Data is not valid NDGRID format.

I would like to know what does it mean "Data is not valid NDGRID format" when I try to use interpn() function to interpolate a function from four variables. I've already checked, a dozen times, my matrixes dimensions. Which one of them agrees in size (4D - 2x11x4x3). Besides, there is no way I can use ndgrid() function to create the mesh, in this case I did this mannualy.

If I have all the function values for which one of the variables, why this error appears?

Thanks in advance,

Answer 1

[dpb]

From the doc for interpn one finds...

"interpn requires that X1,X2,X3,etc. be monotonic and plaid (as if
 they were created using NDGRID).  X1,X2,X3,etc. can be non-uniformly
 spaced."

So, it doesn't matter about the dimension so much, only that the Xn vectors/arrays be monotonic and non-repeating such that they are valid for ndgrid. That you say "there is no way I can use ndgrid() function to create the mesh" pretty much means you can't use interpn to interpolate the data, either (actually, it means it flat out, not just "pretty much" :) ).

You can try some of the other interpolation methods.

Why, specifically, can't you use ndgrid? What is a (smallish) typical subset of the data grid for looksies?

[Victor]

The problem is solved, but this is awkward... I did exactly what the function ngrid() does, but mannualy.

I checked and I noticed that my previous matrixes were exactly the same as the matrixes generated by the ndgrid function, but the fact that I didn't use the ngrid function to generate them caused the problem.

Anyway, I found a way to use the ndgrid function and it did work.

Thanks,

[dpb]

...my previous matrixes were exactly the same as the matrixes generated by the ndgrid function, but the fact that I didn't use the ngrid function to generate them caused the problem.

I don't believe that for a minute! :) I'm sure it looks like that, but I'm certain there's something different in the two cases. Matlab doesn't "know" from whence a set of arrays came; all it knows is whether the input meets the input checks run to validate inputs or not and your inputs didn't.

Unfortunately, W/O actual code/data, no way to tell exactly what the mismatch was...

[Colton Fruhling]

I have a similar issue that I cannot use ndgrid. I have 3D data where I only want to interpolate along the pages ( 3rd dimension). Slicing the matrix can take considerable time so I thought interpn could essentially parallelize it. But the time points where I have the data are not the same for each page and thus I cannot use ndgrid. Any help would be most appreciated.

[dpb]

I suggest instead of a comment to a closed question that you post this as new question and supply a (small!) dataset that illustrates the problem.

[Colton Fruhling]

ok thanks