numpy.maximum() in Python

numpy.maximum() is a NumPy function that compares two arrays (or scalars) element-wise and returns a new array containing the maximum value at each position. If any compared element is NaN, the NaN is returned. If both elements are NaN, the first NaN is returned.

Example: This example shows how numpy.maximum() compares two numbers and returns the larger one.

import numpy as np

a = 10
b = 21
print(np.maximum(a, b))

21

Syntax

numpy.maximum(arr1, arr2, /, out=None, *, where=True, casting='same_kind', order='K', dtype=None)

Parameters:

• arr1: First input array (or scalar).
• arr2: Second input array (or scalar).
• out (optional): Array to store the result.
• where: Boolean mask; True positions are computed.
• dtype (optional): Data type of the output.
• casting / order: Controls data casting and memory layout (rarely used).

Note: / -> Parameters before / are positional-only (must be passed without argument names).
* -> Parameters after * are keyword-only (must be passed using their names).

Examples of numpy.maximum()

Example 1: This example compares two 1D arrays and returns the element-wise maximum values.

import numpy as np

a = np.array([2, 8, 125])
b = np.array([3, 3, 15])
print(np.maximum(a, b))

[  3   8 125]

Explanation: np.maximum(a, b) compares each index, max(2, 3) -> 3, max(8, 3) -> 8 and max(125, 15) -> 125.

Example 2: This example shows how numpy.maximum() behaves when the arrays contain NaN values.

import numpy as np

a = np.array([np.nan, 0, np.nan])
b = np.array([np.nan, np.nan, 0])
print(np.maximum(a, b))

[nan nan nan]

Explanation:

• When one element is np.nan, the result is np.nan.
• When both elements are np.nan, the first np.nan is returned.

Example 3: This example compares two arrays of different shapes using broadcasting and returns element-wise maxima.

import numpy as np

a = np.array([[1, 4, 7], [2, 5, 8]])
b = np.array([3, 3, 3])
print(np.maximum(a, b))

[[3 4 7]
 [3 5 8]]

Explanation:

• b is broadcast across rows.
• np.maximum(a, b) compares each column with [3, 3, 3].