edu.ucla.belief.tree
Class Trees

Object
  extended by Trees

public final class Trees
extends Object

A collection of static routines for creation and conversion of the various structured tree representations.

Author:
unknown

Field Summary
static boolean FLAG_DEBUG
           
static CostFunction SUM_SIZE_COST
          A cost function which is the sum of the sizes(# of instantiations) of the two cliques.
static CostFunction UNION_SIZE_COST
          A cost function which is the size of the union of the variables in the cluster.
 
Method Summary
static EliminationTree eliminationTree(TableIndex[] leaves, List eliminationOrder)
          Generates an elimination tree.
static EliminationTree recursiveEliminationTree(TableIndex[] leaves, List eliminationOrder)
           
static JoinTree traditionalJoinTree(BeliefNetwork bn, int reps, Random seed)
           
static JoinTree traditionalJoinTree(BeliefNetwork bn, List eliminationOrder)
          Generates a traditional JoinTree using the method described in [1].
static JoinTree traditionalJoinTree(Graph moralGraph, List eliminationOrder, CostFunction edgeCost)
          Warning: It is safe to use this method only in the absence of a BeliefNetwork.
 
Methods inherited from class Object
equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

FLAG_DEBUG

public static boolean FLAG_DEBUG
Since:
081502

UNION_SIZE_COST

public static final CostFunction UNION_SIZE_COST
A cost function which is the size of the union of the variables in the cluster. This method is intended for use in binary join tree construction.


SUM_SIZE_COST

public static final CostFunction SUM_SIZE_COST
A cost function which is the sum of the sizes(# of instantiations) of the two cliques. This is the method described in [1].

Method Detail

eliminationTree

public static EliminationTree eliminationTree(TableIndex[] leaves,
                                              List eliminationOrder)
Generates an elimination tree.

Parameters:
potentials - The potentials that form the basis for it.
eliminationOrder - A list of variables in the order that they should be removed.
potentialAssignments - A Reference, which if non null will point to a Map which maps from leaf nodes to the potential assigned to it.

recursiveEliminationTree

public static EliminationTree recursiveEliminationTree(TableIndex[] leaves,
                                                       List eliminationOrder)

traditionalJoinTree

public static JoinTree traditionalJoinTree(BeliefNetwork bn,
                                           int reps,
                                           Random seed)

traditionalJoinTree

public static JoinTree traditionalJoinTree(BeliefNetwork bn,
                                           List eliminationOrder)
Generates a traditional JoinTree using the method described in [1]. This method works only for discrete networks.


traditionalJoinTree

public static JoinTree traditionalJoinTree(Graph moralGraph,
                                           List eliminationOrder,
                                           CostFunction edgeCost)
Warning: It is safe to use this method only in the absence of a BeliefNetwork. Generates a JoinTree by generating the clusters through elimination, then choosing the edges so as to minimize the cost. This is described in [2].

Since:
021004


Copyright 2010 UCLA Automated Reasoning Group