STC and Characterisation ranges

[Arnold Rots]

Anita,

Yes, this is an issue that I have pondered and I was very tempted to
add the capability, basically for the reasons you quote.  I didn't do
it since the substitution group polymorphism mechanism is rather
ponderous, but I will be happy to give it another look.

Just a comment, though.  The distinction between whether a particular
item aloows a range or is single valued isn't as arbitrary as you
suggest.  The values of coordinate elements are
single-valued. Anything that specifically refers to a a volume is
specified by a range.

  - Arnold

Anita Richards wrote:
> 
> Ranges in data specifications - Observation/Characterisation and STC:
> 
> At present many elements are allowed to have one or no values but
> others can have a range of values.  I would like to see more things
> take ranges unless there is some problem I have overlooked.  It will
> add a little complexity but one way round this would be to make max
> and min an optional sub-class in addition to central value and
> uncertainty.
> 
> For some things like the coordinate system or the UCD for an element
> this obviously has to be single-valued if any value is given.
> 
> However in other cases it is not clear why some elements are
> restricted to being single-values and others can have sub-elements or
> multiple values;
> 
> e.g. the wavelength range can have max and min but the pixel size is
> fixed.
> 
> However, for example the exact angular resolution and pixel size of an
> image extracted on demand from visibility data is not known until the
> request is made, and maybe I have just come across peculiar x-ray data
> but in quite a few examples I have used the pixel size varies within a
> single
> image (smaller pixels in low signal-to-noise areas are binned in 4's,
> 16's etc. to improve signal-to-noise). This is particularly important
> for e.g. virtual images from visibility data where the user needs to
> select the desired resolution or pixel size (e.g. the NVSS cutout server
> http://www.cv.nrao.edu/nvss/postage.shtml )
> 
> It would be useful to be allowed to have a max and min here, also.
> This cannot be done using uncertainties since we do know exactly how
> large the pixel size(s) are in any one real image, but either the
> image is virtual until requested or is varied.  There is a separate
> situation where one might not know the pixel size in arcsec to high
> precision e.g. in an old digitised image with serious edged
> distortion; here uncertainty is appropriate.  Analogously, the uncertainty
> on the centre of a wavelength interval is not the same as the width of
> that interval.
> 
> Hope this makes sense
> 
> If no-one agrees maybe someone can suggest an alternative means to
> describe the sort of data with varaible pixel size ec. as described above
> in order to allow users to access the data to their exact specifications
> (as in cut-out servers which allow resampling for example and other
> virtual data).
> 
> Looking forward to using the models!
> 
> Anita
> 
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> Dr. Anita M. S. Richards, AVO Astronomer
> MERLIN/VLBI National Facility, University of Manchester,
> Jodrell Bank Observatory, Macclesfield, Cheshire SK11 9DL, U.K.
> tel +44 (0)1477 572683 (direct); 571321 (switchboard); 571618 (fax).
> 
--------------------------------------------------------------------------
Arnold H. Rots                                Chandra X-ray Science Center
Smithsonian Astrophysical Observatory                tel:  +1 617 496 7701
60 Garden Street, MS 67                              fax:  +1 617 495 7356
Cambridge, MA 02138                             arots at head.cfa.harvard.edu
USA                                     http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------