TabularBeliefState

java.lang.Object
- burlap.oomdp.singleagent.pomdp.beliefstate.tabular.TabularBeliefState

All Implemented Interfaces:

State, BeliefState, DenseBeliefVector, EnumerableBeliefState
```
public class TabularBeliefState
extends java.lang.Object
implements BeliefState, EnumerableBeliefState, DenseBeliefVector
```
A class for storing a sparse tabular representation of the belief state. That is, each MDP state is assigned a unique identifier using a StateEnumerator and this class uses a Map to associated the probability mass with each state. MDP states that have zero mass are not stored in the map.
The OO-MDP representation of this state associates with a singleton SADomain that has one ObjectClass named belief, that has one double array attribute named belief. When An OO-MDP state representation is request, an MutableObjectInstance belonging to class belief is created with its belief attribute set to the dense (non-sparse) belief vector that this BeliefState represents.
If using a BeliefMDP solver with a TabularBeliefState, it is recommended that you use HashableTabularBeliefStateFactory which will compute hash codes and perform state equality checks with the sparse Map representation (rather than the dense OO-MDP representation)

Author:

James MacGlashan.

Nested Class Summary
- Nested classes/interfaces inherited from interface burlap.oomdp.singleagent.pomdp.beliefstate.EnumerableBeliefState
  EnumerableBeliefState.StateBelief

Field Summary

Fields
Modifier and Type	Field and Description
`static java.lang.String`	`BELIEFATTNAME`
`static java.lang.String`	`BELIEFCLASSNAME`
`protected java.util.Map<java.lang.Integer,java.lang.Double>`	`beliefValues` The belief vector, stored sparsely with a `Map`.
`protected PODomain`	`domain` The POMDP domain with which this belief state is associated.
`protected StateEnumerator`	`stateEnumerator` A state enumerator for determining the index of MDP states in the belief vector.

Constructor Summary

Constructors
Constructor and Description
`TabularBeliefState(PODomain domain)`
`TabularBeliefState(PODomain domain, StateEnumerator stateEnumerator)` Constructs a new `TabularBeliefState`.
`TabularBeliefState(TabularBeliefState srcBeliefState)` Constructs a new `TabularBeliefState` from a source `TabularBeliefState`.

