Page 1 of 1

Inverse Methods with SSA

Posted: 09 Jul 2015, 06:13
by DavidL
Is it possible to use either the Adjoint or Robin inverse methods for viscosity and basal shear stress with the SSA solver in Elmer?

As best as I can tell there is not since the adjoint is taken with respect the Stokes equations for both those methods, and the stress at the surface is not calculated so the Robin methods cannot be used either. It would be great to get confirmation or find out that I'm not correct though.

Re: Inverse Methods with SSA

Posted: 16 Jul 2015, 18:46
by fgillet
Hi David,

you're right the Adjoint and Robin methods in the Elmer/Ice distribution are for Stokes only.

I have developped the adjoint tools to invert viscosity or/and basal slipperiness with the SSA solver.
You can find an application in this paper in discussion for The Cryosphere: http://www.the-cryosphere-discuss.net/9 ... ssion.html

I haven't pushed the corresponding solvers in the distrib yet.
I still need some thinking on the best way to do it as the adjoint has to evolve in the same time as the direct code and I also wanted to merge some tools that can be common to the adjoints of the Stokes and SSA solvers and thus avoiding to have multiple files to do similar things.

I will try to take the time to do this this automn.
In the meantime you can contact me in private if needed.

Fabien Gillet-Chaulet (LGGE)