nodeDatabase.h

Go to the documentation of this file.

/**
 *
 *   Copyright 2005-2020 Pierre-Henri WUILLEMIN(@LIP6) & Christophe GONZALES(@AMU)
 *   info_at_agrum_dot_org
 *
 *  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 3 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, see <http://www.gnu.org/licenses/>.
 *
 */


/**
 * @file
 * @brief Headers of the NodeDatabase class.
 *
 * @author Jean-Christophe MAGNAN
 */

// =========================================================================
#ifndef GUM_NODE_DATABASE_H
#define GUM_NODE_DATABASE_H
// =========================================================================
#include <agrum/tools/core/hashTable.h>
#include <agrum/tools/core/sequence.h>
// =========================================================================
#include <agrum/FMDP/learning/core/templateStrategy.h>
#include <agrum/FMDP/learning/core/testPolicy/Chi2TestPolicy.h>
#include <agrum/FMDP/learning/core/testPolicy/GTestPolicy.h>
#include <agrum/FMDP/learning/core/testPolicy/leastSquareTestPolicy.h>
#include <agrum/FMDP/learning/observation.h>
// =========================================================================
#include <agrum/tools/variables/discreteVariable.h>
// =========================================================================

namespace gum {

  /**
   * @class NodeDatabase nodeDatabase.h
   * <agrum/FMDP/learning/datastructure/nodeDatabase.h>
   * @brief
   * @ingroup fmdp_group
   *
   */


  template < TESTNAME AttributeSelection, bool isScalar >
  class NodeDatabase {
    typedef typename ValueSelect< isScalar, double, Idx >::type ValueType;

    template < typename GUM_SCALAR >
    using TestPolicy =
       typename TestSelect< AttributeSelection,
                            GTestPolicy< GUM_SCALAR >,
                            Chi2TestPolicy< GUM_SCALAR >,
                            LeastSquareTestPolicy< GUM_SCALAR > >::type;

    public:
    // ==========================================================================
    /// @name Constructor & destructor.
    // ==========================================================================
    /// @{

    // ###################################################################
    /// Default constructor
    // ###################################################################
    NodeDatabase(const Set< const DiscreteVariable* >*,
                 const DiscreteVariable* = nullptr);

    // ###################################################################
    /// Default destructor
    // ###################################################################
    ~NodeDatabase();

    // ============================================================================
    /// Allocators and Deallocators redefinition
    // ============================================================================
    void* operator new(size_t s) {
      return SmallObjectAllocator::instance().allocate(s);
    }
    void operator delete(void* p) {
      SmallObjectAllocator::instance().deallocate(p, sizeof(NodeDatabase));
    }

    /// @}

    // ==========================================================================
    /// @name Observation handling methods
    // ==========================================================================
    /// @{

    // ###################################################################
    /** Updates database with new observation
     *
     * Calls either @fn addObservation__( const Observation*, Int2Type<true>)
     * or @fn addObservation__( const Observation*, Int2Type<false>)
     * depending on if we're learning reward function or transition probability
     **/
    // ###################################################################
    void addObservation(const Observation*);

    private:
    void addObservation__(const Observation*, Int2Type< true >);
    void addObservation__(const Observation*, Int2Type< false >);

    public:
    // ###################################################################
    /// Nb observation taken into account by this instance
    // ###################################################################
    INLINE Idx nbObservation() const { return nbObservation__; }

    /// @}

    // ==========================================================================
    /// @name Variable Test Methods
    // ==========================================================================
    /// @{

    // ###################################################################
    /// Indicates wether or not, node has sufficient observation so that
    /// any statistic is relevant
    // ###################################################################
    INLINE bool isTestRelevant(const DiscreteVariable* var) const {
      return attrTable__[var]->isTestRelevant();
    }

    // ###################################################################
    /// Returns the performance of given variables according to selection
    /// criterion
    // ###################################################################
    INLINE double testValue(const DiscreteVariable* var) const {
      return attrTable__[var]->score();
    }

    // ###################################################################
    /// Returns the performance of given variables according to selection
    /// secondary criterion (to break ties)
    // ###################################################################
    INLINE double testOtherCriterion(const DiscreteVariable* var) const {
      return attrTable__[var]->secondaryscore();
    }

    /// @}

    // ==========================================================================
    /// @name Aggregation Methods
    // ==========================================================================
    /// @{

    // ###################################################################
    /// Merges given NodeDatabase informations into current nDB.
    // ###################################################################
    NodeDatabase< AttributeSelection, isScalar >&
       operator+=(const NodeDatabase< AttributeSelection, isScalar >& src);

    // ###################################################################
    /// Returns a reference to nDB test policy for given variable
    /// (so that test policy information can be merged too)
    // ###################################################################
    const TestPolicy< ValueType >* testPolicy(const DiscreteVariable* var) const {
      return attrTable__[var];
    }

    // ###################################################################
    /// Iterators on value count to recopy correctly its content
    // ###################################################################
    const HashTableConstIteratorSafe< ValueType, Idx > cbeginValues() const {
      return valueCount__.cbeginSafe();
    }
    const HashTableConstIteratorSafe< ValueType, Idx > cendValues() const {
      return valueCount__.cendSafe();
    }

    /// @}

    // ###################################################################
    ///
    // ###################################################################

    Idx effectif(Idx moda) const {
      return valueCount__.exists(ValueType(moda)) ? valueCount__[ValueType(moda)]
                                                  : 0;
    }

    Idx valueDomain() const { return valueDomain__(Int2Type< isScalar >()); }

    private:
    Idx valueDomain__(Int2Type< true >) const { return valueCount__.size(); }
    Idx valueDomain__(Int2Type< false >) const { return value__->domainSize(); }

    std::string toString() const;

    private:
    /// Table giving for every variables its instantiation
    HashTable< const DiscreteVariable*, TestPolicy< ValueType >* > attrTable__;

    /// So does this reference on the value observed
    const DiscreteVariable* value__;

    ///
    Idx nbObservation__;

    ///
    HashTable< ValueType, Idx > valueCount__;
  };


} /* namespace gum */

#include <agrum/FMDP/learning/datastructure/nodeDatabase_tpl.h>

#endif   // GUM_NODE_DATABASE_H