The basic class for computing the next graph changes possible in a structure learning algorithm. More...

#include <graphChangesGeneratorOnSubDiGraph.h>

Inheritance diagram for gum::learning::GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT >:

Collaboration diagram for gum::learning::GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT >:

Public Member Functions
Constructors / Destructors
	GraphChangesGeneratorOnSubDiGraph (STRUCT_CONSTRAINT &constraint)
	default constructor More...

	GraphChangesGeneratorOnSubDiGraph (const GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT > &from)
	copy constructor More...

	GraphChangesGeneratorOnSubDiGraph (GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT > &&from)
	move operator More...

virtual	~GraphChangesGeneratorOnSubDiGraph ()
	destructor More...

Operators
GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT > &	operator= (const GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT > &from)
	copy operator More...

GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT > &	operator= (GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT > &&from)
	move operator More...

Iterators
iterator	begin () const
	returns an (unsafe) iterator on the beginning of the list of operators More...

const iterator &	end () const
	returns an (unsafe) iterator on the end of the list of operators More...

Accessors / Modifiers
STRUCT_CONSTRAINT &	constraint () const noexcept
	returns the constraint that is used by the generator More...

void	setGraph (const DiGraph &graph)
	sets a new graph from which the operator will compute possible changes More...

void	setTargets (const NodeSet &nodes)
	assign a set of target nodes More...

void	addTarget (NodeId node)
	adds a new target node More...

void	eraseTarget (NodeId node)
	removes a target More...

void	setTails (const NodeSet &nodes)
	assign a set of "tail" nodes More...

void	setTails (Size nb_nodes)
	assign a set of "tail" nodes from 0 to nb_nodes - 1 More...

void	addTail (NodeId node)
	adds a new "tail" node More...

void	eraseTail (NodeId node)
	removes a tail node More...

void	modifyGraph (const ArcAddition &change)
	notify the operator set of a change applied to the graph More...

void	modifyGraph (const ArcDeletion &change)
	notify the operator set of a change applied to the graph More...

void	modifyGraph (const ArcReversal &change)
	notify the operator set of a change applied to the graph More...

void	modifyGraph (const GraphChange &change)
	notify the operator set of a change applied to the graph More...

void	clearChanges () noexcept
	empty the set of possible change operators that can be applied More...

void	notifyGetCompleted ()
	notifies the generator that we have parsed all its legal changes More...

void	setMaxNbThreads (Size nb) noexcept
	sets the maximum number of threads used to compute the set of changes More...

Public Types
using	iterator = typename Set< GraphChange >::const_iterator
	the iterator for parsing the list of possible graph change operators More...

using	const_iterator = iterator
	the const iterator for parsing the list of graph change operators More...

Protected Attributes
STRUCT_CONSTRAINT *	constraint_
	a reference on the structural constraint used to restrict the changes More...

NodeSet	target_nodes_
	the set of target nodes More...

NodeSet	tail_nodes_
	the tail nodes (other extremities than the targets) More...

Set< GraphChange >	legal_changes_
	the current set of operators More...

Protected Member Functions
void	createChanges_ ()
	create the set of legal and illegal changes from a given graph More...

Detailed Description

template<typename STRUCT_CONSTRAINT>
class gum::learning::GraphChangesGeneratorOnSubDiGraph< STRUCT_CONSTRAINT >

The basic class for computing the next graph changes possible in a structure learning algorithm.

Structure learning algorithm try different modifications of the graph. Class GraphChangesGeneratorOnSubDiGraph provides a simple way to compute those that we wish to perform when we wish to limit the set of changes over a subgraph of the graph we learn. For instance, we may wish to relearn an already partially learnt structure just around nodes A,B,C. In this case, we can set the set of target nodes of the GraphChangesGeneratorOnSubDiGraph to {A,B,C} and the set of "tail" nodes as all the nodes of the graph. In this case, only the arcs whose heads are A, B or C will be trnasmitted to the learning algorithm.

Basically, the idea is to use method setGraph at the beginning of the structure learning in order to initialize the possible set of operators. Then, parse this set using a for ( auto iter = operator_set.begin (); iter != operator_set.end (); ++iter ) loop and compute the scores induced by these changes. When this is done, flush the operator set by calling method clearChanges. Then iterate changes and after each new change applied, used again the iterator for loop, and so on. Note that, whenever you execute method modifyGraph, this will automatically flush the current list of changes and put into the list only the changes that are affected by the graph modification.

Definition at line 94 of file graphChangesGeneratorOnSubDiGraph.h.