Re: Are the coords for pre-existing candidates calculated differently (and therefore correctly)?

From: Robert A. Knop Jr. (robert.a.knop@vanderbilt.edu)
Date: Sun Apr 04 2004 - 22:30:51 PDT

  • Next message: Vitaliy Fadeyev: "Re: I-band subtractions"

    On Sun, Apr 04, 2004 at 10:26:52PM -0700, Saul Perlmutter wrote:
    >
    > Hi Rob,
    > When we were hunting for the correct coordinate for #049, at one
    > point we tried going to "pre-existing candidate" tiles while looking at
    > the original search. Surprisingly a pre-existing candidate tile came up
    > listing exactly the *correct* GOODS Ra&Dec coordinates AND the correct
    > pixel x&y coords (as opposed to the incorrect RA&Dec coords with which
    > #049 was saved) -- and of course it didn't know that it was near #049.
    > We saved this as #174, for the record. So the question is: Is it
    > possible that the RA&Dec coords are calculated differently (and
    > correctly) for the *pre-existing* candidates based on the pixel x&y
    > coords, even when they were calculated incorrectly the first time the
    > candidate was found?
    >
    > If so, this might be a much easier way for us to find the correct coords
    > for candidates that got wrong RA&Dec's last night -- although I doubt
    > there will be more to do of those.

    This will *usually* work.

    Here's what it does.

      * Ask the database which candidates have RA and Dec within the range
        included in the subtraction.

      * For each of those, ask the database for the transformation from the
        refsys of the existing candidate to the refsys of the
        transformation.

      * Transform the pixel (x,y). RA and Dec does not enter into this
        calcuation at all.

      * Calculate scores

      * Add this to our list of pre-existing candidates

    This means a few things. First, if I haven't loaded a transformation,
    you don't get the pre-existing candidate. (That explains why you see
    all those warnings about the adam03 candidates when you start up
    searchscan.) It also means that it can find pre-existing candidates
    even if the RA and Dec in the database are wrong, so long as the RA and
    the Dec in the database are within the range subtended by the image.

    This is all much too much information about the internals of course.

    -Rob 

    -- 
--Prof. Robert Knop
  Department of Physics & Astronomy, Vanderbilt University
  robert.a.knop@vanderbilt.edu

    This archive was generated by hypermail 2.1.4 : Sun Apr 04 2004 - 22:31:14 PDT