Source Code Cross Referenced for MinimizingMakeChange.java in  » Development » jgap » examples » Java Source Code / Java DocumentationJava Source Code and Java Documentation

Java Source Code / Java Documentation
1. 6.0 JDK Core
2. 6.0 JDK Modules
3. 6.0 JDK Modules com.sun
4. 6.0 JDK Modules com.sun.java
5. 6.0 JDK Modules sun
6. 6.0 JDK Platform
7. Ajax
8. Apache Harmony Java SE
9. Aspect oriented
10. Authentication Authorization
11. Blogger System
12. Build
13. Byte Code
14. Cache
15. Chart
16. Chat
17. Code Analyzer
18. Collaboration
19. Content Management System
20. Database Client
21. Database DBMS
22. Database JDBC Connection Pool
23. Database ORM
24. Development
25. EJB Server geronimo
26. EJB Server GlassFish
27. EJB Server JBoss 4.2.1
28. EJB Server resin 3.1.5
29. ERP CRM Financial
30. ESB
31. Forum
32. GIS
33. Graphic Library
34. Groupware
35. HTML Parser
36. IDE
37. IDE Eclipse
38. IDE Netbeans
39. Installer
40. Internationalization Localization
41. Inversion of Control
42. Issue Tracking
43. J2EE
44. JBoss
45. JMS
46. JMX
47. Library
48. Mail Clients
49. Net
50. Parser
51. PDF
52. Portal
53. Profiler
54. Project Management
55. Report
56. RSS RDF
57. Rule Engine
58. Science
59. Scripting
60. Search Engine
61. Security
62. Sevlet Container
63. Source Control
64. Swing Library
65. Template Engine
66. Test Coverage
67. Testing
68. UML
69. Web Crawler
70. Web Framework
71. Web Mail
72. Web Server
73. Web Services
74. Web Services apache cxf 2.0.1
75. Web Services AXIS2
76. Wiki Engine
77. Workflow Engines
78. XML
79. XML UI
Java
Java Tutorial
Java Open Source
Jar File Download
Java Articles
Java Products
Java by API
Photoshop Tutorials
Maya Tutorials
Flash Tutorials
3ds-Max Tutorials
Illustrator Tutorials
GIMP Tutorials
C# / C Sharp
C# / CSharp Tutorial
C# / CSharp Open Source
ASP.Net
ASP.NET Tutorial
JavaScript DHTML
JavaScript Tutorial
JavaScript Reference
HTML / CSS
HTML CSS Reference
C / ANSI-C
C Tutorial
C++
C++ Tutorial
Ruby
PHP
Python
Python Tutorial
Python Open Source
SQL Server / T-SQL
SQL Server / T-SQL Tutorial
Oracle PL / SQL
Oracle PL/SQL Tutorial
PostgreSQL
SQL / MySQL
MySQL Tutorial
VB.Net
VB.Net Tutorial
Flash / Flex / ActionScript
VBA / Excel / Access / Word
XML
XML Tutorial
Microsoft Office PowerPoint 2007 Tutorial
Microsoft Office Excel 2007 Tutorial
Microsoft Office Word 2007 Tutorial
Java Source Code / Java Documentation » Development » jgap » examples 
Source Cross Referenced  Class Diagram Java Document (Java Doc) 


001:        /*
002:         * This file is part of JGAP.
003:         *
004:         * JGAP offers a dual license model containing the LGPL as well as the MPL.
005:         *
006:         * For licensing information please see the file license.txt included with JGAP
007:         * or have a look at the top of class org.jgap.Chromosome which representatively
008:         * includes the JGAP license policy applicable for any file delivered with JGAP.
009:         */
010:        package examples;
011:
012:        import java.io.*;
013:
014:        import org.jgap.*;
015:        import org.jgap.data.*;
016:        import org.jgap.impl.*;
017:        import org.jgap.xml.*;
018:        import org.w3c.dom.*;
019:
020:        /**
021:         * This class provides an implementation of the classic "Make change" problem
022:         * using a genetic algorithm. The goal of the problem is to provide a
023:         * specified amount of change (from a cash purchase) in the fewest coins
024:         * possible. This example implementation uses American currency (quarters,
025:         * dimes, nickels, and pennies).
026:         * <p>
027:         * This example may be seen as somewhat significant because it demonstrates
028:         * the use of a genetic algorithm in a less-than-optimal problem space.
029:         * The genetic algorithm does best when there is a smooth slope of fitness
030:         * over the problem space towards the optimum solution. This problem exhibits
031:         * a more choppy space with more local optima. However, as can be seen from
032:         * running this example, the genetic algorithm still will get the correct
033:         * (or a very close) answer virtually everytime.
034:         *
035:         * @author Neil Rotstan
036:         * @author Klaus Meffert
037:         * @since 1.0
038:         */
039:        public class MinimizingMakeChange {
040:            /** String containing the CVS revision. Read out via reflection!*/
041:            private final static String CVS_REVISION = "$Revision: 1.22 $";
042:
043:            /**
044:             * The total number of times we'll let the population evolve.
045:             */
046:            private static final int MAX_ALLOWED_EVOLUTIONS = 200;
047:
048:            /**
049:             * Executes the genetic algorithm to determine the minimum number of
050:             * coins necessary to make up the given target amount of change. The
051:             * solution will then be written to System.out.
052:             *
053:             * @param a_targetChangeAmount the target amount of change for which this
054:             * method is attempting to produce the minimum number of coins
055:             * @throws Exception
056:             *
057:             * @author Neil Rotstan
058:             * @author Klaus Meffert
059:             * @since 1.0
060:             */
061:            public static void makeChangeForAmount(int a_targetChangeAmount)
062:                    throws Exception {
063:                // Start with a DefaultConfiguration, which comes setup with the
064:                // most common settings.
065:                // -------------------------------------------------------------
066:                Configuration conf = new DefaultConfiguration();
067:                conf.setPreservFittestIndividual(true);
068:                conf.setKeepPopulationSizeConstant(true);
069:                // Set the fitness function we want to use, which is our
070:                // MinimizingMakeChangeFitnessFunction. We construct it with
071:                // the target amount of change passed in to this method.
072:                // ---------------------------------------------------------
073:                FitnessFunction myFunc = new MinimizingMakeChangeFitnessFunction(
074:                        a_targetChangeAmount);
075:                //    conf.setFitnessFunction(myFunc);
076:                conf
077:                        .setBulkFitnessFunction(new BulkFitnessOffsetRemover(
078:                                myFunc));
079:                // Optionally, this example is working with DeltaFitnessEvaluator.
080:                // See MinimizingMakeChangeFitnessFunction for details!
081:                // ---------------------------------------------------------------
082:                //    conf.setFitnessEvaluator(new DeltaFitnessEvaluator());
083:
084:                // Now we need to tell the Configuration object how we want our
085:                // Chromosomes to be setup. We do that by actually creating a
086:                // sample Chromosome and then setting it on the Configuration
087:                // object. As mentioned earlier, we want our Chromosomes to each
088:                // have four genes, one for each of the coin types. We want the
089:                // values (alleles) of those genes to be integers, which represent
090:                // how many coins of that type we have. We therefore use the
091:                // IntegerGene class to represent each of the genes. That class
092:                // also lets us specify a lower and upper bound, which we set
093:                // to sensible values for each coin type.
094:                // --------------------------------------------------------------
095:                Gene[] sampleGenes = new Gene[4];
096:                sampleGenes[0] = new IntegerGene(conf, 0, 3 * 10); // Quarters
097:                sampleGenes[1] = new IntegerGene(conf, 0, 2 * 10); // Dimes
098:                sampleGenes[2] = new IntegerGene(conf, 0, 1 * 10); // Nickels
099:                sampleGenes[3] = new IntegerGene(conf, 0, 4 * 10); // Pennies
100:                IChromosome sampleChromosome = new Chromosome(conf, sampleGenes);
101:                conf.setSampleChromosome(sampleChromosome);
102:                // Finally, we need to tell the Configuration object how many
103:                // Chromosomes we want in our population. The more Chromosomes,
104:                // the larger number of potential solutions (which is good for
105:                // finding the answer), but the longer it will take to evolve
106:                // the population (which could be seen as bad).
107:                // ------------------------------------------------------------
108:                conf.setPopulationSize(80);
109:
110:                // Create random initial population of Chromosomes.
111:                // Here we try to read in a previous run via XMLManager.readFile(..)
112:                // for demonstration purpose only!
113:                // -----------------------------------------------------------------
114:                Genotype population;
115:                try {
116:                    Document doc = XMLManager.readFile(new File(
117:                            "JGAPExample32.xml"));
118:                    population = XMLManager.getGenotypeFromDocument(conf, doc);
119:                } catch (UnsupportedRepresentationException uex) {
120:                    // JGAP codebase might have changed between two consecutive runs
121:                    population = Genotype.randomInitialGenotype(conf);
122:                } catch (FileNotFoundException fex) {
123:                    population = Genotype.randomInitialGenotype(conf);
124:                }
125:                // Now we initialize the population randomly, anyway!
126:                // If you want to load previous results from file, remove the next line!
127:                population = Genotype.randomInitialGenotype(conf);
128:                // Evolve the population. Since we don't know what the best answer
129:                // is going to be, we just evolve the max number of times.
