el0ps.datafit.KullbackLeibler

class el0ps.datafit.KullbackLeibler(y, e=1e-08)

Kullback-Leibler datafit function.

The function is defined as

\[f(\mathbf{w}) = \sum_{i=1}^m y_i \log(\tfrac{y_i}{w_i + e}) + (w_i + e) - y_i\]

for some \(\mathbf{y} \in \mathbb{R}_+^m\) and \(e \geq 0\). The function returns \(+\infty\) whenever \(w_i + e \leq 0\) for some \(i \in \{1,\dots,m\}\).

Parameters:

yNDArray

Data vector.

efloat = 1e-8

Smoothing parameter.

__init__(y, e=1e-08)

Methods

__init__(y[, e])
conjugate(w)	Value of the convex conjugate of the function at w.
get_spec()	Specify the numba types of the class attributes.
gradient(w)	Value of gradient at w.
gradient_lipschitz_constant()	Lipschitz constant of the gradient.
params_to_dict()	Returns the parameters name and value used to initialize the class instance.
value(w)	Value of the function at w.