mersenneforum.org  

Go Back   mersenneforum.org > Extra Stuff > Miscellaneous Math

Reply
 
Thread Tools
Old 2020-08-26, 14:26   #23
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

22·1,447 Posts
Default

Quote:
Originally Posted by jzakiya View Post
If you bother to read the paper I explain why this is the case, give you the function to exactly determine the number of their residues frequency, and how to estimate them for any region.

Again, if you're not willing to study the paper you won't understand the reasons why the gap structure of primes exist in the totally deterministic structure I show you that they do.

Please do a little reading before commenting.
I was commenting on your assertion that testing up to <some_limit> is enough to confirm or refute. It isn't. It would merely provide evidence to support or not support. It can't confirm or refute anything.
retina is online now   Reply With Quote
Old 2020-08-26, 14:35   #24
jzakiya
 
Jul 2014

1F16 Posts
Default

If you read my statement before, I already said that what you just stated would be what some people would say about counting the (2,n) tuples up to some finite number.

If you aren't going to actually read, and understand, anything I write why comment on it?

Last fiddled with by jzakiya on 2020-08-26 at 14:35
jzakiya is offline   Reply With Quote
Old 2020-08-26, 14:43   #25
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

22·1,447 Posts
Default

Quote:
Originally Posted by jzakiya View Post
If you aren't going to actually read, and understand, anything I write why comment on it?
I only commented on the part I quoted. And that part I quoted was either attempting to mislead, or shows a basic misunderstanding you have. Either way it was wrong to say it IMO.
retina is online now   Reply With Quote
Old 2020-08-26, 15:15   #26
mart_r
 
mart_r's Avatar
 
Dec 2008
you know...around...

2×5×59 Posts
Default

Quote:
Originally Posted by jzakiya View Post
Update: Wed 2020/7/8

I've now also determined the computational form for
gap coefficients a6 and a7, which are included in
the list below. I also computer generated the PGS
gaps for P37, which is also included (which took
19 hrs 42 mins, with a multi-threaded algorithm).
(...)

for Pn: a1 = (a1')(pn - 2)
for Pn: a2 = (a2')(pn - 2)
for Pn: a3 = (a3')(pn - 3) + (a2')(1) + (a1')(1)
for Pn: a4 = (a4')(pn - 4) + (a3')(1)
for Pn: a5 = (a5')(pn - 5) + (a4')(2) + (a3')(1)
for Pn: a6 = (a6')(pn - 5) + (a5')(6) - (a4')(2)
for Pn: a7 = (a7')(pn - 7) + (a6')(3) - (a5')(3) + (a4')(4)


P37
a1 = 217,929,355,875
(...)

I did just about the same calculations two years ago until I concocted a Pari program that lets you find the number of gaps up to gap size = 212 in "prime generators", as I understand it, the ring of integers coprime to p#.

If you want to do some number crunching, I've attached the program (Pari code together with three data files for the code). Note that the algorithm doesn't work when gap > 2*p.

(The data for "a_z.txt" was taken from S. Herzog, I think it was http://zigherzog.net/primes/mathconstants/trk.html, a link which is now dead. Similar data can be found under http://pauillac.inria.fr/~harley/brent.html)

Alas, though, it won't actually prove Polignac's conjecture. Sorry I can't give any details, as I'm not deep enough into the matter. But I'm sure if it was proven, you'd read it in the newspapers (yes, I'm a child of the 20th century...).
Attached Files
File Type: zip gaps_coprimes.zip (33.4 KB, 5 views)
mart_r is offline   Reply With Quote
Old 2020-08-26, 15:27   #27
CRGreathouse
 
CRGreathouse's Avatar
 
Aug 2006

2·2,963 Posts
Default

Quote:
Originally Posted by jzakiya View Post
If you bother to read the paper I explain why this is the case, give you the function to exactly determine the number of their residues frequency, and how to estimate them for any region.

Again, if you're not willing to study the paper you won't understand the reasons why the gap structure of primes exist in the totally deterministic structure I show you that they do.

Please do a little reading before commenting.
There isn't any actual mathematical content in the paper. I would enjoy a discussion of the paper but there simply isn't anything there to discuss. You don't seem to have understood the basic premise of what a proof entails if you think that you've presented one here.

I don't say this to be unkind, but merely to explain why no one will be able to give you what you would like. Normally I'd glance at a proof to see if I could find a tell-tale mistake, but here there isn't even an attempt to prove the proposition at hand.
CRGreathouse is offline   Reply With Quote
Old 2020-08-26, 16:17   #28
bsquared
 
bsquared's Avatar
 
"Ben"
Feb 2007

CDF16 Posts
Default

Quote:
Originally Posted by jzakiya View Post
This is an application of Prime Generator Theory (PGT) I began developing in 2008, and applied to creating the fastest/most efficient prime sieves, as independently verified in the link below, and a Rubygem.

https://github.com/hathix/prime-algorithms

https://github.com/jzakiya/primes-utils
Your SSoZ is not new; sieving over residue classes has been done for a long time. An early example is the (IMO lovely named) black-key sieve in 1993. (It is considered bad form, by the way, to name a technique/algorithm/proof after yourself.) Yafu also sieves by residue class; admittedly by arranging the bits slightly differently than you do (if I understand your algorithm correctly), but performing > 6x faster in most cases. GPU k-generators for GIMPS trial factoring also sieve this way.

