Page 1
GAlib: A C++ Library of Genetic Algorithm Components
version 2.4
Documentation Revision B
August 1996
Matthew Wall
Mechanical Engineering Department
Massachusetts Institute of Technology
http://lancet.mit.edu/ga/
galib-request@mit.edu
Copyright © 1996 Matthew Wall
all rights reserved
GAlib is a C++ library of genetic algorithm objects. The library includes tools for using
genetic algorithms to do optimization in any C++ program using any representation
and any genetic operators. This documentation includes an extensive overview of how
to implement a genetic algorithm, the programming interface for GAlib classes, and
examples illustrating customizations to the GAlib classes.
This work was supported by the Leaders for Manufacturing Program
Page 2
Contents
i
GAlib: A C++ Library of Genetic Algorithm Components
i
Licensing and Copyright Issues
1
GAlib For-profit User/Distributor License Agreement
1
GAlib Not-for-profit User License Agreement
1
The GNU portions of the GAlib distribution
1
The standard MIT copyright notice and disclaimer
2
Features
3
General Features
3
Algorithms, Parameters, and Statistics
3
Genomes and Operators
3
Overview
5
The Genetic Algorithm
6
Defining a Representation
7
The Genome Operators
7
The Population Object
8
Objective Functions and Fitness Scaling
8
So what does it look like in C++?
9
What can the operators do?
10
How do I define my own operators?
11
What about deriving my own genome class?
13
Class Hierarchy
15
GAlib Class Hierarchy - Pictorial
15
GAlib Class Hierarchy - Outline
16
Programming Interface
17
Global Typedefs and Enumerations
17
Global Variables and Global Constants
17
Function Prototypes
17
Parameter Names and Command-Line Options
18
Error Handling
20
Random Number Functions
21
GAGeneticAlgorithm
22
GADemeGA
27
GAIncrementalGA
29
GASimpleGA
31
GASteadyStateGA
32
Terminators
34
Replacement Schemes
35
Page 3
Contents
ii
GAGenome
36
GA1DArrayGenome<T>
39
GA1DArrayAlleleGenome<T>
41
GA2DArrayGenome<T>
42
GA2DArrayAlleleGenome<T>
44
GA3DArrayGenome<T>
45
GA3DArrayAlleleGenome<T>
47
GA1DBinaryStringGenome
48
GA2DBinaryStringGenome
50
GA3DBinaryStringGenome
52
GABin2DecGenome
54
GAListGenome<T>
56
GARealGenome
57
GAStringGenome
58
GATreeGenome<T>
59
GAEvalData
60
GABin2DecPhenotype
61
GAAlleleSet<T>
62
GAAlleleSetArray<T>
64
GAParameter and GAParameterList
65
GAStatistics
67
GAPopulation
70
GAScalingScheme
75
GASelectionScheme
77
GAArray<T>
79
GABinaryString
81
GAList<T> and GAListIter<T>
82
GATree<T> and GATreeIter<T>
85
Customizing GAlib
89
Deriving your own genome class
89
Genome Initialization
91
Genome Mutation
91
Genome Crossover
92
Genome Comparison
92
Genome Evaluation
93
Population Initialization
93
Population Evaluation
93
Scaling Scheme
93
Selection Scheme
94
Genetic Algorithm
96
Termination Function
96
Descriptions of the Examples
98
Page 4
Licensing and Copyright Issues
GAlib Version 2.4, Document Revision B
1
19-Aug-96
Licensing and Copyright Issues
The GAlib source code is not in the public domain, but it is available at no cost for non-profit purposes.
If you would like to use GAlib for commercial purposes, for-profit single user and distributor licenses
are available. All of GAlib (source and documentation) is protected by the Berne Convention. You may
copy and modify GAlib, but by doing so you agree to the terms of the not-for-profit license.
GAlib For-profit User/Distributor License Agreement
Please contact the MIT Technology Licensing Office at 617.253.6966 or tlo@mit.edu.
GAlib Not-for-profit User License Agreement
1. You may copy and distribute copies of the source code for GAlib in any medium provided that you
consipicuously and appropriately give credit to the author and keep intact all copyright and disclaimer
notices in the library.
2. You may modify your copy (copies) of GAlib or any portion thereof, but you may not distribute
modified versions of GAlib. You may distribute patches to the original GAlib as separate files along
with the original GAlib.
3. You may not charge anything for copies of GAlib beyond a fair estimate of the cost of media and
computer/network time required to make and distribute the copies.
4. Incorporation of GAlib or any portion thereof into commercial software, distribution of GAlib for-
profit, or use of GAlib for other for-profit purposes requires a special agreement with the the MIT
technology licensing office (TLO).
5. Any publications of work based upon experiments that use GAlib must include a suitable
acknowledgement of GAlib. A suggested acknowledgement is: "The software for this work used the
GAlib genetic algorithm package, written by Matthew Wall at the Massachusetts Institute of
Technology."
6. The author of GAlib and MIT assume absolutely no responsibility for the use or misuse of GAlib. In
no event shall the author of GAlib or MIT be liable for any damages resulting from use or performance
of GAlib.
The GNU portions of the GAlib distribution
The portions of GAlib (see below) that contain code from the GNU g++ library are covered under the
terms of the GNU Public License. As such they are freely available and do not fall under the terms of
the GAlib licensing conditions above. The portions of GAlib that are based upon GNU code are all in
the 'gnu' directory in the examples directory (in GAlib release 2.3.2 and later).
Page 5
Licensing and Copyright Issues
GAlib Version 2.4, Document Revision B
2
19-Aug-96
The standard MIT copyright notice and disclaimer
As a work developed using MIT resources and MIT funding, the GAlib source code copyright is owned
by the Massachusetts Institute of Technology. All rights are reserved.
Copyright (c) 1995-1996 Massachusetts Institute of Technology
Permission to use, copy, modify, and distribute this software and its documentation for any non-
commercial purpose and without fee is hereby granted, provided that
•
the above copyright notice appear in all copies
•
both the copyright notice and this permission notice appear in supporting documentation
•
the name of M.I.T. not be used in advertising or publicity pertaining to distribution of the software
without specific, written prior permission.
M.I.T. makes no representations about the suitability of this software for any purpose. It is provided "as
is" without express or implied warranty.
M.I.T. DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL M.I.T. BE
LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
Page 6
Features: General Features
GAlib Version 2.4, Document Revision B
3
19-Aug-96
Features
General Features
•
Many examples are included illustrating the use of various GAlib features, class derivations,
parallelization, deterministic crowding, travelling salesman, DeJong, and Royal Road problems.
•
The library has been used on various DOS/Windows, Windows NT/95, MacOS, and UNIX
configurations. GAlib compiles without warnings on most major compilers.
•
Templates are used in some genome classes, but GAlib can be used without templates if your
compiler does not understand them.
•
Four random number generators are included with the library. You can select the one most
appropriate for your system, or use your own.
Algorithms, Parameters, and Statistics
•
GAlib can be used with PVM (parallel virtual machine) to evolve populations and/or individuals in
parallel on multiple CPUs.
•
Genetic algorithm parameters can be configured from file, command-line, and/or code.
•
Overlapping (steady-state GA) and non-overlapping (simple GA) populations are supported. You
can also specify the amount of overlap (% replacement). The distribution includes examples of other
derived genetic algorithms such as a genetic algorithm with sub-populations and another that uses
deterministic crowding.
•
New genetic algorithms can be quickly tested by deriving from the base genetic algorithm classes
in the library. In many cases you need only overide one virtual function.
•
Built-in termination methods include convergence and number-of-generations. The termination
method can be customized for any existing genetic algorithm class or for new classes you derive.
•
Speciation can be done with either DeJong-style crowding (using a replacement strategy) or
Goldberg-style sharing (using fitness scaling).
•
Elitism is optional for non-overlapping genetic algorithms.
•
Built-in replacement strategies (for overlapping populations) include replace parent, replace
random, replace worst. The replacement operator can be customized.
•
Built-in selection methods include rank, roulette wheel, tournament, stochastic remainder sampling,
stochastic uniform sampling, and deterministic sampling. The selection operator can be customized.
•
"on-line" and "off-line" statistics are recorded as well as max, min, mean, standard deviation, and
diversity. You can specify which statistics should be recorded and how often they should be flushed
to file.
Genomes and Operators
•
Chromosomes can be built from any C++ data type. You can use the types built-in to the library
(bit-string, array, list, tree) or derive a chromosome based on your own objects.
Page 7
Features: Genomes and Operators
GAlib Version 2.4, Document Revision B
4
19-Aug-96
•
Built-in chromosome types include real number arrays, list, tree, 1D, 2D, and 3D arrays, 1D, 2D,
and 3D binary string. The binary strings, strings, and arrays can be variable length. The lists and
trees can contain any object in their nodes. The array can contain any object in each element.
•
All chromosome initialization, mutation, crossover, and comparison methods can be customized.
•
Built-in initialization operators include uniform random, order-based random, and initialize-to-zero.
•
Built-in mutation operators include random flip, random swap, Gaussian, destructive, swap subtree,
swap node.
•
Built-in crossover operators include partial match, ordered, cycle, single point, two point, even, odd,
uniform, node- and subtree-single point.
•
Dominance and Diploidy are not explicitly built in to the library, but any of the genome classes in
the library can easily be extended to become diploid chromosomes.
•
Objective function
•
Objective functions can be population- or individual-based.
•
If the built-in genomes adequately represent your problem, a user-specified objective function is the
only problem-specific code that must be written.
Page 8
Overview: Genomes and Operators
GAlib Version 2.4, Document Revision B
5
19-Aug-96
Overview
This document outlines the contents of the library and presents some of the design philosophy behind
the implementation. Some source code samples are provided at the end of the page to illustrate basic
program structure, operator capabilities, operator customization, and derivation of new genome classes.
When you use the library you will work primarily with two classes: a genome and a genetic algorithm.
Each genome instance represents a single solution to your problem. The genetic algorithm object defines
how the evolution should take place. The genetic algorithm uses an objective function (defined by you)
to determine how 'fit' each genome is for survival. It uses the genome operators (built into the genome)
and selection/replacement strategies (built into the genetic algorithm) to generate new individuals.
There are three things you must do to solve a problem using a genetic algorithm:
1. Define a representation
2. Define the genetic operators
3. Define the objective function
GAlib helps you with the first two items by providing many examples and pieces from which you can
build your representation and operators. In many cases you can use the built-in representations and
operators with little or no modification. The objective function is completely up to you. Once you have a
representation, operators, and objective measure, you can apply any genetic algorithm to find better
solutions to your problem.
When you use a genetic algorithm to solve an optimization problem, you must be able to represent a
single solution to your problem in a single data structure. The genetic algorithm will create a population
of solutions based on a sample data structure that you provide. The genetic algorithm then operates on
the population to evolve the best solution. In GAlib, the sample data structure is called a GAGenome
(some people refer to it as a chromosome). The library contains four types of genomes: GAListGenome,
GATreeGenome, GAArrayGenome, and GABinaryStringGenome. These classes are derived from the
base GAGenome class and a data structure class as indicated by their names. For example, the
GAListGenome is derived from the GAList class as well as the GAGenome class. Use a data structure
that works with your problem definition. For example, if you are trying to optimize a function that
depends on 5 real numbers, then use as your genome a 1-dimensional array of floats with 5 elements.
There are many different types of genetic algorithms. GAlib includes three basic types: 'simple',
'steady-state', and 'incremental'. These algorithms differ in the way that they create new individuals and
replace old individuals during the course of an evolution.
GAlib provides two primary mechanisms for extending the capabilities of built-in objects. First of all
(and most preferred, from a C++ point of view), you can derive your own classes and define new
member functions. If you need to make only minor adjustments to the behavior of a GAlib class, in
most cases you can define a single function and tell the existing GAlib class to use it instead of the
default.
Genetic algorithms, when properly implemented, are capable of both exploration (broad search) and
exploitation (local search) of the search space. The type of behavior you'll get depends on how the
operators work and on the 'shape' of the search space.
Page 9
Overview: The Genetic Algorithm
GAlib Version 2.4, Document Revision B
6
19-Aug-96
The Genetic Algorithm
The genetic algorithm object determines which individuals
should survive, which should reproduce, and which
should die. It also records statistics and decides how long
the evolution should continue. Typically a genetic
algorithm has no obvious stopping criterion. You must tell
the algorithm when to stop. Often the number-of-
generations is used as a stopping measure, but you can
use goodness-of-best-solution, convergence-of-population,
or any problem-specific criterion if you prefer.
The library contains four flavors of genetic algorithms. The
first is the standard 'simple genetic algorithm' described
by Goldberg in his book. This algorithm uses non-
overlapping populations and optional elitism. Each
generation the algorithm creates an entirely new
population of individuals. The second is a 'steady-state
genetic algorithm' that uses overlapping populations. In
this variation, you can specify how much of the population
should be replaced in each generation. The third variation
is the 'incremental genetic algorithm', in which each
generation consists of only one or two children. The
incremental genetic algorithms allow custom replacement
methods to define how the new generation should be
integrated into the population. So, for example, a newly generated child could replace its parent,
replace a random individual in the population, or replace an individual that is most like it. The fourth
type is the 'deme' genetic algorithm. This algorithm evolves multiple populations in parallel using a
steady-state algorithm. Each generation the algorithm migrates some of the individuals from each
population to one of the other populations.
In addition to the basic built-in types, GAlib defines the components you'll need to derive your own
genetic algorithm classes. The examples include a few of these derivations including (1) a genetic
algorithm that uses multiple populations and 'migration' between populations on multiple CPUs, and
(2) a genetic algorithm that does 'deterministic crowding' to maintain different species of individuals
during the evolution.
The base genetic algorithm class contains operators and data common to most flavors of genetic
algorithms. When you derive your own genetic algorithm you can use these member data and
functions to keep track of statistics and monitor performance.
The genetic algorithm contains the statistics, replacement strategy, and parameters for running the
algorithm. the population object, a container for genomes, also contains some statistics as well as
selection and scaling operators. A typical genetic algorithm will run forever. The library has built in
functions for specifying when the algorithm should terminate. These include terminate-upon-
generation, in which you specify a certain number of generations for which the algorithm should run,
and terminate-upon-convergence, in which you specify a value to which the best-of-generation score
should converge. You can customize the termination function to use your own stopping criterion.
The number of function evaluations is a good way to compare different genetic algorithms with various
other search methods. The GAlib genetic algorithms keep track of both the number of genome
evaluations and population evaluations.
initialize
population
select individuals
for mating
mate individuals
to produce offspring
are stopping
criteria satisfied?
finish
insert offspring
into population
mutate offspring
Page 10
Overview: Defining a Representation
GAlib Version 2.4, Document Revision B
7
19-Aug-96
Defining a Representation
Use a data structure that is appropriate for your problem. If you are optimizing a function of real
numbers, use real numbers in your genome. If a solution to your problem can be represented with
some imaginary numbers and some integer values, define a genome with these characteristics.
Defining an appropriate representation is part of the art of using genetic algorithms (and at this point, it
is still an art, not a science). Use a representation that is minimal but completely expressive. Your
representation should be able to represent any solution to your problem, but if at all possible you
should design it so that it cannot represent infeasible solutions to your problem. Remember that if the
genome can represent infeasible solutions then the objective function must be designed to give partial
credit to infeasibles.
The representation should not contain information beyond that needed to represent a solution to the
problem. Although there may be merit in using a representation that contains 'extra' genetic material,
unless properly implemented (in concert with the objective function and in full consideration of the type
and characteristics of the search space), this tends to increase the size of the search space and thus hinder
the performance of the genetic algorithm.
The number of possible representations is endless. You may choose a purely numeric representation
such as an array of real numbers. These could be implemented as real numbers, or, in the Goldberg-
style of a string of bits that map to real numbers (beware that using real numbers directly far out-
performs the binary-to-decimal representation for most problems, especially when you use reasonable
crossover operators). Your problem may depend on a sequence of items, in which case an order-based
representation (either list or array) may be more appropriate. In many of these cases, you must choose
operators that maintain the integrity of the sequence; crossover must generate reordered lists without
duplicating any element in the list. Other problems lend themselves to a tree structure. Here you may
want to represent solutions explicitly as trees and perform the genetic operations on the trees directly.
Alternatively, many people encode trees into an array or parsable string, then operate on the string.
Some problems include a mix of continuous and discrete elements, in which case you may need to
create a new structure to hold the mix of information. In these cases you must define genetic operators
that respect the structure of the solution. For example, a solution with both integer and floating parts
might use a crossover that crosses integer parts with integer parts and floating parts with floating parts,
but never mixes floating parts with integer parts.
Whichever representation you choose, be sure to pick operators that are appropriate for your
representation.
The Genome Operators
Each genome has three primary operators: initialization, mutation, and crossover. With these operators
you can bias an initial population, define a mutation or crossover specific to your problem's
representation, or evolve parts of the genetic algorithm as your population evolves. GAlib comes with
these operators pre-defined for each genome type, but you can customize any of them.
The initialization operator determines how the genome is initialized. It is called when you initialize a
population or the genetic algorithm. This operator does not actually create new genomes, rather it 'stuffs'
the genomes with the primordial genetic material from which all solutions will evolve. The population
object has its own initialization operator. By default this simply calls the initialization operators of the
genomes in the population, but you can customize it to do whatever you want.
The mutation operator defines the procedure for mutating each genome. Mutation means different
things for different data types. For example, a typical mutator for a binary string genome flips the bits
in the string with a given probability. A typical mutator for a tree, on the other hand, would swap
subtrees with a given probability. In general, you should define a mutation that can do both exploration
Page 11
Overview: The Population Object
GAlib Version 2.4, Document Revision B
8
19-Aug-96
and exploitation; mutation should be able to introduce new genetic material as well as modify existing
material. You may want to define multiple types of mutation for a single problem.
The crossover operator defines the procedure for generating a child from two parent genomes. Like the
mutation operator, crossover is specific to the data type. Unlike mutation, however, crossover involves
multiple genomes. In GAlib, each genome 'knows' its preferred method of mating (the default crossover
method) but it is incapable of performing crossover itself. Each genetic algorithm 'knows' how to get the
default crossover method from its genomes then use that method to peform the mating. With this model
it is possible to derive new genetic algorithm classes that use mating methods other than the defaults
defined for a genome.
Each of these methods can be customized so that it is specific not only to the data type, but also to the
problem type. This is one way you can put some problem-specific 'intelligence' into the genetic
algorithm (I won't go into a discussion about whether or not this is a good thing to do...)
In addition to the three primary operators, each genome must also contain an objective function and
may also contain a comparator. The objective function is used to evaluate the genome. The comparator
(often referred to as a 'distance function') is used to determine how different one genome is from
another. Every genetic algorithm requires that an objective function is defined - this is how the genetic
algorithm determines which individuals are better than others. Some genetic algorithms require a
comparator.
The library has some basic data types built in, but if you already have an array or list object, for
example, then you can quickly build a genome from it by multiply inheriting from your object and the
genome object. You can then use this new object directly in the GAlib genetic algorithm objects.
In general, a genetic algorithm does not need to know about the contents of the data structures on which
it is operating. The library reflects this generality. You can mix and match genome types with genetic
algorithms. The
genetic algorithm knows how to clone genomes in order to create populations, initialize genomes to start
a run, cross genomes to generate children, and mutate genomes. All of these operations are performed
via the genome member functions.
The Population Object
The population object is a container for genomes. Each population object has its own initializer (the
default simply calls the initializer for each individual in the population) and evaluator (the default
simply calls the evaluator for each individual in the population). It also keeps track of the best, average,
deviation, etc for the population. Diversity can be recorded as well, but since diversity calculations often
require a great deal of additional compuation, the default is to not record diversity.
The selection method is also defined in the population object. This method is used by the genetic
algorithms to choose which individuals should mate.
Each population object has a scaling scheme object associated with it. The scaling scheme object converts
the objective score of each genome to a fitness score that the genetic algorithm uses for selection. It also
caches fitness information for use later on by the selection schemes.
Objective Functions and Fitness Scaling
Genetic algorithms are often more attractive than gradient search methods because they do not require
compilicated differential equations or a smooth search space. The genetic algorithm needs only a single
measure of how good a single individual is compared to the other individuals. The objective function
provides this measure; given a single solution to a problem, how good is it?
Page 12
Overview: So what does it look like in C++?
GAlib Version 2.4, Document Revision B
9
19-Aug-96
It is important to note the distinction between fitness and objective scores. The objective score is the
value returned by your objective function; it is the raw performance evaluation of a genome. The fitness
score, on the other hand, is a possibly-transformed rating used by the genetic algorithm to determine
the fitness of individuals for mating. The fitness score is typically obtained by a linear scaling of the raw
objective scores (but you can define any mapping you want, or no transformation at all). For example, if
you use linear scaling then the fitness scores are derived from the objective scores using the fitness
proportional scaling technique described in Goldberg's book. The genetic algorithm uses the fitness
scores, not the objective scores, to do selection.
You can evaluate the individuals in a population using an individual-based evaluation function (which
you define), or a population-based evaluator (also which you define). If you use an individual-based
objective, then the function is assigned to each genome. A population-based objective function can make
use of individual objective functions, or it can set the individual scores itself.
So what does it look like in C++?
A typical optimization program has the following form. This example creates a one-dimensional binary
string genome with the default operators then uses a simple genetic algorithm to do the evolution.
float Objective(GAGenome&);
main(){
GA1DBinaryStringGenome genome(length, Objective);
// create a genome
GASimpleGA ga(genome);
// create the genetic algorithm
ga.evolve();
// do the evolution
cout << ga.statistics() << endl;
// print out the results
}
float Objective(GAGenome&) {
// your objective function goes here
}
You can very easily change the behaviour of the genetic algorithm by setting various parameters. Some
of the more common ones are set like this:
ga.populationSize(popsize);
ga.nGenerations(ngen);
ga.pMutation(pmut);
ga.pCrossover(pcross);
GASigmaTruncationScaling sigmaTruncation;
ga.scaling(sigmaTruncation);
Alternatively you can have GAlib read the genetic algorithm options from a file or from the command
line. This snippet creates a genetic algorithm, reads the parameters from a file, reads parameters (if any)
from the command line, performs the evolution, then prints out the statistics from the run.
GASteadyStateGA ga(genome);
ga.parameters("settings.txt");
ga.parameters(argc, argv);
ga.evolve();
cout << ga.statistics() << endl;
A typical (albeit simple) objective function looks like this (this one gives a higher score to a binary
string genome that contains all 1s):
float
Objective(GAGenome & g) {
GA1DBinaryStringGenome & genome = (GA1DBinaryStringGenome &)g;
float score=0.0;
for(int i=0; i<genome.length(); i++)
score += genome.gene(i);
return score;
Page 13
Overview: What can the operators do?
