mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Aliquot Sequences (https://www.mersenneforum.org/forumdisplay.php?f=90)
-   -   Decrease in activity? (https://www.mersenneforum.org/showthread.php?t=13602)

10metreh 2010-07-13 17:31

Decrease in activity?
 
I have noticed that when Subproject #5 ended, several people who had contributed to it did not move to Subproject #4 and (apparently) left the project. If any of you are still reading this forum, then please could you give a reason why you didn't join Subproject #4? Is it because the nice sequences have already been worked on, so there will be few terminations? We seem to have noticeably less contributors now.

kar_bon 2010-07-13 18:18

I'm contributing NPLB (my standard project) more the last time, because there's also much less activity the last months.

I will update my Aliquot pages after the errors in the FactorDB (small composites) are solved. It's a waste of time to do so until so much errors are in there.

mdettweiler 2010-07-13 19:03

I would suspect that this is just the usual "summer blahs" that most distributed computing projects go through. During the Northern Hemisphere summer, the higher temperatures induce a number of people to turn some or all of their "nonessential" machines offline--decreasing their crunching ability, or in the case of this project, sometimes even eliminating it entirely (as a lot of people run this project on older machines that are more likely to be turned off in the summer).

In cases like Karsten's, he's moved primarily back to his home project, NPLB, since they too have been hit by the "summer blahs" and they can't spare him so much. :smile: But as he also said, the FactorDB errors may also be a cause of declining interest--I suspect some people just got too frustrated with them.

Quite frankly, if someone here has the time and resources, they might want to consider setting up their own factor database. It wouldn't have to be too fancy--just a central repository for factorizations and Aliquot sequences, with a minimal worker setup to verify the primality of submissions and do minimal TF (or alternatively, something akin to "Quick ECM"). The less "extras" there are, the less that can go wrong.

Andi47 2010-07-13 19:18

[QUOTE=mdettweiler;221274]I would suspect that this is just the usual "summer blahs" that most distributed computing projects go through. During the Northern Hemisphere summer, the higher temperatures induce a number of people to turn some or all of their "nonessential" machines offline--decreasing their crunching ability, or in the case of this project, sometimes even eliminating it entirely (as a lot of people run this project on older machines that are more likely to be turned off in the summer).

In cases like Karsten's, he's moved primarily back to his home project, NPLB, since they too have been hit by the "summer blahs" and they can't spare him so much. :smile: But as he also said, the FactorDB errors may also be a cause of declining interest--I suspect some people just got too frustrated with them.

Quite frankly, if someone here has the time and resources, they might want to consider setting up their own factor database. It wouldn't have to be too fancy--just a central repository for factorizations and Aliquot sequences, with a minimal worker setup to verify the primality of submissions and do minimal TF (or alternatively, something akin to "Quick ECM"). The less "extras" there are, the less that can go wrong.[/QUOTE]

I have switched all my ressouces to a Cunningham factorization (c153), so I temporarely paused my activities for aliquot sequences. I have now switched one box back to ECM on alq4788 cofactors.

I too got a bit frustrated about the FactorDB errors. (and Syd is away since more than one month and has not responded to an email - I'm a bit worried.). Unfortunately I have neiter time nor the ressources to set up an own DB.

BTW: I too will pause most of my computing activies as soon as my cunningham factorization will have finished (due to summer heat).

RichD 2010-07-14 03:15

Subproject #4 is beyond my realistic reach. With my resources it would take over 24 hours to break a mid-c90 counting the 6-8 hours on ECM curves. I would hate to tackle a c106 using msieve. I haven't been able to get GGNFS to compile on my Mac G4.

Plus the two reasons mentioned earlier. Temps (why pay to generate heat and also pay to remove it) & data base issues.

10metreh 2010-07-14 08:00

Personally, I don't think DB issues are a problem. They're annoying, but if you post the .elf files on the forum then they're available on the internet. It's kind of easier to deal with if (like me) you've been working on aliquot sequences for longer than the database has been able to handle them. Then we had to post .elfs on the forum.

kar_bon 2010-07-14 08:58

[QUOTE=10metreh;221309]Personally, I don't think DB issues are a problem.[/QUOTE]

But for my summary-pages (made by hand so far) I also had to download every *.ELF and look at the last index (and perhaps determine the smaller primes and C-part if they not in there). That's too much manually work to handle.

Marking broken seqs. from the Database (like I did for some -> red) is also too much work:
mark them broken and when ok, edit the same seq-summary again! Double work!

smh 2010-07-14 09:38

[QUOTE=10metreh;221309]Personally, I don't think DB issues are a problem[/QUOTE]They are for me. My script (or AliWin) takes the latest status from the DB.

Greebley 2010-07-15 14:49

[quote=kar_bon;221312]But for my summary-pages (made by hand so far) I also had to download every *.ELF and look at the last index (and perhaps determine the smaller primes and C-part if they not in there). That's too much manually work to handle.

Marking broken seqs. from the Database (like I did for some -> red) is also too much work:
mark them broken and when ok, edit the same seq-summary again! Double work![/quote]

I am trying to see if I can compile an up to date set of elf files. As you said, with the db broken it is more work - I have to find each broken one and then fix it by jumping over the broken part. I was waiting for the db fixes, but it has been long enough I am going to try it. I have already been getting the elf sequences.

Finding the broken ones isn't trivial. One method would be to create a version of aliqueit that does the initial test, but doesn't try to extend the sequence. It should then warn me if anything isn't prime which is one thing my current test program doesn't do. Even once I find them, there is the matter of searching the db for further parts of the sequence by by-passing the broken bit.

Assuming I manage to succeed, I will post the sequences in elf or alq format, so they are down-loadable. This would enable you to do an update.

Note that I also figure out stats like the last index, number of digits, etc. I can also create a csv file or something that might be even easier for you to use.


All times are UTC. The time now is 19:44.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.