PDA

View Full Version : where are we, where we going?



Kileran
01-06-2003, 01:58 AM
I was refered to the SB site today, and am glad to help out. for about 6 years i've helped D.net and a handful of other smaller Distributed computing projects (ogr 22 and 23, some factoring projects, etc)

only thing i cant figure out with SB is... what are we doing? the site seems incredibly sparse, even the FAQ isn't up.

so i ask. what are we doing, what numbers are we crunching. what is being done WITH those numbers.

Sean

Mystwalker
01-06-2003, 05:06 AM
Well, sort of introduction is below the news on the main page.
I think that's quite a good start. If someone wants deeper insight into this subject, this forum is the place to ask - and read... ;)

ATM louie seems to be completely occupied with the new version of the software. Maybe we can help him with the FAQ?

Scarblac
01-06-2003, 06:58 AM
Originally posted by Kileran
only thing i cant figure out with SB is... what are we doing? the site seems incredibly sparse, even the FAQ isn't up.

so i ask. what are we doing, what numbers are we crunching. what is being done WITH those numbers.


I wrote a bit about SoB on Everything2 this weekend. The following nodes should explain it a bit:
http://everything2.com/index.pl?node=Seventeen%20or%20bust
http://everything2.com/index.pl?node=Sierpinski%20Conjecture
http://everything2.com/index.pl?node=Sierpinski%20number

I hope those links work (first post here)

jjjjL
01-06-2003, 02:24 PM
Scarblac - wow. nice writeup on the project. i've never seen a better one... including my own. ;)

-Louie

PS - if people want to start compiling an FAQ for the project, that would be awesome. anyone interested is encouraged to email me w/ any details you need to know. lhelm@umich.edu

Kileran
01-06-2003, 04:29 PM
1) so, we've found 5 so far. how long did that take, start to 5th?

2) if we never got faster, and never changed speed (same participants and comps for say...20 years) would the length between each "discovery" DEFINITELY get longer? is this like an upside down pyramid of time? the further we try to go, the longer it will take? (assuming we get NO new applicants)

Sean

jjjjL
01-06-2003, 04:44 PM
i started the project with my friend dave in march. in april, the website went public... we didn't have a domain name or any users at that point except for close friends.

a few months passed and nothing of real consequence happened. a few dozen people joined the project. then dave and i rewrote the client to run 30x faster. we still only had around 100 active users so not really that much was being done. then half way through november, i made it possible to form teams and that made people interested in the project immediately.

That was Nov 15th that teams were first started. Nov 27th was the date of the first prime discovery.... Dec 23rd was the latest.

And yes, w/o new users or faster clients or an improvement in the sieving or other factoring methods, the time between discoveries should increase. I say "should" only because there is no outright proof that the primes will take longer and longer to find, but other similar searches have shown that this is approximately the case and i believe that to be true for us too.

Then again, assuming no new users, no faster clients, no sieving improvements, and no other improvements would be a poor assumption indeed. If you had made those assumptions about SB just six months ago, you would have been wrong on all counts. ;)

-Louie

Mystwalker
01-06-2003, 05:30 PM
As it takes more time to test one k/n pair the bigger n becomes, it would statistically take longer to find a prime even when assuming equipartition of the primes.

But even when the distribution gets sparse with high numbers, you can't say it will definitely take longer. It's possible that we find the remaining 12 primes within the next 24 hours. It's possible that it takes 2,400 years...

But chances are quite good we'll find some more primes in the next weeks and months. :)

Louie:
How about a public creation of the FAQs? We just open a thread and all members can make suggestions about what should be in?

Kileran
01-06-2003, 08:03 PM
woah! you've made all 5 discoveries just in a few short months? that's amazing! if all 17 are truly out there, you should have no problem finding them in no time.

nuutti
01-07-2003, 01:00 AM
the 6th prime is harder to find than the 5th prime and the 7th is harder to find than the 6th. We have found the easy ones. This project will take some time, but we will know more about probable project lenght after we have found few more primes.

Yours,

Nuutti

Cowering
01-07-2003, 12:30 PM
I'd kinda like to pull a Captain Kirk and divert all our power to the forward phasers and punch a hole in just one K. Better chance of hitting it, and the news that generates will get even more people into SB.

Beef up the website, find one more big prime, and get slashdotted. (has SB been /. before??), that will get the speed and interest up!

I'd love to get more people onto my team too (which has grown to #3 in speed now), but its hard to get anyone interested when trying to explain the math.. and the "I'm running it, and you should too" loses strength the farther outside my usual list of contacts go.

Maybe I should find a way to embed the installer into my other retrogaming tools :)

:|ot|: www.retrogames.com/cowering.html

I am even losing members to www.theneoproject.com, where they have a clear (to the non math junkies) and useful purpose.. to spite Micro$oft in order to run homebrew code on the X-Box.

Cowering
01-07-2003, 12:49 PM
hahahahaha.. MS must have scared off the neoproject.com guys... maybe i'll get my team back up to full strength now... i'll love to hear THIS story about lawyers and DC.

Mystwalker
01-07-2003, 01:07 PM
Not really scared off...
They already said at the release of the X-Box Challenge that they'll stop it as soon as M$ starts to intercede. So it was like "Even when you only think of it, we'll backpedal!"

Of course, M$ did, the X-Box Challenge went back to where it came from and the NEO project has gained a lot of publicity and new members. I guess they celebrate an "I love it when a plan comes together!"-party ATM...

SB already got /.-ted (and heised).

Concentrating on 1 k would result in a likely bigger prime, but it's highly likely (given the assumptions mentioned in my last post) that it'll take longer to get there.

Maybe you can try to convince the folks with the big primes that get discovered and that it' quite likely to achieve a result within weeks.

Cowering
01-07-2003, 01:39 PM
I'm just not up on prime theory anymore. I was somehow thinking that since the other K were solved below N= 1.5M there might be some kind of distribution theorem/conjecture that would let us find a prime in the same basic range. Guess that idea is shot to hell now (assuming the client does not have a bug!!)

Scarblac
01-07-2003, 01:58 PM
[QUOTE]Originally posted by Cowering
I'd kinda like to pull a Captain Kirk and divert all our power to the forward phasers and punch a hole in just one K. Better chance of hitting it, and the news that generates will get even more people into SB.

I don't think that's a good idea, I think there are quite a few people who download the client, start it, then leave again before they've finished the block. The block will be reassigned after two weeks.

If that block happens to be a prime, then currently not much was lost, work on all the other k's wasn't affected, only the rest of the work on this k wasn't necessary in hindsight. If everyone was still looking in that same k, then it was two full weeks of work that could have been spent better.

Cowering
01-07-2003, 02:20 PM
I thought the blocked expired after 5 days now

Scarblac
01-07-2003, 02:26 PM
Well, same thing, just not as important then :-)

I think I read the two weeks in the FAQ, back when it existed.

In general I don't think a depth first search will necessarily find a prime before a breadth first search would.

Cowering
01-07-2003, 02:32 PM
hahahahahaha

<Cowering> hey, think about joining www.seventeenorbust.com over the weekend?
<namewithheld> NO, my wife does not like porn on the computer

(sigh)

Mystwalker
01-07-2003, 05:24 PM
:rotfl:

I already thought of that one on some occasions, too. ;)

Maybe a DC project with the aim to find the g-spot? :eek:


What just came to my mind:
We already did a "deep search" on one single k earlier on: 27653
We're at 3M and no prime so far - whereas we found 5 other primes quite fast...

Kileran
01-09-2003, 03:31 AM
hmm, a "deep search" on the "G-Spot"

i could get into that