[SciPy-user] Memory leak in delaunay interpolator

Robert Kern robert.kern@gmail....
Wed Jan 16 19:22:53 CST 2008


william ratcliff wrote:
> Would there be any interest/problem with allowing error calculations for 
> interpolations done with the Delaunay package?  For example if one could 
> send in x,y,z,zerror=myerror (that is, if it's a keyword argument, it 
> won't break existing code) and if zerror is defined, also return an 
> errorbar (calculated in quadrature) based on the neighbors/areas used in 
> the interpolation?

You can do exactly this calculation with the current API. Just interpolate 
z=myerror**2 and take the sqrt() of the result. I would prefer to keep the API 
small and composable rather than try to implement everything that's possibly 
useful. In particular, I dislike changing the number of return values based on 
the inputs. It makes code harder to read and generic code harder to write.

Feel free to add to the documentation pointing out recipes like this, though.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco


More information about the SciPy-user mailing list