Objective functions and constraints

smoother.DerivativeObjective

class smoother.DerivativeObjective(d=1, weight=None) [source]

A Smoother objective function which minimizes the sum of a square derivative.

Parameters and attributes:	d : int, default=1 e.g. `1` means first derivative, `2` means second derivative.

__call__(self, smoother) [source]

Parameters:	smoother : `Smoother` The smoother to which this objective function applies.
Returns:	value : float Approximate mean square derivative over all points of the distribution.

class smoother.MassConstraint(lb, ub, mass, weight=None) [source]

A Smoother constraint that forces a certain amount of probability mass to be within a given range.

Parameters and attributes:

lb : float

Lower bound of the range within which the probability mass must be.

ub : float

Upper bound of the range within which the probability mass must be.

mass : float between 0. and 1.

Amount of probability mass between lb and ub.

weight : float or None, default=None

Weight to place on the constraint. If None the weight will be set automatically when the constraint is called based on the smoother.

__call__(self, smoother) [source]

Parameters:	smoother : Smoother Smoother to which this constraint applies.
Returns:	loss : float

class smoother.MomentConstraint(value, degree, type_='raw', norm=False, weight=None) [source]

A Smoother constraint that forces a moment condition to hold.

Parameters and attributes:

value : float

The target value of the moment.

degree : int

The degree of the moment; e.g. the 1st moment is the mean.

type_ : str, default='raw'

Type of moment: 'raw', 'central' or 'standardized'.

norm : bool, default=False

Indicates whether to apply a norm to the moment.

weight : float or None, default=None

Weight to place on the constraint. If None the weight will be set automatically when the constraint is called based on the smoother.

__call__(self, smoother) [source]

Parameters:	smoother : Smoother Smoother to which this constraint applies.
Returns:	loss : float