GAlib Version 2.4, Document Revision B
10
19-Aug-96
}
You can define the objective function as a static member of a derived class, or just define a function and
use it with the existing GAlib genome classes.
When you write an objective function, you must first cast the generic genome into the type of genome
that your objective function is expecting. From that point on you can work with the specific genome
type. Each objective function returns a single value that represents the objective score of the genome
that was passed to the objective function.
Please see the examples for more samples of the library in action. And see the programming interface
page for a complete list of member functions and built-in operators.
What can the operators do?
Here are some examples of the types of mutation and crossover that can be done using GAlib.
Traditional crossover generates two children from two parents, and mutation is typically applied to a
single individual. However, many other types of crossover and mutation are possible, such as crossover
using three or more parents, asexual crossover or population-based mutation. The following examples
illustrate some of the standard, sexual crossover and individual mutation methods in GAlib.
tree node swap mutation
tree sub-tree swap mutation
tree one point crossover
array one point crossover
array two point crossover
list one point crossover
list order crossover
list node swap mutation
list destructive mutation
list generative mutation
list sequence swap mutation
Page 14
Overview: How do I define my own operators?
GAlib Version 2.4, Document Revision B
11
19-Aug-96
How do I define my own operators?
Defining the operators is only as difficult as figuring out the algorithm you want to implement. As far
as the actual implementation goes, there's not much to it. To assign an operator to a genome, just use the
appropriate member function. For example, the following code snippet assigns 'MyInitializer' as the
initialization function and 'MyCrossover' as the crossover function for a binary string genome.
GA1DBinaryStringGenome genome(20);
genome.initializer(MyInitializer);
genome.crossover(MyCrossover);
If you do this to the first genome (the one you use to create the genetic algorithm) then all of the ones
that the GA clones will have the same operators defined for them.
When you derive your own genome class you will typically hard-code the operators into the genome
like this:
class MyGenome : public GAGenome {
public:
static void RandomInitializer(GAGenome&);
static int JuggleCrossover(const GAGenome&, const GAGenome&, GAGenome*, GAGenome*);
static int KillerMutate(GAGenome&, float);
static float ElementComparator(const GAGenome&, const GAGenome&);
static float ThresholdObjective(GAGenome&);
public:
MyGenome() {
initializer(RandomInitializer);
crossover(JuggleCrossover);
mutator(KillerMutate);
comparator(ElementComparator);
evaluator(ThresholdObjective);
}
// remainder of class definition here
};
Notice how easy it becomes to change operators. You can very easily define a multitude of operators for
a single representation and experiment with them to see which performs better.
Why are the genome operators GAlib not member functions? The primary reason is so that you do not
have to derive a new class in order to change the behaviour of one of the built-in genome types. In
addition, the use of function pointers rather than member functions lets us change operators at run-time
(unlike member functions or templatized classes). And they are faster than virtual functions (OK, so this
the virtual/non-virtual component is a pretty small fraction of actual execution time compared to most
objective functions...). On the down side, they permit you to make some ugly mistakes by improperly
casting.
The definition for the List1PtCrossover looks like this:
This crossover picks a single point in the parents then generates one or two children from the
two halves of each parent.
template <class T> int
OnePointCrossover(const GAGenome& p1, const GAGenome& p2, GAGenome* c1, GAGenome* c2){
GAListGenome<T> &mom=(GAListGenome<T> &)p1;
GAListGenome<T> &dad=(GAListGenome<T> &)p2;
int nc=0;
unsigned int a = GARandomInt(0, mom.size());
unsigned int b = GARandomInt(0, dad.size()); GAList<T> * list;
Page 15
Overview: How do I define my own operators?
GAlib Version 2.4, Document Revision B
12
19-Aug-96
// first do the sister...
if(c1){
GAListGenome<T> &sis=(GAListGenome<T> &)*c1;
sis.GAList<T>::copy(mom);
list = dad.GAList<T>::clone(b);
if(a < mom.size()){
T *site = sis.warp(a);
while(sis.tail() != site)
sis.destroy();
// delete the tail node
sis.destroy();
// trash the trailing node (list[a])
}
else{
sis.tail();
// move to the end of the list
}
sis.insert(list);
// stick the clone onto the end
delete list;
sis.warp(0);
// set iterator to head of list
nc += 1;
}
// ...now do the brother
if(c2){
GAListGenome<T> &bro=(GAListGenome<T> &)*c2;
bro.GAList<T>::copy(dad);
list = mom.GAList<T>::clone(a);
if(b < dad.size()){
T *site = bro.warp(b);
while(bro.tail() != site)
bro.destroy();
// delete the tail node
bro.destroy();
// trash the trailing node (list[a])
}
else{
bro.tail();
// move to the end of the list
}
bro.insert(list);
// stick the clone onto the end
delete list;
bro.warp(0);
// set iterator to head of list
nc += 1;
}
return nc;
}
The definition for FlipMutator for 1DArrayAlleleGenomes looks like this:
This mutator flips the value of a single element of the array to any of the possible allele values.
int
FlipMutator(GAGenome & c, float pmut) {
GA1DArrayAlleleGenome<T> &child=(GA1DArrayAlleleGenome<T> &)c;
register int n, i;
if(pmut <= 0.0) return(0);
float nMut = pmut * (float)(child.length());
if(nMut < 1.0){
// we have to do a flip test on each bit
nMut = 0;
for(i=child.length()-1; i>=0; i--){
if(GAFlipCoin(pmut)){
child.gene(i, child.alleleset().allele());
nMut++;
}
}
}
else{
// only flip the number of bits we need to flip
for(n=0; n<nMut; n++){
i = GARandomInt(0, child.length()-1);
child.gene(i, child.alleleset().allele());
Page 16
Overview: What about deriving my own genome class?
GAlib Version 2.4, Document Revision B
13
19-Aug-96
}
}
return((int)nMut);
}
And the definition for a typical initializer looks like this:
This initializer creates a tree of bounded random size and forkiness.
void
TreeInitializer(GAGenome & c) {
GATreeGenome<Point> &tree=(GATreeGenome<Point> &)c;
tree.root();
tree.destroy(); // destroy any pre-existing tree
Point p(0,0,0);
tree.insert(p,GATreeBASE::ROOT);
int n = GARandomInt(0,MAX_CHILDREN); // limit number of children
for(int i=0; i<n; i++)
DoChild(tree, 0);
}
void
DoChild(GATreeGenome<Point> & tree, int depth) {
if(depth >= MAX_DEPTH) return;
// limit depth of the tree
int n = GARandomInt(0,MAX_CHILDREN); // limit number of children
Point p(GARandomFloat(0,25), GARandomFloat(0,25), GARandomFloat(0,25));
tree.insert(p,GATreeBASE::BELOW);
for(int i=0; i<n; i++)
DoChild(tree, depth+1);
tree.parent();
// move the iterator up one level
}
What about deriving my own genome class?
Here is the definition of a genome that contains an arbitrary number of lists. It could easily be modified
to become a diploid genome. It is used in exactly the same way that the built-in genomes are used. For
a simpler example, see the GNU example which integrates the GNU BitString object with GAlib to form
a new genome class.
class RobotPathGenome : public GAGenome {
public:
GADefineIdentity("RobotPathGenome", 251);
static void Initializer(GAGenome&);
static int Mutator(GAGenome&, float);
static float Comparator(const GAGenome&, const GAGenome&);
static float Evaluator(GAGenome&);
static void PathInitializer(GAGenome&);
public:
RobotPathGenome(int nrobots, int pathlength);
RobotPathGenome(const RobotPathGenome & orig);
RobotPathGenome& operator=(const GAGenome & arg);
virtual ~RobotPathGenome();
virtual GAGenome *clone(GAGenome::CloneMethod) const ;
virtual void copy(const GAGenome & c);
virtual int equal(const GAGenome& g) const ;
virtual int read(istream & is);
virtual int write(ostream & os) const ;
GAListGenome<int> & path(int i){ return *list[i]; }
int npaths() const { return n; }
int length() const { return l; }
protected:
Page 17
Overview: What about deriving my own genome class?
GAlib Version 2.4, Document Revision B
14
19-Aug-96
int n, l;
GAListGenome<int> **list;
};
Page 18
Class Hierarchy: GAlib Class Hierarchy - Pictorial
GAlib Version 2.4, Document Revision B
15
19-Aug-96
Class Hierarchy
Here is an outline of the GAlib class hierarchy. The first section is a graphic map, the second section
contains an outline of the hierarchy.
GAlib Class Hierarchy - Pictorial
Page 19
Class Hierarchy: GAlib Class Hierarchy - Outline
GAlib Version 2.4, Document Revision B
16
19-Aug-96
GAlib Class Hierarchy - Outline
GAGeneticAlgorithm
GASteadyStateGA (overlapping populations)
GASimpleGA (non-overlapping populations)
GAIncrementalGA (overlapping with custom replacement)
GADemeGA (parallel populations with migration)
GAStatistics
GAParameterList
GAPopulation
GAScalingScheme
GANoScaling
GALinearScaling
GASigmaTruncationScaling
GAPowerLawScaling
GASharing
GASelectionScheme
GARankSelector
GARouletteWheelSelector
GATournamentSelector
GAUniformSelector
GASRSSelector
GADSSelector
GAGenome
GA1DBinaryStringGenome
GABin2DecGenome
GA2DBinaryStringGenome
GA3DBinaryStringGenome
GA1DArrayGenome<>
GA1DArrayAlleleGenome<>
GAStringGenome (same as GA1DArrayAlleleGenome<char>)
GARealGenome (same as GA1DArrayAlleleGenome<float>)
GA2DArrayGenome<>
GA2DArrayAlleleGenome<>
GA3DArrayGenome<>
GA3DArrayAlleleGenome<>
GATreeGenome<>
GAListGenome<>
GAArray<>
GAAlleleSetArray<>
GAAlleleSet<>
GABinaryString
GABin2DecPhenotype
GATree<>
GATreeIter<>
GAList<>
GAListIter<>
Page 20
Programming Interface: Global Typedefs and Enumerations
GAlib Version 2.4, Document Revision B
17
19-Aug-96
Programming Interface
This document describes the programming interface for the library. The section for each class contains a
description of the object's purpose followed by the creator signature and member functions. There are
also sections for library constants, typedefs, and function signatures.
Global Typedefs and Enumerations
typedef float GAProbability, GAProb
typedef enum _GABoolean {gaFalse, gaTrue} GABoolean, GABool
typedef enum _GAStatus {gaSuccess, gaFailure} GAStatus
typedef unsigned char GABit
Global Variables and Global Constants
char* gaErrMsg; // globally defined pointer to current error message
int gaDefScoreFrequency1 = 1; // for non-overlapping populations
int gaDefScoreFrequency2 = 100; // for overlapping populations
float gaDefLinearScalingMultiplier = 1.2;
float gaDefSigmaTruncationMultiplier = 2.0;
float gaDefPowerScalingFactor = 1.0005;
float gaDefSharingCutoff = 1.0;
Function Prototypes
GABoolean (*GAGeneticAlgorithm::Terminator)(GAGeneticAlgorithm&)
GAGenome& (*GAIncrementalGA::ReplacementFunction)(GAGenome&, GAPopulation&)
void (*GAPopulation::Initializer)(GAPopulation &)
void (*GAPopulation::Evaluator)(GAPopulation &)
void (*GAGenome::Initializer)(GAGenome &)
float (*GAGenome::Evaluator)(GAGenome &)
int (*GAGenome::Mutator)(GAGenome &, float)
float (*GAGenome::Comparator)(const GAGenome &, const GAGenome&)
int (*GAGenome::SexualCrossover)(const GAGenome&, const GAGenome&, GAGenome*, GAGenome*)
int (*GAGenome::AsexualCrossover)(const GAGenome&, GAGenome*)
int (*GABinaryEncoder)(float& value, GABit* bits, unsigned int nbits, float min, float
max)
int (*GABinaryDecoder)(float& value, const GABit* bits, unsigned int nbits, float min,
float max)
Page 21
Programming Interface: Parameter Names and Command-Line Options
GAlib Version 2.4, Document Revision B
18
19-Aug-96
Parameter Names and Command-Line Options
Parameters may be specified using the full name strings (for example in parameter files), short name
strings (for example on the command line), or explicit member functions (such as those of the genetic
algorithm objects). All of the #defined names are simply the full names declared as #defined strings;
you can use either the string (e.g. number_of_generations) or the #defined name (e.g.
gaNnGenerations), but if you use the #defined name then the compiler will be able to catch your
spelling mistakes.
When you specify GAlib arguments on the command line, they must be in name-value pairs. You can
use either the long or short name. For example, if my program is called optimizer, the command line
for running the program with a population size of 150, mutation rate of 10%, and score filename of
evolve.txt would be:
optimizer popsize 150 pmut 0.1 sfile evolve.txt
#define name
full name
short name
data type and
default value
gaNminimaxi
minimaxi
mm
int
gaDefMiniMaxi = 1
gaNnGenerations
number_of_generations
ngen
int
gaDefNumGen = 250
gaNpConvergence
convergence_percentage
pconv
float
gaDefPConv = 0.99
gaNnConvergence
generations_to_convergence
nconv
int
gaDefNConv = 20
gaNpCrossover
crossover_probability
pcross
float
gaDefPCross = 0.9
gaNpMutation
mutation_probability
pmut
float
gaDefPMut = 0.01
gaNpopulationSize
population_size
popsize
int
gaDefPopSize = 30
gaNnPopulations
number_of_populations
npop
int
gaDefNPop = 10
gaNpReplacement
replacement_percentage
prepl
float
gaDefPRepl = 0.25
gaNnReplacement
replacement_number
nrepl
int
gaDefNRepl = 5
gaNnBestGenomes
number_of_best
nbest
int
gaDefNumBestGenomes = 1
gaNscoreFrequency
score_frequency
sfreq
int
gaDefScoreFrequency1 = 1
gaNflushFrequency
flush_frequency
ffreq
int
gaDefFlushFrequency = 0
Page 22
Programming Interface: Parameter Names and Command-Line Options
GAlib Version 2.4, Document Revision B
19
19-Aug-96
#define name
full name
short name
data type and
default value
gaNscoreFilename
score_filename
sfile
char*
gaDefScoreFilename =
"generations.dat"
gaNselectScores
select_scores
sscores
int
gaDefSelectScores =
GAStatistics::Maximum
gaNelitism
elitism
el
GABoolean
gaDefElitism = gaTrue
gaNnOffspring
number_of_offspring
noffspr
int
gaDefNumOff = 2
gaNrecordDiversity
record_diversity
recdiv
GABoolean
gaDefDivFlag = gaFalse
gaNpMigration
migration_percentage
pmig
float
gaDefPMig = 0.1
gaNnMigration
migration_number
nmig
int
gaDefNMig = 5
Page 23
Programming Interface: Error Handling
GAlib Version 2.4, Document Revision B
20
19-Aug-96
Error Handling
Exceptions are not used in GAlib version 2.x. However, some GAlib functions return a status value to
indicate whether or not their operation was successful. If a function returns an error status, it posts its
error message on the global GAlib error pointer, a global string called gaErrMsg.
By default, GAlib error messages are sent immediately to the error stream. You can disable the
immediate printing of error messages by passing gaFalse to the ::GAReportErrors function. Passing a
value of gaTrue enables the behavior.
If you would like to redirect the error messages to a different stream, use the ::GASetErrorStream
function to assign a new stream. The default stream is the system standard error stream, cerr.
Here are the error control functions and variables:
extern char gaErrMsg[];
void GAReportErrors(GABoolean flag);
void GASetErrorStream(ostream&);
Page 24
Programming Interface: Random Number Functions
GAlib Version 2.4, Document Revision B
21
19-Aug-96
Random Number Functions
GAlib includes the following functions for generating random numbers:
void GARandomSeed(unsigned s = 0)
int GARandomInt()
int GARandomInt(int low, int high)
double GARandomDouble()
double GARandomDouble(double low, double high)
float GARandomFloat()
float GARandomFloat(float low, float high)
int GARandomBit()
GABoolean GAFlipCoin(float p)
int GAGaussianInt(int stddev)
float GAGaussianFloat(float stddev)
double GAGaussianDouble(double stddev)
double GAUnitGaussian()
If you call it with no argument, the GARandomSeed function uses the current time multiplied by the
process ID (on systems that have PIDs) as the seed for a psuedo-random number generator. On systems
with no process IDs it uses only the time. You can specify your own random seed if you like by passing
a value to this function. Once a seed has been specified, subsequent calls to GARandomSeed with the
same value have no effect. Subsequent calls to GARandomSeed with a different value will re-initialize
the random number generator using the new value.
The functions that take low and high as argument return a random number from low to high, inclusive.
The functions that take no arguments return a value in the interval [0,1]. GAFlipCoin returns a boolean
value based on a biased coin toss. If you give it a value of 1 it will return a 1, if you give it a value of
0.75 it will return a 1 with a 75% chance.
The GARandomBit function is the most efficient way to do unbiased coin tosses. It uses the random bit
generator described in Numerical Recipes in C.
The Gaussian functions return a random number from a Gaussian distribution with deviation that you
specify. The GAUnitGaussian function returns a number from a unit Gaussian distribution with mean 0
and deviation of 1.
GAlib uses a single random number generator for the entire library. You may not change the random
number generator on the fly - it can be changed only when GAlib is compiled. See the config.h and
random.h header files for details. By default, GAlib uses the ran2 generator described in Numerical
Recipes in C.
Page 25
Programming Interface: GAGeneticAlgorithm
GAlib Version 2.4, Document Revision B
22
19-Aug-96
GAGeneticAlgorithm
This is an abstract class that cannot be instantiated. Each genetic algorithm, when instantiated, will have
default operators defined for it. See the documentation for the specific genetic algorithm type for details.
The base genetic algorithm class keeps track of evolution statistics such as number of mutations, number
of crossovers, number of evaluations, best/mean/worst in each generation, and initial/current
population statistics. It also defines the terminator, a member function that specifies the stopping
criterion for the algorithm.
You can maximize or minimize by calling the appropriate member function. If you derive your own
genetic algorithm, remember that users of your algorithm may need either type of optimization.
Statistics can be written to file each generation or periodically by specifying a flush frequency.
Generational scores can be recorded each generation or less frequently by specifying a score frequency.
Parameters such as generations-to-completion, crossover probability and mutation probability can be set
by member functions, command-line, or from file.
The evolve member function first calls initialize then calls the step member function until the done
member function returns gaTrue. It calls the flushScores member as needed when the evolution is
complete. If you evolve the genetic algorithm without using the evolve member function, be sure to call
initialize before stepping through the evolution. You can use the step member function to evolve a
single generation. You should call flushScores when the evolution is finished so that any buffered scores
are flushed.
The names of the individual parameter member functions correspond to the #defined string names. You
may set the parameters on a genetic algorithm one at a time (for example, using the nGenerations
member function), using a parameter list (for example, using the parameters member function with a
GAParameterList), by parsing the command line (for example, using the parameters member function
with argc and argv), by name-value pairs (for example, using the set member function with a
parameter name and value), or by reading a stream or file (for example, using the parameters member
with a filename or stream).
see also: GAParameterList
see also: GAStatistics
see also: Terminators
class hierarchy
class GAGeneticAlgorithm : public GAID
typedefs and constants
GABoolean (*GAGeneticAlgorithm::Terminator)(GAGeneticAlgorithm&)
enum { MINIMIZE = -1, MAXIMIZE = 1 };
member function index
static GAParameterList& registerDefaultParameters(GAParameterList&)
void * userData()
void * userData(void *)
void initialize(unsigned int seed=0)
void evolve(unsigned int seed=0) void step()
GABoolean done()
GAGeneticAlgorithm::Terminator terminator()
GAGeneticAlgorithm::Terminator terminator(GAGeneticAlgorithm::Terminator)
const GAStatistics & statistics() const
float convergence() const
Page 26
Programming Interface: GAGeneticAlgorithm
GAlib Version 2.4, Document Revision B
23
19-Aug-96
int generation() const
void flushScores()
int minimaxi() const
int minimaxi(int)
int minimize()
int maximize()
int nGenerations() const
int nGenerations(unsigned int)
int nConvergence() const
int nConvergence(unsigned int)
float pConvergence() const
float pConvergence(float)
float pMutation() const
float pMutation(float) float pCrossover() const
float pCrossover(float)
GAGenome::SexualCrossover crossover(GAGenome::SexualCrossover func)
GAGenome::SexualCrossover sexual() const
GAGenome::AsexualCrossover crossover(GAGenome::AsexualCrossover func)
GAGenome::AsexualCrossover asexual() const
const GAPopulation & population() const
const GAPopulation & population(const GAPopulation&)
int populationSize() const
int populationSize(unsigned int n)
int nBestGenomes() const
int nBestGenomes(unsigned int n)
GAScalingScheme & scaling() const
GAScalingScheme & scaling(const GAScalingScheme&)
GASelectionScheme & selector() const
GASelectionScheme & selector(const GASelectionScheme& s)
void objectiveFunction(GAGenome::Evaluator)
void objectiveData(const GAEvalData&)
int scoreFrequency() const
int scoreFrequency(unsigned int frequency)
int flushFrequency() const
int flushFrequency(unsigned int frequency)
char* scoreFilename() const
char* scoreFilename(const char *filename)
int selectScores() const
int selectScores(GAStatistics::ScoreID which)
GABoolean recordDiversity() const
GABoolean recordDiversity(GABoolean flag)
const GAParameterList & parameters()
const GAParameterList & parameters(const GAParameterList &)
const GAParameterList & parameters(int& argc, char** argv, GABoolean flag = gaFalse)
const GAParameterList & parameters(const char* filename, GABoolean flag = gaFalse)
const GAParameterList & parameters(istream&, GABoolean flag = gaFalse);
int set(const char* s, int v)
int set(const char* s, unsigned int v)
int set(const char* s, char v)
int set(const char* s, const char* v)
int set(const char* s, const void* v)
int set(const char* s, double v);
int write(const char* filename)
int write(ostream&)
int read(const char* filename)
int read(ostream&)
member function descriptions
convergence
Returns the current convergence. The convergence is defined as the ratio of the Nth previous best-of-
generation score to the current best-of-generation score.
Page 27
Programming Interface: GAGeneticAlgorithm
GAlib Version 2.4, Document Revision B
24
19-Aug-96
crossover
Specify the mating method to use for evolution. This can be changed during the course of an evolution.
