stheno.random module¶
-
class
stheno.random.Normal[source]¶ Bases:
stheno.random.RandomVectorNormal random variable.
- Parameters
mean (column vector, optional) – Mean of the distribution. Defaults to zero.
var (matrix) – Variance of the distribution.
-
property
dim¶ Dimensionality.
-
property
dtype¶ Data type.
-
kl[source]¶ Compute the KL divergence with respect to another normal distribution.
- Parameters
other (
random.Normal) – Other normal.- Returns
KL divergence.
- Return type
scalar
-
logpdf(x)[source]¶ Compute the log-pdf.
- Parameters
x (input) – Values to compute the log-pdf of.
- Returns
- Log-pdf for every input in x. If it can be
determined that the list contains only a single log-pdf, then the list is flattened to a scalar.
- Return type
list[tensor]
-
property
m2¶ Second moment.
-
marginal_credible_bounds()[source]¶ Get the marginal credible region bounds.
- Returns
- A tuple containing the marginal means and marginal lower and
upper 95% central credible interval bounds.
- Return type
tuple
-
marginals()[source]¶ Get the marginals.
- Returns
A tuple containing the marginal means and marginal variances.
- Return type
tuple
-
property
mean¶ Mean.
-
property
mean_is_zero¶ The mean is zero.
-
sample(num=1, noise=None)[source]¶ Sample from the distribution.
- Parameters
num (int) – Number of samples.
noise (scalar, optional) – Variance of noise to add to the samples. Must be positive.
- Returns
Samples as rank 2 column vectors.
- Return type
tensor
-
property
var¶ Variance.
-
w2[source]¶ Compute the 2-Wasserstein distance with respect to another normal distribution.
- Parameters
other (
random.Normal) – Other normal.- Returns
2-Wasserstein distance.
- Return type
scalar
-
class
stheno.random.RandomProcess[source]¶ Bases:
stheno.random.RandomA random process.
-
class
stheno.random.RandomVector[source]¶ Bases:
stheno.random.RandomA random vector.