Bad Factor?

[nfortino]

I found this factor in the most recent data files (12/12/04):
24742121,3047713721183690569
k = 61589580804

3047713721183690569=53*1009*56991112462997

As 53 is clearly not of the form 2kp+1, 3047713721183690569 cannot divide M24742121. As the leading edge of factoring is at about 28.1M, I'm not sure where this faulty factor came from, and it does not appear in cleared.txt. I checked the 11/30/04 data files, and this factor was not there. Since I know factors are checked for both primality and correctness, I am a bit confused as to how this factor found its way into the data files.

[garo]

Yup! Looks like it is a bad factor. Wonder how it managed to get through.

[Prime95]

Originally Posted by garo

Yup! Looks like it is a bad factor. Wonder how it managed to get through.

Seems to be a bug in WindowsXP cut/paste from a DOS window. I used msieve to factor a composite P-1 result which worked fine. However, cutting
and pasting from a DOS window loses the first few digits. Weird.

Nice catch.

[Prime95]

Msieve v. 0.86
random seeds: 000002bc 41bf0e6b
factoring 555087078799770339769608793045705436436591887
Tue Dec 14 11:01:47 2004
using multiplier of 23
Tue Dec 14 11:01:47 2004
using sieve block of 65536
using a sieve bound of 19421 (1073 primes)
using large prime bound of 776840
restarting with 721 full and 5356 partial relations

found 1476 relations (721 full + 755 partial), need 1201
begin with 5356 relations
reduce to 1399 relations in 2 passes
attempting to read 721 full and 1399 partial relations
recovered 721 full and 1399 partial relations
recovered 620 polynomials
attempting to build 755 cycles
found 755 cycles in 1 passes
distribution of cycle lengths:
length 2 : 755
largest cycle: 2 relations
Tue Dec 14 11:01:47 2004
1073 x 1137 system, weight 22284 (avg 19.60/col)
reduce to 982 x 1046 in 3 passes
lanczos halted after 17 iterations
recovered 64 nontrivial dependencies
Tue Dec 14 11:01:47 2004
probable prime factor: 20563047713721235292639
probable prime factor: 26994397257045406913233
Tue Dec 14 11:01:47 2004

[nfortino]

Originally Posted by Prime95

Seems to be a bug in WindowsXP cut/paste from a DOS window. I used msieve to factor a composite P-1 result which worked fine. However, cutting
and pasting from a DOS window loses the first few digits. Weird.

Nice catch.

That is weird. The factor in the database is not a pure truncation, however. The last 9 digits of the false factor and the true factor are different. I assume this is because a pure truncation does not give a factor of the form 2kp+1, and some processing code fixed that defect for you.

[ET_]

Originally Posted by Prime95

Seems to be a bug in WindowsXP cut/paste from a DOS window. I used msieve to factor a composite P-1 result which worked fine. However, cutting
and pasting from a DOS window loses the first few digits. Weird.

Nice catch.

I found out that cutting and pasting from a dos window leaves part of the digits of the long factor in the next line. I need to import the screenshot into wordpad and realign the number (usually deleting the carriage return char) before using it.

Luigi

[xilman]

Originally Posted by ET_

I found out that cutting and pasting from a dos window leaves part of the digits of the long factor in the next line. I need to import the screenshot into wordpad and realign the number (usually deleting the carriage return char) before using it.

Luigi

Alternatively, and this requires more careful attention but less machine resources, C&P the number in several pieces.

You describe an irritating feature of Windows C&P. It works on what's displayed on the screen rather than the underlying semantic object. If it looks like two words on two separate lines, Windows C&P treats it as such, even if it came from a singe word. Most X11-based tools treat a single word as a single entity, whether or not it appears on one line or several.

Paul