Fitness assignment schemes added
git-svn-id: svn://scm.gforge.inria.fr/svnroot/paradiseo@1620 331e1502-861f-0410-8da2-ba01fb791d7f
This commit is contained in:
jhumeau
parent
64346d7f8f
commit
69c4eb90c9
19 changed files with 1845 additions and 80 deletions
|
|
@ -0,0 +1,109 @@
|
|||
/*
|
||||
* <moeoAchievementScalarizingFunctionDistance.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2006-2007
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Jeremie Humeau
|
||||
* Arnaud Liefooghe
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOASFADIST_H_
|
||||
#define MOEOASFADIST_H_
|
||||
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/**
|
||||
Achievment scalarizing function aproach to calculate a distance
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoAchievementScalarizingFunctionDistance : public moeoObjSpaceDistance< MOEOT >
|
||||
{
|
||||
public:
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
constructor with a normalizer
|
||||
@param _rho
|
||||
@param _weight the weight to apply to each dimension
|
||||
@param _normalizer the normalizer
|
||||
*/
|
||||
moeoAchievementScalarizingFunctionDistance(unsigned int _rho, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer): normalizer(_normalizer), weight(_weight), rho(_rho){}
|
||||
|
||||
/**
|
||||
constructor without a normalizer
|
||||
@param _rho
|
||||
@param _weight the weight to apply to each dimension
|
||||
*/
|
||||
moeoAchievementScalarizingFunctionDistance(unsigned int _rho, const ObjectiveVector& _weight): normalizer(defaultNormalizer), weight(_weight), rho(_rho){}
|
||||
|
||||
/**
|
||||
fonction which calculate the distance
|
||||
@param _obj the point to evaluate
|
||||
@param _reference the reference to calculate the distance from
|
||||
@return the fitness conrresponding to the distance
|
||||
*/
|
||||
const Fitness operator()(const ObjectiveVector& _reference, const ObjectiveVector& _obj){
|
||||
unsigned int dim=_obj.size();
|
||||
ObjectiveVector tmp1(_reference);
|
||||
ObjectiveVector tmp2(_obj);
|
||||
Fitness max=iteration(tmp2,tmp1,0,_obj.minimizing(0));
|
||||
Fitness res=max;
|
||||
for (unsigned i=0;i<dim;i++){
|
||||
res=iteration(_obj,_reference,i,_obj.minimizing(i));
|
||||
if (max<res)
|
||||
max=res;
|
||||
}
|
||||
return max;
|
||||
}
|
||||
|
||||
private:
|
||||
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
const ObjectiveVector &weight;
|
||||
double rho;
|
||||
|
||||
Fitness iteration(const ObjectiveVector &obj, const ObjectiveVector& reference, int dim, bool mini){
|
||||
ObjectiveVector obj_norm=normalizer(obj);
|
||||
ObjectiveVector ref_norm=normalizer(reference);
|
||||
Fitness res;
|
||||
if (mini){
|
||||
res=(obj_norm[dim]-ref_norm[dim]);
|
||||
}else{
|
||||
res=(ref_norm[dim]-obj_norm[dim]);
|
||||
}
|
||||
res=weight[dim]*res;
|
||||
return res;
|
||||
}
|
||||
};
|
||||
#endif
|
||||
|
|
@ -0,0 +1,111 @@
|
|||
/*
|
||||
* <moeoAugmentedAchievmentScalarizingFunctionDistance.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2006-2007
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Jeremie Humeau
|
||||
* Arnaud Liefooghe
|
||||
* Legillon François
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOASFAORDIST_H_
|
||||
#define MOEOASFAORDIST_H_
|
||||
|
||||
|
||||
#include <eo>
|
||||
#include <moeo>
|
||||
#include <cmath>
|
||||
|
||||
/**
|
||||
Order representing Achievment scalarizing function aproach to calculate a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoAugmentedAchievementScalarizingFunctionDistance : public moeoObjSpaceDistance< MOEOT >
|
||||
{
|
||||
public:
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
/**
|
||||
constructor with a normalizer
|
||||
@param _rho
|
||||
@param _weight the weight to apply to each dimansion
|
||||
@param _normalizer the normalizer
|
||||
*/
|
||||
moeoAugmentedAchievementScalarizingFunctionDistance(unsigned int _rho,const ObjectiveVector &_weight,moeoObjectiveVectorNormalizer<MOEOT> &_normalizer): normalizer(_normalizer),weight(_weight),rho(_rho)
|
||||
{}
|
||||
/**
|
||||
constructor without a normalizer
|
||||
@param _rho
|
||||
@param _weight the weight to apply to each dimansion
|
||||
*/
|
||||
moeoAugmentedAchievementScalarizingFunctionDistance(unsigned int _rho,const ObjectiveVector &_weight): normalizer(defaultNormalizer),weight(_weight),rho(_rho)
|
||||
{}
|
||||
|
||||
|
||||
/**
|
||||
fonction which apply the metric to calculate a fitness
|
||||
@param _reference the reference point to calculate the distance
|
||||
@param _obj the point to evaluate
|
||||
@return the fitness conrresponding to the distance
|
||||
*/
|
||||
const Fitness operator()(const ObjectiveVector &_reference,const ObjectiveVector &_obj){
|
||||
unsigned int dim=_obj.size();
|
||||
Fitness res=0;
|
||||
Fitness max=iteration(_obj,_reference,0,_obj.minimizing(0));
|
||||
for (unsigned i=0;i<dim;i++){
|
||||
Fitness tmp=iteration(_obj,_reference,i,_obj.minimizing(i));
|
||||
if (max<tmp)
|
||||
max=tmp;
|
||||
res+=tmp;
|
||||
}
|
||||
return res+rho*max;
|
||||
}
|
||||
|
||||
private:
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
const ObjectiveVector &weight;
|
||||
double rho;
|
||||
Fitness iteration(const ObjectiveVector &obj,const ObjectiveVector &reference,int dim,bool mini){
|
||||
ObjectiveVector obj_norm=normalizer(obj);
|
||||
ObjectiveVector ref_norm=normalizer(reference);
|
||||
Fitness res;
|
||||
if (mini){
|
||||
res=obj_norm[dim]-ref_norm[dim];
|
||||
}else{
|
||||
|
||||
res=ref_norm[dim]-obj_norm[dim];
|
||||
}
|
||||
res=weight[dim]*res;
|
||||
return res;
|
||||
}
|
||||
};
|
||||
#endif
|
||||
|
|
@ -0,0 +1,102 @@
|
|||
/*
|
||||
* <moeoAugmentedWeightedChebychevDistance.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2006-2007
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* François Legillon
|
||||
* Jeremie Humeau
|
||||
* Arnaud Liefooghe
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOCHEBYCHEVORDIST_H_
|
||||
#define MOEOCHEBYCHEVORDIST_H_
|
||||
|
||||
#include <distance/moeoObjSpaceDistance.h>
|
||||
|
||||
/**
|
||||
order representing chebychev distance
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoAugmentedWeightedChebychevDistance : public moeoObjSpaceDistance < MOEOT >
|
||||
{
|
||||
public:
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
constructor with a normalizer
|
||||
@param _rho
|
||||
@param _weight the weight to apply to each dimension
|
||||
@param _normalizer the normalizer
|
||||
*/
|
||||
moeoAugmentedWeightedChebychevDistance(unsigned int _rho, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer): normalizer(_normalizer), weight(_weight), rho(_rho){}
|
||||
|
||||
/**
|
||||
constructor with a dummy normalizer
|
||||
@param _rho
|
||||
@param _weight the weight to apply to each dimension
|
||||
*/
|
||||
moeoAugmentedWeightedChebychevDistance(unsigned int _rho, const ObjectiveVector& _weight): normalizer(defaultNormalizer), weight(_weight), rho(_rho){}
|
||||
|
||||
/**
|
||||
fonction which calculate a fitness
|
||||
@param _reference the reference to calculate the distance from
|
||||
@param _obj the point to evaluate
|
||||
@return the fitness conrresponding to the distance
|
||||
*/
|
||||
const Fitness operator()(const ObjectiveVector& _reference, const ObjectiveVector& _obj){
|
||||
unsigned int dim=_obj.size();
|
||||
Fitness res=iteration(_obj,_reference,0);
|
||||
Fitness max=res*weight[0];
|
||||
for (unsigned i=1;i<dim;i++){
|
||||
Fitness tmp=iteration(_obj,_reference,i);
|
||||
if (tmp*weight[i]>max)
|
||||
max=tmp*weight[i];
|
||||
res+=tmp;
|
||||
}
|
||||
res=res*rho+max;
|
||||
return res;
|
||||
}
|
||||
|
||||
private:
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
const ObjectiveVector &weight;
|
||||
double rho;
|
||||
|
||||
Fitness iteration(const ObjectiveVector &obj,const ObjectiveVector &reference,int dim){
|
||||
ObjectiveVector obj_norm=normalizer(obj);
|
||||
ObjectiveVector ref_norm=normalizer(reference);
|
||||
Fitness res=abs(obj_norm[dim]-ref_norm[dim]);
|
||||
return res;
|
||||
}
|
||||
};
|
||||
#endif
|
||||
|
|
@ -4,6 +4,7 @@
|
|||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Arnaud Liefooghe
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
|
|
@ -39,52 +40,48 @@
|
|||
#define MOEOEUCLIDEANDISTANCE_H_
|
||||
|
||||
#include <math.h>
|
||||
#include <distance/moeoNormalizedDistance.h>
|
||||
#include <distance/moeoObjSpaceDistance.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/**
|
||||
* A class allowing to compute an euclidian distance between two solutions in the objective space with normalized objective values (i.e. between 0 and 1).
