spacenet.pcf.helpers.integrated_poly_finite_kernel#

spacenet.pcf.helpers.integrated_poly_finite_kernel(r, w, d_1, d_2, delta_r=1, n=2)#

Computes the integrated polynomial kernel function for given distances and bandwidth parameters. The integrated kernel is defined as:

\[L_{i} \left( r \right) = \sum_{e_{nm} \in E} \int_{0}^{w_{nm}} \kappa_{\Delta r, n } \left( | d_{in} + \frac{d_{im} - d_{in}}{w_{nm}}x - r | \right) dx\]

The kernel function is integrated over the length of the edge, weighted by the edge weight.

Parameters:
rfloat

The distance at which to evaluate the integrated kernel function.

wnumpy.ndarray

An array of edge weights corresponding to the lengths of the edges in the network.

d_1numpy.ndarray

An array of distances from the reference node to one endpoint of each edge.

d_2numpy.ndarray

An array of distances from the reference node to the other endpoint of each edge.

delta_rfloat, optional

The bandwidth parameter for the polynomial kernel function. Default is 1.

nint, optional

The exponent parameter for the polynomial kernel function. Default is 2.

Returns:
integrated_kernel_valuesnumpy.ndarray

An array of integrated kernel values from node i at all r values.

Notes

See reference paper for details on the derivation of the integrated kernel function and its implementation.