gum::learning::Miic Class Reference

The miic learning algorithm. More...

#include <Miic.h>

Inheritance diagram for gum::learning::Miic:

Collaboration diagram for gum::learning::Miic:

Public Attributes
Signaler3< Size, double, double >	onProgress
	Progression, error and time. More...
Signaler1< std::string >	onStop
	Criteria messageApproximationScheme. More...

Public Member Functions
Miic &	operator= (const Miic &from)
	copy operator More...
Miic &	operator= (Miic &&from)
	move operator More...
Constructors / Destructors
	Miic ()
	default constructor More...
	Miic (int maxLog)
	default constructor with maxLog More...
	Miic (const Miic &from)
	copy constructor More...
	Miic (Miic &&from)
	move constructor More...
	~Miic () override
	destructor More...
Accessors / Modifiers
MixedGraph	learnMixedStructure (CorrectedMutualInformation<> &mutualInformation, MixedGraph graph)
	learns the structure of an Essential Graph More...
DAG	learnStructure (CorrectedMutualInformation<> &I, MixedGraph graph)
	learns the structure of a Bayesian network, i.e. a DAG, by first learning an Essential graph and then directing the remaining edges. More...
template<typename GUM_SCALAR = double, typename GRAPH_CHANGES_SELECTOR , typename PARAM_ESTIMATOR >
BayesNet< GUM_SCALAR >	learnBN (GRAPH_CHANGES_SELECTOR &selector, PARAM_ESTIMATOR &estimator, DAG initial_dag=DAG())
	learns the structure and the parameters of a BN More...
const std::vector< Arc >	latentVariables () const
	get the list of arcs hiding latent variables More...
void	setMiicBehaviour ()
	Sets the orientation phase to follow the one of the MIIC algorithm. More...
void	set3of2Behaviour ()
	Sets the orientation phase to follow the one of the 3off2 algorithm. More...
void	addConstraints (HashTable< std::pair< NodeId, NodeId >, char > constraints)
	Set a ensemble of constraints for the orientation phase. More...
Getters and setters
void	setEpsilon (double eps)
	Given that we approximate f(t), stopping criterion on |f(t+1)-f(t)|. More...
double	epsilon () const
	Returns the value of epsilon. More...
void	disableEpsilon ()
	Disable stopping criterion on epsilon. More...
void	enableEpsilon ()
	Enable stopping criterion on epsilon. More...
bool	isEnabledEpsilon () const
	Returns true if stopping criterion on epsilon is enabled, false otherwise. More...
void	setMinEpsilonRate (double rate)
	Given that we approximate f(t), stopping criterion on d/dt(|f(t+1)-f(t)|). More...
double	minEpsilonRate () const
	Returns the value of the minimal epsilon rate. More...
void	disableMinEpsilonRate ()
	Disable stopping criterion on epsilon rate. More...
void	enableMinEpsilonRate ()
	Enable stopping criterion on epsilon rate. More...
bool	isEnabledMinEpsilonRate () const
	Returns true if stopping criterion on epsilon rate is enabled, false otherwise. More...
void	setMaxIter (Size max)
	Stopping criterion on number of iterations. More...
Size	maxIter () const
	Returns the criterion on number of iterations. More...
void	disableMaxIter ()
	Disable stopping criterion on max iterations. More...
void	enableMaxIter ()
	Enable stopping criterion on max iterations. More...
bool	isEnabledMaxIter () const
	Returns true if stopping criterion on max iterations is enabled, false otherwise. More...
void	setMaxTime (double timeout)
	Stopping criterion on timeout. More...
double	maxTime () const
	Returns the timeout (in seconds). More...
double	currentTime () const
	Returns the current running time in second. More...
void	disableMaxTime ()
	Disable stopping criterion on timeout. More...
void	enableMaxTime ()
	Enable stopping criterion on timeout. More...
bool	isEnabledMaxTime () const
	Returns true if stopping criterion on timeout is enabled, false otherwise. More...
void	setPeriodSize (Size p)
	How many samples between two stopping is enable. More...
Size	periodSize () const
	Returns the period size. More...
void	setVerbosity (bool v)
	Set the verbosity on (true) or off (false). More...
bool	verbosity () const
	Returns true if verbosity is enabled. More...
ApproximationSchemeSTATE	stateApproximationScheme () const
	Returns the approximation scheme state. More...
Size	nbrIterations () const
	Returns the number of iterations. More...
const std::vector< double > &	history () const
	Returns the scheme history. More...
void	initApproximationScheme ()
	Initialise the scheme. More...
bool	startOfPeriod ()
	Returns true if we are at the beginning of a period (compute error is mandatory). More...
void	updateApproximationScheme (unsigned int incr=1)
	Update the scheme w.r.t the new error and increment steps. More...
Size	remainingBurnIn ()
	Returns the remaining burn in. More...
void	stopApproximationScheme ()
	Stop the approximation scheme. More...
bool	continueApproximationScheme (double error)
	Update the scheme w.r.t the new error. More...
Getters and setters
std::string	messageApproximationScheme () const
	Returns the approximation scheme message. More...

Public Types
enum	ApproximationSchemeSTATE : char {   ApproximationSchemeSTATE::Undefined, ApproximationSchemeSTATE::Continue, ApproximationSchemeSTATE::Epsilon, ApproximationSchemeSTATE::Rate,   ApproximationSchemeSTATE::Limit, ApproximationSchemeSTATE::TimeLimit, ApproximationSchemeSTATE::Stopped }
	The different state of an approximation scheme. More...

Protected Attributes
double	current_epsilon_
	Current epsilon. More...
double	last_epsilon_
	Last epsilon value. More...
double	current_rate_
	Current rate. More...
Size	current_step_
	The current step. More...
Timer	timer_
	The timer. More...
ApproximationSchemeSTATE	current_state_
	The current state. More...
std::vector< double >	history_
	The scheme history, used only if verbosity == true. More...
double	eps_
	Threshold for convergence. More...
bool	enabled_eps_
	If true, the threshold convergence is enabled. More...
double	min_rate_eps_
	Threshold for the epsilon rate. More...
bool	enabled_min_rate_eps_
	If true, the minimal threshold for epsilon rate is enabled. More...
double	max_time_
	The timeout. More...
bool	enabled_max_time_
	If true, the timeout is enabled. More...
Size	max_iter_
	The maximum iterations. More...
bool	enabled_max_iter_
	If true, the maximum iterations stopping criterion is enabled. More...
Size	burn_in_
	Number of iterations before checking stopping criteria. More...
Size	period_size_
	Checking criteria frequency. More...
bool	verbosity_
	If true, verbosity is enabled. More...

Protected Member Functions
void	findBestContributor_ (NodeId x, NodeId y, const std::vector< NodeId > &ui, const MixedGraph &graph, CorrectedMutualInformation<> &mutualInformation, Heap< CondRanking, GreaterPairOn2nd > &rank)
	finds the best contributor node for a pair given a conditioning set More...
std::vector< Ranking >	unshieldedTriples_ (const MixedGraph &graph, CorrectedMutualInformation<> &mutualInformation, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet)
	gets the list of unshielded triples in the graph in decreasing value of |I'(x, y, z|{ui})| More...
std::vector< ProbabilisticRanking >	unshieldedTriplesMiic_ (const MixedGraph &graph, CorrectedMutualInformation<> &mutualInformation, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet, HashTable< std::pair< NodeId, NodeId >, char > &marks)
	gets the list of unshielded triples in the graph in decreasing value of |I'(x, y, z|{ui})|, prepares the orientation matrix for MIIC More...
std::vector< ProbabilisticRanking >	updateProbaTriples_ (const MixedGraph &graph, std::vector< ProbabilisticRanking > probaTriples)
	Gets the orientation probabilities like MIIC for the orientation phase. More...
bool	propagatesRemainingOrientableEdges_ (MixedGraph &graph, NodeId xj)
	Propagates the orientation from a node to its neighbours. More...
void	propagatesOrientationInChainOfRemainingEdges_ (MixedGraph &graph)
	heuristic for remaining edges when everything else has been tried More...
bool	isForbidenArc_ (NodeId x, NodeId y) const
bool	isOrientable_ (const MixedGraph &graph, NodeId xi, NodeId xj) const
Main phases
void	initiation_ (CorrectedMutualInformation<> &mutualInformation, MixedGraph &graph, HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet, Heap< CondRanking, GreaterPairOn2nd > &rank)
	Initiation phase. More...
void	iteration_ (CorrectedMutualInformation<> &mutualInformation, MixedGraph &graph, HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet, Heap< CondRanking, GreaterPairOn2nd > &rank)
	Iteration phase. More...
void	orientation3off2_ (CorrectedMutualInformation<> &mutualInformation, MixedGraph &graph, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet)
	Orientation phase from the 3off2 algorithm, returns a CPDAG. More...
void	orientationLatents_ (CorrectedMutualInformation<> &mutualInformation, MixedGraph &graph, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet)
	Modified version of the orientation phase that tries to propagate orientations from both orientations in case of a bidirected arc, not used. More...
void	orientationMiic_ (CorrectedMutualInformation<> &mutualInformation, MixedGraph &graph, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &sepSet)
	Orientation phase from the MIIC algorithm, returns a mixed graph that may contain circles. More...

Detailed Description

The miic learning algorithm.

The miic class implements the miic algorithm based on https://doi.org/10.1371/journal.pcbi.1005662. It starts by eliminating edges that correspond to independent variables to build the skeleton of the graph, and then directs the remaining edges to get an essential graph. Latent variables can be detected using bi-directed arcs.

The variant 3off2 is also implemented as proposed by Affeldt and al. in https://doi.org/10.1186/s12859-015-0856-x. Only the orientation phase differs from miic, with a different ranking method and different propagation rules.

Definition at line 98 of file Miic.h.

Member Enumeration Documentation

ApproximationSchemeSTATE

enum gum::IApproximationSchemeConfiguration::ApproximationSchemeSTATE : char

stronginherited

The different state of an approximation scheme.

Enumerator
Undefined
Continue
Epsilon
Rate
Limit
TimeLimit
Stopped

Definition at line 64 of file IApproximationSchemeConfiguration.h.