Method Summary

Methods
Modifier and Type	Method and Description
`State`	`addAllObjects(java.util.Collection<ObjectInstance> objects)` Adds the collection of objects to the state
`State`	`addObject(ObjectInstance o)` Adds object instance o to this state.
`double`	`belief(int stateId)` Returns the value of the belief vector for the provided index.
`double`	`belief(State s)` Returns the probability density/mass for the input MDP state.
`State`	`copy()` Returns a copy of this state, if mutable the copy should be deep.
`boolean`	`equals(java.lang.Object obj)`
`java.util.List<ObjectInstance>`	`getAllObjects()` Returns the list of observable and hidden object instances in this state.
`java.util.List<java.util.List<ObjectInstance>>`	`getAllObjectsByClass()` Returns a list of list of object instances, grouped by object class
`java.util.Map<java.lang.String,java.util.List<java.lang.String>>`	`getAllUnsetAttributes()` Returns a mapping from object instance names to the list of attributes names that have unset values.
`double[]`	`getBeliefVector()` Returns this belief state as a dense (non-sparse) belief vector.
`java.lang.String`	`getCompleteStateDescription()` Returns a string representation of this state using observable and hidden object instances.
`java.lang.String`	`getCompleteStateDescriptionWithUnsetAttributesAsNull()` Returns a string description of the state with unset attribute values listed as null.
`ObjectInstance`	`getFirstObjectOfClass(java.lang.String oclass)` Returns the first indexed object of the object class named oclass
`ObjectInstance`	`getObject(java.lang.String oname)` Returns the object in this state with the name oname
`java.util.Set<java.lang.String>`	`getObjectClassesPresent()` Returns a set of of the object class names for all object classes that have instantiated objects in this state.
`java.util.Map<java.lang.String,java.lang.String>`	`getObjectMatchingTo(State so, boolean enforceStateExactness)` This method computes a matching from objects in the receiver to value-identical objects in the parameter state so.
`java.util.List<ObjectInstance>`	`getObjectsOfClass(java.lang.String oclass)` Returns all objects that belong to the object class named oclass
`java.util.List<java.util.List<java.lang.String>>`	`getPossibleBindingsGivenParamOrderGroups(java.lang.String[] paramClasses, java.lang.String[] paramOrderGroups)` Given an array of parameter object classes and an array of their corresponding parameter order groups, returns all possible object instance bindings to the parameters, excluding bindings that are equivalent due to the parameter order grouping.
`java.util.List<EnumerableBeliefState.StateBelief>`	`getStatesAndBeliefsWithNonZeroProbability()` Returns the states, and their probability mass, that have non-zero probability mass.
`java.util.List<State>`	`getStateSpace()` Returns the set of underlying MDP states this belief vector spans.
`static SADomain`	`getTabularBeliefMDPDomain()`
`protected double`	`getTransitionProb(State s, GroundedAction ga, State sp)` Returns the probability that the underlying MDP will transition from state s to sp when taking action a in state s.
`BeliefState`	`getUpdatedBeliefState(State observation, GroundedAction ga)` Computes a new belief distribution using this BeliefState as the prior and conditioned on the given POMDP observation and action taken.
`void`	`initializeAllBeliefValuesTo(double initialValue)` Initializes the probability mass of all underlying MDP states to the specified value.
`void`	`initializeBeliefsUniformly()` Initializes this belief state to a uniform distribution
`int`	`numStates()` Returns the size of the observed underlying MDP state space.
`int`	`numTotalObjects()` Returns the number of object instances in this state.
`State`	`removeAllObjects(java.util.Collection<ObjectInstance> objects)` Removes the collection of objects from the state
`State`	`removeObject(ObjectInstance o)` Removes the object instance o from this state.
`State`	`removeObject(java.lang.String oname)` Removes the object instance with the name oname from this state.
`State`	`renameObject(ObjectInstance o, java.lang.String newName)` Renames the identifier for object instance o in this state to newName.
`State`	`renameObject(java.lang.String originalName, java.lang.String newName)` Renames the identifier for the object instance currently named originalName with the name newName.
`State`	`sampleStateFromBelief()` Samples an MDP state state from this belief distribution.
`void`	`setBelief(int stateId, double b)` Sets the probability mass (belief) associated with the underlying MDP state.
`void`	`setBelief(State s, double b)` Sets the probability mass (belief) associated with the underlying MDP state.
`void`	`setBeliefVector(double[] b)` Sets this belief state to the provided.
`<T> State`	`setObjectsValue(java.lang.String objectName, java.lang.String attName, T value)`
`State`	`stateForId(int id)` Returns the corresponding MDP state for the provided unique identifier.
`void`	`zeroOutBeliefVector()` Sets this belief state to have zero probability mass for all underlying MDP states.

Methods inherited from class java.lang.Object
clone, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

- Field Detail
  - BELIEFCLASSNAME
```
public static final java.lang.String BELIEFCLASSNAME
```
    See Also:
    Constant Field Values
  - BELIEFATTNAME
```
public static final java.lang.String BELIEFATTNAME
```
    See Also:
    Constant Field Values
  - stateEnumerator
```
protected StateEnumerator stateEnumerator
```
    A state enumerator for determining the index of MDP states in the belief vector.
  - beliefValues
```
protected java.util.Map<java.lang.Integer,java.lang.Double> beliefValues
```
    The belief vector, stored sparsely with a Map.
  - domain
```
protected PODomain domain
```
    The POMDP domain with which this belief state is associated. Contains the ObservationFunction necessary to perform belief state updates.
- Constructor Detail
  - TabularBeliefState
```
public TabularBeliefState(TabularBeliefState srcBeliefState)
```
    Constructs a new TabularBeliefState from a source TabularBeliefState. Changes to the new state or source state will not affect the other.
    
    Parameters:
    srcBeliefState - the source TabularBeliefState to copy.
  - TabularBeliefState
```
public TabularBeliefState(PODomain domain)
```
  - TabularBeliefState
```
public TabularBeliefState(PODomain domain,
                  StateEnumerator stateEnumerator)
```
    Constructs a new TabularBeliefState.
    