|
||||
* A distance value then lies between 0 and sqrt(nObjectives).
|
||||
*/
|
||||
template < class MOEOT >
|
||||
class moeoEuclideanDistance : public moeoNormalizedDistance < MOEOT >
|
||||
class moeoEuclideanDistance : public moeoObjSpaceDistance < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
|
||||
/** the fitness type of the solutions */
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
* Returns the euclidian distance between _moeo1 and _moeo2 in the objective space
|
||||
* @param _moeo1 the first solution
|
||||
* @param _moeo2 the second solution
|
||||
*/
|
||||
const double operator()(const MOEOT & _moeo1, const MOEOT & _moeo2)
|
||||
{
|
||||
double result = 0.0;
|
||||
double tmp1, tmp2;
|
||||
for (unsigned int i=0; i<ObjectiveVector::nObjectives(); i++)
|
||||
{
|
||||
tmp1 = (_moeo1.objectiveVector()[i] - bounds[i].minimum()) / bounds[i].range();
|
||||
tmp2 = (_moeo2.objectiveVector()[i] - bounds[i].minimum()) / bounds[i].range();
|
||||
result += (tmp1-tmp2) * (tmp1-tmp2);
|
||||
}
|
||||
return sqrt(result);
|
||||
}
|
||||
ctr with a normalizer
|
||||
@param _normalizer the normalizer used for every ObjectiveVector
|
||||
*/
|
||||
moeoEuclideanDistance (moeoObjectiveVectorNormalizer<MOEOT> _normalizer):normalizer(_normalizer)
|
||||
{}
|
||||
/**
|
||||
default ctr
|
||||
*/
|
||||
moeoEuclideanDistance ():normalizer(defaultNormalizer)
|
||||
{}
|
||||
|
||||
/**
|
||||
* Returns the euclidian distance between _obj1 and _obj2
|
||||
* Returns the euclidian distance between _obj1 and _obj2 in the objective space
|
||||
* @param _obj1 the first objective vector
|
||||
* @param _obj2 the second objective vector
|
||||
*/
|
||||
const double operator()(const ObjectiveVector & _obj1, const ObjectiveVector & _obj2)
|
||||
const Fitness operator()(const ObjectiveVector & _obj1, const ObjectiveVector & _obj2)
|
||||
{
|
||||
double result = 0.0;
|
||||
double tmp1, tmp2;
|
||||
Fitness result = 0.0;
|
||||
Fitness tmp1, tmp2;
|
||||
for (unsigned int i=0; i<ObjectiveVector::nObjectives(); i++)
|
||||
{
|
||||
tmp1 = (_obj1[i] - bounds[i].minimum()) / bounds[i].range();
|
||||
tmp2 = (_obj2[i] - bounds[i].minimum()) / bounds[i].range();
|
||||
tmp1 = normalizer(_obj1)[i];
|
||||
tmp2 = normalizer(_obj2)[i];
|
||||
result += (tmp1-tmp2) * (tmp1-tmp2);
|
||||
}
|
||||
return sqrt(result);
|
||||
|
|
@ -93,8 +90,8 @@ class moeoEuclideanDistance : public moeoNormalizedDistance < MOEOT >
|
|||
|
||||
private:
|
||||
|
||||
/** the bounds for every objective */
|
||||
using moeoNormalizedDistance < MOEOT > :: bounds;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
|
||||
};
|
||||
|
||||
|
|
|
|||
|
|
@ -4,6 +4,7 @@
|
|||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Arnaud Liefooghe
|
||||
* Francçois Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
|
|
@ -39,34 +40,48 @@
|
|||
#define MOEOMANHATTANDISTANCE_H_
|
||||
|
||||
#include <math.h>
|
||||
#include <distance/moeoNormalizedDistance.h>
|
||||
#include <distance/moeoObjSpaceDistance.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/**
|
||||
* A class allowing to compute the Manhattan distance between two solutions in the objective space normalized objective values (i.e. between 0 and 1).
|
||||
* A distance value then lies between 0 and nObjectives.
|
||||
*/
|
||||
template < class MOEOT >
|
||||
class moeoManhattanDistance : public moeoNormalizedDistance < MOEOT >
|
||||
class moeoManhattanDistance : public moeoObjSpaceDistance < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
|
||||
/** the fitness type of the solutions */
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
* Returns the Manhattan distance between _moeo1 and _moeo2 in the objective space
|
||||
* @param _moeo1 the first solution
|
||||
* @param _moeo2 the second solution
|
||||
ctr with a normalizer
|
||||
@param _normalizer the normalizer used for every ObjectiveVector
|
||||
*/
|
||||
moeoManhattanDistance (moeoObjectiveVectorNormalizer<MOEOT> &_normalizer):normalizer(_normalizer)
|
||||
{}
|
||||
/**
|
||||
default ctr
|
||||
*/
|
||||
moeoManhattanDistance ():normalizer(defaultNormalizer)
|
||||
{}
|
||||
|
||||
/**
|
||||
* Returns the Manhattan distance between _obj1 and _obj2 in the objective space
|
||||
* @param _obj1 the first objective vector
|
||||
* @param _obj2 the second objective vector
|
||||
*/
|
||||
const double operator()(const MOEOT & _moeo1, const MOEOT & _moeo2)
|
||||
const double operator()(const ObjectiveVector & _obj1, const ObjectiveVector & _obj2)
|
||||
{
|
||||
double result = 0.0;
|
||||
double tmp1, tmp2;
|
||||
for (unsigned int i=0; i<ObjectiveVector::nObjectives(); i++)
|
||||
{
|
||||
tmp1 = (_moeo1.objectiveVector()[i] - bounds[i].minimum()) / bounds[i].range();
|
||||
tmp2 = (_moeo2.objectiveVector()[i] - bounds[i].minimum()) / bounds[i].range();
|
||||
tmp1 = normalizer(_obj1)[i];
|
||||
tmp2 = normalizer(_obj2)[i];
|
||||
result += fabs(tmp1-tmp2);
|
||||
}
|
||||
return result;
|
||||
|
|
@ -75,8 +90,8 @@ class moeoManhattanDistance : public moeoNormalizedDistance < MOEOT >
|
|||
|
||||
private:
|
||||
|
||||
/** the bounds for every objective */
|
||||
using moeoNormalizedDistance < MOEOT > :: bounds;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
|
||||
};
|
||||
|
||||
|
|
|
|||
63
trunk/paradiseo-moeo/src/distance/moeoObjSpaceDistance.h
Normal file
63
trunk/paradiseo-moeo/src/distance/moeoObjSpaceDistance.h
Normal file
|
|
@ -0,0 +1,63 @@
|
|||
/*
|
||||
* <moeoObjSpaceDistance.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2006-2007
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Jeremie Humeau
|
||||
* Arnaud Liefooghe
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOOBJDIST_H_
|
||||
#define MOEOOBJDIST_H_
|
||||
|
||||
|
||||
#include <eo>
|
||||
#include <moeo>
|
||||
#include <cmath>
|
||||
|
||||
/**
|
||||
Distances using Objective vectors to evaluate
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoObjSpaceDistance : public moeoDistance < MOEOT, typename MOEOT::Fitness >, public eoBF<const typename MOEOT::ObjectiveVector&,const typename MOEOT::ObjectiveVector&,const typename MOEOT::Fitness>
|
||||
{
|
||||
public:
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
virtual const Fitness operator()(const MOEOT &_moeot1,const MOEOT &_moeot2){
|
||||
return operator()(_moeot1.objectiveVector(),_moeot2.objectiveVector());
|
||||
}
|
||||
|
||||
virtual const Fitness operator()(const ObjectiveVector &_obj1, const ObjectiveVector &_obj2)=0;
|
||||
};
|
||||
#endif
|
||||
|
|
@ -0,0 +1,101 @@
|
|||
/*
|
||||
* <moeoWeightedChebychevDistance.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2006-2007
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Jeremie Humeau
|
||||
* Arnaud Liefooghe
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOCHEBYCHEVDIST_H_
|
||||
#define MOEOCHEBYCHEVDIST_H_
|
||||
|
||||
|
||||
#include <eo>
|
||||
#include <moeo>
|
||||
#include <cmath>
|
||||
/**
|
||||
* weighted chebychev distance
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoWeightedChebychevDistance : public moeoObjSpaceDistance < MOEOT >
|
||||
{
|
||||
public:
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
* constructor with a normalizer
|
||||
* @param _rho
|
||||
* @param _weight the weight to apply to each dimansion
|
||||
* @param _normalizer the normalizer
|
||||
*/
|
||||
