Free cookie consent management tool by TermsFeed Policy Generator

source: branches/OKBJavaConnector/ECJClient/src/ec/select/BoltzmannSelection.java @ 7759

Last change on this file since 7759 was 6152, checked in by bfarka, 14 years ago

added ecj and custom statistics to communicate with the okb services #1441

File size: 6.4 KB
Line 
1/*
2  Copyright 2006 by Sean Luke
3  Licensed under the Academic Free License version 3.0
4  See the file "LICENSE" for more information
5*/
6
7
8package ec.select;
9import ec.util.*;
10import ec.*;
11
12/*
13 * BoltzmannSelection.java
14 *
15 * Created: Thu May 14 2009
16 * By: Jack Compton
17 */
18
19/**
20 * Similar to FitProportionateSelection, but with a Simulated Annealing style twist. BoltzmannSelection picks individuals of a population in
21 * proportion to an adjusted version of their fitnesses instead of their actual fitnesses as returned by fitness(). The adjusted fitness is
22 * calculated by e^(fitness/current_temperature) where current_temperature is a temperature value that decreases by a constant cooling rate as
23 * generations of evolution pass. The current_temperature is calculated by starting-temperature - (cooling-rate * the_current_generation_number).
24 * When the temperature dips below 1.0, annealing ceases and BoltzmannSelection reverts to normal FitProportionateSelection behavior.
25 *
26 * <p>
27 * Like FitProportionateSelection this is not appropriate for steady-state evolution.
28 * If you're not familiar with the relative advantages of
29 * selection methods and just want a good one,
30 * use TournamentSelection instead. Not appropriate for
31 * multiobjective fitnesses.
32 *
33 * <p><b><font color=red>
34 * Note: Fitnesses must be non-negative.  0 is assumed to be the worst fitness.
35 * </font></b>
36
37 <p><b>Typical Number of Individuals Produced Per <tt>produce(...)</tt> call</b><br>
38 Always 1.
39 
40 <p><b>Parameters</b><br>
41 <table>
42 <tr><td valign=top><i>base.</i><tt>starting-temperature</tt><br>
43 <font size=-1>double = some large number (defaults to 1.0)</font></td>
44 <td valign=top>(the starting temperature for our simulated annealing style adjusted fitness proportions)</td></tr>
45 
46 <tr><td valign=top><i>base.</i><tt>cooling-rate</tt><br>
47 <font size=-1> double = some smaller number (defaults to 0.0 which causes BoltzmannSelection to behave just as FitProportionateSelection would)</font></td>
48 <td valign=top>(how slow, or fast, do you want to cool the annealing fitness proportions?)</td></tr>
49 
50 </table>
51
52 <p><b>Default Base</b><br>
53 select.boltzmann
54
55 *
56 * @author Jack Compton
57 * @version 1.0
58 */
59
60public class BoltzmannSelection extends FitProportionateSelection
61    {
62    /** Default base */
63    public static final String P_BOLTZMANN = "boltzmann";
64               
65    /** Starting temperature parameter */
66    public static final String P_STARTING_TEMPERATURE = "starting-temperature";
67               
68    /** Cooling rate parameter */
69    public static final String P_COOLING_RATE = "cooling-rate";
70               
71    /** Starting temperature **/
72    private double startingTemperature;
73               
74    /** Cooling rate */
75    private double coolingRate;
76               
77    public Parameter defaultBase()
78        {
79        return SelectDefaults.base().push(P_BOLTZMANN);
80        }
81               
82    public void setup(final EvolutionState state, final Parameter base)
83        {
84        super.setup(state,base);
85                       
86        Parameter def = defaultBase();
87                       
88        coolingRate = state.parameters.getDouble(base.push(P_COOLING_RATE),def.push(P_COOLING_RATE)); // default cooling rate of 1.0 per generation
89        startingTemperature = state.parameters.getDouble(base.push(P_STARTING_TEMPERATURE),def.push(P_STARTING_TEMPERATURE)); // default starting temp is 0.0/completely cooled - will act as normal fit proportionate selection
90                       
91        if (coolingRate <= 0)
92            {
93            //Hey! you gotta cool! Set your cooling rate to a positive value!
94            state.output.fatal("Cooling rate should be a positive value.",base.push(P_COOLING_RATE),def.push(P_COOLING_RATE));
95            }
96                       
97        if ((startingTemperature - coolingRate) <= 0) {
98            // C'mon, you should cool slowly if you want boltzmann selection to be effective.
99            state.output.fatal("For best results, try to set your temperature to cool to 0 a more slowly. This can be acheived by increasing your starting-temperature and/or decreasing your cooling rate.\nstarting-temperatire/cooling-rate: " + startingTemperature + " / " + coolingRate);                             
100            }
101                       
102        int total_generations = state.numGenerations;
103        if (total_generations == 0)
104            {
105            //Load from parameter database!!
106            state.output.fatal("Hey now, we gotta load the total_generations from the param DB");
107            }
108                       
109        if ((startingTemperature - (coolingRate * total_generations)) > 0)
110            {
111            //Either your cooling rate is to low, or your starting temp is too high, because at this rate you will never cool to 0! (kind of essential to reaping the benefits of boltzmann selection)
112            state.output.warning("If you want BoltzmannnSelection to be effective, your temperature should cool to 0 before all generations have passed. Make sure that (starting-temperature - (cooling-rate * generations)) <= 0.");
113            }
114       
115        }
116
117    // completely override FitProportionateSelection.prepareToProduce
118    public void prepareToProduce(final EvolutionState s,
119        final int subpopulation,
120        final int thread)
121        {
122        // load fitnesses
123        fitnesses = new float[s.population.subpops[subpopulation].individuals.length];
124        for(int x=0;x<fitnesses.length;x++)
125            {
126            fitnesses[x] = (float) boltzmannExpectedValue(
127                ((Individual)(s.population.subpops[subpopulation].individuals[x])).fitness.fitness(),
128                s); // adjust the fitness proportion according to current temperature.
129            if (fitnesses[x] < 0) // uh oh
130                s.output.fatal("Discovered a negative fitness value.  BoltzmannnSelection requires that all fitness values be non-negative(offending subpopulation #" + subpopulation + ")");
131            }
132       
133        // organize the distribution.  All zeros in fitness is fine
134        RandomChoice.organizeDistribution(fitnesses, true);
135        }
136
137    private double boltzmannExpectedValue(double fitness, final EvolutionState s)
138        {
139        double current_temperature = startingTemperature - (coolingRate * s.generation);
140        if (current_temperature < 1.0)
141            return fitness;
142        return Math.exp(fitness/current_temperature);
143        }
144       
145    }
Note: See TracBrowser for help on using the repository browser.