Re: NaN and Surf function

[David Bateman]

Kyle Snow wrote:
> My question is regarding the handling of NaN values for the Z-coordinate
> in the function
> 
>  
> 
> surf(x,y,Z).
> 
>  
> 
> These values exist because my gridded data set has an irregular
> boundary. The surf command does not seem to handle this well. In matlab,
> these data points are simply not plotted, and the surf plot for the
> non-NaN points looks fine. Is there a way I can get this to work using
> Octave’s surf function?
> 
>  
> 
> Note that I am running WindowsXp, and I downloaded the exe file from
> 
> http://superb-east.dl.sourceforge.net/sourceforge/octave/octave-2.1.73-1-inst.exe.
> 
> I have not obtained the Octave source code.

There is a difference between octave 2.1 and matlab, in that the NaN's
are removed from the data before plotting in octave, whereas matlab uses
them to flag unconnected line segments. Octave 2.9 was adapted to have
matlab's behavior. The change is not too major, but does mean a certain
level of backporting from 2.9 to 2.1 and a recompile.. Better to go to a
2.9.x release if you can. Sorry I still haven't gotten back to my mingw
2.9 release, but will one day (if someone doesn't what to take it off my
hands), so you might be limited to linux releases of 2.9 at the moment...

D.