Globals Namespace Reference

Globals contains several definitions global to the MADP toolbox. More...

Typedefs
typedef unsigned int	Index
	A general index. More...
typedef unsigned long long int	LIndex
	A long long index. More...

Enumerations
enum	reward_t { REWARD, COST }
	Inherited from Tony's POMDP file format. More...

Functions
double	CastLIndexToDouble (LIndex i)
Index	CastLIndexToIndex (LIndex i)
bool	EqualProbability (double p1, double p2)
bool	EqualReward (double r1, double r2)

Variables
const size_t	ALL_SOLUTIONS =0
	constant to denote all solutions (e.g., nrDesiredSolutions = ALL_SOLUTIONS ) More...
const Index	INITIAL_JAOHI =0
	The initial (=empty) joint action-observation history index. More...
const Index	INITIAL_JOHI =0
	The initial (=empty) joint observation history index. More...
const unsigned int	MAXHORIZON =999999
	The highest horizon we will consider. More...
const double	PROB_PRECISION =1e-12
	The precision for probabilities. More...
const double	REWARD_PRECISION =1e-12
	Used to determine when two (immediate) rewards are considered equal. More...

Detailed Description

Globals contains several definitions global to the MADP toolbox.

Typedef Documentation

typedef unsigned int Globals::Index

A general index.

typedef unsigned long long int Globals::LIndex

A long long index.

Enumeration Type Documentation

enum Globals::reward_t

Inherited from Tony's POMDP file format.

Enumerator
REWARD
COST

Function Documentation

double Globals::CastLIndexToDouble

(

LIndex

i

)

Referenced by BGIP_SolverBranchAndBound< JP >::GetPerJPRatio(), GMAA_MAAstarCluster::SoftPrintClusteringStats(), GMAA_kGMAACluster::SoftPrintClusteringStats(), BGIP_SolverBruteForceSearch< JP >::Solve(), and BGIP_SolverBFSNonIncremental< JP >::Solve().

Index Globals::CastLIndexToIndex

(

LIndex

i

)