                                        : char
    {
      Undefined,
      Continue,
      Epsilon,
      Rate,
      Limit,
      TimeLimit,
      Stopped
    };

Constructor & Destructor Documentation

Miic() [1/4]

gum::learning::Miic::Miic

(

)

default constructor

Definition at line 44 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

: _maxLog_(100), _size_(0) { GUM_CONSTRUCTOR(Miic); }

Here is the call graph for this function:

Miic() [2/4]

gum::learning::Miic::Miic ( int  maxLog )

explicit

default constructor with maxLog

Definition at line 47 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

: _maxLog_(maxLog), _size_(0) { GUM_CONSTRUCTOR(Miic); }

Here is the call graph for this function:

Miic() [3/4]

gum::learning::Miic::Miic

(

const Miic &

from

)

copy constructor

Definition at line 50 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                               : ApproximationScheme(from), _size_(from._size_) {
      GUM_CONS_CPY(Miic);
    }

Here is the call graph for this function:

Miic() [4/4]

gum::learning::Miic::Miic

(

Miic &&

from

)

move constructor

Definition at line 55 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                          : ApproximationScheme(std::move(from)), _size_(from._size_) {
      GUM_CONS_MOV(Miic);
    }

Here is the call graph for this function:

~Miic()

gum::learning::Miic::~Miic ( )

override

destructor

Definition at line 60 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

{ GUM_DESTRUCTOR(Miic); }

Here is the call graph for this function:

Member Function Documentation

_existsDirectedPath_()

bool gum::learning::Miic::_existsDirectedPath_ ( const MixedGraph &  graph, NodeId  n1, NodeId  n2  )

staticprivate

checks for directed paths in a graph, consider double arcs like edges

Parameters

graph	MixedGraph in which to search the path
n1	tail of the path
n2	head of the path

Definition at line 974 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                             {
      // not recursive version => use a FIFO for simulating the recursion
      List< NodeId > nodeFIFO;
      // mark[node] = successor if visited, else mark[node] does not exist
      Set< NodeId > mark;

      mark.insert(n2);
      nodeFIFO.pushBack(n2);

      NodeId current;

      while (!nodeFIFO.empty()) {
        current = nodeFIFO.front();
        nodeFIFO.popFront();

        // check the parents
        for (const auto new_one: graph.parents(current)) {
          if (graph.existsArc(current,
                              new_one))   // if there is a double arc, pass
            continue;

          if (new_one == n1) { return true; }

          if (mark.exists(new_one))   // if this node is already marked, do not
            continue;                 // check it again

          mark.insert(new_one);
          nodeFIFO.pushBack(new_one);
        }
      }

      return false;
    }

Here is the call graph for this function:

_existsNonTrivialDirectedPath_()

bool gum::learning::Miic::_existsNonTrivialDirectedPath_ ( const MixedGraph &  graph, NodeId  n1, NodeId  n2  )

staticprivate

checks for directed paths in a graph, considering double arcs like edges, not considering arc as a directed path.

Parameters

graph	MixedGraph in which to search the path
n1	tail of the path
n2	head of the path
countArc	bool to know if we

Definition at line 960 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                    {
      for (const auto parent: graph.parents(n2)) {
        if (graph.existsArc(parent,
                            n2))   // if there is a double arc, pass
          continue;
        if (parent == n1)   // trivial directed path => not recognized
          continue;
        if (_existsDirectedPath_(graph, n1, parent)) return true;
      }
      return false;
    }

Here is the call graph for this function:

_isNotLatentCouple_()

bool gum::learning::Miic::_isNotLatentCouple_ ( NodeId  x, NodeId  y  )

private

Definition at line 1173 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                 {
      const auto& lbeg = _latentCouples_.begin();
      const auto& lend = _latentCouples_.end();

      return (std::find(lbeg, lend, Arc(x, y)) == lend)
          && (std::find(lbeg, lend, Arc(y, x)) == lend);
    }

Here is the call graph for this function:

_orientingVstructureMiic_()

void gum::learning::Miic::_orientingVstructureMiic_ ( MixedGraph &  graph, HashTable< std::pair< NodeId, NodeId >, char > &  marks, NodeId  x, NodeId  y, NodeId  z, double  p1, double  p2  )

private

Definition at line 1008 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                             {
      // v-structure discovery
      if (marks[{x, z}] == 'o' && marks[{y, z}] == 'o') {   // If x-z-y
        if (!_existsNonTrivialDirectedPath_(graph, z, x)) {
          graph.eraseEdge(Edge(x, z));
          graph.addArc(x, z);
          GUM_TRACE("1.a Removing edge (" << x << "," << z << ")")
          GUM_TRACE("1.a Adding arc (" << x << "," << z << ")")
          marks[{x, z}] = '>';
          if (graph.existsArc(z, x) && _isNotLatentCouple_(z, x)) {
            GUM_TRACE("Adding latent couple (" << z << "," << x << ")")
            _latentCouples_.emplace_back(z, x);
          }
          if (!_arcProbas_.exists(Arc(x, z))) _arcProbas_.insert(Arc(x, z), p1);
        } else {
          graph.eraseEdge(Edge(x, z));
          GUM_TRACE("1.b Adding arc (" << x << "," << z << ")")
          if (!_existsNonTrivialDirectedPath_(graph, x, z)) {
            graph.addArc(z, x);
            GUM_TRACE("1.b Removing edge (" << x << "," << z << ")")
            marks[{z, x}] = '>';
          }
        }

        if (!_existsNonTrivialDirectedPath_(graph, z, y)) {
          graph.eraseEdge(Edge(y, z));
          graph.addArc(y, z);
          GUM_TRACE("1.c Removing edge (" << y << "," << z << ")")
          GUM_TRACE("1.c Adding arc (" << y << "," << z << ")")
          marks[{y, z}] = '>';
          if (graph.existsArc(z, y) && _isNotLatentCouple_(z, y)) {
            _latentCouples_.emplace_back(z, y);
          }
          if (!_arcProbas_.exists(Arc(y, z))) _arcProbas_.insert(Arc(y, z), p2);
        } else {
          graph.eraseEdge(Edge(y, z));
          GUM_TRACE("1.d Removing edge (" << y << "," << z << ")")
          if (!_existsNonTrivialDirectedPath_(graph, y, z)) {
            graph.addArc(z, y);
            GUM_TRACE("1.d Adding arc (" << z << "," << y << ")")
            marks[{z, y}] = '>';
          }
        }
      } else if (marks[{x, z}] == '>' && marks[{y, z}] == 'o') {   // If x->z-y
        if (!_existsNonTrivialDirectedPath_(graph, z, y)) {
          graph.eraseEdge(Edge(y, z));
          graph.addArc(y, z);
          GUM_TRACE("2.a Removing edge (" << y << "," << z << ")")
          GUM_TRACE("2.a Adding arc (" << y << "," << z << ")")
          marks[{y, z}] = '>';
          if (graph.existsArc(z, y) && _isNotLatentCouple_(z, y)) {
            _latentCouples_.emplace_back(z, y);
          }
          if (!_arcProbas_.exists(Arc(y, z))) _arcProbas_.insert(Arc(y, z), p2);
        } else {
          graph.eraseEdge(Edge(y, z));
          GUM_TRACE("2.b Removing edge (" << y << "," << z << ")")
          if (!_existsNonTrivialDirectedPath_(graph, y, z)) {
            graph.addArc(z, y);
            GUM_TRACE("2.b Adding arc (" << y << "," << z << ")")
            marks[{z, y}] = '>';
          }
        }
      } else if (marks[{y, z}] == '>' && marks[{x, z}] == 'o') {
        if (!_existsNonTrivialDirectedPath_(graph, z, x)) {
          graph.eraseEdge(Edge(x, z));
          graph.addArc(x, z);
          GUM_TRACE("3.a Removing edge (" << x << "," << z << ")")
          GUM_TRACE("3.a Adding arc (" << x << "," << z << ")")
          marks[{x, z}] = '>';
          if (graph.existsArc(z, x) && _isNotLatentCouple_(z, x)) {
            _latentCouples_.emplace_back(z, x);
          }
          if (!_arcProbas_.exists(Arc(x, z))) _arcProbas_.insert(Arc(x, z), p1);
        } else {
          graph.eraseEdge(Edge(x, z));
          GUM_TRACE("3.b Removing edge (" << x << "," << z << ")")
          if (!_existsNonTrivialDirectedPath_(graph, x, z)) {
            graph.addArc(z, x);
            GUM_TRACE("3.b Adding arc (" << x << "," << z << ")")
            marks[{z, x}] = '>';
          }
        }
      }
    }

Here is the call graph for this function:

