Problem with sieve stats

[vjs]

I think the stats are frelled,

http://www.aooq73.dsl.pipex.com/scores.htm

[Death]

OMG 81 ~DeadJDona 5.10 ( 0.00)

whats happened?

Last Update: Fri 23-Jan-2004 19:25 (GMT)

Current 90% sieve point: 71.801T



users stats are normal

Current 90% sieve point: 331.450T (max bias: 8.29)

don't worry....

[vjs]

Humm,

Still not showing correctly for me.... Acutally I'm not on the sieve stats at all????

[Troodon]

Don't worry. I think that's MikeH doing some changes to sync the stats with the new queue system.

[MikeH]

I think the stats are frelled,

Whoops, sorry.
While doing a few updates I ran an update from an old saved copy of the data. When I looked back at the screen it had finished and for one moment I thought "did that upload anything?", but dsmissed it. That'll teach me....

Should be OK in a few minutes. The DC window should be correct at the next update (03:00 GMT+1).

[Death]

Last Update: Tue 22-Jun-2004 03:07 (GMT+1)

Most recent finds
p (T) k n Score Factor found Score changed Score was Score could be Reqd bias
313.331T 19249 15783422 3.133 Mon 21-Jun-2004 243924.612
313.043T 4847 5150223 3.130 Fri 18-Jun-2004 15568.880

and

RATIO: 313063.82G - 313317.67G, size: 253.85G, est fact: 11 (313063821098389-313317667566881) R: 1.535, S: 0.017
( 253.85G) : 313000-313370 Death (ETA: Mid July)

sob.com/sieve says it not new when I try to submit 313063, because I submit it yesterday.

[Death]

313.078T 67607 11161827 3.131 Wed 23-Jun-2004 121891.407
313.331T 19249 15783422 3.133 Mon 21-Jun-2004 243924.612
313.043T 4847 5150223 3.130 Fri 18-Jun-2004 15568.880

well today I found 3131.078, but still no 313.063

where is it??

[MikeH]

Hi Death,

From http://www.aooq73.dsl.pipex.com/resu...duplicates.zip

313063821098389 10223 13931885 3419 44 d1

it's the first duplicate of

192759405970217 10223 13931885 1608 0

found by myself about 5 months ago.

Because the sob.dat file we're working with isn't right up to date, there will always be a small number of factors which have been found since the sob.dat file was created. Taking my last big range as an example

281480 - 290000 8520 452 12 0 549311.704

That's 452 unique factors, and 12 duplicates in an 8.5T range.

Hope that helps

Cheers,
Mike.

[Death]

WHY??? why???????

ermmm... so there's few Q:

how you fing it if this range is untested?

313063821098389 10223 13931885 3419 44 d1 - what does this mean?
192759405970217 10223 13931885 1608 0 - and this (bold).

313.../192.... = 1.624... it's not even two (three) times larger.

and I think that 10223*2^1393...+1 is 313... how it can be that 10332*2^1393... = 192???


now I realize how little I know about sieveing....

and maybe there's a time to update sob.dat?

[larsivi]

Originally posted by Death
and maybe there's a time to update sob.dat?

Well, it's not that important. proth_sieve find them anyway. The only 'problem' is that factors you think are unique turn out to be duplicates. No work is saved anyway, actually I think it takes some time to create new .dat's.

[Mystwalker]

Originally posted by Death
313.../192.... = 1.624... it's not even two (three) times larger.

and I think that 10223*2^1393...+1 is 313... how it can be that 10332*2^1393... = 192???

Those two factors needn't be multiples. Take e.g. 24:
4 and 6 are factors, but 6 / 4 = 1.5

Originally posted by larsivi
No work is saved anyway, actually I think it takes some time to create new .dat's.

That wouldn't really take a lot of time - just take all new unique factors out of the database and erase the coresponding line in the dat file.
My estimates:

- create script: 5 min
- run script: 1 min
- validate result: 4 min

Maybe more time if one got it wrong...
AFAIK, there's probably a very little performance increase, plus memory consumption could be a bit lower...

[royanee]

Originally posted by Mystwalker
Those two factors needn't be multiples. Take e.g. 24:
4 and 6 are factors, but 6 / 4 = 1.5

In fact, they shouldn't be multiples. You would have to multiply two of the factors of a number to get a factor that is evenly divisible by another factor (usually). But because we should really only be trying to divide the numbers with primes (because all composite numbers are nulled by their respective factors) this is always true. Therefore, if there ever were multiples, you have been wasting your CPU time by checking for composite numbers. The prime factors of that composite factor would be the numbers that should have been checked (the smaller of which usually being the unique factor discovered).

[MikeH]

313063821098389 10223 13931885 3419 44 d1 - what does this mean?
192759405970217 10223 13931885 1608 0 - and this (bold).

p, k, n, user id, team id. If there is a field at the end, d=duplicate, e=excluded, the number indicates how many factors of the same k/n have been found before it.

now I realize how little I know about sieveing....

You don't need to understand how sieving works to provide a useful contribution.

and maybe there's a time to update sob.dat?

When the next k is eliminated.

[Keroberts1]

and hopefully soon at that.

[priwo]

on last friday (08/13) i submitted these factors (along with some other factors):

355585545702041 | 21181*2^1273532+1
355620435136117 | 10223*2^6925037+1

I did not get credit for these two factors. The factor-submitting page said "all verifyed, all new"

any idea what happened ?

[Joe O]

Originally posted by priwo
on last friday (08/13) i submitted these factors (along with some other factors):

355585545702041 | 21181*2^1273532+1
355620435136117 | 10223*2^6925037+1

I did not get credit for these two factors. The factor-submitting page said "all verifyed, all new"

any idea what happened ?

355585545702041 21181 1273532 386 9 d1

355620435136117 10223 6925037 386 9 d1

You got credit for them. The d1 means that this k n pair already had a known factor.

A duplicate factor will score as follows:
score = p/100T, capped at 35, or the score (above) as if it were unique, whichever is lower; when a prime is found all duplicates ever found for that k score 0.

EDIT:
266814184928569 21181 1273532
318438111231761 10223 6925037

Were the original factors found.

[royanee]

hehe The bottom one was mine. Sorry about that... It gave me 47,722 points. But don't worry. Now that I won't have access to 1200 kcem/s you'll be more than able to make up for the points deficit. I will become like the sleeping cow...

[priwo]

thank you for the reply, I think I understand what happened