Referenced by ActionObservationHistory::ActionObservationHistory(), BGIPSolution::AddSolution(), MonahanBGPlanner::BackProjectMonahanBG(), BayesianGameWithClusterInfo::BayesianGameWithClusterInfo(), AlphaVectorBG::BeliefBackupExhaustiveStoreAll(), QPOMDP::ComputeRecursively(), QBG::ComputeRecursively(), QHybrid::ComputeRecursively(), QBG::ComputeRecursivelyNoCache(), BG_FactorGraphCreator::Construct_AgentBGPolicy_Variables(), BG_FactorGraphCreator::Construct_LocalPayoff_Factors(), GMAA_MAAstarClassic::ConstructAndValuateNextPolicies(), GMAA_kGMAA::ConstructAndValuateNextPolicies(), GMAA_MAAstar::ConstructAndValuateNextPolicies(), PlanningUnitMADPDiscrete::CreateActionHistoryTree(), PlanningUnitMADPDiscrete::CreateActionObservationHistoryTree(), PlanningUnitMADPDiscrete::CreateObservationHistoryTree(), BGCG_SolverNonserialDynamicProgramming::EliminateAgent(), BayesianGameWithClusterInfo::Extend(), BGforStageCreation::Fill_FirstOHtsI(), BayesianGameForDecPOMDPStage::Fill_FirstOHtsI(), GMAA_MAA_ELSI::Fill_FirstOHtsI(), GMAA_MAA_ELSI::Fill_jaI_Array(), PlanningUnitMADPDiscrete::GetActionHistoryArrays(), PlanningUnitMADPDiscrete::GetActionObservationHistoryArrays(), PlanningUnitMADPDiscrete::GetActionObservationHistoryIndex(), IndividualBeliefJESP::GetAugmentedStateIndex(), PlanningUnitMADPDiscrete::GetJAOHProbGivenPred(), PlanningUnitMADPDiscrete::GetJAOHProbs(), PlanningUnitMADPDiscrete::GetJointActionHistoryIndex(), JPolComponent_VectorImplementation::GetJointActionIndex(), PlanningUnitMADPDiscrete::GetJointActionObservationHistoryArrays(), PlanningUnitMADPDiscrete::GetJointActionObservationHistoryIndex(), PlanningUnitMADPDiscrete::GetJointActionObservationHistoryTree(), PlanningUnitMADPDiscrete::GetJointBeliefInterface(), PlanningUnitMADPDiscrete::GetJointObservationHistoryArrays(), PlanningUnitMADPDiscrete::GetJointObservationHistoryIndex(), BGCG_SolverNonserialDynamicProgramming::GetJpolIndexForBestResponses(), PlanningUnitMADPDiscrete::GetNrPolicyDomainElements(), PlanningUnitMADPDiscrete::GetObservationHistoryArrays(), PlanningUnitMADPDiscrete::GetObservationHistoryIndex(), IndividualBeliefJESP::GetOthersObservationHistIndex(), PlanningUnitMADPDiscrete::GetSuccessorAHI(), PlanningUnitMADPDiscrete::GetSuccessorAOHI(), PlanningUnitMADPDiscrete::GetSuccessorJAHI(), PlanningUnitMADPDiscrete::GetSuccessorJOHI(), PlanningUnitMADPDiscrete::GetSuccessorOHI(), PlanningUnitMADPDiscrete::GetTimeStepForAOHI(), AlphaVectorPlanning::ImportValueFunction(), PlanningUnitMADPDiscrete::InitializeJointActionObservationHistories(), PlanningUnitMADPDiscrete::InitializeJointObservationHistories(), JointObservationHistory::JointObservationHistory(), PlanningUnitMADPDiscrete::JointToIndividualActionObservationHistoryIndicesRef(), LocalBGValueFunctionVector::LocalBGValueFunctionVector(), BayesianGameForDecPOMDPStage::ProbRewardForjoahI(), GMAA_MAA_ELSI::ProbRewardForjoahI(), PlanningUnitMADPDiscrete::RegisterJointActionObservationHistoryTree(), PolicyPureVector::SetIndex(), and FSAOHDist_NECOF::Update().

bool Globals::EqualProbability	(	double	p1,
		double	p2
	)

References PROB_PRECISION.

Referenced by MultiAgentDecisionProcessDiscreteFactoredStates::CacheFlatObservationModel(), MultiAgentDecisionProcessDiscreteFactoredStates::CacheFlatTransitionModel(), DecPOMDPDiscrete::CompareModels(), Problem_CGBG_FF::ComputeLocalUtility(), BG_FactorGraphCreator::Construct_JointType_Factors(), BG_FactorGraphCreator::Construct_LocalJointType_Factors(), BayesianGameWithClusterInfo::ConstructClusteredIndividualTypes(), MultiAgentDecisionProcessDiscreteFactoredStates::MarginalizeTransitionObservationModel(), CPT::SanityCheck(), FSDist_COF::SanityCheck(), FSAOHDist_NECOF::SanityCheck(), BayesianGameBase::SanityCheckBGBase(), BayesianGameWithClusterInfo::TestApproximateEquivalence(), and BayesianGameWithClusterInfo::TestExactEquivalence().

bool Globals::EqualReward	(	double	r1,
		double	r2
	)

References REWARD_PRECISION.

Referenced by GMAA_MAAstar::ConstructAndValuateNextPolicies(), GMAA_MAAstarCluster::ConstructAndValuateNextPolicies(), AlphaVectorPlanning::EqualVS(), OptimalValueDatabase::IsOptimal(), AlphaVectorPruning::LexGreater(), and AlphaVectorPruning::Prune().

Variable Documentation

const size_t Globals::ALL_SOLUTIONS =0

constant to denote all solutions (e.g., nrDesiredSolutions = ALL_SOLUTIONS )

const Index Globals::INITIAL_JAOHI =0

The initial (=empty) joint action-observation history index.

