search

BAMresearch / FenicsXConcrete

MIT License
0 stars 2 forks source link

Adjoint forms definitions #138

Open danielandresarcones opened 1 year ago

danielandresarcones commented 1 year ago

I am looking into implementing adjoint differentiation and it would make sense to me that the adjoint problem with its linear and bilinear forms, and maybe even the gradients directly, were defined in the MaterialProblem, at least for the most common/simplest cases. I would start with the LinearElastic problem.

Is this something desirable in FenicsXConcrete? Has anyone something like this already?

joergfunger commented 1 year ago

I think this is quite interesting, but I'm not yet sure how we would integrate that here. In particular for general problems with a history-dependent model, I'm not fully sure yet how to implement this, since we as far as I understand that we would have to go back in time in the reverse order. Could you make a very short presentation in the next NuTo meeting to explain the concept and then we could discuss the integration afterwards.