This genetic algorithm uses only sexual crossover.
done
Returns gaTrue if the termination criteria have been met, returns gaFalse otherwise. This function
simply calls the completion function that was specified using the terminator member function.
evolve
Initialize the genetic algorithm then evolve it until the termination criteria have been satisfied. This
function first calls initialize then calls the step member function until the done member function returns
gaTrue. It calls the flushScores member as needed when the evolution is complete. You may pass a seed
to evolve if you want to specify your own random seed.
flushFrequency
Use this member function to specify how often the scores should be flushed to disk. A value of 0 means
do not write to disk. A value of 100 means to flush the scores every 100 generations.
flushScores
Force the genetic algorithm to flush its generational data to disk. If you have specified a flushFrequency
of 0 or specified a scoreFilename of nil then calling this function has no effect.
generation
Returns the current generation.
initialize
Initialize the genetic algorithm. If you specify a seed, this function calls GARandomSeed with that
value. If you do not specify a seed, GAlib will choose one for you as described in the random functions
section. It then initializes the population and does the first population evaluation.
nBestGenomes
Specify how many 'best' genomes to record. For example, if you specify 10, the genetic algorithm will
keep the 10 best genomes that it ever encounters. Beware that if you specify a large number here the
algorithm will slow down because it must compare the best of each generation with its current list of
best individuals. The default is 1.
nConvergence
Set/Get the number of generations used for the convergence test.
nGenerations
Set/Get the number of generations.
objectiveData
Set the objective data member on all individuals used by the genetic algorithm. This can be changed
during the course of an evolution.
objectiveFunction
Set the objective function on all individuals used by the genetic algorithm. This can be changed during
the course of an evolution.
Page 28
Programming Interface: GAGeneticAlgorithm
GAlib Version 2.4, Document Revision B
25
19-Aug-96
parameters
Returns a reference to a parameter list containing the current values of the genetic algorithm
parameters.
parameters(GAParameterList&)
Set the parameters for the genetic algorithm. To use this member function you must create a parameter
list (an array of name-value pairs) then pass it to the genetic algorithm.
parameters(int& argc, char** argv, GABoolean flag = gaFalse)
Set the parameters for the genetic algorithm. Use this member function to let the genetic algorithm
parse your command line for arguments that GAlib understands. This method decrements argc and
moves the pointers in argv appropriately to remove from the list the arguments that it understands. If
you pass gaTrue as the third argument then the method will complain about any command-line
arguments that are not recognized by this genetic algorithm.
parameters(char* filename, GABoolean flag = gaFalse)
parameters(istream&, GABoolean flag = gaFalse)
Set the parameters for the genetic algorithm. This version of the parameters member function will parse
the specified file or stream for parameters that the genetic algorithm understands. If you pass gaTrue as
the second argument then the method will complain about any parameters that are not recognized by
this genetic algorithm.
pConvergence
Set/Get the convergence percentage. The convergence is defined as the ratio of the Nth previous best-
of-generation score to the current best-of-generation score. N is defined by the nConvergence member
function.
pCrossover
Set/Get the crossover probability.
pMutation
Set/Get the mutation probability.
population
Set/Get the population. Returns a reference to the current population.
populationSize
Set/Get the population size. This can be changed during the course of an evolution.
recordDiversity
Convenience function for specifying whether or not to calculate diversity. Since diversity calculations
require comparison of each individual with every other, recording this statistic can be expensive. The
default is gaFalse (diversity is not recorded).
registerDefaultParameters
Each genetic algorithm defines this member function to declare the parameters that work with it. Pass a
parameter list to this function and this function will configure the list with the default parameter list and
values for the genetic algorithm class from which you called it. This is a statically defined function, so
invoke it using the class name of the genetic algorithm whose parameters you want to use, for example,
GASimpleGA::registerDefaultParameters(list). The default parameters for the base genetic algorithm
class are:
Page 29
Programming Interface: GAGeneticAlgorithm
GAlib Version 2.4, Document Revision B
26
19-Aug-96
flushFrequency
pConvergence
scoreFilename
minimaxi
pCrossover
scoreFrequency
nBestGenomes
pMutation
selectScores
nGenerations
populationSize
nConvergence
recordDiversity
scaling
Set/Get the scaling scheme. The specified scaling scheme must be derived from the GAScalingScheme
class. This can be changed during the course of an evolution.
scoreFilename
Specify the name of the file to which the scores should be recorded.
scoreFrequency
Specify how often the generational scores should be recorded. The default depends on the type of
genetic algorithm that you are using. You can record mean, max, min, stddev, and diversity for every
n generations.
selector
Set/Get the selection scheme for the genetic algorithm. The selector is used to pick individuals from a
population before mating and mutation occur. This can be changed during the course of an evolution.
selectScores
This function is used to specify which scores should be saved to disk. The argument is the logical OR of
the following values: Mean, Maximum, Minimum, Deviation, Diversity (all defined in the scope of the
GAStatistics object). To record all of the scores, pass GAStatistics::AllScores. When written to file, the
format is as follows:
generation TAB mean TAB max TAB min TAB deviation TAB diversity NEWLINE
set
Set individual parameters for the genetic algorithm. The
first argument should be the full- or short-name of the parameter you wish to set. The second argument
is the value to which you would like to set the parameter.
statistics
Returns a reference to the statistics object in the genetic algorithm. The statistics object maintains
information such as best, worst, mean, and standard deviation, and diversity of each generation as well
as a separate population with the best individuals ever encountered by the genetic algorithm.
step
Evolve the genetic algorithm for one generation.
terminator
Set/Get the termination function. The genetic algorithm is complete when the completion function
returns gaTrue. The function must have the proper signature.
userData
Set/Get the userData member of the genetic algorithm. This member is a generic pointer to any
information that needs to be stored with the genetic algorithm.
Page 30
Programming Interface: GADemeGA
GAlib Version 2.4, Document Revision B
27
19-Aug-96
GADemeGA
(parallel populations with migration)
This genetic algorithm has multiple, independent populations. It creates the populations by cloning the
genome or population that you pass when you create it.
Each population evolves using a steady-state genetic algorithm, but each generation some individuals
migrate from one population to another. The migration algorithm is deterministic stepping-stone; each
population migrates a fixed number of its best individuals to its neighbor. The master population is
updated each generation with best individual from each population.
If you want to experiment with other migration methods, derive a new class from this one and define a
new migration operator. You can change the evolution behavior by defining a new step method in a
derived class.
see also: GAGeneticAlgorithm
class hierarchy
class GADemeGA : public GAGeneticAlgorithm
typedefs and constants
enum { ALL= -1 };
constructors
GADemeGA(const GAGenome&)
GADemeGA(const GAPopulation&)
GADemeGA(const GADemeGA&)
member function index
static GAParameterList& registerDefaultParameters(GAParameterList&);
void migrate()
GADemeGA & operator++()
const GAPopulation& population(unsigned int i) const
const GAPopulation& population(int i, const GAPopulation&)
int populationSize(unsigned int i) const
int populationSize(int i, unsigned int n)
int nReplacement(unsigned int i) const
int nReplacement(int i, unsigned int n)
int nMigration() const
int nMigration(unsigned int i)
int nPopulations() const
int nPopulations(unsigned int i)
const GAStatistics& statistics() const
const GAStatistics& statistics(unsigned int i) const
member function descriptions
nMigration
Specify the number of individuals to migrate each generation. Each population will migrate this many
of its best individuals to the next population (the stepping-stone migration model). The individuals
replace the worst individuals in the receiving population.
nReplacement
Specify a number of individuals to replace each generation. When you specify a number of individuals
to replace, the pReplacement value is set to 0. The first argument specifies which population should be
modified. Use GADemeGA::ALL to apply to all populations.
Page 31
Programming Interface: GADemeGA
GAlib Version 2.4, Document Revision B
28
19-Aug-96
operator++
The increment operator evolves the genetic algorithm's population by one generation by calling the
step member function.
pReplacement
Specify a percentage of the population to replace each generation. When you specify a replacement
percentage, the nReplacement value is set to 0. The first argument specifies which population should be
modified. Use GADemeGA::ALL to apply to all populations.
registerDefaultParameters
This function adds parameters to the specified list that are of interest to this genetic algorithm. The
default parameters for the deme genetic algorithm are the parameters for the base genetic algorithm
class plus the following:
nMigration
nPopulations
Page 32
Programming Interface: GAIncrementalGA
GAlib Version 2.4, Document Revision B
29
19-Aug-96
GAIncrementalGA
(overlapping populations with 1 or 2 children per generation)
This genetic algorithm is similar to those based on the GENITOR model. It uses overlapping
populations, but very little overlap (only one or two individuals get replaced each generation). The
default replacement scheme is WORST. A replacement function is required only if you use CUSTOM or
CROWDING as the replacement scheme. You can do DeJong-style crowding by specifying a distance
function with the CROWDING option. (for best DeJong-style results, derive your own genetic
algorithm)
You can specify the number of children that are generated in each 'generation' by using the nOffspring
member function. Since this genetic algorithm is based on a two-parent crossover model, the number of
offspring must be either 1 or 2. The default is 2.
Use the replacement method to specify which type of replacement the genetic algorithm should use.
The replacement strategy determines how the new children will be inserted into the population. If you
want the new child to replace one of its parents, use the Parent strategy. If you want the child to replace
a random population member, use the Random strategy. If you want the child to replace the worst
population member, use the Worst strategy.
If you specify CUSTOM or CROWDING you must also specify a replacement function with the proper
signature. This function is used to pick which genome will be replaced. The first argument passed to the
replacement function is the individual that is supposed to go into the population. The second argument
is the population into which the individual is supposed to go. The replacement function should return a
reference to the genome that the individual should replace. If no replacement should take place, the
replacement function should return a reference to the individual.
The score frequency for this genetic algorithm defaults to 100 (it records the best-of-generation every
100th generation). The default scaling is Linear, the default selection is RouletteWheel.
see also: GAGeneticAlgorithm
class hierarchy
class GAIncrementalGA : public GAGeneticAlgorithm
typedefs and constants
GAGenome& (*GAIncrementalGA::ReplacementFunction)(GAGenome &, GAPopulation &)
enum ReplacementScheme {
RANDOM = GAPopulation::RANDOM,
BEST = GAPopulation::BEST,
WORST = GAPopulation::WORST,
CUSTOM = -30,
CROWDING = -30,
PARENT = -10
};
constructors
GAIncrementalGA(const GAGenome&)
GAIncrementalGA(const GAPopulation&)
GAIncrementalGA(const GAIncrementalGA&)
member function index
static GAParameterList& registerDefaultParameters(GAParameterList&)
GASteadyStateGA & operator++()
ReplacementScheme replacement()
ReplacementScheme replacement(ReplacementScheme, ReplacementFunction f = NULL)
int nOffspring() const
Page 33
Programming Interface: GAIncrementalGA
GAlib Version 2.4, Document Revision B
30
19-Aug-96
int nOffspring(unsigned int n)
member function descriptions
nOffspring
The incremental genetic algorithm can produce either one or two individuals each generation. Use this
member function to specify how many individuals you would like.
operator++
The increment operator evolves the genetic algorithm's population by one generation by calling the
step member function.
registerDefaultParameters
This function adds to the specified list parameters that are of interest to this genetic algorithm. The
default parameters for the incremental genetic algorithm are the parameters for the base genetic
algorithm class plus the following: nOffspring
replacement
Specify a replacement method. The scheme can be one of:
GAIncrementalGA::RANDOM
GAIncrementalGA::BEST
GAIncrementalGA::CUSTOM
GAIncrementalGA::PARENT
GAIncrementalGA::WORST
GAIncrementalGA::CROWDING
If you specify custom or crowding replacement then you must also specify a function. The replacement
function takes two arguments: the individual to insert and the population into which it will be inserted.
The replacement function should return a reference to the genome that should be replaced. If no
replacement should take place, the replacement function should return a reference to the individual
passed to it.
Page 34
Programming Interface: GASimpleGA
GAlib Version 2.4, Document Revision B
31
19-Aug-96
GASimpleGA
(non-overlapping populations)
This genetic algorithm is the 'simple' genetic algorithm that Goldberg describes in his book. It uses non-
overlapping populations. When you create a simple genetic algorithm, you must specify either an
individual or a population of individuals. The new genetic algorithm will clone the individual(s) that
you specify to make its own population. You can change most of the genetic algorithm behaviors after
creation and during the course of the evolution.
The simple genetic algorithm creates an initial population by cloning the individual or population you
pass when you create it. Each generation the algorithm creates an entirely new population of
individuals by selecting from the previous population then mating to produce the new offspring for the
new population. This process continues until the stopping criteria are met (determined by the
terminator).
Elitism is optional. By default, elitism is on, meaning that the best individual from each generation is
carried over to the next generation. To turn off elitism, pass gaFalse to the elitist member function.
The score frequency for this genetic algorithm defaults to 1 (it records the best-of-generation every
generation). The default scaling is Linear, the default selection is RouletteWheel.
class hierarchy
class GASimpleGA : public GAGeneticAlgorithm
constructors
GASimpleGA(const GAGenome&)
GASimpleGA(const GAPopulation&)
GASimpleGA(const GASimpleGA&)
member function index
static GAParameterList& registerDefaultParameters(GAParameterList&)
GASimpleGA & operator++()
GABoolean elitist() const
GABoolean elitist(GABoolean flag)
member function descriptions
elitist
Set/Get the elitism flag. If you specify gaTrue, the genetic algorithm will copy the best individual from
the previous population into the current population if no individual in the current population is any
better.
operator++
The increment operator evolves the genetic algorithm's population by one generation by calling the
step member function.
registerDefaultParameters
This function adds to the specified list parameters that are of interest to this genetic algorithm. The
default parameters for the simple genetic algorithm are the parameters for the base genetic algorithm
class plus the following: elitism
Page 35
Programming Interface: GASteadyStateGA
GAlib Version 2.4, Document Revision B
32
19-Aug-96
GASteadyStateGA
(overlapping populations)
This genetic algorithm is similar to the algorithms described by DeJong. It uses overlapping populations
with a user-specifiable amount of overlap. The algorithm creates a population of individuals by cloning
the genome or population that you pass when you create it. Each generation the algorithm creates a
temporary population of individuals, adds these to the previous population, then removes the worst
individuals in order to return the population to its original size.
You can select the amount of overlap between generations by specifying the pReplacement parameter.
This is the percentage of the population that should be replaced each generation. Newly generated
offspring are added to the population, then the worst individuals are destroyed (so the new offspring
may or may not make it into the population, depending on whether they are better than the worst in
the population).
If you specify a replacement percentage, then that percentage of the population will be replaced each
generation. Alternatively, you can specify a number of individuals (less than the number in the
population) to replace each generation. You cannot specify both - in a parameter list containing both
parameters, the latter is used.
The score frequency for this genetic algorithm defaults to 100 (it records the best-of-generation every
100th generation). The default scaling is Linear, the default selection is RouletteWheel.
see also: GAGeneticAlgorithm
class hierarchy
class GASteadyStateGA : public GAGeneticAlgorithm
constructors
GASteadyStateGA(const GAGenome&)
GASteadyStateGA(const GAPopulation&)
GASteadyStateGA(const GASteadyStateGA&)
member function index
static GAParameterList& registerDefaultParameters(GAParameterList&)
GASteadyStateGA & operator++()
float pReplacement() const
float pReplacement(float percentage)
int nReplacement() const
int nReplacement(unsigned int)
member function descriptions
nReplacement
Specify a number of individuals to replace each generation. When you specify a number of individuals
to replace, the pReplacement value is set to 0.
operator++
The increment operator evolves the genetic algorithm's population by one generation by calling the
step member function.
pReplacement
Specify a percentage of the population to replace each generation. When you specify a replacement
percentage, the nReplacement value is set to 0.
Page 36
Programming Interface: GASteadyStateGA
GAlib Version 2.4, Document Revision B
33
19-Aug-96
registerDefaultParameters
This function adds to the specified list parameters that are of interest to this genetic algorithm. The
default parameters for the steady-state genetic algorithm are the parameters for the base genetic
algorithm class plus the following:
pReplacement
nReplacement
Page 37
Programming Interface: Terminators
GAlib Version 2.4, Document Revision B
34
19-Aug-96
Terminators
Completion functions are used to determine whether or not a genetic algorithm is finished. The done
member function simply calls the completion function to determine whether or not the genetic
algorithm should continue. The predefined completion functions use generation and convergence to
determine whether or not the genetic algorithm is finished.
The completion function returns gaTrue when the genetic algorithm should finish, and gaFalse when
the genetic algorithm should continue.
In this context, convergence refers to the the similarity of the objective scores, not similarity of
underlying genetic structure. The built-in convergence measures use the best-of-generation scores to
determine whether or not the genetic algorithm has plateaued.
GABoolean GAGeneticAlgorithm::TerminateUponGeneration(GAGeneticAlgorithm &)
GABoolean GAGeneticAlgorithm::TerminateUponConvergence(GAGeneticAlgorithm &)
GABoolean GAGeneticAlgorithm::TerminateUponPopConvergence(GAGeneticAlgorithm &)
TerminateUponGeneration
This function compares the current generation to the specified number of generations. If the current
generation is less than the requested number of generations, it returns gaFalse. Otherwise, it returns
gaTrue.
TerminateUponConvergence
This function compares the current convergence to the specified convergence value. If the current
convergence is less than the requested convergence, it returns gaFalse. Otherwise, it returns gaTrue.
Convergence is a number between 0 and 1. A convergence of 1 means that the nth previous best-of-
generation is equal to the current best-of-generation. When you use convergence as a stopping criterion
you must specify the convergence percentage and you may specify the number of previous generations
against which to compare. The genetic algorithm will always run at least this many generations.
TerminateUponPopConvergence
This function compares the population average to the score of the best individual in the population. If
the population average is within pConvergence of the best individual's score, it returns gaTrue.
Otherwise, it returns gaFalse.
For details about how to write your own termination function, see the customizations section.
Page 38
Programming Interface: Replacement Schemes
GAlib Version 2.4, Document Revision B
35
19-Aug-96
Replacement Schemes
The replacement scheme is used by the incremental genetic algorithm to determine how a new
individual should be inserted into a population. Valid replacement schemes include:
GAIncrementalGA::RANDOM
GAIncrementalGA::CUSTOM
GAIncrementalGA::BEST
GAIncrementalGA::CROWDING
GAIncrementalGA::WORST
GAIncrementalGA::PARENT
In general, replace worst produces the best results. Replace parent is useful for basic speciation, and
custom replacement can be used when you want to do your own type of speciation.
If you specify CUSTOM or CROWDING replacement then you must also specify a replacement function.
The replacement function takes as arguments an individual and the population into which the
individual should be placed. It returns a reference to the genome that the individual should replace. If
the individual should not be inserted into the population, the function should return a reference to the
individual.
Any replacement function must have the following function prototype:
typedef GAGenome& (*GAIncrementalGA::ReplacementFunction)(GAGenome &, GAPopulation &);
The first argument is the genome that will be inserted into the population, the second argument is the
population into which the genome should be inserted. The function should return a reference to the
genome that will be replaced. If no replacement occurs, the function should return a reference to the
original genome.
For details about how to write your own replacement function, see the customizations section.
Page 39
Programming Interface: GAGenome
GAlib Version 2.4, Document Revision B
36
19-Aug-96
GAGenome
The genome is a virtual base class and cannot be instantiated. It defines a number of constants and
function prototypes specific to the genome and its derived classes.
The dimension is used to specify which dimension is being referred to in multi-dimensional genomes.
The clone method specifies whether to clone the entire genome (a new genome with contents identical to
the original will be created) or just the attributes of the genome (a new genome with identical
characteristics will be created). In both cases the caller is responsible for deleting the memory allocated
by the clone member function. The resize constants are used when specifying a resizable genome's
resize behavior.
The genetic operators for genomes are functions that take generic genomes as their arguments. This
makes it possible to define new behaviors for existing genome classes without deriving a new class.
class hierarchy
class GAGenome : public GAID
typedefs and constants
enum GAGenome::Dimension { LENGTH, WIDTH, HEIGHT, DEPTH }
enum GAGenome::CloneMethod { CONTENTS, ATTRIBUTES }
enum { FIXED_SIZE = -1, ANY_SIZE = -10 }
float (*GAGenome::Evaluator)(GAGenome &)
void (*GAGenome::Initializer)(GAGenome &)
int (*GAGenome::Mutator)(GAGenome &, float)
float (*GAGenome::Comparator)(const GAGenome &, const GAGenome&)
int (*GAGenome::SexualCrossover)(const GAGenome&, const GAGenome&, GAGenome*, GAGenome*);
int (*GAGenome::AsexualCrossover)(const GAGenome&, GAGenome*);
member function index
virtual void copy(const GAGenome & c)
virtual GAGenome * clone(CloneMethod flag = CONTENTS)
float score() const
float score(float s)
int nevals()
float evaluate(GABoolean flag = gaFalse) const
GAGenome::Evaluator evaluator() const
GAGenome::Evaluator evaluator(GAGenome::Evaluator func)
void initialize()
GAGenomeInitializer initializer() const
GAGenomeInitializer initializer(GAGenome::Initializer func)
int mutate(float pmutation)
GAGenome::Mutator mutator() const
GAGenome::Mutator mutator(GAGenome::Mutator func)
float compare(const GAGenome& g) const
GAGenome::Comparator comparator() const
GAGenome::Comparator comparator(GAGenome::Comparator c)
GAGenome::SexualCrossover crossover(GAGenome::SexualCrossover f)
GAGenome::SexualCrossover sexual()
GAGenome::AsexualCrossover crossover(GAGenome::AsexualCrossover f)
GAGenome::AsexualCrossover asexual()
GAGeneticAlgorithm * geneticAlgorithm() const
GAGeneticAlgorithm * geneticAlgorithm(GAGeneticAlgorithm &)
void * userData() const
void * userData(void * data)
GAEvalData * evalData() const
GAEvalData * evalData(void * data)
virtual int read(istream &)
virtual int write(ostream &) const
Page 40
Programming Interface: GAGenome
GAlib Version 2.4, Document Revision B
37
19-Aug-96
virtual int equal(const GAGenome &) const
virtual int notequal(const GAGenome &) const
These operators call the corresponding virtual members so that they will work on any properly
derived genome class.
int operator==(const GAGenome&, const GAGenome&)
int operator!=(const GAGenome&, const GAGenome&)
ostream & operator<<(ostream&, const GAGenome&)
istream & operator>>(istream&, GAGenome&)
member function descriptions
clone
This method allocates space for a new genome whereas the copy method uses the space of the genome
to which it belongs.
compare
Compare two genomes. The comparison can be genotype- or phenotype-based. The comparison returns
a value greater than or equal to 0. 0 means the two genomes are identical (no diversity). The exact
meaning of the comparison is up to you.
comparator
Set/Get the comparison method. The comparator must have the correct signature.
copy
The copy member function is called by the base class' operator= and clone members. You can use it to
copy the contents of a genome into an existing genome.
crossover
Each genome class can define its preferred mating method. Use this method to assign the preferred
crossover for a genome instance.
equal
notequal
'equal' and 'notequal' are genome-specific. See the documentation for each genome class for specific
details about what 'equal' means. For example, genomes that have identical contents but different allele
sets may or may not be considered equal. By default, notequal just calls the equal function, but you can
override this in derived classes if you need to optimize the comparison.
evalData
Set/Get the object used to store genome-specific evaluation data. Each genome owns its own evaluation
data object; cloning a genome clones the evaluation data as well.
evaluate
Invoke the genome's evaluation function. If you call this member with gaTrue, the evaluation function
is called no matter what (assuming one has been assigned to the genome). By default, the argument to
this function is gaFalse, so the genome's evaluation function is called only if the state of the genome has
not changed since the last time the evaluator was invoked.
evaluator
Set/Get the function used to evaluate the genome.
