LU Factorization in Network Sensitivity Computations
S. Director · IEEE Transactions on Circuit Theory · 1971
The mutually reciprocal adjoint network associated with a given lumped linear time-invariant network has been shown to be an extremely useful aid for the computation of network sensitivities in automated network design algorithms. The fact that the nodal admittance matrix of the mutually reciprocal adjoint network is the transpose of the nodal admittance matrix of the given network can be employed, along with theLUfactorization of a matrix, to considerably reduce the computational effort necessary to calculate all the firstorder network sensitivities.