From mike@ast.cam.ac.uk Fri Jul 12 10:53:25 2002 Date: Wed, 10 Apr 2002 13:42:22 +0100 (BST) From: Mike Irwin To: Richard McMahon , Weimin Yuan , Simon Hodgkin Subject: fitsio_merge I altered the integer classiciation scheme to allow a bit more flexibility in interpreting merge output. The integer flags are now -1 stellar +1 non-stellar 0 noise-like -2 non-stellar classified objects that are within 3-sigma of the star locus -3 non-stellar classified objects that are 3 to 5-sigma from the star locus I may add more categories to this scheme later as the need arises. This new schema reflects the fact that the classifier is (ideally) Bayesian and up until now I have been assuming the equivalent of a deep high latitude prior (ie. most faint objects are galaxies). Certainly for globular cluster work or things like the halo of M31 or nearer to the Galactic Plane, most faint objects could well be stars and require a prior reflecting this fact. It is also the case that the -2 category even in high latitude fields could be a useful adjunct since it flags objects that are probably compact non-stellar sources such as AGN or compact high SB galaxies. Mike