Page 41
Programming Interface: GAGenome
GAlib Version 2.4, Document Revision B
38
19-Aug-96
geneticAlgorithm
The member function returns a pointer to the genetic algorithm that 'owns' the genome. If this function
returns nil then the genome has no genetic algorithm owner.
initialize
Calls the initialization function for the genome.
initializer
Set/Get the initialization method. The initializer must have the correct signature.
mutate
Calls the mutation method for the genome. The value is typically the mutation likliehood, but the exact
interpretation of this value is up to the designer of the mutation method.
mutator
Set/Get the mutation method. The mutator must have the correct signature.
nevals
Returns the number of objective function evaluations since the genome was initialized.
operator==
operator!=
operator<<
operator>>
These methods call the associated virtual member functions. They can be used on any generic genome.
If the derived class was properly defined, the appropriate derived functions will be called and the
functions will operate on the derived classes rather than the base class.
read
Fill the genome with the data read from the specified stream. sexual
asexual
Returns a pointer to the preferred mating method for this genome. If this function returns nil, no mating
method has been defined for the genome. The genetic algorithm object has ultimate control over the
mating method that is actually used in the evolution.
score
Returns the objective score of the genome using the objective function assigned to the genome. If no
objective function has been assigned and no score has been set, a score of 0 will be returned. If the score
function is called with an argument, the genome's objective score is set to that value (useful for
population-based objective functions in which the population object does the evaluations).
userData
Each genome contains a generic pointer to user-specifiable data. Use this member function to set/get
that pointer. Notice that cloning a genome will cause the cloned genome to refer to the same user data
pointer as the original; the user data is not cloned as well. So all genomes in a population refer to the
same user data.
write
Send the contents of the genome to the specified stream.
Page 42
Programming Interface: GA1DArrayGenome<T>
GAlib Version 2.4, Document Revision B
39
19-Aug-96
GA1DArrayGenome<T>
The 1D array genome is a generic, resizable array of objects. It is a template class derived from the
GAGenome class as well as the GAArray<> class.
Each element in the array is a gene. The values of the genes are determines by the type of the genome.
For example, an array of ints may have integer values whereas an array of doubles may have floating
point values.
see also: GAArray, GAGenome
class hierarchy
class GA1DArrayGenome<T> : public GAArray<T>, public GAGenome
constructors
GA1DArrayGenome(unsigned int length, GAGenome::Evaluator objective = NULL, void * userData
= NULL)
GA1DArrayGenome(const GA1DArrayGenome<T> &)
member function index
const T & gene(unsigned int x=0) const
T & gene(unsigned int x=0)
T & gene(unsigned int x, const T& value) const
T & operator[](unsigned int x) const
T & operator[](unsigned int x)
int length() const
int length(int l)
int resize(int x)
int resizeBehaviour() const
int resizeBehaviour(unsigned int minx, unsigned int maxx)
void copy(const GA1DArrayGenome<T>& original, unsigned int dest, unsigned int src,
unsigned int length)
void swap(unsigned int x1, unsigned int x2)
member function descriptions
copy
Copy the specified bits from the designated genome.
gene
Set/Get the specified element.
length
Set/Get the length.
resize
Set the length.
resizeBehaviour
Set/Get the resize behavior. The min value specifies the minimum allowable length, the max value
specifies the maximum allowable length. If min and max are equal, the genome is not resizable.
Use the resizeBehaviour and resize member functions to control the size of the genome. The default
behavior is fixed size. Using the resizeBehaviour method you can specify minimum and maximum
values for the size of the genome. If you specify minimum and maximum as the same values then fixed
Page 43
Programming Interface: GA1DArrayGenome<T>
GAlib Version 2.4, Document Revision B
40
19-Aug-96
size is assumed. If you use the resize method to specify a size that is outside the bounds set earlier using
resizeBehaviour, the bounds will be 'stretched' to accommodate the value you specify with resize.
Conversely, if the values you specify with resizeBehaviour conflict with the genome's current size, the
genome will be resized to accommodate the new values.
When resizeBehaviour is called with no arguments, it returns the maximum size if the genome is
resizable, or GAGenome::FIXED_SIZE if the size is fixed.
swap
Swap the specified elements.
genetic operators for this class
GA1DArrayGenome<>::SwapMutator
GA1DArrayGenome<>::ElementComparator
GA1DArrayGenome<>::UniformCrossover
GA1DArrayGenome<>::EvenOddCrossover
GA1DArrayGenome<>::OnePointCrossover
GA1DArrayGenome<>::TwoPointCrossover
GA1DArrayGenome<>::PartialMatchCrossover
GA1DArrayGenome<>::OrderCrossover
GA1DArrayGenome<>::CycleCrossover
default genetic operators for this class
initialization:
GAGenome::NoInitializer
comparison:
GA1DArrayGenome<>::ElementComparator
mutation:
GA1DArrayGenome<>::SwapMutator
crossover:
GA1DArrayGenome<>::OnePointCrossover
Page 44
Programming Interface: GA1DArrayAlleleGenome<T>
GAlib Version 2.4, Document Revision B
41
19-Aug-96
GA1DArrayAlleleGenome<T>
The one-dimensional array allele genome is derived from the one-dimensional array genome class. It
shares the same behaviors, but adds the features of allele sets. The value assumed by each element in
an array allele genome depends upon the allele set specified for that element. In the simplest case, you
can create a single allele set which defines the possible values for any element in the array. More
complicated examples can have a different allele set for each element in the array.
If you create the genome with a single allele set, the genome will have a length that you specify and
the allele set will be used for the mapping of each element. If you create the genome using an array of
allele sets, the genome will have a length equal to the number of allele sets in the array and each
element of the array will be mapped using the corresponding allele set.
When you define an allele set for an array genome, the genome makes its own copy. Subsequent clones
of this genome will refer to the original genome's allele set (allele sets do reference counting).
see also: GAArray, GA1DArrayGenome, GAAlleleSet, GAAlleleSetArray
class hierarchy
class GA1DArrayAlleleGenome<T> : public GAArrayGenome<T>
constructors
GA1DArrayAlleleGenome(unsigned int length, const GAAlleleSet<T>& alleleset,
GAGenome::Evaluator objective = NULL, void * userData = NULL)
GA1DArrayAlleleGenome(const GAAlleleSetArray<T>& allelesets, GAGenome::Evaluator objective
= NULL, void * userData = NULL)
GA1DArrayAlleleGenome(const GA1DArrayAlleleGenome<T>&)
member function index
const GAAlleleSet<T>& alleleset(unsigned int i = 0) const
member function descriptions
alleleset
Returns a reference to the allele set for the specified gene. If the genome was created using a single
allele set, the allele set will be the same for every gene. If the genome was created using an allele set
array, each gene may have a different allele set.
genetic operators for this class
GA1DArrayAlleleGenome<>::UniformInitializer
GA1DArrayAlleleGenome<>::OrderedInitializer
GA1DArrayAlleleGenome<>::FlipMutator
default genetic operators for this class
initialization:
GA1DArrayAlleleGenome<>::UniformInitializer
comparison:
GA1DArrayGenome<>::ElementComparator
mutation:
GA1DArrayAlleleGenome<>::FlipMutator
crossover:
GA1DArrayGenome<>::OnePointCrossover
Page 45
Programming Interface: GA2DArrayGenome<T>
GAlib Version 2.4, Document Revision B
42
19-Aug-96
GA2DArrayGenome<T>
The two-dimensional array genome is a generic, resizable array of objects. It is a template class derived
from the GAGenome class as well as the GAArray<> class.
Each element in the array is a gene. The values of the genes are determines by the type of the genome.
For example, an array of ints may have integer values whereas an array of doubles may have floating
point values.
see also: GAArray, GAGenome
class hierarchy
class GA2DArrayGenome<T> : public GAArray<T>, public GAGenome
constructors
GA2DArrayGenome(unsigned int width, unsigned int height, GAGenome::Evaluator objective =
NULL, void * userData = NULL)
GA2DArrayGenome(const GA2DArrayGenome<T> &)
member function index
const T & gene(unsigned int x, unsigned int y) const
T & gene(unsigned int x, unsigned int y)
T & gene(unsigned int x, unsigned int y, const T& value)
int width() const
int width(int w)
int height() const
int height(int h)
int resize(int x, int y)
int resizeBehaviour(GADimension which) const
int resizeBehaviour(GADimension which, unsigned int min, unsigned int max)
int resizeBehaviour(unsigned int minx, unsigned int maxx, unsigned int miny, unsigned int
maxy)
void copy(const GA2DArrayGenome<T>& original, unsigned int xdest, unsigned int ydest,
unsigned int xsrc, unsigned int ysrc, unsigned int width, unsigned int height)
void swap(unsigned int x1, unsigned int y1, unsigned int x2, unsigned int y2)
member function descriptions
copy
Copy the specified bits from the designated genome.
gene
Set/Get the specified element.
height
Set/Get the height.
resize
Change the size to the specified dimensions.
resizeBehaviour
Set/Get the resize behavior. The min value specifies the minimum allowable length, the max value
specifies the maximum allowable length. If min and max are equal, the genome is not resizable.
Page 46
Programming Interface: GA2DArrayGenome<T>
GAlib Version 2.4, Document Revision B
43
19-Aug-96
Use the resizeBehaviour and resize member functions to control the size of the genome. The default
behavior is fixed size. Using the resizeBehaviour method you can specify minimum and maximum
values for the size of the genome. If you specify minimum and maximum as the same values then fixed
size is assumed. If you use the resize method to specify a size that is outside the bounds set earlier using
resizeBehaviour, the bounds will be 'stretched' to accommodate the value you specify with resize.
Conversely, if the values you specify with resizeBehaviour conflict with the genome's current size, the
genome will be resized to accommodate the new values.
When resizeBehaviour is called with no arguments, it returns the maximum size if the genome is
resizable, or GAGenome::FIXED_SIZE if the size is fixed.
The resizeBehaviour function works similarly to that of the 1D array genome. In this case, however, you
must also specify for which dimension you are setting the resize behavior. When resizeBehaviour is
called with no arguments, it returns the maximum size if the genome is resizable, or gaNoResize if the
size is fixed.
swap
Swap the specified elements.
width
Set/Get the width.
genetic operators for this class
GA2DArrayGenome<>::SwapMutator
GA2DArrayGenome<>::ElementComparator
GA2DArrayGenome<>::UniformCrossover
GA2DArrayGenome<>::EvenOddCrossover
GA2DArrayGenome<>::OnePointCrossover
default genetic operators for this class
initialization:
GAGenome::NoInitializer
comparison:
GA2DArrayGenome<>::ElementComparator
mutation:
GA2DArrayGenome<>::SwapMutator
crossover:
GA2DArrayGenome<>::OnePointCrossover
Page 47
Programming Interface: GA2DArrayAlleleGenome<T>
GAlib Version 2.4, Document Revision B
44
19-Aug-96
GA2DArrayAlleleGenome<T>
The two-dimensional array allele genome is derived from the two-dimensional array genome class. It
shares the same behaviors, but adds the features of allele sets. The value assumed by each element in
an array allele genome depends upon the allele set specified for that element. In the simplest case, you
can create a single allele set which defines the possible values for any element in the array. More
complicated examples can have a different allele set for each element in the array.
The genome will have width and height that you specify and the allele set will be used for the
mapping of each element. When you define an allele set for an array genome, the genome makes its
own copy. Subsequent clones of this genome will refer to the original genome's allele set (allele sets do
reference counting).
If you create a genome using an allele set array, the array of alleles will be mapped to the two
dimensions in the order width-then-height.
see also: GAArray, GA2DArrayGenome, GAAlleleSet, GAAlleleSetArray
class hierarchy
class GA1DArrayAlleleGenome<T> : public GAArrayGenome<T>
constructors
GA2DArrayAlleleGenome(unsigned int width, unsigned int height, GAAlleleSet<T>& alleles,
GAGenome::Evaluator objective = NULL, void * userData = NULL)
GA2DArrayAlleleGenome(unsigned int width, unsigned int height, GAAlleleSetArray<T>&
allelesets, GAGenome::Evaluator objective = NULL, void * userData = NULL)
GA2DArrayAlleleGenome(const GA2DArrayAlleleGenome<T> &)
member function index
const GAAlleleSet<T>& alleleset(unsigned int i = 0, unsigned int j = 0) const
member function descriptions
alleleset
Returns a reference to the allele set for the specified gene. If the genome was created using a single
allele set, the allele set will be the same for every gene. If the genome was created using an allele set
array, each gene may have a different allele set.
genetic operators for this class
GA2DArrayAlleleGenome<>::UniformInitializer
GA2DArrayAlleleGenome<>::FlipMutator
default genetic operators for this class
initialization:
GA2DArrayAlleleGenome<>::UniformInitializer
comparison:
GA2DArrayGenome<>::ElementComparator
mutation:
GA2DArrayAlleleGenome<>::FlipMutator
crossover:
GA2DArrayGenome<>::OnePointCrossover
Page 48
Programming Interface: GA3DArrayGenome<T>
GAlib Version 2.4, Document Revision B
45
19-Aug-96
GA3DArrayGenome<T>
The three-dimensional array genome is a generic, resizable array of objects. It is a template class
derived from the GAGenome class as well as the GAArray<> class.
Each element in the array is a gene. The values of the genes are determines by the type of the genome.
For example, an array of ints may have integer values whereas an array of doubles may have floating
point values.
see also: GAArray, GAGenome
class hierarchy
class GA3DArrayGenome<T> : public GAArray<T>, public GAGenome
constructors
GA3DArrayGenome(unsigned int width, unsigned int height, unsigned int depth,
GAGenome::Evaluator objective = NULL, void * userData = NULL)
GA3DArrayGenome(const GA3DArrayGenome<T>&)
member function index
const T & gene(unsigned int x, unsigned int y, unsigned int z) const
T & gene(unsigned int x, unsigned int y, unsigned int z)
T & gene(unsigned int x, unsigned int y, unsigned int z, const T& value)
int width() const
int width(int w)
int height() const
int height(int h)
int depth() const
int depth(int d)
int resize(int x, int y, int z)
int resizeBehaviour(GADimension which) const
int resizeBehaviour(GADimension which, unsigned int min, unsigned int max)
int resizeBehaviour(unsigned int minx, unsigned int maxx, unsigned int miny, unsigned int
maxy, unsigned int minz, unsigned int maxz)
void copy(const GA3DArrayGenome<T>& original, unsigned int xdest, unsigned int ydest,
unsigned int zdest, unsigned int xsrc, unsigned int ysrc, unsigned int zsrc,
unsigned int width, unsigned int height, unsigned int depth)
void swap(unsigned int x1, unsigned int y1, unsigned int z1, unsigned int x2, unsigned int
y2, unsigned int z2)
member function descriptions
copy
Copy the specified bits from the designated genome.
depth
Set/Get the depth.
gene
Set/Get the specified element.
height
Set/Get the height.
resize
Change the size to the specified dimensions.
Page 49
Programming Interface: GA3DArrayGenome<T>
GAlib Version 2.4, Document Revision B
46
19-Aug-96
resizeBehaviour
Set/Get the resize behavior. The min value specifies the minimum allowable length, the max value
specifies the maximum allowable length. If min and max are equal, the genome is not resizable.
Use the resizeBehaviour and resize member functions to control the size of the genome. The default
behavior is fixed size. Using the resizeBehaviour method you can specify minimum and maximum
values for the size of the genome. If you specify minimum and maximum as the same values then fixed
size is assumed. If you use the resize method to specify a size that is outside the bounds set earlier using
resizeBehaviour, the bounds will be 'stretched' to accommodate the value you specify with resize.
Conversely, if the values you specify with resizeBehaviour conflict with the genome's current size, the
genome will be resized to accommodate the new values.
When resizeBehaviour is called with no arguments, it returns the maximum size if the genome is
resizable, or GAGenome::FIXED_SIZE if the size is fixed.
The resizeBehaviour function works similarly to that of the 1D array genome. In this case, however, you
must also specify for which dimension you are setting the resize behavior. When resizeBehaviour is
called with no arguments, it returns the maximum size if the genome is resizable, or gaNoResize if the
size is fixed.
swap
Swap the specified elements.
width
Set/Get the width.
genetic operators for this class
GA3DArrayGenome<>::SwapMutator
GA3DArrayGenome<>::ElementComparator
GA3DArrayGenome<>::UniformCrossover
GA3DArrayGenome<>::EvenOddCrossover
GA3DArrayGenome<>::OnePointCrossover
default genetic operators
initialization:
GAGenome::NoInitializer
comparison:
GA3DArrayGenome<>::ElementComparator
mutation:
GA3DArrayGenome<>::SwapMutator
crossover:
GA3DArrayGenome<>::OnePointCrossover
Page 50
Programming Interface: GA3DArrayAlleleGenome<T>
GAlib Version 2.4, Document Revision B
47
19-Aug-96
GA3DArrayAlleleGenome<T>
The three-dimensional array allele genome is derived from the three-dimensional array genome class. It
shares the same behaviors, but adds the features of allele sets. The value assumed by each element in
an array allele genome depends upon the allele set specified for that element. In the simplest case, you
can create a single allele set which defines the possible values for any element in the array. More
complicated examples can have a different allele set for each element in the array.
The genome will have width, height, and depth that you specify and the allele set will be used for the
mapping of each element. When you define an allele set for an array genome, the genome makes its
own copy. Subsequent clones of this genome will refer to the original genome's allele set (allele sets do
reference counting).
If you create a genome using an allele set array, the array of alleles will be mapped to the three
dimensions in the order width-then-height-then-depth.
see also: GAArray, GA3DArrayGenome, GAAlleleSet, GAAlleleSetArray
class hierarchy
class GA1DArrayAlleleGenome<T> : public GAArrayGenome<T>
constructors
GA3DArrayAlleleGenome(unsigned int width, unsigned int height, unsigned int depth,
GAAlleleSet<T>& alleles, GAGenome::Evaluator objective = NULL, void * userData =
NULL)
GA3DArrayAlleleGenome(unsigned int width, unsigned int height, unsigned int depth,
GAAlleleSet<T>& alleles, GAGenome::Evaluator objective = NULL, void * userData =
NULL)
GA3DArrayAlleleGenome(const GA3DArrayAlleleGenome<T> &)
member function index
const GAAlleleSet<T>& alleleset(unsigned int i = 0, unsigned int j = 0, unsigned int k =
0) const
member function descriptions
alleleset
Returns a reference to the allele set for the specified gene. If the genome was created using a single
allele set, the allele set will be the same for every gene. If the genome was created using an allele set
array, each gene may have a different allele set.
genetic operators for this class
GA3DArrayAlleleGenome<>::UniformInitializer
GA3DArrayAlleleGenome<>::FlipMutator
default genetic operators for this class
initialization:
GA3DArrayAlleleGenome<>::UniformInitializer
comparison:
GA3DArrayGenome<>::ElementComparator
mutation:
GA3DArrayAlleleGenome<>::FlipMutator
crossover:
GA3DArrayGenome<>::OnePointCrossover
Page 51
Programming Interface: GA1DBinaryStringGenome
GAlib Version 2.4, Document Revision B
48
19-Aug-96
GA1DBinaryStringGenome
The binary string genome is derived from the GABinaryString and GAGenome classes. It is a string of
1s and 0s whose length may be fixed or variable. The genes in this genomes are bits. The alleles for
each bit are 0 and 1.
see also: GABinaryString
see also: GAGenome
class hierarchy
class GA1DBinaryStringGenome : public GABinaryString, public GAGenome
constructors
GA1DBinaryStringGenome(unsigned int x, GAGenome::Evaluator objective = NULL, void *
userData = NULL)
GA1DBinaryStringGenome(const GA1DBinaryStringGenome&)
member function index
short gene(unsigned int x = 0) const
short gene(unsigned int, short value)
int length() const
int length(int l)
int resize(int x)
int resizeBehaviour() const
int resizeBehaviour(unsigned int minx, unsigned int maxx)
void copy(const GA1DBinaryStringGenome &, unsigned int xdest, unsigned int xsrc, unsigned
int length)
void set(unsigned int x, unsigned int length)
void unset(unsigned int x, unsigned int length)
member function descriptions
copy
Copy the specified bits from the designated genome.
gene
Set/Get the specified bit.
length
Set/Get the length of the bit string.
resize
Set the length of the bit string.
resizeBehaviour
Set/Get the resize behavior. The min value specifies the minimum allowable length, the max value
specifies the maximum allowable length. If min and max are equal, the genome is not resizable.
Use the resizeBehaviour and resize member functions to control the size of the genome. The default
behavior is fixed size. Using the resizeBehaviour method you can specify minimum and maximum
values for the size of the genome. If you specify minimum and maximum as the same values then fixed
size is assumed. If you use the resize method to specify a size that is outside the bounds set earlier using
resizeBehaviour, the bounds will be 'stretched' to accommodate the value you specify with resize.
Page 52
Programming Interface: GA1DBinaryStringGenome
GAlib Version 2.4, Document Revision B
49
19-Aug-96
Conversely, if the values you specify with resizeBehaviour conflict with the genome's current size, the
genome will be resized to accommodate the new values.
