# Scattering field formulation used in DG-FEM

1. Apr 29, 2013

### discworld

Hello!

Reading up on simulations of electromagnetic scattering with DG-FEM and trying some myself, I got stuck.
In some of papers I have read, a scattering field formulation is used, in which the total field is linearly decomposed in incident field and scattering field:

$E^{T}=E^{S}+E^{I}$

And, the 2D equations for the scattering field in a lossless, isotropic medium are:

$\epsilon_{r} \frac{\partial E^{S}}{\partial t} = \nabla \times H^{S} - (\epsilon_{r} - \epsilon_{r}^{I}) \frac{\partial E^{i}}{\partial t}$
$\mu_{r} \frac{\partial H^{S}}{\partial t} = -\nabla \times E^{S} - (\mu_{r} - \mu_{r}^{I}) \frac{\partial H^{i}}{\partial t}$

My problem is in the interpretation of the "scattering field" and "incident field" in this context. In every use I see of this formulation $\epsilon_{r}$ is space dependent, while $\epsilon_{r}^{I}$ is a constant - specifically, the incident medium's permittivity (same for the permeability). How can this work for multi-substrate cases, where, if I am thinking correctly, the medium considered incident should change?

(I am quite confused with the affair in general, so any clarifications are quite welcome)