Discussion Forum

I have used COMSOL successfully using scattering boundary conditions to simulate axisymmetric models (2-D). However, in general 3-D electromagnetic problems I am confused by several conditions and disclaimers:
Scattering boundary conditions require either plane, cylindrical, or spherical scattered waves, and the wave front must be parallel to the boundary. There is a reference to "Far-Field Calculation" for "a general way of modeling an open boundary"--which is not fully supported by the description of FFC. PML implementation, like SBC, also requires cylindrical waves from a specified axis, spherical waves from a specified center point, etc. I do not see any appropriate boundary condition for general 3-D EM problems.
I would appreciate any comments on the merits of the following three techniques as well as the possibility that they may be implemented in COMSOL
(1) CFS-PMLs to ncrease the absorption of evanescent waves;
(2) Second-order PMLs to absorb both evanescent and propagating waves for all frequencies; or
(3) The novel adiabatic absorbers described in the following paper which have been used in simulaating unusual structures including photonic crystals where PMLs are not appropriate because the material properties are not an analytic function in the direction perpendicular to the boundary: A.F. Oskooi, L. Zhang, Y. Avniel and S.G. Johnson, "The failure of perfectly matched lyers, and towards their redemption by adiabatic absorbers," Optics Express, Vol. 16 (2008) 11376-11392.