Memory-mapped arrays for large data in NumPy - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of memory-mapped array modification

What is the output of the following code snippet that uses a memory-mapped array?

NumPy

import numpy as np
filename = 'test_mmap.dat'
# Create a memmap file with 5 integers
arr = np.memmap(filename, dtype='int32', mode='w+', shape=(5,))
arr[:] = np.arange(5)
arr.flush()
# Reopen memmap in read-write mode
arr2 = np.memmap(filename, dtype='int32', mode='r+', shape=(5,))
arr2[2] = 100
arr2.flush()
print(arr[:])

A [0 1 100 3 4]

B [0 1 2 3 4]

C [0 1 100 100 4]

D [0 1 2 100 4]

💡 Hint

Remember that memmap shares the same file on disk, so changes in one memmap reflect in others after flush.

The first memmap 'arr' writes values 0 to 4. The second memmap 'arr2' modifies index 2 to 100 and flushes. Since both memmaps point to the same file, reading from 'arr' after flush shows the updated value at index 2.

❓ data_output

intermediate

Number of elements in a memory-mapped array

Given a memory-mapped array created with shape (1000, 1000) and dtype float64, how many elements does it contain?

NumPy

import numpy as np
filename = 'large_mmap.dat'
arr = np.memmap(filename, dtype='float64', mode='w+', shape=(1000, 1000))
print(arr.size)

A 2000

B 1000

C 8000

D 1000000

💡 Hint

The size attribute counts total elements in all dimensions.

The array shape is (1000, 1000), so total elements = 1000 * 1000 = 1,000,000.

🔧 Debug

advanced

Identify the error when opening a memmap file

What error will this code raise when trying to open a memmap file that does not exist in read-only mode?

NumPy

import numpy as np
filename = 'nonexistent.dat'
arr = np.memmap(filename, dtype='float32', mode='r', shape=(10,))

A ValueError

B FileNotFoundError

C TypeError

D No error, creates a new file

💡 Hint

Read-only mode requires the file to exist.

Opening a memmap in mode 'r' requires the file to exist. If it does not, Python raises FileNotFoundError.

🧠 Conceptual

advanced

Why use memory-mapped arrays for large data?

Which of the following is the main advantage of using memory-mapped arrays for large datasets?

A They allow processing data larger than RAM by loading only needed parts from disk.

B They automatically compress data to save disk space.

C They speed up CPU computations by using GPU acceleration.

D They convert data into text format for easier reading.

💡 Hint

Think about how memory-mapping helps with limited memory.

Memory-mapped arrays let you work with data larger than your computer's RAM by loading only the parts you access into memory, avoiding full data loading.

🚀 Application

expert

Predict the output after multiple memmap modifications

Consider this code that creates and modifies a memory-mapped array multiple times. What is the final printed output?

NumPy

import numpy as np
filename = 'multi_mod.dat'
# Create memmap and initialize
arr = np.memmap(filename, dtype='int16', mode='w+', shape=(4,))
arr[:] = [10, 20, 30, 40]
arr.flush()
# Open second memmap and modify
arr2 = np.memmap(filename, dtype='int16', mode='r+', shape=(4,))
arr2[1] = 200
arr2.flush()
# Modify original memmap without flush
arr[2] = 300
print(arr2[:])

A [10 20 30 40]

B [10 200 300 40]

C [10 200 30 40]

D [10 20 300 40]

💡 Hint

Changes in one memmap are visible to others only after flush.

arr2 modifies index 1 to 200 and flushes, so arr2 sees that change. arr modifies index 2 to 300 but does not flush, so arr2 still sees old value 30 at index 2.