Home      |       Contents       |       About

Prev: Array indexing       |       Next: Array operations

Array slicing

  • Slicing also works following the same ideas and notation presented in lists.
  • Slices create views (no copies) of the original array: slices produce 'views' of the array, which essenatially means different perspectives on data arangement.
  • So, be careful: if you assign new values in a sliced array you actually assign new values to the original. If this is not the behavior you intended for, consider constructing a copy of the original array first.

1D array

In [1]:
import numpy as np
ar = np.linspace(0,10,8, False) 
print(ar, '\n')

print(ar[::-1])
print(ar[1:len(ar):2])

[ 0.    1.25  2.5   3.75  5.    6.25  7.5   8.75] 

[ 8.75  7.5   6.25  5.    3.75  2.5   1.25  0.  ]
[ 1.25  3.75  6.25  8.75]
  • View vs. copy
In [2]:
import numpy as np
ar = np.linspace(0,10,8, False) 
print(ar, '\n')

# Assigniment of slice member is reflected on the original array 
br = ar[::-1]
br[0]=100
print(br)
print(ar)

[ 0.    1.25  2.5   3.75  5.    6.25  7.5   8.75] 

[ 100.      7.5     6.25    5.      3.75    2.5     1.25    0.  ]
[   0.      1.25    2.5     3.75    5.      6.25    7.5   100.  ]
In [3]:
# Construct a new array by calling copy()
import numpy as np
ar = np.linspace(0,10,8, False) 
print(ar, '\n')

cr = ar.copy()[::-1]
cr[0]=100
print(cr)
print(ar)

[ 0.    1.25  2.5   3.75  5.    6.25  7.5   8.75] 

[ 100.      7.5     6.25    5.      3.75    2.5     1.25    0.  ]
[ 0.    1.25  2.5   3.75  5.    6.25  7.5   8.75]

2D array

  • Slicing in 2D arrays works by independently setting the slices for each axis
In [4]:
import numpy as np
ar = np.arange(1,21).reshape(5,4)
ar
Out[4]:
array([[ 1,  2,  3,  4],
       [ 5,  6,  7,  8],
       [ 9, 10, 11, 12],
       [13, 14, 15, 16],
       [17, 18, 19, 20]])
In [5]:
ar[0,::2]
Out[5]:
array([1, 3])
In [6]:
ar[::2,:2]
Out[6]:
array([[ 1,  2],
       [ 9, 10],
       [17, 18]])
In [7]:
ar[2:5,2:4]
Out[7]:
array([[11, 12],
       [15, 16],
       [19, 20]])
In [8]:
ar[::2,::2]
Out[8]:
array([[ 1,  3],
       [ 9, 11],
       [17, 19]])

3D array

In [9]:
import numpy as np
ar = np.arange(1,17).reshape(2,2,4)
print(ar)

[[[ 1  2  3  4]
  [ 5  6  7  8]]

 [[ 9 10 11 12]
  [13 14 15 16]]]
In [10]:
ar[:,0,1:3]
Out[10]:
array([[ 2,  3],
       [10, 11]])
In [11]:
ar[:,:,0:4:2]
Out[11]:
array([[[ 1,  3],
        [ 5,  7]],

       [[ 9, 11],
        [13, 15]]])
In [12]:
ar[::-1,::-1,::-1]
Out[12]:
array([[[16, 15, 14, 13],
        [12, 11, 10,  9]],

       [[ 8,  7,  6,  5],
        [ 4,  3,  2,  1]]])

Slice with assignment

  • Slicing combined with assignment is a flexible way to change array values. It will operate under the same restrictions as general array assignment, namely:
    • Array homogeneity, and
    • Array broadcasting
In [13]:
import numpy as np
ar = np.linspace(0,10,8, False) 
print(ar, '\n')

ar[2:5] = [1]                   # Works OK

# ar[2:5] = 'X'                 # Does NOT: 'X' can not be converted to integer

ar[2:5] = [100,200,300]         # Works OK: arrays broadcast

# ar[2:5] = [100,200]           # Does NOT: arrays do not broadcast

ar

[ 0.    1.25  2.5   3.75  5.    6.25  7.5   8.75]
Out[13]:
array([   0.  ,    1.25,  100.  ,  200.  ,  300.  ,    6.25,    7.5 ,
          8.75])
In [14]:
import numpy as np
ar = np.arange(1,21).reshape(5,4)
print(ar,'\n')

ar[::2,:2] = '100'            # Works OK
print(ar)

[[ 1  2  3  4]
 [ 5  6  7  8]
 [ 9 10 11 12]
 [13 14 15 16]
 [17 18 19 20]] 

[[100 100   3   4]
 [  5   6   7   8]
 [100 100  11  12]
 [ 13  14  15  16]
 [100 100  19  20]]
In [15]:
import numpy as np
ar = np.arange(1,21).reshape(5,4)
print(ar[::2,:2],'\n')

ar[::2,:2] = [100,200]            # Works OK
print(ar[::2,:2])                 # Works OK because slice and array value broadcast
print('\n',ar)

[[ 1  2]
 [ 9 10]
 [17 18]] 

[[100 200]
 [100 200]
 [100 200]]

 [[100 200   3   4]
 [  5   6   7   8]
 [100 200  11  12]
 [ 13  14  15  16]
 [100 200  19  20]]
In [16]:
import numpy as np
ar = np.arange(1,21).reshape(5,4)
print(ar,'\n')

ar[::2,:2] = [100,200,300]            #Does NOT work: slice and array value do not broadcast

[[ 1  2  3  4]
 [ 5  6  7  8]
 [ 9 10 11 12]
 [13 14 15 16]
 [17 18 19 20]] 


---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-16-4bfcc1d7a60e> in <module>()
      3 print(ar,'\n')
      4 
----> 5 ar[::2,:2] = [100,200,300]            #Does NOT work: slice and array value do not broadcast

ValueError: could not broadcast input array from shape (3) into shape (3,2)

. Free learning material
. See full copyright and disclaimer notice