Instance of one of implementations of abstract class depending on context??

Question

Hi all,
I just wonder if it is possible in Java to call creation of a new instance of an implementation of an abstract class depending on context.
I mean something like:
abstract class Abstract
//1st implementation of Abstract class
class Concrete1 extends Abstract
//2nd implementation of Abstract class
class Concrete2 extends Abstract
}And now, somewhere else in the code, I would really need to call something like this:
Abstract test1 = new ...(); //here I need sometimes to be created instance of Concrete1, sometimes instance of Concrete2
Is there a way how to do this??

Answer 1

Some more concrete code if it helps:
Abstract class:
* Individual.java
* Created on 21. leden 2007, 1:08
package genetics;
* Abstract class defining fields and methods of one individual for genetic algorithms
* This class is supposed to be implemented according to problem to be solved
* @author Richard V�tek
* @version 1.0
abstract public class Individual implements Comparable<Individual>
   * Create random chromosomes for this individual
  protected abstract void createRandomChromosomes();
   * Count fitness of this individual
   * This number says how good is this individual (the higher number the better).
   * Better fitness means that this individual is closer to solution.
   * @return  int   Fitness of this individual
  protected abstract int getFitness();
   * Cross-breed this individual with another individual
   * This leaves untouched number of chromosomes to certain (randomly generated) position.
   * From this position on, it will swap chromosomes between this and another individual.
   * So this individual gets changed (cross-breeded) as well as the other, which is returned
   * as result of this method.
   * @param   other              The other individual to cross-breed with
   * @return  Individual         Hybrid of this and another individual (in fact, the other individual
   *                             after cross-breed (this (source) individual gets changed too after cross-breed)
  protected abstract Individual crossbreed(Individual other);
   * Mutate this individual
   * Mutate chromosomes of this individual; every chromosome is mutated
   * with probability set in settings of evolution.
   * This probability is supposed to be quite low number, roughly 1 %.
  protected abstract void mutate();
   * Check this individual
   * Check if this individual still suits the assignment.
   * If not, repair this individual to suit it again.
  protected abstract void check();
   * Implementation of Comparable: comparing of individuals by fitness
   * @param other Another individual to compare
  public int compareTo(Individual other)
    return this.getFitness() - other.getFitness();
One implementation class:
* KnapsackIndividual.java
* Created on 21. leden 2007, 1:41
package genetics;
import java.util.Random;
import java.util.TreeMap;
import knapsack.KnapsackProblem;
* This is practically the same as KnapsackProblem class but designed specially
* for solving knapsack problem with genetic algorithm so all unnecessary fields
* and methods are removed.
* @author Richard V�tek
* @version 1.0
public class KnapsackIndividual extends Individual
   * Chromosomes of this individual
   * In case of knapsack problem, they are things currentl in knasack
  protected boolean[] arrChromosomes;
   * Cached value of fitness of this individual
   * Used to not to count fitness of this individual everytime when needed
   * (and it is often needed); once counted, it will be read from this cached value
  private int intFitnessCache = Integer.MIN_VALUE;
   * Randomizer for random-driven methods (like mutation, etc.)
  private Random randomizer = new Random();
   * Reference to evolution to read mutation probability from
  protected Evolution evolution;
   * Assignment of problem instance
  protected KnapsackProblem assignment;
   * Create a new Individual instance
   * @param   assignment  Object representing assignment of particular problem
   * @param   evolution   Reference to evolution object to be able to read evolution's settings from
  public KnapsackIndividual(KnapsackProblem assignment, Evolution evolution)
    this.assignment = assignment;
    this.evolution = evolution;
    this.arrChromosomes = new boolean[assignment.getNumberOfThings()];
   * Create random chromosomes for this individual
   * @see Individual#createRandomChromosomes()
  protected void createRandomChromosomes()
    int intChromosomeCount = this.arrChromosomes.length;
    for (int i = 0; i < intChromosomeCount; i++)
      this.arrChromosomes[i] = this.randomizer.nextBoolean();
   * Get fitness of this individual
   * In case of knapsack, fitness is sum of prices of all things currently in knapsack
   * @see Individual#getFitness()
  protected int getFitness()