130:                // ---------------------------------------------------------------
131:                long startTime = System.currentTimeMillis();
132:                for (int i = 0; i < MAX_ALLOWED_EVOLUTIONS; i++) {
133:                    if (!uniqueChromosomes(population.getPopulation())) {
134:                        throw new RuntimeException(
135:                                "Invalid state in generation " + i);
136:                    }
137:                    population.evolve();
138:                }
139:                long endTime = System.currentTimeMillis();
140:                System.out.println("Total evolution time: "
141:                        + (endTime - startTime) + " ms");
142:                // Save progress to file. A new run of this example will then be able to
143:                // resume where it stopped before!
144:                // ---------------------------------------------------------------------
145:
146:                // Represent Genotype as tree with elements Chromomes and Genes.
147:                DataTreeBuilder builder = DataTreeBuilder.getInstance();
148:                IDataCreators doc2 = builder
149:                        .representGenotypeAsDocument(population);
150:                // create XML document from generated tree
151:                XMLDocumentBuilder docbuilder = new XMLDocumentBuilder();
152:                Document xmlDoc = (Document) docbuilder.buildDocument(doc2);
153:                XMLManager.writeFile(xmlDoc, new File("JGAPExample26.xml"));
154:                // Display the best solution we found.
155:                // -----------------------------------
156:                IChromosome bestSolutionSoFar = population
157:                        .getFittestChromosome();
158:                System.out.println("The best solution has a fitness value of "
159:                        + bestSolutionSoFar.getFitnessValue());
160:                System.out.println("It contained the following: ");
161:                System.out.println("\t"
162:                        + MinimizingMakeChangeFitnessFunction
163:                                .getNumberOfCoinsAtGene(bestSolutionSoFar, 0)
164:                        + " quarters.");
165:                System.out.println("\t"
166:                        + MinimizingMakeChangeFitnessFunction
167:                                .getNumberOfCoinsAtGene(bestSolutionSoFar, 1)
168:                        + " dimes.");
169:                System.out.println("\t"
170:                        + MinimizingMakeChangeFitnessFunction
171:                                .getNumberOfCoinsAtGene(bestSolutionSoFar, 2)
172:                        + " nickels.");
173:                System.out.println("\t"
174:                        + MinimizingMakeChangeFitnessFunction
175:                                .getNumberOfCoinsAtGene(bestSolutionSoFar, 3)
176:                        + " pennies.");
177:                System.out.println("For a total of "
178:                        + MinimizingMakeChangeFitnessFunction
179:                                .amountOfChange(bestSolutionSoFar)
180:                        + " cents in "
181:                        + MinimizingMakeChangeFitnessFunction
182:                                .getTotalNumberOfCoins(bestSolutionSoFar)
183:                        + " coins.");
184:            }
185:
186:            /**
187:             * Main method. A single command-line argument is expected, which is the
188:             * amount of change to create (in other words, 75 would be equal to 75
189:             * cents).
190:             *
191:             * @param args amount of change in cents to create
192:             * @throws Exception
193:             *
194:             * @author Neil Rotstan
195:             * @author Klaus Meffert
196:             * @since 1.0
197:             */
198:            public static void main(String[] args) throws Exception {
199:                if (args.length != 1) {
200:                    System.out.println("Syntax: MinimizingMakeChange <amount>");
201:                } else {
202:                    int amount = 0;
203:                    try {
204:                        amount = Integer.parseInt(args[0]);
205:                    } catch (NumberFormatException e) {
206:                        System.out
207:                                .println("The <amount> argument must be a valid integer value");
208:                        System.exit(1);
209:                    }
210:                    if (amount < 1
211:                            || amount >= MinimizingMakeChangeFitnessFunction.MAX_BOUND) {
212:                        System.out
213:                                .println("The <amount> argument must be between 1 and "
214:                                        + (MinimizingMakeChangeFitnessFunction.MAX_BOUND - 1)
215:                                        + ".");
216:                    } else {
217:                        makeChangeForAmount(amount);
218:                    }
219:                }
220:            }
221:
222:            /**
223:             * @param a_pop the population to verify
224:             * @return true if all chromosomes in the populationa are unique
225:             *
226:             * @author Klaus Meffert
227:             * @since 3.3.1
228:             */
229:            public static boolean uniqueChromosomes(Population a_pop) {
230:                // Check that all chromosomes are unique
231:                for (int i = 0; i < a_pop.size() - 1; i++) {
232:                    IChromosome c = a_pop.getChromosome(i);
233:                    for (int j = i + 1; j < a_pop.size(); j++) {
234:                        IChromosome c2 = a_pop.getChromosome(j);
235:                        if (c == c2) {
236:                            return false;
237:                        }
238:                    }
239:                }
240:                return true;
241:            }
242:        }
www.java2java.com | Contact Us
Copyright 2009 - 12 Demo Source and Support. All rights reserved.
All other trademarks are property of their respective owners.