ecga.cpp

这是遗传算法的源代码

// -*- c++ -*-
//
//  File:         ecga.cpp
//
//  Description:  C++ implementation of the class ecga.
//                Contains the ECGA loop.
//
//  Author:       Fernando Lobo
//
//  Date:         June/1999
//
//  Extended to deal with chi-ary problems by Luis de la Ossa
//  GCC 3.4 and 4 series compliance by Kumara Sastry 
//
//  Date:         March/2006

#include <iostream>
#include <iomanip>
#include <fstream>

#include "ecga.hpp"
#include "parameter.hpp"
#include "population.hpp"
#include "mpm.hpp"
#include "utility.hpp"

extern randomG RANDOM;

//----------------------------------------------------------

void ecga::report( std::ofstream &outfile, population *pop, int gen )
{
  //
  // get information about the best individual in the population
  //
  chromosome bestChrom;
  bestChrom = (*pop)[ pop->best() ];

  //
  // print it
  //
  std::cout << "generation      : " << gen << std::endl
	    << "best fitness    : " << pop->maxfit() << std::endl
	    << "avg fitness     : " << pop->avgfit() << std::endl
	    << "best chromosome : " << bestChrom << std::endl
	    << "------------------------------------------------------------"
	    << std::endl;

  outfile << "Generation " << gen << std::endl;
  outfile << *pop;
}

bool ecga::done( population *pop, int gen )
{
  switch( parameter::stop_criteria )
   {
     case ALLELE_CONVERGENCE : return sc_allele_conv( *pop, gen );
     case MAX_GENERATIONS    : return sc_maxgen( *pop, gen );
     default: 
        error("This should never happen.");
   }
}

//
// runs the ECGA and sends output information to the 'outfile'.
//
void ecga::run( std::ofstream &outfile )
{
  int gen = 0;

  //
  // first population with random individuals
  //
  population *pop = new population( parameter::popsize );       
  pop->random();  
  pop->evaluate();
  pop->statistics();   
  report( outfile, pop, gen );

  while( !done( pop, gen ) ) { 
    gen++;
    //
    // apply selection
    //
    population *temp_pop = new population( pop->popsize() );
    pop->selection( temp_pop );      
    delete pop;     // delete unnecessary temporary
    pop = temp_pop;
    
    //
    // model the population with a greedy MPM search
    //
    mpm MPM( parameter::lchrom );   
    MPM.model( pop, outfile );  
    if( parameter::report_MPM ) 
      std::cout << "MPM: " << MPM << std::endl;
    
    //
    // generate a new population using the MPM
    //
    temp_pop = new population( pop->popsize() ); 
    MPM.generate( pop, temp_pop );  
    delete pop;
    pop = temp_pop;
    
    //
    // evaluate individuals and do the report
    //
    pop->evaluate();                  
    pop->statistics();                
    report( outfile, pop, gen );
  }
  
  //
  // cleanup
  //
  delete pop;
}