Module org.chocosolver.solver
Package org.chocosolver.solver

Class Settings

  • public class Settings
extends Object
    Settings for Model and Solver. Can be modified programmatically and can be defined in a Model only on creation.

    Project: choco-solver.

    Since:
    14/12/2017.
    Author:
    Charles Prud'homme

    • Method Detail

      • init

        public static Settings init()
        Create a new instance of `Settings` which can then be adapted to requirements.
        Returns:
        a Settings with default values
        See Also:
        dev(), prod()

      • prod

        public static Settings prod()
        Define and returns settings adapted to production environment. All checks and warnings are turned off.
        Returns:
        a settings adapted to production environment.

      • dev

        public static Settings dev()
        Define and returns settings adapted to development environment. All checks and warnings are turned on.
        Returns:
        a settings adapted to development environment.

      • checkModel

        public boolean checkModel​(Solver solver)
        Parameters:
        solver - the solver
        Returns:
        true if the model is OK wrt the checker, false otherwise

      • setModelChecker

        public Settings setModelChecker​(Predicate<Solver> modelChecker)
        Define what to do when a solution is found. By default, it makes a weak check of the model: 
             
         return !ESat.FALSE.equals(solver.isSatisfied());
        A hard check of the model can be done like this: 
             
     return ESat.TRUE.equals(solver.isSatisfied());
        Parameters:
        modelChecker - a predicate to check the solution
        Returns:
        the current instance

      • cloneVariableArrayInPropagator

        public boolean cloneVariableArrayInPropagator()
        Returns:
        true if all propagators should clone the input variable array instead of simply referencing it.

      • setCloneVariableArrayInPropagator

        public Settings setCloneVariableArrayInPropagator​(boolean cloneVariableArrayInPropagator)
        If this setting is set to true (default value), a clone of the input variable array is made in any propagator constructors. This prevents, for instance, wrong behavior when permutations occurred on the input array (e.g., sorting variables). Setting this to false may limit the memory consumption during modelling.
        Parameters:
        cloneVariableArrayInPropagator - true to clone variables array on constructor
        Returns:
        the current instance

      • enableViews

        public boolean enableViews()
        Returns:
        true if views are enabled.

      • setEnableViews

        public Settings setEnableViews​(boolean enableViews)
        Set to 'true' to allow the creation of views in the Model. Creates new variables with channeling constraints otherwise.
        Parameters:
        enableViews - true to enable views
        Returns:
        the current instance

      • getMaxDomSizeForEnumerated

        public int getMaxDomSizeForEnumerated()
        Returns:
        maximum domain size threshold to force integer variable to be enumerated

      • setMaxDomSizeForEnumerated

        public Settings setMaxDomSizeForEnumerated​(int maxDomSizeForEnumerated)
        Define the minimum number of cardinality threshold to a sum/scalar constraint to be decomposed in intermediate sub-sums.
        Parameters:
        maxDomSizeForEnumerated - cardinality threshold
        Returns:
        the current instance

      • getMinCardForSumDecomposition

        public int getMinCardForSumDecomposition()
        Returns:
        minimum number of cardinality threshold to a sum constraint to be decomposed

      • setMinCardinalityForSumDecomposition

        public Settings setMinCardinalityForSumDecomposition​(int defaultMinCardinalityForSumDecomposition)
        Define the default minimum number of cardinality threshold to a sum/scalar constraint to be decomposed into intermediate sub-sums.
        Parameters:
        defaultMinCardinalityForSumDecomposition - cardinality threshold
        Returns:
        the current instance

      • enableTableSubstitution

        public boolean enableTableSubstitution()
        Returns:
        true if some intension constraints can be replaced by extension constraints

      • setEnableTableSubstitution

        public Settings setEnableTableSubstitution​(boolean enableTableSubstitution)
        Define whether some intension constraints are replaced by extension constraints
        Parameters:
        enableTableSubstitution - enable table substitution
        Returns:
        the current instance

      • getMaxTupleSizeForSubstitution

        public int getMaxTupleSizeForSubstitution()
        Returns:
        maximum domain size threshold to replace intension constraints by extension constraints

      • setMaxTupleSizeForSubstitution

        public Settings setMaxTupleSizeForSubstitution​(int maxTupleSizeForSubstitution)
        Define the maximum domain size threshold to replace intension constraints by extension constraints Only checked when enableTableSubstitution() returns true
        Parameters:
        maxTupleSizeForSubstitution - threshold to substitute intension constraint by table one.
        Returns:
        the current instance

      • sortPropagatorActivationWRTPriority

        public boolean sortPropagatorActivationWRTPriority()
        Returns:
        true if propagators are sorted wrt their priority on initial activation.