_propagatingOrientationMiic_()

void gum::learning::Miic::_propagatingOrientationMiic_ ( MixedGraph &  graph, HashTable< std::pair< NodeId, NodeId >, char > &  marks, NodeId  x, NodeId  y, NodeId  z, double  p1, double  p2  )

private

Definition at line 1101 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                                {
      // orientation propagation
      if (marks[{x, z}] == '>' && marks[{y, z}] == 'o' && marks[{z, y}] != '-') {
        graph.eraseEdge(Edge(z, y));
        // std::cout << "4. Removing edge (" << z << "," << y << ")" <<
        // std::endl;
        if (!_existsDirectedPath_(graph, y, z) && graph.parents(y).empty()) {
          graph.addArc(z, y);
          GUM_TRACE("4.a Adding arc (" << z << "," << y << ")")
          marks[{z, y}] = '>';
          marks[{y, z}] = '-';
          if (!_arcProbas_.exists(Arc(z, y))) _arcProbas_.insert(Arc(z, y), p2);
        } else if (!_existsDirectedPath_(graph, z, y) && graph.parents(z).empty()) {
          graph.addArc(y, z);
          GUM_TRACE("4.b Adding arc (" << y << "," << z << ")")
          marks[{z, y}] = '-';
          marks[{y, z}] = '>';
          _latentCouples_.emplace_back(y, z);
          if (!_arcProbas_.exists(Arc(y, z))) _arcProbas_.insert(Arc(y, z), p2);
        } else if (!_existsDirectedPath_(graph, y, z)) {
          graph.addArc(z, y);
          GUM_TRACE("4.c Adding arc (" << z << "," << y << ")")
          marks[{z, y}] = '>';
          marks[{y, z}] = '-';
          if (!_arcProbas_.exists(Arc(z, y))) _arcProbas_.insert(Arc(z, y), p2);
        } else if (!_existsDirectedPath_(graph, z, y)) {
          graph.addArc(y, z);
          GUM_TRACE("4.d Adding arc (" << y << "," << z << ")")
          _latentCouples_.emplace_back(y, z);
          marks[{z, y}] = '-';
          marks[{y, z}] = '>';
          if (!_arcProbas_.exists(Arc(y, z))) _arcProbas_.insert(Arc(y, z), p2);
        }
      } else if (marks[{y, z}] == '>' && marks[{x, z}] == 'o' && marks[{z, x}] != '-') {
        graph.eraseEdge(Edge(z, x));
        GUM_TRACE("5. Removing edge (" << z << "," << x << ")")
        if (!_existsDirectedPath_(graph, x, z) && graph.parents(x).empty()) {
          graph.addArc(z, x);
          GUM_TRACE("5.a Adding arc (" << z << "," << x << ")")
          marks[{z, x}] = '>';
          marks[{x, z}] = '-';
          if (!_arcProbas_.exists(Arc(z, x))) _arcProbas_.insert(Arc(z, x), p1);
        } else if (!_existsDirectedPath_(graph, z, x) && graph.parents(z).empty()) {
          graph.addArc(x, z);
          GUM_TRACE("5.b Adding arc (" << x << "," << z << ")")
          marks[{z, x}] = '-';
          marks[{x, z}] = '>';
          _latentCouples_.emplace_back(x, z);
          if (!_arcProbas_.exists(Arc(x, z))) _arcProbas_.insert(Arc(x, z), p1);
        } else if (!_existsDirectedPath_(graph, x, z)) {
          graph.addArc(z, x);
          GUM_TRACE("5.c Adding arc (" << z << "," << x << ")")
          marks[{z, x}] = '>';
          marks[{x, z}] = '-';
          if (!_arcProbas_.exists(Arc(z, x))) _arcProbas_.insert(Arc(z, x), p1);
        } else if (!_existsDirectedPath_(graph, z, x)) {
          graph.addArc(x, z);
          GUM_TRACE("5.d Adding arc (" << x << "," << z << ")")
          marks[{z, x}] = '-';
          marks[{x, z}] = '>';
          _latentCouples_.emplace_back(x, z);
          if (!_arcProbas_.exists(Arc(x, z))) _arcProbas_.insert(Arc(x, z), p1);
        }
      }
    }

Here is the call graph for this function:

addConstraints()

void gum::learning::Miic::addConstraints

(

HashTable< std::pair< NodeId, NodeId >, char >

constraints

)

Set a ensemble of constraints for the orientation phase.

Definition at line 956 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                      {
      this->_initialMarks_ = constraints;
    }

Here is the call graph for this function:

continueApproximationScheme()

INLINE bool gum::ApproximationScheme::continueApproximationScheme ( double  error )

inherited

Update the scheme w.r.t the new error.

Test the stopping criterion that are enabled.

Parameters

error

The new error value.

Returns

false if state become != ApproximationSchemeSTATE::Continue

Exceptions

OperationNotAllowed

Raised if state != ApproximationSchemeSTATE::Continue.

Definition at line 208 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                                           {
    // For coherence, we fix the time used in the method

    double timer_step = timer_.step();

    if (enabled_max_time_) {
      if (timer_step > max_time_) {
        stopScheme_(ApproximationSchemeSTATE::TimeLimit);
        return false;
      }
    }

    if (!startOfPeriod()) { return true; }

    if (current_state_ != ApproximationSchemeSTATE::Continue) {
      GUM_ERROR(OperationNotAllowed,
                "state of the approximation scheme is not correct : "
                   + messageApproximationScheme());
    }

    if (verbosity()) { history_.push_back(error); }

    if (enabled_max_iter_) {
      if (current_step_ > max_iter_) {
        stopScheme_(ApproximationSchemeSTATE::Limit);
        return false;
      }
    }

    last_epsilon_    = current_epsilon_;
    current_epsilon_ = error;   // eps rate isEnabled needs it so affectation was
    // moved from eps isEnabled below

    if (enabled_eps_) {
      if (current_epsilon_ <= eps_) {
        stopScheme_(ApproximationSchemeSTATE::Epsilon);
        return false;
      }
    }

    if (last_epsilon_ >= 0.) {
      if (current_epsilon_ > .0) {
        // ! current_epsilon_ can be 0. AND epsilon
        // isEnabled can be disabled !
        current_rate_ = std::fabs((current_epsilon_ - last_epsilon_) / current_epsilon_);
      }
      // limit with current eps ---> 0 is | 1 - ( last_eps / 0 ) | --->
      // infinity the else means a return false if we isEnabled the rate below,
      // as we would have returned false if epsilon isEnabled was enabled
      else {
        current_rate_ = min_rate_eps_;
      }

      if (enabled_min_rate_eps_) {
        if (current_rate_ <= min_rate_eps_) {
          stopScheme_(ApproximationSchemeSTATE::Rate);
          return false;
        }
      }
    }

    if (stateApproximationScheme() == ApproximationSchemeSTATE::Continue) {
      if (onProgress.hasListener()) {
        GUM_EMIT3(onProgress, current_step_, current_epsilon_, timer_step);
      }

      return true;
    } else {
      return false;
    }
  }

Here is the call graph for this function:

currentTime()

INLINE double gum::ApproximationScheme::currentTime ( ) const

virtualinherited

Returns the current running time in second.

Returns

Returns the current running time in second.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 115 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return timer_.step(); }

Here is the call graph for this function:

disableEpsilon()

INLINE void gum::ApproximationScheme::disableEpsilon ( )

virtualinherited

Disable stopping criterion on epsilon.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 53 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_eps_ = false; }

Here is the call graph for this function:

disableMaxIter()

INLINE void gum::ApproximationScheme::disableMaxIter ( )

virtualinherited

Disable stopping criterion on max iterations.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 94 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_max_iter_ = false; }

Here is the call graph for this function:

disableMaxTime()

INLINE void gum::ApproximationScheme::disableMaxTime ( )

virtualinherited

Disable stopping criterion on timeout.

Returns

Disable stopping criterion on timeout.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 118 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_max_time_ = false; }

Here is the call graph for this function:

disableMinEpsilonRate()

INLINE void gum::ApproximationScheme::disableMinEpsilonRate ( )

virtualinherited

Disable stopping criterion on epsilon rate.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 74 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_min_rate_eps_ = false; }

Here is the call graph for this function:

enableEpsilon()

INLINE void gum::ApproximationScheme::enableEpsilon ( )

virtualinherited

Enable stopping criterion on epsilon.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 56 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_eps_ = true; }

Here is the call graph for this function:

enableMaxIter()

INLINE void gum::ApproximationScheme::enableMaxIter ( )

virtualinherited

Enable stopping criterion on max iterations.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 97 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_max_iter_ = true; }

Here is the call graph for this function:

enableMaxTime()

INLINE void gum::ApproximationScheme::enableMaxTime ( )

virtualinherited

Enable stopping criterion on timeout.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 121 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_max_time_ = true; }

Here is the call graph for this function:

enableMinEpsilonRate()

INLINE void gum::ApproximationScheme::enableMinEpsilonRate ( )

virtualinherited

Enable stopping criterion on epsilon rate.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 77 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ enabled_min_rate_eps_ = true; }

Here is the call graph for this function:

epsilon()

INLINE double gum::ApproximationScheme::epsilon ( ) const

virtualinherited

Returns the value of epsilon.

Returns

Returns the value of epsilon.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 50 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return eps_; }

Here is the call graph for this function:

findBestContributor_()

void gum::learning::Miic::findBestContributor_ ( NodeId  x, NodeId  y, const std::vector< NodeId > &  ui, const MixedGraph &  graph, CorrectedMutualInformation<> &  mutualInformation, Heap< CondRanking, GreaterPairOn2nd > &  rank  )

protected

finds the best contributor node for a pair given a conditioning set

Parameters

x	first node
y	second node
ui	conditioning set
mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	containing the assessed nodes
rank	the heap of ranks of the algorithm

Definition at line 574 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                 {
      double maxP = -1.0;
      NodeId maxZ = 0;

      // compute N
      // __N = I.N();
      const double Ixy_ui = mutualInformation.score(x, y, ui);

      for (const NodeId z: graph) {
        // if z!=x and z!=y and z not in ui
        if (z != x && z != y && std::find(ui.begin(), ui.end(), z) == ui.end()) {
          double Pnv;
          double Pb;

          // Computing Pnv
          const double Ixyz_ui    = mutualInformation.score(x, y, z, ui);
          double       calc_expo1 = -Ixyz_ui * M_LN2;
          // if exponentials are too high or to low, crop them at | __maxLog|
          if (calc_expo1 > _maxLog_) {
            Pnv = 0.0;
          } else if (calc_expo1 < -_maxLog_) {
            Pnv = 1.0;
          } else {
            Pnv = 1 / (1 + std::exp(calc_expo1));
          }

          // Computing Pb
          const double Ixz_ui = mutualInformation.score(x, z, ui);
          const double Iyz_ui = mutualInformation.score(y, z, ui);

          calc_expo1        = -(Ixz_ui - Ixy_ui) * M_LN2;
          double calc_expo2 = -(Iyz_ui - Ixy_ui) * M_LN2;

          // if exponentials are too high or to low, crop them at  _maxLog_
          if (calc_expo1 > _maxLog_ || calc_expo2 > _maxLog_) {
            Pb = 0.0;
          } else if (calc_expo1 < -_maxLog_ && calc_expo2 < -_maxLog_) {
            Pb = 1.0;
          } else {
            double expo1, expo2;
            if (calc_expo1 < -_maxLog_) {
              expo1 = 0.0;
            } else {
              expo1 = std::exp(calc_expo1);
            }
            if (calc_expo2 < -_maxLog_) {
              expo2 = 0.0;
            } else {
              expo2 = std::exp(calc_expo2);
            }
            Pb = 1 / (1 + expo1 + expo2);
          }

          // Getting max(min(Pnv, pb))
          const double min_pnv_pb = std::min(Pnv, Pb);
          if (min_pnv_pb > maxP) {
            maxP = min_pnv_pb;
            maxZ = z;
          }
        }   // if z not in (x, y)
      }     // for z in graph.nodes
      // storing best z in rank_
      CondRanking final;
      auto        tup = new CondThreePoints{x, y, maxZ, ui};
      final.first     = tup;
      final.second    = maxP;
      rank.insert(final);
    }

Here is the call graph for this function:

history()

INLINE const std::vector< double > & gum::ApproximationScheme::history ( ) const

virtualinherited

Returns the scheme history.

