Bases 2 & 4 reservations/statuses/primes

[Jean Penné]

Hi,

k=9267 is now tested up to n = 8,806,303 (2,650,966 decimal digits) no prime, continuing...

Best Regards,
Jean

[MisterBitcoin]

Originally Posted by Puzzle-Peter

R2, k=9519 finished to 2^24. No prime. Base released. R366 will pick up some speed now...

I´ve made an sieve file for this sequence up to N=25M.
The calc prime results are looking good.

using base 2, sieve depth 50000000, and reading 781255 candidates from t17_b2_k9519.npg

expected # of primes: 1.7263706830854946
odds of at least one prime: 82.20709997186222%
(or, in other words) 1 in 1.2164399429517392 chance of at least one prime
on average, 1 in 452541.8600156512 candidates should be prime

I´ll input the file into yoyo@home. Not sure if yoyo is going with my pmax=25e15.

After the file is done, around 03/2018, I´ll go ahead and start some testing. Everyone is welcomed to support.

[gd_barnes]

Originally Posted by MisterBitcoin

I´ve made an sieve file for this sequence up to N=25M.
The calc prime results are looking good.

using base 2, sieve depth 50000000, and reading 781255 candidates from t17_b2_k9519.npg

expected # of primes: 1.7263706830854946
odds of at least one prime: 82.20709997186222%
(or, in other words) 1 in 1.2164399429517392 chance of at least one prime
on average, 1 in 452541.8600156512 candidates should be prime

I´ll input the file into yoyo@home. Not sure if yoyo is going with my pmax=25e15.

After the file is done, around 03/2018, I´ll go ahead and start some testing. Everyone is welcomed to support.

Please be careful when taking on such large efforts. I seriously doubt that this is correct. Otherwise we would have easily found a prime already.

Are you sieving only the even n? If not, you are outside of the scope of the project.

The easiest way to accomplish sieving even n only for 9519*2^n-1 n=2^24 to n=25M is to sieve base 4 instead for half the n-range; that is sieve 9519*4^n-1 for n=2^12 to n=12.5M.

My guess is that at most, there is a 20% chance of prime for even n for n=2^24-25M. Are you certain that you really want to do this? It would be extremely low priority on our sieving needs at this point.

[MisterBitcoin]

Originally Posted by gd_barnes

Please be careful when taking on such large efforts. I seriously doubt that this is correct. Otherwise we would have easily found a prime already.

Are you sieving only the even n? If not, you are outside of the scope of the project.

The easiest way to accomplish sieving even n only for 9519*2^n-1 n=2^24 to n=25M is to sieve base 4 instead for half the n-range; that is sieve 9519*4^n-1 for n=2^12 to n=12.5M.

My guess is that at most, there is a 20% chance of prime for even n for n=2^24-25M. Are you certain that you really want to do this? It would be extremely low priority on our sieving needs at this point.

I did a file for 9519*2^n-1 and submitted it to yoyo.
But it looks like the number of expected primes was wrong, so I´m dropping this effort.

[gd_barnes]

Originally Posted by MisterBitcoin

I did a file for 9519*2^n-1 and submitted it to yoyo.
But it looks like the number of expected primes was wrong, so I´m dropping this effort.

I'm still not sure that you understand.

The number of expected primes that you came up with is likely correct for ALL n but that is not what CRUS needs. You would need to sieve only even n. Sieving 9519*4^n-1 accomplishes that. Plenty of primes have already been found for base 2 k=9519 odd n. 9519*2^1-1, 9519*2^11-1, and 9519*2^15-1 are all prime and those are just the smallest 3 primes.

Regardless the effort is apparently not what you expected so it's best to drop it at this point. Few people want to run tests that take more than 1 day each.

[Puzzle-Peter]

Originally Posted by gd_barnes

Regardless the effort is apparently not what you expected so it's best to drop it at this point. Few people want to run tests that take more than 1 day each.

I do enjoy those long tests, so I started a sieve for n=2^24 to 2^26.

[MisterBitcoin]

Originally Posted by Puzzle-Peter

I do enjoy those long tests, so I started a sieve for n=2^24 to 2^26.

Send me that file, I´ll sieve up to 25e15.
Either in my "blog" or via the mail I pm´ed you.

[VBCurtis]

Looks like fun- I'll donate at least a haswell-core-month to LLR when the sieve file is ready.

[Puzzle-Peter]

Originally Posted by MisterBitcoin

Send me that file, I´ll sieve up to 25e15.
Either in my "blog" or via the mail I pm´ed you.

Ist this deep enough? You were planning to go to 25e15 for N=25M. This sieve file goes more than twice as high.

[MisterBitcoin]

Originally Posted by Puzzle-Peter

Ist this deep enough? You were planning to go to 25e15 for N=25M. This sieve file goes more than twice as high.

I can also extend the search limit. Since it´s done with yoyo@home I can go as high as you´d like to have it. :)
You might also ask PG, so far I know they did some sieve files with way deeper P´s than yoyo@home would possibly do.

[gd_barnes]

I think you guys are talking about two different things. This communication confuses me so I'm wondering if you guys are understanding one another. :-)

Peter, I believe that you are saying that a larger n-range needs to be sieved. I see that you have started on n=2^24-2^26 (n=16777216 to 67108864). That range vs. the previously suggested range by Bitcoin of n=2^24 to n=25M.

Bitcoin, I believe that you are saying that you can sieve much more deeply as is needed.

...so two different lines of communication.

I do not think that Primegrid would have a sieve file for this n-range. To the best of my knowledge they have only sieved all k<=10000 for n=9M-12M to P=20P (20e15). [They've also done all n<9M to deeper sieve depths.]

To clear things up, Peter, I would suggest that you just send Bitcoin your entire sieve file and he can have Yoyo sieve as deeply as they can. I have no idea what that limit should be but I feel like P=25P (25e15) as suggested by Bitcoin would be a good starting point since it's just low weight even-n.

One more thing to avoid some previous confusion: Please make sure that you guys are only sieving even-n. :-)

This may possibly be the highest n-range ever sieved or factored outside of GIMPS.