mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Aliquot Sequences

Reply
 
Thread Tools
Old 2014-11-28, 00:18   #1
RobertS
 
RobertS's Avatar
 
Aug 2009
somewhere

C516 Posts
Arrow What do we do next?

Quote:
Originally Posted by fivemack View Post
Unconnected took that number four posts ago.

If you are running elves, please don't run them on reserved numbers, and please put reservations here on the numbers the elves are looking at!
Sorry, overseen (again).

My "elves" are looking at all seqs that are below 120 digits, do I really have to reserve them all?
If so:
Reserving all seqs starting < 1M, that are below 120 digit in size, and that are not reserved.

Last fiddled with by RobertS on 2014-11-28 at 00:20 Reason: A really really big grin
RobertS is offline   Reply With Quote
Old 2014-11-28, 01:15   #2
richs
 
richs's Avatar
 
"Rich"
Aug 2002
Benicia, California

11·103 Posts
Default

RobertS, where your elves the ones which hijacked my reserved 230916 until it reached C120?
richs is offline   Reply With Quote
Old 2014-11-28, 02:33   #3
RobertS
 
RobertS's Avatar
 
Aug 2009
somewhere

C516 Posts
Default

Quote:
Originally Posted by richs View Post
RobertS, where your elves the ones which hijacked my reserved 230916 until it reached C120?
It might be:
Long long time ago (>8 month) I told my "elves" to do aliqueit on all seqs below 1M, and below 120 digits and by using an increasing cofactor limit. Right now they are doing the C116s.
If your 230916 did not have a "hard" cofactor of >= C116 ... my elves probably did it.
I don't have any logs. If it it was me, I'm sorry. Along with the "new" reservation script, I have now implemented reservations to my elves, but.... of course, too late.

PS: By saying "elves" I mean: 8 cores on an oc I7.
I also did not follow this forum thoroughly, because there is no clear guidance any more. Where are the subprojects? Why is the Aliquot Sequence Summary Page (the starting point for anyone who enters "Aliquot") not updated? Why is factoring the M991 more important than factoring 4788.5232? (of course: "mersenneforum", but...)

Last fiddled with by RobertS on 2014-11-28 at 03:15
RobertS is offline   Reply With Quote
Old 2014-11-29, 19:22   #4
henryzz
Just call me Henry
 
henryzz's Avatar
 
"David"
Sep 2007
Cambridge (GMT/BST)

131328 Posts
Default

Quote:
Originally Posted by Dubslow View Post
Henryzz note that the script doesn't check for toe stepping, so you might need to manually reserve it back to unconnected.

Unfortunately, I might not be able to fix it even tomorrow; it depends on how much my laptop charger acts up.
Done.

Quote:
Originally Posted by RobertS View Post
It might be:
Long long time ago (>8 month) I told my "elves" to do aliqueit on all seqs below 1M, and below 120 digits and by using an increasing cofactor limit. Right now they are doing the C116s.
If your 230916 did not have a "hard" cofactor of >= C116 ... my elves probably did it.
I don't have any logs. If it it was me, I'm sorry. Along with the "new" reservation script, I have now implemented reservations to my elves, but.... of course, too late.

PS: By saying "elves" I mean: 8 cores on an oc I7.
I also did not follow this forum thoroughly, because there is no clear guidance any more. Where are the subprojects? Why is the Aliquot Sequence Summary Page (the starting point for anyone who enters "Aliquot") not updated? Why is factoring the M991 more important than factoring 4788.5232? (of course: "mersenneforum", but...)
The Aliquot Sequence Summary Page has sort of been made obsolete by Dubslow's page(linked in the first post).

Subprojects: Does anyone have have any specific ideas? It would be nice to get all the sequences upto 120 digits. Most of those have drivers now so there isn't that much chance of terminations. Also Dubslow's page makes it so easy to find them which makes a thread seems a bit redundant.

I personally like the idea of taking all the sequences without a driver or a 3 up as high as we can. There are only ~400 of them and less still if we exclude guides like 2^3*5 and 2^5*7. It might be nice to see how many we can eliminate.
In the past we ran drives with sequences that stood a good chance of losing their driver on the next iteration. Do people want to do that again?

