Hi Dave, I would imagine that the astrometric information is generated when an object is created from many detetctions. There will be some variability in the objects RA and dec for its various detections and this variability should be fit for proper motion and parallax as well as just determining some S/N weighted mean position. I think objects get created at the archive by doing some giant joint fit to a stack of images.
Kem > Maybe I have missed the point of Jeff's recent e-mail on DM Use Cases, but > my "vision thing" for the astrometric columns in the LSST database is that > all columns are filled for all stars as part of the Observatory > processing. > That is to say, we will solve for (position, proper motion, parallax, and > astrometric wiggles) for all stars, and put the results (significant or > not) > into the DB. So I don't expect queries like "compute the parallax for all > stars with g-r colors in the range of -0.5 to 2.0". > > But then again, I am not sure (i.e., I haven't read the document) on how > or where such processing will take place. The part of the pipeline that > keeps up with the camera isn't the place to do this. Presumably that > pipeline has a calibration model and produces the stream of detections. > Some other task will have the responsibility of cycling through the > entire DB and doing processing on the ensemble of measurements for > all objects. The astrometric task seems similar to the photometric task > of detecting variability and trying to characterize it in terms of known > objects (eg. is this a Cepheid? an RR Lyr?, etc.) > > It would seem that a crawler (or whatever the correct word may be) that > goes through the DB in an optimized manner would be far more efficient > that having this done at the user query level since we know that > obnoxious users (i.e., me) will make such a request. > > -Dave > _______________________________________________ > LSST-data mailing list > [email protected] > http://www.lsstmail.org/mailman/listinfo/lsst-data > _______________________________________________ LSST-data mailing list [email protected] http://www.lsstmail.org/mailman/listinfo/lsst-data