When resizeBehaviour is called with no arguments, it returns the maximum size if the genome is
resizable, or GAGenome::FIXED_SIZE if the size is fixed.
set
unset
Set/Unset the bits in the specified range. If you specify a range that is not represented by the genome,
the range that you specified will be clipped to fit the genome.
genetic operators for this class
GA1DBinaryStringGenome::UniformInitializer
GA1DBinaryStringGenome::SetInitializer
GA1DBinaryStringGenome::UnsetInitializer
GA1DBinaryStringGenome::FlipMutator
GA1DBinaryStringGenome::BitComparator
GA1DBinaryStringGenome::UniformCrossover
GA1DBinaryStringGenome::EvenOddCrossover
GA1DBinaryStringGenome::OnePointCrossover
GA1DBinaryStringGenome::TwoPointCrossover
default genetic operators for this class
initialization:
GA1DBinaryStringGenome::UniformInitializer
comparison:
GA1DBinaryStringGenome::BitComparator
mutation:
GA1DBinaryStringGenome::FlipMutator
crossover:
GA1DBinaryStringGenome::OnePointCrossover
Page 53
Programming Interface: GA2DBinaryStringGenome
GAlib Version 2.4, Document Revision B
50
19-Aug-96
GA2DBinaryStringGenome
The binary string genome is derived from the GABinaryString and GAGenome classes. It is a matrix of
1s and 0s whose width and height may be fixed or variable. The genes in this genomes are bits. The
alleles for each bit are 0 and 1.
see also: GABinaryString
see also: GAGenome
class hierarchy
class GA2DBinaryStringGenome : public GABinaryString, public GAGenome
constructors
GA2DBinaryStringGenome(unsigned int x, unsigned int y, GAGenome::Evaluator objective =
NULL, void * userData = NULL)
GA2DBinaryStringGenome(const GA2DBinaryStringGenome &)
member function index
short gene(unsigned int x, unsigned int y) const
short gene(unsigned int x, unsigned int y, const short value)
int width() const
int width(int w)
int height() const
int height(int h)
int resize(int x, int y)
int resizeBehaviour(GADimension which) const
int resizeBehaviour(GADimension which, unsigned int min, unsigned int max)
int resizeBehaviour(unsigned int minx, unsigned int maxx, unsigned int miny, unsigned int
maxy)
void copy(const GA2DBinaryStringGenome &, unsigned int xdest, unsigned int ydest, unsigned
int xsrc, unsigned int ysrc, unsigned int width, unsigned int height)
void set(unsigned int, unsigned int, unsigned int, unsigned int)
void unset(unsigned int, unsigned int, unsigned int, unsigned int)
member function descriptions
copy
Copy the specified bits from the designated genome. If you specify a range that is not represented by
the genome, the range that you specified will be clipped to fit the genome.
gene
Set/Get the specified bit.
height
Set/Get the height of the bit string.
resize
Set the size of the genome to the specified dimensions.
resizeBehaviour
Set/Get the resize behavior. The min value specifies the minimum allowable length, the max value
specifies the maximum allowable length. If min and max are equal, the genome is not resizable.
Use the resizeBehaviour and resize member functions to control the size of the genome. The default
behavior is fixed size. Using the resizeBehaviour method you can specify minimum and maximum
Page 54
Programming Interface: GA2DBinaryStringGenome
GAlib Version 2.4, Document Revision B
51
19-Aug-96
values for the size of the genome. If you specify minimum and maximum as the same values then fixed
size is assumed. If you use the resize method to specify a size that is outside the bounds set earlier using
resizeBehaviour, the bounds will be 'stretched' to accommodate the value you specify with resize.
Conversely, if the values you specify with resizeBehaviour conflict with the genome's current size, the
genome will be resized to accommodate the new values.
When resizeBehaviour is called with no arguments, it returns the maximum size if the genome is
resizable, or GAGenome::FIXED_SIZE if the size is fixed.
set
unset
Set/Unset the bits in the specified range. If you specify a range that is not represented by the genome,
the range that you specified will be clipped to fit the genome.
width
Set/Get the width of the bit string.
genetic operators for this class
GA2DBinaryStringGenome::UniformInitializer
GA2DBinaryStringGenome::SetInitializer
GA2DBinaryStringGenome::UnsetInitializer
GA2DBinaryStringGenome::FlipMutator
GA2DBinaryStringGenome::BitComparator
GA2DBinaryStringGenome::UniformCrossover
GA2DBinaryStringGenome::EvenOddCrossover
GA2DBinaryStringGenome::OnePointCrossover
default genetic operators for this class
initialization:
GA2DBinaryStringGenome::UniformInitializer
comparison:
GA2DBinaryStringGenome::BitComparator
mutation:
GA2DBinaryStringGenome::FlipMutator
crossover:
GA2DBinaryStringGenome::OnePointCrossover
Page 55
Programming Interface: GA3DBinaryStringGenome
GAlib Version 2.4, Document Revision B
52
19-Aug-96
GA3DBinaryStringGenome
The binary string genome is derived from the GABinaryString and GAGenome classes. It is a three-
dimensional block of 1s and 0s whose width, height, and depth can be fixed or variable. The genes in
this genomes are bits. The alleles for each bit are 0 and 1.
see also: GABinaryString
see also: GAGenome
class hierarchy
class GA3DBinaryStringGenome : public GABinaryString, public GAGenome
constructors
GA3DBinaryStringGenome(unsigned int x, unsigned int y, unsigned int z, GAGenome::Evaluator
objective = NULL, void * userData = NULL)
GA3DBinaryStringGenome(const GA3DBinaryStringGenome&)
member function index
short gene(unsigned int x, unsigned int y, unsigned int z) const
short gene(unsigned int x, unsigned int y, unsigned int z, short value)
int width() const
int width(int w)
int height() const
int height(int h)
int depth() const
int depth(int d)
int resize(int x, int y, int z)
int resizeBehaviour(GADimension which) const
int resizeBehaviour(GADimension which, unsigned int min, unsigned int max)
int resizeBehaviour(unsigned int minx, unsigned int maxx, unsigned int miny, unsigned int
maxy, unsigned int minz, unsigned int maxz)
void copy(const GA3DBinaryStringGenome &, unsigned int xdest, unsigned int ydest, unsigned
int zdest, unsigned int xsrc, unsigned int ysrc, unsigned int zsrc, unsigned int
width, unsigned int height, unsigned int depth);
void set(unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned
int);
void unset(unsigned int, unsigned int, unsigned int, unsigned int, unsigned int, unsigned
int);
member function descriptions
copy
Copy the specified bits from the designated genome. If you specify a range that is not represented by
the genome, the range that you specified will be clipped to fit the genome.
depth
Set/Get the depth of the bit string.
gene
Set/Get the specified bit.
height
Set/Get the height of the bit string.
resize
Set the size of the genome to the specified dimensions.
Page 56
Programming Interface: GA3DBinaryStringGenome
GAlib Version 2.4, Document Revision B
53
19-Aug-96
resizeBehaviour
Set/Get the resize behavior. The min value specifies the minimum allowable length, the max value
specifies the maximum allowable length. If min and max are equal, the genome is not resizable.
Use the resizeBehaviour and resize member functions to control the size of the genome. The default
behavior is fixed size. Using the resizeBehaviour method you can specify minimum and maximum
values for the size of the genome. If you specify minimum and maximum as the same values then fixed
size is assumed. If you use the resize method to specify a size that is outside the bounds set earlier using
resizeBehaviour, the bounds will be 'stretched' to accommodate the value you specify with resize.
Conversely, if the values you specify with resizeBehaviour conflict with the genome's current size, the
genome will be resized to accommodate the new values.
When resizeBehaviour is called with no arguments, it returns the maximum size if the genome is
resizable, or GAGenome::FIXED_SIZE if the size is fixed.
set
unset
Set/Unset the bits in the specified range. If you specify a range that is not represented by the genome,
the range that you specified will be clipped to fit the genome.
width
Set/Get the width of the bit string.
genetic operators for this class
GA3DBinaryStringGenome::UniformInitializer
GA3DBinaryStringGenome::SetInitializer
GA3DBinaryStringGenome::UnsetInitializer
GA3DBinaryStringGenome::FlipMutator
GA3DBinaryStringGenome::BitComparator
GA3DBinaryStringGenome::UniformCrossover
GA3DBinaryStringGenome::EvenOddCrossover
GA3DBinaryStringGenome::OnePointCrossover
default genetic operators for this class
initialization:
GA3DBinaryStringGenome::UniformInitializer
comparison:
GA3DBinaryStringGenome::BitComparator
mutation:
GA3DBinaryStringGenome::FlipMutator
crossover:
GA3DBinaryStringGenome::OnePointCrossover
Page 57
Programming Interface: GABin2DecGenome
GAlib Version 2.4, Document Revision B
54
19-Aug-96
GABin2DecGenome
This genome is an implementation of the traditional method for converting binary strings to decimal
values. It contains a mechanism for customized encoding of the bit string; binary-to-decimal and one
form of Gray coding are built in to the library. The default is binary-to-decimal mapping (counting in
base 2). To use this genome, you must create a mapping of bits to decimal values by specifying how
many bits will be used to represent what bounded numbers. The binary-to-decimal genome is derived
from the 1DBinaryStringGenome class.
You must create a phenotype before you can instantiate this genome. The phenotype defines how bits
should map into decimal values and vice versa. A single binary-to-decimal phenotype contains the
number of bits used to represent the decimal value and the minimum and maximum decimal values to
which the set of bits will map.
see also: GA1DBinaryStringGenome
see also: GABin2DecPhenotype
see also: GACrossover
class hierarchy
class GABin2DecGenome : public GA1DBinaryStringGenome
constructors
GABin2DecGenome(const GABin2DecPhenotype &, GAGenome::Evaluator objective = NULL, void *
userData = NULL)
GABin2DecGenome(const GABin2DecGenome&)
member function index
const GABin2DecPhenotype& phenotypes(const GABin2DecPhenotype &)
const GABin2DecPhenotype& phenotypes() const
int nPhenotypes() const
float phenotype(unsigned int n) const
float phenotype(unsigned int n, float value)
void encoder(GABinaryEncoder)
void decoder(GABinaryDecoder)
member function descriptions
encoder
decoder
Use these member functions to set the encoder/decoder for the genome. These functions determine
what method will be used for converting the binary bits to decimal numbers. The functions that you
specify here must have the proper signature.
nPhenotype
Returns the number of phenotypes (i.e. the number of decimal values represented) in the genome.
phenotypes
Set/Get the mapping from binary to decimal numbers.
phenotype
Set/Get the specified phenotype.
Page 58
Programming Interface: GABin2DecGenome
GAlib Version 2.4, Document Revision B
55
19-Aug-96
default genetic operators for this class
initialization:
GA1DBinaryStringGenome::UniformInitializer
comparison:
GA1DBinaryStringGenome::BitComparator
mutation:
GA1DBinaryStringGenome::FlipMutator
crossover:
GA1DBinaryStringGenome::OnePointCrossover
de/encoding:
GABinaryEncode/GABinaryDecode
additional information
Conversion functions are defined for transforming strings of bits to decimal values and vice versa. The
function prototypes for the encoding (decimal-to-binary) and decoding (binary-to-decimal) are defined as
follows:
typedef int (*GABinaryEncoder)(float& value, GABit* bits, unsigned int nbits, float min,
float max);
typedef int (*GABinaryDecoder)(float& value, const GABit* bits, unsigned int nbits, float
min, float max);
The library includes the following binary-to-decimal/decimal-to-binary converters:
GABinaryEncode/GABinaryDecode
Convert using a binary coding scheme.
GAGrayEncode/GAGrayDecode
Convert using a Gray coding scheme.
Page 59
Programming Interface: GAListGenome<T>
GAlib Version 2.4, Document Revision B
56
19-Aug-96
GAListGenome<T>
The list genome is a template class. It is derived from the GAGenome class as well as the GAList<>
class. It can be used for order-based representations or variable length sequences as well as traditional
applications of lists.
You must define an initialization operator for this class. The default initializer is NoInitializer - if you do
not assign an initialization operator then you'll get errors about no initializer defined when you try to
initialize the genome.
see also: GAList
see also: GAGenome
class hierarchy
class GAListGenome<T> : public GAList<T>, public GAGenome
constructors
GAListGenome(GAGenome::Evaluator objective = NULL, void * userData = NULL)
GAListGenome(const GAListGenome<T> &)
genetic operators for this class
GAListGenome<>::DestructiveMutator
GAListGenome<>::SwapMutator
GAListGenome<>::OnePointCrossover
GAListGenome<>::PartialMatchCrossover
GAListGenome<>::OrderCrossover
GAListGenome<>::CycleCrossover
default genetic operators for this class
initialization:
GAGenome::NoInitializer
comparison:
GAGenome::NoComparator
mutation:
GAListGenome<>::SwapMutator
crossover:
GAListGenome<>::OnePointCrossover
Page 60
Programming Interface: GARealGenome
GAlib Version 2.4, Document Revision B
57
19-Aug-96
GARealGenome
The real number genome was designed to be used for applications whose representation requires an
array of (possibly bounded) real number parameters. The elements of the array can assume bounded
values, discretized bounded values, or enumerated values, depending on the type of allele set that is
used to create the genome. You can mix the bounding within the genome by specifying an appropriate
array of allele sets. The allele set defines the possible values that each element in the genome may
assume.
The real number genome is a specialization of the array genome with alleles. The specialization is of
type float. You must create an allele set or array of allele sets before you can instantiate this genome. If
you create a real number genome using a single allele set, each element in the genome will use that
allele set to determine its value. If you create a real number genome using an allele set array, the
genome will have a length equal to the number of elements in the array and each element of the real
number will be governed by the allele set corresponding to its location in the genome.
To use the real genome in your code, you must include the real genome header file in each of your files
that uses the real genome. You must also include the real genome source file (it contains template
specialization code) in one (and only one) of your source files. Including the real genome source file will
force the compiler to use the real specializations. If you do not include the real genome source file you
will get the generic array routines instead (and some of the allele methods will not work as expected).
see also: GA1DArrayAlleleGenome, GAAlleleSet, GAAlleleSetArray
class hierarchy
typedef GAAlleleSet<float>
GARealAlleleSet
typedef GAAlleleSetCore<float>
GARealAlleleSetCore
typedef GAAlleleSetArray<float>
GARealAlleleSetArray
typedef GA1DArrayAlleleGenome<float>
GARealGenome
constructors
GARealGenome(unsigned int length, const GARealAlleleSet &, GAGenome::Evaluator objective =
NULL, void * userData = NULL)
GARealGenome(const GARealAlleleSetArray &, GAGenome::Evaluator objective = NULL, void *
userData = NULL)
GARealGenome(const GARealGenome&)
genetic operators for this class
GARealGenome::UniformInitializer
GARealGenome::OrderedInitializer
GARealGenome::FlipMutator
GARealGenome::SwapMutator
GARealGaussianMutator
GARealGenome::UniformCrossover
GARealGenome::EvenOddCrossover
GARealGenome::OnePointCrossover
GARealGenome::TwoPointCrossover
GARealGenome::PartialMatchCrossover
GARealGenome::OrderCrossover
GARealGenome::CycleCrossover
default genetic operators for this class
initialization:
GARealGenome::UniformInitializer
comparison:
GARealGenome::ElementComparator
mutation:
GARealGaussianMutator
crossover:
GARealGenome::UniformCrossover
Page 61
Programming Interface: GAStringGenome
GAlib Version 2.4, Document Revision B
58
19-Aug-96
GAStringGenome
The string genome can be used for order-based applications, variable length string applications, or non-
binary allele set alphabets. The allele set defines the possible values that each element in the string
may assume.
The string genome is a specialization of the array genome with alleles. The specialization is of type
char. You must create an allele set or array of allele sets before you can instantiate this genome.
If you create a string genome using a single allele set, each element in the genome will use that allele
set to determine its value. If you create a string genome using an allele set array, the string will have a
length equal to the number of elements in the array and each element of the string will be governed by
the allele set corresponding to its location in the string.
To use the string genome in your code, you must include the string genome header file in each of your
files that uses the string genome. You must also include the string genome source file (it contains
template specialization code) in one (and only one) of your source files. Including the string genome
source file will force the compiler to use the string specializations. If you do not include the string
genome source file you will get the generic array routines instead (and some of the allele methods will
not work as expected).
see also: GA1DArrayAlleleGenome, GAAlleleSet, GAAlleleSetArray
class hierarchy
typedef GAAlleleSet<char>
GAStringAlleleSet
typedef GAAlleleSetCore<char>
GAStringAlleleSetCore
typedef GAAlleleSetArray<char>
GAStringAlleleSetArray
typedef GA1DArrayAlleleGenome<char>
GAStringGenome
constructors
GAStringGenome(unsigned int length, const GAStringAlleleSet &, GAGenome::Evaluator
objective = NULL, void * userData = NULL)
GAStringGenome(const GAStringAlleleSetArray &, GAGenome::Evaluator objective = NULL, void
* userData = NULL)
GAStringGenome(const GAStringGenome&)
genetic operators for this class
GAStringGenome::UniformInitializer
GAStringGenome::OrderedInitializer
GAStringGenome::FlipMutator
GAStringGenome::SwapMutator
GAStringGenome::UniformCrossover
GAStringGenome::EvenOddCrossover
GAStringGenome::OnePointCrossover
GAStringGenome::TwoPointCrossover
GAStringGenome::PartialMatchCrossover
GAStringGenome::OrderCrossover
GAStringGenome::CycleCrossover
default genetic operators for this class
initialization:
GAStringGenome::UniformInitializer
comparison:
GAStringGenome::ElementComparator
mutation:
GAStringGenome::FlipMutator
crossover:
GAStringGenome::UniformCrossover
Page 62
Programming Interface: GATreeGenome<T>
GAlib Version 2.4, Document Revision B
59
19-Aug-96
GATreeGenome<T>
The tree genome is a template class. It is derived from the GAGenome class as well as the GATree<>
class. The tree genome can be used for direct manipulation of tree objects. It can be used to represent
binary trees as well as non-binary trees.
You must define an initialization operator for this class. The default initializer is NoInitializer - if you do
not assign an initialization operator then you'll get errors about no initializer defined when you try to
initialize the genome.
see also: GATree
see also: GAGenome
class hierarchy
class GATreeGenome<T> : public GATree<T>, public GAGenome
constructors
GATreeGenome(GAGenome::Evaluator objective = NULL, void * userData = NULL)
GATreeGenome(const GATreeGenome<T> &)
genetic operators for this class
GATreeGenome<>::DestructiveMutator
GATreeGenome<>::SwapSubtreeMutator
GATreeGenome<>::SwapNodeMutator
GATreeGenome<>::OnePointCrossover
default genetic operators for this class
initialization:
GAGenome::NoInitializer
comparison:
GAGenome::NoComparator
mutation:
GATreeGenome<>::SwapSubtreeMutator
crossover:
GATreeGenome<>::OnePointCrossover
Page 63
Programming Interface: GAEvalData
GAlib Version 2.4, Document Revision B
60
19-Aug-96
GAEvalData
The evaluation data object is a generic base class for genome- and/or population-specific data. Whereas
the userData member of the genome is shared by all genomes in a population, the evalData member is
unique to each genome. The base class defines the copy/clone interface for the evaluation data object.
Your derived classes should use this mechanism. Any derived class must define a clone and copy
member function. These will be called by the base class when the evaluation data is cloned/copied by
the genomes/populations.
class hierarchy
class GAEvalData : public GAID
constructors
GAEvalData()
GAEvalData(const GAEvalData&)
member functions
GAEvalData* clone() const
void copy(const GAEvalData&)
Page 64
Programming Interface: GABin2DecPhenotype
GAlib Version 2.4, Document Revision B
61
19-Aug-96
GABin2DecPhenotype
The binary-to-decimal phenotype defines the mapping from binary string to decimal values. A
mapping for a single binary-to-decimal conversion consists of a range of decimal values and a number
of bits. For example, a map of 8 bits and range of [0,255] would use 8 bits to represent the numbers
from 0 to 255, inclusive. This object does reference counting in order to minimize the memory overhead
imposed by instantiating binary-to-decimal mappings.
constructors
GABin2DecPhenotype()
GABin2DecPhenotype(const GABin2DecPhenotype&)
member function index
void add(unsigned int nbits, float min, float max)
void remove(unsigned int which)
int size() const
int nPhenotypes() const
float min(unsigned int which) const
float max(unsigned int which) const
int length(unsigned int which) const
int offset(unsigned int which) const
void link(GABin2DecPhenotype&)
void unlink()
member function descriptions
add
Create a mapping that tells the phenotype that nbits should be used to represent a floating point
number from min to max, inclusive.
link
unlink
The phenotype object does reference counting to reduce the number of instantiated objects. Use the link
member to make a phenotype object refer to another. Use the unlink member to remove the connection.
When you unlink, the phenotype makes its own copy of the mapping information.
length
Returns the number of bits required for the specified mapping.
max
min
Returns the maximum/minimum decimal value for the specified mapping.
offset
Returns the offset (in bits) for the specified mapping.
remove
Removes a single binary-to-decimal from the phenotype.
size
Returns the number of bits that the set of mappings requires for converting a decimal value to binary
and back again.
Page 65
Programming Interface: GAAlleleSet<T>
GAlib Version 2.4, Document Revision B
62
19-Aug-96
GAAlleleSet<T>
The allele set class is a container for the different values that a gene may assume. It can contain objects
of any type as long as the object has the =, ==, and != operators defined.
Allele sets may be enumerated, bounded, or bounded with discretization. For example, an integer
allele set may be defined as {1,3,5,2,99,-53} (an enumerated set). A bounded float set may be defined
such as [2,743) (the set of numbers from 2, inclusive, to 743, exclusive). A bounded, discretized set may
defined such as [4.5,7.05](0.05) (the set of numbers from 4.5 to 7.5, inclusive, with increment of 0.05).
If you call the allele member function with no argument, the allele set picks randomly from the alleles it
contains and returns one of them.
constructors
GAAlleleSet()
GAAlleleSet(unsigned int n, const T a[])
GAAlleleSet(const T& lower, const T& upper, GAAllele::BoundType
lowerbound=GAAllele::INCLUSIVE, GAAllele::BoundType upperbound=GAAllele::INCLUSIVE)
GAAlleleSet(const T& lower, const T& upper, const T& increment, GAAllele::BoundType
lowerbound=GAAllele::INCLUSIVE, GAAllele::BoundType upperbound=GAAllele::INCLUSIVE)
GAAlleleSet(const GAAlleleSet<T>& set)
member function index
GAAlleleSet<T> * clone() const
T add(const T& allele)
T remove(T& allele)
T allele() const
T allele(unsigned int i)
int size() const
T lower() const
T upper() const
T inc() const
GAAllele::BoundType lowerBoundType() const
GAAllele::BoundType upperBoundType() const
GAAllele::Type type() const
void link(GAAlleleSet<T>&) void unlink()
member function descriptions
add
remove
Add/Remove the indicated allele from the set. This method works only for enumerated allele sets. Both
functions return zero if the operation was successful, non-zero status otherwise.
lower
upper
Returns the lower/upper bounds on the allele set. If the allele set is enumerated, lower returns the first
element of the set and upper returns the last element of the set.
inc
Returns the increment of the allele set. If the set is not discretized, the first element or lower bounds of
the set is returned.
