paradiseo/branches/paradiseo-moeo-1.0/src/moeoIteratedIBMOLS.h

// -*- mode: c++; c-indent-level: 4; c++-member-init-indent: 8; comment-column: 35; -*-

//-----------------------------------------------------------------------------
// moeoIteratedIBMOLS.h
// (c) OPAC Team (LIFL), Dolphin Project (INRIA), 2007
/*
    This library...

    Contact: paradiseo-help@lists.gforge.inria.fr, http://paradiseo.gforge.inria.fr
 */
//-----------------------------------------------------------------------------

#ifndef MOEOITERATEDIBMOLS_H_
#define MOEOITERATEDIBMOLS_H_

#include <eoContinue.h>
#include <eoEvalFunc.h>
#include <eoOp.h>
#include <eoPop.h>
#include <utils/rnd_generators.h>
#include <moMove.h>
#include <moMoveInit.h>
#include <moNextMove.h>
#include <moeoMoveIncrEval.h>
#include <moeoArchive.h>
#include <moeoLS.h>
#include <moeoIndicatorBasedFitnessAssignment.h>
#include <moeoIndicatorBasedLS.h>



//#include <rsCrossQuad.h>



/**
 * Iterated version of IBMOLS as described in
 * Basseur M., Burke K. : "Indicator-Based Multi-Objective Local Search" (2007).
 */
template < class MOEOT, class Move >
class moeoIteratedIBMOLS : public moeoLS < MOEOT, eoPop < MOEOT > & >
{
public:

    /** The type of objective vector */
    typedef typename MOEOT::ObjectiveVector ObjectiveVector;


    /**
     * Ctor.
     * @param _moveInit the move initializer
     * @param _nextMove the neighborhood explorer
     * @param _eval the full evaluation
     * @param _moveIncrEval the incremental evaluation
     * @param _fitnessAssignment the fitness assignment strategy
     * @param _continuator the stopping criteria
     * @param _monOp the monary operator
     * @param _randomMonOp the random monary operator (or random initializer)
     * @param _nNoiseIterations the number of iterations to apply the random noise
     */
    moeoIteratedIBMOLS(
        moMoveInit < Move > & _moveInit,
        moNextMove < Move > & _nextMove,
        eoEvalFunc < MOEOT > & _eval,
        moeoMoveIncrEval < Move > & _moveIncrEval,
        moeoIndicatorBasedFitnessAssignment < MOEOT > & _fitnessAssignment,
        eoContinue < MOEOT > & _continuator,
        eoMonOp < MOEOT > & _monOp,
        eoMonOp < MOEOT > & _randomMonOp,
        unsigned _nNoiseIterations=1
    ) :
            ibmols(_moveInit, _nextMove, _eval, _moveIncrEval, _fitnessAssignment, _continuator),
            eval(_eval),
            continuator(_continuator),
            monOp(_monOp),
            randomMonOp(_randomMonOp),
            nNoiseIterations(_nNoiseIterations)
    {}


    /**
     * Apply the local search iteratively until the stopping criteria is met.
     * @param _pop the initial population
     * @param _arch the (updated) archive
     */
    void operator() (eoPop < MOEOT > & _pop, moeoArchive < MOEOT > & _arch)
    {

        _arch.update(_pop);
        cout << endl << endl << "***** IBMOLS 1" << endl;
        unsigned counter = 2;
        ibmols(_pop, _arch);
        while (continuator(_arch))
        {
            // generate new solutions from the archive
            generateNewSolutions(_pop, _arch);
            cout << endl << endl << "***** IBMOLS " << counter++ << endl;
            // apply the local search (the global archive is updated in the sub-function)
            ibmols(_pop, _arch);
        }

    }


private:

    /** the local search to iterate */
    moeoIndicatorBasedLS < MOEOT, Move > ibmols;
    /** the full evaluation */
    eoEvalFunc < MOEOT > & eval;
    /** the stopping criteria */
    eoContinue < MOEOT > & continuator;
    /** the monary operator */
    eoMonOp < MOEOT > & monOp;
    /** the random monary operator (or random initializer) */
    eoMonOp < MOEOT > & randomMonOp;
    /** the number of iterations to apply the random noise */
    unsigned nNoiseIterations;


    /**
     * Creates new population randomly initialized and/or initialized from the archive _arch.
     * @param _pop the output population
     * @param _arch the archive
     */
    void generateNewSolutions(eoPop < MOEOT > & _pop, const moeoArchive < MOEOT > & _arch)
    {
        // shuffle vector for the random selection of individuals
        vector<unsigned> shuffle;
        shuffle.resize(std::max(_pop.size(), _arch.size()));
        // init shuffle
        for (unsigned i=0; i<shuffle.size(); i++)
        {
            shuffle[i] = i;
        }
        // randomize shuffle
        UF_random_generator <unsigned int> gen;
        std::random_shuffle(shuffle.begin(), shuffle.end(), gen);
        // start the creation of new solutions
        for (unsigned i=0; i<_pop.size(); i++)
        {
            if (shuffle[i] < _arch.size())
                // the given archive contains the individual i
            {
                // add it to the resulting pop
                _pop[i] = _arch[shuffle[i]];
                // then, apply the operator nIterationsNoise times
                for (unsigned j=0; j<nNoiseIterations; j++)
                {
                    monOp(_pop[i]);
                }
            }
            else
                // a randomly generated solution needs to be added
            {
                // random initialization
                randomMonOp(_pop[i]);
            }
            // evaluation of the new individual
            _pop[i].invalidate();
            eval(_pop[i]);
        }
    }





///////////////////////////////////////////////////////////////////////////////////////////////////////
// A DEVELOPPER RAPIDEMENT POUR TESTER AVEC CROSSOVER //
    /*
    	void generateNewSolutions2(eoPop < MOEOT > & _pop, const moeoArchive < MOEOT > & _arch)
    	{
    		// here, we must have a QuadOp !
    		//eoQuadOp < MOEOT > quadOp;
    		rsCrossQuad quadOp;
    		// shuffle vector for the random selection of individuals
    		vector<unsigned> shuffle;
    		shuffle.resize(_arch.size());
    		// init shuffle
    		for (unsigned i=0; i<shuffle.size(); i++)
    		{
    			shuffle[i] = i;
    		}
    		// randomize shuffle
    		UF_random_generator <unsigned int> gen;
    		std::random_shuffle(shuffle.begin(), shuffle.end(), gen);
    		// start the creation of new solutions
    		unsigned i=0;
    		while ((i<_pop.size()-1) && (i<_arch.size()-1))
    		{
    			_pop[i] = _arch[shuffle[i]];
    			_pop[i+1] = _arch[shuffle[i+1]];
    			// then, apply the operator nIterationsNoise times
    			for (unsigned j=0; j<nNoiseIterations; j++)
    			{
    				quadOp(_pop[i], _pop[i+1]);
    			}
    			eval(_pop[i]);
    			eval(_pop[i+1]);
    			i=i+2;
    		}
    		// do we have to add some random solutions ?
    		while (i<_pop.size())
    		{
    			randomMonOp(_pop[i]);
    			eval(_pop[i]);
    			i++;
    		}
    	}
    	*/
///////////////////////////////////////////////////////////////////////////////////////////////////////





};

#endif /*MOEOITERATEDIBMOLS_H_*/