SADE
You can donwload from here the implementation of the algorithm described in the paper Dieletric spectra analysis: reliable parameter estimation using interval analysis. It is released under the GNU General Public License. Under UNIX-like system, just extract the archive, move to the ‘SADE’ directory and type ‘make’. Then, typing ‘make install’ will copy the executabe file into your /usr/bin directory if you want. Please make sure to have the Boost library installed, which is used for the multi-threading management. It should be available as an easy-to-install package for most of the linux distributions. For unlucky people running a Windows system, well, there is an installer providing it for the quite expansive IDE Visual Studio only. If you want to use a free IDE like this one you probably have to compile the library by yourself.
- Download: SADE.tgz
Toy program
A small python script which was used to generate some of the illustrations of the above paper. It fit an artificial set of data using the “nonlinear model” $$ f(x,A,B)=A e^{Bx} $$ It generate a TeX file containing TikZ/PGF code which, after compilation, shows graphically the result of the algorithm. The area covered by red square corresponds to parameters which leads to a ‘good’ fit (click to see the full size image):
- Download: interval.py