This is a demonstration version of GAMBIT: a Genetic Algorithm for Model-Based mixed-Integer opTimization.

Ongoing research on GAMBIT is actively under way, and a stable, fully open-source and up-to-date  version will be available soon.

Please contact k.l.sadowski@uu.nl with any questions or inquiries about GAMBIT, or the demo version.

In this Demo version, it is possible to optimize problems F1-F6:

F1: Onemax+Sphere
F2: Onemax+45DegRotatedEllipsoid
F3: DeceptiveTrap+Sphere
F4: DeceptiveTrap+45DegRotatedEllipsoid
F5: FullyDependent
F6: MixedPairsDepenent

For details on GAMBIT and the F1-F6 problems please refer to our published work, eg.:
http://homepages.cwi.nl/~bosman/publications/2015_aclusteringbased.pdf
 

It is also possible to customize the number of discrete and continuous variables in the problem, as well as set the evaluation limit.

Parameters such as population size or number of clusters are internally handled by a paramterless mechanism.

To run, execute the following command:
java -jar GAMBIT_demo.jar [problemID] [numberDvaraibles] [numberCvariables] [evaluationLimit] 

Example:

java -jar GAMBIT_demo.jar 2 20 20 1000000

Dependencies: apache.commons

Make sure apache.commons Math library is added in the java path