[QUOTE=LaurV;278451]You should consider the fact that not all people have the "power" to go higher with factoring. It does not matter where I do reserve a sequence and where I do abandon it. Is it 130 digits? 115? 100? 95 digits? It does not really matter. What matters is to [B]reserve it properly[/B] (to avoid duplicating the efforts) and to [B]unreserve it properly[/B] when is done, or when is getting boring/difficult-to-handle/blabla, so anybody else could take over, if he wants. THAT is important. If it is 110 digits or just 95 digits, it is no problem. Some people have already difficulties at 100 digits. Abandoning the sequence without saying anything, [B]THAT is bad[/B].[/QUOTE]

That makes sense but there is counterpoint: Keep in mind I'm not suggesting that people go higher with a sequence than where they started at. If they reserve the sequence in the first place, by default it was most likely at >= 110 digits. So since they were searching >= 110 digits to begin with, then they believed they had the firepower to search at that level and if a downdriver is encountered, then they should have the firepower to search the sequence until it is back up to that level.

Perhaps a better way to state it would be: Whatever height one reserves a sequence at is the minimum height that one should release such a sequence (or until it terminates/merges). If one can't handle the initial height to begin with, one shouldn't start such a sequence.

As a personal example: I've never bothered to set up ggfns on my machine (I'd had errors that I could never figure out) so all that I've searched with in the past is the aliquot program with ECM/yafu in the folder. When I hit a 105+ digit factor, if ECM doesn't knock it out, it usually ends up spending 1-2 days using yafu to factor it (this is a single iteration). Even doing that arduous process on just 1 core, I've rarely if ever spent more than one CPU week searching a sequence from 100 to 110 digits...this is on a 2.9 Ghz machine. Even people with one core can do this.

This is all mostly a moot point. By my observation, almost everyone here searches them back to 110+ digits if they have a downdriver run. It only got brought up because it appeared that someone was being encouraged to only search a sequence back to 100 digits after it was reserved at 110+ digits just because it caught a nasty driver on the way back up.

[QUOTE=gd_barnes;278455] Whatever height one reserves a sequence at is the minimum height that one should release such a sequence (or until it terminates/merges).[/QUOTE]

This I agree 100%! It will avoid some "downdriver hunters", or people with fast machines waiting for me to bring the sequence up, and if the downdriver appears they take over and bring it down faster then me, duplicating my work (in fact, I will duplicate their work, because I don't know), just to take the glory in case the sequence terminates.... Well this logic is a bit snakish, but you understand what I want to say.

[QUOTE=LaurV;278459]<snip> and if the downdriver appears they take over and bring it down faster then me, duplicating my work (in fact, I will duplicate their work, because I don't know), just to take the glory in case the sequence terminates.... Well this logic is a bit snakish, but you understand what I want to say.[/QUOTE]

If this actually happened, the person who had the sequence reserved would be credited with the termination.

I like 'magic elves', that's a much better name than just 'workers'. :smile:

In theory the elves are factoring up to c108 now (or more if there's a downdriver), except that recently some kind soul must have run some minimal ecm on every open sequence and cracked many of the composites. Maybe it was the DB itself... it doesn't seem to have been aliquot-focused since most sequences were advanced by only one or two terms. This has left the DB with over 1000 sequences with composites <c108, so the elves are currently busy clearing this backlog. Because of this there are currently 10 unreserved sequences <110 digits, but that's not typical. In general these days anything left unreserved in the DB will quickly be pushed to at least c109, which is very likely to be >= 110 digits. Normally no unreserved sequence will stay below 110 digits for more than 2-3 days.

I see the point that it's polite to leave the sequence as well-off as when you found it, but let me assure you that the elves don't mind at all picking up where you left off. My goal is to add as many terms as possible as quickly as possible -- drivers come and go and if you drop a sequence at 100 digits because it's got a driver, the elves may get lucky. This has happened several times, so beware -- I feel a little guilty terminating a sequence that someone just dropped, but not *that* guilty. :smile:

When I'm bored late at night (lol), I will sometimes go into the DB and do a few scans on some unreserved sequences. It's entertaining to see if I can advance a few of them anywhere from 1 to 10 iterations. On a rare occassion, I've caught a downdriver and will reserve it. But there's no way I'd do 1000 of them though...perhaps 50-100 of them.

At least that's part of the reason why your magic elves picked up some more sequences to search. My guess is that the DB automatically factors some of them after a period of time.

I'll take 277344.

You're all correct: I was a little rash in advocating bailing on a sequence early. The only defense I can offer is that I was a little worried about keeping the motivation of someone new to the aliquot project alive. I was afraid that the grind of taking a sequence back up to at least the point it was picked up might be a little daunting.

