choco.palm.benders.search
Class SubOptimizer

java.lang.Object
  choco.AbstractEntity
      choco.AbstractSolver
          choco.search.AbstractGlobalSearchSolver
              choco.search.Solve
                  choco.palm.cbj.search.JumpGlobalSearchSolver
                      choco.palm.benders.search.SubSearchSolver
                          choco.palm.benders.search.SubOptimizer

All Implemented Interfaces:

Entity

public class SubOptimizer
extends SubSearchSolver

A searchsolver for optimization

Field Summary

boolean	doMaximize a boolean indicating whether we want to maximize (true) or minize (false) the objective variable
int	lowerBound the lower bound of the objective value.
protected int[]	msol Master solution
ExplainedIntVar	objective the variable modelling the objective value
int	targetLowerBound a tentative lower bound
int	targetUpperBound a tentative upper bound
int	upperBound the upper bound of the objective value This value comes from the problem definition; it is strengthened by the search history (solutions found & no-goods)

Fields inherited from class choco.palm.benders.search.SubSearchSolver

currentCtx, slave

Fields inherited from class choco.palm.cbj.search.JumpGlobalSearchSolver

currentFail

Fields inherited from class choco.search.AbstractGlobalSearchSolver

baseWorld, currentTraceIndex, DOWN_BRANCH, encounteredLimit, INIT_SEARCH, limits, loggingMaxDepth, mainGoal, nbSolutions, nextMove, OPEN_NODE, stopAtFirstSol, traceStack, UP_BRANCH

Fields inherited from class choco.AbstractSolver

logger, maxNbSolutionStored, solutions

Fields inherited from class choco.AbstractEntity

hook, problem

Constructor Summary

protected

SubOptimizer(IntDomainVar obj, boolean maximize, boolean slave) constructor

Method Summary

protected void	changeGoal(AbstractIntBranching branching, IntDomainVar newObjective)
int	getBestObjectiveValue()
int	getObjectiveTarget() the target for the objective function: we are searching for a solution at least as good as this (tentative bound)
int	getObjectiveValue() v1.0 accessing the objective value of an optimization problem (note that the objective value may not be instantiated, while all other variables are)
void	initBounds() initialization of the optimization bound data structure
void	logIntermediateSol()
java.lang.Boolean	nextOptimalSolution(int masterWorld) get the next solution of the master problem
void	postDynamicCut() we use targetBound data structures for the optimization cuts
void	postKnownBound()
void	postLowerBound(int lb)
void	postTargetBound() propagating the optimization cuts from the new target bounds
void	postUpperBound(int ub)
void	recordSolution() called when a solution is encountered: printing and, if needed, storing the solution
void	restoreBestSolutionBySearch()
void	setBound() resetting the optimization bounds
void	setTargetBound() resetting the values of the target bounds (bounds for the remaining search)
protected void	setTargetLowerBound()
protected void	setTargetUpperBound()
void	solutionFound(IntBranchingTrace ctx)
void	storeMasterSolution()

Methods inherited from class choco.palm.benders.search.SubSearchSolver

changeGoal, fusionGoal, nextSolution, setCurrentFail

Methods inherited from class choco.search.AbstractGlobalSearchSolver

endTreeNode, endTreeSearch, getEncounteredLimit, incrementalRun, isEncounteredLimit, newTreeNode, newTreeSearch, popTrace, popTraceUntil, printRuntimeStatistics, pushTrace, run, setLoggingMaxDepth, topTrace

Methods inherited from class choco.AbstractSolver

existsSolution, makeSolutionFromCurrentState, restoreBestSolution, showSolution, storeSolution

Methods inherited from class choco.AbstractEntity

getProblem, pretty

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

doMaximize

public boolean doMaximize

a boolean indicating whether we want to maximize (true) or minize (false) the objective variable

objective

public ExplainedIntVar objective

the variable modelling the objective value

lowerBound

public int lowerBound

the lower bound of the objective value. This value comes from the problem definition; it is strengthened by the search history (solutions found & no-goods)

upperBound

public int upperBound

the upper bound of the objective value This value comes from the problem definition; it is strengthened by the search history (solutions found & no-goods)

targetUpperBound

public int targetUpperBound

a tentative upper bound

targetLowerBound

public int targetLowerBound

a tentative lower bound

msol

protected int[] msol

Master solution

Constructor Detail

SubOptimizer

protected SubOptimizer(IntDomainVar obj,
                       boolean maximize,
                       boolean slave)

constructor

Parameters:

obj - the objective variable

maximize - maximization or minimization ?

Method Detail

changeGoal

protected void changeGoal(AbstractIntBranching branching,
                          IntDomainVar newObjective)

nextOptimalSolution

public java.lang.Boolean nextOptimalSolution(int masterWorld)

Description copied from class: SubSearchSolver

get the next solution of the master problem

Overrides:

nextOptimalSolution in class SubSearchSolver

Returns:

solutionFound

public void solutionFound(IntBranchingTrace ctx)

Overrides:

solutionFound in class SubSearchSolver

getObjectiveValue

public int getObjectiveValue()

v1.0 accessing the objective value of an optimization problem (note that the objective value may not be instantiated, while all other variables are)

Returns:

the current objective value

getBestObjectiveValue

public int getBestObjectiveValue()

getObjectiveTarget

public int getObjectiveTarget()

the target for the objective function: we are searching for a solution at least as good as this (tentative bound)

initBounds

public void initBounds()

initialization of the optimization bound data structure

recordSolution

public void recordSolution()

Description copied from class: AbstractGlobalSearchSolver

called when a solution is encountered: printing and, if needed, storing the solution

Overrides:

recordSolution in class AbstractGlobalSearchSolver

setBound

public void setBound()

resetting the optimization bounds

setTargetBound

public void setTargetBound()

resetting the values of the target bounds (bounds for the remaining search)

setTargetLowerBound

protected void setTargetLowerBound()

setTargetUpperBound

protected void setTargetUpperBound()

postTargetBound

public void postTargetBound()
                     throws ContradictionException

propagating the optimization cuts from the new target bounds

Throws:

ContradictionException

postKnownBound

public void postKnownBound()
                    throws ContradictionException

Throws:

ContradictionException

postLowerBound

public void postLowerBound(int lb)
                    throws ContradictionException

Throws:

ContradictionException

postUpperBound

public void postUpperBound(int ub)
                    throws ContradictionException

Throws:

ContradictionException

postDynamicCut

public void postDynamicCut()
                    throws ContradictionException

we use targetBound data structures for the optimization cuts

Overrides:

postDynamicCut in class AbstractGlobalSearchSolver

Throws:

ContradictionException

storeMasterSolution

public void storeMasterSolution()

restoreBestSolutionBySearch

public void restoreBestSolutionBySearch()

logIntermediateSol

public void logIntermediateSol()