ec.app.sum.steadysum.params Maven / Gradle / Ivy

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!

# Copyright 2006 by Sean Luke and George Mason University
# Licensed under the Academic Free License version 3.0
# See the file "LICENSE" for more information

parent.0 = ../../steadystate/steadystate.params
parent.1 = sum.params

# The stuff above is all you need to do single-population
# steady-state evolution.  But there are some additional
# gizmos you should be aware of.

# 0. By default, the "deselector" (the selection operator
# responsible for determining who in the population will
# die) is tournament selection of size 1 -- that is,
# random selection.  Let's say you'd like it to be more
# aggressive, say, tournament selection of size 2, picking
# the worse individual.  You can do that like this:
# 
# steady.deselector.0.size = 	2	
#
# Or much more aggressive:
#
# steady.deselector.0.size =	7
#
# You can change the deselector entirely but it needs
# to be of SteadyStateBreedingSourceForm, and the most
# obvous choice (the classic for Steady State GA's,
# BestSelection set to pick-worst) is actually not of
# this form because it's expensive to scan through the
# whole population each time.

# 1. By default, the steady-state evolution runs for some
# number of *generations*, where a generation is defined
# as the time in which N individuals are evaluated, where
# N is the size of subpopulation 0.  The first generation
# is of course special as all the individuals are
# evaluated together.  Subsequent "generations" are just
# however long it takes for N more individuals to be
# evaluated.  You can change the run-length from being
# based on generations to being based on number of
# evaluations if you want something more precise.  To do
# that you'd set:
#
#evaluations = 100 
#
# Which says how many evaluations you want.  Obviously
# this must be larger than the initial population size...

# 2. When you run in its current form, this parameter
# file will cause the system to complain that you didn't
# provide a SteadyStateStatistics object.  This is okay,
# but you may not get the statistics you expect.  For
# more information, see SteadyStateEvolutionState

# statistics =		...

# 3. You want to specify whether or not to allow any
# children to enter the population if they are identical
# to some existing member of the population.  Disallowing
# duplicates in this way is the traditional approach to
# steady-state evolution.  At present, each time we 
# need to make an individual, we try up to 100 times to
# create one that's not a duplicate.  You can change
# the number of times to retry (or set to 0 to not
# retry) by changing this parameter:

# steady.duplicate-retries = 		...