Portion of a BN identified by the list of nodes and a BayesNet. More...

#include <agrum/BN/BayesNetFragment.h>

Inheritance diagram for gum::BayesNetFragment< GUM_SCALAR >:

Collaboration diagram for gum::BayesNetFragment< GUM_SCALAR >:

Public Member Functions
bool	operator== (const IBayesNet< GUM_SCALAR > &from) const
	This operator compares 2 BNs ! More...

bool	operator!= (const IBayesNet< GUM_SCALAR > &from) const

Size	dim () const
	Returns the dimension (the number of free parameters) in this bayes net. More...

Size	maxVarDomainSize () const

GUM_SCALAR	minParam () const

GUM_SCALAR	maxParam () const

GUM_SCALAR	minNonZeroParam () const

GUM_SCALAR	maxNonOneParam () const

std::string	toString () const

NodeSet	minimalCondSet (NodeId target, const NodeSet &soids) const

NodeSet	minimalCondSet (const NodeSet &targets, const NodeSet &soids) const

double	log10DomainSize () const

bool	hasSameStructure (const DAGmodel &other)

void	attachSignal__ (__sig__::ISignaler *sender)

void	detachSignal__ (__sig__::ISignaler *sender)

Constructors / Destructors
	BayesNetFragment ()=delete

	BayesNetFragment (const BayesNetFragment< GUM_SCALAR > &fragment)=delete

	BayesNetFragment (BayesNetFragment< GUM_SCALAR > &&fragment)=delete

	BayesNetFragment (const IBayesNet< GUM_SCALAR > &bn)

virtual	~BayesNetFragment ()

signals
virtual void	whenNodeAdded (const void *src, NodeId id) noexcept override
	the action to take when a new node is inserted into the graph More...

virtual void	whenNodeDeleted (const void *src, NodeId id) noexcept override
	the action to take when a node has just been removed from the graph More...

virtual void	whenArcAdded (const void *src, NodeId from, NodeId to) noexcept override
	the action to take when a new arc is inserted into the graph More...

virtual void	whenArcDeleted (const void *src, NodeId from, NodeId to) noexcept override
	the action to take when an arc has just been removed from the graph More...

IBayesNet interface
virtual const Potential< GUM_SCALAR > &	cpt (NodeId varId) const override
	Returns the CPT of a variable. More...

virtual const VariableNodeMap &	variableNodeMap () const override
	Returns a constant reference to the VariableNodeMap of this BN. More...

virtual const DiscreteVariable &	variable (NodeId id) const override
	Returns a constant reference over a variabe given it's node id. More...

virtual NodeId	nodeId (const DiscreteVariable &var) const override
	Return id node from discrete var pointer. More...

virtual NodeId	idFromName (const std::string &name) const override
	Getter by name. More...

virtual const DiscreteVariable &	variableFromName (const std::string &name) const override
	Getter by name. More...

virtual std::string	toDot () const override
	creates a dot representing the whole referred BN hilighting the fragment. More...

API for Fragment
bool	isInstalledNode (NodeId id) const noexcept
	check if a certain NodeId exists in the fragment More...

void	installNode (NodeId id)
	install a node referenced by its nodeId More...

void	installAscendants (NodeId id)
	install a node and all its ascendants More...

void	uninstallNode (NodeId id) noexcept
	uninstall a node referenced by its nodeId More...

void	installMarginal (NodeId id, const Potential< GUM_SCALAR > *pot)
	install a local marginal for a node into the fragment. More...

void	installCPT (NodeId id, const Potential< GUM_SCALAR > *pot)
	install a local cpt for a node into the fragment. More...

void	uninstallCPT (NodeId id) noexcept
	uninstall a local CPT. More...

bool	checkConsistency (NodeId id) const
	returns true if the nodeId's (local or not) cpt is consistent with its parents in the fragment More...

bool	checkConsistency () const noexcept
	returns true if all nodes in the fragment are consistent More...

Joint Probability manipulation methods
GUM_SCALAR	jointProbability (const Instantiation &i) const
	Compute a parameter of the joint probability for the BN (given an instantiation of the vars) More...

GUM_SCALAR	log2JointProbability (const Instantiation &i) const
	Compute a parameter of the log joint probability for the BN (given an instantiation of the vars) More...

Getter and setters
const std::string &	property (const std::string &name) const
	Return the value of the property name of this DAGModel. More...

const std::string &	propertyWithDefault (const std::string &name, const std::string &byDefault) const
	Return the value of the property name of this DAGModel. More...

void	setProperty (const std::string &name, const std::string &value)
	Add or change a property of this DAGModel. More...

Variable manipulation methods.
const DAG &	dag () const
	Returns a constant reference to the dag of this Bayes Net. More...

Size	size () const
	Returns the number of variables in this Directed Graphical Model. More...

Size	sizeArcs () const
	Returns the number of arcs in this Directed Graphical Model. More...

bool	empty () const
	Retursn true if this Directed Graphical Model is empty. More...

const NodeGraphPart &	nodes () const
	Returns a constant reference to the dag of this Bayes Net. More...

virtual Instantiation	completeInstantiation () const final
	Get an instantiation over all the variables of the model. More...

Arc manipulation methods.
const ArcSet &	arcs () const
	returns the set of nodes with arc ingoing to a given node More...

const NodeSet &	parents (const NodeId id) const
	returns the set of nodes with arc ingoing to a given node More...

const NodeSet &	parents (const std::string &name) const
	returns the set of nodes with arc ingoing to a given node More...

const NodeSet &	children (const NodeId id) const
	returns the set of nodes with arc outgoing from a given node More...

const NodeSet &	children (const std::string &name) const
	returns the set of nodes with arc ingoing to a given node More...

Graphical methods
const UndiGraph &	moralGraph (bool clear=true) const
	The node's id are coherent with the variables and nodes of the topology. More...

const Sequence< NodeId > &	topologicalOrder (bool clear=true) const
	The topological order stays the same as long as no variable or arcs are added or erased src the topology. More...

Protected Attributes
DAG	_dag
	The DAG of this Directed Graphical Model. More...

DiGraph *	_graph
	the graph to listen to More...

Protected Member Functions
void	_uninstallArc (NodeId from, NodeId to) noexcept

void	_installArc (NodeId from, NodeId to) noexcept

void	_installCPT (NodeId id, const Potential< GUM_SCALAR > *pot) noexcept

void	_uninstallCPT (NodeId id) noexcept
	uninstall a local CPT. More...

Detailed Description

template<typename GUM_SCALAR>
class gum::BayesNetFragment< GUM_SCALAR >

Portion of a BN identified by the list of nodes and a BayesNet.

Author: Pierre-Henri WUILLEMIN

This class is a decorator of a BayesNet implementing the IBayesNet interface. CPTs can be shared with the BN or can be specific to the Fragment if different.

BayesNetFragment is a DiGraphListener in order to be synchronized (especially when removing nodes or arcs).

In a BayesNetFragment, one can install or remove nodes. An arc can be in the fragment if and only if its head and tail are installed in the fragment. *When installing a node, all the arcs that can be added in the fragment are *effectively installed (resp. *when uninstalling a node, etc.).

A BayesNetFragment can redefine potential for node. The main reason is to be able to install a node without installing all its parents (and its ascendants). So local CPT to the node can be installed. However, it is not done automatically.

If a cpt is not locally defined, the fragment uses the cpt defined in the referred BN. The checkConsistency() method verifies that, for all installed nodes, either all the parents are installed or a local CPT is defined.

