ec.gp.breed.README Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of ecj Show documentation

ECJ, A Java-based Evolutionary Computation Research System. ECJ is a research EC system written in Java. It was designed to be highly flexible, with nearly all classes (and all of their settings) dynamically determined at runtime by a user-provided parameter file. All structures in the system are arranged to be easily modifiable. Even so, the system was designed with an eye toward efficiency. ECJ is developed at George Mason University's ECLab Evolutionary Computation Laboratory. The software has nothing to do with its initials' namesake, Evolutionary Computation Journal. ECJ's sister project is MASON, a multi-agent simulation system which dovetails with ECJ nicely.

The newest version!

This directory contains various breeding pipelines made for GP trees.  The
more common ones are historically elsewhere:  CrossoverPipeline and
MutationPipeline are in the 'koza' directory, and the more general 
ReproductionPipeline and MultiBreedingPipelien are in the 'ec/breed' directory.

Special-purpose pipelines in this directory:



ec.gp.breed.InternalCrossoverPipeline

Two points are selected within the same individual, such that neither is a
subtree of the other.  The points are then crossed over.


ec.gp.breed.MutateDemotePipeline

A random non-leaf GPNode is created. A tree is removed from the individual and 
replaced with the GPNode.  The tree then is reattached as a child to that
GPNode.  The remaining child slots of the GPNode, if any, are filled with
randomly generated trees (using a GPNodeBuilder).


ec.gp.breed.MutatePromotePipeline

A tree is replaced with one of its children subtrees.


ec.gp.breed.MutateSwapPipeline

Two children subtrees of a random GPNode are swapped.


ec.gp.breed.MutateOneNodePipeline

A single ERC node in the tree is mutated.


ec.gp.breed.MutateAllNodesPipeline

All ERCs in the tree are mutated.


ec.gp.breed.RehangPipeline

A nonleaf GPNode in the tree is chosen as the new root.  Call it N.  We begin 
by removing a random child subtree from that node (call it T).  The node is 
then placed at he root.  The node's former parent is placed in the position 
formerly occupied by T.  The parent's parent is placed in the position 
formerly occupied by N.  The parent's parent's parent is placed in the
position formerly occupied by N's parent.  And so on, until all parents are
exhausted.  The remaining open slot is then filled with T.  This technique
is highly randomizing and really only useful for odd tests here or there.
It also cannot be used with a typed GP problem.

ec.gp.breed.SizeFairCrossoverPipeline

Crossover is performe using either "Size Fair" or "Homologous" crossover as
defined in "Size Fair and Homologous Tree Genetic Programming Crossovers" 
by Bill Langdon (1998).  The explanation behind this code is somewhat
involved: see the ECJ manual.  This implementation is due to Uday Kamath.