Returns

Returns the scheme history.

Exceptions

OperationNotAllowed

Raised if the scheme did not performed or if verbosity is set to false.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 157 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                                       {
    if (stateApproximationScheme() == ApproximationSchemeSTATE::Undefined) {
      GUM_ERROR(OperationNotAllowed, "state of the approximation scheme is udefined")
    }

    if (verbosity() == false) { GUM_ERROR(OperationNotAllowed, "No history when verbosity=false") }

    return history_;
  }

Here is the call graph for this function:

initApproximationScheme()

INLINE void gum::ApproximationScheme::initApproximationScheme ( )

inherited

Initialise the scheme.

Definition at line 168 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                           {
    current_state_   = ApproximationSchemeSTATE::Continue;
    current_step_    = 0;
    current_epsilon_ = current_rate_ = -1.0;
    history_.clear();
    timer_.reset();
  }

Here is the call graph for this function:

initiation_()

void gum::learning::Miic::initiation_ ( CorrectedMutualInformation<> &  mutualInformation, MixedGraph &  graph, HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet, Heap< CondRanking, GreaterPairOn2nd > &  rank  )

protected

Initiation phase.

We go over all edges and test if the variables are marginally independent. If they are, the edge is deleted. If not, the best contributor is found.

Parameters

mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	the MixedGraph we start from for the learning
sepSet	the separation set for independent couples, here set to {}
rank	the heap of ranks of the algorithm

Definition at line 144 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                                  {
      NodeId  x, y;
      EdgeSet edges      = graph.edges();
      Size    steps_init = edges.size();

      for (const Edge& edge: edges) {
        x          = edge.first();
        y          = edge.second();
        double Ixy = mutualInformation.score(x, y);

        if (Ixy <= 0) {   //< K
          graph.eraseEdge(edge);
          sepSet.insert(std::make_pair(x, y), _emptySet_);
        } else {
          findBestContributor_(x, y, _emptySet_, graph, mutualInformation, rank);
        }

        ++current_step_;
        if (onProgress.hasListener()) {
          GUM_EMIT3(onProgress, (current_step_ * 33) / steps_init, 0., timer_.step());
        }
      }
    }

Here is the call graph for this function:

isEnabledEpsilon()

INLINE bool gum::ApproximationScheme::isEnabledEpsilon ( ) const

virtualinherited

Returns true if stopping criterion on epsilon is enabled, false otherwise.

Returns

Returns true if stopping criterion on epsilon is enabled, false otherwise.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 60 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return enabled_eps_; }

Here is the call graph for this function:

isEnabledMaxIter()

INLINE bool gum::ApproximationScheme::isEnabledMaxIter ( ) const

virtualinherited

Returns true if stopping criterion on max iterations is enabled, false otherwise.

Returns

Returns true if stopping criterion on max iterations is enabled, false otherwise.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 101 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return enabled_max_iter_; }

Here is the call graph for this function:

isEnabledMaxTime()

INLINE bool gum::ApproximationScheme::isEnabledMaxTime ( ) const

virtualinherited

Returns true if stopping criterion on timeout is enabled, false otherwise.

Returns

Returns true if stopping criterion on timeout is enabled, false otherwise.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 125 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return enabled_max_time_; }

Here is the call graph for this function:

isEnabledMinEpsilonRate()

INLINE bool gum::ApproximationScheme::isEnabledMinEpsilonRate ( ) const

virtualinherited

Returns true if stopping criterion on epsilon rate is enabled, false otherwise.

Returns

Returns true if stopping criterion on epsilon rate is enabled, false otherwise.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 81 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return enabled_min_rate_eps_; }

Here is the call graph for this function:

isForbidenArc_()

bool gum::learning::Miic::isForbidenArc_ ( NodeId  x, NodeId  y  ) const

protected

Definition at line 1181 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                      {
      return (_initialMarks_.exists({x, y}) && _initialMarks_[{x, y}] == '-');
    }

Here is the call graph for this function:

isOrientable_()

bool gum::learning::Miic::isOrientable_ ( const MixedGraph &  graph, NodeId  xi, NodeId  xj  ) const

protected

Definition at line 835 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                {
      // no cycle
      if (_existsDirectedPath_(graph, xj, xi)) {
        GUM_TRACE("cycle(" << xi << "-" << xj << ")")
        return false;
      }

      // R1
      if (!(graph.parents(xi) - graph.adjacents(xj)).empty()) {
        GUM_TRACE("R1(" << xi << "-" << xj << ")")
        return true;
      }

      // R2
      if (_existsDirectedPath_(graph, xi, xj)) {
        GUM_TRACE("R2(" << xi << "-" << xj << ")")
        return true;
      }

      // R3
      int nbr = 0;
      for (const auto p: graph.parents(xj)) {
        if (!graph.mixedOrientedPath(xi, p).empty()) {
          nbr += 1;
          if (nbr == 2) {
            GUM_TRACE("R3(" << xi << "-" << xj << ")")
            return true;
          }
        }
      }
      return false;
    }

Here is the call graph for this function:

iteration_()

void gum::learning::Miic::iteration_ ( CorrectedMutualInformation<> &  mutualInformation, MixedGraph &  graph, HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet, Heap< CondRanking, GreaterPairOn2nd > &  rank  )

protected

Iteration phase.

As long as we find important nodes for edges, we go over them to see if we can assess the conditional independence of the variables.

Parameters

mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	the MixedGraph returned from the previous phase
sepSet	the separation set for independent couples, built during the iterations of the phase
rank	the heap of ranks of the algorithm

Definition at line 177 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                                 {
      // if no triples to further examine pass
      CondRanking best;

      Size steps_init = current_step_;
      Size steps_iter = rank.size();

      try {
        while (rank.top().second > 0.5) {
          best = rank.pop();

          const NodeId          x  = std::get< 0 >(*(best.first));
          const NodeId          y  = std::get< 1 >(*(best.first));
          const NodeId          z  = std::get< 2 >(*(best.first));
          std::vector< NodeId > ui = std::move(std::get< 3 >(*(best.first)));

          ui.push_back(z);
          const double i_xy_ui = mutualInformation.score(x, y, ui);
          if (i_xy_ui < 0) {
            graph.eraseEdge(Edge(x, y));
            sepSet.insert(std::make_pair(x, y), std::move(ui));
          } else {
            findBestContributor_(x, y, ui, graph, mutualInformation, rank);
          }

          delete best.first;

          ++current_step_;
          if (onProgress.hasListener()) {
            GUM_EMIT3(onProgress,
                      (current_step_ * 66) / (steps_init + steps_iter),
                      0.,
                      timer_.step());
          }
        }
      } catch (...) {}   // here, rank is empty
      current_step_ = steps_init + steps_iter;
      if (onProgress.hasListener()) { GUM_EMIT3(onProgress, 66, 0., timer_.step()); }
      current_step_ = steps_init + steps_iter;
    }

Here is the call graph for this function:

latentVariables()

const std::vector< Arc > gum::learning::Miic::latentVariables

(

)

const

get the list of arcs hiding latent variables

Definition at line 941 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

{ return _latentCouples_; }

Here is the call graph for this function:

learnBN()

template<typename GUM_SCALAR , typename GRAPH_CHANGES_SELECTOR , typename PARAM_ESTIMATOR >

BayesNet< GUM_SCALAR > gum::learning::Miic::learnBN	(	GRAPH_CHANGES_SELECTOR &	selector,
		PARAM_ESTIMATOR &	estimator,
		DAG	initial_dag = DAG()
	)

learns the structure and the parameters of a BN

Parameters

selector	A selector class that computes the best changes that can be applied and that enables the user to get them very easily. Typically, the selector is a GraphChangesSelector4DiGraph<SCORE, STRUCT_CONSTRAINT, GRAPH_CHANGES_GENERATOR>.
estimator	A estimator.
names	The variables names.
modal	the domain sizes of the random variables observed in the database
translator	The cell translator to use.
initial_dag	the DAG we start from for our learning

Definition at line 945 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                              {
      return DAG2BNLearner<>::createBN< GUM_SCALAR >(estimator,
                                                     learnStructure(selector, initial_dag));
    }

Here is the call graph for this function:

learnMixedStructure()

MixedGraph gum::learning::Miic::learnMixedStructure	(	CorrectedMutualInformation<> &	mutualInformation,
		MixedGraph	graph
	)

learns the structure of an Essential Graph

learns the structure of a MixedGraph

Parameters

mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	the MixedGraph we start from for the learning

Definition at line 106 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                              {
      timer_.reset();
      current_step_ = 0;

      // clear the vector of latent arcs to be sure
      _latentCouples_.clear();

      Heap< CondRanking, GreaterPairOn2nd > rank;

      HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > sep_set;

      initiation_(mutualInformation, graph, sep_set, rank);

      iteration_(mutualInformation, graph, sep_set, rank);

      // std::cout << "Le graphe contient: " << graph.sizeEdges() << " edges." <<
      // std::endl; std::cout << "En voici la liste: " << graph.edges() <<
      // std::endl;

      if (_useMiic_) {
        orientationMiic_(mutualInformation, graph, sep_set);
      } else {
        orientation3off2_(mutualInformation, graph, sep_set);
      }

      return graph;
    }

Here is the call graph for this function:

learnStructure()

DAG gum::learning::Miic::learnStructure	(	CorrectedMutualInformation<> &	I,
		MixedGraph	graph
	)

learns the structure of a Bayesian network, i.e. a DAG, by first learning an Essential graph and then directing the remaining edges.

learns the structure of an Bayesian network, ie a DAG, from an Essential graph.