Definition at line 70 of file BayesNetFragment.h.

Constructor & Destructor Documentation

◆ BayesNetFragment() [1/4]

template<typename GUM_SCALAR>

gum::BayesNetFragment< GUM_SCALAR >::BayesNetFragment ( )

delete

◆ BayesNetFragment() [2/4]

template<typename GUM_SCALAR>

gum::BayesNetFragment< GUM_SCALAR >::BayesNetFragment ( const BayesNetFragment< GUM_SCALAR > & fragment )

delete

◆ BayesNetFragment() [3/4]

template<typename GUM_SCALAR>

gum::BayesNetFragment< GUM_SCALAR >::BayesNetFragment ( BayesNetFragment< GUM_SCALAR > && fragment )

delete

◆ BayesNetFragment() [4/4]

template<typename GUM_SCALAR >

gum::BayesNetFragment< GUM_SCALAR >::BayesNetFragment ( const IBayesNet< GUM_SCALAR > & bn )

explicit

Definition at line 32 of file BayesNetFragment_tpl.h.

                                         :
       DiGraphListener(&bn.dag()),
       __bn(bn) {
     GUM_CONSTRUCTOR(BayesNetFragment);
   }

◆ ~BayesNetFragment()

template<typename GUM_SCALAR >

gum::BayesNetFragment< GUM_SCALAR >::~BayesNetFragment ( )

virtual

Definition at line 40 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__localCPTs, gum::BayesNetFragment< GUM_SCALAR >::_uninstallCPT(), and gum::DAGmodel::nodes().

                                                     {
     GUM_DESTRUCTOR(BayesNetFragment);
 
     for (auto node : nodes())
       if (__localCPTs.exists(node)) _uninstallCPT(node);
   }

Here is the call graph for this function:

Member Function Documentation

◆ _installArc()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::_installArc	(	NodeId	from,
		NodeId	to
	)

protectednoexcept

Definition at line 185 of file BayesNetFragment_tpl.h.

References gum::DAGmodel::_dag, and gum::DAG::addArc().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), and gum::BayesNetFragment< GUM_SCALAR >::uninstallCPT().

                                                                               {
     this->_dag.addArc(from, to);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ _installCPT()

template<typename GUM_SCALAR >

void gum::BayesNetFragment< GUM_SCALAR >::_installCPT	(	NodeId	id,
		const Potential< GUM_SCALAR > *	pot
	)

protectednoexcept

Definition at line 191 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::__localCPTs, gum::BayesNetFragment< GUM_SCALAR >::_installArc(), gum::BayesNetFragment< GUM_SCALAR >::_uninstallArc(), gum::BayesNetFragment< GUM_SCALAR >::_uninstallCPT(), gum::Set< Key, Alloc >::beginSafe(), gum::Set< Key, Alloc >::endSafe(), gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), and gum::DAGmodel::parents().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::installCPT(), and gum::BayesNetFragment< GUM_SCALAR >::installMarginal().

                                                              {
     // topology
     const auto& parents = this->parents(id);
     for (auto node_it = parents.beginSafe(); node_it != parents.endSafe();
          ++node_it)   // safe iterator needed here
       _uninstallArc(*node_it, id);
 
     for (Idx i = 1; i < pot->nbrDim(); i++) {
       NodeId parent = __bn.idFromName(pot->variable(i).name());
 
       if (isInstalledNode(parent)) _installArc(parent, id);
     }
 
     // local cpt
     if (__localCPTs.exists(id)) _uninstallCPT(id);
 
     __localCPTs.insert(id, pot);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ _uninstallArc()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::_uninstallArc	(	NodeId	from,
		NodeId	to
	)

protectednoexcept

Definition at line 179 of file BayesNetFragment_tpl.h.

References gum::DAGmodel::_dag, and gum::ArcGraphPart::eraseArc().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), and gum::BayesNetFragment< GUM_SCALAR >::whenArcDeleted().

                                                                                 {
     this->_dag.eraseArc(Arc(from, to));
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ _uninstallCPT()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::_uninstallCPT ( NodeId id )

protectednoexcept

uninstall a local CPT.

Does nothing if no local CPT for this nodeId No check. No change in the topology. Checks are made in public methods.

Definition at line 236 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__localCPTs.

Referenced by gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), gum::BayesNetFragment< GUM_SCALAR >::uninstallCPT(), and gum::BayesNetFragment< GUM_SCALAR >::~BayesNetFragment().

                                                                               {
     delete __localCPTs[id];
     __localCPTs.erase(id);
   }

Here is the caller graph for this function:

◆ arcs()

INLINE const ArcSet & gum::DAGmodel::arcs ( ) const

inherited

returns the set of nodes with arc ingoing to a given node

Note that the set of arcs returned may be empty if no arc within the ArcGraphPart is ingoing into the given node.

Parameters

id	the node toward which the arcs returned are pointing

Definition at line 101 of file DAGmodel_inl.h.

References gum::DAGmodel::_dag, and gum::ArcGraphPart::arcs().

Referenced by gum::EssentialGraph::__buildEssentialGraph(), gum::DAGmodel::__moralGraph(), gum::MarkovBlanket::hasSameStructure(), and gum::DAGmodel::hasSameStructure().

101 { return _dag.arcs(); }

gum::DAGmodel::_dag

DAG _dag

The DAG of this Directed Graphical Model.

Definition: DAGmodel.h:200

gum::ArcGraphPart::arcs

const ArcSet & arcs() const

returns the set of arcs stored within the ArcGraphPart

Definition: arcGraphPart_inl.h:36

Here is the call graph for this function:

Here is the caller graph for this function:

◆ attachSignal__()

INLINE void gum::Listener::attachSignal__ ( __sig__::ISignaler * sender )

inherited

Definition at line 33 of file listener_inl.h.

References gum::Listener::__senders.

                                                                {
     __senders.push_back(sender);
   }

◆ checkConsistency() [1/2]

template<typename GUM_SCALAR >

bool gum::BayesNetFragment< GUM_SCALAR >::checkConsistency ( NodeId id ) const

returns true if the nodeId's (local or not) cpt is consistent with its parents in the fragment

Exceptions

NotFound if the id is not in the fragment

Definition at line 278 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::cpt(), GUM_ERROR, gum::Set< Key, Alloc >::insert(), gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), and gum::DAGmodel::parents().

                                                                        {
     if (!isInstalledNode(id))
       GUM_ERROR(NotFound, "The node " << id << " is not part of this fragment");
 
     const auto& cpt = this->cpt(id);
     NodeSet     cpt_parents;
 
     for (Idx i = 1; i < cpt.nbrDim(); i++) {
       cpt_parents.insert(__bn.idFromName(cpt.variable(i).name()));
     }
 
     return (this->parents(id) == cpt_parents);
   }

Here is the call graph for this function:

◆ checkConsistency() [2/2]

template<typename GUM_SCALAR >

INLINE bool gum::BayesNetFragment< GUM_SCALAR >::checkConsistency ( ) const

noexcept

returns true if all nodes in the fragment are consistent

Definition at line 293 of file BayesNetFragment_tpl.h.

References gum::DAGmodel::nodes().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::toDot().

                                                                               {
     for (auto node : nodes())
       if (!checkConsistency(node)) return false;
 
     return true;
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ children() [1/2]

INLINE const NodeSet & gum::DAGmodel::children ( const NodeId id ) const

inherited

returns the set of nodes with arc outgoing from a given node

Note that the set of arcs returned may be empty if no arc within the ArcGraphPart is outgoing from the given node.