I don't have a huge amount of time which limits the ideas I come up with. Now we have the script I can deal with a good throughput of reservations. If the project members suggest ideas then I will try to help putting them into place.

In terms of factoring 4788.5232 I think we may as well get started at least with finding a poly. Then those who aren't participating on M991 can start.
I never got an answer when I asked whether people still wanted to work on 3366 or 3408. If I get no response this time they will be unreserved and unstickied.

The project used to be so popular because there was quick work to do and progress was quickly visible. Anyone got any ideas how to get back to that? Extending to >1M might do it but that seems like we are just extending for the sake of it. Searching for sociable numbers as EdH is doing could be quite interesting if it can be distributed easily.
henryzz is offline   Reply With Quote
Old 2014-11-30, 10:37   #5
fivemack
(loop (#_fork))
 
fivemack's Avatar
 
Feb 2006
Cambridge, England

143238 Posts
Default

Getting all the sequences to 120 digits is a job too small to be worth considering for the group - either RobertS or I could do it before Easter.

If we're hunting terminations, which I think is the obvious right thing to be working on, the correct strategy is to work with the smallest sequences and be willing to give up very quickly - 'get all the 120-digit sequences to either 121 digits or termination' - which again doesn't work well for a group project unless we have good task-allocation software.

Following large sequences down longer and narrower rabbit-holes just isn't that appealing.
fivemack is offline   Reply With Quote
Old 2014-11-30, 16:14   #6
EdH
 
EdH's Avatar
 
"Ed Hall"
Dec 2009
Adirondack Mtns

24·11·19 Posts
Default

Quote:
Originally Posted by henryzz View Post
...
The project used to be so popular because there was quick work to do and progress was quickly visible. Anyone got any ideas how to get back to that? Extending to >1M might do it but that seems like we are just extending for the sake of it. Searching for sociable numbers as EdH is doing could be quite interesting if it can be distributed easily.
As I am currently interested in finding 4 cycle sociables, I will tell you it does not meet the first description above. I am searching rather narrowly, in a vast field and only after a couple years did I see a result. My involvement with my machines is basically just checking a few times a day to see if any died (since they are all recycled). I can see how far toward infinity (not very close) the latest machine's assignment was and I can see if anything was found.

Edit: I am not trying to dissuade anyone or keep this to myself, but merely noting that this endeavor is boring.

Last fiddled with by EdH on 2014-11-30 at 16:16 Reason: Trying not to discourage.
EdH is offline   Reply With Quote
Old 2015-01-30, 01:26   #7
schickel
 
schickel's Avatar
 
"Frank <^>"
Dec 2004
CDP Janesville

2×1,061 Posts
Default

Quote:
Originally Posted by fivemack View Post
Getting all the sequences to 120 digits is a job too small to be worth considering for the group - either RobertS or I could do it before Easter.

If we're hunting terminations, which I think is the obvious right thing to be working on, the correct strategy is to work with the smallest sequences and be willing to give up very quickly - 'get all the 120-digit sequences to either 121 digits or termination' - which again doesn't work well for a group project unless we have good task-allocation software.
Back when we first got this sub-forum started, I was kind of dreading the day that we hit this point. Unfortunately after everything was brought up to 100 digits, all the easiest work was done. Terminations/merges are getting harder to find, and unless or until someone proves that all sequences terminate or there is one or more sequence(s) that don't terminate, the only thing to do is keep lifting the floor of the current crop of OE sequences.
Quote:
Following large sequences down longer and narrower rabbit-holes just isn't that appealing.
Unfortunately this is probably the most likely fate of the project, with just a lot of scut work of factor, rinse, and repeat being the rule, and, what, 200 digits being the ultimate barrier?

