A robust approach for finding all well-separated solutions
of sparse systems of nonlinear equations
An illustrative numerical example
A very detailed, illustrative numerical example is available here, in which the steps of the algorithm are explained with many figures, and also compared to traditional tearing.
Download the PDF file
The entire application was re-written in Python, please visit the GitHub repo for the latest version.
The unmaintained source code below (from 2016) serves for documentation purposes only. We intended to put a self-contained, usable application here; unfortunately, due to licensing issues with third party libraries, we cannot publicly distribute the entire Java application. If you must, download it together with the AMPL models here.
- src folder: the Java source code of the proof-of-concept
implementation (at revision 5889; third party libraries at 2014-May-09) is in
here. Algorithm 2 of the manuscript starts at Algorithm.java,
fixedLagIteration()function on line 300; Algorithm 4 of the manuscript starts at Algorithm.java,
reparameterize()function on line 491.
The source code serves for documentation purposes only. A major code clean-up was started but later abandoned; as a result, there are funny looking big clumps in the code (e.g. function arguments that were to be moved to their own classes with the corresponding functions, etc). The entire application was re-written in Python, please visit the GitHub repo for the latest version.
- data folder: the AMPL model files can be found here; the human
readable models are under the subfolder
originalin the corresponding directory. The other files are machine generated (with the COCONUT converters) and are not meant to be read by human beings.
- config folder: configuration files; you can safely ignore them.
The research was funded by the Austrian Science Fund (FWF): P23554, and P27891-N32. Support by the Austrian Research Promotion Agency (FFG) under project number 846920 is gratefully acknowledged.