1: \begin{abstract}

2: Vector Fitting is a popular method of  constructing rational approximants designed to fit given frequency response measurements.

3: The original method, which we refer to as \textsf{VF}, is based on a least-squares fit to the measurements by a rational function, using an iterative reallocation of the poles of the approximant.

4: We show that one can improve the performance  of \textsf{VF} significantly,

5: by using a particular choice of frequency sampling points and properly weighting their contribution

6: based on quadrature rules that connect the least squares objective with an $\Hardy_2$  error measure.

7: Our modified approach, designated here as \textsf{QuadVF},  helps recover the original transfer function

8: with better global fidelity  (as measured with respect to the $\Hardy_2$ norm),

9: than the localized least squares approximation implicit in \textsf{VF}.

10: We extend the new framework also to incorporate derivative information, leading to rational approximants that minimize system error with respect to a discrete Sobolev norm. 

11: We  consider the convergence behavior of both \textsf{VF} and \textsf{QuadVF} as well, and evaluate potential numerical ill-conditioning of the

12: underlying least-squares problems. 

13: We  investigate  briefly \textsf{VF} in the case of noisy measurements and propose a new formulation

14:  for the resulting approximation problem.

15: Several numerical examples are provided to support the theoretical discussion.

16: \end{abstract}

17: