Distributions
gpjax.distributions
tfd = tfp.distributions
module-attribute
DistrT = TypeVar('DistrT', bound=tfd.Distribution)
module-attribute
__all__ = ['GaussianDistribution']
module-attribute
GaussianDistribution
Bases: Distribution
Multivariate Gaussian distribution with a linear operator scale matrix.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
loc |
Optional[Float[Array, ' N']]
|
The mean of the distribution. Defaults to None. |
None
|
scale |
Optional[LinearOperator]
|
The scale matrix of the distribution. Defaults to None. |
None
|
Returns
GaussianDistribution: A multivariate Gaussian distribution with a linear operator scale matrix.
loc = loc
instance-attribute
scale = cola.PSD(scale)
instance-attribute
event_shape: Tuple
property
Returns the event shape.
__init__(loc: Optional[Float[Array, ' N']] = None, scale: Optional[LinearOperator] = None) -> None
Initialises the distribution.
mean() -> Float[Array, ' N']
Calculates the mean.
median() -> Float[Array, ' N']
Calculates the median.
mode() -> Float[Array, ' N']
Calculates the mode.
covariance() -> Float[Array, 'N N']
Calculates the covariance matrix.
variance() -> Float[Array, ' N']
Calculates the variance.
stddev() -> Float[Array, ' N']
Calculates the standard deviation.
entropy() -> ScalarFloat
Calculates the entropy of the distribution.
log_prob(y: Float[Array, ' N']) -> ScalarFloat
Calculates the log pdf of the multivariate Gaussian.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
y |
Optional[Float[Array, ' N']]
|
the value of which to calculate the log probability. |
required |
Returns
ScalarFloat: The log probability of the value.
sample(seed: KeyArray, sample_shape: Tuple[int, ...])
See Distribution.sample.