Population Class Reference

A population of individual solutions; this is the core of a Pareto-based memetic algorithm. More...

#include <Population.hpp>

List of all members.

Public Member Functions
	Population (int nobj)
	Create a population for individuals that have nobj objectives.
void	fromStream (istream &ist, const IndividualReader *rdr)
	Insert individuals from a stream, using a concrete reader.
void	toStream (ostream &ost) const
	Write all individuals to a stream.
double	getObjectiveValue (int individualIndex, int objectiveIndex) const
	Get one objective value from an indexed individual.
vector< double >	getTrace (int individualIndex) const
	Get the vector of traced values from an indexed individual.
int	getIndividualID (int individualIndex) const
	Get the unique ID of an individual.
void	setIndividualID (int individualIndex, int newID)
	Set the unique ID of an individual.
size_t	size () const
	Get the current size of the population.
void	adopt (unique_ptr< Individual > indiv)
	Insert a new individual, obtaining ownership.
void	clear ()
	Clear contents; individuals become deleted.
void	adoptFrom (Population &p1)
	Insert new individuals whose ownership is taken from the other pop.
void	adoptFromTwo (Population &p1, Population &p2)
	Insert new individuals whose ownership is taken from two other pops.
void	deepCopyFrom (Population &p1)
	Insert new individuals that are cloned from the other pop.
void	evaluate ()
	Command all individuals to evaluate their objectives.
void	mutate ()
	Command all individuals to mutate, according to their (possibly individually evolving) rules.
void	improve (mt19937 *mt)
	Command all individuals to do local improvement, according to their own, possibly individually evolving or stochastic, rules.
int	binaryTournament (int ii1, int ii2, mt19937 *mt)
	Return the index of the winner of a binary tournament.
void	assignRankAndCrowdingDistance ()
	Assign ranks and crowding distances to all individuals.
void	fillNondominatedSortFrom (Population &parent)
	Insert copies of individuals from a parent population, as per the nondominated fill of NSGA-II.
void	selectAndCrossFrom (Population &parent, mt19937 *mt)
Protected Member Functions
void	quicksortFrontObj (int objcount, int *obj_array, int front_size) const
void	quicksortDist (int *dist, int front_size) const
void	crowdingFillFrom (Population &mixed, int count, int front_size, void *plstelite)
void	assignCrowdingDistance (int *dist, int **obj_array, int front_size)
void	assignCrowdingDistanceList (void *plst, int front_size)
void	assignCrowdingDistanceIndices (int c1, int c2)
int	binaryTournament (int a, int b, mt19937 *mt) const
Protected Attributes
vector< unique_ptr< Individual > >	_individuals
	Pointers to individuals.
size_t	_nobj
	Number of objectives.

---

Detailed Description

A population of individual solutions; this is the core of a Pareto-based memetic algorithm.

The population is responsible for the storage of individuals, and dispatching the the operations that produce the next population. The population can also be used as an archive of any number of historical solutions.

---

Constructor & Destructor Documentation

Population::Population

(

int

nobj

)

Create a population for individuals that have nobj objectives.

---

Member Function Documentation

void Population::adopt

(

unique_ptr< Individual >

indiv

)

Insert a new individual, obtaining ownership.

void Population::adoptFrom

(

Population &

p1

)

Insert new individuals whose ownership is taken from the other pop.

void Population::adoptFromTwo	(	Population &	p1,
		Population &	p2
	)

Insert new individuals whose ownership is taken from two other pops.

void Population::assignRankAndCrowdingDistance

(

)

Assign ranks and crowding distances to all individuals.

int Population::binaryTournament	(	int	ii1,
		int	ii2,
		mt19937 *	mt
	)

Return the index of the winner of a binary tournament.

Needs a random generator to resolve ties.

void Population::clear

(

)

Clear contents; individuals become deleted.

void Population::deepCopyFrom

(

Population &

p1

)

Insert new individuals that are cloned from the other pop.

void Population::evaluate

(

)

Command all individuals to evaluate their objectives.

void Population::fromStream	(	istream &	ist,
		const IndividualReader *	rdr
	)

Insert individuals from a stream, using a concrete reader.

int Population::getIndividualID

(

int

individualIndex

)

const

Get the unique ID of an individual.

double Population::getObjectiveValue	(	int	individualIndex,
		int	objectiveIndex
	)			const

Get one objective value from an indexed individual.

vector< double > Population::getTrace

(

int

individualIndex

)

const

Get the vector of traced values from an indexed individual.

void Population::improve

(

mt19937 *

mt

)

Command all individuals to do local improvement, according to their own, possibly individually evolving or stochastic, rules.

A thread-specific random generator must always be given, because some operators may require randomization.

void Population::setIndividualID	(	int	individualIndex,
		int	newID
	)

Set the unique ID of an individual.

size_t Population::size

(

)

const

Get the current size of the population.

void Population::toStream

(

ostream &

ost

)

const

Write all individuals to a stream.

---

Member Data Documentation

vector<unique_ptr<Individual> > Population::_individuals [protected]

Pointers to individuals.

FIXME: Expect some brainwork with the pointers and the issue of who should own each individual!

size_t Population::_nobj [protected]

Number of objectives.

FIXME: Shouldn't require this in the general Population, but only in some friend class that operates on the population?

---

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

• src/Population.hpp
• src/Population.cpp
• src/Population_Old.cpp