nojhan/paradiseo
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Add multi-Binomial distrib operators in EDO

Browse source 
So as to model vector<vector<bool>> individuals with 2D binomial distributions (as Eigen matrix).
This commit is contained in:
Johann Dreo 2013-03-07 20:33:09 +01:00
commit b06250dc39
6 changed files with 321 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

3
edo/src/edo
View file

@ -39,6 +39,7 @@ Authors:
#include "edoNormalMulti.h"
#include "edoNormalAdaptive.h"
#include "edoBinomial.h"
#include "edoBinomialMulti.h"
#include "edoEstimator.h"
#include "edoEstimatorUniform.h"
@ -47,6 +48,7 @@ Authors:
#include "edoEstimatorAdaptive.h"
#include "edoEstimatorNormalAdaptive.h"
#include "edoEstimatorBinomial.h"
#include "edoEstimatorBinomialMulti.h"
#include "edoModifier.h"
#include "edoModifierDispersion.h"
@ -61,6 +63,7 @@ Authors:
#include "edoSamplerNormalMulti.h"
#include "edoSamplerNormalAdaptive.h"
#include "edoSamplerBinomial.h"
#include "edoSamplerBinomialMulti.h"
#include "edoVectorBounds.h"

59
edo/src/edoBinomialMulti.h Normal file
View file

@ -0,0 +1,59 @@
/*
The Evolving Distribution Objects framework (EDO) is a template-based,
ANSI-C++ evolutionary computation library which helps you to write your
own estimation of distribution algorithms.
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Copyright (C) 2013 Thales group
*/
/*
Authors:
Johann Dréo <johann.dreo@thalesgroup.com>
*/
#ifndef _edoBinomialMulti_h
#define _edoBinomialMulti_h
#include "edoBinomial.h"
#ifdef WITH_EIGEN // FIXME: provide an uBLAS implementation
#include <Eigen/Dense>
/** A 2D binomial distribution modeled as a matrix.
*
* i.e. a container of binomial distribution.
*
* @ingroup Distributions
* @ingroup Binomial
*/
template<class EOT, class T=Eigen::MatrixXd>
class edoBinomialMulti : public edoDistrib<EOT>, public T
{
public:
/** This constructor takes an initial matrix of probabilities.
* Use it if you have prior knowledge.
*/
edoBinomialMulti( T initial_probas )
: T(initial_probas) {}
/** Constructor without any assumption.
*/
edoBinomialMulti() {}
};
#endif // WITH_EIGEN
#endif // !_edoBinomialMulti_h

97
edo/src/edoEstimatorBinomialMulti.h Normal file
View file

@ -0,0 +1,97 @@
/*
The Evolving Distribution Objects framework (EDO) is a template-based,
ANSI-C++ evolutionary computation library which helps you to write your
own estimation of distribution algorithms.
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Copyright (C) 2013 Thales group
*/
/*
Authors:
Johann Dréo <johann.dreo@thalesgroup.com>
*/
#ifndef _edoEstimatorBinomialMulti_h
#define _edoEstimatorBinomialMulti_h
#include "edoBinomialMulti.h"
#include "edoEstimator.h"
#ifdef WITH_EIGEN
#include <Eigen/Dense>
/** An estimator for edoBinomialMulti
*
* @ingroup Estimators
* @ingroup Binomial
*/
template< class EOT, class D = edoBinomialMulti<EOT> >
class edoEstimatorBinomialMulti : public edoEstimator<D>
{
protected:
D eot2d( EOT from, unsigned int rows, unsigned int cols ) // FIXME maybe more elegant with Eigen::Map?
{
assert( rows > 0 );
assert( from.size() == rows );
assert( cols > 0 );
D to( Eigen::MatrixXd(rows, cols) );
for( unsigned int i=0; i < rows; ++i ) {
assert( from[i].size() == cols );
for( unsigned int j=0; j < cols; ++j ) {
to(i,j) = from[i][j];
}
}
return to;
}
public:
/** The expected EOT interface is the same as an Eigen3::MatrixXd.
*/
D operator()( eoPop<EOT>& pop )
{
unsigned int popsize = pop.size();
assert(popsize > 0);
unsigned int rows = pop[0].size();
assert( rows > 0 );
unsigned int cols = pop[0][0].size();
assert( cols > 0 );
D probas( D::Zero(rows, cols) );
// We still need a loop over pop, because it is an eoVector
for (unsigned int i = 0; i < popsize; ++i) {
D indiv = eot2d( pop[i], rows, cols );
assert( indiv.rows() == rows && indiv.cols() == cols );
// the EOT matrix should be filled with 1 or 0 only
assert( indiv.sum() <= popsize );
probas += indiv / popsize;
// sum and scalar product, no size pb expected
assert( probas.rows() == rows && probas.cols() == cols );
}
return probas;
}
};
#endif // WITH_EIGEN
#endif // !_edoEstimatorBinomialMulti_h

