aGrUM 2.3.2
a C++ library for (probabilistic) graphical models
ShaferShenoyInference.h
Go to the documentation of this file.
1/****************************************************************************
2 * This file is part of the aGrUM/pyAgrum library. *
3 * *
4 * Copyright (c) 2005-2025 by *
5 * - Pierre-Henri WUILLEMIN(_at_LIP6) *
6 * - Christophe GONZALES(_at_AMU) *
7 * *
8 * The aGrUM/pyAgrum library is free software; you can redistribute it *
9 * and/or modify it under the terms of either : *
10 * *
11 * - the GNU Lesser General Public License as published by *
12 * the Free Software Foundation, either version 3 of the License, *
13 * or (at your option) any later version, *
14 * - the MIT license (MIT), *
15 * - or both in dual license, as here. *
16 * *
17 * (see https://agrum.gitlab.io/articles/dual-licenses-lgplv3mit.html) *
18 * *
19 * This aGrUM/pyAgrum library is distributed in the hope that it will be *
20 * useful, but WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, *
21 * INCLUDING BUT NOT LIMITED TO THE WARRANTIES MERCHANTABILITY or FITNESS *
22 * FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE *
23 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER *
24 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, *
25 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR *
26 * OTHER DEALINGS IN THE SOFTWARE. *
27 * *
28 * See LICENCES for more details. *
29 * *
30 * SPDX-FileCopyrightText: Copyright 2005-2025 *
31 * - Pierre-Henri WUILLEMIN(_at_LIP6) *
32 * - Christophe GONZALES(_at_AMU) *
33 * SPDX-License-Identifier: LGPL-3.0-or-later OR MIT *
34 * *
35 * Contact : info_at_agrum_dot_org *
36 * homepage : http://agrum.gitlab.io *
37 * gitlab : https://gitlab.com/agrumery/agrum *
38 * *
39 ****************************************************************************/
40
41
49#ifndef GUM_SHAFER_SHENOY_INFERENCE_H
50#define GUM_SHAFER_SHENOY_INFERENCE_H
51
52#include <utility>
53
54#include <agrum/agrum.h>
55
62
63namespace gum {
64
65
66 // the function used to combine two tables
67 template < typename GUM_SCALAR >
68 INLINE static Tensor< GUM_SCALAR > SSNewmultiTensor(const Tensor< GUM_SCALAR >& t1,
69 const Tensor< GUM_SCALAR >& t2) {
70 return t1 * t2;
71 }
72
73 // the function used to combine two tables
74 template < typename GUM_SCALAR >
75 INLINE static Tensor< GUM_SCALAR > SSNewprojTensor(const Tensor< GUM_SCALAR >& t1,
76 const gum::VariableSet& del_vars) {
77 return t1.sumOut(del_vars);
78 }
79
87 template < typename GUM_SCALAR >
89 public JointTargetedInference< GUM_SCALAR >,
90 public EvidenceInference< GUM_SCALAR >,
91 public ScheduledInference {
92 public:
93 // ############################################################################
95 // ############################################################################
97
102 FindBarrenNodesType barren_type
104 bool use_binary_join_tree = true);
105
108
111 = delete;
112
115
117
118
119 // ############################################################################
121 // ############################################################################
123
125 void setTriangulation(const Triangulation& new_triangulation);
126
128
137
139
145
147
151
153
158
160 GUM_SCALAR evidenceProbability() final;
161
163
164
165 protected:
167 void onEvidenceAdded_(const NodeId id, bool isHardEvidence) final;
168
170 void onEvidenceErased_(const NodeId id, bool isHardEvidence) final;
171
173 void onAllEvidenceErased_(bool has_hard_evidence) final;
174
182 void onEvidenceChanged_(const NodeId id, bool hasChangedSoftHard) final;
183
185
186 void onMarginalTargetAdded_(const NodeId id) final;
187
189
190 void onMarginalTargetErased_(const NodeId id) final;
191