Referenced by QFunctionJAOHTree::ComputeQ(), QHybrid::ComputeQ(), and PlanningUnitMADPDiscrete::GetJAOHProbs().

const Index Globals::INITIAL_JOHI =0

The initial (=empty) joint observation history index.

Referenced by ValueFunctionDecPOMDPDiscrete::CalculateV(), SimulationDecPOMDPDiscrete::RunSimulation(), and SimulationDecPOMDPDiscrete::RunSimulationClusteredBG().

const unsigned int Globals::MAXHORIZON =999999

The highest horizon we will consider.

When the horizon of a problem is set to this value, we consider it an infinite-horizon problem.

Referenced by BeliefSetNonStationary::BeliefSetNonStationary(), Perseus::CheckConvergence(), JPolComponent_VectorImplementation::Construct(), PlanningUnitMADPDiscrete::Deinitialize(), Perseus::GetInitialValueFunction(), BayesianGameBase::GetNrPolicyDomainElements(), PlanningUnitMADPDiscrete::GetNrPolicyDomainElements(), PlanningUnitMADPDiscrete::GetTimeStepForJAOHI(), MDPValueIteration::Initialize(), MDPPolicyIteration::Initialize(), MDPPolicyIterationGPU::Initialize(), PlanningUnitMADPDiscrete::Initialize(), SimulationDecPOMDPDiscrete::Initialize(), MDPSolver::Print(), AlphaVectorPlanning::SampleBeliefs(), AlphaVectorPlanning::SampleBeliefsNonStationary(), PlanningUnitDecPOMDPDiscrete::SanityCheck(), PartialJointPolicyPureVector::SetDepth(), JointPolicyPureVector::SetDepth(), and ArgumentHandlers::solutionMethodOptions_parse_argument().

const double Globals::PROB_PRECISION =1e-12

The precision for probabilities.

Used to determine when two probabilities are considered equal, for instance when converting full beliefs to sparse beliefs.

Referenced by QPOMDP::ComputeRecursively(), QBG::ComputeRecursively(), QHybrid::ComputeRecursively(), BayesianGameWithClusterInfo::ConstructClusteredIndividualTypes(), EqualProbability(), AlphaVectorPruning::FindBeliefAccelerated(), AlphaVectorPlanning::GetDuplicateIndices(), MultiAgentDecisionProcessDiscreteFactoredStates::Initialize2DBN(), FactoredDecPOMDPDiscrete::MarginalizeISD(), MultiAgentDecisionProcessDiscreteFactoredStates::MarginalizeTransitionObservationModel(), BG_FactorGraphCreator::PerturbationTerm(), BGIP_SolverBranchAndBound< JP >::ReSolve(), PlanningUnitDecPOMDPDiscrete::SanityCheck(), MultiAgentDecisionProcessDiscrete::SanityCheck(), Belief::SanityCheck(), BeliefSparse::SanityCheck(), MultiAgentDecisionProcessDiscreteFactoredStates::SanityCheckObservations(), MultiAgentDecisionProcessDiscreteFactoredStates::SanityCheckTransitions(), ObservationModelMappingSparse::Set(), TransitionModelMappingSparse::Set(), EventObservationModelMappingSparse::Set(), MADPComponentDiscreteStates::SetInitialized(), BGIP_SolverBruteForceSearch< JP >::Solve(), BGIP_SolverBFSNonIncremental< JP >::Solve(), JointBeliefSparse::Update(), and JointBeliefSparse::UpdateSlow().

const double Globals::REWARD_PRECISION =1e-12

Used to determine when two (immediate) rewards are considered equal.

Referenced by PerseusConstrainedPOMDPPlanner::BackupStage(), FactoredDecPOMDPDiscrete::CacheFlatRewardModel(), FactoredDecPOMDPDiscrete::ClipRewardModel(), EqualReward(), AlphaVectorPruning::FindBeliefAccelerated(), RewardModelMappingSparse::Set(), RewardModelMappingSparseMapped::Set(), and AlphaVectorPlanning::VectorIsInValueFunction().