Twin prime search?

[pacionet]

OK , so can you started a new thread somewhere in this forum where you put:

1) The free ranges that you would like somebody test
2) The recommended sieving depth for each range

And who wants to test can post a thread with the reserved range.

?
This way we have a place where we can coordinate the search (maybe in the future I can edit a web page replacing the thread)

[MooooMoo]

Originally Posted by pacionet

OK , so can you started a new thread somewhere in this forum where you put:

1) The free ranges that you would like somebody test
2) The recommended sieving depth for each range

And who wants to test can post a thread with the reserved range.

?
This way we have a place where we can coordinate the search (maybe in the future I can edit a web page replacing the thread)

I prefer to wait until there is a new subforum for the project (I posted a request for a new forum in the "forum feedback" section a few days ago, and I'll probably e-mail the admins soon.)

As for now, the n is 195,000, with k searched for all values below 100,000. You or anyone else is free to continue the search from there, but please post your reservations and progress in this thread. People will get to choose as large or as small a range as they want.

I'd suggest to sieve until the rate at which k's are eliminated is equal to the time for one LLR test.

[victor]

If you explain a little more on how to participate (software), I'd put 1-2 CPU on this project ;)

[MooooMoo]

Originally Posted by victor

If you explain a little more on how to participate (software), I'd put 1-2 CPU on this project ;)

I'm happy to assist you

Instructions:

1.) Download NewPGen from

http://primes.utm.edu/programs/NewPGen/

and LLR from


http://www.mersenne.org/gimps/

(choose llr370.zip)

2.) The n is fixed at 195000, and the base is 2. Choose any range of k, and use NewPGen's "twin search" feature to begin sieving. However, post your chosen range in this thread before starting on it, and choose the lowest range of k available (currently, all ranges above 100,000 are available). The ranges can be as large or as small as you want.

3.) Sieve until the rate at which k's are eliminated is equal to the time for one LLR test (about two minutes, but varies greatly depending on your computer's speed).

4.) NewPGen will generate a file after sieving is complete. Cut and paste that file in the folder where LLR is in. Run the program, which will do the rest.

5.) After your range is complete, check LLR's output file to see whether you've found any primes. Post any primes here, even if they are not twin primes, and I'll do the rest.

If you need additional help, feel free to post here.

P.S. Please do me a favor by posting in this thread

http://www.mersenneforum.org/showthread.php?t=5744

and saying that the twin prime search is a good idea. If enough people post, we'll get a new subforum :)

[pacionet]

What is the limit that the software can handle for k in our search ?

How n will change in our search (first is 195000 ,if somebody wants to change it he can choose 195001, then 195002 and so on ? Of course I mean that the search is always "fixed n" )

[axn]

Shouldn't this thread be moved to the new forum?

[victor]

Originally Posted by axn1

Shouldn't this thread be moved to the new forum?

Yes, it should. :)

MooooMoo, we now need a "range reservation" thread, and some posting rules (such as posting 1 thread/finished range) for the new subforum :)

[edorajh]

Congrats on new subforum and new prime searching project. Wish you many Top-5000 primes. I hope I will have an opportunity to take part in searching for Top-5000 primes for this project as well.

[pacionet]

I repost the question because it seems to me important in our search.

What is the maximum k we will test for each n ? I think we will test until:

1) a prime is found
2) we arrive to test maximum k that software can handle (which is this value ?)

I think we must find what is the value for point 2) .

[MooooMoo]

Originally Posted by pacionet

I repost the question because it seems to me important in our search.

What is the maximum k we will test for each n ? I think we will test until:

1) a prime is found
2) we arrive to test maximum k that software can handle (which is this value ?)

I think we must find what is the value for point 2) .

We'll stick with n=195000 until a new twin prime is found. The maximum value of k for LLR and NewPGen is at least 10^12, so we won't need to worry about that for awhile. However, if we do exceed the maximum k value, we'll just go to n=195001, 195002, 195003..., etc.

[pacionet]

I moved my answer here because I don't want to waste space in "Reserve ranges" thread.

pacionet and victor,

It'd be really nice if one of you created a web site for this, but it's not that necccessary now. It might be needed as the project expands, though, so if you want to create one, I'd suggest the format of distributed.net. I don't really like wiki style for websites.

Where can I find some models for web site to modify ? Or I have to create the site from scratch ? At the beginning the site would be some HTML pages (HOME, SOFTWARE (link to download), STATUS, HELP, CONTACT US), with a frame on the left with the previous links.