Page 66
Programming Interface: GAAlleleSet<T>
GAlib Version 2.4, Document Revision B
63
19-Aug-96
lowerBoundType
upperBoundType
Returns GAAllele::INCLUSIVE or GAAllele::EXCLUSIVE to indicate the type of bound on the limits of
the allele set. If no bounds have been defined, these method return GAAllele::NONE.
link
unlink
The alleleset object does reference counting to reduce the number of instantiated objects. Use the link
member to make an alleleset object refer to the data in another. Use the unlink member to remove the
connection. When you unlink, the alleleset makes its own copy of the set data.
size
Returns the number of elements in the allele set. This member is meaningful only for the enumerated
allele set.
type
Returns GAAllele::ENUMERATED, GAAllele::BOUNDED, or GAAllele::DISCRETIZED to indicate the
type of allele set that has been defined. The type of the allele set is specified by the creator used to
instantiate the allele set.
Page 67
Programming Interface: GAAlleleSetArray<T>
GAlib Version 2.4, Document Revision B
64
19-Aug-96
GAAlleleSetArray<T>
The GAAlleleSetArray is a container object with an array of allele sets.
constructors
GAAlleleSetArray()
GAAlleleSetArray(const GAAlleleSet<T>&)
GAAlleleSetArray(const GAAlleleSetArray<T>&)
member function index
int size() const
const GAAlleleSet<T>& set(unsigned int i) const
int add(const GAAlleleSet<T>& s)
int add(unsigned int n, const T a[])
int add(const T& lower, const T& upper, GAAllele::BoundType lb=GAAllele::INCLUSIVE,
GAAllele::BoundType ub=GAAllele::INCLUSIVE)
int add(const T& lower, const T& upper, const T& increment, GAAllele::BoundType
lb=GAAllele::INCLUSIVE, GAAllele::BoundType ub=GAAllele::INCLUSIVE)
int remove(unsigned int)
member function descriptions
add
Use the add members to append an allele set to the end of the array. Each of the overloaded add
members invokes a corresponding allele set creator, so you can use the appropriate add member for
your particular allele set application.
remove
Remove the indicated allele set from the array. Returns zero if successful, non-zero otherwise.
size
Returns the number of allele sets in the array.
Page 68
Programming Interface: GAParameter and GAParameterList
GAlib Version 2.4, Document Revision B
65
19-Aug-96
GAParameter and GAParameterList
The parameter list object contains information about how genetic algorithms should behave. Each
parameter list contains an array of parameters. Each parameter is a name-value pair, where the name is
a string (e.g. "number_of_generations") and the value is an int, float, double, char, string, boolean, or
pointer.
Each parameter is uniquely identified by a pair of names: the full name and the short name. Associated
with the names is a value. Each parameter also has a type from the enumerated list of types shown
above. The GAParameter object automatically does type coercion of the pointer that is passed to it based
upon the type that is passed to it upon its creation. The type cannot be changed once the parameter has
been created.
typedefs and constants
enum GAParameter::Type {BOOLEAN, CHAR, STRING, INT, FLOAT, DOUBLE, POINTER};
constructors
GAParameter(const char* fn, const char* sn, Type tp, const void* v)
GAParameter(const GAParameter& orig)
member function index
void copy(const GAParameter&)
char* fullname() const
char* shrtname() const
const void* value() const
const void* value(const void* v) Type type() const
constructors
GAParameterList()
GAParameterList(const GAParameterList&)
member function index
int size() const
int get(const char*, void*) const
int set(const char*, const void*)
int set(const char* s, int v)
int set(const char* s, unsigned int v)
int set(const char* s, char v)
int set(const char* s, char* v)
int set(const char* s, double v)
int add(const char*, const char*, GAParameter::Type, const void*)
int remove();
GAParameter& operator[](unsigned int i) const
GAParameter& next()
GAParameter& prev()
GAParameter& current() const
GAParameter& first()
GAParameter& last()
GAParameter* operator()(const char* name)
int parse(int& argc, char **argv, GABoolean flag = gaFalse)
int write(const char* filename) const
int write(ostream& os) const
int read(const char* filename)
int read(istream& is)
ostream& operator<<(ostream& os, const GAParameterList& plist)
istream& operator>>(istream& is, GAParameterList& plist)
Page 69
Programming Interface: GAParameter and GAParameterList
GAlib Version 2.4, Document Revision B
66
19-Aug-96
member function descriptions
add
Add a parameter with specified name, type, and default value to the parameter list. This becomes the
current parameter.
current
Return a reference to the current parameter in the list.
first
Return a reference to the first parameter in the list. This becomes the current parameter.
get
Fills the contents of the space pointed to by ptr with the current value of the named parameter. Returns
0 if the parameter was found, non-zero otherwise.
last
Return a reference to the last parameter in the list. This becomes the current parameter.
next
Return a reference to the next parameter in the list. This becomes the current parameter.
parse
Parse an argument list (in command-line format) for recognized name-value pairs. If you pass gaTrue as
the third argument then this method will post warnings about names that it does not recognize.
prev
Return a reference to the next parameter in the list. This becomes the current parameter.
read
Read a parameter list from the specified file or stream. set
Set the named parameter to the specified value. Returns 0 if the paramter was found and successfully
set, non-zero otherwise. You can use either the full or short name to specify a parameter.
size
Returns the number of parameters in the parameter list.
remove
Remove the current parameter from the parameter list.
write
Write the parameter list to the specified file or stream.
Page 70
Programming Interface: GAStatistics
GAlib Version 2.4, Document Revision B
67
19-Aug-96
GAStatistics
The statistics object contains information about the current state of the genetic algorithm objects. Every
genetic algorithm contains a statistics object.
The statistics object defines the following enumerated constants for use by the selectScores member.
They can be bitwise-ORed to specify desired combinations of components. Use the class name to refer to
the values, for example GAStatistics::Mean | GAStatistics::Deviation
typedefs and constants
enum { NoScores, Mean, Maximum, Minimum, Deviation, Diversity, AllScores }
constructors
GAStatistics()
GAStatistics(const GAStatistics&)
member function index
void copy(const GAStatistics &);
float online() const
float offline() const
float initial(ScoreID w=Maximum) const
float current(ScoreID w=Maximum) const
float worstEver() const
float bestEver() const
int generation() const
float convergence() const
int selections() const
int crossovers() const
int mutations() const
int replacements() const
int nConvergence(unsigned int)
int nConvergence() const
int nBestGenomes(const GAGenome&, unsigned int)
int nBestGenomes() const
int scoreFrequency(unsigned int x)
int scoreFrequency() const
int flushFrequency(unsigned int x)
int flushFrequency() const
char* scoreFilename(const char *filename)
char* scoreFilename() const
int selectScores(int whichScores)
int selectScores() const
GABoolean recordDiversity(GABoolean flag)
GABoolean recordDiversity() const
void flushScores()
void update(const GAPopulation& pop)
void reset(const GAPopulation& pop)
const GAPopulation& bestPopulation() const
const GAGenome& bestIndividual(unsigned int n=0) const
int scores(const char* filename, ScoreID which=NoScores)
int scores(ostream& os, ScoreID which=NoScores)
int write(const char* filename) const
int write(ostream& os) const;
ostream& operator<<(ostream&, const GAStatistics&)
Page 71
Programming Interface: GAStatistics
GAlib Version 2.4, Document Revision B
68
19-Aug-96
member function descriptions
bestEver
Returns the score of the best individual ever encountered.
bestIndividual
This function returns a reference to the best individual encountered by the genetic algorithm.
bestPopulation
This function returns a reference to a population containing the best individuals encountered by the
genetic algorithm. The size of this population is specified using the nBestGenomes member function.
convergence
Returns the current convergence. Here convergence means the ratio of the nth previous best-of-
generation to the current best-of-generation.
crossovers
Returns the number of crossovers that have occurred since initialization.
current
Returns the specified score from the current population.
flushFrequency
Set/Get the frequency at which the generational scores should be flushed to disk. A score frequency of
100 means that at every 100th recorded score the scores buffer will be appended to the scores file.
flushScores
Force a flush of the scores buffer to the score file.
generation
Returns the current generation number.
initial
Returns the specified score from the initial population.
mutations
Returns the number of mutations that have occurred since initialization.
nBestGenomes
Set/Get the number of unique best genomes to keep.
nConvergence
Set/Get the number of generations to use for the convergence measure. A value of 10 means the best-of-
generation from 10 generations previous will be used for the convergence test.
offline
Returns the average of the best-of-generation scores.
online
Returns the average of all scores.
Page 72
Programming Interface: GAStatistics
GAlib Version 2.4, Document Revision B
69
19-Aug-96
recordDiversity
This boolean option determines whether or not the diversity of the population will be calculated each
generation. By default, this option is set to false.
replacements
Returns the number of replacements that have occurred since initialization.
reset
Reset the contents of the statistics object using the contents of the specified population.
scoreFilename
Set the name of the file to which the scores should be output. If the filename is set to nil, the scores will
not be written to disk. The default filename is "generations.dat".
scoreFrequency
Set/Get the frequency at which the generational scores should be recorded. A score frequency of 1
means the scores will be recorded each generation. The default depends on the type of genetic
algorithm that is being used.
scores
Print the generational scores to the specified stream. Output is tab-delimited with each line containing
the generation number and the specified scores. You can specify which score you would like by
logically ORing one of the score identifiers listed above. The order of the tab-delimited scores is as
follows:
generation TAB mean TAB max TAB min TAB deviation TAB diversity NEWLINE
selections
Returns the number of selections that have occurred since initialization.
selectScores
This function is used to specify which scores should be saved to disk. The argument is the logical OR of
the following values: Mean, Maximum, Minimum, Deviation, Diversity (all defined in the scope of the
GAStatistics object). To record all of the scores, pass GAStatistics::AllScores.
update
Update the contents of the statistics object to reflect the state of the specified population.
worstEver
Returns the score of the worst individual ever encountered.
Page 73
Programming Interface: GAPopulation
GAlib Version 2.4, Document Revision B
70
19-Aug-96
GAPopulation
The population object is a container for the genomes. It also contains population statistics such as
average, maximum, and minimum genome objective scores. Each population contains a scaling object
that is used to determine the fitness of its genomes. The population also contains a function used for
selecting individuals from the population.
Whenever possible, the population caches the statistics. This means that the first call to one of the
statistics members will be slower than subsequent calls.
You can customize the initialization, evaluation, and sort methods. Use the appropriate member
function. Your customized functions must have the appropriate signature.
The default scaling scheme is linear scaling. The default evaluator invokes the objective function for
each genome. The default selector is roulette wheel and uses the scaled (fitness) scores for its selections.
typedefs and constants
void (*GAPopulation::Initializer)(GAPopulation &)
void (*GAPopulation::Evaluator)(GAPopulation &)
enum SortBasis { RAW, SCALED };
enum SortOrder { LOW_IS_BEST, HIGH_IS_BEST };
enum Replacement { BEST = -1, WORST = -2, RANDOM = -3 };
constructors
GAPopulation()
GAPopulation(const GAGenome&, unsigned int popsize = gaDefPopSize)
GAPopulation(const GAPopulation&)
member function index
GAPopulation * clone() const
void copy(const GAPopulation&)
int size(unsigned int popsize)
int size() const
float sum() const
float ave() const
float var() const
float dev() const
float max() const
float min() const
float div() const
float div(unsigned int i, unsigned int j) const
float fitsum() const
float fitave() const
float fitmax() const
float fitmin() const
float fitvar() const
float fitdev() const
float psum(unsigned int i) const
int nevals() const
void touch()
void statistics(GABoolean flag = gaFalse) const
void diversity(GABoolean flag = gaFalse) const
void prepselect(GABoolean flag = gaFalse) const
GAGenome& select()
GASelectionScheme& selector() const
GASelectionScheme& selector(const GASelectionScheme&)
void scale(GABoolean flag = gaFalse) const
GAScalingScheme& scaling() const
GAScalingScheme& scaling(const GAScalingScheme&)
Page 74
Programming Interface: GAPopulation
GAlib Version 2.4, Document Revision B
71
19-Aug-96
void sort(GABoolean flag = gaFalse, SortBasis basis = RAW) const
SortOrder order() const
SortOrder order(SortOrder flag)
void evaluate(GABoolean flag = gaFalse) const
GAPopulation::Evaluator evaluator(GAPopulation::Evaluator func)
GAPopulation::Evaluator evaluator(GAPopulation::Evaluator func)
void initialize()
GAPopulation::Initializer initializer(GAPopulation::Initializer func)
GAPopulation::Initializer initializer(GAPopulation::Initializer func)
GAGeneticAlgorithm * geneticAlgorithm() const
GAGeneticAlgorithm * geneticAlgorithm(GA&)
void * userData() const
void * userData(void * u)
GAEvalData * evalData() const
GAEvalData * evalData(const GAEvalData&)
GAGenome& individual(unsigned int x, SortBasis basis = RAW) const
GAGenome& best(unsigned int i = 0, SortBasis basis = RAW) const
GAGenome& worst(unsigned int i = 0, SortBasis basis = RAW) const
GAGenome * add(GAGenome *)
GAGenome * add(const GAGenome&)
GAGenome * remove(unsigned int i, SortBasis basis = RAW)
GAGenome * remove(GAGenome *)
GAGenome * replace(GAGenome *, int which = gaPopReplaceRandom, SortBasis basis = RAW)
GAGenome * replace(GAGenome *, GAGenome *)
void destroy(int w = WORST, SortBasis basis = RAW)
virtual void read(istream &)
virtual void write(ostream &) const
ostream& operator<<(ostream &, const GAPopulation &)
istream& operator>>(istream &, GAPopulation &)
member function descriptions
add
Add the specified individual to the population. If you call this method with a reference to a genome,
the population will clone the genome. If you call this method with a pointer to a genome, the population
will use the genome pointed to by the pointer. From then on the population is responsible for deleting
the genome.
ave
Returns the average of the objective scores.
best
Returns a reference to the best individual in the population. Use the SortBasis flag to specify whether
you want the best in terms of raw objective score or scaled (fitness) score.
destroy
Remove the specified individual from the population and free the memory used by that individual. Use
the SortBasis flag to specify whether to use raw objective score or scaled (fitness) score when determining
which genome to destroy.
dev
Returns the standard deviation of the objective scores.
div
Returns the diversity of the population. Diversity is a number between 0 and 1 where 0 indicates that
each individual is completely different than every other individual. If you specify two indices, this
Page 75
Programming Interface: GAPopulation
GAlib Version 2.4, Document Revision B
72
19-Aug-96
member function returns the diversity of the specified individuals (it invokes the comparison function
for those individuals).
evalData
Set/Get the evaluation data for the population. This object is unrelated to any evaluation data objects
used by the genomes in the population.
evaluate
Evaluate the population using the method set by the evaluator function. The default evaluator simply
calls the evaluate member of each genome in the population. If you call this function with gaTrue then
the population performs the evaluation even if it has already cached the evaluation results.
evaluator
Specifies which function to use to evaluate the population. The specified function must have the proper
signature.
fitave
Returns the average of the fitness scores.
fitdev
Returns the standard deviation of the fitness scores.
fitmax
Returns the maximum fitness score.
fitmin
Returns the minimum fitness score.
fitsum
Returns the sum of the fitness scores.
fitvar
Returns the variance of the fitness scores.
geneticAlgorithm
Set/Get the genetic algorithm that 'owns' this population. A return value of nil indicates that the
population is owned by no genetic algorithm.
individual
Returns a reference to the specified individual. Indices for individuals in the population start at 0 and go
to size()-1. the 0th individual is the best individual when the population has been sorted. Use the
SortBasis flag to specify whether you want the ith individual based upon the raw objective score or
scaled (fitness) score.
initialize
Initialize the population using the method set by initializer. The default initializer simply calls the
initialize method of each genome in the population.
Page 76
Programming Interface: GAPopulation
GAlib Version 2.4, Document Revision B
73
19-Aug-96
initializer
Specifies which function to use to initialize the population. The specified function must have the proper
signature.
max
Returns the maximum objective score in the population.
min
Returns the minimum objective score in the population.
order
Set/Get the sort order. A population may be sorted in two ways, highest-score-is-best or lowest-score-is-
best.
prepselect
The function calls the selector's update method. It is typically called by the population before it does a
selection.
psum
Returns the partial sum of the ith fitness score in the array of (sorted) fitness scores.
remove
Remove the specified individual from the population. The genome to be replaced can be specified by
either an index or by pointer. This function returns a pointer to the genome that was removed from the
population. The caller is responsible for the memory used by the returned genome. Use the SortBasis
flag to specify whether to use raw objective score or scaled (fitness) score when determining which
genome to remove.
replace
Replace the specified individual with the first argument. The genome to be replaced can be specified by
either an index or by pointer. This function returns a pointer to the genome that was replaced. If no
genome was replaced or the specified index or pointer is bogus, it returns nil. Use the SortBasis flag to
specify whether to use raw objective score or scaled (fitness) score when determining which genome to
replace.
scale
Scale the raw (objective) scores in the population using the scaling method. If you call this function with
gaTrue then the scaled scores are recalculated even if the population has already cached them.
scaling
Set/Get the scaling method for this population.
select
Returns a reference to a genome from the population using the selection scheme associated with the
population.
selector
Set/Get the selection method for this population.
Page 77
Programming Interface: GAPopulation
GAlib Version 2.4, Document Revision B
74
19-Aug-96
size
Set/Get the number of individuals in the population. If you resize to a larger size, the new individuals
will be initialized but not evaluated. If you resize to a smaller size, the best individuals will be kept.
sort
Sort the individuals in the population. Individuals may be sorted based upon their raw or scaled scores.
statistics
Calculate the population statistics. This method is automatically invoked whenever any of the
population statistics are requested. If you call this function with gaTrue then the statistics are
recalculated even if the population has already cached them.
sum
Returns the sum of the objective scores.
touch
The population object remembers its state so that it does not execute the evaluate or sort methods unless
its state has been changed. If you want to force the population to execute any of its methods the next
time they are invoked, invoke this method.
userData
Set/Get the user data pointer for the population. You can use the user data member to store a pointer to
any object.
var
Returns the variance of the objective scores.
worst
Returns a reference to the worst individual in the population. Use the SortBasis flag to specify whether
you want the worst in terms of raw objective score or scaled (fitness) score.
Page 78
Programming Interface: GAScalingScheme
GAlib Version 2.4, Document Revision B
75
19-Aug-96
GAScalingScheme
The scaling object is embedded in the population object. The base scaling object is not instantiable. This
object keeps track of the fitness scores (not the objective scores) of each individual in the population. The
genomes that it returns are the genomes in the population to which it is linked; it does not make its own
copies.
For details about how to write your own scaling scheme, see the customizations section.
constructors
GAScalingScheme()
GAScalingScheme(const GAScalingScheme& s)
member function index
virtual GAScalingScheme * clone() const
virtual void copy(const GAScalingScheme &)
virtual void evaluate(const GAPopulation & p)
built-in scaling schemes
GAlib contains a number of instantiable scaling objects derived from the base class. Here are the
constructors for these scaling schemes:
GANoScaling()
The fitness scores are identical to the objective scores. No scaling takes place.
GALinearScaling(float c = gaDefLinearScalingMultiplier)
The fitness scores are derived from the objective scores using the linear scaling method described in
Goldberg's book. You can specify the scaling coefficient. Negative objective scores are not allowed with
this method. Objective scores are converted to fitness scores using the relation
f = a • obj + b
where a and b are calculated based upon the objective scores of the individuals in the population as
described in Goldberg's book.
GASigmaTruncationScaling(float c = gaDefSigmaTruncationMultiplier)
Use this scaling method if your objective scores will be negative. It scales based on the variation from
the population average and truncates arbitrarily at 0. The mapping from objective to fitness score for
each individual is given by
f = obj - (obj_ave - c • obj_dev)
GAPowerLawScaling(int k = gaDefPowerScalingFactor)
Power law scaling maps objective scores to fitness scores using an exponential relationship defined as
f = objk
GASharing(GAGenome::Comparator func = 0,
float cutoff = gaDefSharingCutoff, float alpha = 1)
This scaling method is used to do speciation. The fitness score is derived from its objective score by
comparing the individual against the other individuals in the population. If there are other similar
individuals then the fitness is derated. The distance function is used to specify how similar to each other
Page 79
Programming Interface: GAScalingScheme
GAlib Version 2.4, Document Revision B
76
19-Aug-96
two individuals are. A distance function must return a value of 0 or higher, where 0 means that the two
individuals are identical (no diversity). For a given individual,
f =
obj
s(dj)
j=0
n
∑
s(dj ) =
dj <σ→ 1−
dj
σ
 
α
dj ≥σ→ 0
 
dj = distance between current individual and individual j
n = number of individuals in the population
The default sharing object uses the triangular sharing function described in Goldberg's book. You can
specify the cutoff value (sigma in Goldberg's book) using the sigma member function. The curvature of
the sharing function is controlled by the alpha value. When alpha is 1.0 the sharing function is a
straight line (triangular sharing). If you specify a comparator, that function will be used as the distance
function for all comparisons. If you do not specify a comparator, the sharing object will use the default
comparator of each genome.
Notice that the sharing scaling differs depending on whether the objective is to maximized or
minimized. If the goal is to maximize the objective score, the raw scores will be divided by the sharing
factor. If the goal is to minimize the objective score, the raw scores will be multiplied by the sharing
factor. You can explicitly tell the sharing object to do minimize- or maximize-based scaling by using the
minimaxi member function. By default, it uses the min/max settings of the genetic algorithm that is
using it (based on information in the population with which the sharing object is associated). If the
scaling object is associated with a population that has been created independently of any genetic
algorithm object, the sharing object will use the population's order to decide whether to multiply or
divide to do its scaling.
Page 80
Programming Interface: GASelectionScheme
GAlib Version 2.4, Document Revision B
77
19-Aug-96
GASelectionScheme
Selection schemes are used to pick genomes from a population for mating. The GASelectionScheme
object defines the basic selector behavior. It is an abstract class and cannot be instantiated. Each selector
object may be linked to a population from which it will make its selections. The select member returns a
reference to a single genome. A selector may operate on the scaled objective scores or the raw objective
scores. Default behavior is to operate on the scaled (fitness) scores.
For details about how to write your own selection scheme, see the customizations section.
typedefs and constants
enum { RAW, SCALED };
constructors
GASelectionScheme(int which = FITNESS)
GASelectionScheme(const GASelectionScheme&)
member function index
virtual GASelectionScheme* clone() const;
virtual void copy(const GASelectionScheme& orig)
virtual void assign(GAPopulation& pop)
virtual void update()
virtual GAGenome& select() const;
built-in selection schemes
GAlib contains a number of instantiable scaling objects derived from the base class. Here are the
constructors for these scaling schemes:
GARankSelector(int w=GASelectionScheme::SCALED)
The rank selector picks the best member of the population every time.
