Python - tensorflow.math.reduce_logsumexp()

TensorFlow is open-source Python library designed by Google to develop Machine Learning models and deep learning  neural networks.

reduce_logsumexp() is used to compute log sum exp of elements across dimensions of a tensor.

Syntax: tensorflow.math.reduce_logsumexp( input_tensor, axis, keepdims, name)

Parameters:

• input_tensor: It is numeric tensor to reduce.
• axis(optional): It represent the dimensions to  reduce. It's value should be in range [-rank(input_tensor), rank(input_tensor)). If no value is given for this all dimensions are reduced.
• keepdims(optional): It's default value is False. If it's set to True it will retain the reduced dimension with length 1.
• name(optional): It defines the name for the operation.

Returns: It returns a tensor.

Example 1:

# importing the library
import tensorflow as tf

# Initializing the input tensor
a = tf.constant([1, 2, 3, 4], dtype = tf.float64)

# Printing the input tensor
print('Input: ', a)

# Calculating result
res = tf.math.reduce_logsumexp(a)

# Printing the result
print('Result: ', res)

Output:

Input:  tf.Tensor([1. 2. 3. 4.], shape=(4, ), dtype=float64)
Result:  tf.Tensor(4.440189698561196, shape=(), dtype=float64)

Example 2:

# importing the library
import tensorflow as tf

# Initializing the input tensor
a = tf.constant([[1, 2], [3, 4]], dtype = tf.float64)

# Printing the input tensor
print('Input: ', a)

# Calculating result
res = tf.math.reduce_logsumexp(a, axis = 1, keepdims = True)

# Printing the result
print('Result: ', res)

Output:

Input:  tf.Tensor(
[[1. 2.]
 [3. 4.]], shape=(2, 2), dtype=float64)
Result:  tf.Tensor(
[[2.31326169]
 [4.31326169]], shape=(2, 1), dtype=float64)