[SciPy-dev] interp2d and sandbox/delaunay

[Travis Oliphant]

John Travers wrote:

>Hi all,
>
>As you may have guessed from my noise (tell me to shut up if i'm
>irritating), i'm taking a serious look at the state of
>scipy.interpolate
>  
>
We are all very glad you are looking at it.  In my mind, 
scipy.interpolate is very weak and needs help.

>As previously stated, I think it needs to be reorganised/split into
>several parts, but I'll leave that to other threads.
>
>In this post I'm wondering about the status of interp2d.
>  
>
interp2d is not at all what it should be.

>At the moment it is based on netlib->dierckx->surfit. This is bad as
>this routine should not be used for interpolation. I think others have
>had problems with this code (reading about a year ago on scipy-user) -
>it segfaults mine.
>  
>
Please change it from using surfit.  It shouldn't.

>I have recently implemented netlib->dierckx->regrid. Which
>interpolates nicely for regular rectangular grids -> this could be
>used in interp2d, except that it doesn't work on scattered data (is
>this required? I think so yes).
>  
>
Yes, interp2d should work on scattered data, I think.

>The solution would be to call regrid for regular data and
><some_other_method> for scattered data.
>
>The best <some_other_method> appears to be delaunay triangulation
>followed by natural neighbour interpolation or regrid. (this is what
>MATLAB does).
>
>So my questions are:
>
>1. can I fiddle with interp2d and the other scipy.interpolate stuff?
>2. what is the status of the sanbox/delaunay stuff (why is it not in
>the main scipy) if there is a problem with it, can I package a
>different package that does this?
>  
>
It's just the review process that has kept it out, I think.   Ask Robert 
(if he's not already preparing his response).


-Travis