EA

This page describes the evolutionary algorithm provided by EvoAl. Most information can be found on this page, while some special topics, such as grammatical evolution, or model-driven optimisation are documented in separate sub-pages.

Mapping domain data to the EA

The evolutionary-algorithm configuration allows mapping domain data to the EA. This is done using the genotype property for which EvoAl offers different implementations depending on the concrete problem data. Which combinations to use depends on the domain data and the existing constraints. The base idea of genotypes is depicted in the following diagram:

classDiagram
  Genotype *-- "1..*" Chromosome
  Chromosome *-- "1..*" Gene

As mentioned, Evoal offers different implementations of genotypes, chromosomes, and genes. The following diagram gives an overview of some of the existing chromosomes and genes.

classDiagram
  class Genotype
  class vGenotype ["vector-genotype"]
  class Chromosome ["chromosome"]
  class Gene ["gene"]
  class BitChromosome ["bit-chromosome"]
  class GrayChromosome ["gray-chromosome"]
  class DoubleChromosome ["double-chromosome"]
  class IntegerChromosome ["integer-chromosome"]

  vGenotype  --|> Genotype
  vGenotype *-- Chromosome : chromosomes
  Chromosome *-- Gene : genes
  BitChromosome --|> Chromosome
  BitChromosome : +int scale
  GrayChromosome --|> BitChromosome
  DoubleChromosome --|> Chromosome
  IntegerChromosome --|> Chromosome

  Gene : +data gene

The most commonly used and most simple genotype version is the vector-genotype.

de.evoal.core.ea.vector-genotype

The basic idea of this genotype is that the domain data consists of different values that are unstructured. Every single data item will be mapped to a gene depending on the type and requirements.

de.evoal.core.ea.bit-chromosome

Maps integer- and real-valued data to a bit representation. In this case, each bit of the value corresponds to a gene. Real-valued data is represented in a fixed-point precision since the IEEE-754 representation is not really useful for this purpose

de.evoal.core.ea.gray-chromosome

A special version of a bit encoding where the data is encoded using the gray code. This leads to different mutation behaviour.

de.evoal.core.ea.double-chromosome

In this chromosome, a gene corresponds to a single double value.

de.evoal.core.ea.integer-chromosome

In this chromosome, a gene corresponds to a single integer value.