Be the first to rate this file! 19 Downloads (last 30 days) File Size: 29.3 KB File ID: #36455

Automated Eigenvalue analysis of SD models

by

 

Identifying dominant behavior patterns, links and loops: Automated Eigenvalue analysis of SD models.

| Watch this File

File Information
Description

The method presented in this paper allows for an investigation of how model behavior is created from the underlying model structure and how this behavior feeds back to change the relative significance of the model behavior. The method also allows us to identify the dynamics of the relative significance of the various parameters that governs the gains of the links and loops of the model. The method has been implemented mathematically using Matlab based software developed for the purpose of facilitating an Eigenvalue analysis of models representing complex, dynamic systems. This work is based on control theory as well as the previous work on eigenvalue analysis in the system dynamics. It summarizes the work by Ahmed AbdelTawab AbdelGawad (2004) and Bahaa E. Aly Abdel Aleem (2004) on automating the process of analyzing the structural origin of various modes of behavior exhibited by a system dynamics model.
The method outline and code developed in preparation for this paper is publicly available, allows us to analyze the relationship between model structure and behavior and may be implemented as par of any simulation package.
The code was designed and written by Ahmed A. AbdelGawad and Bahaa E. Aly Abdel Aleem, and supervised by Mohamed M. Saleh and Pål I. Davidsen.
The code is using 3 functions ("allcycsn", "allpathn", and "reachabi") from the "Digraph toolbox" (Bahar, M.; Jantzen, J., 1995).
Read the paper at: http://www.systemdynamics.org/conferences/2005/proceed/papers/ABDEL373.pdf

Required Products Symbolic Math Toolbox
MATLAB
MATLAB release MATLAB 7 (R14)
Other requirements The code depends on the Symbolic Toolbox having Maple engine inside.
Tags for This File   Please login to tag files.
Please login to add a comment or rating.

Contact us