scipy.special.pseudo_huber#

scipy.special.pseudo_huber(delta, r, out=None) = <ufunc 'pseudo_huber'>#

Pseudo-Huber loss function.

\[\mathrm{pseudo\_huber}(\delta, r) = \delta^2 \left( \sqrt{ 1 + \left( \frac{r}{\delta} \right)^2 } - 1 \right)\]
Parameters
deltaarray_like

Input array, indicating the soft quadratic vs. linear loss changepoint.

rarray_like

Input array, possibly representing residuals.

outndarray, optional

Optional output array for the function results

Returns
resscalar or ndarray

The computed Pseudo-Huber loss function values.

Notes

This function is convex in \(r\).

New in version 0.15.0.