Convergence of the fictitious-current model

Abstract

When the fictitious-current model is applied for an electromagnetic-scattering problem, the accuracy and convergence of numerical results are sensitive to the distance delta(t) between the physical surface and the mathematical surface on which fictitious current sources are placed. In this model, delta(t) must be neither too small nor too large, since a small delta(t) degrades the accuracy of numerical solution, and a large delta(t) makes the moment matrix unsuited to numerical computation. Thus, when iterative methods are used, there is a trade-off between the convergence rate and the accuracy of numerical solution. Also, it is found that there exist resonant peaks depending on sampling distance delta(s), and the geometry of the scatterer. Therefore, when this model is used for CNR frequencies, spurious frequencies can be obtained. It is found that there is a linear relationship between delta(t) and delta(s) which maintain a constant condition number. Using this relation, a useful modelling method for electromagnetic problems is introduced.