Parameters

id	the node which is the tail of the arcs returned

Definition at line 108 of file DAGmodel_inl.h.

References gum::DAGmodel::_dag, and gum::ArcGraphPart::children().

Referenced by gum::MarkovBlanket::__buildMarkovBlanket(), gum::DAGmodel::parents(), gum::prm::InstanceBayesNet< GUM_SCALAR >::toDot(), and gum::prm::ClassBayesNet< GUM_SCALAR >::toDot().

                                                                 {
     return _dag.children(id);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ children() [2/2]

const NodeSet& gum::DAGmodel::children ( const std::string & name ) const

inlineinherited

returns the set of nodes with arc ingoing to a given node

Note that the set of arcs returned may be empty if no arc within the ArcGraphPart is ingoing into the given node.

Parameters

id	the node toward which the arcs returned are pointing

Definition at line 162 of file DAGmodel.h.

References gum::DAGmodel::hasSameStructure(), gum::DAGmodel::idFromName(), gum::DAGmodel::log10DomainSize(), gum::DAGmodel::moralGraph(), gum::DAGmodel::operator=(), gum::DAGmodel::parents(), and gum::DAGmodel::topologicalOrder().

                                                          {
       return parents(idFromName(name));
     };

Here is the call graph for this function:

◆ completeInstantiation()

INLINE Instantiation gum::DAGmodel::completeInstantiation ( ) const

finalvirtualinherited

Get an instantiation over all the variables of the model.

Definition at line 83 of file DAGmodel_inl.h.

References gum::DAGmodel::dag(), and gum::DAGmodel::variable().

                                                       {
     Instantiation I;
 
     for (const auto node : dag())
       I << variable(node);
 
     return I;
   }

Here is the call graph for this function:

◆ cpt()

template<typename GUM_SCALAR >

INLINE const Potential< GUM_SCALAR > & gum::BayesNetFragment< GUM_SCALAR >::cpt ( NodeId varId ) const

overridevirtual

Returns the CPT of a variable.

Exceptions

NotFound If no variable's id matches varId.

Implements gum::IBayesNet< GUM_SCALAR >.

Definition at line 77 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::__localCPTs, GUM_ERROR, and gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode().

Referenced by gum::ImportanceSampling< GUM_SCALAR >::_unsharpenBN(), gum::BayesNetFragment< GUM_SCALAR >::checkConsistency(), and gum::BayesNetFragment< GUM_SCALAR >::uninstallCPT().

                                                                   {
     if (!isInstalledNode(id)) GUM_ERROR(NotFound, id << " is not installed");
 
     if (__localCPTs.exists(id))
       return *__localCPTs[id];
     else
       return __bn.cpt(id);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ dag()

INLINE const DAG & gum::DAGmodel::dag ( ) const

inherited

Returns a constant reference to the dag of this Bayes Net.

Definition at line 60 of file DAGmodel_inl.h.

References gum::DAGmodel::_dag.

Referenced by gum::DAGmodel::__moralGraph(), gum::DAGmodel::completeInstantiation(), gum::learning::DAG2BNLearner< ALLOC >::createBN(), gum::learning::genericBNLearner::Database::Database(), gum::BayesNetFragment< GUM_SCALAR >::installCPT(), gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), gum::MarginalTargetedInference< GUM_SCALAR >::MarginalTargetedInference(), gum::BayesBall::relevantPotentials(), gum::dSeparation::relevantPotentials(), gum::DAGmodel::size(), gum::BayesNetFragment< GUM_SCALAR >::toDot(), gum::DAGmodel::topologicalOrder(), gum::InfluenceDiagram< GUM_SCALAR >::toString(), and gum::BayesNetFragment< GUM_SCALAR >::whenArcDeleted().

60 { return _dag; }

gum::DAGmodel::_dag

DAG _dag

The DAG of this Directed Graphical Model.

Definition: DAGmodel.h:200

Here is the caller graph for this function:

◆ detachSignal__()

INLINE void gum::Listener::detachSignal__ ( __sig__::ISignaler * sender )

inherited

Definition at line 37 of file listener_inl.h.

References gum::Listener::__senders.

                                                                {
     auto del = std::remove(__senders.begin(), __senders.end(), sender);
 
     if (del != __senders.end()) __senders.erase(del, __senders.end());
   }

◆ dim()

template<typename GUM_SCALAR >

Size gum::IBayesNet< GUM_SCALAR >::dim ( ) const

inherited

Returns the dimension (the number of free parameters) in this bayes net.

\( dim(G)=\sum_{i \in nodes} ((r_i-1)\cdot q_i) \) where \( r_i \) is the number of instantiations of node \( i \) and \( q_i \) is the number of instantiations of its parents.

Definition at line 76 of file IBayesNet_tpl.h.

                                           {
     Size dim = 0;
 
     for (auto node : nodes()) {
       Size q = 1;
 
       for (auto parent : parents(node))
         q *= variable(parent).domainSize();
 
       dim += (variable(node).domainSize() - 1) * q;
     }
 
     return dim;
   }

◆ empty()

INLINE bool gum::DAGmodel::empty ( ) const

inherited

Retursn true if this Directed Graphical Model is empty.

Definition at line 96 of file DAGmodel_inl.h.

References gum::DAGmodel::size().

96 { return size() == 0; }

gum::DAGmodel::size

Size size() const

Returns the number of variables in this Directed Graphical Model.

Definition: DAGmodel_inl.h:93

Here is the call graph for this function:

◆ hasSameStructure()

bool gum::DAGmodel::hasSameStructure ( const DAGmodel & other )

inherited

Returns: true if all the named node are the same and all the named arcs are the same

Definition at line 119 of file DAGmodel.cpp.

References gum::DAGmodel::arcs(), gum::Set< Key, Alloc >::exists(), gum::DAGmodel::idFromName(), gum::DAGmodel::nodes(), gum::DAGmodel::size(), gum::DAGmodel::sizeArcs(), and gum::DAGmodel::variable().

Referenced by gum::DAGmodel::children().

                                                        {
     if (this == &other) return true;
 
     if (size() != other.size()) return false;
 
     if (sizeArcs() != other.sizeArcs()) return false;
 
     for (const auto& nid : nodes()) {
       try {
         other.idFromName(variable(nid).name());
       } catch (NotFound) { return false; }
     }
 
     for (const auto& arc : arcs()) {
       if (!other.arcs().exists(Arc(other.idFromName(variable(arc.tail()).name()),
                                    other.idFromName(variable(arc.head()).name()))))
         return false;
     }
 
     return true;
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ idFromName()

template<typename GUM_SCALAR >

INLINE NodeId gum::BayesNetFragment< GUM_SCALAR >::idFromName ( const std::string & name ) const

overridevirtual

Getter by name.

Exceptions

NotFound if no such name exists in the graph.

Implements gum::IBayesNet< GUM_SCALAR >.

Definition at line 114 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, GUM_ERROR, and gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode().

                                                                                {
     NodeId id = __bn.idFromName(name);
 
     if (!isInstalledNode(id))
       GUM_ERROR(NotFound, "variable " << name << " is not installed");
 
     return id;
   }

Here is the call graph for this function:

◆ installAscendants()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::installAscendants ( NodeId id )

install a node and all its ascendants

Exceptions

NotFound if the node does not exist in the referred BN

Warning: nothing happens if the node is already installed

Definition at line 162 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, and gum::BayesNetFragment< GUM_SCALAR >::installNode().

