[QUOTE=bsquared;289773]No, as you found it is not strictly necessary. But it is highly typical to find num relations > unique ideals before the filtering will succeed. It would be a better question for jasonp (or others) as to why...[/QUOTE]
It is necessary for the number of relations to exceed the number of ideals *after* singleton removal, but if you are very close to the point where that will happen then the initial dataset may have lots of ideals that will be removed immediately, when singleton removal starts.

have small trouble downloading last version of yafu, is something wrong with my internet or is [strike]softpedia[/strike] [sourceforge, problem solved, I am idiot!] down?

edit:
and by the way, I think yafu worth a sub-subforum here in the factoring subforum. It would be easier to find, and we could split this HUGE thread in the future into different sub-subjects, this thread got 50 pages already, is impossible to find some old discussion related to yafu even if we are sure is here in this thread, and at last, why should msieve or other less used :razz: programs have their own subforum and yafu not have one of the same?

any mod here around? are ye hearing this?

Cool, looks like we have a new home.

I'll see if I can organize this place a bit - but please be patient.

[QUOTE=cgy606;289717]I just got gmp-ecm 6.4 for my mac. I ran a couple hundred curves for a factor at b1=11M and I am pretty sure the program found a factor after 23 curves. However, the screen didn't display the factor. I know the program found a factor because the time required to perform step 1 and step 2 is displayed after each run and the time to perform decreases after step 23. However, the factor did not print out on the screen. Is the output save to some logfile like it is in yafu that I can access? Thanks[/QUOTE]

How much did the time decrease by? If it found a factor I believe it will always print it to the screen. They are not stored in a log file unless you redirect the output to one, AFAIK.

[QUOTE=bsquared;290206]How much did the time decrease by? If it found a factor I believe it will always print it to the screen. They are not stored in a log file unless you redirect the output to one, AFAIK.[/QUOTE]

I stand corrected. It does print out a factor in the log file when it finds one. The time per factor was reduced because another thread of ecm finished factoring another number on the machine which freed up memory. This brings up another interesting question: Given a machine (with x clock speed, y GB of ram and other parameters that are relevant to this) running ecm on one thread at a particular B1 and B2, how long does it take per curve to run an input composite of n digits. My observations have found that the larger the factor the longer it takes per curve to run. What is the functional form?

Trying to learn snfs method, I've prepared a simple ("toy") example for my training. I used a small enough number n = C107 = 2^355+9. It has two small factors and a sufficiently large composite factor, C107 = 17.2339.C103.
I've already factored the C103 by QS and GNFS/yafu methods (=P46.P57), but cann't repeat this by SNFS/yafu.

Firstly I had prepared a job-file ("snfs.job") for the "natural" polynomial x^5+9=0 and m=2^71:
[quote]
n: 73391955711682288371546268649666782105490079653384995959602842860381532034831513858240593699524021969747977
Y1: 1
Y0: -2361183241434822606848
c5: 1
c0: 9
skew: 1.0
rlim: 800000
alim: 699999
lpbr: 25
lpba: 25
mfbr: 44
mfba: 44
rlambda: 2.2
alambda: 2.2
q0: 700000
qintsize: 50000
[/quote]

Further I started sieving: yafu nfs("my n=C107") -ns -job snfs.job
... and obtained the following answer:
[quote]
nfs: commencing gnfs on C103: 184...979
poly selection mast be performed before sieving is possible
***factors found***
P2 = 17
P4 = 2339
**co-factor**
C103 = 184...979
[/quote]

As far as I understand, yafu
- excluded two small factors from C107 and obtained C103;
- compared the C103 with the number from the job-file (C107) and found a discrepancy;
- began a new job with the C103 (i.e. the existing job-file was ignored).
The work was finished because a job-file for the new number C103 was absent.

If so, can I ask yafu to run the only nfs method and forbide any preliminary analysis? I cann't find an appropriate key in documentation.

Thank you in advance,
Victor

The value for n should be the C103, not the C107.

[QUOTE=wblipp;303561]The value for n should be the C103, not the C107.[/QUOTE] Thank you. It works!

Will we see YAFU using GMP-ECM 6.4.3 anytime soon?

[URL="http://gmplib.org/#FUTURE"]GMP 5.1.0 is planned for 2012 btw.[/URL]

[QUOTE=lorgix;317714]Will we see YAFU using GMP-ECM 6.4.3 anytime soon?

[URL="http://gmplib.org/#FUTURE"]GMP 5.1.0 is planned for 2012 btw.[/URL][/QUOTE]

You are free to link to any version of GMP-ECM you like with the ecm_path option in yafu.ini. I have no plans at this point to upgrade the linked-in version of GMP-ECM because nominally it only runs for very small B1's and I wouldn't expect a noticeable difference there.

[QUOTE=lorgix;317714]Will we see YAFU using GMP-ECM 6.4.3 anytime soon?[/QUOTE]

Here's a YAFU 1.33 (for Linux x64) built with GMP-ECM 6.4.3, as attached in the other thread. (Though, as B[sup]2[/sup] just pointed out, after I'd compiled/compressed everything :razz:, YAFU uses an external GMP-ECM binary in almost all cases anyways.)