moeoWeightedChebychevDistance(unsigned int _rho, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer): normalizer(_normalizer), weight(_weight), rho(_rho){}
|
||||
|
||||
/**
|
||||
* constructor with a normalizer
|
||||
* @param _rho
|
||||
* @param _weight the weight to apply to each dimansion
|
||||
*/
|
||||
moeoWeightedChebychevDistance(unsigned int _rho, const ObjectiveVector& _weight): normalizer(defaultNormalizer), weight(_weight), rho(_rho){}
|
||||
|
||||
/**
|
||||
* fonction which apply the metric to calculate a fitness
|
||||
* @param _obj the point to evaluate
|
||||
* @param _reference the reference to calculate the distance from
|
||||
* @return the fitness conrresponding to the distance
|
||||
*/
|
||||
const Fitness operator()(const ObjectiveVector& _reference, const ObjectiveVector& _obj){
|
||||
unsigned int dim=_obj.size();
|
||||
Fitness res=0;
|
||||
ObjectiveVector obj_norm=normalizer(_obj);
|
||||
ObjectiveVector ref_norm=normalizer(_reference);
|
||||
for (unsigned i=0;i<dim;i++){
|
||||
res+=iteration(obj_norm,ref_norm,i);
|
||||
}
|
||||
return res;
|
||||
}
|
||||
|
||||
private:
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
const ObjectiveVector &weight;
|
||||
double rho;
|
||||
|
||||
Fitness iteration(const ObjectiveVector &obj,const ObjectiveVector &reference,int dim){
|
||||
Fitness res=abs(obj[dim]-reference[dim]);
|
||||
res=weight[dim]*pow(res,rho);
|
||||
res=pow(res,1/rho);
|
||||
return res;
|
||||
}
|
||||
};
|
||||
#endif
|
||||
|
|
@ -0,0 +1,146 @@
|
|||
/*
|
||||
* <moeoScalarizationFunctionMetricFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoScalarizationFunctionMetricFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOASFAFITNESSASSIGNMENT_H_
|
||||
#define MOEOASFAFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <eoPop.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <distance/moeoAchievementScalarizingFunctionDistance.h>
|
||||
#include <metric/moeoDistanceMetric.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which use a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoAchievementScalarizingFunctionMetricFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
typedef typename ObjectiveVector::Type Type;
|
||||
|
||||
/**
|
||||
* ctor with normalizer
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
*/
|
||||
moeoAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer) : normalizer(_normalizer), eval(defaultEval), distance(_rho, _weight), metric( distance, _reference, defaultNormalizer){}
|
||||
|
||||
/**
|
||||
* ctor with an evaluing fonction, applied if give moeot is invalid
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
*/
|
||||
moeoAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, ObjectiveVector& _reference, const ObjectiveVector& _weight, eoEvalFunc<MOEOT>& _eval): eval(_eval), normalizer(defaultNormalizer), distance(_rho, _weight), metric( distance, _reference, defaultNormalizer){}
|
||||
|
||||
/**
|
||||
* ctor with an evaluing fonction, applied if give moeot is invalid, and a noramlizer, applied to ObjectiveVectors
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
*/
|
||||
moeoAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer, eoEvalFunc<MOEOT>& _eval) : normalizer(_normalizer), eval(_eval), distance(_rho, _weight), metric(distance, _reference, _normalizer){}
|
||||
|
||||
/**
|
||||
default constructor
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight) : eval(defaultEval), normalizer(defaultNormalizer), distance(_rho, _weight), metric(distance, _reference, defaultNormalizer){}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for a moeot
|
||||
* @param _mo the MOEOT
|
||||
*/
|
||||
void operator()(MOEOT & _mo){
|
||||
if (_mo.invalidObjectiveVector())
|
||||
eval(_mo);
|
||||
_mo.fitness(operator()(_mo.objectiveVector()));
|
||||
}
|
||||
|
||||
/**
|
||||
return the fitness of a valid objectiveVector
|
||||
@param _mo the objectiveVector
|
||||
@return the fitness value of _mo
|
||||
*/
|
||||
typename MOEOT::Fitness operator()(const typename MOEOT::ObjectiveVector& _mo){
|
||||
return -metric(_mo);
|
||||
}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the populing _pop (and in the archive)
|
||||
* @param _pop the populing
|
||||
*/
|
||||
void operator()(eoPop < MOEOT > & _pop)
|
||||
{
|
||||
for (unsigned int k=0; k < _pop.size(); k++)
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
|
||||
/**
|
||||
* @param _pop the populing
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec){}
|
||||
|
||||
private:
|
||||
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoAchievementScalarizingFunctionDistance<MOEOT> distance;
|
||||
moeoDistanceMetric<MOEOT> metric;
|
||||
eoEvalFunc<MOEOT> &eval;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){
|
||||
}
|
||||
} defaultEval;
|
||||
};
|
||||
|
||||
#endif /*moeoAugmentedScalarizingFunctionMetricFitnessASSIGNMENT_H_*/
|
||||
|
|
@ -0,0 +1,129 @@
|
|||
/*
|
||||
* <moeoAggregationFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoAggregationFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOAGGREGATIONFITNESSASSIGNMENT_H_
|
||||
#define MOEOAGGREGATIONFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <eoPop.h>
|
||||
#include <eoEvalFunc.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which use weight foreach objectives
|
||||
*/
|
||||
template < class MOEOT >
|
||||
class moeoAggregationFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
* Default ctor
|
||||
* @param _weight vectors contains all weights.
|
||||
* @param _eval a eval function, to revalidate the objectiveVector if needed
|
||||
*/
|
||||
moeoAggregationFitnessAssignment(std::vector<double> & _weight,eoEvalFunc<MOEOT> &_eval) : weight(_weight),eval(_eval){}
|
||||
|
||||
/**
|
||||
* Ctor with a dummy evaluation function
|
||||
* @param _weight vectors contains all weights.
|
||||
*/
|
||||
moeoAggregationFitnessAssignment(std::vector<double> & _weight) : weight(_weight),eval(defaultEval){}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for _moeot
|
||||
* @param _moeot the MOEOT
|
||||
*/
|
||||
virtual void operator()(MOEOT & _moeot){
|
||||
if (_moeot.invalidObjectiveVector())
|
||||
eval(_moeot);
|
||||
_moeot.fitness(operator()(_moeot.objectiveVector()));
|
||||
}
|
||||
|
||||
/**
|
||||
* function which calculate the fitness from an objectiveVector (which has troi be valid.)
|
||||
* @param _mo an valid objectiveVector
|
||||
* @return the fitness value of _mo
|
||||
*/
|
||||
virtual Fitness operator()(const typename MOEOT::ObjectiveVector & _mo){
|
||||
unsigned int dim=_mo.nObjectives();
|
||||
Fitness res=0;
|
||||
if (dim>weight.size()){
|
||||
std::cout<<"moeoAggregationFitnessAssignmentFitness: Error -> given weight dimension is smaller than MOEOTs"<<std::endl;
|
||||
return res;
|
||||
}
|
||||
for(unsigned int l=0; l<dim; l++){
|
||||
if (_mo.minimizing(l))
|
||||
res-=(_mo[l]) * weight[l];
|
||||
else
|
||||
res+=(_mo[l]) * weight[l];
|
||||
}
|
||||
return res;
|
||||
}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the population _pop (and in the archive)
|
||||
* @param _pop the population
|
||||
*/
|
||||
virtual void operator()(eoPop < MOEOT > & _pop){
|
||||
for (unsigned int k=0; k < _pop.size(); k++)
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
|
||||
/**
|
||||
* Warning: no yet implemented: Updates the fitness values of the whole population _pop by taking the deletion of the objective vector _objVec into account.
