Python equivalent of Matlab code

[kwokmaster]

I am converting the interp2 function of matlab to python.
interp2

However, after searching through the equivalents like interpolate.interp2d, they all only allow 1-D array input, while my function requires a 2-D array input. I have no knowledge about interpolation and I can't handwrite an equivalent for it. It would be great if anyone can help.

Thank you in advance!

[scidam]

If you have a meshgrid data, e.g.

1 2 3 4

import numpy as np # x, y -- 1D arrays X, Y = np.meshgrid(x, y) # Z is a matrix,

you can always do

1	interpolate.interp2d(X.ravel(), Y.ravel(), Z.ravel())

Another option, that should also work, is

1	interpolate.interp2d(x, y, Z)

It seems (from the docs) that MatLab's interp2 and scipy's interp2d do not differ significantly.

[kwokmaster]

Hi scidam, thanks for the reply!

I tried the way you mentioned like the following

ip = interpolate.interp2d(Ui,Vi,im,kind='linear')
out = ip(U,V)

where U,V is a 1000*1000 array and im is a 701*701 array.

Then I return with the following error:
ValueError: Invalid length for input z for non rectangular grid

As I have no idea about interpolation, what is this error about and how should I solve it?

Thank you!

[scidam]

I suspect that the problem is in the line ip = interpolate.interp2d(Ui,Vi,im,kind='linear'). Could you provide a minimal reproducible example? What are shapes of Ui, Vi? It is hard to answer without additional information.

[kwokmaster]

I suspect that the problem is in the line ip = interpolate.interp2d(Ui,Vi,im,kind='linear'). Could you provide a minimal reproducible example? What are shapes of Ui, Vi? It is hard to answer without additional information.

The shapes of Ui and Vi are both 701*701

[scidam]

Here is minimal working example,

1 2 3 4 5

>>> from scipy.interpolate import interp2d >>> import numpy as np >>> x, y = np.random.rand(701), np.random.rand(701) >>> z = np.random.rand(701, 701) >>> inp = interp2d(x, y, z)

So, if you call

1	inp([1,2,3],[4,5,6])

you get a matrix of size 3x3 with values for all combinations of points.
Try to restructure U and V into 1D arrays (probably of shape (1000,)).