Factoring

bayanne · 2002-11-16, 11:35

Quote:

I wonder whether it would be worth anyone's while following on from a trial factoring test with a p-1 testing of the same exponent. Certainly those who have dedicated P4s to the task (along with a good slug of memory) may find their strike rate in finding factors might increase quite a bit.

Has anyone actually tried doing this, or are there good reasons not to?

Ok, I did try this out and no I didn't find a factor.

However what it did make me wonder is if there is a mechanism in place, such that if a factor is found under normal trial-factoring, that this would then stop the p-1 factoring from then taking place?

George, can you advise .....

Deamiter · 2002-11-16, 14:58

just out of curiosity, how can you automatically get the program to follow factoring with a p-1 test?

bayanne · 2002-11-16, 15:01

Factor=20346577,58
Pfactor=20346577,66,0

However this is not automatic. You have to manually place the Pfactor command into the worktodo.ini file. :)

Deamiter · 2002-11-16, 21:06

right. that makes sense. Thanks a lot.

Now I'd assume that a successful factoring would write something unique to the results file. Seems to me that it wouldn't be too difficult to write up a script to delete all mention of 'xxxxxx' in the worktodo file if the expression "xxxxxx is not prime" (or whatever) is found in the results file.

binarydigits · 2002-11-17, 06:25

If the client does a "send new expected completion dates to server" after finding the trial factor but before starting the P-1, it will get an "error 11 exponent already tested" and remove the "pfactor=" from the worktodo. But it would undoubtedly have already started the P-1 before checking in (as it does not usually do a checkin when sending results), and it will not delete a line from worktodo if the work has already started.

You could stagger them in the worktodo as:
Factor=xxx,58
Factor=yyy,58
Pfactor=xxx,66,0
Factor=zzz,58
Pfactor=yyy,66,0
etc., and set "days between sending new end dates" to 1. Then it should take care of itself without having to worry about a script.

bayanne · 2002-11-17, 08:00

Yes, that's a good to way to go 8)

Certainly it should increase the strike rate in finding factors ....

Deamiter · 2002-11-17, 22:44

I don't know. Does straight factoring, or P-1 find more factors per time alloted. Seems to me that you'd be more likely to find a factor in basic factoring than in p-1, but that'd just intuition.

philmoore · 2002-11-17, 23:30

According to George, it actually makes sense to do all but the final one or two bits of trial factoring first, then do P-1 factoring, and then finish the trial factoring. So the answer depends upon how far you have gotten in the trial factoring process.

2002-11-17, 23:30	#8
philmoore "Phil" Sep 2002 Tracktown, U.S.A. 10001100001₂ Posts	P-1 vs. trial factoring According to George, it actually makes sense to do all but the final one or two bits of trial factoring first, then do P-1 factoring, and then finish the trial factoring. So the answer depends upon how far you have gotten in the trial factoring process.

2002-11-16, 14:58	#2
Deamiter Sep 2002 165₈ Posts	just out of curiosity, how can you automatically get the program to follow factoring with a p-1 test?

2002-11-16, 15:01	#3
bayanne "Tony Gott" Aug 2002 Yell, Shetland, UK 373₁₀ Posts	Factor=20346577,58 Pfactor=20346577,66,0 However this is not automatic. You have to manually place the Pfactor command into the worktodo.ini file. :)

2002-11-16, 21:06	#4
Deamiter Sep 2002 165₈ Posts	right. that makes sense. Thanks a lot. Now I'd assume that a successful factoring would write something unique to the results file. Seems to me that it wouldn't be too difficult to write up a script to delete all mention of 'xxxxxx' in the worktodo file if the expression "xxxxxx is not prime" (or whatever) is found in the results file.

2002-11-17, 06:25	#5
binarydigits Aug 2002 2²·13 Posts	If the client does a "send new expected completion dates to server" after finding the trial factor but before starting the P-1, it will get an "error 11 exponent already tested" and remove the "pfactor=" from the worktodo. But it would undoubtedly have already started the P-1 before checking in (as it does not usually do a checkin when sending results), and it will not delete a line from worktodo if the work has already started. You could stagger them in the worktodo as: Factor=xxx,58 Factor=yyy,58 Pfactor=xxx,66,0 Factor=zzz,58 Pfactor=yyy,66,0 etc., and set "days between sending new end dates" to 1. Then it should take care of itself without having to worry about a script.

2002-11-17, 08:00	#6
bayanne "Tony Gott" Aug 2002 Yell, Shetland, UK 373 Posts	Yes, that's a good to way to go 8) Certainly it should increase the strike rate in finding factors ....

2002-11-17, 22:44	#7
Deamiter Sep 2002 3²·13 Posts	I don't know. Does straight factoring, or P-1 find more factors per time alloted. Seems to me that you'd be more likely to find a factor in basic factoring than in p-1, but that'd just intuition.