Low Weight Subsequences

[geoff]

Originally Posted by michaf

I'm in for a little challenge... is there a way to create a list of those numbers?
(204394*5^n-1: n = 16*m+11 with n = 51 (mod 60): 243 terms)?

Easy with AWK. If you have a NewPGen file 204394.in:

awk '$2%60==51' 204394.in > 204394.out

will write the 51 (mod 60) terms to 204394.out.


Removing a subsequence is probably most useful when sieving just one lightweight sequence, it will have very little effect when there are 6000 other subsequences to sieve.

Note that the -pentium4 builds of sr5sieve/sr1sieve process the subsequences two at a time to make use of SSE2. If there are an odd number of subsequences then eliminating one of them will have some extra benefit.

[masser]

I think B2 should be between 14 and 18 times B1 for the P-1 step... then the times spent on stage 1 and stage 2 are about equal. It has been noted elsewhere that having an equal amount of time spent in the two stages leads to the most efficient removal of factors.

Originally Posted by michaf

First result:
204394*5^202251-1 has a factor: 159148635697417

(I am running ECM with B1=1000 and B2=100000 10 curves and P-1 with B1=10000 and B2=100000 on every number (which have not been prp'd yet))
If those parameters are way off, let me know please :>

[michaf]

Where can I find some more information on P-1 factoring.
As in, what are the chances of finding a factor of certain size, with certain B1 and B2 limits?

Google isn't my friend in this

Oh, doing B1=100000 and B2=1500000 now for P-1; about an hour per run (on the smaller numbers). But factors ARE being found :)
Doing P-1 and ECM this way is quite inefficient to get numbers eliminated, but still quite some fun:)

[tnerual]

do you have a list of all the remaining numbers
it's the week end and i have some time to play with them

[michaf]

Originally Posted by tnerual

do you have a list of all the remaining numbers
it's the week end and i have some time to play with them

I have a list at home, am at work now.
Some 222 values remained (above 170k or so, the rest is already prp'd), 4 factors were found yesterday evening.

I will send you the list sometime this evening (I hope to remember, since I have a wedding to go to first, and am not sure if I will be sober enough on return
I should be able to fit in a few seconds to send though :)

[tnerual]

you can remove this one:

204394*5^1995051-1 is not prime.

took me 9 days, 0 hours and 58 minutes to test it on one core of a centrino duo 1666 mhz

i will sen residues (with others ...) to the email address

[jasong]

How about giving these monsters to the regular server, but on a different port number?

Edit: Or how about an ecm server? Things would be automatically organized properly.