Bug report: Graphics`ArgColors`
- To: mathgroup at yoda.physics.unc.edu
- Subject: Bug report: Graphics`ArgColors`
- From: lsf at holmes.astro.nwu.edu (Sam Finn)
- Date: Sat, 19 Oct 91 15:32:36 CDT
Mathematica 2.0 Standard Package Graphics`ArgColors`
L. S. Finn//lsf at holmes.astro.nwu.edu
License L2086-7328
There are (at least) three (count'em) bugs in this package.
1) ArgColor[0] evaluates to an RGBColor[]; otherwise ArgColor[] evaluates to
a Hue[]. This can produce a pcolor error if the surface you are shading
passes through zero.
2) The domain of Hue[] is [0..1], while the range of Arg[] is [-Pi, Pi].
Thus, the mapping that ArgColor and ColorCircle provide from complex z
to Hue[] is, in principle, undefined for half the complex plane. In fact,
Hue seems to do the right thing, but this should not be counted on. Note
that ArgShade is properly protected against this sort of thing.
3) The rule for non-zero arguments to ArgColor[] has a typo: it is
ArgColor[z_] := ColorCircle[ Arg[z] ] /; NumberQ[N[arg]]
and it ought to be
ArgColor[z_] := ColorCircle[ Arg[z] ] /; NumberQ[N[z]]
Recommended replacements for ArgColor[] and ColorCircle[] follow below:
(* don't mix Hue[] and RGBColor[] *)
ArgColor[z_] := Hue[1,0,1] /; z == 0.0
(* typo in package *)
ArgColor[z_] := ColorCircle[ Arg[z] ] /; NumberQ[N[z]]
(* Adjust range of ColorCircle[] to match domain of Hue[] *)
ColorCircle[arg_, light_:1.0] :=
Hue[ N[Mod[arg/2/Pi,1]], 1, light] /; NumberQ[N[arg]]
Sincerely,
L. S. Finn