genetic_algorithm 1.0.0

It is fast, easy to use and extensable genetic algorithm implementaion for dart.

Genetic Algorithm

It is fast, easy to use and extensable genetic algorithm implementaion for dart.

Get started:

add the package to pubspec.yaml file.

genetic_algorithm: ^1.0.0

Evolution Engine

Evolution is the proccess of produces new candidate solution (generation) from old generation.

There are tow type of engine implemented:

1. GenerationEvolutionEngine.
2. PlusSelectionStrategyEngine.

here is example of creating generation evoliton engine that evolve List<int> chromosomes.

final engine = GenerationEvolutionEngine<List<int>>(...);

to start evolution process call evolve() method of engine instance which accept EngineSetting and return the best soultion.

List<int> result = await engine.evolve(settings);

You can extends EvolutionEngine class to create your own behavior of evolution process.

Population factory

Its is factory used to create initial population and random chromosomes.

/// This factory is used to create bit (zero or one) chromosomes with specific [length].
class AllOnesFactory extends PopulationFactory<Iterable<int>> {
  final int length;

  AllOnesFactory(this.length);

  @override
  Uint8List generateRandomChromosome(Random rnd) {
    final data = List.generate(length, (index) => rnd.nextInt(2));

    return Uint8List.fromList(data);
  }
}

Fitness evalutor

its the judje of candidate soultion.

// The problem is to generate soultion with only one.
class AllOnesFitnessEvaluator implements FitnessEvaluator<Iterable<int>> {
  @override
  double evaluate(Iterable<int> chromosome) {
    int correct = 0;
    for (final gene in chromosome) {
      if (gene == 1) {
        correct++;
      }
    }

    return correct / chromosome.length;
  }
}

Genetic Operator

it is operator that applied to population to generate new population. there are tow type of operators.

1. Crossover.
2. Mutation.

The package contains many built-in operators like:

1. SinglePointCrossover.
2. TowPointCrossover.
3. UniformCrossover.
4. SwapMutation.
5. etc.

You can create your own operator by extends Crossover or Mutation class or directly implement GeneticOperator class.

Termination conditions

The engine will evalute the generations until one of termination conditions is met.

There are serveral terminations:

1. Target value termination.
2. Elapsed time termination.
3. Generation count termination.
4. User cancel termination.
5. Stagination termination.

Engine Setting

Its settings that control the evolution process.

final settings = EngineSettigs(
    // The the of chromosomes in the each generation.
    populationSize: 100,
    // The number of chromosomes that will be moved to next generation without alteration,
    eliteCount: 2,
    // The list of chromosomes that will be added to first generation.
    seeds: [],
    // The list of conditions used to terminate the evalution process.
    terminations: [],
);