|
||||
* @param _pop the population
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec){}
|
||||
|
||||
private:
|
||||
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){}
|
||||
}defaultEval;
|
||||
|
||||
//the vector of weight
|
||||
std::vector<double>& weight;
|
||||
eoEvalFunc<MOEOT>& eval;
|
||||
|
||||
};
|
||||
#endif /*MOEOAGGREGATIONFITNESSASSIGNMENT_H_*/
|
||||
|
|
@ -0,0 +1,160 @@
|
|||
/*
|
||||
* <moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoAugmentedScalarizingFunctionMetricFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOAUGASFAFITNESSASSIGNMENT_H_
|
||||
#define MOEOAUGASFAFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <moeo>
|
||||
#include <eo>
|
||||
#include <vector>
|
||||
#include <eoPop.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
#include <distance/moeoAugmentedAchievementScalarizingFunctionDistance.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which use a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
typedef typename ObjectiveVector::Type Type;
|
||||
|
||||
|
||||
/**
|
||||
* ctor with normalizer
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho,const ObjectiveVector &_reference,const ObjectiveVector &_weight,moeoObjectiveVectorNormalizer<MOEOT> &_normalizer) : normalizer(_normalizer),eval(defaultEval), distance(_rho,_weight), metric(distance,_reference,normalizer)
|
||||
{}
|
||||
/**
|
||||
* ctor with an evaluing fonction, applied if give moeot is invalid
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector &_reference,const ObjectiveVector &_weight,eoEvalFunc<MOEOT> &_eval) : eval(_eval),normalizer(defaultNormalizer), distance(_rho,_weight), metric(distance,_reference,normalizer)
|
||||
{}
|
||||
/**
|
||||
* ctor with an evaluing fonction, applied if give moeot is invalid, and a noramlizer, applied to ObjectiveVectors
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector &_reference,const ObjectiveVector &_weight,moeoObjectiveVectorNormalizer<MOEOT> &_normalizer,eoEvalFunc<MOEOT> &_eval) :normalizer(_normalizer),eval(_eval), distance(_rho,_weight), metric(distance,_reference,normalizer)
|
||||
{}
|
||||
/**
|
||||
default constructor
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector &_reference,const ObjectiveVector &_weight) : eval(defaultEval),normalizer(defaultNormalizer), distance(_rho,_weight), metric(distance,_reference,normalizer)
|
||||
{}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for a moeot
|
||||
* @param _mo the MOEOT
|
||||
*/
|
||||
void operator()(MOEOT & _mo){
|
||||
if (_mo.invalidObjectiveVector()) eval(_mo);
|
||||
_mo.fitness(operator()(_mo.objectiveVector()));
|
||||
|
||||
}
|
||||
|
||||
/**
|
||||
return the fitness of a valid objectiveVector
|
||||
@param _mo the objectiveVector
|
||||
@return the fitness value of _mo
|
||||
*/
|
||||
typename MOEOT::Fitness operator()(const typename MOEOT::ObjectiveVector & _mo){
|
||||
return -metric(_mo);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the populing _pop (and in the archive)
|
||||
* @param _pop the populing
|
||||
*/
|
||||
void operator()(eoPop < MOEOT > & _pop)
|
||||
{
|
||||
unsigned int pop_size= _pop.size();
|
||||
for (unsigned int k=0; k<pop_size; k++){
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
||||
|
||||
/**
|
||||
* @param _pop the populing
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec)
|
||||
{
|
||||
}
|
||||
|
||||
|
||||
private:
|
||||
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoAugmentedAchievementScalarizingFunctionDistance<MOEOT> distance;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
moeoDistanceMetric<MOEOT> metric;
|
||||
eoEvalFunc<MOEOT> &eval;
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){
|
||||
}
|
||||
} defaultEval;
|
||||
|
||||
};
|
||||
|
||||
#endif /*moeoAugmentedScalarizingFunctionMetricFitnessASSIGNMENT_H_*/
|
||||
|
|
@ -0,0 +1,144 @@
|
|||
/*
|
||||
* <moeoAugmentedWeightedChebychevMetricFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoAugmentedWeightedChebychevMetricFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOAUGCHEVMETRICFITNESSASSIGNMENT_H_
|
||||
#define MOEOAUGCHEVMETRICFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <eoPop.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which use a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoAugmentedWeightedChebychevMetricFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
typedef typename ObjectiveVector::Type Type;
|
||||
|
||||
/**
|
||||
* ctor with normalizer
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
*/
|
||||
moeoAugmentedWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer) : normalizer(_normalizer), eval(defaultEval), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
* ctor with an evaluation fonction, applied if give moeot is invalid
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
*/
|
||||
moeoAugmentedWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, eoEvalFunc<MOEOT>& _eval) : eval(_eval), normalizer(defaultNormalizer), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
* ctor with an evaluation fonction, applied if give moeot is invalid, and a noramlizer, applied to ObjectiveVectors
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
*/
|
||||
moeoAugmentedWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer, eoEvalFunc<MOEOT>& _eval) : normalizer(_normalizer), eval(_eval), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
default constructor
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoAugmentedWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight) : eval(defaultEval), normalizer(defaultNormalizer), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for a moeot
|
||||
* @param _mo the MOEOT
|
||||
*/
|
||||
void operator()(MOEOT & _mo){
|
||||
if (_mo.invalidObjectiveVector())
|
||||
eval(_mo);
|
||||
_mo.fitness(operator()(_mo.objectiveVector()));
|
||||
}
|
||||
|
||||
/**
|
||||
return the fitness of a valid objectiveVector
|
||||
@param _mo the objectiveVector
|
||||
@return the fitness value of _mo
|
||||
*/
|
||||
typename MOEOT::Fitness operator()(const typename MOEOT::ObjectiveVector & _mo){
|
||||
return -metric(_mo);
|
||||
}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the population _pop (and in the archive)
|
||||
* @param _pop the population
|
||||
*/
|
||||
void operator()(eoPop < MOEOT > & _pop){
|
||||
for (unsigned int k=0; k<_pop.size(); k++)
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
|
||||
/**
|
||||
* @param _pop the population
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec){}
|
||||
|
||||
private:
|
||||
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){
|
||||
}
|
||||
} defaultEval;
|
||||
|
||||
moeoAugmentedWeightedChebychevDistance<MOEOT> distance;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> &normalizer;
|
||||
moeoDistanceMetric<MOEOT> metric;
|
||||
eoEvalFunc<MOEOT> &eval;
|
||||
|
||||
};
|
||||
|
||||
#endif /*moeoAugmentedWeightedChebychevMetricFitnessASSIGNMENT_H_*/
|
||||
|
|
@ -0,0 +1,171 @@
|
|||
/*
|
||||
* <moeoConstraintFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoConstraintFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOCONSTRAINTFITNESSASSIGNMENT_H_
|
||||
#define MOEOCONSTRAINTFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <eoPop.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which give a penalty if MOEOT does not respect constraints
|
||||
*/
|
||||
template < class MOEOT >
|
||||
class moeoConstraintFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
/** the fitness type of the solutions */
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
/** the type of the solutions */
|
||||
typedef typename ObjectiveVector::Type Type;
|
||||
|
||||
/**
|
||||
* Default ctor
|
||||
* @param _weight vectors contains all weights to apply for not respecting the contraint in each dimension.
|
||||
* @param _constraint vector containing the constraints, normalizer is applied to it
|
||||
* @param _to_optimize dimension in which we ignore the constraint
|
||||
* @param _normalizer normalizer to apply to each objective
|
||||
*/
|
||||
moeoConstraintFitnessAssignment(std::vector<double> & _weight, ObjectiveVector &_constraint, int _to_optimize, moeoObjectiveVectorNormalizer<MOEOT> &_normalizer, eoEvalFunc<MOEOT> &_eval) : weight(_weight),constraint(_constraint),to_optimize(_to_optimize),normalizer(_normalizer),eval(_eval),to_eval(true){}
|
||||
|
||||
/**
|
||||
* Ctor with a dummy eval
|
||||
* @param _weight vectors contains all weights to apply for not respecting the contraint in each dimension.
|
||||
* @param _constraint vector containing the constraints, normalizer is applied to it
|
||||
* @param _to_optimize dimension in which we ignore the constraint
|
||||
* @param _normalizer normalizer to apply to each objective
|
||||
*/
|
||||
moeoConstraintFitnessAssignment(std::vector<double> & _weight, ObjectiveVector &_constraint, int _to_optimize, moeoObjectiveVectorNormalizer<MOEOT> &_normalizer) : weight(_weight), constraint(_constraint), to_optimize(_to_optimize), normalizer(_normalizer), eval(defaultEval), to_eval(false){}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the population _pop (and in the archive)
|
||||
* @param _mo the MOEOT
|
||||
*/
|
||||
void operator()(MOEOT & _mo){
|
||||
if (to_eval && _mo.invalidObjectiveVector())
|
||||
eval(_mo);
|
||||
_mo.fitness(operator()(_mo.objectiveVector()));
|
||||
}
|
||||
|
||||
/**
|
||||
* Calculate a fitness from a valid objectiveVector
|
||||
* @param _mo a valid objectiveVector
|
||||
* @return the fitness of _mo
|
||||
*/
|
||||
Fitness operator()(const typename MOEOT::ObjectiveVector & _mo){
|
||||
unsigned int dim=_mo.nObjectives();
|
||||
Fitness res=0;
|
||||
if (dim>weight.size()){
|
||||
std::cout<<"moeoAggregationFitnessAssignmentFitness: ouch, given weight dimension is smaller than MOEOTs"<<std::endl;
|
||||
}
|
||||
else{
|
||||
for(unsigned int l=0; l<dim; l++){
|
||||
if (l==to_optimize)
|
||||
if (_mo.minimizing(l))
|
||||
res-=(normalizer(_mo)[l]) * weight[l];
|
||||
else
|
||||
res+=(normalizer(_mo)[l]) * weight[l];
|
||||
else{
|
||||
if(_mo.minimizing(l)){
|
||||
if (normalizer(_mo)[l]>normalizer(constraint)[l])
|
||||
res-=(normalizer(_mo)[l]-normalizer(constraint)[l])*weight[l];
|
||||
}
|
||||
else{
|
||||
if (normalizer(_mo)[l]<normalizer(constraint)[l])
|
||||
//negative so we add it instead of removing it
|
||||
res+=(normalizer(_mo)[l]-normalizer(constraint)[l])*weight[l];
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
return res;
|
||||
}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the population _pop (and in the archive)
|
||||
* @param _pop the population
|
||||
*/
|
||||
void operator()(eoPop < MOEOT > & _pop)
|
||||
{
|
||||
for(unsigned int k=0; k<_pop.size(); k++)
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
|
||||
/**
|
||||
* Warning: no yet implemented: Updates the fitness values of the whole population _pop by taking the deletion of the objective vector _objVec into account.
