Polynomial EVD Toolbox

This toolbox contains a number of Matlab implementations of iterative algorithms to approximate the polynomial eigenvalue decomposition (EVD) of a parahermitian matrix. Parahermitian matrices arise e.g. when formulating covariance matrices for broadband array signals, and the term parahermitian hints as an extension of the (narrowband) Hermitian property to an generalised symmetry property of the polynomial matrix case.

The toolbox files are organised in four subdirectories:

• "General" contains a number of utility functions to generate, manipulate and display polynomial matrices;
• "Decompositions" contains the decomposition algorithms; these include SBR2, SMD, and DFT-based algorithms for the extraction of an analytic EVD; these algorithms continue to evolve with various options, and are provided within this toolbox as p-code, i.e. are executable but cannot be viewed;
• "Demos" provides a number of examples of how to apply the PEVD algorithms to a number of applications;
• "Docs" contains an auto-generated html documentation of the tool box. This manual can be navigated using the lower menu in the left sidebar.

To find out more about the PEVD, the contained algorithms, the toolbox license and related issues, please follow the links provided on the left.

We hope that you find this toolbox useful, and we look forward to any comments or feedback.

Stephan Weiss, Ian Proudler, and John McWhirter (University of Strathclyde, Glasgow, Scotland)
previous contributors: Jamie Corr, Fraser Coutts