special_priors

Specialized priors that behave like the Prior class.

The Prior class has certain design constraints that prevent it from covering all cases. So this module contains a collection of priors that do not inherit from the Prior class but have many of the same methods.

Functions

deserialize_alternative_prior(data)	Alternative deserializer that recursively handles all nested parameters.
is_alternative_prior(data)	Check if the data is a dictionary representing a Prior (alternative check).

Classes

LaplacePrior([dims, centered])	A Laplace prior parameterized by a location and a scale parameter.
LogNormalPrior([dims, centered])	Lognormal prior parameterized by positive-scale mean and std.
MaskedPrior(prior, mask[, active_dim])	Create variables from a prior over only the active entries of a boolean mask.
SpecialPrior([dims, centered])	A base class for specialized priors.