Fræser: a Framework for Estimating Errors-in-Variables Systems

Installation and setup

No preliminary installation is required to use the estimation framework. Run the self-extracting installer from any empty directory or extract the contents of the installation package to any directory of your choice.

Some functions use external functions (.mex functions), implemented in C. To ensure that these functions work properly, compilation of external functions is necessary. In order to do so, issue


after starting the framework, as explained in Quick start. It is assumed that the MEX compiler is properly configured. If not, see the documentation for the command mex on how this can be accomplished. The release version has precompiled external functions for Win32 and Win64 platforms.

The graphical environment makes use of user interface components written in Java. These components are bundled in the Java archive (jar) file JPropertySheet.jar. In order to work property, the Java static class path in MatLab has to be extended with the full path of this file, which can be accomplished by adding it at the end of the MatLab configuration file classpath.txt. Issuing

which classpath.txt

at the MatLab command prompt will return the location of this file or

edit classpath.txt

will open it for editing. If the framework is started without the appropriate jar files on the static class path, an error message will be issued, which lists the necessary steps to take to remedy the problem.


The framework has been developed using MatLab release 2008a. As the framework extensively uses new-style classes (defined using the classdef keyword), it is incompatible with previous MatLab releases. The following toolboxes are expected to be installed:

The Optimization Toolbox may provide faster convergence for certain algorithms.

The Symbolic Math Toolbox or the symbolic polynomial class is used only in precomputing covariance matrices for nonlinear system identification with the generalized Koopmans-Levin method. (See examples with option Covariance.) The symbolic covariance matrices are persisted as regular MatLab functions with input parameters. Once the code has been generated, the Symbolic Math Toolbox or the symbolic polynomial class is no longer necessary. Likewise, it is not necessary for trying the examples for which covariance matrices are supplied as generated .m files.

The symbolic polynomial class (sympoly) is included with the framework (in directory math).

Quick start

On Windows systems, the framework is launched using


which adds any nested directories below the current path. This is necessary in order to invoke functions in descendant directories using only their name rather than their full path.

On other platforms, the nested directories have to be manually added to the MatLab path using


after starting MatLab.

In order to use the graphical interface, type


Documentation within the .m files themselves can help you get started.

Issue tracking

The software is currently under active development, you may experience abnormal operation during use. Any problems should be reported directly to the author.


Copyright © 2007-2009 Levente Hunyadi
Copyright © 2007-2009 BME Department of Automation and Applied Informatics

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

The program contains portions derived from external sources: