Scattering-Matrix Based Inexact-Newton Harmonic Balance with Incomplete Preconditioning

In harmonic-balance (HB) analysis based on Krylov-subspace methods, an effective exploitation of the available computer resources may be achieved by incomplete factorization of the preconditioner, after artificially generating a suitable sparsity pattern in the Jacobian matrix. The algorithm works best in conjunction with a new formulation of the HB errors based on the linear subnetwork scattering parameters. Important reductions of both memory occupation and CPU time may be obtained in this way, without compromising the power-handling capabilities of the analysis algorithm.