Parameters

I	A mutual information instance that will do the computations and has loaded the database
graph	the MixedGraph we start from for the learning

Definition at line 768 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                     {
      MixedGraph essentialGraph = learnMixedStructure(I, initialGraph);
      // orientate remaining edges

      const Sequence< NodeId > order = essentialGraph.topologicalOrder();

      // first, forbidden arcs force arc in the other direction
      for (NodeId x: order) {
        const auto nei_x = essentialGraph.neighbours(x);
        for (NodeId y: nei_x)
          if (isForbidenArc_(x, y)) {
            essentialGraph.eraseEdge(Edge(x, y));
            if (isForbidenArc_(y, x)) {
              GUM_TRACE("Neither arc allowed for edge (" << x << "," << y << ")")
            } else {
              GUM_TRACE("Forced orientation : " << y << "->" << x)
              essentialGraph.addArc(y, x);
            }
          } else if (isForbidenArc_(y, x)) {
            essentialGraph.eraseEdge(Edge(x, y));
            GUM_TRACE("Forced orientation : " << x << "->" << y)
            essentialGraph.addArc(x, y);
          }
      }
      GUM_TRACE(essentialGraph.toDot());

      // first, propagate existing orientations
      bool newOrientation = true;
      while (newOrientation) {
        newOrientation = false;
        for (NodeId x: order) {
          if (!essentialGraph.parents(x).empty()) {
            newOrientation |= propagatesRemainingOrientableEdges_(essentialGraph, x);
          }
        }
      }
      GUM_TRACE(essentialGraph.toDot());
      propagatesOrientationInChainOfRemainingEdges_(essentialGraph);
      GUM_TRACE(essentialGraph.toDot());

      // then decide the orientation for double arcs
      for (NodeId x: order)
        for (NodeId y: essentialGraph.parents(x))
          if (essentialGraph.parents(y).contains(x)) {
            GUM_TRACE(" + Resolving double arcs (poorly)")
            essentialGraph.eraseArc(Arc(y, x));
          }

      // std::cout << "Le mixed graph après une deuxième propagation mesdames et
      // messieurs: "
      //<< essentialGraph.toDot() << std::endl;
      // std::cout << "Le graphe contient maintenant : " <<
      // essentialGraph.sizeArcs() << " arcs."
      //<< std::endl;
      // std::cout << "Que voici: " << essentialGraph.arcs() << std::endl;
      // turn the mixed graph into a dag
      DAG dag;
      for (auto node: essentialGraph) {
        dag.addNodeWithId(node);
      }
      for (const Arc& arc: essentialGraph.arcs()) {
        dag.addArc(arc.tail(), arc.head());
      }

      return dag;
    }

Here is the call graph for this function:

maxIter()

INLINE Size gum::ApproximationScheme::maxIter ( ) const

virtualinherited

Returns the criterion on number of iterations.

Returns

Returns the criterion on number of iterations.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 91 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return max_iter_; }

Here is the call graph for this function:

maxTime()

INLINE double gum::ApproximationScheme::maxTime ( ) const

virtualinherited

Returns the timeout (in seconds).

Returns

Returns the timeout (in seconds).

Implements gum::IApproximationSchemeConfiguration.

Definition at line 112 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return max_time_; }

Here is the call graph for this function:

messageApproximationScheme()

INLINE std::string gum::IApproximationSchemeConfiguration::messageApproximationScheme ( ) const

inherited

Returns the approximation scheme message.

Returns

Returns the approximation scheme message.

Definition at line 38 of file IApproximationSchemeConfiguration_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                                                {
    std::stringstream s;

    switch (stateApproximationScheme()) {
      case ApproximationSchemeSTATE::Continue:
        s << "in progress";
        break;

      case ApproximationSchemeSTATE::Epsilon:
        s << "stopped with epsilon=" << epsilon();
        break;

      case ApproximationSchemeSTATE::Rate:
        s << "stopped with rate=" << minEpsilonRate();
        break;

      case ApproximationSchemeSTATE::Limit:
        s << "stopped with max iteration=" << maxIter();
        break;

      case ApproximationSchemeSTATE::TimeLimit:
        s << "stopped with timeout=" << maxTime();
        break;

      case ApproximationSchemeSTATE::Stopped:
        s << "stopped on request";
        break;

      case ApproximationSchemeSTATE::Undefined:
        s << "undefined state";
        break;
    };

    return s.str();
  }

Here is the call graph for this function:

minEpsilonRate()

INLINE double gum::ApproximationScheme::minEpsilonRate ( ) const

virtualinherited

Returns the value of the minimal epsilon rate.

Returns

Returns the value of the minimal epsilon rate.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 71 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return min_rate_eps_; }

Here is the call graph for this function:

nbrIterations()

INLINE Size gum::ApproximationScheme::nbrIterations ( ) const

virtualinherited

Returns the number of iterations.

Returns

Returns the number of iterations.

Exceptions

OperationNotAllowed

Raised if the scheme did not perform.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 148 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                       {
    if (stateApproximationScheme() == ApproximationSchemeSTATE::Undefined) {
      GUM_ERROR(OperationNotAllowed, "state of the approximation scheme is undefined")
    }

    return current_step_;
  }

Here is the call graph for this function:

operator=() [1/2]

Miic & gum::learning::Miic::operator=

(

const Miic &

from

)

copy operator

Definition at line 63 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                          {
      ApproximationScheme::operator=(from);
      return *this;
    }

Here is the call graph for this function:

operator=() [2/2]

Miic & gum::learning::Miic::operator=

(

Miic &&

from

)

move operator

Definition at line 69 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                     {
      ApproximationScheme::operator=(std::move(from));
      return *this;
    }

Here is the call graph for this function:

orientation3off2_()

void gum::learning::Miic::orientation3off2_ ( CorrectedMutualInformation<> &  mutualInformation, MixedGraph &  graph, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet  )

protected

Orientation phase from the 3off2 algorithm, returns a CPDAG.

Parameters

mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	the MixedGraph returned from the previous phase
sepSet	the separation set for independent couples, built during the previous phase

Definition at line 226 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                  {
      std::vector< Ranking > triples      = unshieldedTriples_(graph, mutualInformation, sepSet);
      Size                   steps_orient = triples.size();
      Size                   past_steps   = current_step_;

      // marks always correspond to the head of the arc/edge. - is for a forbidden
      // arc, > for a mandatory arc
      // we start by adding the mandatory arcs
      for (auto iter = _initialMarks_.begin(); iter != _initialMarks_.end(); ++iter) {
        if (graph.existsEdge(iter.key().first, iter.key().second) && iter.val() == '>') {
          graph.eraseEdge(Edge(iter.key().first, iter.key().second));
          graph.addArc(iter.key().first, iter.key().second);
        }
      }

      NodeId i = 0;
      // list of elements that we shouldnt read again, ie elements that are
      // eligible to
      // rule 0 after the first time they are tested, and elements on which rule 1
      // has been applied
      while (i < triples.size()) {
        // if i not in do_not_reread
        Ranking triple = triples[i];
        NodeId  x, y, z;
        x = std::get< 0 >(*triple.first);
        y = std::get< 1 >(*triple.first);
        z = std::get< 2 >(*triple.first);

        std::vector< NodeId >       ui;
        std::pair< NodeId, NodeId > key     = {x, y};
        std::pair< NodeId, NodeId > rev_key = {y, x};
        if (sepSet.exists(key)) {
          ui = sepSet[key];
        } else if (sepSet.exists(rev_key)) {
          ui = sepSet[rev_key];
        }
        double Ixyz_ui = triple.second;
        bool   reset{false};
        // try Rule 0
        if (Ixyz_ui < 0) {
          // if ( z not in Sep[x,y])
          if (std::find(ui.begin(), ui.end(), z) == ui.end()) {
            if (!graph.existsArc(x, z) && !graph.existsArc(z, x)) {
              // when we try to add an arc to the graph, we always verify if
              // we are allowed to do so, ie it is not a forbidden arc an it
              // does not create a cycle
              if (!_existsDirectedPath_(graph, z, x) && !isForbidenArc_(x, z)) {
                reset = true;
                graph.eraseEdge(Edge(x, z));
                graph.addArc(x, z);
              } else if (_existsDirectedPath_(graph, z, x) && !isForbidenArc_(z, x)) {
                reset = true;
                graph.eraseEdge(Edge(x, z));
                // if we find a cycle, we force the competing edge
                graph.addArc(z, x);
                if (std::find(_latentCouples_.begin(), _latentCouples_.end(), Arc(z, x))
                    == _latentCouples_.end()) {
                  _latentCouples_.emplace_back(z, x);
                }
              }
            } else if (!graph.existsArc(y, z) && !graph.existsArc(z, y)) {
              if (!_existsDirectedPath_(graph, z, y) && !isForbidenArc_(x, z)) {
                reset = true;
                graph.eraseEdge(Edge(y, z));
                graph.addArc(y, z);
              } else if (_existsDirectedPath_(graph, z, y) && !isForbidenArc_(z, y)) {
                reset = true;
                graph.eraseEdge(Edge(y, z));
                // if we find a cycle, we force the competing edge
                graph.addArc(z, y);
                if (std::find(_latentCouples_.begin(), _latentCouples_.end(), Arc(z, y))
                    == _latentCouples_.end()) {
                  _latentCouples_.emplace_back(z, y);
                }
              }
            } else {
              // checking if the anti-directed arc already exists, to register a
              // latent variable
              if (graph.existsArc(z, x) && _isNotLatentCouple_(z, x)) {
                _latentCouples_.emplace_back(z, x);
              }
              if (graph.existsArc(z, y) && _isNotLatentCouple_(z, y)) {
                _latentCouples_.emplace_back(z, y);
              }
            }
          }
        } else {   // try Rule 1
          if (graph.existsArc(x, z) && !graph.existsArc(z, y) && !graph.existsArc(y, z)) {
            if (!_existsDirectedPath_(graph, y, z) && !isForbidenArc_(z, y)) {
              reset = true;
              graph.eraseEdge(Edge(z, y));
              graph.addArc(z, y);
            } else if (_existsDirectedPath_(graph, y, z) && !isForbidenArc_(y, z)) {
              reset = true;
              graph.eraseEdge(Edge(z, y));
              // if we find a cycle, we force the competing edge
              graph.addArc(y, z);
              if (std::find(_latentCouples_.begin(), _latentCouples_.end(), Arc(y, z))
                  == _latentCouples_.end()) {
                _latentCouples_.emplace_back(y, z);
              }
            }
          }
          if (graph.existsArc(y, z) && !graph.existsArc(z, x) && !graph.existsArc(x, z)) {
            if (!_existsDirectedPath_(graph, x, z) && !isForbidenArc_(z, x)) {
              reset = true;
              graph.eraseEdge(Edge(z, x));
              graph.addArc(z, x);
            } else if (_existsDirectedPath_(graph, x, z) && !isForbidenArc_(x, z)) {
              reset = true;
              graph.eraseEdge(Edge(z, x));
              // if we find a cycle, we force the competing edge
              graph.addArc(x, z);
              if (std::find(_latentCouples_.begin(), _latentCouples_.end(), Arc(x, z))
                  == _latentCouples_.end()) {
                _latentCouples_.emplace_back(x, z);
              }
            }
          }
        }   // if rule 0 or rule 1

