Hi everyone,
I’m hoping the community can help me figure something out, or at the least point me in the right direction.
As a way to do a convolution during my simulations, I want to be able to take a Fourier transform (forward and inverse) of my data. To do this, I think I will need to get an ordered (row-major or something) array from a function, give it to a python-scipy routine to perform the transforms, and then give it back to fenics in it’s preferred order.
Assuming this is a reasonable and efficient approach, I’m having difficulties even beginning to approach this. The built in vertex and dof mapping is a bit confusing to me. Not to mention the headaches I’m having trying to figure out how they work across multiple processors.
If this is not a reasonable and efficient approach in the fenics framework, I would like to hear some alternatives or ideas from the community. Has anyone ever attempted something like this?
Thanks,
-Nana