                                                                          {
     installNode(id);
 
     // bn is a dag => this will have an end ...
     for (auto pa : this->__bn.parents(id))
       installAscendants(pa);
   }

Here is the call graph for this function:

◆ installCPT()

template<typename GUM_SCALAR >

void gum::BayesNetFragment< GUM_SCALAR >::installCPT	(	NodeId	id,
		const Potential< GUM_SCALAR > *	pot
	)

install a local cpt for a node into the fragment.

This function will change the arcs from the parents to the node in order to be consistent with the new local potential.

Parameters

id	the nodeId
pot	the potential<

Exceptions

NotFound	if the id is not in the fragment
OperationNotAllowed	if the potential is not compliant with the variable or if a variable in the CPT is not a parent in the referred bn.

Definition at line 212 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), gum::Set< Key, Alloc >::contains(), gum::DAGmodel::dag(), GUM_ERROR, gum::MultiDimDecorator< GUM_SCALAR >::nbrDim(), gum::DAGmodel::parents(), gum::MultiDimDecorator< GUM_SCALAR >::variable(), and gum::BayesNetFragment< GUM_SCALAR >::variable().

Referenced by gum::ImportanceSampling< GUM_SCALAR >::_onContextualize(), and gum::ImportanceSampling< GUM_SCALAR >::_unsharpenBN().

                                                     {
     if (!dag().existsNode(id))
       GUM_ERROR(NotFound, "Node " << id << " is not installed in the fragment");
 
     if (&(pot->variable(0)) != &(variable(id))) {
       GUM_ERROR(OperationNotAllowed,
                 "The potential is not a marginal for __bn.variable <"
                    << variable(id).name() << ">");
     }
 
     const NodeSet& parents = __bn.parents(id);
 
     for (Idx i = 1; i < pot->nbrDim(); i++) {
       if (!parents.contains(__bn.idFromName(pot->variable(i).name())))
         GUM_ERROR(OperationNotAllowed,
                   "Variable <" << pot->variable(i).name()
                                << "> is not in the parents of node " << id);
     }
 
     _installCPT(id, pot);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ installMarginal()

template<typename GUM_SCALAR >

void gum::BayesNetFragment< GUM_SCALAR >::installMarginal	(	NodeId	id,
		const Potential< GUM_SCALAR > *	pot
	)

install a local marginal for a node into the fragment.

This function will remove all the arcs from the parents to the node.

Parameters

id	the nodeId
pot	the potential

Exceptions

NotFound	if the id is not in the fragment
OperationNotAllowed	if the potential is not compliant with the variable (or is not a marginal)

Definition at line 258 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), GUM_ERROR, gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), gum::MultiDimDecorator< GUM_SCALAR >::nbrDim(), and gum::MultiDimDecorator< GUM_SCALAR >::variable().

                                                     {
     if (!isInstalledNode(id)) {
       GUM_ERROR(NotFound, "The node " << id << " is not part of this fragment");
     }
 
     if (pot->nbrDim() > 1) {
       GUM_ERROR(OperationNotAllowed, "The potential is not a marginal :" << pot);
     }
 
     if (&(pot->variable(0)) != &(__bn.variable(id))) {
       GUM_ERROR(OperationNotAllowed,
                 "The potential is not a marginal for __bn.variable <"
                    << __bn.variable(id).name() << ">");
     }
 
     _installCPT(id, pot);
   }

Here is the call graph for this function:

◆ installNode()

template<typename GUM_SCALAR >

void gum::BayesNetFragment< GUM_SCALAR >::installNode ( NodeId id )

install a node referenced by its nodeId

Exceptions

NotFound if the node does not exist in the referred BN

Warning: nothing happens if the node is already installed

Definition at line 143 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::DAGmodel::_dag, gum::DAG::addArc(), gum::NodeGraphPart::addNodeWithId(), GUM_ERROR, and gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::installAscendants().

                                                             {
     if (!__bn.dag().existsNode(id))
       GUM_ERROR(NotFound, "Node " << id << " does not exist in referred BayesNet");
 
     if (!isInstalledNode(id)) {
       this->_dag.addNodeWithId(id);
 
       // adding arcs with id as a tail
       for (auto pa : this->__bn.parents(id)) {
         if (isInstalledNode(pa)) this->_dag.addArc(pa, id);
       }
 
       // addin arcs with id as a head
       for (auto son : this->__bn.children(id))
         if (isInstalledNode(son)) this->_dag.addArc(id, son);
     }
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ isInstalledNode()

template<typename GUM_SCALAR >

INLINE bool gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode ( NodeId id ) const

noexcept

check if a certain NodeId exists in the fragment

Definition at line 137 of file BayesNetFragment_tpl.h.

References gum::DAGmodel::dag(), and gum::NodeGraphPart::existsNode().

               {
     return dag().existsNode(id);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ jointProbability()

template<typename GUM_SCALAR >

GUM_SCALAR gum::IBayesNet< GUM_SCALAR >::jointProbability ( const Instantiation & i ) const

inherited

Compute a parameter of the joint probability for the BN (given an instantiation of the vars)

Warning: a variable not present in the instantiation is assumed to be instantiated to 0.

Definition at line 217 of file IBayesNet_tpl.h.

                                                                            {
     auto value = (GUM_SCALAR)1.0;
 
     GUM_SCALAR tmp;
 
     for (auto node : nodes()) {
       if ((tmp = cpt(node)[i]) == (GUM_SCALAR)0) { return (GUM_SCALAR)0; }
 
       value *= tmp;
     }
 
     return value;
   }

◆ log10DomainSize()

INLINE double gum::DAGmodel::log10DomainSize ( ) const

inherited

Returns: Returns the log10 domain size of the joint probabilty for the Directed Graphical Model

Definition at line 72 of file DAGmodel_inl.h.

References gum::DAGmodel::nodes(), and gum::DAGmodel::variable().

Referenced by gum::DAGmodel::children(), and gum::InfluenceDiagram< GUM_SCALAR >::toString().

                                          {
     double dSize = 0.0;
 
     for (const auto node : nodes()) {
       dSize += std::log10(variable(node).domainSize());
     }
 
     return dSize;
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ log2JointProbability()

template<typename GUM_SCALAR >

GUM_SCALAR gum::IBayesNet< GUM_SCALAR >::log2JointProbability ( const Instantiation & i ) const

inherited

Compute a parameter of the log joint probability for the BN (given an instantiation of the vars)

Compute a parameter of the joint probability for the BN (given an instantiation of the vars)

Warning: a variable not present in the instantiation is assumed to be instantiated to 0.

Definition at line 236 of file IBayesNet_tpl.h.

                                                                                {
     auto value = (GUM_SCALAR)0.0;
 
     GUM_SCALAR tmp;
 
     for (auto node : nodes()) {
       if ((tmp = cpt(node)[i]) == (GUM_SCALAR)0) {
         return (GUM_SCALAR)(-std::numeric_limits< double >::infinity());
       }
 
       value += log2(cpt(node)[i]);
     }
 
     return value;
   }

◆ maxNonOneParam()

template<typename GUM_SCALAR >

GUM_SCALAR gum::IBayesNet< GUM_SCALAR >::maxNonOneParam ( ) const

inherited

Returns: the biggest value (not equal to 1) in the CPTs of *this

Definition at line 132 of file IBayesNet_tpl.h.

