[MITgcm-devel] ML for ML

[Jean-Michel Campin]

Hi Matt,

I think pkg/mypackage is a good place to start (some tendency terms, some state vars ?
plus loading parameters from file) and then refine it to make it to work. 

But as it get more "mature" you may want to have a proper new package for this
so that you could add it to the main repos (whereas mypackage in the main repos 
will always remain like an empty shell).
I think I started this way with with the gray-atmosphere physics (pkg/atm_phys).

But I don't know if this really answer your question.

Cheers,
Jean-Michel

On Mon, Apr 25, 2022 at 08:58:04PM -0700, Matthew Mazloff wrote:
> Hi Jean-Michel et al.,
> 
> We at SIO are aiming to use a neural net and https://github.com/scientific-computing/FKB <https://github.com/scientific-computing/FKB> for a mixed layer parameterization in the MITgcm. Does it make sense to put this in ???mypackage??? with tendency applied in 
> https://github.com/MITgcm/MITgcm/blob/master/pkg/mypackage/mypackage_tendency_apply.F <https://github.com/MITgcm/MITgcm/blob/master/pkg/mypackage/mypackage_tendency_apply.F>
> 
> Let me know if there is a better place,
> Thanks!
> Matt
> 

> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-devel