mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   PrimeNet (https://www.mersenneforum.org/forumdisplay.php?f=11)
-   -   Advice on choice of targets? (https://www.mersenneforum.org/showthread.php?t=27992)

WillowW 2022-08-04 15:08

Advice on choice of targets?
 
Hi all!

Another small question from a not-quite-so-newbie-as-last-time:

Since last we spoke, I finished up the remaining PrP checks on 123456xxx, and now I've been casting about for some charming numbers to work on.

I'd like to continue working in the 123xxxxxx range (yes, it's a whim, humor me), so I've been doing P-1 on numbers of the form 12345xxxx, since my little computer seems pretty happy and efficient with such tasks and it's a helpful stepping-stone for others to then do PrP.

I noticed, however, that there are 14 unverified LL and PrPs in the range 123xxxxxx, such as 123000023 („reliable“) and 123000169 („unverified“). I feel the urge to verify them, as a nice gesture to the people who made the effort to test them the first time.

[B]So my question is:[/B] Is it better for the project to do *new* PrPs, rather than finishing up the once-verified? Or does it not make really any difference?

Any way, I'd appreciate your advice, even if maybe the question isn't answerable.

Thanks! Willow


P.S. Team NerdyGirls is a welcoming place for girls who love number theory. Just saying!! :D

Uncwilly 2022-08-04 15:22

First, you should do what pleases you.

Second, if you want to run the tests to do the double checks, sure do that. The PRP on [M]123000023[/M] does not have the proof file, so it needs to be rerun in full to verify it. The LL's obviously will too. Just be sure that once you add them to your queue, that you do a manual communication to server, so you get them assigned to you and avoid anyone also trying to get them assigned.

Does, it make a difference to the project which you do first? Seeing as how there are only 14, I would say clean them up (do the DC's), then do the others in the subrange you want (doing them as PRP's that generate the proof files.) This will make it such that when the wave front hits the 123M range, everything should get verified as the PRP's are turned in and the range will be all tidy. 123M is not too far above the wave front as to be to fanciful a range to work on.

kriesel 2022-08-04 16:41

(Seconding)

LL first tests are more likely to be in error than PRP with GEC. So DC of those early would be good.

DC by PRP with proof generation (of PRP or LL first test) is the preferred method, both because it is more reliable and faster on average after considering error rates.
There's a lot of first testing occurring, and less DC effort. The wavefront of DC lags the wavefront of first testing by several years, ~48M exponent difference currently. Helping catch up on the DC backlog is very welcome. All productive efforts are welcome.

Experiment with different approaches, find what you like, & do that. Running PRP or PRP-DC with GEC is an effective way of checking the reliability of your hardware also.

WillowW 2022-08-05 07:28

Target acquired :D
 
Thanks so much for your kind answers!

As soon as the P-1 runs on 12345xxxx wrap up, I'll start in on DC'ing the 123xxxxxx tests. With any luck, I might be able to get them done by Christmas.

'Til the next small question,
Willow

PS. 322222223 is a pretty cool prime number, even if it isn't a Mersenne prime. Thanks for that!

mrh 2022-09-21 23:03

[QUOTE=Uncwilly;610799]First, you should do what pleases you.

Second, if you want to run the tests to do the double checks, sure do that. The PRP on [M]123000023[/M] does not have the proof file, so it needs to be rerun in full to verify it. .[/QUOTE]

I wonder how that mrh guy didn't upload a proof? sorry about that!

Uncwilly 2022-09-22 05:05

[QUOTE=mrh;613864]I wonder how that mrh guy didn't upload a proof? sorry about that![/QUOTE]
You are assuming mrh's gender. Also proof files might not have been a thing at the time.

WillowW 2022-09-22 06:13

[QUOTE=mrh;613864]I wonder how that mrh guy didn't upload a proof? sorry about that![/QUOTE]

Dear mrh, not to worry! I had fun doing such a lovely number as 123000023. If you have more numbers ending in 23 that might need a touch-up, send them my way!

Speaking of gender, assumptions and having fun, I'm going to go full White Queen and prepare breakfast while imagining that mrh is *definitely* (and secretly) the legendary number theoretician, the honourable Mr. Hermes Hadrian Hanahan, Habile Head of Hyperbolic Hieroglyphs at Harvard, whose only hindrance in life has been, of course, his Cockney accent. :D

Happy hunting, mrh and all of us!

Willow

kruoli 2022-09-22 09:10

[QUOTE=Uncwilly;613879]You are assuming mrh's gender. Also proof files might not have been a thing at the time.[/QUOTE]

Take a look at the post's content and author again, please.

mrh 2022-09-22 13:21

[QUOTE=Uncwilly;613879]You are assuming mrh's gender. Also proof files might not have been a thing at the time.[/QUOTE]

mrh is me :) I'm normally pretty careful to make sure proofs are uploaded. I was apologizing for missing this one.

mrh 2022-09-22 14:33

[QUOTE=WillowW;613885]Dear mrh, not to worry! I had fun doing such a lovely number as 123000023. If you have more numbers ending in 23 that might need a touch-up, send them my way!

Speaking of gender, assumptions and having fun, I'm going to go full White Queen and prepare breakfast while imagining that mrh is *definitely* (and secretly) the legendary number theoretician, the honourable Mr. Hermes Hadrian Hanahan, Habile Head of Hyperbolic Hieroglyphs at Harvard, whose only hindrance in life has been, of course, his Cockney accent. :D

Happy hunting, mrh and all of us!

Willow[/QUOTE]

I liked that number as well. It doesn't end in 23, but do you want to factor [M]5891453[/M] for me? It looks like I did the LL in 1998, it would be nice to finish it off :)

slandrum 2022-09-22 15:49

[QUOTE=mrh;613907]I liked that number as well. It doesn't end in 23, but do you want to factor [M]5891453[/M] for me? It looks like I did the LL in 1998, it would be nice to finish it off :)[/QUOTE]

This one has already been rerun with a cert. There are now 4 tests proving it's composite, 2 matching LL runs, and 2 matching PRP runs (one with cert).


All times are UTC. The time now is 05:48.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.