Parameter#

class scarlet2.Parameter(node, name=None, constraint=None, prior=None, stepsize=None)[source]#

Bases: object

Class representing a single optimizable parameter

Definition of optimizable parameter

Parameters:

node (array) – Data portion of a member of Module
name (str, optional) – Name to assign to this parameter If not set, uses varname to determine the name node has within its module.
constraint (numpyro.distributions.constraints.Constraint, optional) – Region over which the parameter value is valid. Contains a bijective transformation to reach this region. Cannot be used at the same time as prior.
prior (numpyro.distributions.distribution.Distribution, optional) – Distribution to determine the probability of a parameter value. This is used by the optimization in scarlet2.Scene.fit() and scarlet2.Scene.sample().
stepsize ((float, callable)) – Step size, or function to determine it (e.g. relative_step()) for parameter updates. This is used by the optimization in scarlet2.Scene.fit().

Parameter