查找失败的索引numpy.assert_almost_equal

Question

I'm dealing with large 3D arrays like (110,80,817) and wanting to compare two arrays in some unit tests. However, the default output from numpy.assert_almost_equal doesn't help me track down the errors very easily. For example:

>                   raise AssertionError(msg)
E                   AssertionError:
E                   Arrays are not almost equal to 7 decimals
E
E                   (mismatch 0.0314621119395%)
E                    x: array([[[ 0.,  0.,  0., ...,  0.,  0.,  0.],
E                           [ 0.,  0.,  0., ...,  0.,  0.,  0.],
E                           [ 0.,  0.,  0., ...,  0.,  0.,  0.],...
E                    y: array([[[ 0.,  0.,  0., ...,  0.,  0.,  0.],
E                           [ 0.,  0.,  0., ...,  0.,  0.,  0.],
E                           [ 0.,  0.,  0., ...,  0.,  0.,  0.],...

Is there a way to easily see which 3D indexes are failing this assertion?

Answer 1

You can use np.isclose combined with np.where for this

idx = zip(*np.where(~np.isclose(a, b, atol=0, rtol=1e-7)))

Now idx will be a list of all the indices (x,y,z) where the assertion fails.