|
||||
* @param _pop the population
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec)
|
||||
{
|
||||
//std::cout << "WARNING : updateByDeleting not implemented in moeoAssignmentFitnessAssignment" << std::endl;
|
||||
}
|
||||
|
||||
private:
|
||||
|
||||
//dummy evaluation function
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){
|
||||
}
|
||||
} defaultEval;
|
||||
|
||||
//the vector of weight
|
||||
std::vector<double> weight;
|
||||
|
||||
//the vector of constraints
|
||||
ObjectiveVector constraint;
|
||||
|
||||
//index of the objective to optimize
|
||||
int to_optimize;
|
||||
|
||||
//the normalizer
|
||||
moeoObjectiveVectorNormalizer<MOEOT>& normalizer;
|
||||
|
||||
//the evaluation function
|
||||
eoEvalFunc<MOEOT> &eval;
|
||||
|
||||
//true if the evaluation has to be done
|
||||
bool to_eval;
|
||||
|
||||
};
|
||||
|
||||
#endif /*MOEOAGGREGATIONFITNESSASSIGNMENT_H_*/
|
||||
140
trunk/paradiseo-moeo/src/fitness/moeoMetricFitnessAssignment.h
Normal file
140
trunk/paradiseo-moeo/src/fitness/moeoMetricFitnessAssignment.h
Normal file
|
|
@ -0,0 +1,140 @@
|
|||
/*
|
||||
* <moeoMetricFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoMetrucFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOMETRICFITNESSASSIGNMENT_H_
|
||||
#define MOEOMETRICFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <moeo>
|
||||
#include <eo>
|
||||
#include <vector>
|
||||
#include <eoPop.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which use a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoMetricFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
typedef typename ObjectiveVector::Type Type;
|
||||
|
||||
|
||||
/**
|
||||
* Default ctor
|
||||
* @param _metric a metric function to calculate fitness
|
||||
* @param _negate true if fitness should be maxed
|
||||
*/
|
||||
moeoMetricFitnessAssignment(moeoUnaryMetric<ObjectiveVector,Fitness> &_metric, bool _negate=true) : metric(_metric),eval(defaultEval),negate(_negate)
|
||||
{}
|
||||
/**
|
||||
* ctor with an evaluation fonction, applied if give moeot is invalid
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
* @param _metric a metric function to calculate fitness
|
||||
* @param _negate true if fitness should be maxed
|
||||
*/
|
||||
moeoMetricFitnessAssignment(moeoUnaryMetric<ObjectiveVector,Fitness> &_metric,eoEvalFunc<MOEOT> &_eval, bool _negate=false) : metric(_metric),eval(_eval),negate(_negate)
|
||||
{}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for a moeot
|
||||
* @param _mo the MOEOT
|
||||
*/
|
||||
void operator()(MOEOT & _mo){
|
||||
if (_mo.invalidObjectiveVector()) eval(_mo);
|
||||
_mo.fitness(operator()(_mo.objectiveVector()));
|
||||
|
||||
}
|
||||
|
||||
/**
|
||||
return the fitness of a valid objectiveVector
|
||||
@param _mo the objectiveVector
|
||||
@return the fitness value of _mo
|
||||
*/
|
||||
typename MOEOT::Fitness operator()(const typename MOEOT::ObjectiveVector & _mo){
|
||||
if (negate)
|
||||
return 1/metric(_mo);
|
||||
else return metric(_mo);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the population _pop (and in the archive)
|
||||
* @param _pop the population
|
||||
*/
|
||||
void operator()(eoPop < MOEOT > & _pop)
|
||||
{
|
||||
unsigned int pop_size= _pop.size();
|
||||
for (unsigned int k=0; k<pop_size; k++){
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
||||
|
||||
/**
|
||||
* Warning: no yet implemented: Updates the fitness values of the whole population _pop by taking the deletion of the objective vector _objVec into account.
|
||||
* @param _pop the population
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec)
|
||||
{
|
||||
//std::cout << "WARNING : updateByDeleting not implemented in moeoAssignmentFitnessAssignment" << std::endl;
|
||||
}
|
||||
|
||||
|
||||
private:
|
||||
|
||||
moeoUnaryMetric<typename MOEOT::ObjectiveVector,Fitness> &metric;
|
||||
eoEvalFunc<MOEOT> &eval;
|
||||
bool negate;
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){
|
||||
}
|
||||
} defaultEval;
|
||||
|
||||
};
|
||||
|
||||
#endif /*MOEOMETRICFITNESSASSIGNMENT_H_*/
|
||||
67
trunk/paradiseo-moeo/src/fitness/moeoSingleObjectivization.h
Normal file
67
trunk/paradiseo-moeo/src/fitness/moeoSingleObjectivization.h
Normal file
|
|
@ -0,0 +1,67 @@
|
|||
/*
|
||||
* <moeoSingleObjectivization.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2009
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
|
||||
#ifndef MOEOSINGLEOBJECTIVIZATION_H_
|
||||
#define MOEOSINGLEOBJECTIVIZATION_H_
|
||||
|
||||
#include <fitness/moeoFitnessAssignment.h>
|
||||
#include <eoEvalFunc.h>
|
||||
/**
|
||||
* Class to adapt multiobjective problems to single objective algorithms
|
||||
*/
|
||||
template < class MOEOT >
|
||||
class moeoSingleObjectivization : public moeoFitnessAssignment < MOEOT > , public eoEvalFunc < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/**
|
||||
* herited from moeoFitnessAssignment
|
||||
* @param _pop the population
|
||||
*/
|
||||
virtual void operator () (eoPop < MOEOT > & _pop)=0;
|
||||
|
||||
/**
|
||||
herited from eoEvalFunc
|
||||
@param _moeot
|
||||
*/
|
||||
virtual void operator() (MOEOT & _moeot)=0;
|
||||
|
||||
virtual typename MOEOT::Fitness operator() (const typename MOEOT::ObjectiveVector & _obj)=0;
|
||||
|
||||
};
|
||||
|
||||
#endif /*MOEOSINGLEOBJECTIVIZATION_H_*/
|
||||
|
|
@ -0,0 +1,149 @@
|
|||
/*
|
||||
* <moeoWeightedChebychevMetricFitnessAssignment.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Lille-Nord Europe, 2006-2008
|
||||
* (C) OPAC Team, LIFL, 2002-2008
|
||||
*
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
// moeoWeightedChebychevMetricFitnessAssignment.h
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEOCHEVMETRICFITNESSASSIGNMENT_H_
|
||||
#define MOEOCHEVMETRICFITNESSASSIGNMENT_H_
|
||||
|
||||
#include <moeo>
|
||||
#include <eo>
|
||||
#include <vector>
|
||||
#include <eoPop.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <utils/moeoObjectiveVectorNormalizer.h>
|
||||
#include <distance/moeoWeightedChebychevDistance.h>
|
||||
|
||||
/*
|
||||
* Fitness assignment scheme which use a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoWeightedChebychevMetricFitnessAssignment : public moeoSingleObjectivization < MOEOT >
|
||||
{
|
||||
public:
|
||||
|
||||
/** the objective vector type of the solutions */
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
typedef typename ObjectiveVector::Type Type;
|
||||
|
||||
/**
|
||||
* ctor with normalizer
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
*/
|
||||
moeoWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer) : normalizer(_normalizer), eval(defaultEval), distance(_rho, _weight), metric(distance, _reference, _normalizer){}
|
||||
|
||||
/**
|
||||
* ctor with an evaluation fonction, applied if give moeot is invalid
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
*/
|
||||
moeoWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, eoEvalFunc<MOEOT>& _eval) : eval(_eval), normalizer(defaultNormalizer), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
* ctor with an evaluation fonction, applied if give moeot is invalid, and a noramlizer, applied to ObjectiveVectors
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
* @param _normalizer the normalizer to apply to objectiveVectors
|
||||
* @param _eval a evalFunc to regenerate the objectiveVector if needed
|
||||
*/
|
||||
moeoWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight, moeoObjectiveVectorNormalizer<MOEOT>& _normalizer, eoEvalFunc<MOEOT>& _eval) : normalizer(_normalizer), eval(_eval), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
default constructor
|
||||
* @param _rho
|
||||
* @param _reference the reference point
|
||||
* @param _weight the weights applied to the objectives
|
||||
*/
|
||||
moeoWeightedChebychevMetricFitnessAssignment(unsigned int _rho, const ObjectiveVector& _reference, const ObjectiveVector& _weight) : eval(defaultEval), normalizer(defaultNormalizer), distance(_rho, _weight), metric(distance, _reference, normalizer){}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for a moeot
|
||||
* @param _mo the MOEOT
|
||||
*/
|
||||
void operator()(MOEOT & _mo){
|
||||
if (_mo.invalidObjectiveVector())
|
||||
eval(_mo);
|
||||
_mo.fitness(operator()(_mo.objectiveVector()));
|
||||
}
|
||||
|
||||
/**
|
||||
return the fitness of a valid objectiveVector
|
||||
@param _mo the objectiveVector
|
||||
@return the fitness value of _mo
|
||||
*/
|
||||
typename MOEOT::Fitness operator()(const typename MOEOT::ObjectiveVector & _mo){
|
||||
return -metric(_mo);
|
||||
}
|
||||
|
||||
/**
|
||||
* Sets the fitness values for every solution contained in the population _pop (and in the archive)
|
||||
* @param _pop the population
|
||||
*/
|
||||
void operator()(eoPop < MOEOT > & _pop)
|
||||
{
|
||||
for (unsigned int k=0; k < _pop.size(); k++)
|
||||
operator()(_pop[k]);
|
||||
}
|
||||
|
||||
/**
|
||||
* @param _pop the population
|
||||
* @param _objVec the objective vector
|
||||
*/
|
||||
void updateByDeleting(eoPop < MOEOT > & _pop, ObjectiveVector & _objVec){}
|
||||
|
||||
private:
|
||||
|
||||
moeoWeightedChebychevDistance<MOEOT> distance;
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT>& normalizer;
|
||||
moeoDistanceMetric<MOEOT> metric;
|
||||
eoEvalFunc<MOEOT> &eval;
|
||||
|
||||
class DummyEval: public eoEvalFunc<MOEOT>{
|
||||
void operator()(MOEOT &moeo){
|
||||
}
|
||||
} defaultEval;
|
||||
|
||||
};
|
||||
|
||||
#endif /*moeoWeightedChebychevMetricFitnessASSIGNMENT_H_*/
|
||||
86
trunk/paradiseo-moeo/src/metric/moeoDistanceMetric.h
Normal file
86
trunk/paradiseo-moeo/src/metric/moeoDistanceMetric.h
Normal file
|
|
@ -0,0 +1,86 @@
|
|||
/*
|
||||
* <moeoDistanceMetric.h>
|
||||
* Copyright (C) DOLPHIN Project-Team, INRIA Futurs, 2006-2007
|
||||
* (C) OPAC Team, LIFL, 2002-2007
|
||||
*
|
||||
* Jeremie Humeau
|
||||
* Arnaud Liefooghe
|
||||
* François Legillon
|
||||
*
|
||||
* This software is governed by the CeCILL license under French law and
|
||||
* abiding by the rules of distribution of free software. You can use,
|
||||
* modify and/ or redistribute the software under the terms of the CeCILL
|
||||
* license as circulated by CEA, CNRS and INRIA at the following URL
|
||||
* "http://www.cecill.info".