      • setSortPropagatorActivationWRTPriority

        public Settings setSortPropagatorActivationWRTPriority​(boolean sortPropagatorActivationWRTPriority)
        Set whether propagators are sorted wrt their priority in PropagationEngine when dealing with propagator activation.
        Parameters:
        sortPropagatorActivationWRTPriority - true to allow sorting static propagators.
        Returns:
        the current instance

      • getMaxPropagatorPriority

        public int getMaxPropagatorPriority()
        Returns:
        the maximum priority any propagators can have (default is 7)

      • setMaxPropagatorPriority

        public Settings setMaxPropagatorPriority​(int maxPropagatorPriority)
        Increase the number of priority for propagators (default is PropagatorPriority.VERY_SLOW). This directly impacts the number of queues to schedule propagators in the propagation engine.
        Parameters:
        maxPropagatorPriority - the new maximum prioirity any propagator can declare
        Returns:
        the current instance

      • makeDefaultSearch

        public AbstractStrategy<?> makeDefaultSearch​(Model model)
        Creates a default search strategy for the input model
        Parameters:
        model - a model requiring a default search strategy
        Returns:
        a default search strategy for model
        See Also:
        Search.defaultSearch(Model)

      • setDefaultSearch

        public Settings setDefaultSearch​(Function<Model,​AbstractStrategy<?>> defaultSearch)
        Define a default search strategy for the input model
        Parameters:
        defaultSearch - what default search strategy should be
        Returns:
        the current instance

      • getEnvironmentHistorySimulationCondition

        @Deprecated
public ICondition getEnvironmentHistorySimulationCondition()
        Deprecated.

      • setEnvironmentHistorySimulationCondition

        @Deprecated
public Settings setEnvironmentHistorySimulationCondition​(ICondition environmentHistorySimulationCondition)
        Deprecated.

      • warnUser

        public boolean warnUser()
        Returns:
        true if warnings detected during modeling/solving are output.

      • setWarnUser

        public Settings setWarnUser​(boolean warnUser)
        To be informed of warnings detected during modeling/solving
        Parameters:
        warnUser - true to be print warnings on console
        Returns:
        the current instance

      • enableDecompositionOfBooleanSum

        public boolean enableDecompositionOfBooleanSum()
        Returns:
        true if boolean sum should be decomposed into an equality constraint and an arithmetic constraint, falseif a single constraint should be used instead.

      • setEnableDecompositionOfBooleanSum

        public Settings setEnableDecompositionOfBooleanSum​(boolean enableDecompositionOfBooleanSum)
        Define if boolean sums should be decomposed into an equality constraint + arithmetic constraint
        Parameters:
        enableDecompositionOfBooleanSum - true to enable decomposition
        Returns:
        the current instance

      • enableIncrementalityOnBoolSum

        public boolean enableIncrementalityOnBoolSum​(int nbvars)
        Parameters:
        nbvars - number of variables in the constraint
        Returns:
        true if the incrementality is enabled on boolean sum, based on the number of variables involved.

      • setEnableIncrementalityOnBoolSum

        public Settings setEnableIncrementalityOnBoolSum​(IntPredicate enableIncrementalityOnBoolSum)
        Define the predicate to choose incremental sum, based on number variables declared
        Parameters:
        enableIncrementalityOnBoolSum - predicate to pick declare sum
        Returns:
        the current instance

      • enableSAT

        public boolean enableSAT()
        Returns:
        true when an underlying SAT solver is used to manage clauses declared through ISatFactory, false when clauses are managed with CSP constraints only.

      • setEnableSAT

        public Settings setEnableSAT​(boolean enableSAT)
        Indicate if clauses are managed by a unique SAT solver.
        Parameters:
        enableSAT - true to rely on SAT Solver to handle clauses
        Returns:
        the current instance

      • swapOnPassivate

        public boolean swapOnPassivate()
        Returns:
        true if, on propagator passivation, the propagator is swapped from active to passive in its variables' propagators list. false if, on propagator passivation, only the propagator's state is set to PASSIVE.

      • setSwapOnPassivate

        public Settings setSwapOnPassivate​(boolean swapOnPassivate)
        Define if passivation of propagator swap it in variables' list
        Parameters:
        swapOnPassivate - true to enable swapping
        Returns:
        the current instance

      • checkDeclaredConstraints

        public boolean checkDeclaredConstraints()
        Returns:
        true (default value) to check if all declared constraints are not free anymore, that is either posted or reified, before running the resolution. false to skip the control.

      • setCheckDeclaredConstraints

        public Settings setCheckDeclaredConstraints​(boolean checkDeclaredConstraints)
        Indicate if the declared constraints are either posted or reified.
        Parameters:
        checkDeclaredConstraints - true to check constraints before resolution
        Returns:
        the current instance

      • printAllUndeclaredConstraints

        public boolean printAllUndeclaredConstraints()
        Returns:
        true to list all undeclared constraint, false (default value) otherwise. Only active when checkDeclaredConstraints() is on.