The work order for the immediate future would seem to be:
  • work all sequences without a driver, partial driver or 3 until termination or driver capture, developing any resulting downdrivers at top priority,
  • work sequences with partial driver guides without a 3 the same way,
  • work sequences with a 3,
  • work driver controlled sequences in some order, smallest to largest, maybe?
  • work other sequences that are interesting.
In fact, if you look at my reservations, that's what I've been doing. I used to troll the status looking for downdrivers, but there is so much firepower beyond what I have now that I've concentrated on just a few sequecnes myself. Any "full" 2^9 drivers were my priority and currently 7044 is the only one that falls into that category. Of the others, 363270, 611156, and 700128 are former 2^9 sequences, 689652 was one of the "Project 400" sequences that turned interesting, turning in a couple of nice plateaus on its way to 160 digits (I picked it up because it had a partial 2^9 driver when I noticed it), and 689652 being the only sequence over 5400+ lines without a driver or 3 (I'm hoping that it can turn in a nice downdriver drop, potentially bringing it up over 10k lines.....)
schickel is offline   Reply With Quote
Old 2015-01-30, 20:33   #8
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3·29·83 Posts
Default

What about things like 2^2 * 7^2 * C? Or similar for other drivers? May get a few easy driver breaks that way. Such sequences are easily findable via my webpage, either the HTML/JS interface or by scripting through the text/JSON data also provided.
Dubslow is offline   Reply With Quote
Old 2015-01-31, 05:10   #9
schickel
 
schickel's Avatar
 
"Frank <^>"
Dec 2004
CDP Janesville

2×1,061 Posts
Default

Quote:
Originally Posted by Dubslow View Post
What about things like 2^2 * 7^2 * C? Or similar for other drivers? May get a few easy driver breaks that way. Such sequences are easily findable via my webpage, either the HTML/JS interface or by scripting through the text/JSON data also provided.
We tried that a while back, I looked for sequences that had that form and several forumites worked through several succeeding lines (or more); unfortunately it's harder than it looks. The lower drivers like 2 * 3 and 2^2 * 7 keep the even power on the 3 or 7 easily, but the "1 or 2 factors with the proper form" lines are unfortunately too rare. The higher drivers like 2^4 * 31 and 2^6 * 127 tend to acquire the needed even power less often, and it only hangs around for a line or two when it does show up.

Searching, looks like it was just 2^3 * 3 and 2^4 * 31....of course, your "more current summary" makes it a little easier to search, so just like the cycle search, this is not to say no to anyone, just a note about how difficult it will potentially be.

The other potential expansion of the project I forgot to list was extending the search range to starting numbers above 1M, but that is a whole other can of worms that we may not want to open up.
schickel is offline   Reply With Quote
Old 2015-02-01, 05:08   #10
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3·29·83 Posts
Default

I actually meant specifically those that have the even power and only one cofactor, not any extra crap. After also filtering for the modulo requirement mod 4 of the cofactor, we surely ought to get several driver breaks on the immediate next line?
Dubslow is offline   Reply With Quote
Old 2015-02-01, 10:33   #11
schickel
 
schickel's Avatar
 
"Frank <^>"
Dec 2004
CDP Janesville

2×1,061 Posts
Default

Quote:
Originally Posted by Dubslow View Post
I actually meant specifically those that have the even power and only one cofactor, not any extra crap. After also filtering for the modulo requirement mod 4 of the cofactor, we surely ought to get several driver breaks on the immediate next line?
That sounds like a potential project, especially if you can add a module to your status spider to do the finding for you.

Again, I'm not trying to actively discourage, I'm just saying that even if you do find some number of candiates to look at, it's not neccessarily a slam dunk. I did quite a few trials on my own before proposing the two projects, and I did manage a couple of escapes, so maybe this would be something that could be expanded to a viable subproject. In fact if you can get an escape to a stable guide, there's always a good chance of a nice long downdriver run can happen. And as Tom said, the prospect of plodding through ever larger factorizations can be discouraging. (In fact I'm currently engaged in a c170 job that is ~30 days into probably 90 days....trudge, trudge, trudge.)
schickel is offline   Reply With Quote
Reply

Thread Tools


All times are UTC. The time now is 10:15.

Sat Sep 19 10:15:56 UTC 2020 up 9 days, 7:26, 0 users, load averages: 2.22, 1.88, 1.71

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.