Registering DC'd P-1 assignments >10M
 

There has been scattered discussion of the problem in various threads (e.g. [url=http://www.mersenneforum.org/showthread.php?p=253956]1[/url],[url=http://www.mersenneforum.org/showthread.php?p=270847]2[/url]) but no official answer that I've seen:

[color=blue]It seems that all exponents >10M that have been double-checked are not available to register via PrimeNet. Can we please undo this lockout?[/color]

Those of us who are working in that range will continue to work in that range whether it's locked or not, but the chance of a lot of wasted duplicated work is significantly higher if we can't register the assignments as normal.

Seems to me that the assignments page should let us know (a warning) that we are doing P-1 on something that has been DC'ed.

Those numbers have already been proven as composite. Why would you want to reserve them?

[QUOTE=ixfd64;271362]Those numbers have already been proven as composite. Why would you want to reserve them?[/QUOTE]

Because he wants to find factors of them.

More specifically, because factors are a much more directly verifiable proof of compositeness than LL tests.
Otherwise, what are we doing ECM for?

[QUOTE=Christenson;271381]More specifically, because factors are a much more directly verifiable proof of compositeness than LL tests.
Otherwise, what are we doing ECM for?[/QUOTE]

To make progress towards fully factoring the numbers (though, arguably, we are so far from fully factoring any non-trivial number that large it's pointless). Or to find factors just for the factors themselves (like we search for Mersenne primes largely for the primes themselves).
While, obviously, you're right that factors are a much more directly and easily verifiable proof of compositeness, matching LL residues leave practically zero doubt. I personally see very little point in searching for factors for the sake of a more certain/verifiable proof.

There's a small contingent of us who likes to find factors for the sake of finding factors. Some search for P-1 factors where the original P-1 test was skipped or poorly done; some search for factors when one (or more) small factors are already known. Does it contribute to the advancement of the LL wavefront? No. Does it contribute to GIMPS progress? I'd say yes if you consider the 3 aims:
1) find Mersenne Primes (no)
2) prove all other Mersenne number composite (somewhat)
3) fully factor all Mersenne numbers (yes)

But the main goal is not for GIMPS credit, or fame and fortune, but because we like finding factors. At least the above applies for me, I'm sure we all have slightly different reasons for doing what we're doing.

I guess you don't really care that much about finding a new Mersenne prime then. :P

[QUOTE=ixfd64;271406]I guess you don't really care that much about finding a new Mersenne prime then. :P[/QUOTE]Not really, no. :smile:
In the last 365, PrimeNet tells me I've spent:
* P-1 = 5153 GHz-days
* TF = 4758 GHz-days (mostly on my GPU)
* ECM = 150 GHz-days
* DC = 25 GHz-days
* LL = zero

James, I just checked the primenet status ([url]http://mersenne.org/primenet/[/url]) and it looks like the line that used to divide "sub-10M" from "10M+" has become a line to divide "sub-20M" from "20M+". I haven't tried grabbing an assignment between 10-20M, but they are now available for ECM work. I suspect that means they are available for generic assignment.

I am now able to register the 10M-range exponents I'd been working on.