Numpy lexicographic ordering

Question

I'd like to lexicographically sort the following array a (get index positions), but, I'm having problems understanding the numpy results:

>>> a = np.asarray([[1, 1, 1, 2, 1, 2], [2, 1, 2, 3, 1, 0], [1, 2, 3, 3, 2, 2]])
>>> a
array([[1, 1, 1, 2, 1, 2],
       [2, 1, 2, 3, 1, 0],
       [1, 2, 3, 3, 2, 2]])
>>> np.lexsort(a)
array([0, 5, 1, 4, 2, 3])

For instance, I don't understand why [1, 2, 1] (a[:,0]) is sort-index 0 while [1, 1, 2] (a[:,1]) is index 5, even thought it should be samller than [1, 2, 1].

Answer 1

The order of significance for keys is opposite to what you expected. In order to get expected result just flip the matrix upside down

>>> np.lexsort(np.flipud(a))
array([1, 4, 0, 2, 5, 3])