DS24 analysis

[Andrew Bennett <andrew.bennett@ns.sympatico.ca>]

Tom sent me 3 sets of the 7x7 data. I have run
them through my PSF fitting routines, after a
struggle. Preliminary results:

First the good news.

Putting the same star in 49 different places on
the CCD gives lots of variation in the magnitude
estimate: my Ensemble calibration program takes
a lot of it out again. This is exactly the same
Ensemble code that was previously driving me to
distraction by failing to make the slightest
impression on large and obviously systematic
errors in previous data sets. The sytematic
coverage of the entire CCD gives the Ensemble
code something to get its teeth into ...
the improvement in W-S statistic is impressive.

And then the rest of the news.

New data set = new bugs. For your amusement here
are some of them.

My routine to read in the previous estimate of the
astrometry inadvertently also set the base magnitude
(about 21) to zero. Then the program read the previous
estimate of the base magnitude ... if the program had
never got as far as estimating one, there wasn't one to
read so this value zero was used instead. All the
magnitudes then came out low by 21 magnitudes and none
matched the catalog. I have no idea how I previously
failed to hit this bug!

I recently "improved" my error estimates in RA by
putting Cos(Dec) factors in appropriate places ...
except that one of them was a Cos(RA) factor instead.
This did horrible things for RA>90 degrees as the
code assumed that negative errors just didn't happen.
Took 2 days to find out why the program refused to
process the 3rd data set (RA 105 to 113). Some of the
funny things I found on the way to fixing this bug
are still unexplained.

Any significant computer program contains at least
one bug.

*****************

There's still lots to think about. The PSF varies quite
a bit from one image to the next and I'm not sure if the
whole set should be analysed with the same average PSF.
Obviously, there is a trade-off here. Using the same PSF
is "more consistent" but the changes are, I think, real
and must affect the photometry.

Andrew Bennett, Avondale Vineyard