This is a read only copy of the old FEniCS QA forum. Please visit the new QA forum to ask questions

Perfectly matched layer

+4 votes

I would like to solve 2D/3D scattering problems using fenics. For this I would need to use a perfectly matched layer, however, I have not yet found any example where PML has been considered so far with fenics. Therefore my short question: Is it easily possible to extend fenics with PML using a complex coordinate transformation or does one need to dig deeper into the code for this.

asked Jun 12, 2013 by cevito FEniCS User (5,550 points)

1 Answer

+2 votes
 
Best answer

There are no plug and play PMLs available in FEniCS (yet).

Are you looking to implement PMLs for frequency domain or time domain?

answered Jun 12, 2013 by stevenvdk FEniCS User (1,590 points)
selected Jun 12, 2013 by cevito

Thanks for your answer.
At the moment I'm interested in seeing this implemented for the frequency domain.

commented Jun 12, 2013 by cevito FEniCS User (5,550 points)

Complex coordinate stretching in frequency domain is (as far as I know) just a matter of changing your material parameters/tensors, which is easily done in FEniCS...

What wave equation are you considering?
Implementing a PML for electromagnetics in time domain happens to be a task on my TODO-list (however with average priority).

commented Jun 12, 2013 by stevenvdk FEniCS User (1,590 points)

Scalar Helmholtz in 2D and curl curl Helmholtz in 3D. Changing the material parameters is of course an option, but the best thing would be to simply be able to define an outer circular/spherical/rectangular layer together with some parameters like the estimated wavelength in the PML, which takes care of the absorption.

Thanks for your answers!

commented Jun 12, 2013 by cevito FEniCS User (5,550 points)
...