|
||||
*
|
||||
* As a counterpart to the access to the source code and rights to copy,
|
||||
* modify and redistribute granted by the license, users are provided only
|
||||
* with a limited warranty and the software's author, the holder of the
|
||||
* economic rights, and the successive licensors have only limited liability.
|
||||
*
|
||||
* In this respect, the user's attention is drawn to the risks associated
|
||||
* with loading, using, modifying and/or developing or reproducing the
|
||||
* software by the user in light of its specific status of free software,
|
||||
* that may mean that it is complicated to manipulate, and that also
|
||||
* therefore means that it is reserved for developers and experienced
|
||||
* professionals having in-depth computer knowledge. Users are therefore
|
||||
* encouraged to load and test the software's suitability as regards their
|
||||
* requirements in conditions enabling the security of their systems and/or
|
||||
* data to be ensured and, more generally, to use and operate it in the
|
||||
* same conditions as regards security.
|
||||
* The fact that you are presently reading this means that you have had
|
||||
* knowledge of the CeCILL license and that you accept its terms.
|
||||
*
|
||||
* ParadisEO WebSite : http://paradiseo.gforge.inria.fr
|
||||
* Contact: paradiseo-help@lists.gforge.inria.fr
|
||||
*
|
||||
*/
|
||||
//-----------------------------------------------------------------------------
|
||||
#ifndef MOEODISTANCEMETRIC_H_
|
||||
#define MOEODISTANCEMETRIC_H_
|
||||
|
||||
#include <cmath>
|
||||
#include <distance/moeoObjSpaceDistance.h>
|
||||
|
||||
/**
|
||||
Adapter to use Distances as a metric
|
||||
*/
|
||||
template < class MOEOT>
|
||||
class moeoDistanceMetric : public moeoUnaryMetric < typename MOEOT::ObjectiveVector , typename MOEOT::Fitness >
|
||||
{
|
||||
public:
|
||||
typedef typename MOEOT::ObjectiveVector ObjectiveVector;
|
||||
typedef typename MOEOT::Fitness Fitness;
|
||||
|
||||
/**
|
||||
constructor with a normalizer
|
||||
@param _distance the distance
|
||||
@param _referencePoint the point from which we evaluate the distance
|
||||
@param _normalizer the normalizer
|
||||
*/
|
||||
moeoDistanceMetric(moeoObjSpaceDistance<MOEOT> &_distance, const ObjectiveVector &_referencePoint,moeoObjectiveVectorNormalizer<MOEOT>& _normalizer): distance(_distance), reference(_referencePoint),normalizer(_normalizer){}
|
||||
|
||||
/**
|
||||
constructor with a dummy normalizer
|
||||
@param _distance the distance
|
||||
@param _referencePoint the point from which we evaluate the distance
|
||||
*/
|
||||
moeoDistanceMetric(moeoObjSpaceDistance<MOEOT> &_distance, const ObjectiveVector &_referencePoint): distance(_distance), reference(_referencePoint),normalizer(defaultNormalizer){}
|
||||
|
||||
/**
|
||||
fonction which apply the metric to calculate a fitness
|
||||
@param _obj the point to evaluate
|
||||
@return the fitness conrresponding to the distance
|
||||
*/
|
||||
Fitness operator()(ObjectiveVector _obj){
|
||||
return distance(normalizer(reference), normalizer(_obj));
|
||||
}
|
||||
|
||||
private:
|
||||
moeoObjectiveVectorNormalizer<MOEOT> defaultNormalizer;
|
||||
moeoObjectiveVectorNormalizer<MOEOT>& normalizer;
|
||||
moeoObjSpaceDistance<MOEOT>& distance;
|
||||
const ObjectiveVector& reference;
|
||||
|
||||
};
|
||||
#endif
|
||||
|
|
@ -61,7 +61,7 @@
|
|||
#include <archive/moeoUnboundedArchive.h>
|
||||
#include <archive/moeoImprOnlyBoundedArchive.h>
|
||||
#include <archive/moeoFitDivBoundedArchive.h>
|
||||
#include <archive/moeoEpsilonHyperboxArchive.h>
|
||||
//#include <archive/moeoEpsilonHyperboxArchive.h>
|
||||
#include <archive/moeoQuadTreeArchive.h>
|
||||
|
||||
|
||||
|
|
@ -91,9 +91,14 @@
|
|||
|
||||
#include <distance/moeoDistance.h>
|
||||
#include <distance/moeoDistanceMatrix.h>
|
||||
#include <distance/moeoObjSpaceDistance.h>
|
||||
#include <distance/moeoEuclideanDistance.h>
|
||||
#include <distance/moeoManhattanDistance.h>
|
||||
#include <distance/moeoNormalizedDistance.h>
|
||||
#include <distance/moeoAchievementScalarizingFunctionDistance.h>
|
||||
#include <distance/moeoAugmentedAchievementScalarizingFunctionDistance.h>
|
||||
#include <distance/moeoAugmentedWeightedChebychevDistance.h>
|
||||
#include <distance/moeoWeightedChebychevDistance.h>
|
||||
|
||||
#include <diversity/moeoCrowdingDiversityAssignment.h>
|
||||
#include <diversity/moeoDiversityAssignment.h>
|
||||
|
|
@ -104,8 +109,10 @@
|
|||
#include <diversity/moeoSharingDiversityAssignment.h>
|
||||
|
||||
#include <fitness/moeoAchievementFitnessAssignment.h>
|
||||
#include <fitness/moeoAggregationFitnessAssignment.h>
|
||||
#include <fitness/moeoAggregativeFitnessAssignment.h>
|
||||
#include <fitness/moeoBinaryIndicatorBasedFitnessAssignment.h>
|
||||
#include <fitness/moeoConstraintFitnessAssignment.h>
|
||||
#include <fitness/moeoCriterionBasedFitnessAssignment.h>
|
||||
#include <fitness/moeoDominanceBasedFitnessAssignment.h>
|
||||
#include <fitness/moeoDominanceCountFitnessAssignment.h>
|
||||
|
|
@ -117,7 +124,13 @@
|
|||
#include <fitness/moeoFitnessAssignment.h>
|
||||
#include <fitness/moeoIndicatorBasedFitnessAssignment.h>
|
||||
#include <fitness/moeoScalarFitnessAssignment.h>
|
||||
#include <fitness/moeoSingleObjectivization.h>
|
||||
#include <fitness/moeoUnaryIndicatorBasedFitnessAssignment.h>
|
||||
#include <fitness/moeoAchievementScalarizingFunctionMetricFitnessAssignment.h>
|
||||
#include <fitness/moeoAugmentedAchievementScalarizingFunctionMetricFitnessAssignment.h>
|
||||
#include <fitness/moeoWeightedChebychevMetricFitnessAssignment.h>
|
||||
#include <fitness/moeoAugmentedWeightedChebychevMetricFitnessAssignment.h>
|
||||
|
||||
|
||||
#include <metric/moeoAdditiveEpsilonBinaryMetric.h>
|
||||
#include <metric/moeoContributionMetric.h>
|
||||
|
|
@ -130,6 +143,7 @@
|
|||
#include <metric/moeoVecVsVecEpsilonBinaryMetric.h>
|
||||
#include <metric/moeoVecVsVecAdditiveEpsilonBinaryMetric.h>
|
||||
#include <metric/moeoVecVsVecMultiplicativeEpsilonBinaryMetric.h>
|
||||
#include <metric/moeoDistanceMetric.h>
|
||||
|
||||
#include <replacement/moeoElitistReplacement.h>
|
||||
#include <replacement/moeoEnvironmentalReplacement.h>
|
||||
|
|
|
|||
|
|
@ -55,9 +55,10 @@ class moeoObjectiveVectorNormalizer
|
|||
|
||||
/**
|
||||
constructor with a supplied scale, usefull if you tweak your scale
|
||||
@param scale the scale for noramlzation
|
||||
@param _scale the scale for noramlzation
|
||||
@param max_param the returned values will be between 0 and max
|
||||
*/
|
||||
moeoObjectiveVectorNormalizer(Scale &_scale,Type max_param=100):scale(_scale),max(max_param)
|
||||
moeoObjectiveVectorNormalizer(Scale _scale=make_dummy_scale(),Type max_param=100):scale(_scale),max(max_param)
|
||||
{}
|
||||
/**
|
||||
