mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   Best Work for Finding Primes (https://www.mersenneforum.org/showthread.php?t=16929)

Unregistered 2012-06-23 03:13

Best Work for Finding Primes
 
Hello,

Right now I have my computers "trial factoring to low limits" by using the "Lucas-Lehmer test." I don't know what any of that means to be quite honest but I just wanted to make sure I'm doing the right thing so I can help test a new prime! Sorry to be a bother but just making sure. Is there anything else I should be doing? Thanks!

Brain 2012-06-23 10:08

Welcome
 
Trial factoring (TF) is a preliminary step that eliminates possible candidates by finding small factors. Factor found = not prime.
For proving a prime you must run "first time Lucas Lehmer tests" (LL) which you seem to have demanded from Prime95. But if your PC is relatively slow PrimeNet will assign less demanding work (for example TF).
Could you add your PC's hardware?
Last but not least: Welcome to the fun world of finding primes and factors.

Brian-E 2012-06-23 11:02

Just to add to what Brain writes:

Trial factoring, and Lucas-Lehmer testing, are two necessary types of work. They are not the same, and there are other work types as well. You can set a work type preference if you wish on your PrimeNet account to choose what you want to do, or you can simply select the "what makes sense" option which will make PrimeNet decide what work to give you on the basis of your hardware's particular strengths plus what the project most needs at the moment.

If you want to choose your own work type, and if you want a chance of discovering a mersenne prime yourself then, as Brain says, first time LL testing is the way to go. (DC, which is doing an LL test that someone else has already done as a double check, provides a smaller chance of discovering a prime too.) An LL test will take weeks or months, though. Factoring work is faster, just as important to the project, and much more likely to give a positive result (factor found); but it will not lead to you being the discoverer of a new prime.

science_man_88 2012-06-23 12:57

[QUOTE=Brian-E;303077]Just to add to what Brain writes:

Trial factoring, and Lucas-Lehmer testing, are two necessary types of work. They are not the same, and there are other work types as well. You can set a work type preference if you wish on your PrimeNet account to choose what you want to do, or you can simply select the "what makes sense" option which will make PrimeNet decide what work to give you on the basis of your hardware's particular strengths plus what the project most needs at the moment.

If you want to choose your own work type, and if you want a chance of discovering a mersenne prime yourself then, as Brain says, first time LL testing is the way to go. (DC, which is doing an LL test that someone else has already done as a double check, provides a smaller chance of discovering a prime too.) An LL test will take weeks or months, though. Factoring work is faster, just as important to the project, and much more likely to give a positive result (factor found); but it will not lead to you being the discoverer of a new prime.[/QUOTE]

basically what is being pointed out here is that gimps does not do TF by LL.

Unregistered 2012-06-23 16:52

[QUOTE=science_man_88;303083]basically what is being pointed out here is that gimps does not do TF by LL.[/QUOTE]

Okay thank you. When I go to test -> status it says that both workers are doing Lucas-Lehmer test so from what I now understand, that's good! I have a 2.4 Ghz Intel Core 2 Duo. I think I'm good then if I'm not mistaken.

Unregistered 2012-06-23 16:54

[QUOTE=science_man_88;303083]basically what is being pointed out here is that gimps does not do TF by LL.[/QUOTE]

As well I selected the choose the work for me option since I thought it would be best.

davieddy 2012-06-24 03:47

[QUOTE=science_man_88;303083]basically what is being pointed out here is that gimps does not do TF by LL.[/QUOTE]
It's not GIMPS, it is what an LL test tells you:
Either the number is prime or it's composite, but
if composite gives no indication as to what the factors are.

(Although one might conjecture that the actual non-zero residue
could potentially help in finding the factors).

It is worth checking for factors below 2^73.
GIMPS typically searches up to 2^71,
because of an administrative cock-up.

science_man_88 2012-06-24 12:29

[QUOTE=davieddy;303130](Although one might conjecture that the actual non-zero residue
could potentially help in finding the factors).[/QUOTE]

funny you say that because I've brought a way up before, I believe. it's just I'm almost positive it's more expensive than TF. PM me if you want details.

Dubslow 2012-06-24 13:26

EWMayer has explicitly said that the (full) residue can be used to effect a primality test on the cofactor after finding a first factor.

science_man_88 2012-06-24 13:50

[QUOTE=Dubslow;303155]EWMayer has explicitly said that the (full) residue can be used to effect a primality test on the cofactor after finding a first factor.[/QUOTE]

the first m*y+z y>0 can be used so for 2^11-1 that means (194^2-2)%(2^11-1) = 788 can be used.

788%23 = 6 =((194%23)^2-2)%23 since these equal 23 is a factor of m*y at least and in this case a factor of m.


All times are UTC. The time now is 14:06.

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