|
TRSCustomGeneticProgramming<T>
|
The TRSCustomGeneticProgramming<T> component is the base class for including genetic programming (GP) in your programs. The descendant classes (TRSBinaryGeneticProgramming, TRSGeneticProgramming, etc) instantiate this generic type for floating point values and publishe the properties of this class and are what you will drop on your forms or data module.
Genetic programming is a computer science method, inspired by evolutionary biology, for automatically solving problems, without having to know or define the form or structure of optimum problem structure beforehand. You define the basic building blocks (functions, constants, and variables) of the problem and then the component does the rest. Genetic programming solves problems by evolving a group or population of candidate individuals through successive generations, selecting fitter (or better) child individuals for each generation, until a solution is found. It uses evolutionary biology techniques such as inheritance, mutation, selection, and crossover (also called recombination).
The TRSCustomGeneticProgramming<T> class contains a population of individuals (TRSGPPopulation<T>), which contain genetic program trees as DNA. These DNA are used to represent the solution to the search problem. They represent the functions, constants, and variables of each candidate solution (or individual), represented by a TRSGPIndividual<T> class.
After setting up the Instructions that your genetic programs can use, you set the InitialPopulation , define a fitness function ( OnEvaluateFitness event) to properly "score" each individual, and then Evolve your solution.
 Note
|
This is a generics-based implementation of a genetic programming component. For the RSGeneticProgramming|TRSGeneticProgramming class, the generics class is instantiated as floating point values. However, you can define your own genetic programming component using whatever type you want.
|
|
|
TRSCustomGeneticProgrammingExecutor<T>
|
Defines the base class for a component that can read and execute a genetic program. Use this class when you have already evolved a solution to your problem and need to execute the genetic program.
Note
|
This is a generics-based implementation of a genetic programming executor component. For the RSGeneticProgramming|TRSGeneticProgrammingExecutor class, the generics class is instantiated as floating point values. However, you can define your own genetic programming executor component using whatever type you want.
|
|
|
TRSGPIndividual<T>
|
Represents one individual, or candidate solution, in the Population of solutions in a genetic programming problem. In genetic programming ( RSGenerics.GeneticProgramming.TRSCustomGeneticProgramming<T> component), there is a population of individuals (TRSGPPopulation<T>), which contain tree-based DNA programs. Each candidate individual (represented by a TRSGPIndividual<T> class) has "DNA", which contains one program containing functions and terminals (constants and variables)
Genetic Programming works by evolving your population towards the solution of the problem. Parents are selected from the current generation to reproduce the children of the next generation. Evolving a new generation involves:
| • | Selecting 2 parents (Individuals) to reproduce |
| • | Splicing the DNA program trees of the 2 parents together to make a child (the child inherits the parents trees or sub-trees) |
| • | Optionally Mutating and Inverting the DNA trees of the child to provide randomness |
| • | Repeating the above steps until the population of the new generation has been produced. |
|