      • setPrintAllUndeclaredConstraints

        public Settings setPrintAllUndeclaredConstraints​(boolean printAllUndeclaredConstraints)
        Indicate if all undeclared constraints are listed on console when checkDeclaredConstraints() is on.
        Parameters:
        printAllUndeclaredConstraints - true to list all undeclared constraints
        Returns:
        the current instance

      • checkDeclaredViews

        public boolean checkDeclaredViews()
        Returns:
        true (default value) to check prior to creation if a view already semantically exists.

      • setCheckDeclaredViews

        public Settings setCheckDeclaredViews​(boolean checkDeclaredViews)
        Check if a view already semantically exists before creating it.
        Parameters:
        checkDeclaredViews - true to check views before creation
        Returns:
        the current instance

      • setCheckDeclaredMonitors

        public Settings setCheckDeclaredMonitors​(boolean check)

      • checkDeclaredMonitors

        public boolean checkDeclaredMonitors()

      • initSolver

        public Solver initSolver​(Model model)
        This method is called in Model(IEnvironment, String, Settings) to create the solver to associate with a model.
        Parameters:
        model - a model to initialize with a solver
        Returns:
        the new solver

      • setInitSolver

        public Settings setInitSolver​(Function<Model,​Solver> initSolver)
        Define the solver initialization
        Parameters:
        initSolver - function to initialize the solver
        Returns:
        the current instance

      • enableHybridizationOfPropagationEngine

        public byte enableHybridizationOfPropagationEngine()
        Returns:
        0b00 if constraint-oriented propagation engine, 0b01 if hybridization between variable and constraint oriented and 0b10 if variable-oriented.

      • setHybridizationOfPropagationEngine

        public Settings setHybridizationOfPropagationEngine​(byte hybrid)
        Define behavior of the propagation engine.
        Parameters:
        hybrid - When set to '0b00', this works as a constraint-oriented propagation engine; when set to '0b01', this workds as an hybridization between variable and constraint oriented propagation engine. when set to '0b10', this workds as a variable- oriented propagation engine.
        Returns:
        the current instance

      • getLearntClausesDominancePerimeter

        public int getLearntClausesDominancePerimeter()
        When a clause is learnt from a conflict, it may happen that it dominates previously learnt ones. The dominance will be evaluated with the n last learnt clauses. n = 0 means no dominance check, n = Integer.MAX_VALUE means checking all clauses with the last one.
        Returns:
        dominance perimeter

      • setLearntClausesDominancePerimeter

        public Settings setLearntClausesDominancePerimeter​(int n)
        When a clause is learnt from a conflict, it may happen that it dominates previously learnt ones. The dominance will be evaluated with the n last learnt clauses. n = 0 means no dominance check, n = Integer.MAX_VALUE means checking all clauses with the last one.
        Returns:
        dominance perimeter

      • explainGlobalFailureInSum

        public boolean explainGlobalFailureInSum()
        Returns:
        true if additional clauses can be learned from sum's global failure

      • explainGlobalFailureInSum

        public Settings explainGlobalFailureInSum​(boolean b)
        Set to true to allow additional clauses to be learned from sum's global failure

      • getIbexContractionRatio

        public double getIbexContractionRatio()
        Returns:
        the ratio that a domains must be contracted by ibex to compute the constraint.

      • setIbexContractionRatio

        public void setIbexContractionRatio​(double ibexContractionRatio)
        Defines the ratio that real domains must be contracted by ibex to compute the constraint. A contraction is considered as significant when at least {@param ratio} of a domain has been reduced. If the contraction is not meet, then it is considered as insufficient and therefore ignored. A too small ratio can degrade the ibex performance. The default value is 1% (0.01). See issue #653.

        Example: given x = [0.0, 100.0], y = [0.5,0.5] and CSTR(x > y) - When the ratio is 1% (0.01) bounds of X are kept as [0.0, 100.0] because it's contraction is less than 1%. - When the ratio is 0.1% (0.001) bounds of X are update to [0.5, 100.0] because it's contraction is greater than 0.1%.

        Parameters:
        ibexContractionRatio - defines the ratio that a domains must be contract to compute the constraint.

      • setIbexRestoreRounding

        public Settings setIbexRestoreRounding​(boolean ibexRestoreRounding)
        If preserve_rounding is true, Ibex will restore the default Java rounding method when coming back from Ibex, which is transparent for Java but causes a little loss of efficiency. To improve the running time, ibex changes the rounding system for double values during contraction. In Linux/MACOS environments it leads to different results in calculations like `Math.pow(10, 6)`. See issue #740.
        Parameters:
        ibexRestoreRounding - either Java or ibex rounding method

      • getIbexRestoreRounding

        public boolean getIbexRestoreRounding()
        Returns:
        if ibex must restore java rounding mode when returning a call.