1

Can any one tell how to interpret this kind of addressing : a[2::2, ::2]? I have no idea how to use this kind of indexing. 

a = np.random.random((6,6))
print(a)
[[0.17948771 0.61436323 0.48768101 0.20540278 0.15331527 0.17766787]
 [0.40028608 0.63915391 0.92719741 0.56604286 0.92959065 0.92707981]
 [0.27554561 0.09253335 0.73841082 0.00840638 0.33683454 0.89065058]
 [0.25030938 0.37093169 0.70789081 0.95205777 0.60483874 0.81425781]
 [0.14250593 0.56916738 0.45440191 0.21140548 0.72945015 0.59313599]
 [0.68116512 0.45349473 0.23057609 0.36349226 0.622701   0.07951557]]

a[2::2, ::2]
array([[0.27554561, 0.73841082, 0.33683454],
       [0.14250593, 0.45440191, 0.72945015]])
s.ouchene
  • 1,682
  • 13
  • 31
Flying pig
  • 572
  • 1
  • 10
  • 27

1 Answers1

6

a[2:] will return a slice of a from index 2 to the end of the list. a[2::2] will do the same, but will only return every second value, a[2::3] every third etc.. So if

a = [0,1,2,3,4]

then a[2::2] will be [2,4] (2 to 4, but only every second). For numpy arrays a[2:,:] will return a slice of a starting from index 2 of the first dimension and the entirety of the second dimension. a[2::2, ::2] will thus return only every second element compared to a[2:,:].

In your example, the returned elements are from the following indices

[[(2,0), (2,2), (2,4)], [(4,0), (4,2), (4,4)]]
user2653663
  • 2,818
  • 1
  • 18
  • 22