    //if cached value exist, return it
    if (this.intFitnessCache != Integer.MIN_VALUE)
      return this.intFitnessCache;
    //otherwise, count fitness of this individual
    int intChromosomeCount = this.arrChromosomes.length;
    int intSumOfValues = 0;
    //in case of knapsack, fitness is value of all things currently in knapsack
    //(sum of values of all things in knapsack)
    for (int i = 0; i < intChromosomeCount; i++)
      intSumOfValues = this.assignment.arrPrices;
//save counted fitness to cache
this.intFitnessCache = intSumOfValues;
return intSumOfValues;
* Cross-breed two individuals
* @param other The other individual for cross-breed
* @return The other individual after cross-breed (but this individual is affected too)
* @see Individual#crossbreed()
protected Individual crossbreed(Individual other)
int intChromosomeCount = this.arrChromosomes.length;
//position from which on swap chromosomes of this and the other individual
int intCrossbreedPosition = this.randomizer.nextInt(intChromosomeCount);
boolean booTemp;
//swap chromosomes from cross-breed position on
for (int i = intCrossbreedPosition; i < intChromosomeCount; i++)
booTemp = ((KnapsackIndividual) this).arrChromosomes[i];
((KnapsackIndividual) this).arrChromosomes[i] = ((KnapsackIndividual) other).arrChromosomes[i];
((KnapsackIndividual) other).arrChromosomes[i] = booTemp;
return other;
* Mutate individual chromosomes of this individual with certain probability
* In case of knapsack, particular thing is just inserted/taken out of the knapsack
* @see Individual#mutate()
protected void mutate()
//probability of mutation (in percents)
int intMutationProbability = this.evolution.getMutationProbability();
int intChromosomeCount = this.arrChromosomes.length;
//iterate through all chromosomes, mutating them with certain (set) probability
for (int i = 0; i < intChromosomeCount; i++)
//mutation probability passed => mutate this chromosome
if (this.randomizer.nextInt(100) < intMutationProbability)
this.arrChromosomes[i] = !this.arrChromosomes[i];
//when mutation finished, we must check if this individual still suits the assignment;
//if not, repait it
this.check();
* Check if this individual still suits the assignment; if not, repair it
* In case of knapsack it means that sum of weights of things currently in knapsack
* will not exceed capacity of backpack; if exceeds, take out as many things as necessary
* to not to exceed again; choose things to be taken out according to worst weight to price ratio
* @see Individual#check()
protected void check()
int intSumOfWeights = 0;
//list of things in the knapsack sorted by weight to price ratio
//key: index of thing in knapsack
//value: weight/price ratio
TreeMap<Integer, Float> things = new TreeMap<Integer, Float>();
for (int i = 0; i < this.arrChromosomes.length; i++)
//thing in the knapsack
if (this.arrChromosomes[i] == true)
//add its weight to sum of weights
intSumOfWeights += this.assignment.arrWeights[i];
//add it to the list of things sorted by weight to price ratio
things.put(i, (((float) this.assignment.arrWeights[i]) / ((float) this.assignment.arrPrices[i])));
//sum of weights exceeds knapsack capacity => take out as many things as necessary
while (intSumOfWeights > this.assignment.getKnapsackCapacity())
//take out thing with worst weight/price ratio from all things currently in knapsack
this.arrChromosomes[things.lastKey()] = false;
//update sum of weights of things currently in knapsack
intSumOfWeights -= things.get(things.lastKey());
//also remove this thing from list of things
things.remove(things.lastKey());
And another class, where i need this feature (tried to use generics for that, but they can't be used in this way):
* Evolution.java
* Created on 21. leden 2007, 2:47
package genetics;
import java.io.BufferedReader;
import java.io.FileReader;
import java.io.IOException;
import java.util.Arrays;
* Class for algorithms using simulated evolution to deal with a problem
* Tried to be designed general enough to allow to be used for every genetic algotihm.
* If true, only class Individual must be implemented according to problem to be solved,
* the rest of genetic algorithm stays the same.
* @author Richard V�tek
* @version
public class Evolution<Problem, IndividualClass extends Individual>
   * Number of generations of evolution to finish
  protected int intGenerationCount;
   * Number of individuals in each generation
  protected int intGenerationSize;
   * Elite individual count
   * All elite individuals are just copied from one generation to another with no change
  protected int intGenerationEliteCount;
   * Number of individuals participating a tournament
   * To select an good individual for next generation, tournaments are hold.
   * This affects number of individuals which one good individual is selected
   * from in one tournament.
