---

Selector Class Reference

Selection handler of an Individual. More...

#include <selection.h>

Inheritance diagram for Selector:

List of all members.

Public Methods
void	read (const Genome &g)
double	select (const SelectionSituation &situation, int self_i, int other_j) const
int	score () const
void	virginize ()
Static Public Methods
void	addGenesTo (Gentainer &g, const StringMap &params)
Protected Methods
virtual double	selectWithMethod (const SelectionSituation &situation, int smno, int i, int j) const
double	muLambdaSelection (const SelectionSituation &situation, int i, int j) const
double	linearRanking (const SelectionSituation &situation, int i, int j) const
double	proportionalSelection (const SelectionSituation &situation, int i, int j) const
double	tournamentSelection (const SelectionSituation &situation, int i, int j) const
Protected Attributes
int	mScore

---

Detailed Description

Selection handler of an Individual.

Each individual has one instance of this class to handle all it's selection tasks.

This class provides some additional functionality for SelectionPrms. It is kept separate because we don't want to give these methods if not necessary.

Definition at line 239 of file selection.h.

---

Member Function Documentation

void addGenesTo (  Gentainer &    g, const StringMap &    params )  [static]

Inserts genes that control selection into the given genome, if the selection parameters are set to be self-adaptive. Parameters: params  Dynamic parameters in a String Map. See various methods of SelectionPrms and SelectionMatrix for further descriptions of the parameters. params  ["mu"] mu-parameter for (mu,+lambda)-Selection. params  ["q"] q-parameter for tournament selection. params  ["eta+"] eta+-parameter for proportional selection. params  ["adaptMu"] Should mu be self-adapted? [Default:0] params  ["adaptQ"] Should q be self-adapted? [Default:0] params  ["adaptEta+"] Should eta+ be self-adapted? [Default:0] Definition at line 212 of file selection.cc. References Gentainer::add(), and Genstruct::hide(). Referenced by Individual::addGenesTo().

double linearRanking (  const SelectionSituation &    situation, int    i, int    j )  const [protected]

Selection method where the individuals are ranked according to their fitness, and the selection probability is a linear function of this rank. Definition at line 304 of file selection.cc. References SelectionPrms::mEtaPlus, SimplePopulation::mUseGlobalEtaPlus, SelectionSituation::population(), SimplePopulation::selParams(), and SimplePopulation::size(). Referenced by selectWithMethod().

double muLambdaSelection (  const SelectionSituation &    situation, int    i, int    j )  const [protected]

Evolution Strategies (mu,+lambda)-Selection, where lambda is the number fittest individuals in the population that can be potential parents, and lambda is the number of offspring. Definition at line 293 of file selection.cc. References SelectionPrms::muFor(), SimplePopulation::mUseGlobalMu, SelectionSituation::population(), SimplePopulation::selParams(), and SimplePopulation::size(). Referenced by selectWithMethod().

double proportionalSelection (  const SelectionSituation &    situation, int    i, int    j )  const [protected]

Proportional selection. Definition at line 312 of file selection.cc. References FitnessStats::avgFitness(), Individual::getfitness(), SelectionSituation::getOrdered(), SimplePopulation::mFitnessStats, and SelectionSituation::population(). Referenced by selectWithMethod().

void read (  const Genome &    g )

Reads selection parameters from the given genome. Definition at line 233 of file selection.cc. References Gentainer::getGene(), SelectionPrms::mEtaPlus, SelectionPrms::mMuPart, SelectionPrms::mQ, and SelectionPrms::mSelMethodW. Referenced by Individual::incarnate().

int score (    )  const [inline]

Tells how many times the individual has got some. (Hmm, shouldn't this kind of information be private??). Definition at line 274 of file selection.h. References mScore.

double select (  const SelectionSituation &    situation, int    self_i, int    other_j )  const

Returns the selection affinity of individual i to individual j, combined over all different selection methods. Definition at line 258 of file selection.cc. References SelectionPrms::mSelMethodW, and selectWithMethod(). Referenced by SelectionMatrix::calculateMatrix().

double selectWithMethod (  const SelectionSituation &    situation, int    smno, int    i, int    j )  const [protected, virtual]

Returns the selection probability for individuals i and j. Parameters: smno  Selection method number (see the enum 'smconsts'). i  The individual making the selection choise. j  The target of selection. Returns: Selection probability (not affinity) of Individual j by i. Definition at line 279 of file selection.cc. References linearRanking(), muLambdaSelection(), proportionalSelection(), and tournamentSelection(). Referenced by select().

double tournamentSelection (  const SelectionSituation &    situation, int    i, int    j )  const [protected]

Tournament selection. Definition at line 316 of file selection.cc. References SelectionPrms::mQ, SimplePopulation::mUseGlobalQ, SelectionSituation::population(), SimplePopulation::selParams(), and SimplePopulation::size(). Referenced by selectWithMethod().

void virginize (    )  [inline]

Makes the individual virgin again (resets it's score). Wow. * Definition at line 277 of file selection.h. References mScore.

---

Member Data Documentation

int mScore [protected]

Some selection methods, like the method the canonic GA, select some of the best individuals with 100% propability. To achieve this, we have to count how many times an individual has got some. Definition at line 319 of file selection.h. Referenced by score(), and virginize().

---

The documentation for this class was generated from the following files:

• selection.h
• selection.cc

---