Understanding array memory layout in NumPy - Practice Questions & Exercises

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of array strides in NumPy

What is the output of the following code showing the strides of a 2D NumPy array?

NumPy

import numpy as np
arr = np.array([[1, 2, 3], [4, 5, 6]], dtype=np.int32)
print(arr.strides)

A (12, 4)

B (3, 1)

C (8, 4)

D (4, 12)

💡 Hint

Remember that strides show the number of bytes to step in each dimension.

The array has shape (2, 3) with dtype int32 (4 bytes). The stride for the first dimension is 3 elements * 4 bytes = 12 bytes. The stride for the second dimension is 4 bytes.

❓ data_output

intermediate

Effect of transpose on array strides

What is the output of the strides attribute after transposing the array?

NumPy

import numpy as np
arr = np.array([[1, 2, 3], [4, 5, 6]], dtype=np.int32)
arr_t = arr.T
print(arr_t.strides)

A (4, 12)

B (12, 4)

C (3, 1)

D (8, 4)

💡 Hint

Transposing swaps the dimensions and their strides.

Transposing swaps the shape and strides. Original strides were (12, 4), so after transpose they become (4, 12).

❓ visualization

advanced

Visualizing memory layout of a reshaped array

Which option correctly describes the memory layout visualization of the following reshaped array?

NumPy

import numpy as np
arr = np.arange(6, dtype=np.int32)
arr_reshaped = arr.reshape((2, 3))
print(arr_reshaped.strides)

A The array is stored non-contiguously with gaps between elements.

B The array is stored with strides (4, 12), columns stored one after another.

C The array is stored in a contiguous block with strides (12, 4), rows stored one after another.

D The array is stored in Fortran order with strides (4, 8).

💡 Hint

Check the strides and how reshape affects memory layout.

Reshape does not change the memory layout if possible. The array is contiguous with strides (12, 4) meaning rows are stored one after another.

🧠 Conceptual

advanced

Understanding C-contiguous vs F-contiguous arrays

Which statement correctly describes the difference between C-contiguous and F-contiguous arrays in NumPy?

A C-contiguous arrays are always 1D; F-contiguous arrays are always 2D.

B C-contiguous arrays store columns sequentially; F-contiguous arrays store rows sequentially.

C Both C-contiguous and F-contiguous arrays store data randomly in memory.

D C-contiguous arrays store rows sequentially in memory; F-contiguous arrays store columns sequentially.

💡 Hint

Think about how multi-dimensional arrays are stored in C and Fortran languages.

C-contiguous means rows are stored one after another (row-major order). F-contiguous means columns are stored one after another (column-major order).

🔧 Debug

expert

Identifying cause of unexpected array memory layout

Given the code below, why does arr.flags['C_CONTIGUOUS'] return False?

NumPy

import numpy as np
arr = np.arange(12).reshape((3,4))
arr_sliced = arr[:, ::2]
print(arr_sliced.flags['C_CONTIGUOUS'])

A Because the array dtype is not compatible with C-contiguity.

B Because slicing with a step creates a non-contiguous view in memory.

C Because the array has only one dimension.

D Because reshape always returns a non-contiguous array.

💡 Hint

Consider how slicing with steps affects memory layout.

Slicing with a step (like ::2) creates a view that skips elements, breaking the contiguous memory layout.