On the Convergence of the Vector-Fitting Algorithm

Vector fitting (VF), first published by Gustavesen and Semlyen in 1999, is widely used for constructing rational models from measured or full-wave electromagnetic simulated frequency-domain responses (<formula formulatype="inline"> <tex Notation="TeX">$S$</tex></formula>-, <formula formulatype="inline"><tex Notation="TeX">$Y$</tex></formula>-, or <formula formulatype="inline"><tex Notation="TeX">$Z$</tex></formula>-parameters). As pointed out by Grivet-Talocia and Bandinu in 2006, to date there is no convergence analysis of the pole relocation iteration in VF. The goal of this paper is to elucidate this issue. It will be shown that the iteration seeks the roots of a set of coupled multivariate rational equations. For noise-free measurements, it is shown that there is no iteration involved, assuming that the number of starting poles is chosen greater or equal to the order of the underlying system. For noisy data, the VF iteration may not find any solution due to the fact that all stationary points of the fixed point iteration are repelling. Therefore, in case the iteration does not converge, we propose to incorporate the Newton step in the VF iteration, thus guaranteeing local convergence. Lastly, we provide a short review of variable projection as an alternative to VF.

[1]  L. Mcbride,et al.  A technique for the identification of linear systems , 1965 .

[2]  T. Söderström,et al.  The Steiglitz-McBride identification algorithm revisited--Convergence analysis and accuracy aspects , 1981 .

[3]  Gene H. Golub,et al.  The differentiation of pseudo-inverses and non-linear least squares problems whose variables separate , 1972, Milestones in Matrix Computation.

[4]  T. Dhaene,et al.  Orthonormal Vector Fitting: A Robust Macromodeling Tool for Rational Approximation of Frequency Domain Responses , 2007, IEEE Transactions on Advanced Packaging.

[5]  T. Dhaene,et al.  Robust Parametric Macromodeling Using Multivariate Orthonormal Vector Fitting , 2008, IEEE Transactions on Microwave Theory and Techniques.

[6]  B. Gustavsen,et al.  Improving the pole relocating properties of vector fitting , 2006, 2006 IEEE Power Engineering Society General Meeting.

[7]  E. C. Levy.  Complex-curve fitting , 1959, IRE Transactions on Automatic Control.

[8]  Tom Dhaene,et al.  Vector Fitting vs. Levenberg-Marquardt : Some experiments , 2009, 2009 IEEE Workshop on Signal Propagation on Interconnects.

[9]  Athanasios C. Antoulas,et al.  A New Approach to Modeling Multiport Systems From Frequency-Domain Data , 2010, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[10]  Phillip A. Regalia,et al.  On the existence of stationary points for the Steiglitz-McBride algorithm , 1997, IEEE Trans. Autom. Control..

[11]  S. Grivet-Talocia,et al.  Improving the convergence of vector fitting for equivalent circuit extraction from noisy frequency responses , 2006, IEEE Transactions on Electromagnetic Compatibility.

[12]  G. Golub,et al.  Separable nonlinear least squares: the variable projection method and its applications , 2003 .

[13]  C. Sanathanan,et al.  Transfer function synthesis as a ratio of two complex polynomials , 1963 .

[14]  A. Semlyen,et al.  Rational approximation of frequency domain responses by vector fitting , 1999 .

[15]  P. Stoica,et al.  Convergence properties of "A method for the indentification of linear systems using the generalized least squares principle" , 1979 .