Class TEA
Unit
EA
Declaration
type TEA = class(TObject)
Description
Class implementing a general evolutionary algorithm (EA). TEA has the following features:
- Replacement schema
- Generational
- Steady-state
- Selection schema
- Tournament selection (arbitrary tournament size)
- Roulette-wheel selection
- Elitism
- Sub-populations
- Tunneling of individuals between sub-populations
- Support of arbitrary mutation and crossover operators
- Crossover schema
- Creep methods
- Encoding types
It should be noted that in the steady-state replacement scheme, the two worst individuals are replaced by new individuals.
In addition, TEA relies on the procedure TAgent.Mutate for performing mutations. This procedure needs to be implemented in a user-defined descendant of TAgent. See also TRobot.Mutate.
Hierarchy
TEA > TObjectMethods
Overview
Description
 |
constructor Create; |
 |
destructor Destroy; override; |
 |
procedure AdjustFitnessValues; |
 |
procedure GenerateCumulativeNormalizedFitnessSum(Pop: TPopulation); |
 |
procedure IncrementEvaluated; |
 |
procedure MakeGenerationalReplacement; |
 |
procedure MakeNewGeneration; |
 |
procedure MakeSteadyStateReplacement; |
 |
function RouletteWheelSelect(Pop: TPopulation): integer; |
Properties
Overview
Description
 |
property EvaluatedIndividuals: integer; |
 |
property Generation: integer; |
Provides acccess to the population. Initially Population is empty and the user must populate it by, for instance, using the procedure TPopulation.AddIndividual.
Generated by PasDoc 0.8.8.3 on 2005-05-28 08:13:49