Factoring 43rd Term of Euclid-Mullin sequence

[grandpascorpion]

150 more for a total of 4700 at B1=11M

[ValerieVonck]

Originally Posted by grandpascorpion

Cedric,

Does the input indicate it ran 200 curves? The input file needs to be in Unix format. I bumped into this problem myself.

Folks,

I'll try to close out the B1=11000000 factoring

To be honest: I do not know this.
I have put my concerns into the software section of the mersenneforum.
I have also tried to read everything on ecm what I could find.
I never found any complete tutorial, example.

So please ignore my results...

Best of luck with your factoring project.

Regards
Cedric Vonck

[patrik]

Normal output could (if you don't find any factors) look like:

GMP-ECM 5.0.3 [powered by GMP 4.1.4] [ECM]
Input number is 2784908410762794071887414394815651793559267258537102016323316206
42982062389901741890579963524423782637435949041666525000702723914662388812510545
494307250950777886431051612811386531 (180 digits)
Using B1=11000000, B2=25577181640, polynomial Dickson(12), sigma=1058443219
Step 1 took 338410ms
Step 2 took 231680ms
Run 2 out of 1000:
Using B1=11000000, B2=25577181640, polynomial Dickson(12), sigma=762516686
Step 1 took 338160ms
Step 2 took 233180ms
Run 3 out of 1000:
[---]
Run 999 out of 1000:
Using B1=11000000, B2=25577181640, polynomial Dickson(12), sigma=252573442
Step 1 took 334870ms
Step 2 took 230990ms
Run 1000 out of 1000:
Using B1=11000000, B2=25577181640, polynomial Dickson(12), sigma=2738065587
Step 1 took 334500ms
Step 2 took 231170ms

Finding a factor (for another number, of course) could look like this:

Step 2 took 85320ms
********** Factor found in step 2: 310280425886984887694911580049011296782214691
8598407
Found probable prime factor of 52 digits: 31028042588698488769491158004901129678
22146918598407
Composite cofactor 2115042966390179978303718113827180116042610311426696576586244
13023205204173795594453958188167116829523236801468161403289317068522085489 has 1
35 digits

[ValerieVonck]

Never saw this before....

my batch file look like this:

ecm5 11e6 25577181640 -c 100 -save save.txt -resume resume.txt < factor.txt

factor.txt contains the factor. I stripped the CR (carriage return)
Nothing more nothing less.

[akruppa]

Command line options such as -c, -save etc must be placed before the B1, B2 values. Gmp-ecm stops parsing the command line after B2. The options -save and -resume have no meaning for what you're trying to do anyways, you can use them to do part of stage 1 now and the rest later, or import stage 1 residues from other programs. Since you want to do complete curves in one go, you don't really want saving or resuming of residues.

Alex

[ValerieVonck]

Originally Posted by akruppa

Command line options such as -c, -save etc must be placed before the B1, B2 values. Gmp-ecm stops parsing the command line after B2. The options -save and -resume have no meaning for what you're trying to do anyways, you can use them to do part of stage 1 now and the rest later, or import stage 1 residues from other programs. Since you want to do complete curves in one go, you don't really want saving or resuming of residues.

Alex

So my _exact_ syntax should be:

ecm5 -c 100 11e6 25577181640 < factor.txt
?????

[patrik]

Originally Posted by CedricVonck

So my _exact_ syntax should be:

ecm5 -c 100 11e6 25577181640 < factor.txt
?????

Yes, except that you could skip 25577181640, since that is the default value. If you want to save your output in a file you could type:

ecm5 -c 100 11e6 < factor.txt > out.txt

I don't know how the -c option works when you find a factor. Does it stop then? Otherwise your found factor could scroll above the top of the window if you don't send the output to a file. (Earlier I have only used gmp-ecm on numbers of the form 2^n-1, where Prime95 is faster for step 1, and then I don't have to use -c when I run step 2 using gmp-ecm. That is an example of a case when -resume is used.)

[akruppa]

Originally Posted by patrik

I don't know how the -c option works when you find a factor. Does it stop then?

It stops if the cofactor is a probable prime, but continues working on composite cofactors.

Alex

[ValerieVonck]

Will try....

Sorry guys, problem remains.
I leaving this "factoring" project.....
Good results...

[patrik]

800 curves with B1=43e6, default B2 (=178426462988).

I have one more idea that might explain how ecm behaved for CedricVonck, and that is if LF or CR LF is missing completely in factor.txt. (Both these two works for me.) I.e. if the size is 180 bytes it won't work. The size should be 181 or 182 bytes for this 180 digit factor.

Windows notepad works a bit different regarding this compared to e.g. the UNIX program vi. In notepad, if you type a line without hitting return in the end, no CR LF will be appended, but in vi you get a LF. If you do hit return and the end of a notepad line you will get CR LF, while in vi you then would get two lines (terminated with LF).

[Mystwalker]

I also had some hard time figuring it out. Here is comes:

command line:
ecm5 -c 100 43e6 < factor.txt

factor.txt has to contain the factor, followed by a \n
Then, you have to start a new line (inside the file).

Now it works for me...