Re: INTERP2 / GRIDDATA : problems using those functions

[Pierre Baldensperger]

Hello,

Thank you very much, Gottfried! This indeed solves the "griddata" 
problem and
the result looks OK. Never encountered this transposition problem 
before. It's
something worth remembering!

Concerning the "interp2" problem, I don't think it would make sense to
transpose any of the arguments. However, I had a quick look at the source of
"interp2.m" and it looks to me like the algorithm assumes XI and YI are in
"meshgrid" format, which is not the case in my example, and which is 
also not
clearly specified in the routine's help messages. In my opinion, requiring
"meshgridded" XI and YI would seriously restrict the field of application of
that function, so it is likely to be a mistake...

I'm not sure I have fully understood how it works, but the following 
calls to
the "lookup" function only consider the first line (resp. first column) 
of XI
(resp. YI), as if the other lines (resp. columns) were identical, which may
not always be the case (unless they are in "meshgrid" format).

xidx = lookup(X(1, 2:end-1), XI(1,:))' + 1;
yidx = lookup(Y(2:end-1, 1), YI(:,1)) + 1;

Same strong suspicion for the subsequent lines (including "nearest" method).
Anybody out there has an idea how to fix it? Do you know how I can 
contact the
author / maintainer of that routine, so I can submit this problem to him
directly?

Thanks again,
-Pierre.



Gottfried Necker wrote:

> Hello,
>
> I don't think your octave installation is corrupt. I can see the same 
> problems. For the griddata case I would suggest to transpose the first 3 
> arguments to griddata, i.e.:
> posx_1d=reshape(posx,size(posx,1)*size(posx,2),1);
> posy_1d=reshape(posy,size(posy,1)*size(posy,2),1);
> funcz_1d=reshape(funcz,size(funcz,1)*size(funcz,2),1);
>
> Using these lines, the problem with cross goes away. But I don't know, whether 
> it gives the correct result.
> If I remember correctly  octave sometimes returns/expects  a row vector where 
> matlab returns a column vector and/or vice versa. This might also cause the 
> problems with the interp2 case, but I don't know.
>
> Gottfried




-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------