warp.norm_pseudo_huber#

warp.norm_pseudo_huber(v: Any, delta: float) → float#

  • Kernel

  • Differentiable

Compute the “pseudo” Huber norm of a vector v with a given delta.

\[H^\prime(v) = \delta \sqrt{1 + \frac{\|v\|^2}{\delta^2}}\]
../../_images/norm_pseudo_huber.svg
Parameters:

  • v (Vector[Float, Any]) – The vector to compute the Huber norm of.

  • delta (float) – The threshold value, defaults to 1.0.

Returns:

The Huber norm of the vector.

Return type:

float