[R] Boundaries and deldir

[Rolf Turner]

On 02/02/15 16:26, p_connolly wrote:

> Just what is meant by dummy points as referred to by the help for the
> deldir() function?  I understood they indicated the boundary beyond
> which triangulation would cease.
>
> I thought I would need the x/y elements (as described in the help file
> at the end of the description of the use of the dpl argument) to
> describe ad hoc dummy points as way to define a polygon or two as a
> boundary.  However, it gives this error:
>
> Error in xd[-drop] : only 0's may be mixed with negative subscripts
>
> Something internal is doing the negative subscripting.
> I tried ndx/ndy instead of x/y but it evidently refers only to a
> rectangle so not what I need.
>
> Am I barking up the wrong tree altogether?  Is the boundary defined
> somewhere else entirely?  I need to get that clear before I am able to
> provide
> useful example code.

The dummy points have nothing to do with any "boundary".  In fact they 
have nothing much to do with anything, really! :-)  They are a hangover 
from the original purpose of deldir which was to assist in a numerical 
integration needed for the maximum likelihood estimation of the 
intensity function of an inhomogeneous Poisson process.  I really should 
get rid of them, but that would require a bit of work and re-writing of 
code and help files, and they do no real harm so I have decided to apply 
the "If it ain't broke don't fix it." principle.

The deldir function creates a Delaunay triangulation/Dirichlet 
tessellation inside a "rectangular window" (denoted by "rw" in the 
argument list).  This is the only boundary invoked or involved.

The function plot.tile.list() will *plot* the Dirichlet tessellation 
"clipped" to a specified polygon.  But that is just for *plotting*.  I 
am not sure that I really understand the idea of a "boundary beyond 
which the triangulation ceases".  The Delaunay triangulation is a finite
structure; its outer boundary is the convex hull of the set of points 
being triangulated.  You cannot confine it to a smaller region without 
losing some of those points.

If you can explain what you really want to do, perhaps I can help.

cheers,

Rolf

-- 
Rolf Turner
Technical Editor ANZJS
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276
Home phone: +64-9-480-4619