled with by danaj on 2019-03-09 at 14:48
danaj is offline   Reply With Quote
Old 2019-04-16, 22:11   #6
Pietro Maiorana
 
Feb 2019

3 Posts
Default

(Start)
A100319+1 (Even numbers m such that at least one of m-1 and m+1 is composite)
can be obtained as the union of:

3*A005818;

5*A038179 without 2;


7*A007310 without 1;


A038511


A025584 without 2, 3. (End)

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

The sequence A038511 can be obtained by multiplying each term of A008364 , except {1}, by itself and by all subsequent terms. Rewrite the terms in ascending order.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
(Start)
To obtain A025584 (Primes p such that p-2 is not a prime):
Write all terms of the form 2k + 7 and delete all terms of the form (6k + (-1) ^ k + 13)/2 [without A121764] and all terms of A092256
Finally, add 2 and 3.
(End)



To calculate A121764 (Single, or isolated or non-twin primes of form 6n + 1)
consider all term of A092256(n) + 2 except those in common with:
A038511, 5*A038179, and 7*A007310.



A092256 (Nonprimes of form 6k+5) is the union of:
numbers of the form 5*(6k + 1) , numbers of the form 7*(6k + 5), and terms of A038511 of the form 3*k+2
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Last fiddled with by Pietro Maiorana on 2019-04-16 at 22:15
Pietro Maiorana is offline   Reply With Quote
Old 2019-09-26, 21:03   #7
Pietro Maiorana
 
Feb 2019

3 Posts
Thumbs up

Each pair of odd twin primes (Oeis A077800) has an even median value.

The non-median even values ​​(complementary to Oeis A014574) are found in the sequence Oeis A100319

(Even numbers m such that at least one of m-1 and m+1 is composite)

According to my theory, A100319+1 can be divided into 5 infinite sub-sequences

a) 3*A005818; b) 5*A038179 without 2; c) 7*A007310 without 1; d) A038511; e) A025584 without 2, 3.

The first three subsequences can be rewritten respectively as:

a) 9+6*(n-1)

b) 5*(floor((41/21 - (3 mod n))^(-3*n+5)) + 3*n - 4)

c) 7/2*(6*k+(-1)^k+9)

The sequence

d) A038511 can be obtained by multiplying each term of A008364, except {1}, by itself and by all subsequent terms. Rewrite the terms in ascending order.

where A008364 is equal to

a(n) = 35n/8 + O(1). - Charles R Greathouse IV, Sep 14 2015 (see Oeis)

e) To obtain A025584 (Primes p such that p-2 is not a prime):

Write all terms of the form 2k + 7 and delete: all terms of A092256 and all terms of the form (6k + (-1) ^ k + 13)/2 [without A121764]

(Start)

A092256 (Nonprimes of form 6k+5) is the union of:

numbers of the form 5*(6k + 1), numbers of the form 7*(6k + 5), and terms of A038511 of the form 3*k+2 (End)

(Start)

To calculate A121764 (Single, or isolated or non-twin primes of form 6n + 1) consider all term of A092256+2 except those in common with:

d) A038511,
b) 5*(floor((41/21 - (3 mod n))^(-3*n+5)) + 3*n - 4),
c) 7/2*(6*k+(-1)^k+9);
(End)

QUESTION: Why did I calculate A100319 + 1 and not A100319?

ANSWER: Because some sub-sequences used in A100319 + 1 as it was possible to see in the reasoning have been re-used in the construction for a sub-sequence in A025584 (exactly Oeis A121764 Single, or isolated or non-twin primes in the form 6n + 1)

---------------------------------------------------------------------------------------------------------------------
SECOND METHOD
---------------------------------------------------------------------------------------------------------------------

Consider only the sequence A025584 (Primes p such that p-2 is not a prime):

To obtain A025584 (without 2, 3)
Write all terms of the form 2k + 7 and delete: all terms of A092256 and all terms of the form (6k + (-1) ^ k + 13)/2 [without A121764]

(Start)

A092256 (Nonprimes of form 6k+5) is the union of:


numbers of the form 5*(6k + 1),
numbers of the form 7*(6k + 5),
and terms of A038511 of the form 3*k+2
(End)

(Start)

To calculate A121764 (Single, or isolated or non-twin primes of form 6n + 1)
consider all term of A092256+2 except those in common with:


a) A038511,
b) 5*(floor((41/21 - (3 mod n))^(-3*n+5)) + 3*n - 4),
c) 7/2*(6*k+(-1)^k+9);
(End)

While A038511 can be obtained by multiplying each term of 35n/8 + O(1) , except {1}, by itself and by all subsequent terms. Rewrite the terms in ascending order.

Well, if you remove from the infinite list of prime numbers, the values of A025584 can be obtained A006512 "Greater of twin primes"
Pietro Maiorana is offline   Reply With Quote
Old 2019-09-26, 22:55   #8
rudy235
 
rudy235's Avatar
 
Jun 2015
Vallejo, CA/.

312 Posts
Default

......

Last fiddled with by rudy235 on 2019-09-26 at 23:08 Reason: duplicate
rudy235 is offline   Reply With Quote
Old 2019-09-26, 23:07   #9
rudy235
 
rudy235's Avatar
 
Jun 2015
Vallejo, CA/.

312 Posts
Default

Quote:
Originally Posted by danaj View Post
Sigh, I now realize this was a test for the OP to show how well his method works. Well, ok um you can use the rest of this as a comparison I guess.




10^1000 + 9705091 is the smaller of the pair.

I'm sure it's online somewhere. It took about 2 minutes for my Macbook to find it. They are ES BPSW, and then I let Pari/GP prove them, a little over 1 minute each.
yup, there are there at factordb both proven prime in 08/2015
rudy235 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
find very easy twin prime in the infamy twin primes hal1se Miscellaneous Math 13 2018-11-05 16:34
Search for all amicable pairs up to 2^64 Sergei Chernykh Aliquot Sequences 37 2018-01-27 19:23
Top 3 twin found! MooMoo2 Twin Prime Search 4 2015-07-13 23:30
Twin: new sieve JohnFullspeed Twin Prime Search 6 2011-06-04 06:07
Twin prime search? MooooMoo Twin Prime Search 115 2010-08-29 17:38

All times are UTC. The time now is 16:35.

Thu Oct 1 16:35:04 UTC 2020 up 21 days, 13:46, 0 users, load averages: 1.23, 1.60, 1.62

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.