RE: Minimum spanning circles and Voronoi diagrams

[Tom Droege <tdroege2@earthlink.net>]

Let me try to discuss this another way.

We have lots of measurements of each star.  Blended stars are a special 
case.  Blended stars are bound to be of poor measurement quality.  I think 
just the subtraction will introduce a SQRT(2) into the error.  Then there 
is the question of how to do it perfectly.  So another factor.  Perhaps we 
can detect gross variations, but they will always be suspect.

For stars that are not blended, we still have problems depending on other 
factors.  Passing clouds, satellite tracks, cosmic rays, bad pixels, etc..

For these, we want to make a cut that removes the questionable 
measurements.  We have lots of measurements of each star, so we can afford 
a large cut.

Well, it is not so easy.  There is always a blended star in there 
somewhere, or a galaxy, or ...  Are there any 7" squares of sky that 
contain nothing?  But for most stars other stuff will be well down from our 
measurement.

Tom Droege