mcmc_sampler.hpp

Go to the documentation of this file.

/*
-------------------------------------------------------------------------
   This file is part of BayesOpt, an efficient C++ library for 
   Bayesian optimization.

   Copyright (C) 2011-2015 Ruben Martinez-Cantin <rmcantin@unizar.es>
 
   BayesOpt is free software: you can redistribute it and/or modify it 
   under the terms of the GNU Affero General Public License as published by
   the Free Software Foundation, either version 3 of the License, or
   (at your option) any later version.

   BayesOpt 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 Affero General Public License for more details.

   You should have received a copy of the GNU Affero General Public License
   along with BayesOpt.  If not, see <http://www.gnu.org/licenses/>.
------------------------------------------------------------------------
*/


#ifndef  _MCMC_SAMPLER_HPP_
#define  _MCMC_SAMPLER_HPP_

#include <boost/scoped_ptr.hpp>
#include "randgen.hpp"
#include "optimizable.hpp"

namespace bayesopt {

  // We plan to add more in the future 
  typedef enum {
    SLICE_MCMC           
  } McmcAlgorithms;


  class MCMCSampler
  {
  public:
    MCMCSampler(RBOptimizable* rbo, size_t dim, randEngine& eng);
    virtual ~MCMCSampler();

    void setAlgorithm(McmcAlgorithms newAlg);

    void setNParticles(size_t nParticles);

    void setNBurnOut(size_t nParticles);

    void run(vectord &Xnext);

    vectord getParticle(size_t i);

    void printParticles();

  private:
    void randomJump(vectord &x);
    void burnOut(vectord &x);
    void sliceSample(vectord &x);

    boost::scoped_ptr<RBOptimizableWrapper> obj;

    McmcAlgorithms mAlg;
    size_t mDims;
    size_t nBurnOut;
    size_t nSamples;
    bool mStepOut;

    vectord mSigma;
    vecOfvec mParticles;
    randEngine& mtRandom;

  private: //Forbidden
    MCMCSampler();
    MCMCSampler(MCMCSampler& copy);
  };

  inline void MCMCSampler::setAlgorithm(McmcAlgorithms newAlg)
  { mAlg = newAlg; };

  inline void MCMCSampler::setNParticles(size_t nParticles)
  { nSamples = nParticles; };

  inline void MCMCSampler::setNBurnOut(size_t nParticles)
  { nBurnOut = nParticles; };

  inline vectord MCMCSampler::getParticle(size_t i)
  { return mParticles[i]; };

  inline void MCMCSampler::printParticles()
  {
    for(size_t i=0; i<mParticles.size(); ++i)
      { 
    FILE_LOG(logDEBUG) << i << "->" << mParticles[i] 
               << " | Log-lik " << -obj->evaluate(mParticles[i]);
      }
  }

} //namespace bayesopt


#endif