    Parameters:
    domain - the PODomain domain to which the belief state is associated.
    stateEnumerator - a StateEnumerator to index the states in the belief vector.
- Method Detail
  - getTabularBeliefMDPDomain
```
public static SADomain getTabularBeliefMDPDomain()
```
  - belief
```
public double belief(State s)
```
    Description copied from interface: BeliefState
    
    Returns the probability density/mass for the input MDP state.
    
    Specified by:
    
    belief in interface BeliefState
    
    Parameters:
    s - the the input MDP state defined by a State instance.
    
    Returns:
    the probability density/mass of the input MDP state in this belief distribution.
  - belief
```
public double belief(int stateId)
```
    Returns the value of the belief vector for the provided index.
    
    Parameters:
    stateId - the index (state identification number) of the belief vector to return.
    
    Returns:
    the value of the belief vector for the provided index.
  - sampleStateFromBelief
```
public State sampleStateFromBelief()
```
    Description copied from interface: BeliefState
    
    Samples an MDP state state from this belief distribution.
    
    Specified by:
    
    sampleStateFromBelief in interface BeliefState
    
    Returns:
    an MDP state defined by a State instance.
  - getUpdatedBeliefState
```
public BeliefState getUpdatedBeliefState(State observation,
                                GroundedAction ga)
```
    Description copied from interface: BeliefState
    
    Computes a new belief distribution using this BeliefState as the prior and conditioned on the given POMDP observation and action taken.
    
    Specified by:
    
    getUpdatedBeliefState in interface BeliefState
    
    Parameters:
    observation - the conditioned POMDP observation defined by a State instance.
    ga - the conditioned action selection in the previous time step.
    
    Returns:
    the new belief state distribution represented by a new BeliefState instance.
  - getStatesAndBeliefsWithNonZeroProbability
```
public java.util.List<EnumerableBeliefState.StateBelief> getStatesAndBeliefsWithNonZeroProbability()
```
    Description copied from interface: EnumerableBeliefState
    
    Returns the states, and their probability mass, that have non-zero probability mass. States that are not included in the returned listed are assumed to have probability mass zero.
    
    Specified by:
    
    getStatesAndBeliefsWithNonZeroProbability in interface EnumerableBeliefState
    
    Returns:
    a List of EnumerableBeliefState.StateBelief objects specifying the enumerated probability mass function.
  - numStates
```
public int numStates()
```
    Returns the size of the observed underlying MDP state space.
    
    Returns:
    the size of the observed underlying MDP state space.
  - stateForId
```
public State stateForId(int id)
```
    Returns the corresponding MDP state for the provided unique identifier.
    
    Parameters:
    id - the MDP state identifier
    
    Returns:
    the corresponding MDP state, defined by a State, for the provided unique identifier.
  - setBelief
```
public void setBelief(State s,
             double b)
```
    Sets the probability mass (belief) associated with the underlying MDP state. If this object has not yet assigned a unique identifier to the provided MDP state, then it will first create one. Note that using this method will not ensure that the total probability mass across this belief state sums to 1, so other changes will have to be specified manually.
    
    Parameters:
    s - the underlying MDP state defined as a State
    b - the probability mass to assigned to the underlying MDP state.
  - setBelief
```
public void setBelief(int stateId,
             double b)
```
    Sets the probability mass (belief) associated with the underlying MDP state. Note that using this method will not ensure that the total probability mass across this belief state sums to 1, so other changes will have to be specified manually.
    
    Parameters:
    stateId - the unique numeric identifier of the underlying MDP state defined.
    b - the probability mass to assigned to the underlying MDP state.
  - getTransitionProb
```
protected double getTransitionProb(State s,
                       GroundedAction ga,
                       State sp)
```
    Returns the probability that the underlying MDP will transition from state s to sp when taking action a in state s.
    
    Parameters:
    s - the previous MDP state defined by a State
    ga - the taken action defined by a GroundedAction
    sp - The next MDP state observed defined by a State.
    
    Returns:
    the probability that the underlying MDP will transition from state s to sp when taking action a in state s.
  - setBeliefVector
```
public void setBeliefVector(double[] b)
```
    Sets this belief state to the provided. Dense belief vector. If the belief vector dimensionality does not match this objects dimensionality then a runtime exception will be thrown.
    
