Re: Ensemble analysis: "new variables"

[<aah@nofs.navy.mil>]

Andrew, my automatic ensemble pipeline does not use
the 60 brightest stars, but some smaller number of well-exposed stars.
When you use the brightest ones, they tend to be saturated
and usually will not give good results (even the centroid
tends to get messed up, which may be why you get such
poor matches there).  In addition, if they are bright in
one passband, they tend to be even brighter in the other
passband and the saturation will often pull (V-I) out of
range.
  A well-exposed star is one that is no more than half-well.
Within a half degree (I assume that is radius, so you are
talking a degree-sized area) you typically have 300 Tycho2
stars, all of which should be well-exposed.  I don't know
why an average of 60 such stars should give you poor
statistics; sqrt(60) is a big factor and they should all
have Poisson statistics under one percent.
Arne