                                                            {
     GUM_SCALAR res = 0.0;
     for (auto node : nodes()) {
       auto v = cpt(node).maxNonOne();
       if (v > res) { res = v; }
     }
     return res;
   }

◆ maxParam()

template<typename GUM_SCALAR >

GUM_SCALAR gum::IBayesNet< GUM_SCALAR >::maxParam ( ) const

inherited

Returns: the biggest value in the CPTs of *this

Definition at line 112 of file IBayesNet_tpl.h.

                                                      {
     GUM_SCALAR res = 1.0;
     for (auto node : nodes()) {
       auto v = cpt(node).max();
       if (v > res) { res = v; }
     }
     return res;
   }

◆ maxVarDomainSize()

template<typename GUM_SCALAR >

Size gum::IBayesNet< GUM_SCALAR >::maxVarDomainSize ( ) const

inherited

Returns: the biggest domainSize among the variables of *this

Definition at line 92 of file IBayesNet_tpl.h.

Referenced by gum::ImportanceSampling< GUM_SCALAR >::_onContextualize().

                                                        {
     Size res = 0;
     for (auto node : nodes()) {
       auto v = variable(node).domainSize();
       if (v > res) { res = v; }
     }
     return res;
   }

Here is the caller graph for this function:

◆ minimalCondSet() [1/2]

template<typename GUM_SCALAR >

NodeSet gum::IBayesNet< GUM_SCALAR >::minimalCondSet	(	NodeId	target,
		const NodeSet &	soids
	)		const

inherited

Definition at line 352 of file IBayesNet_tpl.h.

                                                                               {
     if (soids.contains(target)) return NodeSet({target});
 
     NodeSet res;
     NodeSet alreadyVisitedUp;
     NodeSet alreadyVisitedDn;
     alreadyVisitedDn << target;
     alreadyVisitedUp << target;
 
     for (auto fath : _dag.parents(target))
       __minimalCondSetVisitUp(
          fath, soids, res, alreadyVisitedUp, alreadyVisitedDn);
     for (auto chil : _dag.children(target))
       __minimalCondSetVisitDn(
          chil, soids, res, alreadyVisitedUp, alreadyVisitedDn);
     return res;
   }

◆ minimalCondSet() [2/2]

template<typename GUM_SCALAR >

NodeSet gum::IBayesNet< GUM_SCALAR >::minimalCondSet	(	const NodeSet &	targets,
		const NodeSet &	soids
	)		const

inherited

Definition at line 372 of file IBayesNet_tpl.h.

                                                                               {
     NodeSet res;
     for (auto node : targets) {
       res += minimalCondSet(node, soids);
     }
     return res;
   }

◆ minNonZeroParam()

template<typename GUM_SCALAR >

GUM_SCALAR gum::IBayesNet< GUM_SCALAR >::minNonZeroParam ( ) const

inherited

Returns: the smallest value (not equal to 0) in the CPTs of *this

Definition at line 122 of file IBayesNet_tpl.h.

Referenced by gum::ImportanceSampling< GUM_SCALAR >::_onContextualize().

                                                             {
     GUM_SCALAR res = 1.0;
     for (auto node : nodes()) {
       auto v = cpt(node).minNonZero();
       if (v < res) { res = v; }
     }
     return res;
   }

Here is the caller graph for this function:

◆ minParam()

template<typename GUM_SCALAR >

GUM_SCALAR gum::IBayesNet< GUM_SCALAR >::minParam ( ) const

inherited

Returns: the smallest value in the CPTs of *this

Definition at line 102 of file IBayesNet_tpl.h.

                                                      {
     GUM_SCALAR res = 1.0;
     for (auto node : nodes()) {
       auto v = cpt(node).min();
       if (v < res) { res = v; }
     }
     return res;
   }

◆ moralGraph()

const UndiGraph & gum::DAGmodel::moralGraph ( bool clear = true ) const

inherited

The node's id are coherent with the variables and nodes of the topology.

Parameters

clear If false returns the previously created moral graph.

Definition at line 99 of file DAGmodel.cpp.

References gum::DAGmodel::__moralGraph(), gum::DAGmodel::__mutableMoralGraph, and gum::UndiGraph::clear().

Referenced by gum::prm::SVED< GUM_SCALAR >::__eliminateNodes(), gum::prm::SVE< GUM_SCALAR >::__eliminateNodes(), gum::prm::SVED< GUM_SCALAR >::__eliminateNodesWithEvidence(), gum::prm::SVE< GUM_SCALAR >::__eliminateNodesWithEvidence(), gum::prm::SVED< GUM_SCALAR >::__initLiftedNodes(), gum::prm::SVE< GUM_SCALAR >::__initLiftedNodes(), and gum::DAGmodel::children().

                                                         {
     if (clear
         || (__mutableMoralGraph == nullptr)) {   // we have to call _moralGraph
       if (__mutableMoralGraph == nullptr) {
         __mutableMoralGraph = new UndiGraph();
       } else {
         // clear is True ,__mutableMoralGraph exists
         __mutableMoralGraph->clear();
       }
 
       __moralGraph();
     }
 
     return *__mutableMoralGraph;
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ nodeId()

template<typename GUM_SCALAR >

INLINE NodeId gum::BayesNetFragment< GUM_SCALAR >::nodeId ( const DiscreteVariable & var ) const

overridevirtual

Return id node from discrete var pointer.

Exceptions

NotFound If no variable matches var.

Implements gum::IBayesNet< GUM_SCALAR >.

Definition at line 103 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, GUM_ERROR, gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), and gum::Variable::name().

                                                                                  {
     NodeId id = __bn.nodeId(var);
 
     if (!isInstalledNode(id))
       GUM_ERROR(NotFound, "variable " << var.name() << " is not installed");
 
     return id;
   }

Here is the call graph for this function:

◆ nodes()

INLINE const NodeGraphPart & gum::DAGmodel::nodes ( ) const

inherited

Returns a constant reference to the dag of this Bayes Net.

Definition at line 112 of file DAGmodel_inl.h.

References gum::DAGmodel::_dag.

Referenced by gum::credal::CredalNet< GUM_SCALAR >::__bnCopy(), gum::EssentialGraph::__buildEssentialGraph(), gum::MarkovBlanket::__buildMarkovBlanket(), gum::DAGmodel::__moralGraph(), gum::credal::CredalNet< GUM_SCALAR >::__sort_varType(), gum::credal::CNMonteCarloSampling< GUM_SCALAR, BNInferenceEngine >::__verticesSampling(), gum::ImportanceSampling< GUM_SCALAR >::_unsharpenBN(), gum::BayesNetFactory< GUM_SCALAR >::BayesNetFactory(), gum::BayesNetFragment< GUM_SCALAR >::checkConsistency(), gum::MCBayesNetGenerator< GUM_SCALAR, ICPTGenerator, ICPTDisturber >::disturbBN(), gum::Estimator< GUM_SCALAR >::Estimator(), gum::getMaxModality(), gum::DAGmodel::hasSameStructure(), gum::DAGmodel::log10DomainSize(), gum::prm::InstanceBayesNet< GUM_SCALAR >::modalities(), gum::prm::ClassBayesNet< GUM_SCALAR >::modalities(), gum::Estimator< GUM_SCALAR >::setFromBN(), gum::prm::InstanceBayesNet< GUM_SCALAR >::toDot(), gum::prm::ClassBayesNet< GUM_SCALAR >::toDot(), gum::credal::CredalNet< GUM_SCALAR >::toString(), and gum::BayesNetFragment< GUM_SCALAR >::~BayesNetFragment().

