Riesel base 16 - team drive #2

[mdettweiler]

Originally Posted by gd_barnes

Not a problem. I knew LLRnet was new at the time. I would foward you the entire sieved file for n=110K-200K except for one issue. I remove k's as we find primes but not k's from the LLRnet server files. I only remove them from k's on files that are ABOVE the highest range reserved by the LLRnet server files.

Carlos initially indicated that it would be too big of a hassle to have a rally here due to have to remove k's where primes were found. I responded that I thought it was more hassle than what it was worth to remove k's from LLRnet files and that it wouldn't be necessary for a rally. After all, if we keep a k in there for a little too long, we might find a 2nd top-5000 prime on it! So we have decided to not mess with it, at least at these lower n-ranges. Now, if n>1M base 2, that may be a different story!

Okay, that sounds good--that makes things MUCH easier on my end when I compare results with input files!

I just thought of something. CRUS is certainly not the first Conjectures project to use LLRnet. Both base 5 and the prime Sierp project deal with it too. You might ask Masser at base 5 how they deal with the removal of k's while still matching up the original sieved files with results files. I'm wondering if there is even any matching done on the larger projects like RieselSieve and SOB. Perhaps not and that is part of why they end up with missing primes.

Big projects such as those use MySQL databases for their LLRnet servers, so they don't need to worry about comparing results files and input files, since that is essentially done automatically.

And because I know you'll ask, no, it probably wouldn't be possible for small projects like CRUS and NPLB to switch over to a completely MySQL-based system easily and quickly.

Anon

[gd_barnes]

Originally Posted by Anonymous

Okay, that sounds good--that makes things MUCH easier on my end when I compare results with input files!


Big projects such as those use MySQL databases for their LLRnet servers, so they don't need to worry about comparing results files and input files, since that is essentially done automatically.

And because I know you'll ask, no, it probably wouldn't be possible for small projects like CRUS and NPLB to switch over to a completely MySQL-based system easily and quickly.

Anon

Interestingly, I would not have asked about a MySQL database. But if I did, it would be off in the future and it would be for THIS project, not NPLB. Where one is needed the most is on a project where we know we have to search ridiculously deep before the project even gets a little off the ground.

Imagine what we'll have to do here just for Sierp base 6! Now take a look at base 3 and then there's base 16 on both sides. Even though there are fewer k's left, I anticipate that we'd have to search further to prove the Sierp conjecture than SOB or RieselSieve will, simply because there's only 1/4th as many n-values base 2 'available' for base 16.

As the above demonstrates, there will come a time on CRUS where there will be certain bases that we have to have a MySQL database set up for but that is long off. This project is theoretically exponentially larger than SOB and RieselSieve combined. I would expect it or it's successor project(s) to still be operating over 100 years from now even if bases are not expanded beyond 32 and power of 2 up to 256.

I expect to expand bases on this up to 100 perhaps sometime in 2009 (and maybe even powers of 2 bases to 1024 or 4096 depending on what the next 'interesting' base is). There's already one person that contacted me via Email who may be searching higher bases now or at least determining their lowest conjectured values with numeric covering sets.


Gary

[mdettweiler]

Originally Posted by gd_barnes

Interestingly, I would not have asked about a MySQL database. But if I did, it would be off in the future and it would be for THIS project, not NPLB. Where one is needed the most is on a project where we know we have to search ridiculously deep before the project even gets a little off the ground.

Imagine what we'll have to do here just for Sierp base 6! Now take a look at base 3 and then there's base 16 on both sides. Even though there are fewer k's left, I anticipate that we'd have to search further to prove the Sierp conjecture than SOB or RieselSieve will, simply because there's only 1/4th as many n-values base 2 'available' for base 16.

As the above demonstrates, there will come a time on CRUS where there will be certain bases that we have to have a MySQL database set up for but that is long off. This project is theoretically exponentially larger than SOB and RieselSieve combined. I would expect it or it's successor project(s) to still be operating over 100 years from now even if bases are not expanded beyond 32 and power of 2 up to 256.

I expect to expand bases on this up to 100 perhaps sometime in 2009 (and maybe even powers of 2 bases to 1024 or 4096 depending on what the next 'interesting' base is). There's already one person that contacted me via Email who may be searching higher bases now or at least determining their lowest conjectured values with numeric covering sets.


Gary

You know, considering the scale of this effort, who knows, maybe once Riesel Sieve and SoB prove their respective conjectures, they might want to team up with CRUS to help us get our conjectures proven so much faster.

Just think of all the resources we'd have--not least of which would be an already-set-up BOINC server with lots of active users!

Anon

[gd_barnes]

Testing has been completed to n=110K on LLRnet. Verification of all k/n pairs is now in progress.

[mdettweiler]

Here's the lresults for LLRnet's range 100K-110K (already marked as complete).

[gd_barnes]

Originally Posted by Anonymous

Here's the lresults for LLRnet's range 100K-110K (already marked as complete).

Thanks for finishing that up Anon!

Meanwhile:

n=110K-114K complete; no primes

reserving n=114K-118K (2 files)

[gd_barnes]

reserving n=118K-122K

[gd_barnes]

n=118K-122K complete; no primes

[VBCurtis]

I'm not sure where to put this update:
k=443 riesel is tested to 740k base 2, 185k base 16. LLR is active to 1.2M, sieving active on 1.2M-3M base 2.

-Curtis

[gd_barnes]

Originally Posted by VBCurtis

I'm not sure where to put this update:
k=443 riesel is tested to 740k base 2, 185k base 16. LLR is active to 1.2M, sieving active on 1.2M-3M base 2.

-Curtis

Thanks for the update Curtis. This is as good a place as any to report it. It can also be reported in the general reservations/statuses thread since the k is not in the team drive.

This also puts Riesel base 256 k=7088 at n=92.5K. I have both the base 16 and base 256 k's reserved for you.

It seems doubtful that we'll find a prime for Riesel base 256 k=7088 in our lifetime, but who knows? We might get lucky.


Gary

[gd_barnes]

Reserving n=122K-124K