mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Aliquot Sequences (https://www.mersenneforum.org/forumdisplay.php?f=90)
-   -   Sugg. for work distr. based on Syd's database (https://www.mersenneforum.org/showthread.php?t=11713)

hhh 2009-04-11 13:49

Sugg. for work distr. based on Syd's database
 
Dear friends,

I have an idea for a more automatized, yet more distributed approach to aliquot sequences.

It basically consists in choosing a few dozen of low priority sequences, to put them into Syd's database, and to set the database to automatically run Very High Limit ECM on all composites that show up in these sequences. Perhaps it could even be set to run yafu on everything that deserves this treatment. At some point, inevitably, the c9*+ GGNFS jobs show up. Here, the database does nothin but to list the composites on some special page, with a reserve button next to them.

If I want to contribute, I can choose the one I like to factor, click the button, and have the number reserved for some reasonable time, depending on the size of the composite. Later, for the really big composites, the database could even provide the polynomial file, which would be provided and approved by some of the factoring professionals. But that's for later.

The purpose of all this is to let the machine do the easy work, and to leave the more difficult and most time-consuming to the machines with manual interface. I think it would be not too difficult to implement either. Also, it would enable every user to run factorizations which are within the limits of his possibilities, for me with my laptop perhaps doing stuff below c115, without having to drop a whole sequence just because some c125 shows up, which can be tackled by the guy with the i7 who gets now bored with factoring c105s.

Drawback is that the aspect of having "his own" sequence is lost, the waiting for the downdriver and enjoying it and so on. But those who want this are still to reserve their own sequence. Also, if no membership system is implemented (which I think should not be done, because too complicated to maintain), there would be no credit whatsoever.

What do other people think?

Cheers, anyway, H.

10metreh 2009-04-11 17:29

What exactly do you mean by "low priority" sequences? I would have thought these are the ones with 2^3*3*5 and its variants, which are not best for this treatment. I would go for a sequence that is stable or driverless in the 120-130 digit range, which are large to tackle on one's own but could be done with the help of the database, and have a possibility of acquiring the downdriver.

hhh 2009-04-12 02:26

[QUOTE=10metreh;168930]What exactly do you mean by "low priority" sequences?[/QUOTE]

I mean just any ordinary sequence; ordinary meaning with no particular interest. Say, not being element of any top ten. By the way, I think sequences with driver should be elegible as well. Their composites will stockpile somewhere down the list, but that's no problem, is it?


All times are UTC. The time now is 12:20.

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