/*

     dsh-evolve  Framework for evolutionary algorithms.

     Copyright (c) 2005-2013 held jointly by the individual authors.

     This library is free software; you can redistribute it and/or modify it

     under the terms of the GNU Lesser General Public License as published

     by the Free Software Foundation; either version 3 of the License, or (at

     your option) any later version.

     This library is distributed in the hope that it will be useful, but WITHOUT

     ANY WARRANTY; with out even the implied warranty of MERCHANTABILITY or

     FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public

     License for more details.

     You should have received a copy of the GNU Lesser General Public License

     along with this library;  if not, write to the Free Software Foundation,

     Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307  USA.

     > http://www.fsf.org/licensing/licenses/lgpl.html

     > http://www.opensource.org/licenses/lgpl-license.php

 */

 package org.dishevelled.evolve.select;

 import java.util.Map;

 import java.util.List;

 import java.util.ArrayList;

 import java.util.Comparator;

 import java.util.Collection;

 import java.util.Collections;

 import org.dishevelled.weighted.WeightedMap;

 import org.dishevelled.weighted.HashWeightedMap;

 import org.dishevelled.evolve.Selection;

 /**

  * Elitist selection function.

  *

  * @param <I> individual type

  * @author  Michael Heuer

  */

 public final class ElitistSelection<I>

     implements Selection<I>

 {

     /** Number of individuals. */

     private final int individuals;

     /** Default hash map load factor. */

     private static final float DEFAULT_LOAD_FACTOR = 0.75f;

     /** WeightedMap entry comparator. */

     private final WeightedMapEntryComparator weightedMapEntryComparator = new WeightedMapEntryComparator();

     /**

      * Create a new elitist selection function with the specified

      * number of individuals.

      *

      * @param individuals number of individuals for this elitist

      *    selection function, must be <code>&gt;= 0</code>

      */

     public ElitistSelection(final int individuals)

     {

         if (individuals < 0)

         {

             throw new IllegalArgumentException("individuals must be greater than or equal to zero");

         }

         this.individuals = individuals;

     }

     /**

      * Return the number of individuals for this elitist selection function.

      *

      * @return the number of individuals for this elitist selection function

      */

     public int getIndividuals()

     {

         return individuals;

     }

     /** {@inheritDoc} */

     public Collection<I> select(final Collection<I> population,

                                 final WeightedMap<I> scores)

     {

         if (scores.totalWeight() == 0.0d)

         {

             throw new IllegalStateException("scores total weight must be greater than zero");

         }

         int size = population.size();

         List<I> selected = new ArrayList<I>(size);

         // sort keys by score

         List<Map.Entry<I, Double>> sortedKeys = new ArrayList<Map.Entry<I, Double>>(scores.entrySet());

         Collections.sort(sortedKeys, weightedMapEntryComparator);

         // take the top n individuals

         List<Map.Entry<I, Double>> eliteKeys = sortedKeys.subList(0, Math.min(individuals, size));

         // create intermediate map of the top n individuals

         WeightedMap<I> intermediate = new HashWeightedMap<I>(individuals, DEFAULT_LOAD_FACTOR);

         for (Map.Entry<I, Double> e : eliteKeys)

         {

             intermediate.put(e.getKey(), Double.valueOf(0.0d));

         }

         // update fitness based on number of top n individuals

         for (I i : intermediate.keySet())

         {

             intermediate.put(i, Double.valueOf(1.0d / individuals));

         }

         // fitness proportional selection on intermediate map

         for (int i = 0; i < size; i++)

         {

             selected.add(intermediate.sample());

         }

         sortedKeys = null;

         eliteKeys = null;

         intermediate = null;

         return selected;

     }

     /**

      * WeightedMap entry comparator.

      */

     private class WeightedMapEntryComparator

         implements Comparator<Map.Entry<I, Double>>

     {

         /** {@inheritDoc} */

         public int compare(final Map.Entry<I, Double> e1, final Map.Entry<I, Double> e2)

         {

             // reverse order of comparison to get decreasing sort

             return e2.getValue().compareTo(e1.getValue());

         }

     }

 }