    Specified by:
    
    setBeliefVector in interface DenseBeliefVector
    
    Parameters:
    b - the belief vector to set this belief state to.
  - getBeliefVector
```
public double[] getBeliefVector()
```
    Returns this belief state as a dense (non-sparse) belief vector.
    
    Specified by:
    
    getBeliefVector in interface DenseBeliefVector
    
    Returns:
    a double array specifying this belief state as a dense (non-sparse) belief vector.
  - getStateSpace
```
public java.util.List<State> getStateSpace()
```
    Returns the set of underlying MDP states this belief vector spans.
    
    Returns:
    the set of underlying MDP states this belief vector spans.
  - zeroOutBeliefVector
```
public void zeroOutBeliefVector()
```
    Sets this belief state to have zero probability mass for all underlying MDP states.
  - initializeBeliefsUniformly
```
public void initializeBeliefsUniformly()
```
    Initializes this belief state to a uniform distribution
  - initializeAllBeliefValuesTo
```
public void initializeAllBeliefValuesTo(double initialValue)
```
    Initializes the probability mass of all underlying MDP states to the specified value. Note that this will not enforce summing to one, so you will have to manually modify the value of other states if initialValue is not 1/n where n is the number of underlying MDP states.
    
    Parameters:
    initialValue - the probability mass to assign to each state.
  - copy
```
public State copy()
```
    Description copied from interface: State
    
    Returns a copy of this state, if mutable the copy should be deep.
    
    Specified by:
    
    copy in interface State
    
    Returns:
    a copy of this state.
  - addObject
```
public State addObject(ObjectInstance o)
```
    Description copied from interface: State
    
    Adds object instance o to this state.
    
    Specified by:
    
    addObject in interface State
    
    Parameters:
    o - the object instance to be added to this state.
    
    Returns:
    the modified state
  - addAllObjects
```
public State addAllObjects(java.util.Collection<ObjectInstance> objects)
```
    Description copied from interface: State
    
    Adds the collection of objects to the state
    
    Specified by:
    
    addAllObjects in interface State
    
    Returns:
    the modified state
  - removeObject
```
public State removeObject(java.lang.String oname)
```
    Description copied from interface: State
    
    Removes the object instance with the name oname from this state.
    
    Specified by:
    
    removeObject in interface State
    
    Parameters:
    oname - the name of the object instance to remove.
    
    Returns:
    the modified state
  - removeObject
```
public State removeObject(ObjectInstance o)
```
    Description copied from interface: State
    
    Removes the object instance o from this state.
    
    Specified by:
    
    removeObject in interface State
    
    Parameters:
    o - the object instance to remove from this state.
    
    Returns:
    the modified state
  - removeAllObjects
```
public State removeAllObjects(java.util.Collection<ObjectInstance> objects)
```
    Description copied from interface: State
    
    Removes the collection of objects from the state
    
    Specified by:
    
    removeAllObjects in interface State
    
    Returns:
    the modified state
  - renameObject
```
public State renameObject(java.lang.String originalName,
                 java.lang.String newName)
```
    Description copied from interface: State
    
    Renames the identifier for the object instance currently named originalName with the name newName.
    
    Specified by:
    
    renameObject in interface State
    
    Parameters:
    originalName - the original name of the object instance to be renamed in this state
    newName - the new name of the object instance
    
    Returns:
    the modified state
  - renameObject
```
public State renameObject(ObjectInstance o,
                 java.lang.String newName)
```
    Description copied from interface: State
    
    Renames the identifier for object instance o in this state to newName.
    
    Specified by:
    
    renameObject in interface State
    
    Parameters:
    o - the object instance to rename in this state
    newName - the new name of the object instance
    
    Returns:
    the modified state
  - getObjectMatchingTo
```
public java.util.Map<java.lang.String,java.lang.String> getObjectMatchingTo(State so,
                                                                   boolean enforceStateExactness)
```
    Description copied from interface: State
    
    This method computes a matching from objects in the receiver to value-identical objects in the parameter state so. The matching is returned as a map from the object names in the receiving state to the matched objects in state so. If enforceStateExactness is set to true, then the returned matching will be an empty map if the two states are not OO-MDP-wise identical (i.e., if there is a not a bijection between value-identical objects of the two states). If enforceExactness is false and the states are not identical, the the method will return the largest matching between objects that can be made.
    
