How to index and slice a NumPy array

NumPy array indexing selects values by position across one or more axes. It is the normal way to take one element, a row, a column, or a rectangular window from an ndarray before passing that subset into a calculation.

For two-dimensional arrays, the first index addresses rows and the second index addresses columns. A comma separates axes, while a colon keeps all positions on that axis or defines a start, stop, and step range.

Basic integer and slice selections use zero-based positions. Slices stop before the ending index, and basic slices return views when NumPy can point back into the same data buffer; call .copy() before changing a selected block independently.

1. Create a script that selects one element, one row, one column, a rectangular window, and a stepped reversed slice.

   array-index-slice.py

   import numpy as np
    
   readings = np.arange(1, 21).reshape(4, 5)
    
   single = readings[2, 3]
   row = readings[2, :]
   column = readings[:, 3]
   window = readings[1:3, 2:5]
   stepped = readings[::2, ::-1]
    
   print("readings:")
   print(readings)
   print("single readings[2, 3]:", single)
   print("row readings[2, :]:", row)
   print("column readings[:, 3]:", column)
   print("window readings[1:3, 2:5]:")
   print(window)
   print("stepped readings[::2, ::-1]:")
   print(stepped)
   print("window shares memory:", np.shares_memory(readings, window))
    
   assert single == 14
   assert row.tolist() == [11, 12, 13, 14, 15]
   assert column.tolist() == [4, 9, 14, 19]
   assert window.tolist() == [[8, 9, 10], [13, 14, 15]]
   assert stepped.tolist() == [[5, 4, 3, 2, 1], [15, 14, 13, 12, 11]]
   assert np.shares_memory(readings, window)
    
   print("indexing checks passed")

2. Run the script.

   $ python3 array-index-slice.py
   readings:
   [[ 1  2  3  4  5]
    [ 6  7  8  9 10]
    [11 12 13 14 15]
    [16 17 18 19 20]]
   single readings[2, 3]: 14
   row readings[2, :]: [11 12 13 14 15]
   column readings[:, 3]: [ 4  9 14 19]
   window readings[1:3, 2:5]:
   [[ 8  9 10]
    [13 14 15]]
   stepped readings[::2, ::-1]:
   [[ 5  4  3  2  1]
    [15 14 13 12 11]]
   window shares memory: True
   indexing checks passed

   readings[2, 3] selects one scalar. readings[2, :] keeps all columns in row 2, and readings[:, 3] keeps column 3 from every row.

3. Check the rectangular slice.

   readings[1:3, 2:5] keeps rows 1 and 2, then columns 2 through 4. The stop value is excluded on both axes.

4. Check the stepped slice.

   readings[::2, ::-1] keeps every second row and reverses each selected row's columns.

5. Confirm that the final line shows indexing checks passed.

   The assertions prove that the scalar, row, column, window, stepped slice, and memory-sharing checks matched the expected arrays. Use .copy() before editing window when changes should not affect readings.