This library implements a framework for an evolutionary algorithm. It aims at including different techniques from the area of evolutionary computation, such as genetic algorithms and genetic programming, in one framework. It is setup such that new projects can be implemented by adjusting only the necessary parts in the library.
Get it from
SourceForge or download below.
Changelog
Revisions: problems with latex3 fixed, allmusic options works again, some new templates, new option template_list.
Description
This problem generator for dynamic routing problems can be used to study different aspects of realtime routing by changing parameters of the problem. It is written in Perl and has extensive documentation.
Downloads
These are travelling salesman problems that were created by an evolutionary algorithm with the objective function to maximise the time it takes to solve these problems by one of two LinKernighan heuristic solvers.
See these papers for a detailed description:
 Discovering the suitability of optimisation algorithms by learning from evolved instances (K. SmithMiles, J.I. van Hemert), In Annals of Mathematics and Artificial Intelligence, volume 61, 2011.
 Evolving combinatorial problem instances that are difficult to solve (J.I. van Hemert), In Evolutionary Computation, volume 14, 2006.
 Property analysis of symmetric travelling salesman problem instances acquired through evolution (J.I. van Hemert), In Evolutionary Computation in Combinatorial Optimization (G. Raidl, J. Gottlieb, eds.), Springer, 2005.
 Phase transition properties of clustered travelling salesman problem instances generated with evolutionary computation (J.I. van Hemert, N.B. Urquhart), In Parallel Problem Solving from Nature (Xin Yao, Edmund Burke, Jose A. Lozano, Jim Smith, Juan J. MereloGuervós, John A. Bullinaria, Jonathan Rowe, Peter Ti\vno Ata Kabán, HansPaul Schwefel, eds.), Springer, volume 3242, 2004.
These are travelling salesman problems that were created by an evolutionary algorithm such that they contain clusters of cities.
See these papers for a detailed description:
 Discovering the suitability of optimisation algorithms by learning from evolved instances (K. SmithMiles, J.I. van Hemert), In Annals of Mathematics and Artificial Intelligence, volume 61, 2011.
 Evolving combinatorial problem instances that are difficult to solve (J.I. van Hemert), In Evolutionary Computation, volume 14, 2006.
 Property analysis of symmetric travelling salesman problem instances acquired through evolution (J.I. van Hemert), In Evolutionary Computation in Combinatorial Optimization (G. Raidl, J. Gottlieb, eds.), Springer, 2005.
 Phase transition properties of clustered travelling salesman problem instances generated with evolutionary computation (J.I. van Hemert, N.B. Urquhart), In Parallel Problem Solving from Nature (Xin Yao, Edmund Burke, Jose A. Lozano, Jim Smith, Juan J. MereloGuervós, John A. Bullinaria, Jonathan Rowe, Peter Ti\vno Ata Kabán, HansPaul Schwefel, eds.), Springer, volume 3242, 2004.
These are binary constraint satisfaction problems created to test the performance of algorithms. The first set covers the landscape of solvable to nonsolvable problems and covers the phase transition of easy to hard to easy. The second set comprises small to large problems.
See these papers for a detailed description:
 Comparing Evolutionary Algorithms on Binary Constraint Satisfaction Problems (B.G.W. Craenen, A.E. Eiben, J.I. van Hemert), In IEEE Transactions on Evolutionary Computation, volume 7, 2003.
 Application of Evolutionary Computation to Constraint Satisfaction and Data Mining (J.I. van Hemert), PhD thesis, Leiden University, 2002. (ISBN: 906734057X)
Goals

To facilitate as a suit of programs to create and analyse randomly created binary constraint satisfaction problems,
for those who do not like to work with libraries for any reason a suite of programs is included that provides a simple interface to the library. With these programs it is easy to create a set of random problem instances, to verify solutions and to analyse instances on a number of features.

To be used as a library to implement and test new or existing constraint satisfaction solving techniques,
the library part has an extended documentation of its class hierarchy that helps new developers on their way creating new tools or solving techniques for binary constraint satisfaction. At the same time the library allows testing using a number of theoretical models.

To be freely available for anyone,
the library and programs that come with it are all licensed under the Gnu Public License, ensuring free use forever.
Changelog
Changes with previous (1.7.0) version: Added Model RB by Xu and Li; works now with GCC3.3.2; changed default output from list to matrix
Downloads