what constitutes a validate?

Message boards : Number crunching : what constitutes a validate?
Message board moderation

To post messages, you must log in.

AuthorMessage
merle van osdol

Send message
Joined: 23 Oct 02
Posts: 809
Credit: 1,980,117
RAC: 0
United States
Message 1595672 - Posted: 1 Nov 2014, 19:44:38 UTC

A validate means what? The same number of spikes, pulses, gaussians, etc, or what?
merle - vote yes for freedom of speech
ID: 1595672 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22160
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1595683 - Posted: 1 Nov 2014, 20:26:54 UTC

A valid result is determined by the two tasks (remember each Work Unit is sent out at least twice to random crunchers) being returned, and having results that are near enough identical. That is they have the same number of features, and that those features are at the same times and frequencies.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1595683 · Report as offensive
merle van osdol

Send message
Joined: 23 Oct 02
Posts: 809
Credit: 1,980,117
RAC: 0
United States
Message 1595689 - Posted: 1 Nov 2014, 20:42:46 UTC - in response to Message 1595683.  

A valid result is determined by the two tasks (remember each Work Unit is sent out at least twice to random crunchers) being returned, and having results that are near enough identical. That is they have the same number of features, and that those features are at the same times and frequencies.


Thanks Rob
merle - vote yes for freedom of speech
ID: 1595689 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1595706 - Posted: 1 Nov 2014, 21:28:27 UTC

I've asked this question a few times myself over the years, and I still don't have a complete answer. There are a few veterans here that keep trying to explain it, but basically..

As long as more than half of the signals found by the two wingmates (or more, in the case of inconclusive) are very close in value (for example, a CPU application may find a value of 20.00001 and a GPU app may find 19.99998), it is considered to be "close enough" to being 20.00000. As long as more than half of the signals found are "close enough," then the results are considered "weakly similar" and both get credit and are validated.

I believe the one that found more signals, or was the first to report back, is the one that become canonical and gets put into the database.

The most recent time I asked about this (Oct. 15), was for an AP task. I found 4 single and 10 repetitive pulses, and my wingmate found 0/10, but both were valid. The problem was the drivers they were using on that GPU would not find any single pulses, but 10 is more than half of the 14 total, so it still validated. Therefore, the two (actually, three of us in that situation..I was the third party tie-breaker) were "close enough" to be "weakly similar" and therefore valid.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1595706 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1595710 - Posted: 1 Nov 2014, 21:45:28 UTC

What you see on the webpage for the results in the <stderr_txt> is not the result file being used for comparison. It is just some extra info we can look at & sometimes use to determine what is going on.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1595710 · Report as offensive
David S
Volunteer tester
Avatar

Send message
Joined: 4 Oct 99
Posts: 18352
Credit: 27,761,924
RAC: 12
United States
Message 1596985 - Posted: 4 Nov 2014, 15:07:36 UTC - in response to Message 1595706.  

I've asked this question a few times myself over the years, and I still don't have a complete answer. There are a few veterans here that keep trying to explain it, but basically..

As long as more than half of the signals found by the two wingmates (or more, in the case of inconclusive) are very close in value (for example, a CPU application may find a value of 20.00001 and a GPU app may find 19.99998), it is considered to be "close enough" to being 20.00000. As long as more than half of the signals found are "close enough," then the results are considered "weakly similar" and both get credit and are validated.

I believe the one that found more signals, or was the first to report back, is the one that become canonical and gets put into the database.

The most recent time I asked about this (Oct. 15), was for an AP task. I found 4 single and 10 repetitive pulses, and my wingmate found 0/10, but both were valid. The problem was the drivers they were using on that GPU would not find any single pulses, but 10 is more than half of the 14 total, so it still validated. Therefore, the two (actually, three of us in that situation..I was the third party tie-breaker) were "close enough" to be "weakly similar" and therefore valid.

I presume one of the others pretty closely matched you with 4/10. One of you was chosen as canonical. The oddball was called weakly similar and granted credit for having done good work, even if it was considered to be "wrong."
David
Sitting on my butt while others boldly go,
Waiting for a message from a small furry creature from Alpha Centauri.

ID: 1596985 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1597024 - Posted: 5 Nov 2014, 4:13:17 UTC - in response to Message 1596985.  

The most recent time I asked about this (Oct. 15), was for an AP task. I found 4 single and 10 repetitive pulses, and my wingmate found 0/10, but both were valid. The problem was the drivers they were using on that GPU would not find any single pulses, but 10 is more than half of the 14 total, so it still validated. Therefore, the two (actually, three of us in that situation..I was the third party tie-breaker) were "close enough" to be "weakly similar" and therefore valid.

I presume one of the others pretty closely matched you with 4/10. One of you was chosen as canonical. The oddball was called weakly similar and granted credit for having done good work, even if it was considered to be "wrong."

Yeah, that's what Josef ended up saying, too.

Stock Linux app was the canonical result. The GPU was the original wingmate, and it was inconclusive, so it got sent to me to be the tie-breaker. Mine matched the stock linux app, but because the GPU's 10 signals is more than 50% of the total of 14, it was given credit.

The point I was trying to make by adding that scenario into this thread as my explanation is that it doesn't necessarily require that you have the same number of signals found to be valid. If there are two results that have a different number of signals, then a third result is necessary, and the third result will (hopefully) match one of the two that have already submitted their results. In the case of my example, if someone else with a pre-Fermi GPU used drivers newer than 340.x, they would have also reported 0/10, and the stock Linux app would not have been the canonical result, but still given credit, but then the 4 single pulses that were legitimately there, would not have ended up in the database.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1597024 · Report as offensive

Message boards : Number crunching : what constitutes a validate?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.