Small Bug with non-constant density contrast about invert4geom

[mdtanker]

Yeah that sounds great, happy to implement that at some point! Thanks for sharing that paper, don't think I had seen that one yet.

[mdtanker]

Hey @LL-Geo thanks for checking out the package! Yes that's a good catch, I've only been using constant densities so far, but yes we should allow variable densities. The only other place it may need some work to implement would be the density cross validation. I think we would just need to clarify in the documentation that the density CV only works for constant density values. Maybe there could be some scheme for a CV for finding the optimal density values for different regions, i.e. do a CV to find optimal density for 1 region based on a mask, then do another CV for the remaining areas.

I'll try and add some documentation for now.

[LL-Geo]

Hey @LL-Geo thanks for checking out the package! Yes that's a good catch, I've only been using constant densities so far, but yes we should allow variable densities. The only other place it may need some work to implement would be the density cross validation. I think we would just need to clarify in the documentation that the density CV only works for constant density values. Maybe there could be some scheme for a CV for finding the optimal density values for different regions, i.e. do a CV to find optimal density for 1 region based on a mask, then do another CV for the remaining areas.

I'll try and add some documentation for now.

Yeh~ We can try something like this.. 😁
https://academic.oup.com/gji/article/221/3/1896/5808816?login=true
Basically, divide the region into several blocks and get the density contrast. That might be a bit tricky for bathymetry inversion as we don't have much control points, haha. (but we can do a clustering to separate domains based on gravity and mag).

[LL-Geo]

Yeah that sounds great, happy to implement that at some point! Thanks for sharing that paper, don't think I had seen that one yet.

haha.. looking forward to it 😉