Minutes for UA MOPS meeting, April 13, 2006
In attendance:
Tim Axelrod, Kobus Barnard, Alon Efrat, Scott Morris, Bongki Moon, Andy
Connolly (via telecom).
Main points:
1. Andy updated us on U.Pitt/CMU activities. They are running tracklet finding
on full density data. This is 10 times more detections per image than
previous studies. The false detection rate is now set at 50%. They report
that they are running into memory problems as a result of the increase in
density. This issue is under investigation.
2. Tim updated us on the status of the memorandum of understanding between LSST
and PANSTARS. Andy suggested that the agreed upon 1% of image data, should
perhaps be checked or amended to ensure that a certain percentage of certain
parts of the sky could be available as some of the LSST (perhaps it 100
square degrees? I did not catch this). Regardless, we are all keen on
collaborating with PANSTARS.
3. Kobus explained phantoms better. We introduce phantom orbits into graph
matching to give false detections a place to go, and we introduce phantom
detections to include our knowledge about expected missing detections for
various orbits. This discussion was in the context of evaluation of graph
matching verses the greedy algorithm which requires proper handling of
phantoms; otherwise one is assuming that something that is likely to be
noise is very unlikely to be observed, simply because it is not near any
known orbits. Kobus recommended that we included sufficient phantoms so that
every orbit and every detection can, in theory, be matched to a phantom.
4. We discussed possible content for the upcoming SPIE paper.
--
Kobus Barnard
[EMAIL PROTECTED]
http://kobus.ca
730 Gould Simpson, U. Arizona (520-621-4237)
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