   * @see <a href="http://cs.felk.cvut.cz/%7Exobitko/ga/example_f.html">Genetic Algorithm Example Applet</a>
  protected int intGenerationTournamentSize;
   * Probability of mutation (in percents)
  protected int intMutationProbability;
   * Current generation of individuals in evolution
  protected Generation<IndividualClass> thisGeneration;
   * Next generation of individuals in evolution
  protected Generation<IndividualClass> nextGeneration;
   * Fitness of best individual in this generation
  private int intIndividualBestFitness;
   * Sum of fitnesses of all individuals in this generation
  private int intIndividualsSumFitness;
   * Fitness of worst individual in this generation
  private int intIndividualWorstFitness;
   * Fitness of best elite individual in (every) generation
   * Auxilliary variable to not to count statistics for elite individuals
   * in each generation as well (not needed - elite individuals don't change themselves)
  private int intIndividualEliteBestFitness;
   * Sum of fitnesses of all elite individuals in (every) generation
   * Auxilliary variable to not to count statistics for elite individuals
   * in each generation as well (not needed - elite individuals don't change themselves)
  private int intIndividualElitesSumFitness;
   * Fitness of worst elite individual in (every) generation
   * Auxilliary variable to not to count statistics for elite individuals
   * in each generation as well (not needed - elite individuals don't change themselves)
  private int intIndividualEliteWorstFitness;
   * Create a new instance of Evolution (settings passed through parameters)
   * @param   intGenerationCount            Number of generation of evolution to finish
   * @param   intGenerationSize             Number of individuals in each generation
   * @param   intGenerationEliteRatio       Elite individuals to generation size ratio (in percents)
   * @param   intGenerationTournamentRatio  Members of tournament to generation size ratio (in percents)
   * @param   intMutationProbability        Probability of mutation of each chromosome of each individual of generation (in percents)
   * @see     #intGenerationEliteCount
   * @see     #intGenerationTournamentSize
  public Evolution(
    int intGenerationCount, int intGenerationSize,
    int intGenerationEliteRatio, int intGenerationTournamentRatio,
    int intMutationProbability)
    this.intGenerationCount = intGenerationCount;
    this.intGenerationSize = intGenerationSize;
    this.intGenerationEliteCount = (int) (intGenerationSize * (intGenerationEliteRatio / 100.0));
    this.intGenerationTournamentSize = (int) (intGenerationSize * (intGenerationTournamentRatio / 100.0));
    this.intMutationProbability = intMutationProbability;
   * Create a new instance of Evolution (settings loaded from settings file)
   * @param   strSettingFile  Name of file containing settings for evolution
   * @throws  IOException     File does not exist, cannot be read, etc.
   * @throws  Exception       Another exception occured during loading of file
  public Evolution(String strSettingFile)
    BufferedReader settings;
    String settingsLine;
    int intLineCounter = 0;
    int intSetting;
    try
      settings = new BufferedReader(new FileReader(strSettingFile));
      while ((settingsLine = settings.readLine()) != null)
        intLineCounter++;
        settingsLine = settingsLine.substring(0, settingsLine.indexOf("\t"));
        intSetting = Integer.parseInt(settingsLine);
        switch (intLineCounter)
          case 1:
            this.intGenerationCount = intSetting;
            break;
          case 2:
            this.intGenerationSize = intSetting;
            break;
          case 3:
            this.intGenerationEliteCount = (int) (this.intGenerationSize * (intSetting / 100.0));
            break;
          case 4:
            this.intGenerationTournamentSize = (int) (this.intGenerationSize * (intSetting / 100.0));
            break;
          case 5:
            this.intMutationProbability = intSetting;
            break;
        } //switch
      } //while
      //after reading has been completed, let's close the stream
      settings.close();
    } //try
    //IO exception - file not found, cannot be read, etc.
    catch (IOException ioe)
      System.out.println("Vyskytl se I/O probl�m p&#345;i na&#269;�t�n� zad�n� ze souboru " + strSettingFile);
    //Exception - another problem during reading of file
    catch (Exception e)
      System.out.printf("Vyskytl se n&#283;jak� divn� probl�m p&#345;i na&#269;�t�n� zad�n� ze souboru %s. V�pis vyj�mky:\n", strSettingFile);
      e.printStackTrace();
   * Vivify first generation for evolution
   * Necessary number of individuals is created with random chromosomes.
   * Their chromosomes must then be checked if they suit the assignment
   * and if not so, repaired.
  private Generation<IndividualClass> vivifyFirstGeneration()
    //create a brand-new generation
    Generation generation = new Generation<IndividualClass>(this);
    int intTemp;
    //for all individual of this generation
    for (int i = 0; i < this.intGenerationSize; i++)
//create an individual with no chromosomes
generation.arrIndividuals[i] = new IndividualClass(this, Problem);
      //create a set of random chromosomes
      neration.arrIndividuals.createRandomChromosomes();
//if these chromosomes does not suit assignment, repair them
generation.arrIndividuals[i].check();
//sort Individuals by fitness so elite individuals get to first positions of array
Arrays.sort(generation.arrIndividuals);
//now count statistics for elite individuals (it is enough to count them once,
//elite don't get changed so their statistics don't get changed either)
this.intIndividualEliteBestFitness = Integer.MIN_VALUE;
this.intIndividualElitesSumFitness = 0;
this.intIndividualEliteWorstFitness = Integer.MAX_VALUE;
//count statistics for elite individuals
for (int i = 0; i < this.intGenerationEliteCount; i++)
intTemp = generation.arrIndividuals[i].getFitness();
//better fitness than best fitness so far
if (intTemp > this.intIndividualEliteBestFitness)
this.intIndividualEliteBestFitness = intTemp;
//worse fitness than worst fitness so far
else if (intTemp < this.intIndividualEliteWorstFitness)
this.intIndividualEliteWorstFitness = intTemp;
this.intIndividualElitesSumFitness += intTemp;
//reset generation's statistics
this.intIndividualBestFitness = this.intIndividualEliteBestFitness;
this.intIndividualsSumFitness = this.intIndividualElitesSumFitness;
this.intIndividualWorstFitness = this.intIndividualEliteWorstFitness;
//count generation's statistics also from rest of individuals
for (int i = this.intGenerationEliteCount; i < this.intGenerationSize; i++)
updateGenerationStatistics(generation.arrIndividuals[i].getFitness());
return generation;
* Get next generation in evolution
* Core method for all evolution; Through this method, new generation is obtained.
