NumPy fmax() – Maximum of Two Arrays Element-wise Ignoring NaN

NumPy fmax()

The numpy.fmax() function computes the element-wise maximum of two input arrays. Unlike numpy.maximum(), fmax() ignores NaN values whenever possible, returning the non-NaN value if available.

Syntax

numpy.fmax(x1, x2, /, out=None, *, where=True, casting='same_kind', order='K', dtype=None, subok=True)

Parameters

Parameter	Type	Description
x1, x2	array_like	Arrays containing elements to compare. If their shapes differ, they must be broadcastable.
out	ndarray, None, or tuple of ndarray and None, optional	Optional output array for storing results. If None, a new array is created.
where	array_like, optional	Boolean mask specifying where to compute the maximum. Unselected elements retain original values.
casting	str, optional	Defines the casting behavior of the function.
order	str, optional	Memory layout of the output array.
dtype	data-type, optional	Specifies the desired data type of the output array.
subok	bool, optional	Determines if subclasses of ndarray are preserved in the output.

Return Value

Returns an array containing the element-wise maximum of x1 and x2. If both inputs are scalars, a scalar is returned.

---

Examples

1. Element-wise Maximum of Two Arrays

Finds the maximum value for each element between two arrays.

import numpy as np

# Define two input arrays
x1 = np.array([3, 5, 7, 2])
x2 = np.array([4, 2, 8, 10])

# Compute the element-wise maximum
result = np.fmax(x1, x2)

# Print the results
print("Element-wise maximum:", result)

Output:

Element-wise maximum: [ 4  5  8 10]

2. Handling NaN Values in fmax()

Unlike numpy.maximum(), fmax() ignores NaN values whenever possible.

import numpy as np

# Define two arrays with NaN values
x1 = np.array([np.nan, 3, np.nan, 7])
x2 = np.array([1, np.nan, 5, np.nan])

# Compute the element-wise maximum ignoring NaN
result = np.fmax(x1, x2)

# Print the results
print("Element-wise maximum (ignoring NaN):", result)

Output:

Element-wise maximum (ignoring NaN): [ 1.  3.  5.  7.]

3. Using the out Parameter

Storing the result in a pre-allocated output array.

import numpy as np

# Define two input arrays
x1 = np.array([1, 4, 6, 9])
x2 = np.array([3, 2, 8, 5])

# Pre-allocate an output array
output_array = np.empty_like(x1)

# Compute element-wise maximum and store in output_array
np.fmax(x1, x2, out=output_array)

# Print the results
print("Element-wise maximum with output array:", output_array)

Output:

Element-wise maximum with output array: [3 4 8 9]

4. Using the where Parameter

Computing the element-wise maximum only for selected elements using a condition.

import numpy as np

# Define two input arrays
x1 = np.array([2, 8, 1, 6])
x2 = np.array([5, 3, 7, 4])

# Define a mask (condition)
mask = np.array([True, False, True, False])

# Compute element-wise maximum only where mask is True
result = np.fmax(x1, x2, where=mask)

# Print the results
print("Element-wise maximum with condition:", result)

Output:

Element-wise maximum with condition: [5 0 7 0]

Only elements where mask=True are affected, while the others retain their original values.