Sierp base 3 - mini-drive II

[Flatlander]

Originally Posted by Mini-Geek

...
This time I'm using PFGW with the automatic stop-k-on-prime feature that I just learned about.

How do I do that?

[mdettweiler]

Originally Posted by Flatlander

How do I do that?

First, you need to make sure the file is in ABC format. A NewPGen file can actually be converted to ABC quite easily without the need for conversion tools: for example on Sierp. base 3, you'd simply change the first like of the file to "ABC $a*3^$b+1" rather than the NewPGen header (the thing with all the colons in it).

Next, you need to specify the "number_primes" option as a comment in the file. To do this, you need to change the ABC header to something like this: "ABC $a*3^$b+1 // {number_primes,$a,1}". That tells PFGW that it should stop searching a particular $a value (that is, the k) if 1 prime is found for that value.

At this point, you're all ready to run it through PFGW. Navigate to the PFGW directory through the command prompt and run the following command (or, use WinPFGW and give it the following command line):

pfgw -l name-of-sieve-file.txt

Note that the -l is quite important--without it, PFGW will not log its composite results to a file (in this case, pfgw.out; to change the name, specify the -l option as something like "-lfile.txt" instead).

One note on stopping/restarting with the number_primes option: if you stop and restart, PFGW will "forget" which k's it's stopped searching. It will still pick up exactly where it's left off, but it won't remember to leave out the k's which it's found a prime for prior to the restart. Thus, if you need to restart, you'll have to remove prior-primed k's from the sieve file (be sure to adjust the line # accordingly in pfgw.ini).

[kar_bon]

Originally Posted by mdettweiler

Next, you need to specify the "number_primes" option as a comment in the file. To do this, you need to change the ABC header to something like this: "ABC $a*3^$b+1 // {number_primes,$a,1}". That tells PFGW that it should stop searching a particular $a value (that is, the k) if 1 prime is found for that value.

pfgw -l name-of-sieve-file.txt

this call will test the file for PRP's, and it won't stop on finding a PRP!

i've tested this with the Sierp Base 3 file with the header

ABC $a*3^$b+1 // {number_primes,$a,1}

i tried $a or $b, the same.

WinPFGW won't stop!

[mdettweiler]

Originally Posted by kar_bon

this call will test the file for PRP's, and it won't stop on finding a PRP!

i've tested this with the Sierp Base 3 file with the header

ABC $a*3^$b+1 // {number_primes,$a,1}

i tried $a or $b, the same.

WinPFGW won't stop!

Really? It worked just fine for me when I tried it on Riesel base 39.

Edit: Possibly it doesn't work with WinPFGW? Try it with command-line PFGW and see if it works then. It *should* work with both, but it's possible there's a bug in WinPFGW.

[Flatlander]

Thanks guys!

[kar_bon]

Originally Posted by mdettweiler

Edit: Possibly it doesn't work with WinPFGW? Try it with command-line PFGW and see if it works then. It *should* work with both, but it's possible there's a bug in WinPFGW.

out in WinPFGW:

PFGW Version 20090723.Win_Dev (Beta 'caveat utilitor') [GWNUM 25.12]

Output logging to file pfgw.out
Recognized ABC Sieve file: 
ABC File Processing for at most 1 Primes

196996994*3^25147+1 is 3-PRP! (2.6909s+0.0006s)
PRP: 196853336*3^25152+1 10000/39892

found a PRP but continues!

same in pfgw.exe!

[mdettweiler]

Originally Posted by kar_bon

out in WinPFGW:

PFGW Version 20090723.Win_Dev (Beta 'caveat utilitor') [GWNUM 25.12]
 
Output logging to file pfgw.out
Recognized ABC Sieve file: 
ABC File Processing for at most 1 Primes
 
196996994*3^25147+1 is 3-PRP! (2.6909s+0.0006s)
PRP: 196853336*3^25152+1 10000/39892

found a PRP but continues!

same in pfgw.exe!

Hmm...that's weird. Mark, if you're reading this, have you got any idea why this is happening?

[gd_barnes]

I know for a fact that the stop on prime works for the new version of Linux PFGW. I haven't tried it with the new version of WinPFGW.

Max and Karsten, I think there may be a miscommunication based on the two lines that Karsten displayed.

Karsten, it should stop processing that k-value for all future occurrences in the file but should continue processing the file for all other k's. That's what's so cool about PFGW. The two lines that you displayed have different k-values. I guess my question is: Why have you showed lines for 2 different k-values? It doesn't demonstrate the problem that you appear to be talking about.

Also, please note what Max said. If you stop and restart PFGW, it won't remember what k-values it has found primes for. You'll need to manually remove them from the sieve file. Personally, I never stop it. I just pick the range that I want and let it rip until it's done. But you have to plan your work amount carefully ahead of time.


Gary

[kar_bon]

Originally Posted by gd_barnes

Max and Karsten, I think there may be a miscommunication based on the two lines that Karsten displayed.

Karsten, it should stop processing that k-value for all future occurrences in the file but should continue processing the file for all other k's.

oh.. that's quite different i thought!

if so, i was wrong. i had in mind the StopOnSuccess option from LLR/cllr when it really stops if a prime was found!

i'll check this now!

[kar_bon]

ok, pfgw stops testing the same k-value a prime/PRP was found for, not stop testing at all!

so my fault, sorry!

[gd_barnes]

I'm surprised at how little is known about PFGW. How have people been starting new bases? Manually with srsieve and LLR or Phrot? I hope not. That is hard and takes too much time!

IMHO, PFGW with the stop-on-prime option set on is the only reasonable way to start a new base. You could sieve ALL k's below the conjecture immediately and let srsieve eliminate the k's with trivial factors but you'd still have a file that you'd want LLR/Phrot/PFGW/whatever to stop processing a k once it found a prime. Else, you'd have to eliminate tons of k's manually yourself if the conjecture was big...not something that I want to do.

I'm asking because is there some other reasonable way to start a new base that's possibly good that I'm not aware of?