mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Conjectures 'R Us (https://www.mersenneforum.org/forumdisplay.php?f=81)
-   -   Report top-5000 primes here (https://www.mersenneforum.org/showthread.php?t=9782)

rogue 2008-04-30 12:34

[QUOTE=Anonymous;132422]I just noticed this part in the first post of this thread, where it instructs users to credit their primes as LLR and PFGW for top-5000 primes on a non-power-of-2 base. Actually, because LLR's code for non-k*2^n+-1 numbers is taken directly from PRP, users should enter "PRP" under additional software, not LLR, if LLR only found a probable prime. In fact, LLR will instruct users to do so with a note in the lresults.txt file--it will say "such-and-such is a probable prime. Please credit George Woltman's PRP for this result!". Also, according to the top-5000 site, when you list multiple programs in a prover-code, they all get full credit, not half credit as this thread says--this is done to encourage reporting of all programs involved.[/QUOTE]

IIRC, there was a similar discussion a while ago in the SR5 project. It came down to the point of whether or not you used LLR or PRP to do the test. Even if they share the same code, all you need to do is credit the program(s) you used. There are a number of fuzzy lines when you start going down this path. There will be a desire to credit anyone's code that was involved. One example would be GMP, which is used by PFGW or pieces of code used by George's FFT library that was borrowed from other sources. Then you can start talking about giving people credit for writing the code and the testers, etc. IMO, limit the prover codes to the applications used to find the prime.

mdettweiler 2008-04-30 15:08

[quote=rogue;132440]IIRC, there was a similar discussion a while ago in the SR5 project. It came down to the point of whether or not you used LLR or PRP to do the test. Even if they share the same code, all you need to do is credit the program(s) you used. There are a number of fuzzy lines when you start going down this path. There will be a desire to credit anyone's code that was involved. One example would be GMP, which is used by PFGW or pieces of code used by George's FFT library that was borrowed from other sources. Then you can start talking about giving people credit for writing the code and the testers, etc. IMO, limit the prover codes to the applications used to find the prime.[/quote]
But, the LLR application even tells says in the results file "Please credit George Woltman's PRP for this result"--seemed pretty clear to me.

Maybe we should ask Jean Penne how he prefers to have them reported?

michaf 2008-04-30 15:35

Riesel base 31
 
Finally, a long due result for Riesel base 31:

131994*31^68109-1 is prime! :smile:

After a 2 months wait, it's a rather satisfying top 5000 prime too! :flex:

This leaves 9 k's for Riesel base 31.

rogue 2008-04-30 17:09

[QUOTE=Anonymous;132450]But, the LLR application even tells says in the results file "Please credit George Woltman's PRP for this result"--seemed pretty clear to me.

Maybe we should ask Jean Penne how he prefers to have them reported?[/QUOTE]

I don't believe that Jean cares, but I won't speak for him.

mdettweiler 2008-04-30 17:14

[quote=rogue;132464]I don't believe that Jean cares, but I won't speak for him.[/quote]
Okay, I posted a message about this over in the LLR 3.7.1c thread in the Software forum. :smile:

gd_barnes 2008-05-02 22:19

Interesting discussion here.

I have to lean towards what Rogue suggested despite the message in Jean's LLR program. I believe the message was put there not thinking of the ramifications of a single program pulling code from multiple sources, which as Rogue implies, gets into a very slippery slope of crediting any # of contributing programs or testers.

I was not aware that each program gets full credit if more than one is used. Thanks for correcting that Anon.

I was aware of the display in Jean's program but had chosen to ignore it for the same 'slippery slope' reason.

If it comes back that PRP should be credited, I'll create a new proof code if I find a non-power-of-2 top-5000 prime. Fortunately I haven't found a non-power-of-2 top-5000 prime yet or I'd be stuck with a messy situation of moving some of them over to a new proof code, which perhaps isn't too messy for Prof. Caldwell to move a few primes over to.


Gary

gd_barnes 2008-05-23 01:52

BANG!!
 
Ba da bing...ba da boom...

The odd-n conjectures finally hit pay dirt on the Riesel side and knock a k-value out of 3 different bases...

6927*2^743481-1 is prime!!:george::george:

This also slays k=13854 for Riesel base 4 and 16. Primes for those reported in the main prime thread.


Gary

gd_barnes 2008-05-23 02:02

There is now one less k-value remaining for Riesel base 4 and 16:

13854*4^371740-1 is prime
-and-
13854*16^185870-1 is prime

submitted as:
6927*2^743481-1

Siemelink 2008-05-23 14:17

Base 19 prime
 
Hidiho,

I've been taking some of my k's to much higher n's then I usually do. With success:

8216*19^91041-1 is prime!

The only base 19 prime in the top 5000.

Having fun, Willem.

Jean Penné 2008-05-24 15:54

[QUOTE=gd_barnes;134050]Ba da bing...ba da boom...

The odd-n conjectures finally hit pay dirt on the Riesel side and knock a k-value out of 3 different bases...

6927*2^743481-1 is prime!!:george::george:

This also slays k=13854 for Riesel base 4 and 16. Primes for those reported in the main prime thread.


Gary[/QUOTE]

Very nice result, Gary!
Jean

Xentar 2008-06-12 07:25

Just saw this one in the log file:
160*17^166048+1 is a probable prime. Time: 4811.857 sec.

Place 604 :)
[url]http://primes.utm.edu/primes/page.php?id=85139[/url]

only 2 more for sierp b17


All times are UTC. The time now is 08:04.

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