        // if what we want to add already exists : pass to the next triplet
        if (reset) {
          i = 0;
        } else {
          ++i;
        }
        if (onProgress.hasListener()) {
          GUM_EMIT3(onProgress,
                    ((current_step_ + i) * 100) / (past_steps + steps_orient),
                    0.,
                    timer_.step());
        }
      }   // while

      // erasing the the double headed arcs
      for (const Arc& arc: _latentCouples_) {
        graph.eraseArc(Arc(arc.head(), arc.tail()));
      }
    }

Here is the call graph for this function:

orientationLatents_()

void gum::learning::Miic::orientationLatents_ ( CorrectedMutualInformation<> &  mutualInformation, MixedGraph &  graph, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet  )

protected

Modified version of the orientation phase that tries to propagate orientations from both orientations in case of a bidirected arc, not used.

varient trying to propagate both orientations in a bidirected arc

Parameters

mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	the MixedGraph returned from the previous phase
sepSet	the separation set for independent couples, built during the previous phase

Definition at line 371 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                  {
      std::vector< Ranking > triples      = unshieldedTriples_(graph, mutualInformation, sepSet);
      Size                   steps_orient = triples.size();
      Size                   past_steps   = current_step_;

      NodeId i = 0;
      // list of elements that we shouldnt read again, ie elements that are
      // eligible to
      // rule 0 after the first time they are tested, and elements on which rule 1
      // has been applied
      while (i < triples.size()) {
        // if i not in do_not_reread
        Ranking triple = triples[i];
        NodeId  x, y, z;
        x = std::get< 0 >(*triple.first);
        y = std::get< 1 >(*triple.first);
        z = std::get< 2 >(*triple.first);

        std::vector< NodeId >       ui;
        std::pair< NodeId, NodeId > key     = {x, y};
        std::pair< NodeId, NodeId > rev_key = {y, x};
        if (sepSet.exists(key)) {
          ui = sepSet[key];
        } else if (sepSet.exists(rev_key)) {
          ui = sepSet[rev_key];
        }
        double Ixyz_ui = triple.second;
        // try Rule 0
        if (Ixyz_ui < 0) {
          // if ( z not in Sep[x,y])
          if (std::find(ui.begin(), ui.end(), z) == ui.end()) {
            // if what we want to add already exists : pass
            if ((graph.existsArc(x, z) || graph.existsArc(z, x))
                && (graph.existsArc(y, z) || graph.existsArc(z, y))) {
              ++i;
            } else {
              i = 0;
              graph.eraseEdge(Edge(x, z));
              graph.eraseEdge(Edge(y, z));
              // checking for cycles
              if (graph.existsArc(z, x)) {
                graph.eraseArc(Arc(z, x));
                try {
                  std::vector< NodeId > path = graph.directedPath(z, x);
                  // if we find a cycle, we force the competing edge
                  _latentCouples_.emplace_back(z, x);
                } catch (gum::NotFound) { graph.addArc(x, z); }
                graph.addArc(z, x);
              } else {
                try {
                  std::vector< NodeId > path = graph.directedPath(z, x);
                  // if we find a cycle, we force the competing edge
                  graph.addArc(z, x);
                  _latentCouples_.emplace_back(z, x);
                } catch (gum::NotFound) { graph.addArc(x, z); }
              }
              if (graph.existsArc(z, y)) {
                graph.eraseArc(Arc(z, y));
                try {
                  std::vector< NodeId > path = graph.directedPath(z, y);
                  // if we find a cycle, we force the competing edge
                  _latentCouples_.emplace_back(z, y);
                } catch (gum::NotFound) { graph.addArc(y, z); }
                graph.addArc(z, y);
              } else {
                try {
                  std::vector< NodeId > path = graph.directedPath(z, y);
                  // if we find a cycle, we force the competing edge
                  graph.addArc(z, y);
                  _latentCouples_.emplace_back(z, y);

                } catch (gum::NotFound) { graph.addArc(y, z); }
              }
              if (graph.existsArc(z, x) && _isNotLatentCouple_(z, x)) {
                _latentCouples_.emplace_back(z, x);
              }
              if (graph.existsArc(z, y) && _isNotLatentCouple_(z, y)) {
                _latentCouples_.emplace_back(z, y);
              }
            }
          } else {
            ++i;
          }
        } else {   // try Rule 1
          bool reset{false};
          if (graph.existsArc(x, z) && !graph.existsArc(z, y) && !graph.existsArc(y, z)) {
            reset = true;
            graph.eraseEdge(Edge(z, y));
            try {
              std::vector< NodeId > path = graph.directedPath(y, z);
              // if we find a cycle, we force the competing edge
              graph.addArc(y, z);
              _latentCouples_.emplace_back(y, z);
            } catch (gum::NotFound) { graph.addArc(z, y); }
          }
          if (graph.existsArc(y, z) && !graph.existsArc(z, x) && !graph.existsArc(x, z)) {
            reset = true;
            graph.eraseEdge(Edge(z, x));
            try {
              std::vector< NodeId > path = graph.directedPath(x, z);
              // if we find a cycle, we force the competing edge
              graph.addArc(x, z);
              _latentCouples_.emplace_back(x, z);
            } catch (gum::NotFound) { graph.addArc(z, x); }
          }

          if (reset) {
            i = 0;
          } else {
            ++i;
          }
        }   // if rule 0 or rule 1
        if (onProgress.hasListener()) {
          GUM_EMIT3(onProgress,
                    ((current_step_ + i) * 100) / (past_steps + steps_orient),
                    0.,
                    timer_.step());
        }
      }   // while

      // erasing the the double headed arcs
      for (const Arc& arc: _latentCouples_) {
        graph.eraseArc(Arc(arc.head(), arc.tail()));
      }
    }

Here is the call graph for this function:

orientationMiic_()

void gum::learning::Miic::orientationMiic_ ( CorrectedMutualInformation<> &  mutualInformation, MixedGraph &  graph, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet  )

protected

Orientation phase from the MIIC algorithm, returns a mixed graph that may contain circles.

varient using the orientation protocol of MIIC

Parameters

mutualInformation	A mutual information instance that will do the computations and has loaded the database.
graph	the MixedGraph returned from the previous phase
sepSet	the separation set for independent couples, built during the previous phase

Definition at line 501 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                  {
      // structure to store the orientations marks -, o, or >,
      // Considers the head of the arc/edge first node -* second node
      HashTable< std::pair< NodeId, NodeId >, char > marks = _initialMarks_;

      // marks always correspond to the head of the arc/edge. - is for a forbidden
      // arc, > for a mandatory arc
      // we start by adding the mandatory arcs
      for (auto iter = marks.begin(); iter != marks.end(); ++iter) {
        if (graph.existsEdge(iter.key().first, iter.key().second) && iter.val() == '>') {
          graph.eraseEdge(Edge(iter.key().first, iter.key().second));
          graph.addArc(iter.key().first, iter.key().second);
        }
      }

      std::vector< ProbabilisticRanking > proba_triples
         = unshieldedTriplesMiic_(graph, mutualInformation, sepSet, marks);

      const Size steps_orient = proba_triples.size();
      Size       past_steps   = current_step_;

      ProbabilisticRanking best;
      if (steps_orient > 0) { best = proba_triples[0]; }

      while (!proba_triples.empty() && std::max(std::get< 2 >(best), std::get< 3 >(best)) > 0.5) {
        const NodeId x = std::get< 0 >(*std::get< 0 >(best));
        const NodeId y = std::get< 1 >(*std::get< 0 >(best));
        const NodeId z = std::get< 2 >(*std::get< 0 >(best));

        const double i3 = std::get< 1 >(best);

        const double p1 = std::get< 2 >(best);
        const double p2 = std::get< 3 >(best);
        if (i3 <= 0) {
          _orientingVstructureMiic_(graph, marks, x, y, z, p1, p2);
        } else {
          _propagatingOrientationMiic_(graph, marks, x, y, z, p1, p2);
        }

        delete std::get< 0 >(best);
        proba_triples.erase(proba_triples.begin());
        // actualisation of the list of triples
        proba_triples = updateProbaTriples_(graph, proba_triples);

        if (!proba_triples.empty()) best = proba_triples[0];

        ++current_step_;
        if (onProgress.hasListener()) {
          GUM_EMIT3(onProgress,
                    (current_step_ * 100) / (steps_orient + past_steps),
                    0.,
                    timer_.step());
        }
      }   // while

      // erasing the double headed arcs
      for (auto iter = _latentCouples_.rbegin(); iter != _latentCouples_.rend(); ++iter) {
        graph.eraseArc(Arc(iter->head(), iter->tail()));
        if (_existsDirectedPath_(graph, iter->head(), iter->tail())) {
          // if we find a cycle, we force the competing edge
          graph.addArc(iter->head(), iter->tail());
          graph.eraseArc(Arc(iter->tail(), iter->head()));
          *iter = Arc(iter->head(), iter->tail());
        }
      }

      if (onProgress.hasListener()) { GUM_EMIT3(onProgress, 100, 0., timer_.step()); }
    }

Here is the call graph for this function:

periodSize()

INLINE Size gum::ApproximationScheme::periodSize ( ) const

virtualinherited

Returns the period size.

