What's new
Added a new download which shows how to interface rational models with circuit solvers via a Norton equivalent and discrete convolution. The approach is applicable for models representing Y-parameters, Z-parameters, S-parameters and general transfer functions.
Major update!
New Matrix Fitting Toolbox available, for use with Y-parameter and S-parameter multi-port data sets. Includes rational fitting and passivity enforcement.
New version of Vector Fitting made available in routine vectfit3.m. Available in package VFIT3.zip. Implements the Fast Relaxed Vector Fitting (FRVF). Gives major speed improvement over vectfit¨2.m (RVF), for the fitting of long vectors.
Update to QPpassive.zip:
Added option for using sparse solver (CPLEX) with QPpassive.m. (Gives much faster computations but requires installment of CPLEX).
Updates to QPpassive.zip:
  • Removed a major bug in QPpassive.m. (The bug resulted in that parameter weightfactor was always unity, leading to too large perturbations inside fitting band.)
  • Removed a bug in QPpassive.m related to memory preallocation (wasted memory).
  • Updated user’s instructions (QPpassive.pdf).
  • Updated examples to reflect above changes
  • Removed a tiny bug from VFIT2.zip (vectfit2.m)
  • Minor updates to web pages.
  • Modified package VFIT2.zip
  • Modified the call structure (vectfit2.m)
  • Removed one crash situation (vectfit2.m)
  • Made relaxation optional (vectfit2.m)
  • Updated examples to reflect above changes
  • Added IEEE paper describing the relaxation of the least squares problem that is used in vectfit2.m (ref. [1.11]).
  • Minor modifications to the web pages
An error has been corrected in vfit2.zip, in examples ex4a.m and ex4b.m. The error was related to matrix dimensioning and gave a crash in Matlab v6.5 (but not in v7.0).
A “Links” page has been created. Included link to the IdEM software page.
Version 2.0 of vectfit.m is available in download package vfit2.zip. The new routine (vectfit2.m) has much better convergence properties than the old one (vectfit.m). It has also options for sparsity and Normal Equations to reduce computation time and memory requirements. A new set of examples is included.
The circuit drawing (fig. 2) in the user-manual of mtrxfit.zip has been corrected to match the matlab example in tutorial.m. (Thanks to Antonio Carlos Siqueira de Lima, Brazil, for spotting the error)
Several bugs in QPpassive.m (QPpassive.zip) have been corrected. (The errors were related to array cindexsub and sometimes caused the routine to produce bad results or fail).
  • Minor corrections to the vfit.zip
  • Corrections for mtrxfit.zip: Specification of real starting poles did not work correctly.
  • Minor corrections to QPpassive.zip
  • Added description to calculate zeros in vector fitting, see linkbar "Algoritm"
Passivity enforcement package made available.

Published March 20, 2013

In cooperation with:

Bjørn Gustavsen