ex7.c

麻省理工开发的免费遗传算法类库GAlib,很好用

/* ----------------------------------------------------------------------------
  ex7.C
  mbwall 19jan95
  Copyright 1995-1996  Massachusetts Institute of Technology

 DESCRIPTION:
   This example functions the same way as example 3, but this example shows
how to use some of the other member functions in the GA library.  We also do
a few more fancy things with the genome (ie use the read/write methods).
---------------------------------------------------------------------------- */
#include <stdio.h>
#include <stdlib.h>
#include <iostream.h>
#include <fstream.h>
#include <ga/ga.h>

float objective(GAGenome &);

int
main(int argc, char *argv[])
{
  cout << "Example 7\n\n";
  cout << "This program reads in a data file then runs a steady-state GA \n";
  cout << "whose objective function tries to match the pattern of bits that\n";
  cout << "are in the data file.\n\n";

// See if we've been given a seed to use (for testing purposes).  When you
// specify a random seed, the evolution will be exactly the same each time
// you use that seed number.

  for(int ii=1; ii<argc; ii++) {
    if(strcmp(argv[ii++],"seed") == 0) {
      GARandomSeed((unsigned int)atoi(argv[ii]));
    }
  }

// Set the default values of the parameters.

  int i,j;
  GAParameterList params;
  GASteadyStateGA::registerDefaultParameters(params);
  params.set(gaNpopulationSize, 50);    // number of individuals in population
  params.set(gaNpCrossover, 0.8);       // likelihood of doing crossover
  params.set(gaNpMutation, 0.001);	// probability of mutation
  params.set(gaNnGenerations, 200);	// number of generations
  params.set(gaNscoreFrequency, 20);	// how often to record scores
  params.set(gaNflushFrequency, 50);    // how often to flush scores to file
  params.set(gaNscoreFilename, "bog.dat");
  params.parse(argc, argv, gaFalse);

  char datafile[128] = "smiley.txt";
  char parmfile[128] = "";

// Parse the command line for arguments.  We look for two possible arguments
// (after the parameter list has grabbed everything it recognizes).  One is the
// name of a data file from which to read, the other is the name of a 
// parameters file from which to read.  Notice that any parameters in the 
// parameters file will override the defaults above AND any entered on the
// command line.

  for(i=1; i<argc; i++){
    if(strcmp("dfile", argv[i]) == 0){
      if(++i >= argc){
        cerr << argv[0] << ": the data file option needs a filename.\n";
        exit(1);
      }
      else{
	sprintf(datafile, argv[i]);
        continue;
      }
    }
    else if(strcmp("pfile", argv[i]) == 0){
      if(++i >= argc){
        cerr << argv[0] << ": the parameters file option needs a filename.\n";
        exit(1);
      }
      else{
	sprintf(parmfile, argv[i]);
	params.read(parmfile);
        continue;
      }
    }
    else if(strcmp("seed", argv[i]) == 0){
      if(++i < argc) continue;
      continue;
    }
    else {
      cerr << argv[0] << ":  unrecognized arguement: " << argv[i] << "\n\n";
      cerr << "valid arguements include GAlib arguments plus:\n";
      cerr << "  dfile\tdata file from which to read (" << datafile << ")\n";
      cerr << "  pfile\tparameters file (" << parmfile << ")\n\n";
      cerr << "default parameters are:\n" << params << "\n\n";
      exit(1);
    }
  }

// Read in the pattern from the specified file.  File format is pretty simple:
// two integers that give the height then width of the matrix, then the matrix
// of 1's and 0's (with whitespace inbetween).
//   Here we use a binary string genome to store the desired pattern.  This
// shows how you can read in directly from a stream into a genome.  (This can 
// be useful in a population initializer when you want to bias your population)

  ifstream inStream(datafile, ios :: in);
  if(!inStream){
    cerr << "Cannot open " << datafile << " for input.\n";
    exit(1);
  }

  int height, width;
  inStream >> height >> width;
  GA2DBinaryStringGenome target(width, height);
  inStream >> target;
  inStream.close();

// Print out the pattern to be sure we got the right one.

  cout << "input pattern:\n";
  for(j=0; j<height; j++){
    for(i=0; i<width; i++)
      cout << (target.gene(i,j) == 1 ? '*' : ' ') << " ";
    cout << "\n";
  }
  cout << "\n"; cout.flush();

// Now create the first genome and the GA.  When we create the genome, we give
// it not only the objective function but also 'user data'.  In this case, the
// user data is a pointer to our target pattern.  From a C++ point of view it
// would be better to derive a new genome class with its own data, but here we
// just want a quick-and-dirty implementation, so we use the user-data.

  GA2DBinaryStringGenome genome(width, height, objective, (void *)&target);
  GASteadyStateGA ga(genome);

// When you use a GA with overlapping populations, the default score
// frequency (how often the best of generation score is recorded) defaults
// to 100.  We use the parameters member function to change this value (along
// with all of the other parameters we set above).  You can also change the 
// score frequency using the scoreFrequency member function of the GA.  Each of
// the parameters can be set individually if you like.
//   Here we just use the values that were set in the parameter list.

  ga.parameters(params);

// The default selection method is RouletteWheel.  Here we set the selection
// method to TournamentSelection.

  GATournamentSelector selector;
  ga.selector(selector);

// The following member functions override the values that were set using the
// parameter list.  They are commented out here so that you can see how they 
// would be used.

// We can control the amount of overlap from generation to generation using the
// pReplacement member function.  If we specify a value of 1 (100%) then the
// entire population is replaced each generation.  Notice that the percentage
// must be high enough to have at least one individual produced in each 
// generation.  If not, the GA will post a warning message.

//  ga.pReplacement(0.3);

// Often we use the number of generations as the criterion for terminating the
// GA run.  Here we override that and tell the GA to use convergence as a 
// termination criterion.  Note that you can pass ANY function as the stopping
// criterion (as long as it has the correct signature).
//   Notice that the values we set here for p- and n-convergence override those
// that we set in the parameters object.

  ga.terminator(GAGeneticAlgorithm::TerminateUponConvergence);

//  ga.pConvergence(0.99);	        // converge to within 1%
//  ga.nConvergence(100);		// within the last 100 generations

// Evolve the GA 'by hand'.  When you use this method, be sure to initialize
// the GA before you start to evolve it.  You can print out the status of the
// current population by using the ga.population() member function.  This is
// also how you would print the status of the GA to disk along the way (rather
// than waiting to the end then printing the scores, for example).

  ga.initialize();
  while(!ga.done()){
    ++ga;
  }
  ga.flushScores();

// Now that the GA is finished, we set our default genome to equal the contents
// of the best genome that the GA found.  Then we print it out.

  genome = ga.statistics().bestIndividual();
  cout << "the ga generated:\n";
  for(j=0; j<height; j++){
    for(i=0; i<width; i++)
      cout << (genome.gene(i,j) == 1 ? '*' : ' ') << " ";
    cout << "\n";
  }
  cout << "\n"; cout.flush();

  cout << "best of generation data are in '" << ga.scoreFilename() << "'\n";

  return 0;
}
 



// This objective function just tries to match the genome to the pattern in the
// user data.
float
objective(GAGenome & c)
{
  GA2DBinaryStringGenome & genome = (GA2DBinaryStringGenome &)c;
  GA2DBinaryStringGenome * pattern = 
    (GA2DBinaryStringGenome *)c.userData();

  float value=0.0;
  for(int i=0; i<genome.width(); i++)
    for(int j=0; j<genome.height(); j++)
      value += (float)(genome.gene(i,j) == pattern->gene(i,j));
  return(value);
}