Efficient Preconditioning Technique for Frequency Domain Finite Element Simulation of the Darwin Model

In electromagnetic field analysis, the Darwin model enables us to consider the effects of inductance and capacitance. In this study, we examine two interrelated Darwin model methods for accelerating the convergence of an iterative solution. One method introduces redundant variables, enlarging the size of the coefficient matrix, whereas the other method, the subspace correction (SC) method, does not. Numerical results reveal that both methods significantly improve the convergence of the Darwin model equations. In particular, the SC method is more advantageous in terms of computational cost than solving by introducing redundant variables. Moreover, it is estimated that even with redundant variables, block preconditioning can reduce the overall computational cost.