Returns

Returns the period size.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 134 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return period_size_; }

Here is the call graph for this function:

propagatesOrientationInChainOfRemainingEdges_()

void gum::learning::Miic::propagatesOrientationInChainOfRemainingEdges_ ( MixedGraph &  graph )

protected

heuristic for remaining edges when everything else has been tried

Definition at line 868 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                       {
      // then decide the orientation for remaining edges
      while (!essentialGraph.edges().empty()) {
        const auto& edge               = *(essentialGraph.edges().begin());
        NodeId      root               = edge.first();
        Size        size_children_root = essentialGraph.children(root).size();
        NodeSet     visited;
        NodeSet     stack{root};
        // check the best root for the set of neighbours
        while (!stack.empty()) {
          NodeId next = *(stack.begin());
          stack.erase(next);
          if (visited.contains(next)) continue;
          if (essentialGraph.children(next).size() > size_children_root) {
            size_children_root = essentialGraph.children(next).size();
            root               = next;
          }
          for (const auto n: essentialGraph.neighbours(next))
            if (!stack.contains(n) && !visited.contains(n)) stack.insert(n);
          visited.insert(next);
        }
        // orientation now
        visited.clear();
        stack.clear();
        stack.insert(root);
        while (!stack.empty()) {
          NodeId next = *(stack.begin());
          stack.erase(next);
          if (visited.contains(next)) continue;
          const auto nei = essentialGraph.neighbours(next);
          for (const auto n: nei) {
            if (!stack.contains(n) && !visited.contains(n)) stack.insert(n);
            GUM_TRACE(" + amap reasonably orientation for " << n << "->" << next);
            essentialGraph.eraseEdge(Edge(n, next));
            essentialGraph.addArc(n, next);
          }
          visited.insert(next);
        }
      }
    }

Here is the call graph for this function:

propagatesRemainingOrientableEdges_()

bool gum::learning::Miic::propagatesRemainingOrientableEdges_ ( MixedGraph &  graph, NodeId  xj  )

protected

Propagates the orientation from a node to its neighbours.

Definition at line 910 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                               {
      bool       res        = false;
      const auto neighbours = graph.neighbours(xj);
      for (auto& xi: neighbours) {
        bool i_j = isOrientable_(graph, xi, xj);
        bool j_i = isOrientable_(graph, xj, xi);
        if (i_j || j_i) {
          GUM_TRACE(" + Removing edge (" << xi << "," << xj << ")")
          graph.eraseEdge(Edge(xi, xj));
          res = true;
        }
        if (i_j) {
          GUM_TRACE(" + add arc (" << xi << "," << xj << ")")
          graph.addArc(xi, xj);
          propagatesRemainingOrientableEdges_(graph, xj);
        }
        if (j_i) {
          GUM_TRACE(" + add arc (" << xi << "," << xj << ")")
          graph.addArc(xj, xi);
          propagatesRemainingOrientableEdges_(graph, xi);
        }
        if (i_j && j_i) {
          GUM_TRACE(" + add arc (" << xi << "," << xj << ")")
          _latentCouples_.emplace_back(xi, xj);
        }
      }

      return res;
    }

Here is the call graph for this function:

remainingBurnIn()

INLINE Size gum::ApproximationScheme::remainingBurnIn ( )

inherited

Returns the remaining burn in.

Returns

Returns the remaining burn in.

Definition at line 191 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                   {
    if (burn_in_ > current_step_) {
      return burn_in_ - current_step_;
    } else {
      return 0;
    }
  }

Here is the call graph for this function:

set3of2Behaviour()

void gum::learning::Miic::set3of2Behaviour

(

)

Sets the orientation phase to follow the one of the 3off2 algorithm.

Definition at line 954 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

{ this->_useMiic_ = false; }

Here is the call graph for this function:

setEpsilon()

INLINE void gum::ApproximationScheme::setEpsilon ( double  eps )

virtualinherited

Given that we approximate f(t), stopping criterion on |f(t+1)-f(t)|.

If the criterion was disabled it will be enabled.

Parameters

eps	The new epsilon value.

Exceptions

OutOfLowerBound

Raised if eps < 0.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 42 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                        {
    if (eps < 0.) { GUM_ERROR(OutOfLowerBound, "eps should be >=0") }

    eps_         = eps;
    enabled_eps_ = true;
  }

Here is the call graph for this function:

setMaxIter()

INLINE void gum::ApproximationScheme::setMaxIter ( Size  max )

virtualinherited

Stopping criterion on number of iterations.

If the criterion was disabled it will be enabled.

Parameters

max	The maximum number of iterations.

Exceptions

OutOfLowerBound

Raised if max <= 1.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 84 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                      {
    if (max < 1) { GUM_ERROR(OutOfLowerBound, "max should be >=1") }
    max_iter_         = max;
    enabled_max_iter_ = true;
  }

Here is the call graph for this function:

setMaxTime()

INLINE void gum::ApproximationScheme::setMaxTime ( double  timeout )

virtualinherited

Stopping criterion on timeout.

If the criterion was disabled it will be enabled.

Parameters

timeout

The timeout value in seconds.

Exceptions

OutOfLowerBound

Raised if timeout <= 0.0.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 105 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                            {
    if (timeout <= 0.) { GUM_ERROR(OutOfLowerBound, "timeout should be >0.") }
    max_time_         = timeout;
    enabled_max_time_ = true;
  }

Here is the call graph for this function:

setMiicBehaviour()

void gum::learning::Miic::setMiicBehaviour

(

)

Sets the orientation phase to follow the one of the MIIC algorithm.

Definition at line 952 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

{ this->_useMiic_ = true; }

Here is the call graph for this function:

setMinEpsilonRate()

INLINE void gum::ApproximationScheme::setMinEpsilonRate ( double  rate )

virtualinherited

Given that we approximate f(t), stopping criterion on d/dt(|f(t+1)-f(t)|).

If the criterion was disabled it will be enabled

Parameters

rate	The minimal epsilon rate.

Exceptions

OutOfLowerBound

if rate<0

Implements gum::IApproximationSchemeConfiguration.

Definition at line 63 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                                {
    if (rate < 0) { GUM_ERROR(OutOfLowerBound, "rate should be >=0") }

    min_rate_eps_         = rate;
    enabled_min_rate_eps_ = true;
  }

Here is the call graph for this function:

setPeriodSize()

INLINE void gum::ApproximationScheme::setPeriodSize ( Size  p )

virtualinherited

How many samples between two stopping is enable.

Parameters

p	The new period value.

Exceptions

OutOfLowerBound

Raised if p < 1.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 128 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                       {
    if (p < 1) { GUM_ERROR(OutOfLowerBound, "p should be >=1") }

    period_size_ = p;
  }

Here is the call graph for this function:

setVerbosity()

INLINE void gum::ApproximationScheme::setVerbosity ( bool  v )

virtualinherited

Set the verbosity on (true) or off (false).

Parameters

v	If true, then verbosity is turned on.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 137 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ verbosity_ = v; }

Here is the call graph for this function:

startOfPeriod()

INLINE bool gum::ApproximationScheme::startOfPeriod ( )

inherited

Returns true if we are at the beginning of a period (compute error is mandatory).

Returns

Returns true if we are at the beginning of a period (compute error is mandatory).

Definition at line 178 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                 {
    if (current_step_ < burn_in_) { return false; }

    if (period_size_ == 1) { return true; }

    return ((current_step_ - burn_in_) % period_size_ == 0);
  }

Here is the call graph for this function:

stateApproximationScheme()

INLINE IApproximationSchemeConfiguration::ApproximationSchemeSTATE gum::ApproximationScheme::stateApproximationScheme ( ) const

virtualinherited

Returns the approximation scheme state.

Returns

Returns the approximation scheme state.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 143 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                             {
    return current_state_;
  }

Here is the call graph for this function:

stopApproximationScheme()

INLINE void gum::ApproximationScheme::stopApproximationScheme ( )

inherited

Stop the approximation scheme.

