A plehtora of Quantities
dsb at ast.man.ac.uk
Fri May 14 04:24:46 PDT 2004
> - Angle: naturally a ratio of arc length to radius; if you measure both
> lengths in the same unit, then the ratio doesn't need units; degrees of arc
> are a perversion (and _so_ third-millennium-BC :) ).
> I agree that we have to use a unit for angle since there is so much historical
> bad practice. However, inside the VObs, where the machines live, we could use
> implied radians throughout and only translate to pervy units for display to
> I suggest "count" and "ratio" for the others. I don't think that putting
> quasi-units like "CCD" in a Q is going to work; better to write "count" and
> infer the quasi-unit from context.
I could live with that - the Frame could contain other meta-data saying
"we are counting CCDs here" - possible a UCD?
> BTW: "pixel" is not a count but a length in arbitrary units.
I would need to disagree here I think. True - a pixel has a size, but
then so do forests and you wouldn't want to measure distance in "forests".
The point is that the size of a pixel (and a forest) is variable, and so
does not correspond directly to a measure of length. What you do is you
use your WCS information to map a pixel "count" into a physical "world
coordinate". [Of course you can gave fractional pixel counts as well!]
Just for the oldies amongst us, here's a bit of fortran:
DOUBLE PRECISION FRED( 10 )
DO I = 1, 10
WRITE(*,*) FRED( I )
What are the units of I? "Pixels"?
More information about the dm