mindspore.nn.probability.distribution.Gumbel

class mindspore.nn.probability.distribution.Gumbel(loc, scale, seed=0, dtype=mindspore.float32, name='Gumbel')[source]

Gumbel distribution.

Parameters
  • loc (float, list, numpy.ndarray, Tensor) – The location of Gumbel distribution.

  • scale (float, list, numpy.ndarray, Tensor) – The scale of Gumbel distribution.

  • seed (int) – the seed used in sampling. The global seed is used if it is None. Default: None.

  • dtype (mindspore.dtype) – type of the distribution. Default: mstype.float32.

  • name (str) – the name of the distribution. Default: ‘Gumbel’.

Supported Platforms:

Ascend GPU

Note

scale must be greater than zero. dist_spec_args are loc and scale. dtype must be a float type because Gumbel distributions are continuous. kl_loss and cross_entropy are not supported on GPU backend.

Examples

>>> import mindspore
>>> import mindspore.context as context
>>> import mindspore.nn as nn
>>> import mindspore.nn.probability.distribution as msd
>>> from mindspore import Tensor
>>> context.set_context(mode=1)
>>> # To initialize a Gumbel distribution of `loc` 3.0 and `scale` 4.0.
>>> gumbel = msd.Gumbel(3.0, 4.0, dtype=mindspore.float32)
>>> # Private interfaces of probability functions corresponding to public interfaces, including
>>> # `prob`, `log_prob`, `cdf`, `log_cdf`, `survival_function`, and `log_survival`, have the same
>>> # arguments as follows.
>>> # Args:
>>> #     value (Tensor): the value to be evaluated.
>>> # Examples of `prob`.
>>> # Similar calls can be made to other probability functions
>>> # by replacing 'prob' by the name of the function.
>>> value = Tensor([1.0, 2.0, 3.0], dtype=mindspore.float32)
>>> ans = gumbel.prob(value)
>>> print(ans.shape)
(3,)
>>> # Functions `mean`, `mode`, sd`, `var`, and `entropy` do not take in any argument.
>>> ans = gumbel.mean()
>>> print(ans.shape)
()
>>> # Interfaces of 'kl_loss' and 'cross_entropy' are the same:
>>> # Args:
>>> #     dist (str): the type of the distributions. Only "Gumbel" is supported.
>>> #     loc_b (Tensor): the loc of distribution b.
>>> #     scale_b (Tensor): the scale distribution b.
>>> # Examples of `kl_loss`. `cross_entropy` is similar.
>>> loc_b = Tensor([1.0], dtype=mindspore.float32)
>>> scale_b = Tensor([1.0, 1.5, 2.0], dtype=mindspore.float32)
>>> ans = gumbel.kl_loss('Gumbel', loc_b, scale_b)
>>> print(ans.shape)
(3,)
>>> # Examples of `sample`.
>>> # Args:
>>> #     shape (tuple): the shape of the sample. Default: ()
>>> ans = gumbel.sample()
>>> print(ans.shape)
()
>>> ans = gumbel.sample((2,3))
>>> print(ans.shape)
property loc

Return the location of the distribution after casting to dtype.

property scale

Return the scale of the distribution after casting to dtype.