GARouletteWheelSelector(int w=GASelectionScheme::SCALED)
This selection method picks an individual based on the magnitude of the fitness score relative to the rest
of the population. The higher the score, the more likely an individual will be selected. Any individual
has a probability p of being chosen where p is equal to the fitness of the individual divided by the sum
of the fitnesses of each individual in the population.
GATournamentSelector(int w=GASelectionScheme::SCALED)
The tournament selector uses the roulette wheel method to select two individuals then picks the one
with the higher score. The tournament selector typically chooses higher valued individuals more often
than the RouletteWheelSelector.
GADSSelector(int w=GASelectionScheme::SCALED)
The deterministic sampling selector (DS) uses a two-staged selection procedure. In the first stage, each
individual's expected representation is calculated. A temporary population is filled using the
individuals with the highest expected numbers. Any remaining positions are filled by first sorting the
original individuals according to the decimal part of their expected representation, then selecting those
highest in the list. The second stage of selection is uniform random selection from the temporary
population.
Page 81
Programming Interface: GASelectionScheme
GAlib Version 2.4, Document Revision B
78
19-Aug-96
GASRSSelector(int w=GASelectionScheme::SCALED)
The stochastic remainder sampling selector (SRS) uses a two-staged selection procedure. In the first
stage, each individual's expected representation is calculated. A temporary population is filled using the
individuals with the highest expected numbers. Any fractional expected representations are used to
give the individual more likeliehood of filling a space. For example, an individual with e of 1.4 will
have 1 position then a 40% chance of a second position. The second stage of selection is uniform random
selection from the temporary population.
GAUniformSelector(int w=GASelectionScheme::SCALED)
The stochastic uniform sampling selector picks randomly from the population. Any individual in the
population has a probability p of being chosen where p is equal to 1 divided by the population size.
Page 82
Programming Interface: GAArray<T>
GAlib Version 2.4, Document Revision B
79
19-Aug-96
GAArray<T>
The GAArray<T> object is defined for your convenience so that you do not have to create your own
array object. It is a template-ized container class whose elements can contain objects of any type. The 1-,
2-, and 3-dimensional arrays used in GAlib are all based upon this single-dimensional array object. This
object is defined in the file arraytmpl.h.
The squares are elements in the array. Arrays are 1
dimensional, but derived classes can have 2 or more
dimensions. Each element contains a user-specified object.
Any object in the array must have the following methods defined and publicly available:
copy constructor
operator =
operator ==
operator !=
The elements in an array are indexed starting with 0 (the first element in the array is element number
0). The last element in array with n elements is element n-1.
constructors
GAArray(unsigned int)
GAArray(const GAArray<T>&)
member function index
GAArray<T> & operator=(const GAArray<T>& orig)
GAArray<T> & operator=(const T array [])
GAArray<T> * clone()
const T & operator[](unsigned int i)
const T & operator[](unsigned int i)
void copy(const GAArray<T>& orig)
void copy(const GAArray<T>& orig, unsigned int dest, unsigned int src, unsigned int
length)
void move(unsigned int dest, unsigned int src, unsigned int length)
void swap(unsigned int i, unsigned int j)
int size() const
int size(unsigned int n)
int equal(const GAArray<T>& b, unsigned int dest, unsigned int src, unsigned int length)
const
int operator==(const GAArray<T>& a, const GAArray<T>& b)
int operator!=(const GAArray<T>& a, const GAArray<T>& b)
member function descriptions
clone
Return a pointer to an exact duplicate of the original array. The caller is responsible for the memory
allocated by the call to this function.
copy
Duplicate the specified array or part of the specified array. If duplicating a part of the specified array,
length elements starting at position src in the original are copied into position dest in the copy. If there
is not enough space in the copy, the extra elements are not copied.
equal
Return 1 if the specified portion of the two arrays is identical, return 0 otherwise.
Page 83
Programming Interface: GAArray<T>
GAlib Version 2.4, Document Revision B
80
19-Aug-96
move
Move the number of elements specified with length from position src to position dest.
operator[]
Return a reference to the contents of the ith element of the array.
size
Return the number of elements in the array.
swap
Swap the contents of element i with the contents of element j.
Page 84
Programming Interface: GABinaryString
GAlib Version 2.4, Document Revision B
81
19-Aug-96
GABinaryString
The binary string object is a simple implementation of a string of bits. Each bit is represented by a
single word of memory (no fancy bit-munging happens here). The binary string class defines the
following member functions. Binary strings are resizable.
constructors
GABinaryString(unsigned int length)
GABinaryString(const GABinaryString&)
member function index
void copy(const GABinaryString&)
int resize(unsigned int)
int size() const
short bit(unsigned int a) const
short bit(unsigned int a, short val)
int equal(const GABinaryString& b, unsigned int dest, unsigned int src, unsigned int
length) const
void copy(const GABinaryString& orig, unsigned int dest, unsigned int src, unsigned int
length)
void move(unsigned int dest, unsigned int src, unsigned int length)
void set(unsigned int a, unsigned int length)
void unset(unsigned int a, unsigned int length)
void randomize(unsigned int a, unsigned int length)
member function descriptions
copy
Makes an exact copy of the specified string. If invoked with a range of bits then copies the specified
range of bits.
bit
Set/Get the specified bit.
equal
Returns 1 if the specified range of bits are equal, 0 otherwise.
move
Move length bits starting at src to dest.
set/unset
Set/Unset length bits starting at a
size
resize
Set/Get the length of the bit string.
randomize
Set to random values length bits starting at a
Page 85
Programming Interface: GAList<T> and GAListIter<T>
GAlib Version 2.4, Document Revision B
82
19-Aug-96
GAList<T> and GAListIter<T>
The GAList<T> object is defined for your convenience so that you do not have to create your own list
object. It is a template-ized container class whose nodes can contain objects of any type. The GAList<T>
object is circular and doubly-linked. A list iterator object is also defined to be used when moving around
the list to keep track of the current, next, previous, or whichever node. Iterators do not change the state
of the list.
The circles are nodes in the list. Each node contains
a user-specified object; the initialization method
determines the size of the list and the contents of
each node. The list is circular and doubly linked.
The template-ized GAList<T> is derived from a generic list base class called GAListBASE. The template
list is defined in listtmpl.h, the list base class is defined in listbase.h
Any object used in the nodes must have the following methods defined and publicly available:
copy constructor
operator =
operator ==
operator !=
Each list object contains an iterator. The list's traversal member functions (next, prev, etc) simply call the
member functions on the internal iterator. You can also instantiate iterators external to the list object so
that you can traverse the list without modifying its state.
The list base class defines constants for specifying where insertions should take place (these are relative
to the node to which the iterator is currently pointing).
Nodes in the list are numbered from 0 to 1 less than the list size. The head node is node 0.
When you do an insertion, the list makes a copy of the specified object (allocating space for it in the
process). The internal iterator is left pointing to the node which was just inserted. The insertion function
uses the copy constructor member to do this, so the objects in your list must have a copy constructor
defined. The new node is inserted relative to the current location of the list's internal iterator. Use the
where flag to determine whether the new node will be inserted before or after the current node, or if
the new node should become the head node of the list.
The remove member returns a pointer to the object that was in the specified node. You are responsible
for deallocating the memory for this object! The destroy member deallocates the memory used by the
object in the current node. In both cases the iterator is left pointing to the node previous to the one that
was deleted.
All of the list traversal functions (prev, next, current, etc) return a pointer to the contents of the node on
which they are operating. You should test the pointer to see if it is NULL before you dereference it.
When you call any of the traversal functions, the list's internal iterator is left pointing to the node to
which traversal function moved. You can create additional iterators (external to the list) to keep track of
multiple positions in the list.
typedefs and constants
GAListBASE::HEAD
GAListBASE::TAIL
GAListBASE::BEFORE
GAListBASE::AFTER
constructors
GAListIter(const GAList<T> &)
Page 86
Programming Interface: GAList<T> and GAListIter<T>
GAlib Version 2.4, Document Revision B
83
19-Aug-96
member function index
T * current()
T * head()
T * tail()
T * next()
T * prev()
T * warp(const GAList<T>& t)
T * warp(const GAListIter<T>& i)
T * warp(unsigned int i)
constructors
GAList()
GAList(const T& t)
GAList(const GAList<T>& orig)
member function index
GAList<T> * clone()
void copy(const GAList<T>& orig)
void destroy()
void swap(unsigned int, unsigned int)
T * remove()
void insert(GAList<T> * t, GAListBASE::Location where=AFTER)
void insert(const T& t, GAListBASE::Location where=AFTER)
T * current()
T * head()
T * tail()
T * next()
T * prev()
T * warp(unsigned int i)
T * warp(const GAListIter<T>& i)
T * operator[](unsigned int i)
int size() const
member function descriptions
These functions change the state of the list.
clone
Return a pointer to an exact duplicate of the original list. The caller is responsible for the memory
allocated by the call to this function.
copy
Duplicate the specified list.
destroy
Destroy the current node in the list. This function uses the location of the internal iterator to determine
which node should be destroyed. If the head node is destroyed, the next node in the list becomes the
head node.
insert
Add a node or list to the list. The insertion is made relative to the location of the internal iterator. The
where flag specifies whether the insertion should be made before or after the current node.
remove
Returns a pointer to the contents of the current node and removes the current node from the list. The
iterator moves to the previous node. The caller is responsible for the memory used by the contents.
Page 87
Programming Interface: GAList<T> and GAListIter<T>
GAlib Version 2.4, Document Revision B
84
19-Aug-96
swap
Swap the positions of the two specified nodes. The internal iterator is not affected. If the iterator was
pointing to one of the nodes before the swap it will still point to that node after the swap, even if that
node was swapped.
These functions do not change the contents of the list, but they change the state of the list's
internal iterator (when invoked on a list object).
current
Returns a pointer to the contents of the current node.
head
Returns a pointer to the contents of the first node in the list.
next
Returns a pointer to the contents of the next node.
operator[]
Returns a pointer to the contents of the ith node in the list (same as warp).
prev
Returns a pointer to the contents of the previous node.
tail
Returns a pointer to the contents of the last node in the list.
warp
Returns a pointer to the contents of the ith node in the list, or a pointer to the element in the list pointed
to by the specified iterator. The head node is number 0.
Page 88
Programming Interface: GATree<T> and GATreeIter<T>
GAlib Version 2.4, Document Revision B
85
19-Aug-96
GATree<T> and GATreeIter<T>
The GATree<T> object is defined for your convenience so that you do not have to create your own tree
object. It is a template-ized container class whose nodes can contain objects of any type. Each level in the
GATree<T> object is a circular and doubly-linked list. The eldest child of a level is the head of the
linked list, each child in a level points to its parent, and the parent of those children points to the eldest
child. Any tree can have only one root node. Any node can have any number of children. A tree
iterator is also defined to be used when moving around the list to keep track of the current, next,
parent, or whichever node. Iterators do not change the state of the tree.
The circles are nodes in the tree. Each node contains a user-
specified object; the initialization method determines the tree
topology and the contents of each node. Each tree contains one
(and only one) root node. Each level in the tree is a circular,
doubly linked list. The head of each list is called the 'eldest'
child, each node in a level has a link to its parent, and each
parent has a link to the eldest of its children (if it has any
children).
The template-ized GATree<T> is derived from a generic tree base class called GATreeBASE. The
template tree is defined in treetmpl.h, the tree base class is defined in treebase.h
Any object used in the nodes have the following methods defined and publicly available:
copy constructor
operator =
operator ==
operator !=
Each tree object contains an iterator. The tree's traversal member functions (next, prev, etc) simply call
the member functions on the internal iterator. You can also instantiate iterators external to the tree object
so that you can traverse the tree without modifying its contents.
The tree base class defines constants for specifying where insertions should occur.
Nodes in a tree are numbered starting at 0 then increasing in a depth-first traversal of the tree. The root
node is node 0. A tree can have only one root node, but any node in the tree can have any number of
children.
When you do an insertion, the tree makes a copy of the specified object (allocating space for it in the
process). The internal iterator is left pointing to the node which was just inserted. The insertion function
uses the copy constructor member to do this, so the objects in your tree must have a copy constructor
defined. The new node is inserted relative to the current location of the tree's internal iterator. Use the
where flag to determine whether the new node will be inserted before, after, or below the current node,
or if the new node should become the root node of the tree.
The remove member returns a pointer to a tree. The root node of this tree is the node at which the
iterator was pointing. You are responsible for deallocating the memory for this tree! The destroy
member deallocates the memory used by the object in the current node and completely destroys any
subtree hanging on that node. In both cases, the iterator is left pointing to the elder child or parent of
the node that was removed/destroyed.
All of the tree traversal functions (prev, next, current, etc) return a pointer to the contents of the node on
which they are operating. You should test the pointer to see if it is NULL before you dereference it.
Also, the iterator is left pointing to the node to which you traverse with each traversal function. You can
create additional iterators (external to the tree) to keep track of multiple positions in the tree.
Page 89
Programming Interface: GATree<T> and GATreeIter<T>
GAlib Version 2.4, Document Revision B
86
19-Aug-96
typedefs and constants
GATreeBASE::ROOT
GATreeBASE::BEFORE
GATreeBASE::AFTER
GATreeBASE::BELOW
constructors
GATreeIter(const GATree<T>& t)
member function index
T * current()
T * root()
T * next()
T * prev()
T * parent()
T * child()
T * eldest()
T * youngest()
T * warp(const GATree<T>& t)
T * warp(const GATreeIter<T>& i)
T * warp(unsigned int i)
int size()
int depth()
int nchildren()
int nsiblings()
constructors
GATree()
GATree(const T& t)
GATree(const GATree<T>& orig)
member function index
GATree<T> * clone()
void copy(const GATree<T>& orig)
void destroy()
void swaptree(GATree<T> * t)
void swaptree(unsigned int, unsigned int)
void swap(unsigned int, unsigned int)
GATree<T> * remove()
void insert(GATree<T> * t, GATreeBASE::Location where=BELOW)
void insert(const T& t, GATreeBASE::Location where=BELOW)
T * current()
T * root()
T * next()
T * prev()
T * parent()
T * child()
T * eldest()
T * youngest()
T * warp(unsigned int i)
T * warp(const GATreeIter<T>& i)
int ancestral(unsigned int i, unsigned int j) const
int size()
int depth()
int nchildren()
int nsiblings()
member function descriptions
These functions change the state of the tree.
Page 90
Programming Interface: GATree<T> and GATreeIter<T>
GAlib Version 2.4, Document Revision B
87
19-Aug-96
clone
Return a pointer to an exact duplicate of the original tree. The caller is responsible for the memory
allocated by the call to this function.
copy
Duplicate the specified tree.
destroy
Destroy the current node in the tree. If the node has children, the entire sub-tree connected to the node
is destroyed as well. This function uses the location of the internal iterator to determine which node
should be destroyed. If the root node is destroyed, the entire contents of the tree will be destroyed, but
the tree object itself will not be deleted.
insert
Add a node or tree to the tree. The insertion is made relative to the location of the internal iterator. The
where flag specifies whether the insertion should be made before, after, or below the current node.
remove
Returns a pointer to a new tree object whose root node is the (formerly) current node of the original tree.
Any subtree connected to the node stays with the node. The iterator moves to the previous node in the
current generation, or the parent node if no elder sibling exists. The caller is responsible for the
memory used by the new tree.
swap
Swap the contents of the two specified nodes. Sub-trees connected to either node are not affected; only
the specified nodes are swapped.
swaptree
Swap the contents of the two specified nodes as well as any sub-trees connected to the specified nodes.
These functions do not change the contents of the tree, but they change the state of the tree's
internal iterator (when invoked on a tree object).
ancestral
Returns 1 if one of the two specified nodes is the ancestor of the other, returns 0 otherwise.
child
Returns a pointer to the contents of the eldest child of the current node. If the current node has no
children, this function returns NULL.
current
Returns a pointer to the contents of the current node.
depth
Returns the number of generations (the depth) of the tree. When called as the member function of a tree
iterator, this function returns the depth of the subtree connected to the iterator's current node.
eldest
Returns a pointer to the contents of the eldest node in the current generation. The eldest node is the
node pointed to by the 'child' function in the node's parent.
Page 91
Programming Interface: GATree<T> and GATreeIter<T>
GAlib Version 2.4, Document Revision B
88
19-Aug-96
nchildren
Returns the number of children of the node to which the iterator is pointing.
next
Returns a pointer to the contents of the next node in the current generation.
nsiblings
Returns the number of nodes in the level of the tree as the node to which the iterator is pointing.
parent
Returns a pointer to the contents of the parent of the current node. If the current node is the root node,
this function returns NULL.
prev
Returns a pointer to the contents of the previous node in the current generation.
root
Returns a pointer to the contents of the root node of the tree.
size
Returns the number of nodes in the tree. When called as the member function of a tree iterator, this
function returns the size of the subtree connected to the iterator's current node.
warp
Returns a pointer to the contents of the ith node in the tree, or a pointer to the element in the tree
pointed to by the specified iterator. The head node is number 0 then the count increases as a depth-first
traversal of the tree.
youngest
Returns a pointer to the contents of the youngest node in the current generation.
Page 92
Customizing GAlib: Deriving your own genome class
GAlib Version 2.4, Document Revision B
89
19-Aug-96
Customizing GAlib
This document describes how to extend GAlib's capabilities by defining your own genomes and genetic
operators. The best way to customize the behavior of an object is to derive a new class. If you do not
want to do that much work, GAlib is designed to let you replace behaviors of existing objects by
defining new functions.
Deriving your own genome class
You can create your own genome class by multiply-inheriting from the base genome class and your
own data type. For example, if you have already have an object defined, say MyObject, then you would
derive a new genome class called MyGenome, whose class definition looks like this:
// Class definition for the new genome object, including statically
// defined declarations for default evaluation, initialization,
// mutation, and comparison methods for this genome class.
class MyGenome : public MyObject, public GAGenome {
public:
GADefineIdentity("MyGenome", 201);
static void Init(GAGenome&);
static int Mutate(GAGenome&, float);
static float Compare(const GAGenome&, const GAGenome&);
static float Evaluate(GAGenome&);
static int Cross(const GAGenome&, const GAGenome&, GAGenome*, GAGenome*);
public:
MyGenome() : GAGenome(Init, Mutate, Compare) {
evaluator(Evaluate);
crossover(Cross);
}
MyGenome(const MyGenome& orig) { copy(orig); }
virtual ~MyGenome() {}
MyGenome& operator=(const GAGenome& orig){
if(&orig != this) copy(orig);
return *this;
}
virtual GAGenome* clone(CloneMethod) const
{return new MyGenome(*this);}
virtual void copy(const GAGenome& orig) {
GAGenome::copy(orig); // this copies all of the base genome parts
// copy any parts of MyObject here
// copy any parts of MyGenome here
}
// any data/member functions specific to this new class
};
void
MyGenome::Init(GAGenome&){
// your initializer here
}
int
MyGenome::Mutate(GAGenome&, float){
// your mutator here
}
float
MyGenome::Compare(const GAGenome&, const GAGenome&){
Page 93
Customizing GAlib: Deriving your own genome class
GAlib Version 2.4, Document Revision B
90
19-Aug-96
// your comparison here
}
float
MyGenome::Evaluate(GAGenome&){
// your evaluation here
}
int
MyGenome::Cross(const GAGenome& mom, const GAGenome& dad, GAGenome* sis, GAGenome* bro){
// your crossover here
}
By convention, one of the arguments to a derived genome constructor is the objective function.
Alternatively (as illustrated in this example), you can hard code a default objective function into your
genome - just call the evaluator member somewhere in your constructor and pass the function you want
used as the default.
Once you have defined your genome class, you should define the intialization, mutation, comparison,
and crossover operators for it. The comparison operator is optional, but if you do not define it you will
not be able to use the diversity measures in the genetic algorithms and/or populations.
Note that the genetic comparator is not necessarily the same as the boolean operator== and operator!=
comparators. The genetic comparator returns 0 if the two individuals are the same, -1 if the comparison
fails for some reason, and a real number greater than 0 indicating the degree of difference if the
individuals are not identical but can be compared. It may be based on genotype or phenotype. The
boolean comparators, on the other hand, indicate only whether or not two individuals are identical. In
most cases, the boolean comparator can simply call the genetic comparator, but in some cases it is more
efficient to define different operators (the boolean comparators are called much more often than the
genetic comparators, especially if no diversity is being measured).
To work properly with the GAlib, you must define the following:
MyGenome( -default-args-for-your-genome-constructor )
MyGenome(const MyGenome&)
virtual GAGenome* clone(GAGenome::CloneMethod) const
If your genome adds any non-trivial member data, you must define these:
virtual ~MyGenome()
virtual copy(const GAGenome&)
virtual int equal(const GAGenome&) const
To enable streams-based reading and writing of your genome, you should define these:
virtual int read(istream&)
virtual int write(ostream&) const
When you derive a genome, don't forget to use the _evaluated flag to indicate when the state of the
genome has changed and an evaluation is needed. If a member function changes the state of your
genome, that member function should set the _evaluated flag to gaFalse. GAlib uses the _evaluated
flag to control its caching of the genome scores, so setting and unsetting the flag is critical. If your
member functions do not unset the flag when they modify the contents of the genome, then the genome
will not be (re)evaluated.
Assign a default crossover, mutation, initialization, and comparison method so that users don't have to
assign one unless they want to.
It is a good idea to define an identity for your genome (especially if you will be using it in an
environment with multiple genome types running around). Use the DefineIdentity macro (defined in
id.h) to do this in your class definition. The DefineIdentity macro sets a class ID number and the name
Page 94
Customizing GAlib: Genome Initialization
GAlib Version 2.4, Document Revision B
91
19-Aug-96
that will be used in error messages for the class. You can use any number above 200 for the ID, but be
sure to use a different number for each of your classes.
When run-time type information (RTTI) has stabilized across compilers, GAlib will probably use that
instead of the Define/Declare identity macros.
Genome Initialization
The initializer takes a single argument: the genome to be initialized. The genome has already been
allocated; the intializer only needs to populate it with appropriate contents.
Here is the implementation of an initializer for the GATreeGenome<int> class.
void
TreeInitializer(GAGenome & c) {
GATreeGenome<int> &child=(GATreeGenome<int> &)c;
// destroy any pre-existing tree
child.root();
child.destroy();
// Create a new tree with depth of 'depth' and each eldest node
// containing 'n' children (the other siblings have none).
int depth=2, n=3, count=0;
child.insert(count++,GATreeBASE::ROOT);
for(int i=0; i<depth; i++){ child.eldest();
child.insert(count++);
for(int j=0; j<n; j++) child.insert(count++,GATreeBASE::AFTER);
}
}
Genome Mutation
The genome mutator takes two arguments: the genome that will receive the mutation(s) and a mutation
probability. The exact meaning of the mutation probability is up to the designer of the mutation
operator. The mutator should return the number of mutations that occured.