[edit]
I am intrigued by the thought of packing the residue-class bits contiguously, and then sieving at fixed bit-positions for each class. Yafu makes independent arrays for each class and sieves each in turn. There are pros and cons to each approach.

Independent array pros:
* trivial to parallelize across classes
* fewer segments to represent a given number-line range, per class

Cons:
* variable-bit memory access per sieve step (so that one must compute a bit-offset in the current byte for each step). I'm actually not sure this is a bottleneck though, because the computation is so much faster than the memory access.

Bit-packed pros:
* fixed-bit sieving (perhaps an improvement, see cons above)
* a given segment remains cached throughout class sieving (this is probably a bigger win)
* primes can be generated in-order

Cons:
* more segments to represent a given number-line range
* parallelization would probably be best done across sub-segments of the range

I might have to play around with some code to see how things compare

Last fiddled with by bsquared on 2020-08-26 at 16:54
bsquared is offline   Reply With Quote
Old 2020-08-26, 17:40   #29
jzakiya
 
Jul 2014

378 Posts
Default

Quote:
Originally Posted by bsquared View Post
Your SSoZ is not new; sieving over residue classes has been done for a long time. An early example is the (IMO lovely named) black-key sieve in 1993. (It is considered bad form, by the way, to name a technique/algorithm/proof after yourself.) Yafu also sieves by residue class; admittedly by arranging the bits slightly differently than you do (if I understand your algorithm correctly), but performing > 6x faster in most cases. GPU k-generators for GIMPS trial factoring also sieve this way.
Had never heard or seen anything about the black-key sieve. It obviously hasn't been adopted much. But again, I provide a comprehensive mathematical framework which explains how to use any even number as the modulus for a prime generator, i.e. a generalized method to explain how it works, and why, with coded examples.

Also its interesting that apparently the Sieve of Eratosthenes, Sieve of Atkin, and Sieve of Sundaram are totally good form, but the Sieve of Zakiya somehow is not. Why? Who should my sieves and techniques be named after? So only certain types of people have work named after them?

https://en.wikipedia.org/wiki/Generation_of_primes

Thankfully, thousands of people have downloaded my Rubygems, read my papers, and are using my work.
jzakiya is offline   Reply With Quote
Old 2020-08-26, 18:38   #30
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

210038 Posts
Default

Quote:
Originally Posted by jzakiya View Post
Also its interesting that apparently the Sieve of Eratosthenes, Sieve of Atkin, and Sieve of Sundaram are totally good form, but the Sieve of Zakiya somehow is not. Why? Who should my sieves and techniques be named after? So only certain types of people have work named after them?
To have others name something after you is an honour. To name something for yourself is pride.
Eratosthenes came up with the idea. It was a good one, people called it after him.
Uncwilly is online now   Reply With Quote
Old 2020-08-26, 18:45   #31
kruoli
 
kruoli's Avatar
 
"Oliver"
Sep 2017
Porta Westfalica, DE

3·109 Posts
Default

Quote:
Originally Posted by jzakiya View Post
Thankfully, thousands of people have downloaded my Rubygems, read my papers, and are using my work.
It is nice that you can be comfortable with your work. Nonetheless, we try to base everything here on facts, not on your reach or your "customer ratings".

You could even have had a great breakthrough once and your next proposal could as well be rubbish. But vice versa is of course also possible. In all cases, we need facts, at best a mathematical proof. CRGreathouse is really trying to push you forward in a friendly manner!
kruoli is online now   Reply With Quote
Old 2020-08-26, 19:00   #32
CRGreathouse
 
CRGreathouse's Avatar
 
Aug 2006

172616 Posts
Default

Quote:
Originally Posted by Uncwilly View Post
To have others name something after you is an honour. To name something for yourself is pride.
Eratosthenes came up with the idea. It was a good one, people called it after him.
I have a project tracking self-naming and non-self-naming in the OEIS at
https://oeis.org/wiki/User:Charles_R...ouse_IV/Vanity

It needs vastly more work than I am able to provide, but it does give some kind of an idea of the norms and expectations in the OEIS, which are somewhat more relaxed than those of academic mathematics, which in turn is less cut-throat than many (most?) other academic disciplines.

Incidentally I would be happy to receive feedback (by private message or email) on this or any of my other projects.
CRGreathouse is offline   Reply With Quote
Old 2020-08-26, 19:01   #33
jzakiya
 
Jul 2014

31 Posts
Default

Quote:
Originally Posted by kruoli View Post
It is nice that you can be comfortable with your work. Nonetheless, we try to base everything here on facts, not on your reach or your "customer ratings".
Well, since you aren't going to address any of the "facts" presented in the paper there's little else left for me to say to you. For others who are interested/curious I can answer their questions to help them understand.
jzakiya is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Twin Prime Constellations robert44444uk Prime Gap Searches 21 2019-06-09 12:51
How do you efficiently sieve for prime 3/4-tuples? Puzzle-Peter Software 156 2019-06-03 20:19
find very easy twin prime in the infamy twin primes hal1se Miscellaneous Math 13 2018-11-05 16:34
Highest Prime is also a twin prime... NOT hydeer Lone Mersenne Hunters 9 2018-04-03 22:54
Twin Prime Days, Prime Day Clusters cuBerBruce Puzzles 3 2014-12-01 18:15

All times are UTC. The time now is 16:17.

Wed Oct 21 16:17:35 UTC 2020 up 41 days, 13:28, 1 user, load averages: 2.01, 1.78, 1.78

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.