Mixed discretization schemes for electromagnetic surface integral equations

In developing accurate and stable discretization schemes for integral equations, important steps are the choices of the basis and testing functions. In the electromagnetic surface integral equations, the unknown quantities, the equivalent electric and magnetic surface current densities, are by default divergence conforming and therefore should be expanded with the divergence-conforming basis functions. The choice of the testing functions however is less obvious and depends on the choice of the basis functions and the properties of the integral operators. We develop stable and well-defined discretization procedures where the operators are tested either their range or on the dual of their range. This requires the use of non-conventional finite-elements spaces, called ‘dual spaces’, and non-traditional testing procedures. We show with numerical examples that in most cases, testing the operators on the dual of their range leads to improved solution accuracy compared with testing the operators on their range. For the integral equations of the first kind, this agrees with the conventional Galerkin's scheme, but for the integral equations of the second kind, this requires the use of Petrov–Galerkin type discretization schemes with appropriate curl-conforming (dual) testing functions. Copyright © 2012 John Wiley & Sons, Ltd.