loopyBeliefPropagation_tpl.h

Go to the documentation of this file.

/***************************************************************************
 *   Copyright (C) 2005 by Christophe GONZALES et Pierre-Henri WUILLEMIN   *
 *   {prenom.nom}_at_lip6.fr                                               *
 *                                                                         *
 *   This program is free software; you can redistribute it and/or modify  *
 *   it under the terms of the GNU General Public License as published by  *
 *   the Free Software Foundation; either version 2 of the License, or     *
 *   (at your option) any later version.                                   *
 *                                                                         *
 *   This program 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 General Public License for more details.                          *
 *                                                                         *
 *   You should have received a copy of the GNU General Public License     *
 *   along with this program; if not, write to the                         *
 *   Free Software Foundation, Inc.,                                       *
 *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.             *
 ***************************************************************************/
#ifndef DOXYGEN_SHOULD_SKIP_THIS

#  include <algorithm>
#  include <sstream>
#  include <string>

#  define LBP_DEFAULT_MAXITER 100
#  define LBP_DEFAULT_EPSILON 1e-8
#  define LBP_DEFAULT_MIN_EPSILON_RATE 1e-10
#  define LBP_DEFAULT_PERIOD_SIZE 1
#  define LBP_DEFAULT_VERBOSITY false


// to ease parsing for IDE
#  include <agrum/BN/inference/loopyBeliefPropagation.h>


namespace gum {

  template < typename GUM_SCALAR >
  LoopyBeliefPropagation< GUM_SCALAR >::LoopyBeliefPropagation(
     const IBayesNet< GUM_SCALAR >* bn) :
      ApproximateInference< GUM_SCALAR >(bn) {
    // for debugging purposes
    GUM_CONSTRUCTOR(LoopyBeliefPropagation);

    this->setEpsilon(LBP_DEFAULT_EPSILON);
    this->setMinEpsilonRate(LBP_DEFAULT_MIN_EPSILON_RATE);
    this->setMaxIter(LBP_DEFAULT_MAXITER);
    this->setVerbosity(LBP_DEFAULT_VERBOSITY);
    this->setPeriodSize(LBP_DEFAULT_PERIOD_SIZE);

    __init_messages();
  }

  template < typename GUM_SCALAR >
  INLINE LoopyBeliefPropagation< GUM_SCALAR >::~LoopyBeliefPropagation() {
    GUM_DESTRUCTOR(LoopyBeliefPropagation);
  }


  template < typename GUM_SCALAR >
  void LoopyBeliefPropagation< GUM_SCALAR >::__init_messages() {
    __messages.clear();
    for (const auto& tail : this->BN().nodes()) {
      Potential< GUM_SCALAR > p;
      p.add(this->BN().variable(tail));
      p.fill(static_cast< GUM_SCALAR >(1));

      for (const auto& head : this->BN().children(tail)) {
        __messages.insert(Arc(head, tail), p);
        __messages.insert(Arc(tail, head), p);
      }
    }
  }

  template < typename GUM_SCALAR >
  void LoopyBeliefPropagation< GUM_SCALAR >::_updateOutdatedBNStructure() {
    __init_messages();
  }


  template < typename GUM_SCALAR >
  Potential< GUM_SCALAR >
     LoopyBeliefPropagation< GUM_SCALAR >::__computeProdPi(NodeId X) {
    const auto& varX = this->BN().variable(X);

    auto piX = this->BN().cpt(X);
    for (const auto& U : this->BN().parents(X)) {
      piX *= __messages[Arc(U, X)];
    }
    piX = piX.margSumIn({&varX});

    return piX;
  }

  template < typename GUM_SCALAR >
  Potential< GUM_SCALAR >
     LoopyBeliefPropagation< GUM_SCALAR >::__computeProdPi(NodeId X,
                                                           NodeId except) {
    const auto& varX = this->BN().variable(X);
    const auto& varExcept = this->BN().variable(except);
    auto        piXexcept = this->BN().cpt(X);
    for (const auto& U : this->BN().parents(X)) {
      if (U != except) { piXexcept *= __messages[Arc(U, X)]; }
    }
    piXexcept = piXexcept.margSumIn({&varX, &varExcept});
    return piXexcept;
  }


