Class GrowShrinkTree
java.lang.Object
edu.cmu.tetrad.search.utils.GrowShrinkTree
GrowShrinkTree class.
If the score given is a GraphScore (an m-separation oracle, which supports only score differences), the tree
runs in oracle mode: the grow step adds a variable just in case it is m-connected to the target node given the
parents chosen so far, the shrink step removes a parent just in case it is m-separated from the target node given
the remaining parents, and the value returned by trace is minus the number of parents retained. Summed over the
nodes of a permutation this is minus the number of edges in the DAG built from that permutation, so maximizing it
is the sparsest-permutation criterion of Raskutti and Uhler. For any other score the behavior is unchanged.
- Version:
- $Id: $Id
- Author:
- bryanandrews
-
Constructor Summary
Constructors -
Method Summary
Modifier and TypeMethodDescriptiongetFirstLayer.Getter for the fieldforbidden.Getter for the fieldindex.getNode()Getter for the fieldnode.Getter for the fieldrequired.getVariables.booleanisForbidden(Node node) isForbidden.booleanisRequired(Node node) isRequired.localScore.localScore(int[] X) localScore.voidreset()reset.voidsetKnowledge(List<Node> required, List<Node> forbidden) setKnowledge.doubletrace.doubletrace.
-
Constructor Details
-
GrowShrinkTree
-
-
Method Details
-
trace
-
trace
-
getNode
-
getFirstLayer
-
getIndex
-
localScore
-
localScore
-
isRequired
-
isForbidden
-
getVariables
-
getRequired
-
getForbidden
-
setKnowledge
-
reset
public void reset()reset.
-