Speaking of 2D and higher-dimensional arrays, integer array indexing works the same as in 1D arrays along each axis. If we use only one integer array for indexing, we index along only one axis (axis 0). If we use two arrays separated by a comma, we index along both axes (axis 0 and axis 1).

Indexing only along axis 0 using an array of integers returns a 2D array. When we access elements via such indexing, we group them into a new array. This new array consists of 1D arrays, and grouping them increases the dimensionality by one, resulting in a 2D array.

Indexing along axis 0 and axis 1 using two arrays of integers returns a 1D array.

Note

All integer arrays used for each of the axes must have the same shape.

Now it’s time for an example:

Here is an illustration for clarification:

As you can see, we can also combine basic integer indexing and integer array indexing.

Note

Once again, if at least one of the indices is out of bounds, an IndexError is thrown.