* Every next generation should contain better individuals than the previous one
* (till certain point) so with growing number of iterations in evolution, we
* get better results (till certain point).
* Everytime all elite individuals are copied to next generation, then hold needed number of
* tournaments to choose a pair of good-looking individuals, cross-breed individuals in these
* pairs, mutate them (and repair if necessary) and finally add to next generation.
* @return Generation Next generation in evolution
* @see Generation#tournament()
* @see Generation#crossbreed()
* @see Generation#mutate()
private Generation getNextGeneration()
Generation nextGeneration = new Generation(this);
//number of pairs of individuals to select for next generation
int intIndividualPairsToSelect = (this.intGenerationSize - this.intGenerationEliteCount) / 2;
int intTemp;
//reset generation's statistics
this.intIndividualBestFitness = this.intIndividualEliteBestFitness;
this.intIndividualsSumFitness = this.intIndividualElitesSumFitness;
this.intIndividualWorstFitness = this.intIndividualEliteWorstFitness;
//just copy all elite individuals from this generation to another
//(they are on first positions of array of individuals)
for (int i = 0; i < this.intGenerationEliteCount; i++)
nextGeneration.arrIndividuals[i] = this.thisGeneration.arrIndividuals[i];
//hold as many tournaments as necessary to select remaining number of pairs
//of good-looking individuals for next generation (apart from the elite ones)
for (int i = 0; i < intIndividualPairsToSelect; i++)
this.thisGeneration.tournament();
this.thisGeneration.crossbreed();
this.thisGeneration.mutate();
//add this individual in next generation
nextGeneration.arrIndividuals[2 * i] = this.thisGeneration.nextGenerationIndividual01;
//update statistics of generation
updateGenerationStatistics(this.thisGeneration.nextGenerationIndividual01.getFitness());
//add this individual in next generation
nextGeneration.arrIndividuals[2 * i + 1] = this.thisGeneration.nextGenerationIndividual02;
//update statistics of generation
updateGenerationStatistics(this.thisGeneration.nextGenerationIndividual02.getFitness());
//next generation is complete => return it
return nextGeneration;
* Update statistics of current generations
* @param intFitness Fitness that may possibly update generation's statistics
* (best fitness, worst fitness, sum of fitnesses)
private void updateGenerationStatistics(int intFitness)
//better fitness than best fitness so far
if (intFitness > this.intIndividualBestFitness)
this.intIndividualBestFitness = intFitness;
//worse fitness than worst fitness so far
else if (intFitness < this.intIndividualWorstFitness)
this.intIndividualWorstFitness = intFitness;
//update sum of fitnesses as well (for average fitness)
this.intIndividualsSumFitness += intFitness;
* Execute evolution process
* Vivify first generation and then as many next generations as set in settings of evolution
public void evolution()
this.thisGeneration = vivifyFirstGeneration();
//output generation's statistics
System.out.printf("Generace 0:\t%d\t%d\t%d", this.getIndividualBestFitness(), this.getIndividualAverageFitness(), this.getIndividualWorstFitness());
for (int i = 0; i < this.intGenerationCount; i++)
this.nextGeneration = getNextGeneration();
this.thisGeneration = this.nextGeneration;
//output generation's statistics
System.out.printf("Generace %d:\t%d\t%d\t%d", i, this.getIndividualBestFitness(), this.getIndividualAverageFitness(), this.getIndividualWorstFitness());
* Get best fitness of all individuals in this generation
public int getIndividualBestFitness()
return intIndividualBestFitness;
* Get average fitness of all individuals in this generation
public float getIndividualAverageFitness()
return (this.intIndividualsSumFitness / (float) this.intGenerationSize);
* Get worst fitness of all individuals in this generation
public int getIndividualWorstFitness()
return intIndividualWorstFitness;
* Get probability of mutation
* @return Probability of mutation of each chromosome of every individual in generation (in percents)
public int getMutationProbability()
return intMutationProbability;