[SciPy-Dev] Problem with N-dimensional interpolation using a new griddata function for N>=3

Pauli Virtanen pav@iki...
Tue Sep 21 16:46:26 CDT 2010


Tue, 21 Sep 2010 20:35:06 +0200, Adam Machnik wrote:
[clip]
> griddata() function. I hope that it will work for me when fixed these
> issues with the handling of degenerated
> zero volume cases on interpolation level.
[clip]

I believe it is fixed here:

    http://github.com/pv/scipy-work/tree/bug/griddata-nans

Click the "download source" link to get a fixed tarball. If you can test 
this, that would be useful, since I cannot reproduce the issue you are 
seeing.

If you want to help further, please also do

        >>> from scipy.spatial import Delaunay
	>>> tri = Delaunay(points)
        >>> numpy.savez('delny.npz', tri.__dict__)

and send the 'delny.npz' file to me -- this way it should be possible to 
get a reproducible test case.

    ***

If it is indeed fixed now, the problem was that the directed search for a 
simplex could stop at a degenerate one, in which barycentric coordinates 
and linear interpolation is not defined -- which then would produce nans.

-- 
Pauli Virtanen



More information about the SciPy-Dev mailing list