193 void onModelChanged_(const GraphicalModel* bn) final;
194
196
197 void onJointTargetAdded_(const NodeSet& set) final;
198
200
201 void onJointTargetErased_(const NodeSet& set) final;
202
205
208
211
214
216 void onStateChanged_() final {};
217
219
223
225
229
231
232 void makeInference_() final;
233
234
236
237 const Tensor< GUM_SCALAR >& posterior_(NodeId id) final;
238
240
242 const Tensor< GUM_SCALAR >& jointPosterior_(const NodeSet& set) final;
243
251 const Tensor< GUM_SCALAR >& jointPosterior_(const NodeSet& wanted_target,
252 const NodeSet& declared_target) final;
253
255 Tensor< GUM_SCALAR >* unnormalizedJointPosterior_(NodeId id) final;
256
258 Tensor< GUM_SCALAR >* unnormalizedJointPosterior_(const NodeSet& set) final;
259
260
261 private:
262 using _TensorSet_ = Set< const Tensor< GUM_SCALAR >* >;
264
265 using _TensorSetIterator_ = SetIteratorSafe< const Tensor< GUM_SCALAR >* >;
266
267
271
276 gum::VariableSet& kept_vars);
277
280
282 Tensor< GUM_SCALAR > (*_projection_op_)(const Tensor< GUM_SCALAR >&,
284
286 Tensor< GUM_SCALAR > (*_combination_op_)(const Tensor< GUM_SCALAR >&,
287 const Tensor< GUM_SCALAR >&){SSNewmultiTensor};
288
291
295
297
303
305 JoinTree* _JT_{nullptr};
306
309
311
315
317
324
327
330
332
340
342
347
349
353
355
362
364
366
368
370
376
378
382
384
391
393
398
401
404
408
410 bool _use_schedules_{false};
411
413 static constexpr double _schedule_threshold_{1000000.0};
414
416 static constexpr GUM_SCALAR _one_minus_epsilon_{GUM_SCALAR(1.0 - 1e-6)};
417
418
420 bool _isNewJTNeeded_() const;
421
424
427
430
432 void _setProjectionFunction_(Tensor< GUM_SCALAR > (*proj)(const Tensor< GUM_SCALAR >&,
433 const gum::VariableSet&));
434
436 void _setCombinationFunction_(Tensor< GUM_SCALAR > (*comb)(const Tensor< GUM_SCALAR >&,
437 const Tensor< GUM_SCALAR >&));
438
440 void _diffuseMessageInvalidations_(NodeId from_id, NodeId to_id, NodeSet& invalidated_cliques);
441
444
447
451 gum::VariableSet& kept_vars);
452
456 gum::VariableSet& kept_vars);
457
461 gum::VariableSet& kept_vars);
462
467
471
474 _ScheduleMultiDimSet_& pot_list,
475 gum::VariableSet& del_vars);
476
479
483 _ScheduleMultiDimSet_ pot_list,
484 gum::VariableSet& del_vars,
485 gum::VariableSet& kept_vars);
486
490 gum::VariableSet& del_vars,
491 gum::VariableSet& kept_vars);
492
494 void _produceMessage_(Schedule& schedule, NodeId from_id, NodeId to_id);
495
497 void _produceMessage_(NodeId from_id, NodeId to_id);
498
500 void _collectMessage_(Schedule& schedule, NodeId id, NodeId from);
501
504
506 Tensor< GUM_SCALAR >* _unnormalizedJointPosterior_(Schedule& schedule, NodeId id);
507
509 Tensor< GUM_SCALAR >* _unnormalizedJointPosterior_(NodeId id);
510
512 Tensor< GUM_SCALAR >* _unnormalizedJointPosterior_(Schedule& schedule, const NodeSet& set);
513
515 Tensor< GUM_SCALAR >* _unnormalizedJointPosterior_(const NodeSet& set);
516 };
517
518
519#ifndef GUM_NO_EXTERN_TEMPLATE_CLASS
520 extern template class ShaferShenoyInference< double >;
521#endif
522
523
524} /* namespace gum */
525
527
528#endif /* SHAFER_SHENOY_INFERENCE_H */
Implementation of Shafer-Shenoy's propagation for inference in Bayesian networks.
Detect barren nodes for inference in Bayesian networks.
virtual const IBayesNet< GUM_SCALAR > & BN() const final
Returns a constant reference over the IBayesNet referenced by this class.
EvidenceInference(const IBayesNet< GUM_SCALAR > *bn)
default constructor
Virtual base class for probabilistic graphical models.
The class for generic Hash Tables.
Definition hashTable.h:637
Class representing the minimal interface for Bayesian network with no numerical data.
Definition IBayesNet.h:75
The Table-agnostic base class of scheduleMultiDim.
JointTargetedInference(const IBayesNet< GUM_SCALAR > *bn)
default constructor
Class containing a schedule of operations to perform on multidims.
Definition schedule.h:80
ScheduledInference(Size max_nb_threads=0, double max_megabyte_memory=0.0)
default constructor
Safe iterators for the Set class.
Definition set.h:601
Representation of a set.
Definition set.h:131
Implementation of Shafer-Shenoy's propagation algorithm for inference in Bayesian networks.
void _computeJoinTreeRoots_()
compute a root for each connected component of JT
void onMarginalTargetErased_(const NodeId id) final
fired before a single target is removed
Tensor< GUM_SCALAR > * unnormalizedJointPosterior_(NodeId id) final
returns a fresh tensor equal to P(argument,evidence)
HashTable< NodeSet, const Tensor< GUM_SCALAR > * > _joint_target_posteriors_
the set of set target posteriors computed during the last inference
bool _isNewJTNeeded_() const
check whether a new join tree is really needed for the next inference
void _setProjectionFunction_(Tensor< GUM_SCALAR >(*proj)(const Tensor< GUM_SCALAR > &, const gum::VariableSet &))
sets the operator for performing the projections
static constexpr GUM_SCALAR _one_minus_epsilon_
for comparisons with 1 - epsilon
static constexpr double _schedule_threshold_
minimal number of operations to perform in the JT to use schedules
void _collectMessage_(Schedule &schedule, NodeId id, NodeId from)
perform the collect phase using schedules
void onEvidenceChanged_(const NodeId id, bool hasChangedSoftHard) final
fired after an evidence is changed, in particular when its status (soft/hard) changes
void onEvidenceAdded_(const NodeId id, bool isHardEvidence) final
fired after a new evidence is inserted
const JunctionTree * junctionTree()
returns the current junction tree
void _invalidateAllMessages_()
invalidate all messages, posteriors and created tensors
const IScheduleMultiDim * _marginalizeOut_(Schedule &schedule, _ScheduleMultiDimSet_ pot_list, gum::VariableSet &del_vars, gum::VariableSet &kept_vars)
removes variables del_vars from a list of tensors and returns the resulting list using schedules
NodeProperty< _ScheduleMultiDimSet_ > _clique_tensors_
the list of all tensors stored in the cliques
bool _use_schedules_
indicates whether we should use schedules for inference
void setFindBarrenNodesType(FindBarrenNodesType type)
sets how we determine barren nodes
EvidenceChangeType
the possible types of evidence changes
const Tensor< GUM_SCALAR > & jointPosterior_(const NodeSet &set) final
returns the posterior of a declared target set
void _findRelevantTensorsWithdSeparation3_(_ScheduleMultiDimSet_ &pot_list, gum::VariableSet &kept_vars)
update a set of tensors: the remaining are those to be combined to produce a message on a separator
void _findRelevantTensorsXX_(_ScheduleMultiDimSet_ &pot_list, gum::VariableSet &kept_vars)
update a set of tensors: the remaining are those to be combined to produce a message on a separator
SetIteratorSafe< const Tensor< GUM_SCALAR > * > _TensorSetIterator_
FindBarrenNodesType _barren_nodes_type_
the type of barren nodes computation we wish
void onJointTargetAdded_(const NodeSet &set) final
fired after a new joint target is inserted
GUM_SCALAR evidenceProbability() final
returns the probability of evidence
Tensor< GUM_SCALAR > * _unnormalizedJointPosterior_(Schedule &schedule, const NodeSet &set)
returns a fresh tensor equal to P(argument,evidence) using schedules
Tensor< GUM_SCALAR >(* _projection_op_)(const Tensor< GUM_SCALAR > &, const gum::VariableSet &)
the operator for performing the projections
NodeProperty< const IScheduleMultiDim * > _node_to_hard_ev_projected_CPTs_
the CPTs that were projected due to hard evidence nodes
const Tensor< GUM_SCALAR > & posterior_(NodeId id) final
returns the posterior of a given variable
void _findRelevantTensorsGetAll_(_ScheduleMultiDimSet_ &pot_list, gum::VariableSet &kept_vars)
update a set of tensors: the remaining are those to be combined to produce a message on a separator
void onStateChanged_() final
fired when the state of the inference engine is changed
_ScheduleMultiDimSet_ _removeBarrenVariables_(Schedule &schedule, _ScheduleMultiDimSet_ &pot_list, gum::VariableSet &del_vars)
remove barren variables and return the newly created projected tensors
void _findRelevantTensorsWithdSeparation2_(_ScheduleMultiDimSet_ &pot_list, gum::VariableSet &kept_vars)
update a set of tensors: the remaining are those to be combined to produce a message on a separator
void onAllMarginalTargetsErased_() final
fired before all the single targets are removed
void onEvidenceErased_(const NodeId id, bool isHardEvidence) final
fired before an evidence is removed
void _diffuseMessageInvalidations_(NodeId from_id, NodeId to_id, NodeSet &invalidated_cliques)
invalidate all the messages sent from a given clique
Tensor< GUM_SCALAR >(* _combination_op_)(const Tensor< GUM_SCALAR > &, const Tensor< GUM_SCALAR > &)
the operator for performing the combinations
void(ShaferShenoyInference< GUM_SCALAR >::* _findRelevantTensors_)(Set< const IScheduleMultiDim * > &pot_list, gum::VariableSet &kept_vars)
update a set of tensors: the remaining are those to be combined to produce a message on a separator
NodeProperty< const IScheduleMultiDim * > _node_to_soft_evidence_
the soft evidence stored in the cliques per their assigned node in the BN
~ShaferShenoyInference()
destructor
ShaferShenoyInference< GUM_SCALAR > & operator=(const ShaferShenoyInference< GUM_SCALAR > &)=delete
avoid copy operators
Triangulation * _triangulation_
the triangulation class creating the junction tree used for inference
void onAllJointTargetsErased_() final
fired before all the joint targets are removed
UndiGraph _graph_
the undigraph extracted from the BN and used to construct the join tree
HashTable< NodeSet, NodeId > _joint_target_to_clique_
for each set target, assign a clique in the JT that contains it
void onAllEvidenceErased_(bool has_hard_evidence) final
fired before all the evidence are erased
void onMarginalTargetAdded_(const NodeId id) final
fired after a new single target is inserted
ShaferShenoyInference(const IBayesNet< GUM_SCALAR > *BN, RelevantTensorsFinderType=RelevantTensorsFinderType::DSEP_BAYESBALL_TENSORS, FindBarrenNodesType barren_type=FindBarrenNodesType::FIND_BARREN_NODES, bool use_binary_join_tree=true)
default constructor
ShaferShenoyInference(const ShaferShenoyInference< GUM_SCALAR > &)=delete
avoid copy constructors
JoinTree * _JT_
the join (or junction) tree used to answer the last inference query
NodeProperty< EvidenceChangeType > _evidence_changes_
indicates which nodes of the BN have evidence that changed since the last inference
void updateOutdatedStructure_() final
prepares inference when the latter is in OutdatedStructure state
JunctionTree * _junctionTree_
the junction tree to answer the last inference query
void _produceMessage_(Schedule &schedule, NodeId from_id, NodeId to_id)
creates the message sent by clique from_id to clique to_id using schedules
Tensor< GUM_SCALAR > * _unnormalizedJointPosterior_(const NodeSet &set)
returns a fresh tensor equal to P(argument,evidence) without using schedules
void _initializeJTCliques_(Schedule &schedule)
put all the CPTs into the cliques when creating the JT using a schedule
void _produceMessage_(NodeId from_id, NodeId to_id)
creates the message sent by clique from_id to clique to_id without schedules
void onAllMarginalTargetsAdded_() final
fired after all the nodes of the BN are added as single targets
NodeProperty< GUM_SCALAR > _constants_
the constants resulting from the projections of CPTs defined over only hard evidence nodes @TODO remo...
bool _is_new_jt_needed_
indicates whether a new join tree is needed for the next inference
NodeSet _hard_ev_nodes_
the hard evidence nodes which were projected in CPTs
NodeProperty< const IScheduleMultiDim * > _clique_ss_tensor_
the tensors stored into the cliques by Shafer-Shenoy
void onAllTargetsErased_() final
fired before all single and joint targets are removed
void makeInference_() final
called when the inference has to be performed effectively
void onModelChanged_(const GraphicalModel *bn) final
fired after a new Bayes net has been assigned to the inference engine
ArcProperty< const IScheduleMultiDim * > _arc_to_created_tensors_
the set of tensors created for the last inference messages
Tensor< GUM_SCALAR > * _unnormalizedJointPosterior_(NodeId id)
computes the unnormalized posterior of a node without using schedules
void _createNewJT_()
create a new junction tree as well as its related data structures
ArcProperty< const IScheduleMultiDim * > _separator_tensors_
the list of all tensors stored in the separators after inferences
void setTriangulation(const Triangulation &new_triangulation)
use a new triangulation algorithm
void onJointTargetErased_(const NodeSet &set) final
fired before a joint target is removed
void _collectMessage_(NodeId id, NodeId from)
actually perform the collect phase directly without schedules
RelevantTensorsFinderType _find_relevant_tensor_type_
the type of relevant tensor finding algorithm to be used
Tensor< GUM_SCALAR > * _unnormalizedJointPosterior_(Schedule &schedule, NodeId id)
computes the unnormalized posterior of a node using schedules
void setRelevantTensorsFinderType(RelevantTensorsFinderType type)
sets how we determine the relevant tensors to combine
const JoinTree * joinTree()
returns the current join tree used
Set< const Tensor< GUM_SCALAR > * > _TensorSet_
void _initializeJTCliques_()
put all the CPTs into the cliques when creating the JT without using a schedule
Set< const IScheduleMultiDim * > _ScheduleMultiDimSet_
void _findRelevantTensorsWithdSeparation_(_ScheduleMultiDimSet_ &pot_list, gum::VariableSet &kept_vars)
update a set of tensors: the remaining are those to be combined to produce a message on a separator
NodeSet _roots_
a clique node used as a root in each connected component of JT
void _setCombinationFunction_(Tensor< GUM_SCALAR >(*comb)(const Tensor< GUM_SCALAR > &, const Tensor< GUM_SCALAR > &))
sets the operator for performing the combinations
const IScheduleMultiDim * _marginalizeOut_(_ScheduleMultiDimSet_ &pot_list, gum::VariableSet &del_vars, gum::VariableSet &kept_vars)
removes variables del_vars from a list of tensors and returns the resulting list directly without sch...
ArcProperty< bool > _messages_computed_
indicates whether a message (from one clique to another) has been computed
NodeProperty< const Tensor< GUM_SCALAR > * > _target_posteriors_
the set of single posteriors computed during the last inference
void updateOutdatedTensors_() final
prepares inference when the latter is in OutdatedTensors state
NodeProperty< NodeId > _node_to_clique_
for each node of graph (~ in the Bayes net), associate an ID in the JT
bool _use_binary_join_tree_
indicates whether we should transform junction trees into binary join trees
_TensorSet_ _removeBarrenVariables_(_TensorSet_ &pot_list, gum::VariableSet &del_vars)
remove barren variables without schedules and return the newly created projected tensors
aGrUM's Tensor is a multi-dimensional array with tensor operators.
Definition tensor.h:85
Interface for all the triangulation methods.
Base class for undirected graphs.
Definition undiGraph.h:128
Class for computing default triangulations of graphs.
This file contains the abstract class definition for computing the probability of evidence entered in...
Size NodeId
Type for node ids.
HashTable< Arc, VAL > ArcProperty
Property on graph elements.
HashTable< NodeId, VAL > NodeProperty
Property on graph elements.
Set< NodeId > NodeSet
Some typdefs and define for shortcuts ...
This file contains the abstract inference class definition for computing (incrementally) joint poster...
gum is the global namespace for all aGrUM entities
Definition agrum.h:46
FindBarrenNodesType
type of algorithm to determine barren nodes
Set< const DiscreteVariable * > VariableSet
static INLINE Tensor< GUM_SCALAR > SSNewprojTensor(const Tensor< GUM_SCALAR > &t1, const gum::VariableSet &del_vars)
static INLINE Tensor< GUM_SCALAR > SSNewmultiTensor(const Tensor< GUM_SCALAR > &t1, const Tensor< GUM_SCALAR > &t2)
CliqueGraph JoinTree
a join tree is a clique graph satisfying the running intersection property (but some cliques may be i...
CliqueGraph JunctionTree
a junction tree is a clique graph satisfying the running intersection property and such that no cliqu...
RelevantTensorsFinderType
type of algorithm for determining the relevant tensors for combinations using some d-separation analy...
the type of algorithm to use to perform relevant reasoning in Bayes net inference
The class enabling flexible inferences using schedules.