Mark IV reduction pipeline, release 0.2

[Stupendous Man <richmond@a188-l009.rit.edu>]

  I've been working on my Mark IV reduction pipeline over the
past few weeks.  Since I'll be presenting a poster at the upcoming
meeting of the American Astronomical Society meeting in Pasadena
(June 2 - June 7; let me know privately if you'll be attending!),
I figured it was time to stop playing with the software and
start using it to reduce Disk Set 18.  

  You can find documentation explaining the structure of the
pipeline, the packages which constitute it, and the way it works,
at

     http://spiff.rit.edu/tass/pipeline/pipeline.html

That document describes where you can download all the software
you need.  Here's a terse recap:

        http://spiff.rit.edu/tass/bait/bait.tar.gz
        http://spiff.rit.edu/tass/match/match-0.3.tar.gz
        http://spiff.rit.edu/tass/photom/photom.tar.gz
          http://freshmeat.net/redir/ccmath/1083/url_tgz/ccmath-2.2.0.tar.gz
        http://spiff.rit.edu/tass/pipeline/pipeline.tar.gz
        http://spiff.rit.edu/tass/xvista/xvista.tar.gz

  If you are using a Linux system and like to live dangerously,
you could simply try to download the entire contents (including all
subdirectories) of

        http://spiff.rit.edu/tass/

which include all the executables.  I doubt this will work as well
as one would hope, though.

  You'll also need a reference astrometric and photometric catalog;
I have created one based on Tycho-2:

        http://spiff.rit.edu/tass/tycho

  There is a small subset of images from Data Disk 18a you can use
as a testbed.  The format of the FITS headers in these images is
significantly different than that in older images, so the new 
pipeline should be tested on these new ones.  I've placed 5 dark
frames in V and I, and 7 night-sky images in V and I (for a total
of 10+14 = 24 images) into 

        http://spiff.rit.edu/tass/input

Since each image is about 8 Meg, this comes to about 200 Meg total.
On my machine (670 MHz PIII, 512 Meg RAM), it takes about 24 minutes
for the pipeline to run through the images.

  If you do run the code on these test images, compare your final result
to mine.  I find a photometric solution (in file Mhra2011797.coeff)

    V=V,(V-I)  N    166  a  -8.603  0.032  b  -0.182  0.029   RMS  0.124 
    I=I,(I-V)  N    166  a  -8.258  0.024  b  -0.036  0.021   RMS  0.102 

  Comments, suggestions, bug reports, fixes are all welcome.

----------------------------------------------------------------------------


                     A Cry for Help! 

  In the XVista package, the file "gaussFit.c" contains a routine 
which tries to fit a gaussian to a pair of arrays with "x" and "y"
data values (with uniform weighting), such that

                                            (x - m)^2
                            y   =  A exp [- --------- ]
                                              2*s^2

where
            A  = amplitude of gaussian
            m  = center of gaussian
            s  = width of gaussian

  I wrote this routine myself, back in the dark ages.  Recently, I 
discovered that it does a less than stellar job.  Can someone provide
me with a better one?  I would like

        - a self-contained source code file, so I don't have to 
                    include hundreds of kilobytes of extra code
                    from some numerical library just to do this one task

        - good documentation, both within the source code itself 
                    and as a README or big comment at the top
                    explaining the general method of solution

        - the ability to turn on/off a flood of diagnostics with
                    a simple #define

        - code written in vanilla ANSI C -- though I'll accept code
                    in any language, as long as I can understand it

                    
  If you can send me such a routine, then (with your permission)
I'll put it into XVista and distribute it with the next version.
I would also very much appreciate pointers to good references for
algorithms one can use to solve this non-linear problem.

  Thanks very much!

                                          Michael Richmond