MultivariateNormalish

A multivariate Normal distribution built from and callable on ParameterArrays.

Methods

`MultivariateNormalish.cdf`(x)	Cumulative distribution function.
`MultivariateNormalish.logcdf`(x)	Log of the cumulative distribution function.
`MultivariateNormalish.logpdf`(x)	Log of the probability density function.
`MultivariateNormalish.pdf`(x)	Probability density function.
`MultivariateNormalish.rvs`([size, random_state])	Draw random samples from the distribution.

Attributes

class gumbi.arrays.MultivariateNormalish(mean: ParameterArray, cov: int | float | np.ndarray, **kwargs)

Bases: multivariate_normal_frozen

A multivariate Normal distribution built from and callable on ParameterArrays.

In particular, this class takes care of transforming variables to/from “natural” space as necessary.

Parameters:

mean (ParameterArray) – A ParameterArray containing the distribution mean, must be a single point
cov (float or np.array, default 1) – Covariance matrix of the distribution as a standard array or scalar (default one)
**kwargs – Additional keyword arguments passed to scipy.stats.multivariate_normal

cdf(x: ParameterArray) → float

Cumulative distribution function.

Parameters:: x (ParameterArray) – Quantiles, with the last axis of x denoting the components.
Return type:: array-like

logcdf(x: ParameterArray) → float

Log of the cumulative distribution function.

Parameters:: x (ParameterArray) – Quantiles, with the last axis of x denoting the components.
Return type:: array-like

logpdf(x: ParameterArray) → float

Log of the probability density function.

Parameters:: x (ParameterArray) – Quantiles, with the last axis of x denoting the components.
Return type:: array-like

pdf(x: ParameterArray) → float

Probability density function.

Parameters:: x (ParameterArray) – Quantiles, with the last axis of x denoting the components.
Return type:: array-like

rvs(size=1, random_state=None) → ParameterArray

Draw random samples from the distribution.

Parameters:

Return type:

ParameterArray