please raise limit on p factors II

[Moo_the_cow]

Louie, could you please raise the 2^64 limit on factors? I tried to
submit a factor > 2^64,

552467210540360947721 | 4847*2^304983+1

but the verifier won't accept it.
I know that it won't save a secret test, but as of today (and probably the next week or so) it will save a supersecret test.

[jjjjL]

the 2^64 limit is not arbitrary. the factors are stored as 64-bit integers.

i may be able to make a new page to submit those factors, but what do you think the limit should be on those? i don't want people submitting prime (or composite) cofactors for numbers with small factors. Joe has tried to do this in the past. i'm sure others would consider doing it if there were no limit.

somone come up with an intellegent limit and i'll make a new script.

-Louie

[ceselb]

Well, you could use that program I found in the P-1 factorer thread. Takes almost no time at all to factor the number (and the p-1) submitted.

Then you don't have to worry about a limit at all.

[jjjjL]

well look at 25350293797121 | 4847*2^3949551+1

if someone divided 4847*2^3949551+1 by 25350293797121, they could submit that number for the same k/n even though it likely isn't prime nor is it the smallest factor.

i'm sure that program is fast for small numbers, but i doubt it could handle several hundred or thousands of digits.

so there needs to be some limit. what would be good?

-Louie

[ceselb]

Well, < 1 seconds reveal

C:\>bfactor 25350293797121
25350293797121

C:\>bfactor 25350293797120
2
2
2
2
2
2
2
2
5
19804917029

C:\>

According to the readme it can handle 600 digits, but I really doubt anyone finding a factor that large. You do check for duplicates, right?

[jjjjL]

i'm not worried about people submitting:
25350293797121 | 4847*2^3949551+1

i'm worried about someone submitting:
(4847*2^3949551+1/25350293797121) | 4847*2^3949551+1

you could suggest only accepting the lowest known factor for each pair, but it would only take someone discovering a new factor first and then only reporting the cofactor.

it may be able to factor those numbers sometimes but the point is i don't want to waste the server's cputime figuring that stuff out.

-Louie

[ceselb]

well, 2^256 should be a safe bet then. 2^k gets very big, so you'd have to do a lot of work to do it even with such a high limit.

If anyone finds a factor of more than 77 digits, I'm sure we can submit that in an email or something.

just drop factors n<supersecret or something.

[jjjjL]

http://www.seventeenorbust.com/largesieve/

it will save the factors and prevent future tests. however, it has to save to a different mysql table so these numbers will not show up in the regular result.txt dumps.

-Louie

[wblipp]

so there needs to be some limit. what would be good?

-Louie

The World Record for P-1 Factoring is 47 digits. For ECM factoring it is 54 digits. Something in this range should be high enough.

[Joe O]

Originally posted by jjjjL
i don't want people submitting prime (or composite) cofactors for numbers with small factors. Joe has tried to do this in the past. i'm sure others would consider doing it if there were no limit.
-Louie

Louie, I have never tried to submit a composite factor, or the larger prime factors of the numbers that I have completely factored. Also I was working on the small n when we were sieving 101 < n < 3m and trying to raise the lower range value to make the lower range sieving more efficient.

I don't know what a reasonable value is for the limit. I have a lot of factors around 2^80 (24 digits) and 1 or 2 around 2^170 (51 digits). In each case these are the smallest factors for their expressions, the other factor is slightly larger. These were very difficult to find because both factors were the same size. Im working on one with ECM that will be around 53 digits with its cofactor 53 or 54 digits. I 'll have to look to see if the 51 digit factor was found by P-1, or ECM. I might have a record.

[smh]

These were very difficult to find because both factors were the same size. Im working on one with ECM that will be around 53 digits with its cofactor 53 or 54 digits.

Joe,

Which number are you trying to factor? 50+ factors are hard to find by ECM, but if the number isn't to large, SNFS (or siqs/mpqs) might be able factor this number faster.

What bounds are you currentely using for ECM?

[wblipp]

Originally posted by wblipp
The World Record for P-1 Factoring is 47 digits. For ECM factoring it is 54 digits. Something in this range should be high enough.

The record for ECM just increased to 57 digits