## spike/pulse/triplet/gaussian counts

Message boards : Number crunching : spike/pulse/triplet/gaussian counts
Message board moderation

AuthorMessage
zombie67 [MM]
Volunteer tester

Joined: 22 Apr 04
Posts: 758
Credit: 27,771,894
RAC: 0
Message 618404 - Posted: 13 Aug 2007, 2:25:12 UTC

Spike count: 4
Pulse count: 2
Triplet count: 5
Gaussian count: 0

This WU has a lot more numbers than I usually see. But I have no idea what is normal or significant. I read the definitions on the wiki, but it doesn't talk about numbers at all. Is this result in any way significant? What values are?

http://setiathome.berkeley.edu/result.php?resultid=588097991

Same question for this one:

Spike count: 1
Pulse count: 2
Triplet count: 13
Gaussian count: 1

http://setiathome.berkeley.edu/result.php?resultid=578629128

Or this one?

Spike count: 0
Pulse count: 1
Triplet count: 2
Gaussian count: 5

http://setiathome.berkeley.edu/result.php?resultid=584460219

Dublin, California
Team: SETI.USA
ID: 618404 ·
zombie67 [MM]
Volunteer tester

Joined: 22 Apr 04
Posts: 758
Credit: 27,771,894
RAC: 0
Message 618519 - Posted: 13 Aug 2007, 8:45:56 UTC

okay, looking at the stats...

1 post
43 views

Many people are interested! Not so many with answers...
Dublin, California
Team: SETI.USA
ID: 618519 ·
W-K 666
Volunteer tester

Joined: 18 May 99
Posts: 19060
Credit: 40,757,560
RAC: 67
Message 618533 - Posted: 13 Aug 2007, 9:25:12 UTC

As the signals we are looking at are what we would normally call white noise, as heard on TV or FM radio when not tuned in or transmitter is off, then I assume the number of each in a work unit is completely random. Each spike etc. only becomes significant if it is above a predetermined size or it is persistent (see Near Time Persistency Checker)and seen in most units that look at the same pinpoint of sky.

Andy
ID: 618533 ·
Sutehk
Volunteer tester

Joined: 11 Jun 99
Posts: 42
Credit: 1,443,674
RAC: 0
Message 618568 - Posted: 13 Aug 2007, 11:48:31 UTC

If your willing to read this article it has a very good explanation of the project and what each of the counts are. Doesn't really give average numbers for each. It was written by the heads of seti@home.

Hope this may help a little.

 spelling
ID: 618568 ·
Josef W. Segur
Volunteer developer
Volunteer tester

Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
Message 618629 - Posted: 13 Aug 2007, 16:43:54 UTC - in response to Message 618404.

...I have no idea what is normal or significant. I read the definitions on the wiki, but it doesn't talk about numbers at all. Is this result in any way significant? What values are?
...

As Andy pointed out, we're dealing with a random noise situation. That's not only within a single WU, but also work recorded at different times. There are thresholds passed in the WU header which define what the application should consider a good enough signal to report, and those are set IIRC to give about a 50/50 chance of reporting one false signal of each type when the noise is at it's typical levels. While the app is being tested on Beta, data is gathered and the thresholds are adjusted until the results approximate that goal. Additional trimming can be done at any time if needed.

The long term effect can be seen on the Science status page. For the average result which has been entered into the Master Science Database, there have been 1.346 Spikes, 0.827 Gaussians, 0.866 Pulses, and 0.890 Triplets. I think the reason Spikes are higher is that noisy (result_overflow) WUs more often overflow on Spikes than the other signal types.
`                                                                 Joe`
ID: 618629 ·

Message boards : Number crunching : spike/pulse/triplet/gaussian counts