Efficient computation of network sensitivities

A class of methods for generating network functions and their sensitivities, when only one element is allowed to vary, is presented. This class of methods requires either two or three circuit analyses to obtain expressions for the network function and its first order, higher order, and large-scale sensitivities for all values of the variable component. The class of the two circuit analyses methods reduces moderately the computational effort and makes better use of circuit analysis data as compared to the previous methods which employ at least three circuit analysis. The results are applicable to network functions with bilinear characteristics and can be extended to the general class of bilinear systems.