ModelParameter

ModelParameter(
    parameter,
    value=None,
    lower=None,
    upper=None,
    fixed=None,
    prior=None,
)

Bounds, a fixed flag, a prior, or a starting value for one model parameter.

Supplying a prior replaces the model’s default. Do that together with use_default_flat_priors=False on the model, so a later data assignment cannot overwrite what you set.

Parameters

Name	Type	Description	Default
parameter	str or int	The parameter to constrain, as a name ("scale") or a 1-based position. In a model carrying trends the parameter vector is wider than the distribution’s, so a position is required there.	required
value	float	Starting value. A parameter_values vector on the model wins over this.	None
lower	float	Bounds used by the estimators and samplers.	None
upper	float	Bounds used by the estimators and samplers.	None
fixed	bool	Hold the parameter at its value instead of estimating it.	None
prior	Distribution	The parameter’s prior distribution.	None

Examples

>>> model_parameter("scale", lower=0.5, upper=40)
<ModelParameter scale: lower = 0.5, upper = 40>