Most genetic algorithms invoke the mutation method on each newly generated offspring. So your
mutation operator should base its actions on the value of the mutation probability. For example, an
array of floats could flip a pmut-biased coin for each element in the array. If the coin toss returns true,
the element gets a Gaussian mutation. If it returns false, the element is left unchanged. Alternatively, a
single biased coin toss could be used to determine whether or not the entire genome should be mutated.
Here is an implementation of the flip mutator for the GA1DBinaryString class. This mutator flips a
biased coin for each bit in the string.
int
GA1DBinStrFlipMutator(GAGenome & c, float pmut) {
GA1DBinaryStringGenome &child=(GA1DBinaryStringGenome &)c;
if(pmut <= 0.0) return(0);
int nMut=0;
for(int i=child.length()-1; i>=0; i--){
if(GAFlipCoin(pmut)){
child.gene(i, ((child.gene(i) == 0) ? 1 : 0));
nMut++;
}
}
return nMut;
}
Page 95
Customizing GAlib: Genome Crossover
GAlib Version 2.4, Document Revision B
92
19-Aug-96
Genome Crossover
The crossover method is used by the genetic algorithm to mate individuals from the population to form
new offspring. Each genome should define a default crossover method for the genetic algorithms to use.
The sexual and asexual member functions return a pointer to the preferred sexual and asexual mating
methods, respectively. The crossover member function is used to change the preferred mating method.
The genome does not have a member function to invoke the crossover; only the genetic algorithm can
actually perform the crossover.
Some genetic algorithms use sexual mating, others use asexual mating. If possible, define both so that
your genome will work with either kind of genetic algorithm. If your derived class does not define a
cross method, an error message will be posted whenever crossover is attempted.
Sexual crossover takes four arguments: two parents and two children. If one child is nil, the operator
should be able to generate a single child. The genomes have already been allocated, so the crossover
operator should simply modify the contents of the child genome as appropriate. The crossover function
should return the number of crossovers that occurred. Your crossover function should be able to operate
on one or two children, so be sure to test the child pointers to see if the genetic algorithm is asking you
to create one or two children.
Here is an implementation of the two-parent/one-or-two-child single point crossover operator for fixed-
length genomes of the GA1DBinaryStringGenome class.
int
SinglePointCrossover(const GAGenome& p1, const GAGenome& p2, GAGenome* c1, GAGenome* c2){
GA1DBinaryStringGenome &mom=(GA1DBinaryStringGenome &)p1;
GA1DBinaryStringGenome &dad=(GA1DBinaryStringGenome &)p2;
int n=0;
unsigned int site = GARandomInt(0, mom.length());
unsigned int len = mom.length() - site;
if(c1){
GA1DBinaryStringGenome &sis=(GA1DBinaryStringGenome &)*c1;
sis.copy(mom, 0, 0, site);
sis.copy(dad, site, site, len);
n++;
}
if(c2){
GA1DBinaryStringGenome &bro=(GA1DBinaryStringGenome &)*c2;
bro.copy(dad, 0, 0, site);
bro.copy(mom, site, site, len);
n++;
}
return n;
}
Genome Comparison
The comparison method is used for diversity calculations. It compares two genomes and returns a
number that is greater than or equal to zero. A value of 0 means that the two genomes are identical (no
diversity). There is no maximum value for the return value from the comparator. A value of -1 indicates
that the diversity could not be calculated.
Here is the comparator for the binary string genomes. It simply counts up the number of bits that both
genomes share. In this example, we return a -1 if the genomes are not the same length.
float
GA1DBinStrComparator(const GAGenome& a, const GAGenome& b){
GA1DBinaryStringGenome
&sis=(GA1DBinaryStringGenome &)a;
GA1DBinaryStringGenome
&bro=(GA1DBinaryStringGenome &)b;
if(sis.length() != bro.length()) return -1;
float count = 0.0;
Page 96
Customizing GAlib: Genome Evaluation
GAlib Version 2.4, Document Revision B
93
19-Aug-96
for(int i=sis.length()-1; i>=0; i--)
count += ((sis.gene(i) == bro.gene(i)) ? 0 : 1);
return count/sis.length();
}
Genome Evaluation
The genome evaluator is the objective function for your problem. It takes a single genome as its
argument. The evaluator returns a number that indicates how good or bad the genome is. You must
cast the generic genome to the genome type that you are using. If your objective function works with
different genome types, then use the genome object's className and/or classID member functions to
determine the genome class before you do the casts.
Here is a simple evaluation function for a real number genome with a single element. The function tries
to maximize a sinusoidal.
float
Objective(GAGenome& g){
GARealGenome& genome = (GARealGenome &)g;
return 1 + sin(genome.gene(0)*2*M_PI);
}
Population Initialization
This method is invoked when the population is initialized.
Here is an implemenation that invokes the initializer for each genome in the population.
void
PopInitializer(GAPopulation & p){
for(int i=0; i<p.size(); i++)
p.individual(i).initialize();
}
Population Evaluation
This method is invoked when the population is evaluated. If your objective is population-based, you can
use this method to set the score for each genome rather than invoking an evaluator for each genome.
Here is an implementation that invokes the evaluation method for each genome in the population.
void
PopEvaluator(GAPopulation & p){
for(int i=0; i<p.size(); i++)
p.individual(i).evaluate();
}
Scaling Scheme
The scaling object does the transformation from raw (objective) scores to scaled (fitness) scores. The most
important member function you will have to define for a new scaling object is the evaluate member
function. This function calculates the fitness scores based on the objective scores in the population that is
passed to it.
The GAScalingScheme class is a pure virtual (abstract) class and cannot be instantiated. To make your
derived class non-virtual, you must define the clone and evaluate functions. You should also define the
copy method if your derived class introduces any additional data members that require non-trivial
copy.
Page 97
Customizing GAlib: Selection Scheme
GAlib Version 2.4, Document Revision B
94
19-Aug-96
The scaling class is polymorphic, so you should define the object's identity using the GADefineIdentity
macro. This macro sets a class ID number and the name that will be used in error messages for the class.
You can use any number above 200 for the ID, but be sure to use a different number for each of your
objects.
Here is an implementation of sigma truncation scaling.
class SigmaTruncationScaling : public GAScalingScheme {
public:
GADefineIdentity("SigmaTruncationScaling", 286);
SigmaTruncationScaling(float m=gaDefSigmaTruncationMultiplier) :
c(m) {}
SigmaTruncationScaling(const SigmaTruncationScaling & arg)
{copy(arg);}
SigmaTruncationScaling & operator=(const GAScalingScheme & arg)
{ copy(arg); return *this; }
virtual ~SigmaTruncationScaling() {}
virtual GAScalingScheme * clone() const
{ return new SigmaTruncationScaling(*this); }
virtual void evaluate(const GAPopulation & p);
virtual void copy(const GAScalingScheme & arg){
if(&arg != this && sameClass(arg)){
GAScalingScheme::copy(arg);
c=((SigmaTruncationScaling&)arg).c;
}
}
float multiplier(float fm) { return c=fm; }
float multiplier() const { return c; }
protected:
float c;
// std deviation multiplier
};
void
SigmaTruncationScaling::evaluate(const GAPopulation & p) {
float f;
for(int i=0; i<p.size(); i++){
f = p.individual(i).score() - p.ave() + c * p.dev();
if(f < 0) f = 0;
p.individual(i).fitness(f);
}
}
Selection Scheme
The selection object is used to pick individuals from the population. Before a
selection occurs, the
update method is called. You can use this method to do any pre-selection data transformations for your
selection scheme. When a selection is requested, the select method is called. The select method should
return a reference to a single individual from the population.
A selector may make its selections based either on the scaled (fitness) scores or on the raw (objective)
scores of the individuals in the population. Note also that a population may be sorted either low-to-high
or high-to-low, depending on which sort order was chosen. Your selector should be able to handle either
order (this way it will work with genetic algorithms that maximize or minimize).
The selection scheme class is polymorphic, so you should define the object's identity using the
GADefineIdentity macro. This macro sets a class ID number and the name that will be used in error
messages for the class. You can use any number above 200 for the ID, but be sure to use a different
number for each of your objects.
Page 98
Customizing GAlib: Selection Scheme
GAlib Version 2.4, Document Revision B
95
19-Aug-96
Here is an implementation of a tournament selector. It is based on the roulette wheel selector and shares
some of the roulette wheel selector's functionality. In particular, this tournament selector uses the
roulette wheel selector's update method, so it does not define its own. The select method does two
fitness-proportionate selections then returns the individual with better score.
class TournamentSelector : public GARouletteWheelSelector {
public:
GADefineIdentity("TournamentSelector", 255);
TournamentSelector(int w=GASelectionScheme::FITNESS) :
GARouletteWheelSelector(w) {}
TournamentSelector(const TournamentSelector& orig) { copy(orig); }
TournamentSelector& operator=(const GASelectionScheme& orig)
{ if(&orig != this) copy(orig); return *this; }
virtual ~TournamentSelector() {}
virtual GASelectionScheme* clone() const
{ return new TournamentSelector; }
virtual GAGenome& select() const;
};
GAGenome &
TournamentSelector::select() const {
int picked=0;
float cutoff;
int i, upper, lower;
cutoff = GARandomFloat();
lower = 0; upper = pop->size()-1;
while(upper >= lower){
i = lower + (upper-lower)/2;
if(psum[i] > cutoff)
upper = i-1;
else
lower = i+1;
}
lower = Min(pop->size()-1, lower);
lower = Max(0, lower);
picked = lower;
cutoff = GARandomFloat();
lower = 0; upper = pop->size()-1;
while(upper >= lower){
i = lower + (upper-lower)/2;
if(psum[i] > cutoff)
upper = i-1;
else
lower = i+1;
}
lower = Min(pop->size()-1, lower);
lower = Max(0, lower);
GAPopulation::SortBasis basis =
(which == FITNESS ? GAPopulation::SCALED : GAPopulation::RAW);
if(pop->order() == GAPopulation::LOW_IS_BEST){
if(pop->individual(lower,basis).score() <
pop->individual(picked,basis).score())
picked = lower;
}
else{
if(pop->individual(lower,basis).score() >
pop->individual(picked,basis).score())
picked = lower;
}
return pop->individual(picked,basis);
}
Page 99
Customizing GAlib: Genetic Algorithm
GAlib Version 2.4, Document Revision B
96
19-Aug-96
Genetic Algorithm
Here is a sample derived class that does restricted mating. In this example, one of the parents is selected
as usual. The second individual is select as the first, but it is used only if it is similar to the first
individual. If not, we make another selection. If enough selections fail, we take what we can get.
class RestrictedMatingGA : public GASteadyStateGA {
public:
GADefineIdentity("RestrictedMatingGA", 288);
RestrictedMatingGA(const GAGenome& g) : GASteadyStateGA(g) {}
virtual ~RestrictedMatingGA() {}
virtual void step();
RestrictedMatingGA & operator++() { step(); return *this; }
};
void
RestrictedMatingGA::step() {
int i, k;
for(i=0; i<tmpPop->size(); i++){
mom = &(pop->select());
k=0;
do {
k++; dad = &(pop->select());
} while(mom->compare(*dad) < THRESHOLD && k<pop->size());
stats.numsel += 2;
if(GAFlipCoin(pCrossover()))
stats.numcro += (*scross)(*mom, *dad, &tmpPop->individual(i), 0);
else
tmpPop->individual(i).copy(*mom);
stats.nummut += tmpPop->individual(i).mutate(pMutation());
}
for(i=0; i<tmpPop->size(); i++)
pop->add(tmpPop->individual(i));
pop->evaluate();
// get info about current pop for next
time pop->scale();
// remind the population to do its scaling
for(i=0; i<tmpPop->size(); I++)
pop->destroy(GAPopulation::WORST, GAPopulation::SCALED);
stats.update(*pop);
// update the statistics by one generation
}
Termination Function
The termination function determines when the genetic algorithm should stop evolving. It takes a
genetic algorithm as its argument and returns gaTrue if the genetic algorithm should stop or gaFalse if
the algorithm should continue.
Here are three examples of termination functions. The first compares the current generation to the
desired number of generations. If the current generation is less than the desired number of generations,
it returns gaFalse to signify that the GA is not yet complete.
GABoolean
GATerminateUponGeneration(GAGeneticAlgorithm & ga){
return(ga.generation() < ga.nGenerations() ? gaFalse : gaTrue);
}
The second example compares the average score in the current population with the score of the best
individual in the current population. If the ratio of these exceeds a specified threshhold, it returns
gaTrue to signify that the GA should stop. Basically this means that the entire population has converged
to a 'good' score.
// stop when pop average is 95% of best
const float desiredRatio = 0.95;
Page 100
Customizing GAlib: Termination Function
GAlib Version 2.4, Document Revision B
97
19-Aug-96
GABoolean
GATerminateUponScoreConvergence(GAGeneticAlgorithm & ga){
if(ga.statistics().current(GAStatistics::Mean) /
ga.statistics().current(GAStatistics::Maximum) > desiredRatio)
return gaTrue;
else
return gaFalse;
}
The third uses the population diversity as the criterion for stopping. If the diversity drops below a
specified threshhold, the genetic algorithm will stop.
// stop when population diversity is below this
const float thresh = 0.01;
GABoolean
StopWhenNoDiversity(GAGeneticAlgorithm & ga){
if(ga.statistics().current(GAStatistics::Diversity) < thresh)
return gaTrue;
else
return gaFalse;
}
A faster method of doing a nearly equivalent termination is to use the population's standard deviation
as the stopping criterion (this method does not require comparisons of each individual). Notice that this
judges diversity based upon the genome scores rather than their actual genetic diversity.
// stop when population deviation is below this
const float thresh = 0.01;
GABoolean
StopWhenNoDeviation(GAGeneticAlgorithm & ga){
if(ga.statistics().current(GAStatistics::Deviation) < thresh)
return gaTrue;
else
return gaFalse;
}
Page 101
Descriptions of the Examples
GAlib Version 2.4, Document Revision B
98
19-Aug-96
Descriptions of the Examples
Each of the examples contains comments in the source files with complete description about what is
going on. Here is a short summary of what each one of the examples does:
ex1
Fill a 2DBinaryStringGenome with alternating 0s and 1s using a SimpleGA.
ex2
Generate a sequence of random numbers, then use a Bin2DecChromosome and SimpleGA to try and
match the sequence. This example shows how to use the user-data member of genomes in objective
functions.
ex3
Read a 2D pattern from a data file then try to match the pattern using a 2DBinaryStringGenome and a
SimpleGA. This example also shows how to use the GAParametes object for setting genetic algorithm
parameters and reading command-line arguments.
ex4
Fill a 3DBinaryStringChromosome with alternating 0s and 1s using a SteadyStateGA. This example uses
many member functions of the genetic algorithm to control which statistics are recorded and dumped to
file.
ex5
This example shows how to build a composite genome (a cell?) using a 2DBinaryStringGenome and a
Bin2DecGenome. The composite genome uses behaviors that are defined in each of the genomes that it
contains. The objective is to match a pattern and sequence of numbers.
ex6
Grow a GATreeGenome using a SteadyStateGA. This example illustrates the use of specialized methods
to override the default initialization method and to specialize the output from a tree. It also shows how
to use templatized genome classes. Finally, it shows the use of the parameters object to set default
values then allow these to be modified from the command line. The objective function in this example
tries to grow the tree as large as possible.
ex7
Identical in function to example 3, this example shows how to use the increment operator (++),
completion measure, and other member functions of the GA. It uses a GA with overlapping populations
rather than the non-overlapping GA in example 3 and illustrates the use of many of the GA member
functions. It also illustrates the use of the parameter list for reading settings from a file, and shows how
to stuff a genome with data from an input stream.
ex8
Grow a GAListGenome using a GA with overlapping populations. This shows how to randomly
initialize a list of integers, how to use the sigma truncation scaling object to handle objective scores that
may be positive or negative, and the 'set' member of the genetic algorithm for controlling statistics and
other genetic algorithm parameters.
Page 102
Descriptions of the Examples
GAlib Version 2.4, Document Revision B
99
19-Aug-96
ex9
Find the maximum value of a continuous function in two variables. This example uses a
GABin2DecGenome and simple GA. It also illustrates how to use the GASigmaTruncationScaling object
(rather than the default linear scaling). Sigma truncation is particularly useful for objective functions that
return negative values.
ex10
Find the maximum value of a continuous, periodic function. This example illustrates the use of sharing
to do speciation. It defines a sample distance function (one that does the distance measure based on the
genotype, the other based on phenotype). It uses a binary- to-decimal genome to represent the function
values.
ex11
Generate a sequence of descending numbers using an order-based list. This example illustrates the use
of a GAListGenome as an order-based chromosome. It contains a custom initializer and shows how to
use this custom initializer in the List genome.
ex12
Alphabetize a sequence of characters. Similar to example 11, this example illustrates the use of the
GAStringGenome (rather than a list) as an order-based chromosome.
ex13
This program runs a GA-within-GA. The outer level GA tries to match the pattern read in from a file.
The inner GA tries to match a sequence of randomly generated numbers (the sequence is generated at
the beginning of the program's execution). The inner level GA is run only when the outer GA reaches a
threshhold objective score.
ex14
Another illustration of how to use composite chromosomes. In this example, the composite chromosome
contains a user-specifiable number of lists. Each list behaves differently and is not affected by mutations,
crossovers, or initializations of the other lists.
ex15
The completion function of a GA determines when it is "done". This example uses the convergence to
tell when the GA has reached the optimum (the default completion measure is number-of-generations).
It uses a binary-to-decimal genome and tries to match a sequence of randomly generated numbers.
ex16
Tree chromosomes can contain any kind of object in the nodes. This example shows how to put a point
object into the nodes of a tree to represent a 3D plant. The objective function tries to maximize the size of
the plant.
ex17
Array chromsomes can be used when you need tri-valued alleles. This example uses a 2D array with
trinary alleles.
ex18
This example compares the performance of three different genetic algorithms. The genome and
objective function are those used in example 3, but this example lets you specify which type of GA you
want to use to solve the problem. You can use steady state, simple, or incremental just by specifying
Page 103
Descriptions of the Examples
GAlib Version 2.4, Document Revision B
100
19-Aug-96
one of them on the command line. The example saves the generational data to file so that you can then
plot the convergence data to see how the performance of each genetic algorithm compares to the others.
ex19
The 5 DeJong test problems.
ex20
Holland's royal road function. This example computes Holland's 1993 ICGA version of the Royal Road
problem. Holland posed this problem as a challenge to test the performance of genetic algorithms and
challenged other GA users to match or beat his performance.
ex21
This example illustrates various uses of the allele set in array genomes. The allele set may be an
enumerated list of items or a bounded range of continuous values, or a bounded set of discrete values.
This example shows how each of these may be used in combination with a real number genome.
ex22
This example shows how to derive a new genetic algorithm class in order to customize the replacement
method. Here we derive a new type of steady-state genetic algorithm in which speciation is done more
effectively by not only scaling fitness values but also by controlling the way new individuals are
inserted into the population.
ex23
The genetic algorithm object can either maximize or minimize your objective function. This example
shows how to use the minimize abilities of the genetic algorithm. It uses a real number genome with
one element to find the maximum or minimum of a sinusoid.
ex24
This example shows how to restricted mating using a custom genetic algorithm and custom selection
scheme. The restricted mating in the genetic algorithm tries to pick individuals that are similar (based
upon their comparator). The selector chooses only the upper half of the population (so it cannot choose
very bad individuals, unlike the roulette wheel selector, for example).
ex25
Multiple populations on a single CPU. This example uses the genetic algorithm class called a
'DemeGA'. The genetic algorithm controls the migration behavior for moving individuals between
populations. In this example, the island model is used with a stepping-stone migration behavior in
which the best individuals from each population migrate to their nearest neighboring population. You
can easily modify both the migration algorithm and the population behaviors by deriving a new class
from the DemeGA.
ex26
Travelling Salesperson Problem. Although genetic algorithms are not the best way to solve the TSP, we
include an example of how it can be done. This example uses an order-based list as the genome to
figure out the shortest path that connects a bunch of towns such that each town is visited exactly once. It
uses the edge recombination crossover operator (you can try it with the partial match crossover as well
to see how poorly PMX does on this particular problem).
ex27
Deterministic crowding. Although the algorithms built-in to GAlib allow you to do quite a bit of
customization, sometimes you'll want to derive your own class so that you can really tweak the way the
Page 104
Descriptions of the Examples
GAlib Version 2.4, Document Revision B
101
19-Aug-96
algorithm works. This example shows one way of implementing the deterministic crowding method by
deriving an entirely new genetic algorithm class.
graphic1
You can learn a great deal by watching the genetic algorithm evolve. This example has a simple X
windows interface that lets you start, stop, restart, and incrementally evolve a population of indivdiuals.
You can see the evolution in action, so it becomes very obvious if your operators are not working
correctly or if the algorithm is converging prematurely.
The directory contains two different examples. In the first you can choose between 3 different genetic
algorithms, 2 different genomes (real or binary-to-decimal), and 4 different functions. In the second you
can watch a population of routes evolve for the travelling salesman problem. Both programs use X
resources as well as command-line arguments to control their behavior. You can also use a standard
GAlib settings file. These programs will compile using either the Motif or the athena widget set.
gnu1
This directory contains the code for an example that uses the BitString object from the GNU class library.
The example illustrates how to incorporate an existing object (in this case the BitString) into a GAlib
Genome type. The gnu directory contains the source code needed for the BitString object (taken from the
GNU library) plus the two files (bitstr.h and bitstr.C) needed to define the new genome type and the
example file that runs the GA (gnuex.C).
pvmind1
This directory contains code that illustrates how to use GAlib with PVM in a master-slave configuration
wherein the master process is the genetic algorithm with a single population and each slave process is a
genome evaluator. The master sends individual genomes to the slave processes to be evaluated then the
slaves return the evaluations.
pvmpop1
This directory contains code that illustrates a PVM implementation of parallel populations. The master
process initiates a cluster of slaves each of which contains a single population. The master process
harvests individuals from all of the distributed populations. With a few modifications you can also use
this example with the deme GA from example 25 (it uses migration to distribute diversity between
pops).
randtest
Use this program to verify that the random number generator is generating suitably random numbers
on your machine. This is by no means a comprehensive random number testor, but it will give you
some idea of how well GAlib's random number generator is working.
1 These examples are included only in the UNIX distribution.