constructor to create a normalizer from a given population
|
||||
|
|
@ -68,15 +69,17 @@ class moeoObjectiveVectorNormalizer
|
|||
{}
|
||||
/**
|
||||
constructor to create a normalizer with given boundaries
|
||||
@param boundaries the supplied vectors should have their values between thos boundaries
|
||||
@param _boundaries the supplied vectors should have their values between thos boundaries
|
||||
@param max_param the returned values will be between 0 and max
|
||||
**/
|
||||
moeoObjectiveVectorNormalizer(std::vector<Bounds> &_boundaries, Type max_param=100):scale(make_scale_from_bounds(_boundaries,max_param)), max(max_param)
|
||||
{}
|
||||
/**
|
||||
constructor to create a normalizer from bounds
|
||||
@param bounds the supplied vectors should have their value in those bounds
|
||||
@param _bounds the supplied vectors should have their value in those bounds
|
||||
@param max_param the returned values will be between 0 and max
|
||||
**/
|
||||
moeoObjectiveVectorNormalizer(Bounds &_bounds, Type max_param=100 ):scale(make_scale_from_bounds(_bounds,ObjectiveVector::nObjectives(),max_param)), max(max_param)
|
||||
moeoObjectiveVectorNormalizer(Bounds &_bounds, Type max_param=100 ):scale(make_scale_from_bounds(_bounds,max_param)), max(max_param)
|
||||
{}
|
||||
/**
|
||||
constructor to create a normalizer from a worst vector and a best vector
|
||||
|
|
@ -141,11 +144,11 @@ class moeoObjectiveVectorNormalizer
|
|||
|
||||
/**
|
||||
create a scale from bounds
|
||||
@param boundaries the boundaries
|
||||
@param _boundaries the boundaries
|
||||
@param max the maximum for returned values
|
||||
@return a scale
|
||||
*/
|
||||
static Scale make_scale_from_bounds(std::vector<Bounds> &_boundaries,Type max=100){
|
||||
static Scale make_scale_from_bounds(const std::vector<Bounds> &_boundaries,Type max=100){
|
||||
Scale res;
|
||||
for (unsigned i=0;i<_boundaries.size();i++){
|
||||
std::vector<Type> coeff;
|
||||
|
|
@ -162,8 +165,9 @@ class moeoObjectiveVectorNormalizer
|
|||
@param max the maximum for returned values
|
||||
@return a scale
|
||||
*/
|
||||
static Scale make_scale_from_bounds(Bounds &bounds,int dim,Type max=100){
|
||||
static Scale make_scale_from_bounds(const Bounds &bounds,Type max=100){
|
||||
Scale res;
|
||||
unsigned int dim=MOEOT::ObjectiveVector::nObjectives();
|
||||
for (unsigned i=0;i<dim;i++){
|
||||
std::vector<Type> coeff;
|
||||
coeff.push_back(max/(bounds.maximum()-bounds.minimum()));
|
||||
|
|
@ -190,6 +194,22 @@ class moeoObjectiveVectorNormalizer
|
|||
}
|
||||
return res;
|
||||
}
|
||||
|
||||
/**
|
||||
create a default scale that does nothing when applied
|
||||
@return a dummy scale
|
||||
*/
|
||||
static Scale make_dummy_scale(){
|
||||
unsigned int dim=MOEOT::ObjectiveVector::nObjectives();
|
||||
Scale res;
|
||||
for (unsigned int i=0;i<dim;i++){
|
||||
std::vector<Type> coeff;
|
||||
coeff.push_back(1);
|
||||
coeff.push_back(0);
|
||||
res.push_back(coeff);
|
||||
}
|
||||
return res;
|
||||
}
|
||||
/**
|
||||
* main fonction, normalize a vector. All objective returned vectors will be between 0 and max previously
|
||||
* supplied, be carefull about a possible rounding error.
|
||||
|
|
@ -221,6 +241,7 @@ class moeoObjectiveVectorNormalizer
|
|||
/**
|
||||
fast(to use, not in complexity) function to normalize a population
|
||||
@param pop the population to normalize
|
||||
@param max the returned values will be between 0 and max
|
||||
@return a vector of normalized Objective vectors < max
|
||||
*/
|
||||
static std::vector<ObjectiveVector> normalize(const eoPop<MOEOT> &pop, Type &max){
|
||||
|
|
@ -231,20 +252,33 @@ class moeoObjectiveVectorNormalizer
|
|||
/**
|
||||
Change the scale according to a new pop. Should be called everytime pop is updated
|
||||
@param pop population to analyse
|
||||
@param max_param the worst vector is is set to max_param
|
||||
*/
|
||||
void update_by_pop(eoPop<MOEOT> pop){
|
||||
scale=make_scale_from_pop(pop,max);
|
||||
}
|
||||
|
||||
/** change the scale with the worst point and the best point
|
||||
@param max the worst point
|
||||
@param min the best point
|
||||
@param _max the worst point
|
||||
@param _min the best point
|
||||
*/
|
||||
void update_by_min_max(const ObjectiveVector &_min,const ObjectiveVector &_max){
|
||||
scale=make_scale_from_minmax(_min,_max,max);
|
||||
}
|
||||
|
||||
/** change the scale according to given boundaries
|
||||
@param boundaries a vector of bounds corresponding to the bounds in each dimension
|
||||
*/
|
||||
void update_by_bounds(const std::vector<Bounds> &boundaries){
|
||||
scale=make_scale_from_bounds(boundaries);
|
||||
}
|
||||
/** change the scale according to bounds,them same is used in each dimension
|
||||
@param bounds bounds corresponding to the bounds in each dimension
|
||||
*/
|
||||
void update_by_bounds(const Bounds &bounds){
|
||||
scale=make_scale_from_bounds(bounds);
|
||||
}
|
||||
|
||||
|
||||
/**
|
||||
updates the scale
|
||||
@param _scale the new scale
|
||||
|
|
@ -265,10 +299,7 @@ class moeoObjectiveVectorNormalizer
|
|||
}
|
||||
max=_max;
|
||||
}
|
||||
protected:
|
||||
moeoObjectiveVectorNormalizer()
|
||||
{
|
||||
}
|
||||
|
||||
private:
|
||||
Scale scale;
|
||||
Type max;
|
||||
|
|
|
|||
|
|
@ -74,7 +74,7 @@ public:
|
|||
}
|
||||
|
||||
private:
|
||||
ObjectiveVector& objVec;
|
||||
ObjectiveVector objVec;
|
||||
std::map<unsigned int, QuadTreeNode<ObjectiveVector>*> subTree;
|
||||
};
|
||||
|
||||
|
|
@ -119,10 +119,13 @@ public:
|
|||
if(!(root->getSubTree().empty())){
|
||||
QuadTreeIterator it=root->getSubTree().begin();
|
||||
while(!stop && (it != root->getSubTree().end())){
|
||||
std::cout << "hop" << std::endl;
|
||||
if( ((*it).first < succ) && (((succ ^ bound) & ((*it).first ^ bound))== succ) ){
|
||||
if((*it).second != NULL){
|
||||
std::cout << "hop"<<std::endl;
|
||||
std::cout << "first: " << (*it).first << ", bound: " << bound << ", xor: " << ((*it).first ^ bound) << std::endl;
|
||||
if( ((*it).first < succ) && (((succ ^ bound) & ((*it).first ^ bound)) == (succ ^ bound)) ){
|
||||
|
||||
stop = test1(tmp, (*it).second);
|
||||
stop = test1(tmp, (*it).second);
|
||||
}
|
||||
}
|
||||
it++;
|
||||
}
|
||||
|
|
@ -132,8 +135,10 @@ public:
|
|||
//dominance test2
|
||||
QuadTreeIterator it=root->getSubTree().begin();
|
||||
while(it != root->getSubTree().end()){
|
||||
if( (succ < (*it).first) && ((succ & (*it).first) == succ)){
|
||||
test2(tmp, (*it).second, root, (*it).first);
|
||||
if((*it).second != NULL){
|
||||
if( (succ < (*it).first) && ((succ & (*it).first) == succ)){
|
||||
test2(tmp, (*it).second, root, (*it).first);
|
||||
}
|
||||
}
|
||||
it++;
|
||||
}
|
||||
|
|
@ -141,6 +146,7 @@ public:
|
|||
QuadTreeNode<ObjectiveVector>* tmp = new QuadTreeNode<ObjectiveVector>(_obj);
|
||||