                                                     {
     return (NodeGraphPart&)_dag;
   }

Here is the caller graph for this function:

◆ operator!=()

template<typename GUM_SCALAR>

bool gum::IBayesNet< GUM_SCALAR >::operator!= ( const IBayesNet< GUM_SCALAR > & from ) const

inherited

Returns: Returns false if the src and this are equal.

Definition at line 300 of file IBayesNet_tpl.h.

                                                                       {
     return !this->operator==(from);
   }

◆ operator==()

template<typename GUM_SCALAR>

bool gum::IBayesNet< GUM_SCALAR >::operator== ( const IBayesNet< GUM_SCALAR > & from ) const

inherited

This operator compares 2 BNs !

Warning: To identify nodes between BNs, it is assumed that they share the same name.

Returns: true if the src and this are equal.

Definition at line 253 of file IBayesNet_tpl.h.

                                                                       {
     if (size() != from.size()) { return false; }
 
     if (sizeArcs() != from.sizeArcs()) { return false; }
 
     // alignment of variables between the 2 BNs
     Bijection< const DiscreteVariable*, const DiscreteVariable* > alignment;
 
     for (auto node : nodes()) {
       try {
         alignment.insert(&variable(node),
                          &from.variableFromName(variable(node).name()));
       } catch (NotFound&) {
         // a name is not found in from
         return false;
       }
     }
 
     for (auto node : nodes()) {
       NodeId fromnode = from.idFromName(variable(node).name());
 
       if (cpt(node).nbrDim() != from.cpt(fromnode).nbrDim()) { return false; }
 
       if (cpt(node).domainSize() != from.cpt(fromnode).domainSize()) {
         return false;
       }
 
       Instantiation i(cpt(node));
       Instantiation j(from.cpt(fromnode));
 
       for (i.setFirst(); !i.end(); i.inc()) {
         for (Idx indice = 0; indice < cpt(node).nbrDim(); ++indice) {
           const DiscreteVariable* p = &(i.variable(indice));
           j.chgVal(*(alignment.second(p)), i.val(*p));
         }
 
         if (std::pow(cpt(node).get(i) - from.cpt(fromnode).get(j), (GUM_SCALAR)2)
             > (GUM_SCALAR)1e-6) {
           return false;
         }
       }
     }
 
     return true;
   }

◆ parents() [1/2]

INLINE const NodeSet & gum::DAGmodel::parents ( const NodeId id ) const

inherited

returns the set of nodes with arc ingoing to a given node

Note that the set of arcs returned may be empty if no arc within the ArcGraphPart is ingoing into the given node.

Parameters

id	the node toward which the arcs returned are pointing

Definition at line 103 of file DAGmodel_inl.h.

References gum::DAGmodel::_dag, and gum::ArcGraphPart::parents().

Referenced by gum::MarkovBlanket::__buildMarkovBlanket(), gum::DAGmodel::__moralGraph(), gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), gum::BayesNetFragment< GUM_SCALAR >::checkConsistency(), gum::DAGmodel::children(), gum::BayesNetFragment< GUM_SCALAR >::installCPT(), gum::DAGmodel::parents(), gum::prm::InstanceBayesNet< GUM_SCALAR >::toDot(), and gum::prm::ClassBayesNet< GUM_SCALAR >::toDot().

                                                                {
     return _dag.parents(id);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ parents() [2/2]

const NodeSet& gum::DAGmodel::parents ( const std::string & name ) const

inlineinherited

returns the set of nodes with arc ingoing to a given node

Note that the set of arcs returned may be empty if no arc within the ArcGraphPart is ingoing into the given node.

Parameters

id	the node toward which the arcs returned are pointing

Definition at line 153 of file DAGmodel.h.

References gum::DAGmodel::children(), gum::DAGmodel::idFromName(), and gum::DAGmodel::parents().

                                                         {
       return parents(idFromName(name));
     };

Here is the call graph for this function:

◆ property()

INLINE const std::string & gum::DAGmodel::property ( const std::string & name ) const

inherited

Return the value of the property name of this DAGModel.

Exceptions

NotFound Raised if no name property is found.

Definition at line 34 of file DAGmodel_inl.h.

References gum::DAGmodel::__properties(), and GUM_ERROR.

Referenced by gum::InfluenceDiagram< GUM_SCALAR >::toDot().

                                                                  {
     try {
       return __properties()[name];
     } catch (NotFound&) {
       std::string msg = "The following property does not exists: ";
       GUM_ERROR(NotFound, msg + name);
     }
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ propertyWithDefault()

INLINE const std::string & gum::DAGmodel::propertyWithDefault	(	const std::string &	name,
		const std::string &	byDefault
	)		const

inherited

Return the value of the property name of this DAGModel.

return byDefault if the property name is not found

Definition at line 45 of file DAGmodel_inl.h.

References gum::DAGmodel::__properties().

                                                                      {
     try {
       return __properties()[name];
     } catch (NotFound&) { return byDefault; }
   }

Here is the call graph for this function:

◆ setProperty()

INLINE void gum::DAGmodel::setProperty	(	const std::string &	name,
		const std::string &	value
	)

inherited

Add or change a property of this DAGModel.

Definition at line 53 of file DAGmodel_inl.h.

References gum::DAGmodel::__properties(), and gum::HashTable< Key, Val, Alloc >::insert().

Referenced by gum::BayesNet< double >::fastPrototype().

                                                                           {
     try {
       __properties()[name] = value;
     } catch (NotFound&) { __properties().insert(name, value); }
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ size()

INLINE Size gum::DAGmodel::size ( ) const

inherited

Returns the number of variables in this Directed Graphical Model.

Definition at line 93 of file DAGmodel_inl.h.

References gum::DAGmodel::dag(), and gum::NodeGraphPart::size().

Referenced by gum::credal::CredalNet< GUM_SCALAR >::__initCNNets(), gum::InfluenceDiagram< GUM_SCALAR >::decisionNodeSize(), gum::DAGmodel::empty(), gum::MarkovBlanket::hasSameStructure(), gum::DAGmodel::hasSameStructure(), gum::IBayesNet< double >::operator==(), gum::prm::InstanceBayesNet< GUM_SCALAR >::toDot(), and gum::prm::ClassBayesNet< GUM_SCALAR >::toDot().

93 { return dag().size(); }

gum::NodeGraphPart::size

Size size() const

alias for sizeNodes

Definition: nodeGraphPart_inl.h:281

gum::DAGmodel::dag

const DAG & dag() const

Returns a constant reference to the dag of this Bayes Net.

Definition: DAGmodel_inl.h:60

Here is the call graph for this function:

Here is the caller graph for this function:

◆ sizeArcs()

INLINE Size gum::DAGmodel::sizeArcs ( ) const

inherited

Returns the number of arcs in this Directed Graphical Model.

Definition at line 99 of file DAGmodel_inl.h.

References gum::DAGmodel::_dag, and gum::ArcGraphPart::sizeArcs().

Referenced by gum::MarkovBlanket::hasSameStructure(), gum::DAGmodel::hasSameStructure(), and gum::IBayesNet< double >::operator==().

99 { return _dag.sizeArcs(); }

gum::DAGmodel::_dag

DAG _dag

The DAG of this Directed Graphical Model.

