Lucky gmp-ecm curve...

[WraithX]

I just found a 41 digit factor using a B1=1e6! I was hoping it might make it onto the Top list for 2009, but I see the last entry is already at 42 digits. Oh well, Just wanted to share with everyone.

Has anyone else had a lucky curve, where you found a factor that is "many" digits above what was expected?

echo "10^121-8363" | ./ecm -sigma 242376148 1000000
GMP-ECM 6.1.2 [powered by GMP 4.2.1] [ECM]
Input number is 10^121-8363 (121 digits)
Using B1=1000000, B2=1045563762, polynomial Dickson(6), sigma=242376148
Step 1 took 13469ms
Step 2 took 7797ms
********** Factor found in step 2: 57315926928065111052544509749282072073751
Found probable prime factor of 41 digits: 57315926928065111052544509749282072073751
Composite cofactor (10^121-8363)/57315926928065111052544509749282072073751 has 81 digits

[Batalov]

The probability of a p41 with 904 B1=1e6 curves is ~0.1. Not very rare.

If the c81 splits p41.p41, then I'd say you had something (a 3-brilliant split).

Ah. Alas, no...
Using B1=1000000, B2=1045563762, polynomial Dickson(6), sigma=161802477
Step 1 took 1828ms
********** Factor found in step 1: 1766409430355415794746950731
Found probable prime factor of 28 digits: 1766409430355415794746950731
Probable prime cofactor ((10^121-8363)/57315926928065111052544509749282072073751)/1766409430355415794746950731 has 53 digits

[henryzz]

Originally Posted by WraithX

I just found a 41 digit factor using a B1=1e6! I was hoping it might make it onto the Top list for 2009, but I see the last entry is already at 42 digits. Oh well, Just wanted to share with everyone.

Has anyone else had a lucky curve, where you found a factor that is "many" digits above what was expected?

echo "10^121-8363" | ./ecm -sigma 242376148 1000000
GMP-ECM 6.1.2 [powered by GMP 4.2.1] [ECM]
Input number is 10^121-8363 (121 digits)
Using B1=1000000, B2=1045563762, polynomial Dickson(6), sigma=242376148
Step 1 took 13469ms
Step 2 took 7797ms
********** Factor found in step 2: 57315926928065111052544509749282072073751
Found probable prime factor of 41 digits: 57315926928065111052544509749282072073751
Composite cofactor (10^121-8363)/57315926928065111052544509749282072073751 has 81 digits

why are you still using 6.1.2

[WraithX]

Originally Posted by henryzz

why are you still using 6.1.2

Because I have made additions to it that are useful in my search for brilliant numbers. I actually call my binary 6.1.2.1. But I haven't changed the version number that is output by the program, so 6.1.2 is what shows up.

I should probably diff this against 6.2.1, and see if it will apply cleanly. I made modifications to 6.0.1, and reimplemented those in 6.1, and then reimplented those in 6.1.2. (back before I knew a tool like diff existed, still not sure if it'll work) I didn't know how fast gmp-ecm was going to keep getting updates, so I just stopped at 6.1.2 since it worked well for me.

[MatWur-S530113]

Hello,

this is a curve 11 digits above the nominal size of 30 digits:

 
(08/12/25)
GMP-ECM 6.1.3 [powered by GMP 4.2.2] [ECM]
Input number is ((367#)+((397#)/(367#)))/854170501969645699246152769 (122 digits)
Using B1=250000, B2=128992510, polynomial Dickson(3), sigma=280297040
Step 1 took 3235ms
Step 2 took 1687ms
********** Factor found in step 2: 12277223854594410567719074755920634852313
Found probable prime factor of 41 digits: 12277223854594410567719074755920634852313
Probable prime cofactor (((367#)+((397#)/(367#)))/854170501969645699246152769)/12277223854594410567719074755920634852313 has 82 digits

best regards,

Matthias