Any interest in some low n ECM factoring?
 

Hi,

i am doing some ECM factoring on low n.
Anyone interested to help me out a little bit?

And yes i know it will not bring the project ahead i am doing it just for fun.

Lars

Care to explain a not-mathematician what it's about and how you do it?
I've got my Pentium M 1.6 GHz laptop (not running 24/7) which (when I'm finished with S/R B5) I might consider using for other stuff, maybe this. :whistle:

Could you post a list of k and n you want to factor. I could run some curves. (what bounds you want us to run will be helpful)

Citrix


Ps: I am going to use
[url]http://www.alpertron.com.ar/ECM.HTM[/url]

I will try my best to describe it and hope that somebody else corrects all my errors.

There are several methods to find a factor for a given number.
For example we use sieving to find factors for as many numbers as possible before we do a PRP test to check if a number is prime or not.
At the moment we find factors that are 13 Digits long with sieving.

There are other methods called "p-1","p+1" and ECM.

These methods can find factors that are much longer but it takes also alot of time time compared to do a PRP test. Don't ask me about the math behind it. I just know how to set up a test.

For the fun of finding some large factors (30 digits and larger at the moment)
i started to run p-1 and ECM on all non factored k/n pairs with n less then 2000.
There are 26 n left where i found no factor less then 25 digits. (For the experts B1=50000)

For 13 of them i also finished the 30 digit level (B1=250000) and for the other 13 i should have finished that level in 2 days.

To help you can use either Prime95 or ECM6.
I am using ECM6.


Lars

Where do you get the binaries for ECM6? You only want to find the lowest factor or factorize the whole number?

Citrix

Here is a list of the values that need ECM at B1=1000000.
To complete this B1 level each of the numbers needs 950 curves.

237019*2^1006+1 : 1*10^6 | 60 : done
152267*2^1059+1 : 1*10^6 | 61 : done
168451*2^1116+1 : 1*10^6 | 60 : done
265711*2^1128+1 : 1*10^6 | 60 : done
149183*2^1157+1 : 1*10^6 | 60 : done
258317*2^1167+1
79817*2^1191+1
214519*2^1194+1
222113*2^1221+1
149183*2^1309+1
258317*2^1311+1
237019*2^1330+1
265711*2^1344+1 : 1*10^6 | 1 : done

The other 13 ( hopefully less) will follow when i have finished B1=250000.

237019*2^1006+1 is my most wanted as it is the lowest unfactored n.

@Citrix: As far as i know the the ECM you want to use is slow compared to ECM6.
I have precompiled windows version for P3,P4,Athlon and Athlon64.
( No i did not make the compilation but found the links somewhere here in one of the forums)

Lars

Could you provide me a link to the binary for p4! (or upload it or email it to me). I will try to look in the other forms.


I will start with 152267*2^1059+1

Citrix

I found the binaries under the following link:

[url]http://www.pa.uky.edu/~childers/ecm6.zip[/url]

If you want to read the thread where if found the link:

[url]http://www.mersenneforum.org/showthread.php?t=3766&page=3&pp=50[/url]


I only want to find one factor and not the complete factorisation so you should use the "-one" option.

Lars

I'll guide again
 

I've collected pieces of information around here on the forums and I've come up with [URL=http://www.geocities.com/omboohankvald/compileGMP.html]this[/URL] as a basic guide to how to compile an exe which suits your own system.

I hope it's correct...

I will return my k, I can't get the exe to work. Though I did 1 curve using the applet.

Citrix

Ill try p-1. What bounds have you used for p-1, though it seems unlikely that I will find a factor.

Citrix