Definition: DAGmodel.h:200

gum::ArcGraphPart::sizeArcs

Size sizeArcs() const

indicates the number of arcs stored within the ArcGraphPart

Definition: arcGraphPart_inl.h:34

Here is the call graph for this function:

Here is the caller graph for this function:

◆ toDot()

template<typename GUM_SCALAR >

std::string gum::BayesNetFragment< GUM_SCALAR >::toDot ( ) const

overridevirtual

creates a dot representing the whole referred BN hilighting the fragment.

Returns: Returns a dot representation of this fragment

Reimplemented from gum::IBayesNet< GUM_SCALAR >.

Definition at line 301 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::__localCPTs, gum::BayesNetFragment< GUM_SCALAR >::checkConsistency(), gum::DAGmodel::dag(), and gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode().

                                                         {
     std::stringstream output;
     output << "digraph \"";
 
     std::string bn_name;
 
     static std::string inFragmentStyle = "fillcolor=\"#ffffaa\","
                                          "color=\"#000000\","
                                          "fontcolor=\"#000000\"";
     static std::string styleWithLocalCPT = "fillcolor=\"#ffddaa\","
                                            "color=\"#000000\","
                                            "fontcolor=\"#000000\"";
     static std::string notConsistantStyle = "fillcolor=\"#ff0000\","
                                             "color=\"#000000\","
                                             "fontcolor=\"#ffff00\"";
     static std::string outFragmentStyle = "fillcolor=\"#f0f0f0\","
                                           "color=\"#f0f0f0\","
                                           "fontcolor=\"#000000\"";
 
     try {
       bn_name = __bn.property("name");
     } catch (NotFound&) { bn_name = "no_name"; }
 
     bn_name = "Fragment of " + bn_name;
 
     output << bn_name << "\" {" << std::endl;
     output << "  graph [bgcolor=transparent,label=\"" << bn_name << "\"];"
            << std::endl;
     output << "  node [style=filled];" << std::endl << std::endl;
 
     for (auto node : __bn.nodes()) {
       output << "\"" << __bn.variable(node).name() << "\" [comment=\"" << node
              << ":" << __bn.variable(node) << ", \"";
 
       if (isInstalledNode(node)) {
         if (!checkConsistency(node)) {
           output << notConsistantStyle;
         } else if (__localCPTs.exists(node))
           output << styleWithLocalCPT;
         else
           output << inFragmentStyle;
       } else
         output << outFragmentStyle;
 
       output << "];" << std::endl;
     }
 
     output << std::endl;
 
     std::string tab = "  ";
 
     for (auto node : __bn.nodes()) {
       if (__bn.children(node).size() > 0) {
         for (auto child : __bn.children(node)) {
           output << tab << "\"" << __bn.variable(node).name() << "\" -> "
                  << "\"" << __bn.variable(child).name() << "\" [";
 
           if (dag().existsArc(Arc(node, child)))
             output << inFragmentStyle;
           else
             output << outFragmentStyle;
 
           output << "];" << std::endl;
         }
       }
     }
 
     output << "}" << std::endl;
 
     return output.str();
   }

Here is the call graph for this function:

◆ topologicalOrder()

const Sequence< NodeId > & gum::DAGmodel::topologicalOrder ( bool clear = true ) const

inherited

The topological order stays the same as long as no variable or arcs are added or erased src the topology.

Parameters

clear If false returns the previously created topology.

Definition at line 115 of file DAGmodel.cpp.

References gum::DAGmodel::dag(), and gum::DiGraph::topologicalOrder().

Referenced by gum::EssentialGraph::__buildEssentialGraph(), gum::InfluenceDiagramGenerator< GUM_SCALAR >::__checkTemporalOrder(), gum::DAGmodel::children(), gum::InfluenceDiagram< GUM_SCALAR >::decisionOrderExists(), and gum::InfluenceDiagram< GUM_SCALAR >::getDecisionOrder().

                                                                        {
     return this->dag().topologicalOrder(clear);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ toString()

template<typename GUM_SCALAR >

INLINE std::string gum::IBayesNet< GUM_SCALAR >::toString ( ) const

inherited

Returns: Returns a string representation of this IBayesNet.

Definition at line 142 of file IBayesNet_tpl.h.

Referenced by gum::operator<<().

                                                            {
     Size   param = 0;
     double dSize = log10DomainSize();
 
     for (auto node : nodes())
       param += cpt(node).content()->realSize();
 
     double compressionRatio = log10(1.0 * param) - dSize;
 
     std::stringstream s;
     s << "BN{nodes: " << size() << ", arcs: " << dag().sizeArcs() << ", ";
 
     if (dSize > 6)
       s << "domainSize: 10^" << dSize;
     else
       s << "domainSize: " << std::round(std::pow(10.0, dSize));
 
     s << ", parameters: " << param << ", compression ratio: ";
 
     if (compressionRatio > -3)
       s << trunc(100.0 - std::pow(10.0, compressionRatio + 2.0));
     else
       s << "100-10^" << compressionRatio + 2.0;
 
     s << "% }";
 
     return s.str();
   }

Here is the caller graph for this function:

◆ uninstallCPT()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::uninstallCPT ( NodeId id )

noexcept

uninstall a local CPT.

Warning: Nothing happens if no local CPT for this nodeId or if the node is not installed.

Definition at line 242 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, gum::BayesNetFragment< GUM_SCALAR >::__localCPTs, gum::BayesNetFragment< GUM_SCALAR >::_installArc(), gum::BayesNetFragment< GUM_SCALAR >::_uninstallCPT(), gum::BayesNetFragment< GUM_SCALAR >::cpt(), gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), gum::MultiDimDecorator< GUM_SCALAR >::nbrDim(), and gum::MultiDimDecorator< GUM_SCALAR >::variable().

Referenced by gum::ImportanceSampling< GUM_SCALAR >::_onContextualize(), and gum::BayesNetFragment< GUM_SCALAR >::uninstallNode().

                                                                              {
     if (__localCPTs.exists(id)) {
       _uninstallCPT(id);
 
       // re-create arcs from referred potential
       const Potential< GUM_SCALAR >& pot = cpt(id);
 
       for (Idx i = 1; i < pot.nbrDim(); i++) {
         NodeId parent = __bn.idFromName(pot.variable(i).name());
 
         if (isInstalledNode(parent)) _installArc(parent, id);
       }
     }
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ uninstallNode()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::uninstallNode ( NodeId id )

noexcept

uninstall a node referenced by its nodeId

Warning: nothing happens if the node is not installed

Definition at line 171 of file BayesNetFragment_tpl.h.

References gum::DAGmodel::_dag, gum::DiGraph::eraseNode(), gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode(), and gum::BayesNetFragment< GUM_SCALAR >::uninstallCPT().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::whenNodeDeleted().

                                                                               {
     if (isInstalledNode(id)) {
       this->_dag.eraseNode(id);
       uninstallCPT(id);
     }
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ variable()

template<typename GUM_SCALAR >

INLINE const DiscreteVariable & gum::BayesNetFragment< GUM_SCALAR >::variable ( NodeId id ) const

overridevirtual

Returns a constant reference over a variabe given it's node id.

Exceptions

NotFound If no variable's id matches varId.

Implements gum::IBayesNet< GUM_SCALAR >.

Definition at line 95 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, GUM_ERROR, and gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode().

