Select "corner" elements of a 2D NumPy array

You can do:

data[[0, 0, -1, -1], [0, -1, 0, -1]]

Here are two possibilities. (Ok, first one isn't actually fancy):

>>> a = np.arange(9).reshape(3, 3)
>>> 
>>> m, n = a.shape
>>> a[::m-1, ::n-1]                                                                                  
array([[0, 2],                                                                                                      
       [6, 8]])                                                                                                     
>>>                                                                                                                 
>>> a[np.ix_((0,-1), (0,-1))]                                                   
array([[0, 2],                                                                                                      
       [6, 8]])                                                                                                     

More explicitly:

>>> idx = np.ix_((0,-1), (0,-1))
>>> idx
(array([[ 0],
       [-1]]), array([[ 0, -1]]))
>>> a[idx]
array([[0, 2],
       [6, 8]])

The trick is to leverage broadcasting on the indices. np.ix_ knows the details of how to do it.