D:/Projekt/ECF_trunk/ECF/binary/Copy of BinaryMutSimple.cpp

#include "../ECF_base.h"
#include "Binary.h"
#include <cmath>

uint extraTime;

void BinaryMutSimple::registerParameters(StateP state)
{
    myGenotype_->registerParameter(state, "mut.simple", (voidP) new double(0), DOUBLE);
    myGenotype_->registerParameter(state, "mut.simple.bitprob", (voidP) new double(0.001), DOUBLE);
    myGenotype_->registerParameter(state, "mut.simple.extra", (voidP) new uint(1), UINT);
}


bool BinaryMutSimple::initialize(StateP state)
{
    voidP sptr = myGenotype_->getParameterValue(state, "mut.simple");
    probability_ = *((double*)sptr.get());

    sptr = myGenotype_->getParameterValue(state, "mut.simple.bitprob");
    bitProb_ = *((double*)sptr.get());

    bUseBitProb_ = false;
    if(myGenotype_->isParameterDefined(state, "mut.simple.bitprob"))
        bUseBitProb_ = true;

    sptr = myGenotype_->getParameterValue(state, "mut.simple.extra");
    extraTime = *((uint*)sptr.get());

    return true;
}


bool BinaryMutSimple::mutate(GenotypeP gene)
{
    Binary* bin = (Binary*) (gene.get());

    //
    // trosenje vremena - simulacija Marinove mutacije koja prolazi sve bitove
    //
    bool bit = true;
    for(uint i = 0; i < extraTime * bin->variables.size() * bin->getNumBits(); i++) 
        if(state_->getRandomizer()->getRandomDouble() < bitProb_)
            bit = !bit;


    // invert all bits with 'bitProb_' probability
    if(bUseBitProb_) {
        for(uint i = 0; i < bin->variables.size(); i++)
            for(uint j = 0; j < bin->getNumBits(); j++)
                if(state_->getRandomizer()->getRandomDouble() < bitProb_)
                    bin->variables[i][j] = !(bin->variables[i][j]);
    }
    // invert a single random bit in the genotype
    else {
        uint iBit = state_->getRandomizer()->getRandomInteger((uint) bin->getNumBits());
        uint dimension = state_->getRandomizer()->getRandomInteger((uint) bin->variables.size());
        bin->variables[dimension][iBit] = !(bin->variables[dimension][iBit]);
    }

    bin->update();

    //
    // izvedba jednostavnog lokalnog operatora
    //
    IndividualP myInd = state_->getAlgorithm()->mutation_->currentInd;
    FitnessP current, modified;
    double precision = 1 / pow(10., (int) bin->nDecimal_);

    current = state_->getAlgorithm()->evalOp_->evaluate(myInd);

    for(uint i = 0; i < bin->variables.size(); i++) {
        double newVal = bin->realValue[i];

        bin->realValue[i] += precision;
        modified = state_->getAlgorithm()->evalOp_->evaluate(myInd);
        if(modified->isBetterThan(current)) {
            current = modified;
            newVal = bin->realValue[i];
        }

        bin->realValue[i] -= 2 * precision;
        modified = state_->getAlgorithm()->evalOp_->evaluate(myInd);
        if(modified->isBetterThan(current)) {
            current = modified;
            newVal = bin->realValue[i];
        }

        bin->realValue[i] = newVal;
    }

    // update genotype according to realValue
    for(uint iVar = 0; iVar < bin->nDimension_; iVar++) {
        bin->decValue[iVar] = static_cast<long int> ((bin->realValue[iVar] - bin->minValue_) / (bin->maxValue_ - bin->minValue_) * bin->potention_);

        long dec = bin->decValue[iVar];
        for (int iBit = bin->nBits_; iBit > 0; dec = dec/2, iBit--) {
            bin->vBool_[iBit - 1] = (dec % 2) ? true:false;
        }
        bin->variables[iVar] = bin->vBool_;

        if(bin->bRounding_) {
            bin->realValue[iVar] = bin->round(bin->realValue[iVar], bin->nDecimal_);
        }
    }

    return true;
}