[eriks]

Does anyone have any experience with/has anyone tried applying automatic differentiation to OpenFoam to get parameter sensitivities for a solution?

I would like to use OpenFoam to solve a flowfield with a movable boundary. The thing I really need though, is the partial derivative of the integrated pressure on the surface with respect to movement of the surface (i.e., a linearized stiffness).

The easy answer is a finite difference approximation, but automatic differentiation seems like a more elegant (possibly more efficient) approach.

Thanks

[hjasak]

Yup, I have done work along those lines with good results. The biggest problem is funding: this is a kind of work with wide implications and applicability but requires proficiency along many lines. If you have backing to set up a serious project or fund a PhD student, please talk to me.

I have to say that doing this requires profound knowledge of C++, OpenFOAM, mathematics involved etc. and I would not jump into it lightly.

Hrv

[eriks]

Thanks for the reply. I'm not quite to the point of getting serious about this project - I'm still trying to develop funding and(or) a huge block of time.

If I get something going, I will probably be in touch.

Thanks,
Erik