mersenneforum.org > Data Thinking out loud about getting under 20M unfactored exponents
 Register FAQ Search Today's Posts Mark Forums Read

 2021-10-01, 14:18 #782 chalsall If I May     "Chris Halsall" Sep 2002 Barbados 2×5,023 Posts BOINC Users... Update your CA file(s)! For anyone working with Reb's SRBase doing TF'ing, or anyone using BOINC in general... It appears this is one of the domains where Let's Encrypt expiry of a root cert is affecting clients. Please see this thread explaining how to fix this. And for any Sys Admins out there... Be sure to update your "ca-certificates" package on your servers. This doesn't affect any (common) clients hitting your websites, but it /can/ affect server-side IPC going out to other sites.
2021-10-01, 17:12   #783
masser

Jul 2003
Behind BB

3×587 Posts

Quote:
 Originally Posted by petrw1 So he is talking about ECM ... does that make a difference?
He was pushing ECM as an alternative to deeper P-1. I don't think it makes a difference for the point about the conditional probability of finding a factor with a deeper P-1, given an earlier P-1 attempt. When a lot of ECM has been run on a candidate, that does have some effect on the probability of subsequent P-1 attempts, but that might be tricky to quantify.

The probabilities associated with ECM curves work differently than the P-1 probabilities. With P-1 you're searching the same number space: Factors F, with F-1 smooth. With ECM, each curve should be a different space.

2021-10-01, 18:01   #784
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

23·5·112 Posts

Quote:
 Originally Posted by masser He was pushing ECM as an alternative to deeper P-1. I don't think it makes a difference for the point about the conditional probability of finding a factor with a deeper P-1, given an earlier P-1 attempt. When a lot of ECM has been run on a candidate, that does have some effect on the probability of subsequent P-1 attempts, but that might be tricky to quantify. The probabilities associated with ECM curves work differently than the P-1 probabilities. With P-1 you're searching the same number space: Factors F, with F-1 smooth. With ECM, each curve should be a different space.
Oh, wait .... ding ding ... my light just went on.
(1- Old-Prob) is only slightly less than 1 with the typical old-probs we have here: 2-3%.
Even 1-3% is still 0.97.

This says I can expect slightly better success than new-old.

In all my attempts so far (well over 50,000) I have been choosing new Bounds based simply on New-Prob - Old-Prob.
And I have been consistently exceeding these estimates (though not by a lot; maybe by 0.1% - 0.2%).
If it had been using the RDS formula then my results would be that much more closer to this.

Old-Prob 2%; New-Prob 5%
Using 5%-2% I was expecting a 3% success rate.
Using (5%-2%)/(1-2% ) = 3.06%

Thanks

2021-10-02, 20:40   #785
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

23×5×112 Posts

Quote:
 Originally Posted by De Wandelaar Yes, +/- 150 to go, work for 6-7 days; After 85 days working on 30.5, only 17 factors found until now (but I came after a first "normal" P-1 wave) I'll take 34.4.
Thanks...do you have any problem with me helping out here.
We got any extra unanticipated boost of TF in 3xM; your range is the only one remaining.
If you plan to start at the top again, I'll start at the bottom.

Wayne

2021-10-02, 21:13   #786
De Wandelaar

"Yves"
Jul 2017
Belgium

2×3×13 Posts

Quote:
 Originally Posted by petrw1 Thanks...do you have any problem with me helping out here. We got any extra unanticipated boost of TF in 3xM; your range is the only one remaining. If you plan to start at the top again, I'll start at the bottom. Wayne
No problem, Wayne.
But I would prefer you to start at the top (my worktodo file is already populated).
Yves

2021-10-02, 21:26   #787
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

23×5×112 Posts

Quote:
 Originally Posted by De Wandelaar No problem, Wayne. But I would prefer you to start at the top (my worktodo file is already populated). Yves
Chris: Can you make this happen.
I'm about 9 days from there.

A good start might be to curate the top 1000 for me.

Wayne

Last fiddled with by petrw1 on 2021-10-02 at 21:28

2021-10-02, 23:52   #788
chalsall
If I May

"Chris Halsall"
Sep 2002

2×5,023 Posts

Quote:
 Originally Posted by petrw1 Chris: Can you make this happen.
Yes.

2021-10-03, 18:50   #789
petrw1
1976 Toyota Corona years forever!

"Wayne"
Nov 2006

23×5×112 Posts

Quote:
 Originally Posted by De Wandelaar No problem, Wayne. But I would prefer you to start at the top (my worktodo file is already populated). Yves
Just to clarify: "start at the top" means highest?

2021-10-04, 05:28   #790
De Wandelaar

"Yves"
Jul 2017
Belgium

2·3·13 Posts

Quote:
 Originally Posted by petrw1 Just to clarify: "start at the top" means highest?
Sorry, Im meant from the bottom thus from the lowest (and I wrote from the top).
If this is a problem for you, it is yet possible for me to adapt my worktodo.

2021-10-04, 13:58   #791
chalsall
If I May

"Chris Halsall"
Sep 2002

2×5,023 Posts

Quote:
 Originally Posted by De Wandelaar If this is a problem for you, it is yet possible for me to adapt my worktodo.
OK... Have you started "working" 34.4M yet? Because Wayne has already completed 6, and has 34 assigned to him "top-down".

I could release what hasn't yet been assigned, and bring in the candidates from the bottom instead.

I'm not going to make any changes until you two come to convergence, and then please tell me what you would like to have happen.

2021-10-04, 14:58   #792
De Wandelaar

"Yves"
Jul 2017
Belgium

2×3×13 Posts

Quote:
 Originally Posted by chalsall OK... Have you started "working" 34.4M yet? Because Wayne has already completed 6, and has 34 assigned to him "top-down". I could release what hasn't yet been assigned, and bring in the candidates from the bottom instead. I'm not going to make any changes until you two come to convergence, and then please tell me what you would like to have happen.

I have not yet started, I'm finishing the 30.5 range.
Since Wayne is working top-down, I'll work bottom-up.
Again, sorry for the disturbance.
Yves

 Similar Threads Thread Thread Starter Forum Replies Last Post jschwar313 GPU to 72 3 2016-01-31 00:50 Batalov Factoring 6 2011-12-27 22:40 jasong jasong 1 2008-11-11 09:43 devarajkandadai Math 4 2007-07-25 03:01 WraithX GMP-ECM 1 2006-03-19 22:16

All times are UTC. The time now is 00:27.

Fri Dec 3 00:27:28 UTC 2021 up 132 days, 18:56, 1 user, load averages: 1.27, 1.39, 1.35