[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Soliciting criticism (Yikes!)
Sorry for the jumbled up mess. My PC has a hardware problem.
I suspect an intermittent memory error in the L2 cache.
The result was a hashed e-mail being sent and a crashed PC.
It runs OK for days then crashes twice within an hour.
What I intended to say was that sps2x's PSF fitting looks
to produce a graph with better slope, I think because it
can make use of energy out in the wings of the PSF or in
other words "more pixels per star". The Mk IV's
PSF is not a figure of revolution and it is also not constant
over the image. But SPS assume it is both. I think this is
a source of scatter: the stars fall on different parts of the
frame and have a different PSF each time they are seen. What
is needed is a better PSF model. One the uses a non-constant,
two dimensional PSFs.
If I am correct then processing data from just the central
portion of each image should have less scatter, but you will
have fewer observations of each star.
Chris Albertson wrote:
>
> Just looking at the thumbnail pictures of the two graphs
> I can see the slopes are diferent. I think you have a good,
> graphic and intuitive way to compare data reduction pipelines.
> You want all the green and red points to lie on a straight
> horizontal line on the bottom of the graph.
> sps2x uses,sps2x uses, I think PSF fitting while sextrator does aperature
> photometery. I think PSF fitting while sextrator does aperature
> photometery.
> One thing required by the psf fitting technique is a good psf
> model.
>
> "Creager, Robert S" wrote:
> >
> > > Hey folks,
> > >
> > > I'm hoping some of you have the desire to critique, and offer suggestion
> > > about the graphs here: http://robertcreager.mystarband.net/ . The are all
> > > from Tom's data, night 2157 (from a few weeks ago). As the graph titles
> > > imply, one is reduced from source extractor, and one from sps2x.
> > >
> > > Speaking of source extractor, does anyone have a set of values which might
> > > be appropriate for reducing Mark IV images? I grabbed Glenn Gombert setup
> > > from 1996, but at this point, I'm shooting in the dark as to how to
> > > improve the reduction. If anyone wants to direct, I'll be happy to make
> > > changes and keep re-running the reduction. It takes about 4 hours from
> > > start to finish for reducing night 2157, so I can get two runs per day at
> > > home.
> > > The basic steps are:
> > > * Create master dark and flats from median data
> > > * Adjust the object data by: mean * (object[i][i] - dark[i][j] ) /
> > > flat[i][j] where mean is the mean of a central box of the flat field
> > > * Do the reduction (sps2x, source extractor) in V-I pairs
> > > * Match the reduction data to transform into ra/dec
> > > * Try to match both to subset of Tycho2 data
> > > * If both match, use in next step
> > > * If only one matches, match other band to the matching one to
> > > calculate ra/dec
> > > * Collate the V-I pairs from previous
> > > * Solve for magnitude adjustment from collated data
> > > * Use stars which match the Tycho2 subset
> > > * Fit a line to the raw color (V-I) vs. magnitude delta (Tycho2 -
> > > measured) for each V and I
> > > * Iterate over the line fit, removing 4 sigma data, until either no
> > > more data was removed, we've removed too much data (10% removed) or the
> > > RMS error is below 0.05
> > > * Import star list into database, and match stars within 0.004
> > > degrees.
> > > * Produce graphs included from the previous data
> > > The majority of the stars in the graph have 30+ measurements.
>
> --
>
> --
> Chris Albertson
> Redondo Beach, California
> home: 310-376-1029 chrisalbertson90278@yahoo.com
> cell: 310-990-7550
> office: 310-336-5189 Christopher.J.Albertson@aero.org
--
--
Chris Albertson
Redondo Beach, California
home: 310-376-1029 chrisalbertson90278@yahoo.com
cell: 310-990-7550
office: 310-336-5189 Christopher.J.Albertson@aero.org