mindspore.ops.logsumexp

mindspore.ops.logsumexp(x, axis, keep_dims=False)[source]

Reduces a dimension of a tensor by calculating exponential for all elements in the dimension, then calculate logarithm of the sum.

\[logsumexp(x) = \log(\sum(e^(x-x_{max}))) + x_{max}\]

Note

The dimension of input Tensor on Ascend should be less than or equal to 8, and the dimension of input Tensor on CPU should be less than 8.

Parameters

x (Tensor) – The input tensor. With float16 or float32 data type.
axis (Union[int, tuple(int), list(int)]) – The dimensions to reduce. Default: (), reduce all dimensions. Only constant value is allowed.
keep_dims (bool) – If True, keep these reduced dimensions and the length is 1. If False, don’t keep these dimensions. Default : False.

Returns

Tensor, has the same dtype as the x.

If axis is (), and keep_dims is False, the output is a 0-D tensor representing the sum of all elements in the input tensor.
If axis is int, set as 2, and keep_dims is False, the shape of output is \((x_1, x_3, ..., x_R)\).
If axis is tuple(int), set as (2, 3), and keep_dims is False, the shape of output is \((x_1, x_4, ..., x_R)\).

Examples

>>> x = Tensor(np.random.randn(3, 4, 5, 6).astype(np.float32))
>>> output = ops.logsumexp(x, 1, keep_dims=True)
>>> print(output.shape)
(3, 1, 5, 6)