How to extract the covariance matrix of a defined model

[Miorat]

Hi,

Could anyone assist me with how I can extract the covariance matrix of any generated model?

Thank you in advance,

[LSchueler]

Hi,

what exactly do you mean by covariance matrix? - The Gaussian fields used here are completely defined by their covariance model. The fields are statistically homogeneous, except for an anisotropy factor along the main axes.

[MuellerSeb]

Hey there,

my assumption is, you mean the covariance matrix used in simple-kriging (or any other kriging routine).

I really though it was implemented there, but it seems, that this attribute is missing in the kriging classes.

A temporary workaround to calculate cov_mat could look like this:

import gstools as gs
from scipy.spatial.distance import pdist, squareform

# 2D x-y pos pairs for unstructured data
pos = ([2,3,4], [1,3,5])
model = gs.Gaussian(dim=2, nugget=0.5)
iso_pos = model.isometrize(pos)

cov_mat = model.cov_nugget(squareform(pdist(iso_pos.T)))
print(cov_mat)

Maybe this helps in the meantime.

Cheers, Sebastian