posterior_mcmc.cpp

/*
-------------------------------------------------------------------------
   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/>.
------------------------------------------------------------------------
*/
#include "log.hpp"
#include "posterior_mcmc.hpp"

namespace bayesopt
{
  MCMCModel::MCMCModel(size_t dim, Parameters parameters, 
               randEngine& eng):
    PosteriorModel(dim,parameters,eng), nParticles(10)
  {
    //TODO: Take nParticles from parameters
    
    // Configure Surrogate and Criteria Functions
    setSurrogateModel(eng);
    setCriteria(eng);

    // Seting MCMC for kernel hyperparameters...
    // We use the first GP as the "walker" to get the particles. Then,
    // we will use a whole vector of GPs to avoid recomputing the
    // kernel matrices after every data point.
    size_t nhp = mGP[0].nHyperParameters();
    kSampler.reset(new MCMCSampler(&mGP[0],nhp,eng));

    kSampler->setNParticles(nParticles);
    kSampler->setNBurnOut(100);
  }

  MCMCModel::~MCMCModel()
  { } // Default destructor


  void MCMCModel::updateHyperParameters()
  {
    // We take the initial point as the last particle from previous update.
    size_t last = mGP.size()-1;
    vectord lastTheta = mGP[last].getHyperParameters();

    FILE_LOG(logDEBUG) << "Initial kernel parameters: " << lastTheta;
    kSampler->run(lastTheta);
    for(size_t i = 0; i<nParticles; ++i)
      {
    mGP[i].setHyperParameters(kSampler->getParticle(i));
      }
    FILE_LOG(logDEBUG) << "Final kernel parameters: " << lastTheta; 
  };


  void MCMCModel::setSurrogateModel(randEngine& eng)
  {
    for(size_t i = 0; i<nParticles; ++i)
      {
    mGP.push_back(NonParametricProcess::create(mDims,mParameters,
                           mData,mMean,eng));
      } 
  } // setSurrogateModel

  void MCMCModel::setCriteria(randEngine& eng)
  {
    CriteriaFactory mCFactory;

    for(size_t i = 0; i<nParticles; ++i)
      {
    mCrit.push_back(mCFactory.create(mParameters.crit_name,&mGP[i]));
    mCrit[i].setRandomEngine(eng);

    if (mCrit[i].nParameters() == mParameters.crit_params.size())
      {
        mCrit[i].setParameters(mParameters.crit_params);
      }
    else // If the number of parameters is different, use default.
      {
        if (mParameters.crit_params.size() != 0)
          {
        FILE_LOG(logERROR) << "Expected " << mCrit[i].nParameters() 
                   << " parameters. Got " 
                   << mParameters.crit_params.size() << " instead.";
          }
        FILE_LOG(logINFO) << "Using default parameters for criteria.";
      }
      }
  } // setCriteria




}// namespace bayesopt