Bases 33-100 reservations/statuses/primes

[Mini-Geek]

Originally Posted by gd_barnes

[(13.5-9) / 9 = 50%]

Wouldn't it be easier to calculate 13.5/9-1=50%?

Originally Posted by mdettweiler

Just today I discovered how useful the stop-on-prime option is, compared to Phrot's comparatively simpler stop-on-prime setting which stops the whole program on a prime (not just the k that found the prime).

I didn't know PFGW had a stop-on-prime option! (I found the documentation for it now, and I'm not surprised I didn't know before...shouldn't it be more visible than a comment for the ABC file format?) What a handy feature, wish I knew about it when I was running that base 3 work the other day.
Is there a way to automatically verify that a PRP is prime?

[mdettweiler]

Originally Posted by Mini-Geek

Is there a way to automatically verify that a PRP is prime?

Not directly, though PRPnet will have PFGW verify any PRPs it finds.

[Mini-Geek]

Originally Posted by mdettweiler

Not directly, though PRPnet will have PFGW verify any PRPs it finds.

Ok, thanks. Good enough for me, it's not a big deal to not verify immediately. I'm sure the odds are astronomical that any of them are actually composite (at least for the numbers from the base 3 work I'm running right now).

I got to n=26400 on my new CRUS base 3 reservation with no primes, which I thought was very odd, because I'd expect over 5 primes by that time and the odds of having 0 was under 1%, when I realized I had accidentally set it to run base 2 for those k and n's by mistake.
Since those numbers weren't sieved, I should've expected about 0.2 primes for the numbers I processed, which is much more in line with what I found: 0.
I've restarted it now with the correct base. I'm just glad I noticed it after a few hours instead of after a few days (when it would finish).

[KEP]

Just a quick question, maybe a rather stupid one, but is k*b^0+/-1 considered valid primes when dealing with conjectures or is a k first considered to be primed if n=>1? Just need to ask, since it can mean a difference of 25371 k's for Sierp. base 63. Hope to here from anyone who has the prober knowledge

Regards

Kenneth

[gd_barnes]

Originally Posted by KEP

Just a quick question, maybe a rather stupid one, but is k*b^0+/-1 considered valid primes when dealing with conjectures or is a k first considered to be primed if n=>1? Just need to ask, since it can mean a difference of 25371 k's for Sierp. base 63. Hope to here from anyone who has the prober knowledge

Regards

Kenneth

n=0 is not considered. Otherwise many specific k's for all bases could be easily eliminated. Example: k=4 would never remain for either the Riesel or Sierp side because 4*b^0-1 = 3 and 4*b^0+1 = 5. Both 3 and 5 are prime. n must be >= 1. That is shown on the web pages. There has to be an "effect" of the base. If n=0, the base would have no effect.

[gd_barnes]

Originally Posted by mdettweiler

Not directly, though PRPnet will have PFGW verify any PRPs it finds.

I don't think this is clear. You have to define "directly".

Tim, PFGW can prove primes when not running through PRPnet. You just have to "re-run" the PRP's that it finds back through it when you are done with the -tp option for Riesel PRP's and -t option for Sierp PRP's. It's slower to do the entire sieve file with one of those options so it's best to do a 2nd run on only the PRP's.

Yes, always use the stop on prime option in PFGW for the conjectures now. There is clearly no way to do any manual CRUS testing any faster at this point in time (that I am aware of). PFGW is a very powerful program and it is by far my favorite now that it is so fast. If you go through the various README and other helpful files, you'll see how powerful it is.


Gary

[rogue]

Originally Posted by gd_barnes

I don't think this is clear. You have to define "directly".

Tim, PFGW can prove primes when not running through PRPnet. You just have to "re-run" the PRP's that it finds back through it when you are done with the -tp option for Riesel PRP's and -t option for Sierp PRP's. It's slower to do the entire sieve file with one of those options so it's best to do a 2nd run on only the PRP's.

PRPNet will use PFGW to do a primality test if the number is PRP. That is handled automatically by the client. And, of course, it can send an e-mail of the find directly to whomever is administering the project.

[mdettweiler]

Originally Posted by rogue

PRPNet will use PFGW to do a primality test if the number is PRP. That is handled automatically by the client. And, of course, it can send an e-mail of the find directly to whomever is administering the project.

@Gary: yes, this is what I meant. Perhaps "automatically" would have been a better choice of word than "directly".

[mdettweiler]

Riesel base 39 k=474 is complete to n=50K. 596 was knocked out rather early in the 10K-50K range:

596*39^10649-1 is prime!

Results for 10K-50K are attached; continuing onward with 474.

[gd_barnes]

Riesel base 60 is at n=21K; 62 k's remaining; continuing to n=25K.

Sierp base 60 is complete to n=25K; 37 k's remaining; now unreserved.

New reservations:

Riesel bases 42 and 48 for n=10K-25K.

Sierp bases 52, 67, and 91 to n=25K.

These reservations should just about complete all of the "low hanging fruit" for bases <= 100; that is bases that are likely to have <= ~50 k's remaining at n=25K. Everything else not started will be a fair amount tougher likely ranging anywhere from > 50 k's remaining to millions of k's remaining at n=25K. (bases 7, 15, and 71 may fit the latter category)

After all of these are done, I'll likely start on tough Sierp base 25 up to n=25K. Having to convert base 5 primes to base 25, determine which base 5 k's remaining are the equivalent of ours, and list their converted high search limits from the base 5 project will be time consuming. But it is the last base <= 32 that is not b=2^q-1 (the really tough ones!) that has not been started yet so it'll be about time to get going with it starting in the next 2-3 months.


Gary

[mdettweiler]

Riesel base 39 k=474 is complete to n=100K, no primes. Results for 50K-100K are attached.

Releasing this base. However, I did sieve all the way up to n=150K, so I've also included a sieve file for 100K-150K in the attachment.