TASS_DBMS_11JAN98.tgz posted

[Chris Albertson <chrisa@wavenet.com>]

Hello,

I posted the latest database software suite to 
ftp.tass-survey.org/pub/incoming/TASS_DBMS_11JAN98.tgz  
No, I'm not done with it yet. 
I posted it just in case anyone wanted to see
what I'm doing or look at the latest table definitions.
I have converted over to milli-arcseconds and integer
Julian date/time as well as other minor changes.
Documentation and READs will come later, so e-mail me
if you want to find something.

I have reached a minor milestone.  The program will process
test data into a collated table.

I'm calling this a "development snapshot", not an alpha
or pre-release or anything as far along as that.  It runs
on some test data I have.  I hand picked 17 of Glenn's
data files that I knew would have overlapping coverage.  The
program seems to find the stars in common between these files.

If you are developing a Web interface this software will
generate good enough test data and is quick enough that
you can wait while it runs.  I do not intend to change the
structure of the database further except to correct an error.
The changed table have new names so they will not over write
existing tables.  Name changes are as follows:
  tass_catalog --> tass_cat
  observetion  --> observ 

I have purposely left out dealing with the RA=0 problem and
frame overlap.  So this is not ready for production use.

Other questions:  Does it find all the matches it should?
Does it match those it should not?  Does it blow up or
take all day if you give it a million line data set?  I
don't know yet.

Here is a kind of histogram of the collated "master catalog"

  Num. Obs.  Stars 
    1  26531
    2  6919
    3  2425
    4  1591
    5  769
    6  962
    7  189
    8  5
    9  0

Notice that there are non zero counts for Num Obs = 7 and 8.
All my test data came from only two nights and Glenn only has
three cameras.  The extras are from overlapping frames and
noise-on-noise correlation.  Glenn was running at a sigma=2 
cutoff.  The above run took about ten minutes on a P100 but
run times now do not reflect production run times.

Looks like I'm in the market for a bigger disk now as I'll
need one to do larger scale tests.

-- 
   --Chris Albertson             home: chrisa@wavenet.com           
     Redondo Beach, California   work: chris@topdog.logicon.com