Re: Delaunay and Boundaries

• To: Kevin Manross <kevin.manross@xxxxxxxx>
• Subject: Re: Delaunay and Boundaries
• From: Bill Hibbard <billh@xxxxxxxxxxxxx>
• Date: Wed, 14 Jan 2004 13:51:13 -0600 (CST)

Hi Kevin,

> I'm trying to plot values at specific points between to boundaries.  If
> you look at:
>
> http://www.cimms.ou.edu/~kmanross/VCPRPE/VisADSamples/Gridded2DSet.gif
>
> You can see that I have been able to get the points and set them up in a
> Gridded2DSet(MathType, samples[][], NUM_POINTS, 1).  (Note this area is
> not "filled" - if you zoom in, you will see individual points.)

This Gridded2DSet is sort of degenerate, as a grid of
NUM_POINTS x 1.

> Since this is an irregular shape, I also tried creating an
> Irregular2DSet from the same points.  The Irregular2DSet follows the
> bottom boundary fairly well, but does not do so well with the top boundary.
>
> Please see:
>
> http://www.cimms.ou.edu/~kmanross/VCPRPE/VisADSamples/Delaunay.gif
>
> Why is there such a discrepancy between the way the Delaunay fits to the
> bottom boundary and the way it doesn't fit the top boundary?

The automated Delaunay construction fills the convex hull of
your points, hence draws a line across the concave top curve.

> I would like to
>
> A) Get the Irregular2DSet to follow the top boundary as well as it does
> the bottom

To do this, you need to construct a DelaunayCustom with your
own triangulation. If you can think of how you would draw a
bunch of triangles filling the region you want, then you can
do this.

> or
>
> B) Possibly use a Gridded2DSet (man dim = 2).
>
> Can I do B given the irregularity of the data.  (I have 28700 points
> arranged in 400 columns = 71.75 rows?)

You'll need a non-degenerate 2-D grid (i.e., neither factor = 1).
Perhaps you can add some dummy points to make it a real 2-D grid?

Good luck,
Bill

• References:
  • Delaunay and Boundaries
    • From: Kevin Manross