sift.kernels.LaplacianKernel#

class sift.kernels.LaplacianKernel(length_scale=1.0, ignore_self=False, **kwargs)[source]#

L1-exponential kernel.

It is parameterized by a length scale parameter \(l>0\), The kernel is given by:

\[k(x_i, x_j) = \exp\left(- \frac{| x_i, x_j|_{1}}{2l^2} \right)\]

where \(l\) is the length scale of the kernel.

Parameters:

Attributes table#

`backend`	The kernel's `pykeops` backend.
`device`	The kernel's device.
`dtype`	The kernel's data type.
`k`	The instantiated kernel object.

LaplacianKernel.backend#

The kernel’s pykeops backend.

Can be either torch or numpy.