Fast passivity enforcement for large and sparse macromodels

This work presents a fast technique for testing and enforcing passivity of linear macromodels characterized by a large and sparse structure. An optimized algorithm is proposed for the computation of the imaginary eigenvalues of the associated Hamiltonian matrix, which are iteratively perturbed until passivity is enforced. Each iteration of the proposed scheme requires a small computational cost that scales only linearly with the size of the problem.