CODEWITHSUNDEEP
pythonnumpy
tryingtosolve
tryingtosolve

Reputation: 803

Ways to convert cartesian indexing to matrix indexing as in numpy.meshgrid

I'm building some software that involves a meshgrid created using np.meshgrid and the matrix indexing (that is, setting indexing = `ij`). However, the function that I'm working on could potentially take meshgrids created using the cartesian indexing (that is, setting indexing = `xy`). Is there a quick way to go from indexing = `xy` to indexing = `ij`). When it's just 2 dimensional, I can simply transpose. However, when I go up to 3D, 4D, etc., transposition wouldn't do the job.

Upvotes: 0

Views: 1172

Answers (1)

unutbu
unutbu

Reputation: 879739

indexing only affects the order of the first two axes. swapaxes(0, 1) can be used to reverse the order:

In [106]: I,J,K,L = np.meshgrid([1,2,3],['A','B'],[10,20,30,40],['X','Y','Z'], indexing='ij')

In [107]: X,Y,Z,W = np.meshgrid([1,2,3],['A','B'],[10,20,30,40],['X','Y','Z'], indexing='xy')

In [108]: (I.swapaxes(0,1) == X).all()
Out[108]: True

In [109]: (J.swapaxes(0,1) == Y).all()
Out[109]: True

In [110]: (K.swapaxes(0,1) == Z).all()
Out[110]: True

In [111]: (L.swapaxes(0,1) == W).all()
Out[111]: True

swapaxes(0, 1) converts xy to ij indexing too (of course):

In [112]: (X.swapaxes(0,1) == I).all()
Out[112]: True

Upvotes: 1

Related Questions