I [I]was[/I] also leaning on bchaffin's workers. I think that aside from the factorDB, his workers would be what I would point out as the most important thing to come out of this project the last two years. The fact that every single unreserved sequence gets work on it, with no regard to the current state of the sequence, means that we can survey everything and pick out sequences that might profit from more concerted effort. And in case I don't mention it often enough, thank you, thank you, thank you to bchaffin for running the workers.

[QUOTE=gd_barnes;278455]As a personal example: I've never bothered to set up ggfns on my machine (I'd had errors that I could never figure out) so all that I've searched with in the past is the aliquot program with ECM/yafu in the folder. When I hit a 105+ digit factor, if ECM doesn't knock it out, it usually ends up spending 1-2 days using yafu to factor it (this is a single iteration). Even doing that arduous process on just 1 core, I've rarely if ever spent more than one CPU week searching a sequence from 100 to 110 digits...this is on a 2.9 Ghz machine. Even people with one core can do this.[/quote]Have you ever re-considered getting GNFS up and running. If it's Windows, I might be able to help you get it up and running....[quote]This is all mostly a moot point. By my observation, almost everyone here searches them back to 110+ digits if they have a downdriver run. It only got brought up because it appeared that someone was being encouraged to only search a sequence back to 100 digits after it was reserved at 110+ digits just because it caught a nasty driver on the way back up.[/QUOTE]Well, after trying some escape attempts where a sequence had a driver with the proper form, I just don't want anyone to get discouraged and give up because the feel like they're not getting anywhere.....and if someone catches the factoring bug, I hope that they get the urge to throw the jobs our way. When I first got interested in extending sequences, I added some lines to 115302. At the time Wolfgang had this note on his site:[quote=wolfcrey]The calculations of the sequence [URL="http://factordb.com/sequences.php?se=1&eff=2&aq=115302&action=last20&fr=0&to=100"]115302[/URL] have been stopped at index 5415 (C100).[/quote]I figured, what the heck, an escape can't be that hard, can it? Well, you can see where that kind of thinking (and a reluctance to give up on something you've already spent too much time on) can lead to.....

[SIZE="1"]Thinking about that makes me think that where it is now is a whole lot less than stuff I've done in the last 12 months.....maybe I should pick it back up (must resist, must resist!)[/SIZE]

I am thinking the same in this very moment about [URL="http://factorization.ath.cx/aliquot.php?type=1&aq=1100000000462900393"]this sequence[/URL] which I am working it for Jean Luc, I said to myself a couple of terms ago "it looks like it started to show that is tired, it must fall...." :smile:, and since then am am cursing my days factoring only C100+ and being unable to get rid of 2*3. Raising very slow, but still rising... you see me after few years, at some C180, with exactly the same slope of the graphic...

By the way, [URL="http://factorization.ath.cx/sequences.php?se=1&eff=2&aq=996666&action=last20&fr=0&to=100"]996666[/URL] got a downdriver at T1599 after being quite undecided for a while, at 134 digits. Let's see for how long. I am still crunching it, but a couple of C129-C130 made my last days quite boring. If it will go to the ground, then this could be a new record for the highest peak before termination, couldn't be? :P

[QUOTE=LaurV;278927]By the way, [URL="http://factorization.ath.cx/sequences.php?se=1&eff=2&aq=996666&action=last20&fr=0&to=100"]996666[/URL] got a downdriver at T1599 after being quite undecided for a while, at 134 digits. Let's see for how long. I am still crunching it, but a couple of C129-C130 made my last days quite boring. If it will go to the ground, then this could be a new record for the highest peak before termination, couldn't be? :P[/QUOTE]The record right now looks like this (from Wolfgang's page):[code]
62850: i3227 ([COLOR="Red"]C121[/COLOR]: lg 120.98) terminates in 41 04-09-2008 Stern
707016: i3396 ([COLOR="red"]C124[/COLOR]: lg 123.81) terminates in 41 12-11-2010 Batalov
921232: i5510 ([COLOR="red"]C127[/COLOR]: lg 126.67) terminates in 11 12-04-2010 Creyaufmüller/unconnected [/code]So, yes, a [COLOR="red"]c134[/COLOR] maximum height would certainly set a new high....

[QUOTE=LaurV;278927]I am thinking the same in this very moment about [URL="http://factorization.ath.cx/aliquot.php?type=1&aq=1100000000462900393"]this sequence[/URL] which I am working it for Jean Luc, I said to myself a couple of terms ago "it looks like it started to show that is tired, it must fall...." :smile:, and since then am am cursing my days factoring only C100+ and being unable to get rid of 2*3. Raising very slow, but still rising... you see me after few years, at some C180, with exactly the same slope of the graphic... [/QUOTE]
In a similar vein, last night [URL="http://www.factordb.com/sequences.php?se=1&eff=2&aq=277344&action=last20&fr=0&to=100"]277344[/URL] was at 127 digits with 2^4*3^2 and quickly rising. I considered releasing it, but decided to give it one more day. At 129 digits it lost the 3 and started falling, then acquired the downdriver. Let's see how far it falls...