Function Differences with tf.math.divide

tf.math.divide

tf.math.divide(x, y, name=None) -> Tensor

For more information, see tf.math.divide.

mindspore.ops.div

mindspore.ops.div(input, other, rounding_mode=None) -> Tensor

For more information, see mindspore.ops.div.

Differences

TensorFlow: The quotient is obtained by dividing two Tensors element-wise.

MindSpore: When the parameter rounding_mode of MindSpore API is None by default, MindSpore implements the same function as TensorFlow.

Categories	Subcategories	TensorFlow	MindSpore	Differences
Parameters	Parameter 1	x	input	Same function, different parameter names
	Parameter 2	y	other	Same function, different parameter names
	Parameter 3	-	rounding_mode	This parameter is not available in TensorFlow. MindSpore is an optional parameter that determines the rounding type of the result, and the default value is None
	Parameter 4	name	-	Not involved

Code Example

When the parameter rounding_mode of MindSpore is not specified, the two APIs achieve the same function and have the same usage.

# TensorFlow
import tensorflow as tf
import numpy

x = tf.constant([[2, 4, 6, 8], [1, 2, 3, 4]])
y = tf.constant([5, 8, 8, 16])
out = tf.math.divide(x, y).numpy()
print(out)
# [[0.4   0.5   0.75  0.5  ]
#  [0.2   0.25  0.375 0.25 ]]

# MindSpore
import mindspore
from mindspore import Tensor
import mindspore.ops as ops
import numpy as np

x_ = np.array([[2, 4, 6, 8], [1, 2, 3, 4]])
y_ = np.array([5, 8, 8, 16])
x = Tensor(x_, mindspore.float64)
y = Tensor(y_, mindspore.float64)
output = ops.div(x, y)
print(output)
# [[0.4   0.5   0.75  0.5  ]
#  [0.2   0.25  0.375 0.25 ]]