Definition at line 200 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                           {
    if (current_state_ == ApproximationSchemeSTATE::Continue) {
      stopScheme_(ApproximationSchemeSTATE::Stopped);
    }
  }

Here is the call graph for this function:

unshieldedTriples_()

std::vector< Ranking > gum::learning::Miic::unshieldedTriples_ ( const MixedGraph &  graph, CorrectedMutualInformation<> &  mutualInformation, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet  )

protected

gets the list of unshielded triples in the graph in decreasing value of |I'(x, y, z|{ui})|

Definition at line 650 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                  {
      std::vector< Ranking > triples;
      for (NodeId z: graph) {
        for (NodeId x: graph.neighbours(z)) {
          for (NodeId y: graph.neighbours(z)) {
            if (y < x && !graph.existsEdge(x, y)) {
              std::vector< NodeId >       ui;
              std::pair< NodeId, NodeId > key     = {x, y};
              std::pair< NodeId, NodeId > rev_key = {y, x};
              if (sepSet.exists(key)) {
                ui = sepSet[key];
              } else if (sepSet.exists(rev_key)) {
                ui = sepSet[rev_key];
              }
              // remove z from ui if it's present
              const auto iter_z_place = std::find(ui.begin(), ui.end(), z);
              if (iter_z_place != ui.end()) { ui.erase(iter_z_place); }

              double  Ixyz_ui = mutualInformation.score(x, y, z, ui);
              Ranking triple;
              auto    tup   = new ThreePoints{x, y, z};
              triple.first  = tup;
              triple.second = Ixyz_ui;
              triples.push_back(triple);
            }
          }
        }
      }
      std::sort(triples.begin(), triples.end(), GreaterAbsPairOn2nd());
      return triples;
    }

Here is the call graph for this function:

unshieldedTriplesMiic_()

std::vector< ProbabilisticRanking > gum::learning::Miic::unshieldedTriplesMiic_ ( const MixedGraph &  graph, CorrectedMutualInformation<> &  mutualInformation, const HashTable< std::pair< NodeId, NodeId >, std::vector< NodeId > > &  sepSet, HashTable< std::pair< NodeId, NodeId >, char > &  marks  )

protected

gets the list of unshielded triples in the graph in decreasing value of |I'(x, y, z|{ui})|, prepares the orientation matrix for MIIC

Definition at line 687 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                   {
      std::vector< ProbabilisticRanking > triples;
      for (NodeId z: graph) {
        for (NodeId x: graph.neighbours(z)) {
          for (NodeId y: graph.neighbours(z)) {
            if (y < x && !graph.existsEdge(x, y)) {
              std::vector< NodeId >       ui;
              std::pair< NodeId, NodeId > key     = {x, y};
              std::pair< NodeId, NodeId > rev_key = {y, x};
              if (sepSet.exists(key)) {
                ui = sepSet[key];
              } else if (sepSet.exists(rev_key)) {
                ui = sepSet[rev_key];
              }
              // remove z from ui if it's present
              const auto iter_z_place = std::find(ui.begin(), ui.end(), z);
              if (iter_z_place != ui.end()) { ui.erase(iter_z_place); }

              const double         Ixyz_ui = mutualInformation.score(x, y, z, ui);
              auto                 tup     = new ThreePoints{x, y, z};
              ProbabilisticRanking triple{tup, Ixyz_ui, 0.5, 0.5};
              triples.push_back(triple);
              if (!marks.exists({x, z})) { marks.insert({x, z}, 'o'); }
              if (!marks.exists({z, x})) { marks.insert({z, x}, 'o'); }
              if (!marks.exists({y, z})) { marks.insert({y, z}, 'o'); }
              if (!marks.exists({z, y})) { marks.insert({z, y}, 'o'); }
            }
          }
        }
      }
      triples = updateProbaTriples_(graph, triples);
      std::sort(triples.begin(), triples.end(), GreaterTupleOnLast());
      return triples;
    }

Here is the call graph for this function:

updateApproximationScheme()

INLINE void gum::ApproximationScheme::updateApproximationScheme ( unsigned int  incr = 1 )

inherited

Update the scheme w.r.t the new error and increment steps.

Parameters

incr	The new increment steps.

Definition at line 187 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

                                                                              {
    current_step_ += incr;
  }

Here is the call graph for this function:

updateProbaTriples_()

std::vector< ProbabilisticRanking > gum::learning::Miic::updateProbaTriples_ ( const MixedGraph &  graph, std::vector< ProbabilisticRanking >  probaTriples  )

protected

Gets the orientation probabilities like MIIC for the orientation phase.

Definition at line 728 of file Miic.cpp.

References gum::learning::genericBNLearner::Database::Database().

                                                                                 {
      for (auto& triple: probaTriples) {
        NodeId x, y, z;
        x                 = std::get< 0 >(*std::get< 0 >(triple));
        y                 = std::get< 1 >(*std::get< 0 >(triple));
        z                 = std::get< 2 >(*std::get< 0 >(triple));
        const double Ixyz = std::get< 1 >(triple);
        double       Pxz  = std::get< 2 >(triple);
        double       Pyz  = std::get< 3 >(triple);

        if (Ixyz <= 0) {
          const double expo = std::exp(Ixyz);
          const double P0   = (1 + expo) / (1 + 3 * expo);
          // distinguish betweeen the initialization and the update process
          if (Pxz == Pyz && Pyz == 0.5) {
            std::get< 2 >(triple) = P0;
            std::get< 3 >(triple) = P0;
          } else {
            if (graph.existsArc(x, z) && Pxz >= P0) {
              std::get< 3 >(triple) = Pxz * (1 / (1 + expo) - 0.5) + 0.5;
            } else if (graph.existsArc(y, z) && Pyz >= P0) {
              std::get< 2 >(triple) = Pyz * (1 / (1 + expo) - 0.5) + 0.5;
            }
          }
        } else {
          const double expo = std::exp(-Ixyz);
          if (graph.existsArc(x, z) && Pxz >= 0.5) {
            std::get< 3 >(triple) = Pxz * (1 / (1 + expo) - 0.5) + 0.5;
          } else if (graph.existsArc(y, z) && Pyz >= 0.5) {
            std::get< 2 >(triple) = Pyz * (1 / (1 + expo) - 0.5) + 0.5;
          }
        }
      }
      std::sort(probaTriples.begin(), probaTriples.end(), GreaterTupleOnLast());
      return probaTriples;
    }

Here is the call graph for this function:

verbosity()

INLINE bool gum::ApproximationScheme::verbosity ( ) const

virtualinherited

Returns true if verbosity is enabled.

Returns

Returns true if verbosity is enabled.

Implements gum::IApproximationSchemeConfiguration.

Definition at line 139 of file approximationScheme_inl.h.

References gum::Set< Key, Alloc >::emplace().

{ return verbosity_; }

Here is the call graph for this function:

Member Data Documentation

_arcProbas_

ArcProperty< double > gum::learning::Miic::_arcProbas_

private

Storing the propabilities for each arc set in the graph.

Definition at line 335 of file Miic.h.

_emptySet_

const std::vector< NodeId > gum::learning::Miic::_emptySet_

private

an empty conditioning set

Definition at line 325 of file Miic.h.

_initialMarks_

HashTable< std::pair< NodeId, NodeId >, char > gum::learning::Miic::_initialMarks_

private

Initial marks for the orientation phase, used to convey constraints.

Definition at line 338 of file Miic.h.

_latentCouples_

std::vector< Arc > gum::learning::Miic::_latentCouples_

private

an empty vector of arcs

Definition at line 327 of file Miic.h.

_maxLog_

int gum::learning::Miic::_maxLog_ = 100

private

Fixes the maximum log that we accept in exponential computations.

Definition at line 323 of file Miic.h.

_size_

Size gum::learning::Miic::_size_

private

size of the database

Definition at line 330 of file Miic.h.

_useMiic_

bool gum::learning::Miic::_useMiic_ {false}

private

wether to use the miic algorithm or not

Definition at line 332 of file Miic.h.

burn_in_

Size gum::ApproximationScheme::burn_in_

protectedinherited

Number of iterations before checking stopping criteria.

Definition at line 413 of file approximationScheme.h.

current_epsilon_

double gum::ApproximationScheme::current_epsilon_

protectedinherited

Current epsilon.

Definition at line 368 of file approximationScheme.h.

current_rate_

double gum::ApproximationScheme::current_rate_

protectedinherited

Current rate.

Definition at line 374 of file approximationScheme.h.

current_state_

ApproximationSchemeSTATE gum::ApproximationScheme::current_state_

protectedinherited

The current state.

Definition at line 383 of file approximationScheme.h.

current_step_

Size gum::ApproximationScheme::current_step_

protectedinherited

The current step.

Definition at line 377 of file approximationScheme.h.

enabled_eps_

bool gum::ApproximationScheme::enabled_eps_

protectedinherited

If true, the threshold convergence is enabled.

Definition at line 392 of file approximationScheme.h.

enabled_max_iter_

bool gum::ApproximationScheme::enabled_max_iter_

protectedinherited

If true, the maximum iterations stopping criterion is enabled.

Definition at line 410 of file approximationScheme.h.

enabled_max_time_

bool gum::ApproximationScheme::enabled_max_time_

protectedinherited

If true, the timeout is enabled.

Definition at line 404 of file approximationScheme.h.

enabled_min_rate_eps_

bool gum::ApproximationScheme::enabled_min_rate_eps_

protectedinherited

If true, the minimal threshold for epsilon rate is enabled.

Definition at line 398 of file approximationScheme.h.

eps_

double gum::ApproximationScheme::eps_

protectedinherited

Threshold for convergence.

Definition at line 389 of file approximationScheme.h.

history_

std::vector< double > gum::ApproximationScheme::history_

protectedinherited

The scheme history, used only if verbosity == true.

Definition at line 386 of file approximationScheme.h.

last_epsilon_

double gum::ApproximationScheme::last_epsilon_

protectedinherited

Last epsilon value.

Definition at line 371 of file approximationScheme.h.

max_iter_

Size gum::ApproximationScheme::max_iter_

protectedinherited

The maximum iterations.

Definition at line 407 of file approximationScheme.h.

max_time_

double gum::ApproximationScheme::max_time_

protectedinherited

The timeout.

Definition at line 401 of file approximationScheme.h.

min_rate_eps_

double gum::ApproximationScheme::min_rate_eps_

protectedinherited

Threshold for the epsilon rate.

Definition at line 395 of file approximationScheme.h.

onProgress

Signaler3< Size, double, double > gum::IApproximationSchemeConfiguration::onProgress

inherited

Progression, error and time.

Definition at line 58 of file IApproximationSchemeConfiguration.h.

onStop

Signaler1< std::string > gum::IApproximationSchemeConfiguration::onStop

inherited

Criteria messageApproximationScheme.

Definition at line 61 of file IApproximationSchemeConfiguration.h.

period_size_

Size gum::ApproximationScheme::period_size_

protectedinherited

Checking criteria frequency.

Definition at line 416 of file approximationScheme.h.

timer_

Timer gum::ApproximationScheme::timer_

protectedinherited

The timer.

Definition at line 380 of file approximationScheme.h.

verbosity_

bool gum::ApproximationScheme::verbosity_

protectedinherited

If true, verbosity is enabled.

Definition at line 419 of file approximationScheme.h.

---

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

• agrum/BN/learning/Miic.h
• agrum/BN/learning/Miic.cpp