StochasticTensor¶

class tfsnippet.StochasticTensor(distribution, tensor, n_samples=None, group_ndims=0, is_reparameterized=None, flow_origin=None, log_prob=None)¶

Bases: tfsnippet.utils.tensor_wrapper.TensorWrapper

Samples or observations of a stochastic variable.

StochasticTensor is a tensor-like object, carrying samples or observations of a random variable, following some distribution of a specific Distribution type.

It mimics the interface of zhusuan.model.StochasticTensor, except that it does not carry a name, and does not add itself to any BayesianNet context automatically.

Attributes Summary

`distribution`	Get the `Distribution` of this `StochasticTensor`.
`flow_origin`	Get the original stochastic tensor from the base distribution of a `tfsnippet.FlowDistribution`.
`group_ndims`	Get the number of dimensions to be considered as events group.
`is_continuous`	Whether or not this `StochasticTensor` is continuous?
`is_reparameterized`	Whether or not this `StochasticTensor` is re-parameterized?
`n_samples`	Get the number of samples taken in `Distribution.sample`.
`tensor`	Get the samples or observations tensor.

Methods Summary

`log_prob`([group_ndims, name])	Compute the log-densities of this `StochasticTensor`.
`prob`([group_ndims, name])	Compute the densities of this `StochasticTensor`.

Attributes Documentation

distribution¶

Get the Distribution of this StochasticTensor.

Returns:	The distribution instance.
Return type:	Distribution

flow_origin¶

Get the original stochastic tensor from the base distribution of a tfsnippet.FlowDistribution.

Returns:	The original stochastic tensor, or `None` if there is no original stochastic tensor.
Return type:	StochasticTensor or None

group_ndims¶

Get the number of dimensions to be considered as events group.

Returns:	The configured group_ndims.
Return type:	int or tf.Tensor

is_continuous¶

Whether or not this StochasticTensor is continuous?

Returns:	Equivalent to `self.distribution.is_continuous`.
Return type:	bool

is_reparameterized¶

Whether or not this StochasticTensor is re-parameterized?

Returns:	A boolean indicating whether it is re-parameterized.
Return type:	bool

n_samples¶

Get the number of samples taken in Distribution.sample.

Returns:	The number of samples.
Return type:	int or tf.Tensor or None

tensor¶

Get the samples or observations tensor.

Returns:	The tensor specified at construction.
Return type:	tf.Tensor

Methods Documentation

log_prob(group_ndims=None, name=None)¶

Compute the log-densities of this StochasticTensor.

Parameters:	group_ndims (int or tf.Tensor) – If specified, overriding the configured group_ndims. name – TensorFlow name scope of the graph nodes.
Returns:	The log-densities.
Return type:	tf.Tensor

prob(group_ndims=None, name=None)¶

Compute the densities of this StochasticTensor.

Parameters:	group_ndims (int or tf.Tensor) – If specified, overriding the configured group_ndims. name – TensorFlow name scope of the graph nodes.
Returns:	The densities.
Return type:	tf.Tensor