Thread Subject: 2D nonmonotonic interpolation

hi,
i am trying to find a 2D nonmonotonic interpolation.
my data is z(x,y), but again the x and the y are
nonmonotonic data.
I have the same problem when i try to fit 2D curve to this
data.
thank,
n.

In article <g5nq2h$kea$1@fred.mathworks.com>,
nir zag <nirzag@technion.ac.il> wrote:

>i am trying to find a 2D nonmonotonic interpolation.
>my data is z(x,y), but again the x and the y are
>nonmonotonic data.

is there a reason you could not use

[sx, sxidx] = sort(x);
[sy, syidx] = sort(y);

sz = z(sxidx, syidx);

Then, presuming that x and y were vectors that formed a grid of (x,y)
pairs, sz would be z reordered onto a monotonic grid whose
coordinates were (sx, sy)

After the sorting, the sx would be monotonic, but possibly not
strictly increasing. If the sorted x are not strictly increasing
and the sorted y are not either, then you have a logistic problem
with interpolation: the standard interpolation routines need one of
the two coordinates to be strictly increasing in order to produce
valid results (otherwise, they have trouble deciding which of the
several y values for each repeated x should be used, and tend to pick
amongst them in non-intuitive ways.)

"nir zag" <nirzag@technion.ac.il> wrote in message
<g5nq2h$kea$1@fred.mathworks.com>...
> hi,
> i am trying to find a 2D nonmonotonic interpolation.
> my data is z(x,y), but again the x and the y are
> nonmonotonic data.
> I have the same problem when i try to fit 2D curve to this
> data.

Use griddata, or you may find my own gridfit
to be of service.

http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?
objectId=8998&objectType=FILE

John

In article <g5nq2h$kea$1@fred.mathworks.com>,
nir zag <nirzag@technion.ac.il> wrote:

>i am trying to find a 2D nonmonotonic interpolation.
>my data is z(x,y), but again the x and the y are
>nonmonotonic data.

is there a reason you could not use

[sx, sxidx] = sort(x);
[sy, syidx] = sort(y);

sz = z(sxidx, syidx);

Then, presuming that x and y were vectors that formed a grid of (x,y)
pairs, sz would be z reordered onto a monotonic grid whose
coordinates were (sx, sy)

After the sorting, the sx would be monotonic, but possibly not
strictly increasing. If the sorted x are not strictly increasing
and the sorted y are not either, then you have a logistic problem
with interpolation: the standard interpolation routines need one of
the two coordinates to be strictly increasing in order to produce
valid results (otherwise, they have trouble deciding which of the
several y values for each repeated x should be used, and tend to pick
amongst them in non-intuitive ways.)

"nir zag" <nirzag@technion.ac.il> wrote in message
<g5nq2h$kea$1@fred.mathworks.com>...
> hi,
> i am trying to find a 2D nonmonotonic interpolation.
> my data is z(x,y), but again the x and the y are
> nonmonotonic data.
> I have the same problem when i try to fit 2D curve to this
> data.

Use griddata, or you may find my own gridfit
to be of service.

http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?
objectId=8998&objectType=FILE

John