Indexing on ndarrays - use Technical Quotient

Dealing with variable numbers of indices within programs

1. Remember

What does the expression z[(1, 1, 1, 1)] return when z is a NumPy array with shape (3, 3, 3, 3)?

A subarray from z
An error
z reshaped to (1, 1, 1, 1)
40

3. Apply

You have a 4D NumPy array z and want to select a 2D slice by fixing the first and last dimensions at index 1 and varying the second and third dimensions. How can you achieve this?

indices = [1, :, :, 1]; z[indices]

Incorrect; the syntax uses a list, which is not how slices are specified, and : cannot be used inside a list or tuple directly.

indices = (1, slice(None), slice(None), 1); z[indices]

Correct; using slice(None) allows selecting all elements along those dimensions, effectively varying them.

indices = (1, ..., 1); z[indices]

Incorrect; the ellipsis ... is used to replace one or more full slices, not to specify varying dimensions explicitly.

indices = [slice(1, None), 1, 1]; z[indices]

4. Analyze

Given an array z = np.arange(81).reshape(3, 3, 3, 3) and a tuple of indices from np.nonzero(z % 40 == 0), why can this tuple be used directly to index z and retrieve specific elements?

The tuple represents direct indices to elements where the condition is true.

Correct; np.nonzero() returns a tuple of arrays, each representing the indices along one dimension where the condition is true, allowing direct indexing into the array.

Indexing with a tuple from np.nonzero() retrieves all elements.
The tuple needs to be converted to a list before indexing.
np.nonzero() returns a list that cannot be used for direct indexing.

Incorrect; np.nonzero() returns a tuple, not a list, specifically for direct indexing.

1. Remember

2. Understand

3. Apply

4. Analyze

Leave a ReplyCancel Reply