Type promotion in operations in NumPy - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of adding int32 and float64 arrays

What is the data type of the resulting array after adding these two NumPy arrays?

NumPy

import numpy as np
arr1 = np.array([1, 2, 3], dtype=np.int32)
arr2 = np.array([1.5, 2.5, 3.5], dtype=np.float64)
result = arr1 + arr2
print(result.dtype)

A float64

B int32

C float32

D object

💡 Hint

When adding integers and floats, NumPy promotes to the more general type.

NumPy promotes the integer array to float64 to match the float64 array, so the result is float64.

❓ Predict Output

intermediate

Result type of multiplying uint8 and int16 arrays

What is the data type of the resulting array after multiplying these two NumPy arrays?

NumPy

import numpy as np
arr1 = np.array([10, 20, 30], dtype=np.uint8)
arr2 = np.array([1000, 2000, 3000], dtype=np.int16)
result = arr1 * arr2
print(result.dtype)

A int16

B int32

C uint8

D float64

💡 Hint

NumPy promotes the narrower dtype to the wider one.

The uint8 array is promoted to int32 to safely hold the result with int16, so the result is int32.

❓ data_output

advanced

Output array and dtype after adding complex64 and float32 arrays

What is the output array and its data type after adding these two arrays?

NumPy

import numpy as np
arr1 = np.array([1+2j, 3+4j], dtype=np.complex64)
arr2 = np.array([1.5, 2.5], dtype=np.float32)
result = arr1 + arr2
print(result)
print(result.dtype)

A [2.5+2.j 5.5+4.j] with dtype float64

B [2.5+2.j 5.5+4.j] with dtype complex128

C [2.5 5.5] with dtype float32

D [2.5+2.j 5.5+4.j] with dtype complex64

💡 Hint

Adding float32 to complex64 promotes float32 to complex64.

NumPy promotes float32 to complex64 to match complex64, so output dtype is complex64.

❓ Predict Output

advanced

Data type after subtracting int64 and uint64 arrays

What is the data type of the resulting array after subtracting these two arrays?

NumPy

import numpy as np
arr1 = np.array([10, 20, 30], dtype=np.int64)
arr2 = np.array([1, 2, 3], dtype=np.uint64)
result = arr1 - arr2
print(result.dtype)

A float64

B int64

C uint64

D object

💡 Hint

Subtracting signed and unsigned 64-bit integers promotes to float64 to avoid overflow.

NumPy promotes int64 and uint64 to float64 to safely represent the result.

❓ visualization

expert

Visualizing type promotion in mixed dtype operations

Which option shows the correct plot of the data types resulting from adding arrays of these dtypes: int8, uint16, float32, complex128?

NumPy

import numpy as np
import matplotlib.pyplot as plt

dtypes = [np.int8, np.uint16, np.float32, np.complex128]
results = []
for i, dt1 in enumerate(dtypes):
    for j, dt2 in enumerate(dtypes):
        a = np.array([1], dtype=dt1)
        b = np.array([1], dtype=dt2)
        c = a + b
        results.append((dt1.__name__, dt2.__name__, c.dtype.name))

# Prepare data for heatmap
import pandas as pd
pivot_data = pd.DataFrame(results, columns=['Type1', 'Type2', 'Result'])
pivot_table = pivot_data.pivot('Type1', 'Type2', 'Result')

plt.figure(figsize=(8,6))
plt.title('Resulting dtype from adding two dtypes')
plt.imshow(pivot_table.applymap(lambda x: {'int8':0,'uint16':1,'float32':2,'complex128':3}[x]), cmap='viridis')
plt.xticks(range(len(dtypes)), [dt.__name__ for dt in dtypes])
plt.yticks(range(len(dtypes)), [dt.__name__ for dt in dtypes])
plt.colorbar(ticks=[0,1,2,3], label='dtype code')
plt.show()

A A heatmap where adding int8 and uint16 results in int8, and adding float32 and complex128 results in float32

B A heatmap where adding int8 and uint16 results in complex128, and adding float32 and complex128 results in uint16

C A heatmap where adding int8 and uint16 results in uint16, and adding float32 and complex128 results in complex128

D A heatmap where all additions result in float64 regardless of input types

💡 Hint

NumPy promotes to the smallest dtype that can hold both inputs without loss.

NumPy promotes types to the more general type that can hold both inputs, e.g., int8 + uint16 → uint16, float32 + complex128 → complex128.