// std::cout << "insert case new son: " << root->getVec() << std::endl;
|
||||
if(root->setChild(succ, tmp)){
|
||||
std::cout << "\n\nthe root changed\n\n";
|
||||
root=root->getSubTree()[succ];
|
||||
}
|
||||
else{
|
||||
|
|
@ -154,6 +160,7 @@ public:
|
|||
stop=true;
|
||||
}
|
||||
}
|
||||
std::cout << "realroot: " << realroot->getVec() << std::endl;
|
||||
root=realroot;
|
||||
}
|
||||
return res;
|
||||
|
|
@ -165,6 +172,7 @@ public:
|
|||
* @param _objVec2
|
||||
*/
|
||||
unsigned int k_succ(const ObjectiveVector& _objVec1, const ObjectiveVector& _objVec2){
|
||||
std::cout << "enter k_succ" << std::endl;
|
||||
unsigned int res=0;
|
||||
for(int i=0; i < ObjectiveVector::nObjectives(); i++){
|
||||
if( (ObjectiveVector::minimizing(i) && ((_objVec1[i] - _objVec2[i]) >= (-1.0 * 1e-6 ))) ||
|
||||
|
|
@ -172,6 +180,7 @@ public:
|
|||
res+=pow(2,ObjectiveVector::nObjectives()-i-1);
|
||||
}
|
||||
}
|
||||
std::cout << "quit k_succ" << std::endl;
|
||||
return res;
|
||||
}
|
||||
|
||||
|
|
@ -184,32 +193,39 @@ public:
|
|||
//create the new root
|
||||
QuadTreeNode<ObjectiveVector>* newroot = new QuadTreeNode<ObjectiveVector>(_newroot);
|
||||
//reconsider each son of the old root
|
||||
if(!(root->getSubTree().empty())){
|
||||
if(!(root->getSubTree().empty())){
|
||||
QuadTreeIterator it;
|
||||
for(it=(root->getSubTree()).begin(); it != (root->getSubTree()).end(); it++){
|
||||
std::cout << "replace: " << (*it).second->getVec() << std::endl;
|
||||
reconsider(newroot, (*it).second);
|
||||
std::cout << "on passe ici" << std::endl;
|
||||
if((*it).second!=NULL){
|
||||
std::cout << "replace: " << (*it).second->getVec() << std::endl;
|
||||
reconsider(newroot, (*it).second);
|
||||
std::cout << "end replacement" << std::endl;
|
||||
}
|
||||
}
|
||||
}
|
||||
std::cout << "replace after reconsider" << std::endl;
|
||||
//replace the old root by the new one
|
||||
delete(root);
|
||||
root = newroot;
|
||||
std::cout << root << " -> "<< root->getVec() << std::endl;
|
||||
|
||||
std::cout << "replace after change the root" << std::endl;
|
||||
std::cout << "quit replace: " << std::endl;
|
||||
}
|
||||
|
||||
void reconsider(QuadTreeNode<ObjectiveVector>* _newroot, QuadTreeNode<ObjectiveVector>* _child){
|
||||
std::cout << "enter reconsider: " << std::endl;
|
||||
unsigned int succ;
|
||||
if(!(_child->getSubTree().empty())){
|
||||
std::cout << "enter reconsider" << std::endl;
|
||||
QuadTreeIterator it;
|
||||
for(it=(_child->getSubTree()).begin(); it != (_child->getSubTree()).end(); it++){
|
||||
std::cout << "reconsider: " << (*it).second->getVec() << std::endl;
|
||||
QuadTreeNode<ObjectiveVector>* tmp=(*it).second;
|
||||
_child->getSubTree()[(*it).first]=NULL;
|
||||
if((*it).second != NULL){
|
||||
std::cout << "reconsider: " << (*it).second->getVec() << std::endl;
|
||||
QuadTreeNode<ObjectiveVector>* tmp=(*it).second;
|
||||
_child->getSubTree()[(*it).first]=NULL;
|
||||
|
||||
reconsider(_newroot, tmp);
|
||||
reconsider(_newroot, tmp);
|
||||
}
|
||||
}
|
||||
}
|
||||
else{
|
||||
|
|
@ -221,27 +237,30 @@ public:
|
|||
if(succ==bound)
|
||||
delete(_child);
|
||||
else if(_newroot->getSubTree()[succ] != NULL){
|
||||
std::cout << "hohoho" << std::endl;
|
||||
// std::cout << "hohoho" << std::endl;
|
||||
reinsert(_newroot->getSubTree()[succ],_child);
|
||||
}
|
||||
else{
|
||||
std::cout << "houhouhou" << std::endl;
|
||||
// std::cout << "houhouhou" << std::endl;
|
||||
_newroot->setChild(succ, _child);
|
||||
}
|
||||
std::cout << "quit reconsider: " << std::endl;
|
||||
}
|
||||
|
||||
void reinsert(QuadTreeNode<ObjectiveVector>* _node1, QuadTreeNode<ObjectiveVector>* _node2){
|
||||
std::cout << "enter reinsert: " << std::endl;
|
||||
if(_node1 != _node2){
|
||||
std::cout << "enter reinsert" << std::endl;
|
||||
std::cout << "node1: " << _node1->getVec() << ", node2: " << _node2->getVec() << std::endl;
|
||||
unsigned int succ;
|
||||
if(!(_node1->getSubTree().empty())){
|
||||
QuadTreeIterator it;
|
||||
for(it=(_node1->getSubTree()).begin(); it != (_node1->getSubTree()).end(); it++){
|
||||
std::cout << "reinsert: " << (*it).second->getVec() << std::endl;
|
||||
QuadTreeNode<ObjectiveVector>* tmp=(*it).second;
|
||||
_node1->getSubTree().erase(it);
|
||||
reinsert(_node1, tmp);
|
||||
if((*it).second != NULL){
|
||||
std::cout << "reinsert: " << (*it).second->getVec() << std::endl;
|
||||
QuadTreeNode<ObjectiveVector>* tmp=(*it).second;
|
||||
_node1->getSubTree().erase(it);
|
||||
reinsert(_node1, tmp);
|
||||
}
|
||||
}
|
||||
}
|
||||
succ=k_succ(_node2->getVec(),_node1->getVec());
|
||||
|
|
@ -252,10 +271,12 @@ public:
|
|||
_node1->setChild(succ, _node2);
|
||||
}
|
||||
}
|
||||
std::cout << "quit reinsert: " << std::endl;
|
||||
}
|
||||
|
||||
void remove(QuadTreeNode<ObjectiveVector>* _node, QuadTreeNode<ObjectiveVector>* _parent, unsigned int _succ){
|
||||
std::cout << "enter remove" << std::endl;
|
||||
std::cout << "enter remove -> " << _node->getVec() << std::endl;
|
||||
printTree();
|
||||
unsigned int k=1;
|
||||
QuadTreeNode<ObjectiveVector>* tmp=NULL;
|
||||
_parent->getSubTree()[_succ]=NULL;
|
||||
|
|
@ -274,6 +295,8 @@ public:
|
|||
k++;
|
||||
}
|
||||
delete(_node);
|
||||
std::cout << "quit remove: " << std::endl;
|
||||
printTree();
|
||||
}
|
||||
|
||||
bool test1(QuadTreeNode<ObjectiveVector>* _node1, QuadTreeNode<ObjectiveVector>* _node2){
|
||||
|
|
@ -287,12 +310,15 @@ public:
|
|||
else{
|
||||
QuadTreeIterator it=_node2->getSubTree().begin();
|
||||
while(!res && (it != _node2->getSubTree().end())){
|
||||
if( ((succ ^ bound) & ((*it).first ^ bound)) == succ){
|
||||
res = res || test1(_node1, (*it).second);
|
||||
if((*it).second!=NULL){
|
||||
if( ((succ ^ bound) & ((*it).first ^ bound)) == (succ^bound)){
|
||||
res = res || test1(_node1, (*it).second);
|
||||
}
|
||||
}
|
||||
it++;
|
||||
}
|
||||
}
|
||||
std::cout << "quit test1" << std::endl;
|
||||
return res;
|
||||
}
|
||||
|
||||
|
|
@ -302,6 +328,7 @@ public:
|
|||
unsigned int succ;
|
||||
succ=k_succ(_node1->getVec(), _node2->getVec());
|
||||
if(succ==0){
|
||||
// std::cout << "\n\n\nPEUT ETRE ICI\n\n\n";
|
||||
remove(_node2, _parent, _succ);
|
||||
if(_parent->getSubTree()[_succ]!=NULL)
|
||||
test2(_node1, _parent->getSubTree()[_succ], _parent, _succ);
|
||||
|
|
@ -309,12 +336,15 @@ public:
|
|||
else{
|
||||
QuadTreeIterator it=_node2->getSubTree().begin();
|
||||
while(it != _node2->getSubTree().end()){
|
||||
if( (succ & (*it).first) == succ){
|
||||
test2(_node1, (*it).second, _node2, (*it).first);
|
||||
if((*it).second!=NULL){
|
||||
if( (succ & (*it).first) == succ){
|
||||
test2(_node1, (*it).second, _node2, (*it).first);
|
||||
}
|
||||
}
|
||||
it++;
|
||||
}
|
||||
}
|
||||
std::cout << "quit test2" << std::endl;
|
||||
}
|
||||
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue