![]() |
aGrUM
0.14.2
|
Base Class for all elimination sequence algorithms that require only the graph to be triangulated and the nodes' domain sizes to produce the node elimination ordering. More...
#include <agrum/graphs/algorithms/triangulations/eliminationStrategies/eliminationSequenceStrategy.h>
Go to the source code of this file.
Classes | |
class | gum::UnconstrainedEliminationSequenceStrategy |
The base class for all elimination sequence algorithms that require only the graph to be triangulated and the nodes' domain sizes to produce the node elimination ordering. More... | |
Namespaces | |
gum | |
gum is the global namespace for all aGrUM entities | |
Base Class for all elimination sequence algorithms that require only the graph to be triangulated and the nodes' domain sizes to produce the node elimination ordering.
In many cases, the graph to be triangulated and the nodes domain sizes are the only information needed to compute an elimination ordering. For instance, this is only what is needed for computing junction trees for Bayesian networks or GAI networks. For such cases, the elimination sequence algorithms derived from UnconstrainedEliminationSequenceStrategy are appropriate. There are however other cases in which additional informations must be passed to the elimination sequence algorithm to produce a correct elimination ordering. For instance, computing a strong junction tree for an influence diagram requires the knowledge of a partial ordering on the nodes to be eliminated. For such cases, classes derived from UnconstrainedEliminationSequenceStrategy are not appropriate and those should be derived from other elimination sequence factories, for from instance PartialOrderedEliminationSequenceStrategy.
Definition in file unconstrainedEliminationSequenceStrategy.h.