Referenced by gum::BayesNetFragment< GUM_SCALAR >::installCPT().

                                                                        {
     if (!isInstalledNode(id)) GUM_ERROR(NotFound, id << " is not installed");
 
     return __bn.variable(id);
   }

Here is the call graph for this function:

Here is the caller graph for this function:

◆ variableFromName()

template<typename GUM_SCALAR >

INLINE const DiscreteVariable & gum::BayesNetFragment< GUM_SCALAR >::variableFromName ( const std::string & name ) const

overridevirtual

Getter by name.

Exceptions

NotFound if no such name exists in the graph.

Implements gum::IBayesNet< GUM_SCALAR >.

Definition at line 124 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::__bn, GUM_ERROR, and gum::BayesNetFragment< GUM_SCALAR >::isInstalledNode().

                                   {
     NodeId id = __bn.idFromName(name);
 
     if (!isInstalledNode(id))
       GUM_ERROR(NotFound, "variable " << name << " is not installed");
 
     return __bn.variable(id);
   }

Here is the call graph for this function:

◆ variableNodeMap()

template<typename GUM_SCALAR >

INLINE const VariableNodeMap & gum::BayesNetFragment< GUM_SCALAR >::variableNodeMap ( ) const

overridevirtual

Returns a constant reference to the VariableNodeMap of this BN.

Implements gum::IBayesNet< GUM_SCALAR >.

Definition at line 88 of file BayesNetFragment_tpl.h.

References GUM_ERROR.

                                                                      {
     GUM_ERROR(FatalError,
               "Not implemented yet. please use referent bayesnet method");
   }

◆ whenArcAdded()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::whenArcAdded	(	const void *	src,
		NodeId	from,
		NodeId	to
	)

overridevirtualnoexcept

the action to take when a new arc is inserted into the graph

Parameters

src	the object that sent the signal
from	the id of tail of the new arc inserted into the graph
to	the id of head of the new arc inserted into the graph

Implements gum::DiGraphListener.

Definition at line 60 of file BayesNetFragment_tpl.h.

                                                                                {
     // nothing to do
   }

◆ whenArcDeleted()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::whenArcDeleted	(	const void *	src,
		NodeId	from,
		NodeId	to
	)

overridevirtualnoexcept

the action to take when an arc has just been removed from the graph

Parameters

src	the object that sent the signal
from	the id of tail of the arc removed from the graph
to	the id of head of the arc removed from the graph

Implements gum::DiGraphListener.

Definition at line 66 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::_uninstallArc(), and gum::DAGmodel::dag().

                                                                                  {
     if (dag().existsArc(from, to)) _uninstallArc(from, to);
   }

Here is the call graph for this function:

◆ whenNodeAdded()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::whenNodeAdded	(	const void *	src,
		NodeId	id
	)

overridevirtualnoexcept

the action to take when a new node is inserted into the graph

Parameters

src	the object that sent the signal
id	the id of the new node inserted into the graph

Implements gum::DiGraphListener.

Definition at line 50 of file BayesNetFragment_tpl.h.

                                                                                 {
     // nothing to do
   }

◆ whenNodeDeleted()

template<typename GUM_SCALAR >

INLINE void gum::BayesNetFragment< GUM_SCALAR >::whenNodeDeleted	(	const void *	src,
		NodeId	id
	)

overridevirtualnoexcept

the action to take when a node has just been removed from the graph

Parameters

src	the object that sent the signal
id	the id of the node has just been removed from the graph

Implements gum::DiGraphListener.

Definition at line 55 of file BayesNetFragment_tpl.h.

References gum::BayesNetFragment< GUM_SCALAR >::uninstallNode().

                                                                                   {
     uninstallNode(id);
   }

Here is the call graph for this function:

Member Data Documentation

◆ __bn

template<typename GUM_SCALAR>

const IBayesNet< GUM_SCALAR >& gum::BayesNetFragment< GUM_SCALAR >::__bn

private

The referred BayesNet.

Definition at line 75 of file BayesNetFragment.h.

◆ __localCPTs

template<typename GUM_SCALAR>

NodeProperty< const Potential< GUM_SCALAR >* > gum::BayesNetFragment< GUM_SCALAR >::__localCPTs

private

Mapping between the variable's id and their CPT specific to this Fragment.

Definition at line 79 of file BayesNetFragment.h.

Referenced by gum::BayesNetFragment< GUM_SCALAR >::_installCPT(), gum::BayesNetFragment< GUM_SCALAR >::_uninstallCPT(), gum::BayesNetFragment< GUM_SCALAR >::cpt(), gum::BayesNetFragment< GUM_SCALAR >::toDot(), gum::BayesNetFragment< GUM_SCALAR >::uninstallCPT(), and gum::BayesNetFragment< GUM_SCALAR >::~BayesNetFragment().

◆ _dag

DAG gum::DAGmodel::_dag

protectedinherited

The DAG of this Directed Graphical Model.

Definition at line 200 of file DAGmodel.h.

◆ _graph

DiGraph* gum::DiGraphListener::_graph

protectedinherited

the graph to listen to

Definition at line 96 of file diGraphListener.h.

Referenced by gum::DiGraphListener::DiGraphListener().

The documentation for this class was generated from the following files:

agrum/BN/BayesNetFragment.h
agrum/BN/BayesNetFragment_tpl.h

Public Member Functions

Protected Attributes

Protected Member Functions

Detailed Description

template<typename GUM_SCALAR> class gum::BayesNetFragment< GUM_SCALAR >

Constructor & Destructor Documentation

◆ BayesNetFragment() [1/4]

◆ BayesNetFragment() [2/4]

◆ BayesNetFragment() [3/4]

◆ BayesNetFragment() [4/4]

◆ ~BayesNetFragment()

Member Function Documentation

◆ _installArc()

◆ _installCPT()

◆ _uninstallArc()

◆ _uninstallCPT()

◆ arcs()

◆ attachSignal__()

◆ checkConsistency() [1/2]

◆ checkConsistency() [2/2]

◆ children() [1/2]

◆ children() [2/2]

◆ completeInstantiation()

◆ cpt()

◆ dag()

◆ detachSignal__()

◆ dim()

◆ empty()

◆ hasSameStructure()

◆ idFromName()

◆ installAscendants()

◆ installCPT()

◆ installMarginal()

◆ installNode()

◆ isInstalledNode()

◆ jointProbability()

◆ log10DomainSize()

◆ log2JointProbability()

◆ maxNonOneParam()

◆ maxParam()

◆ maxVarDomainSize()

◆ minimalCondSet() [1/2]

◆ minimalCondSet() [2/2]

◆ minNonZeroParam()

◆ minParam()

◆ moralGraph()

◆ nodeId()

◆ nodes()

◆ operator!=()

◆ operator==()

◆ parents() [1/2]

◆ parents() [2/2]

◆ property()

◆ propertyWithDefault()

◆ setProperty()

◆ size()

◆ sizeArcs()

◆ toDot()

◆ topologicalOrder()

◆ toString()

◆ uninstallCPT()

◆ uninstallNode()

◆ variable()

◆ variableFromName()

◆ variableNodeMap()

◆ whenArcAdded()

◆ whenArcDeleted()

◆ whenNodeAdded()

◆ whenNodeDeleted()

Member Data Documentation

◆ __bn

◆ __localCPTs

◆ _dag

◆ _graph

template<typename GUM_SCALAR>
class gum::BayesNetFragment< GUM_SCALAR >