test_discretepdf.cpp

一个很全的matlab工具包

// $Id: test_discretepdf.cpp,v 1.14 2004/11/05 15:21:06 wmeeusse Exp $
// KG october 2002: test program for the discrete distribution

#include <pdf/discretepdf.h>
#include <matrix_wrapper.h>
#include <vector_wrapper.h>
#include <iostream>

using namespace std;
using namespace BFL;
using namespace MatrixWrapper;


#define NUMBER_OF_CFS 5

// Mapping between discrete class and simple numbers
#define CF1 0
#define CF2 1
#define CF3 2
#define CF4 3
#define CF5 4

#define SUCCES 0;
#define NUM_SAMPLES 10000

int main ()
{
  cerr << "==================================================\n"
       << "Test of the Discrete PDF class applied to CF\n"
       << "estimation\n"
       << "=================================================="
       << endl;

  DiscretePdf CF_estimate(NUMBER_OF_CFS);
  ColumnVector uniform_vector(NUMBER_OF_CFS);
  uniform_vector = CF_estimate.ProbabilitiesGet();
  cout << "A priori probabilities (5 Contact formations):\n" << uniform_vector
       << endl;

  CF_estimate.ProbabilitySet(CF3,100);
  uniform_vector = CF_estimate.ProbabilitiesGet();
  cout << "After setting the probability of CF3 to 100:\n" << uniform_vector 
       << endl;

  cerr << "Sampling " << NUM_SAMPLES << " samples from PDF" << endl;
  list<Sample<int> > samples(NUM_SAMPLES);
  list<Sample<int> >::iterator it;
  samples = CF_estimate.SampleFrom(NUM_SAMPLES,DEFAULT,NULL);
  
  // Calculating the mean of the samples
  cerr << "Calculating mean and covariance of samples" << endl;
  
  double mean; double sum; mean = 0.0 ; sum = 0.0;
  for (it = samples.begin(); it != samples.end(); it++)
    { 
      // cerr << "sample:\n" << it->ValueGet() << endl;
      sum += it->ValueGet();
    }
  mean = sum * (1.0 / NUM_SAMPLES);
  
  // Calculating the covariance matrix
  double diff; // Temporary storage
  double diffsum; diffsum = 0.0;

  for (it = samples.begin(); it != samples.end(); it++)
    {
      diff = it->ValueGet() - mean;
      diffsum += diff * diff;
    }

  double Covariance;
  Covariance = (( 1.0 / (NUM_SAMPLES-1) ) * diffsum);

  cout << "mean of the samples = \n" << mean << endl;
  cout << "Covariance Matrix of samples =\n" << Covariance  << endl;

  return SUCCES;

}