genetiK::RankingSelection Class Reference
[Common GenetiK framework]

Ranking Selection method. More...

#include <RankingSelection.h>

Inheritance diagram for genetiK::RankingSelection:

[legend]Collaboration diagram for genetiK::RankingSelection:

[legend]List of all members.

Public Member Functions
	RankingSelection (void)
	default constructor
virtual	~RankingSelection (void)
	default destructor
void	init (Population *population)
	initializes the Ranking Selection method with a specific Population
Individual *	select ()
	selects an Individual from the current Population
Individual *	getBest ()
	retrieves the Individual having highest fitness in the Population
double	phi (unsigned int rank)
	used to calculate the probability to select an Invididual

---

Detailed Description

Ranking Selection method.

This selection scheme is designed to select to individuals with a probability that is function of their specific rank in the population, sorted by (ascending) fitness.

More formally, for a generic individual i

This class uses by default an exponential function to calculate probability (). The probability itself is proportional to phi (it is proportional, but not equal, because the class itself encapsulates normalization). It is possible to override phi to define different selection policies.

See also:

Individual::fitness(), SelectionMethod, RouletteWheel and TournamentSelection

---

Member Function Documentation

Individual * genetiK::RankingSelection::getBest (   )  [virtual]

retrieves the Individual having highest fitness in the Population Returns: the best Individual See also: Individual::fitness() Implements genetiK::SelectionMethod.

void genetiK::RankingSelection::init (  Population *  population  )  [virtual]

initializes the Ranking Selection method with a specific Population When initializing, the Cumulative Distribution Function (CDF) associated to the probability distribution defined by the previously mentioned function is calculated. Parameters: population  the Population to select the Individuals from Reimplemented from genetiK::SelectionMethod.

double genetiK::RankingSelection::phi (  unsigned int  rank  )

used to calculate the probability to select an Invididual It is possible to override this function to customize the schema used to calculate probability to select an Individual. This implementation, in particular, defines as . Parameters: rank  the position of the Individual in the sorted Population Returns: the value function described above See also: Individual::fitness()

Individual * genetiK::RankingSelection::select (   )  [virtual]

selects an Individual from the current Population An Individual i will be selected with probability Returns: the selected Individual Exceptions: IllegalStateException  if the Population has not been previously initialized with init(Population* population) See also: init(Population* population), IllegalStateException Implements genetiK::SelectionMethod.

---

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

• devel/genetik/src/RankingSelection.h
• devel/genetik/src/RankingSelection.cpp

---