Generation of immittance functions in symbolic form for lumped distributed active networks

An efficient method for numerically generating the immittance and sensitivity parameters of linear time-invariant lumped distributed active networks as explicit functions of the frequency variable s is presented. It is based on the fast Fourier transform (FFT) algorithm for numerical interpolation, the sparse-matrix method for efficient computation and low storage, and the adjoint-network approach for sensitivity evaluation. The situation when a few of the network element values are to be left as symbols is considered. The influence of various numerical errors is investigated and ways of reducing them are proposed.