74
edo/src/edoSamplerBinomialMulti.h Normal file
View file

@ -0,0 +1,74 @@
/*
The Evolving Distribution Objects framework (EDO) is a template-based,
ANSI-C++ evolutionary computation library which helps you to write your
own estimation of distribution algorithms.
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Copyright (C) 2013 Thales group
*/
/*
Authors:
Johann Dréo <johann.dreo@thalesgroup.com>
*/
#ifndef _edoSamplerBinomialMulti_h
#define _edoSamplerBinomialMulti_h
#include <utils/eoRNG.h>
#include "edoSampler.h"
#include "edoBinomialMulti.h"
#ifdef WITH_EIGEN
#include <Eigen/Dense>
/** A sampler for an edoBinomialMulti distribution.
*
* @ingroup Samplers
* @ingroup Binomial
*/
template< class EOT, class D = edoBinomialMulti<EOT> >
class edoSamplerBinomialMulti : public edoSampler<D>
{
public:
EOT sample( D& distrib )
{
unsigned int rows = distrib.rows();
unsigned int cols = distrib.cols();
assert(rows > 0);
assert(cols > 0);
// The point we want to draw
// x = {x1, x2, ..., xn}
EOT solution;
// Sampling all dimensions
for( unsigned int i = 0; i < rows; ++i ) {
typename EOT::AtomType vec;
for( unsigned int j = 0; j < cols; ++j ) {
// Toss a coin, biased by the proba of being 1.
vec.push_back( rng.flip( distrib(i,j) ) );
}
solution.push_back( vec );
}
return solution;
}
};
#endif // WITH_EIGEN
#endif // !_edoSamplerBinomialMulti_h

1
edo/test/CMakeLists.txt
View file

@ -41,6 +41,7 @@ set(SOURCES
# t-dispatcher-round
t-repairer-modulo
t-binomial
t-binomialmulti
)
foreach(current ${SOURCES})

87
edo/test/t-binomialmulti.cpp Normal file
View file

@ -0,0 +1,87 @@
/*
The Evolving Distribution Objects framework (EDO) is a template-based,
ANSI-C++ evolutionary computation library which helps you to write your
own estimation of distribution algorithms.
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Copyright (C) 2013 Thales group
*/
/*
Authors:
Johann Dréo <johann.dreo@thalesgroup.com>
*/
#include <vector>
#include <iostream>
#include <string>
#include <cmath>
#include <eo>
#include <edo>
#include <ga.h> // for Bools
#ifdef WITH_EIGEN
#include <Eigen/Dense>
// NOTE: a typedef on eoVector does not work, because of readFrom on a vector AtomType
// typedef eoVector<eoMinimizingFitness, std::vector<bool> > Bools;
class Bools : public std::vector<std::vector<bool> >, public EO<double>
{
public:
typedef std::vector<bool> AtomType;
};
int main(int ac, char** av)
{
eoParser parser(ac, av);
std::string section("Algorithm parameters");
unsigned int popsize = parser.createParam((unsigned int)100000, "popSize", "Population Size", 'P', section).value(); // P
unsigned int rows = parser.createParam((unsigned int)2, "lines", "Lines number", 'l', section).value(); // l
unsigned int cols = parser.createParam((unsigned int)3, "columns", "Columns number", 'c', section).value(); // c
double proba = parser.createParam((double)0.5, "proba", "Probability to estimate", 'b', section).value(); // b
if( parser.userNeedsHelp() ) {
parser.printHelp(std::cout);
exit(1);
}
make_help(parser);
std::cout << "Init distrib" << std::endl;
Eigen::MatrixXd initd = Eigen::MatrixXd::Constant(rows,cols,proba);
edoBinomialMulti<Bools> distrib( initd );
std::cout << distrib << std::endl;
edoEstimatorBinomialMulti<Bools> estimate;
edoSamplerBinomialMulti<Bools> sample;
std::cout << "Sample a pop from the init distrib" << std::endl;
eoPop<Bools> pop; pop.reserve(popsize);
for( unsigned int i=0; i < popsize; ++i ) {
pop.push_back( sample( distrib ) );
}
std::cout << "Estimate a distribution from the sampled pop" << std::endl;
distrib = estimate( pop );
std::cout << distrib << std::endl;
std::cout << "Estimated initial proba = " << distrib.mean() << std::endl;
}
#endif // WITH_EIGEN