Choose your own K and work on finding a top-5000 prime!

[mdettweiler]

Originally Posted by Anonymous

Reserving k=1003 from n=10K.

Status for k=1003: now at n=100K. lresults for 10K-100K are attached. Continuing with this k.

One prime was found in the range 10K-100K:
1003*2^45987-1 is prime!

[Kosmaj]

From the Top-5000:

rank           description              digits  who year comment
-----  -------------------------------- ------- ---- ---- --------------
  901  1003*2^584103-1                   175836  L51 2004 
 1354  1003*2^482655-1                   145297  L51 2004 
19266  1003*2^144163-1                    43401 g327 2003

[mdettweiler]

Originally Posted by Kosmaj

From the Top-5000:

rank           description              digits  who year comment
-----  -------------------------------- ------- ---- ---- --------------
  901  1003*2^584103-1                   175836  L51 2004 
 1354  1003*2^482655-1                   145297  L51 2004 
19266  1003*2^144163-1                    43401 g327 2003

Yes, I noticed those when I was first checking out the k--but I decided to start at n=10K anyway to fill any gaps that may be left around. After all, what's the point of all this searching for primes if you're going to do it willy-nilly and leave gaps?

[mdettweiler]

k=1003 is now tested, completely and contiguously, to n=500K. Two primes were confirmed in my range of n=10K-500K, at n=144163 and n=482655; one new prime was found, at n=45987. (All these primes have been reported in the respective post-primes-here threads.)

I am releasing k=1003, though I may come back to it in the future; however, if anyone else wants to search this k, then by no means let that stop you from reserving it, as that only holds things up and is contrary to the best interests of the prime searching community.

Residuals for n=100K-500K are attached. (BTW, in case anyone thinks the times listed for n=390K-500K look funny, that's because the results for that range were converted from the results files from an internal LLRnet server that I ran that range on. LLRnet records the total time the client had the test "checked out", rather than the actual time it took to perform the test. So, don't worry, the funny timings aren't a sign of instability. )

[Buzzo]

hi i'll take k=1121 from n=10k, thanks

[VBCurtis]

Welcome to RPS, Buzzo. Good luck with the search!

If you want help/advice setting up a sieve, PM me. If you are experimenting with the software/methods and want a short project, I recommend searching 10k-400k or lower. You are not very likely to find a top-5000 prime this way, but the project should take just a few CPU-weeks on a modern machine.

Testing 10k-100k will take just a day or so; sieve a day or less, then test 10-100k.

-Curtis

[Buzzo]

Originally Posted by VBCurtis

Welcome to RPS, Buzzo. Good luck with the search!

If you want help/advice setting up a sieve, PM me. If you are experimenting with the software/methods and want a short project, I recommend searching 10k-400k or lower. You are not very likely to find a top-5000 prime this way, but the project should take just a few CPU-weeks on a modern machine.

Testing 10k-100k will take just a day or so; sieve a day or less, then test 10-100k.

-Curtis

okay, thanks for the advice--i've decided to do 10k-300k. i figured out how to get srsieve/sr1sieve going on my own, but thanks for the offer to help anyway!

[grobie]

Reserving k=684255

[Buzzo]

k=1121 done to n=300k, 3 primes for 10k-300k as shown in primes thread. i'll release this k now.

[grobie]

Originally Posted by grobie

Reserving k=684255

I am going to let this k go. If anyone wants the sieve file PM me, it is only sieved to 340G

I am going to reserve k=2385 from n=10k

[sps27]

I'd like to search a k value that hasn't really been done at all yet (or at least not past 10K), so that I can find a few small primes at first to give me something to look at. But I also would like that k value to have reasonable potential for producing a top-5000 prime if I were to continue the search to high n values. (or are top-5000 primes not likely to be found quickly for k's that havent been tested to >10k yet?)

Can you suggest what range of k values I should be looking at. And, within that range, what makes a particular k value appealing over others? Is the number of primes found for n<10k a good predictor for future primes?

I've been glancing through posts on this forum and seen some data on websites, but a lot of it seems old/outdated, so I'm just curious where would be a good place to search given that I'm starting now?