Convenience class to store information associated to a specific state. More...

Public Member Functions
	SearchNode (int numberOfValues, boolean maximize, boolean ignore_inf)
	Constructor.
void	initSampling (UtilitySolutionSpace< V, AddableReal > space, AddableReal infeasibleUtility, String[] contextVariables, V[] context, V[] domain)
	Finds the first feasible local solution, given the current context.
void	visited (Bound< V > b)
	Method called whenever the state is sampled.
int	solveLocalProblem (UtilitySolutionSpace< V, AddableReal > space, AddableReal infeasibleUtility, String[] contextVariables, V[] context, V[] domain)
	Finds the next feasible local solution.
AddableReal	storeCost (int valueIndex, AddableReal costSample, AddableReal infeasibleUtility, final boolean maximize)
	Stores the sum of the costs reported by the children for this particular state.
double	convergenceBound (int index, double delta)
	With probablity 1- `delta` the error between the real value and the estimated value lies in.
String	toString (double delta)
int	getRandomUnknowLocal (UtilitySolutionSpace< V, AddableReal > space, AddableReal infeasibleUtility, String[] contextVariables, V[] context, V[] domain)
	Method to randomly pick a value that has not been sampled yet.

Public Attributes
AddableReal[]	localCosts
	For each domain value, the local costs assocated to it.
ArrayList< Integer >	unknowLocalSolutions
	List of local solutions that have not yet been sampled.
int	nbrFeasibleLocalSolutions
	Counts the number of local infeasible solutions.
boolean	feasible
	`true` when the local problem has at least one feasible solution, `false` otherwise
int	numberOfValues
	the size of the domain
double[]	averageUtil
	For each domain value, the average utility received so far.
double[]	bestUtil
	The best util received so far.
double[]	bounds
	For each domain value, the current bound.
int[]	actionFrequencies
	For each posible domain values, the number of times it has been selected.
int	frequencie
	The number of times the corresponding state has been visited.
double	maxValue
	The value woth the currently best utility.
double	maxBound
	The value with the currently highest upper bound.
int	maxValueIndex
	The index of the currently best value.
int	maxBoundIndex
	The index of the currently highest bound.

Protected Attributes
final boolean	IGNORE_INF
	when `true`, infeasible values should always be ignored, when `false` the should be sampled once
boolean	random
	`true` when not all domain values have been sampled, and `false` otherwise
int	lastVisited
	Last value visited.

Package Functions
String	arrayToString (Object[] arr, int highlight)
	Method for printing an array, highlighting the entry at `highlight`.
String	arrayToString (double[] arr, int highlight)
	Method for printing an array, highlighting the entry at `highlight`.

Package Attributes
ArrayList< Integer >	samplingOrder = new ArrayList<Integer>()
	The sampling order.

Detailed Description

Convenience class to store information associated to a specific state.

Author: Brammert Ottens, 29 aug. 2011

Parameters

<V>	type used for domain values

Constructor & Destructor Documentation

◆ SearchNode()

frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.SearchNode	(	int	numberOfValues,
		boolean	maximize,
		boolean	ignore_inf )

Constructor.

Parameters

numberOfValues	the domain size of the variable
maximize	`true` when maximizing, and `false` otherwise
ignore_inf	`true` when infeasible utilities should be totally ignored, `false` otherwise

References actionFrequencies, averageUtil, bestUtil, bounds, feasible, IGNORE_INF, localCosts, numberOfValues, and random.

Member Function Documentation

◆ arrayToString() [1/2]

String frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.arrayToString	(	double[]	arr,
		int	highlight )

package

Method for printing an array, highlighting the entry at highlight.

Author: Brammert Ottens, Oct 31, 2011

Parameters

arr	the array to be printed
highlight	the index to be highlighted

Returns: string representation of the array

◆ arrayToString() [2/2]

String frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.arrayToString	(	Object[]	arr,
		int	highlight )

package

Method for printing an array, highlighting the entry at highlight.

Author: Brammert Ottens, Oct 31, 2011

Parameters

arr	the array to be printed
highlight	the index to be highlighted

Returns: string representation of the array

Referenced by toString().

◆ convergenceBound()

double frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.convergenceBound	(	int	index,
		double	delta )

With probablity 1- delta the error between the real value and the estimated value lies in.

Author: Brammert Ottens, 29 aug. 2011

Parameters

index	the index of the value for which the bound should be calculated
delta	the probability

Returns: the bound

References actionFrequencies.

Referenced by frodo2.algorithms.duct.termination.TerminateBest< V extends Addable< V > >.converged(), frodo2.algorithms.duct.termination.TerminateBest_Child< V extends Addable< V > >.converged(), frodo2.algorithms.duct.termination.TerminateMean< V extends Addable< V > >.converged(), frodo2.algorithms.duct.termination.TerminateRegret< V extends Addable< V > >.converged(), frodo2.algorithms.duct.termination.TerminateRegret_Child< V extends Addable< V > >.converged(), and toString().

◆ getRandomUnknowLocal()

int frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.getRandomUnknowLocal	(	UtilitySolutionSpace< V, AddableReal >	space,
		AddableReal	infeasibleUtility,
		String[]	contextVariables,
		V[]	context,
		V[]	domain )

Method to randomly pick a value that has not been sampled yet.

Parameters

space	the space owned by the variable
infeasibleUtility	the infeasible utility
contextVariables	the variables in the separator
context	the current context (variable assignment)
domain	the domain of the variable

Returns: an index of an unsampled, feasible solution, or -1 if no such value exists

References frodo2.solutionSpaces.AddableReal.add(), and unknowLocalSolutions.

Referenced by initSampling(), and solveLocalProblem().

Here is the call graph for this function:

◆ initSampling()

void frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.initSampling	(	UtilitySolutionSpace< V, AddableReal >	space,
		AddableReal	infeasibleUtility,
		String[]	contextVariables,
		V[]	context,
		V[]	domain )

Finds the first feasible local solution, given the current context.

Author: Brammert Ottens, Oct 24, 2011

Parameters

space	the local space
infeasibleUtility	the infeasible utility
contextVariables	the context variables
context	the context
domain	the variable domain

References getRandomUnknowLocal(), and lastVisited.

Referenced by frodo2.algorithms.duct.Sampling< V extends Addable< V > >.VariableInfo.setContext(), and frodo2.algorithms.duct.SamplingPruning< V extends Addable< V > >.VariableInfo.setContext().

Here is the call graph for this function:

◆ solveLocalProblem()

int frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.solveLocalProblem	(	UtilitySolutionSpace< V, AddableReal >	space,
		AddableReal	infeasibleUtility,
		String[]	contextVariables,
		V[]	context,
		V[]	domain )

Finds the next feasible local solution.

Author: Brammert Ottens, Oct 24, 2011

Parameters

space	the local space
infeasibleUtility	the infeasible utility
contextVariables	the context variables
context	the context
domain	the variable domain

Returns: the last visited feasible local solution

References getRandomUnknowLocal(), lastVisited, and unknowLocalSolutions.

Here is the call graph for this function:

◆ storeCost()

AddableReal frodo2.algorithms.duct.SearchNode< V extends Addable< V > >.storeCost	(	int	valueIndex,
		AddableReal	costSample,
		AddableReal	infeasibleUtility,
		final boolean	maximize )

Stores the sum of the costs reported by the children for this particular state.

Author: Brammert Ottens, 24 aug. 2011

Parameters

valueIndex	the index of the sampled value
costSample	the sum of the costs reported by the children
infeasibleUtility	utility of infeasible assignments
maximize	`true` when maximizing, and `false` otherwise