Different distances apart from the euclidean

[gsantam]

Hi,

have you considered adding other distances apart from the euclidean, or letting the user customize their own distance? If you are interested in this feature I could start working on it.

[rth]

Currently, we also support great-circle distance.

I think other distance metrics could be interesting but some code re-factoring (#56 and #31) might be beneficial before doing that. Currently, as we saw with the great-circle distance adding a new distance requires repetitive changes all through the code base (also related to #121), and it would be preferable if that could be avoided as much as possile.

You are very welcome to contribute in general though.

[MuellerSeb]

One thing to have in mind: As with the great-circle distance, the meaning of anisotropy depends on your metric. When handling #133 with an isotropic n-dimensional kriging, the distance calculation could be just a plugin.

Since scipys cdist is used to calculate pairwise distances, we could simply provide all their metrics by passing through the metric keyword.

[MuellerSeb]

We will discuss this here: #143. Closing for now. Feel free to Re-open or (better) discuss in the linked issue.