  template < typename GUM_SCALAR >
  Potential< GUM_SCALAR >
     LoopyBeliefPropagation< GUM_SCALAR >::__computeProdLambda(NodeId X) {
    Potential< GUM_SCALAR > lamX;
    if (this->hasEvidence(X)) {
      lamX = *(this->evidence()[X]);
    } else {
      lamX.add(this->BN().variable(X));
      lamX.fill(1);
    }
    for (const auto& Y : this->BN().children(X)) {
      lamX *= __messages[Arc(Y, X)];
    }

    return lamX;
  }

  template < typename GUM_SCALAR >
  Potential< GUM_SCALAR >
     LoopyBeliefPropagation< GUM_SCALAR >::__computeProdLambda(NodeId X,
                                                               NodeId except) {
    Potential< GUM_SCALAR > lamXexcept;
    if (this->hasEvidence(X)) {   //
      lamXexcept = *this->evidence()[X];
    } else {
      lamXexcept.add(this->BN().variable(X));
      lamXexcept.fill(1);
    }
    for (const auto& Y : this->BN().children(X)) {
      if (Y != except) { lamXexcept *= __messages[Arc(Y, X)]; }
    }

    return lamXexcept;
  }


  template < typename GUM_SCALAR >
  GUM_SCALAR LoopyBeliefPropagation< GUM_SCALAR >::__updateNodeMessage(NodeId X) {
    auto piX = __computeProdPi(X);
    auto lamX = __computeProdLambda(X);

    GUM_SCALAR KL = 0;
    Arc        argKL(0, 0);

    // update lambda_par (for arc U->x)
    for (const auto& U : this->BN().parents(X)) {
      auto newLambda =
         (__computeProdPi(X, U) * lamX).margSumIn({&this->BN().variable(U)});
      newLambda.normalize();
      auto ekl = static_cast< GUM_SCALAR >(0);
      try {
        ekl = __messages[Arc(X, U)].KL(newLambda);
      } catch (InvalidArgument&) {
        GUM_ERROR(InvalidArgument, "Not compatible pi during computation");
      } catch (FatalError&) {   // 0 misplaced
        ekl = std::numeric_limits< GUM_SCALAR >::infinity();
      }
      if (ekl > KL) {
        KL = ekl;
        argKL = Arc(X, U);
      }
      __messages.set(Arc(X, U), newLambda);
    }

    // update pi_child (for arc x->child)
    for (const auto& Y : this->BN().children(X)) {
      auto newPi = (piX * __computeProdLambda(X, Y));
      newPi.normalize();
      GUM_SCALAR ekl = KL;
      try {
        ekl = __messages[Arc(X, Y)].KL(newPi);
      } catch (InvalidArgument&) {
        GUM_ERROR(InvalidArgument, "Not compatible pi during computation");
      } catch (FatalError&) {   // 0 misplaced
        ekl = std::numeric_limits< GUM_SCALAR >::infinity();
      }
      if (ekl > KL) {
        KL = ekl;
        argKL = Arc(X, Y);
      }
      __messages.set(Arc(X, Y), newPi);
    }

    return KL;
  }

  template < typename GUM_SCALAR >
  INLINE void LoopyBeliefPropagation< GUM_SCALAR >::__initStats() {
    __init_messages();
    for (const auto& node : this->BN().topologicalOrder()) {
      __updateNodeMessage(node);
    }
  }


  template < typename GUM_SCALAR >
  void LoopyBeliefPropagation< GUM_SCALAR >::_makeInference() {
    __initStats();
    this->initApproximationScheme();

    std::vector< NodeId > shuffleIds;
    for (const auto& node : this->BN().nodes())
      shuffleIds.push_back(node);

    auto engine = std::default_random_engine{};

    GUM_SCALAR error = 0.0;
    do {
      std::shuffle(std::begin(shuffleIds), std::end(shuffleIds), engine);
      this->updateApproximationScheme();
      for (const auto& node : shuffleIds) {
        GUM_SCALAR e = __updateNodeMessage(node);
        if (e > error) error = e;
      }
    } while (this->continueApproximationScheme(error));
  }


  template < typename GUM_SCALAR >
  INLINE const Potential< GUM_SCALAR >&
               LoopyBeliefPropagation< GUM_SCALAR >::_posterior(NodeId id) {
    auto p = __computeProdPi(id) * __computeProdLambda(id);
    p.normalize();
    __posteriors.set(id, p);

    return __posteriors[id];
  }
} /* namespace gum */

#endif   // DOXYGEN_SHOULD_SKIP_THIS