    Specified by:
    
    getObjectMatchingTo in interface State
    
    Parameters:
    so - the state to whose objects the receiving state's objects should be matched
    enforceStateExactness - whether to require that states are identical to return a matching
    
    Returns:
    a matching from this receiving state's objects to objects in so that have identical values.
  - numTotalObjects
```
public int numTotalObjects()
```
    Description copied from interface: State
    
    Returns the number of object instances in this state.
    
    Specified by:
    
    numTotalObjects in interface State
    
    Returns:
    the number of object instances in this state.
  - getObject
```
public ObjectInstance getObject(java.lang.String oname)
```
    Description copied from interface: State
    
    Returns the object in this state with the name oname
    
    Specified by:
    
    getObject in interface State
    
    Parameters:
    oname - the name of the object instance to return
    
    Returns:
    the object instance with the name oname or null if there is no object in this state named oname
  - getAllObjects
```
public java.util.List<ObjectInstance> getAllObjects()
```
    Description copied from interface: State
    
    Returns the list of observable and hidden object instances in this state.
    
    Specified by:
    
    getAllObjects in interface State
    
    Returns:
    the list of observable and hidden object instances in this state.
  - getObjectsOfClass
```
public java.util.List<ObjectInstance> getObjectsOfClass(java.lang.String oclass)
```
    Description copied from interface: State
    
    Returns all objects that belong to the object class named oclass
    
    Specified by:
    
    getObjectsOfClass in interface State
    
    Parameters:
    oclass - the name of the object class for which objects should be returned
    
    Returns:
    all objects that belong to the object class named oclass
  - getFirstObjectOfClass
```
public ObjectInstance getFirstObjectOfClass(java.lang.String oclass)
```
    Description copied from interface: State
    
    Returns the first indexed object of the object class named oclass
    
    Specified by:
    
    getFirstObjectOfClass in interface State
    
    Parameters:
    oclass - the name of the object class for which the first indexed object should be returned.
    
    Returns:
    the first indexed object of the object class named oclass
  - getObjectClassesPresent
```
public java.util.Set<java.lang.String> getObjectClassesPresent()
```
    Description copied from interface: State
    
    Returns a set of of the object class names for all object classes that have instantiated objects in this state.
    
    Specified by:
    
    getObjectClassesPresent in interface State
    
    Returns:
    a set of of the object class names for all object classes that have instantiated objects in this state.
  - getAllObjectsByClass
```
public java.util.List<java.util.List<ObjectInstance>> getAllObjectsByClass()
```
    Description copied from interface: State
    
    Returns a list of list of object instances, grouped by object class
    
    Specified by:
    
    getAllObjectsByClass in interface State
    
    Returns:
    a list of list of object instances, grouped by object class
  - getCompleteStateDescription
```
public java.lang.String getCompleteStateDescription()
```
    Description copied from interface: State
    
    Returns a string representation of this state using observable and hidden object instances.
    
    Specified by:
    
    getCompleteStateDescription in interface State
    
    Returns:
    a string representation of this state using observable and hidden object instances.
  - getAllUnsetAttributes
```
public java.util.Map<java.lang.String,java.util.List<java.lang.String>> getAllUnsetAttributes()
```
    Description copied from interface: State
    
    Returns a mapping from object instance names to the list of attributes names that have unset values.
    
    Specified by:
    
    getAllUnsetAttributes in interface State
    
    Returns:
    a mapping from object instance names to the list of attributes names that have unset values.
  - getCompleteStateDescriptionWithUnsetAttributesAsNull
```
public java.lang.String getCompleteStateDescriptionWithUnsetAttributesAsNull()
```
    Description copied from interface: State
    
    Returns a string description of the state with unset attribute values listed as null. This avoids runtime exceptions when attributes are unset and informs you which they are.
    
    Specified by:
    
    getCompleteStateDescriptionWithUnsetAttributesAsNull in interface State
    
    Returns:
    a string description of the state with unset attribute values listed as null.
  - getPossibleBindingsGivenParamOrderGroups
```
public java.util.List<java.util.List<java.lang.String>> getPossibleBindingsGivenParamOrderGroups(java.lang.String[] paramClasses,
                                                                                        java.lang.String[] paramOrderGroups)
```
    Description copied from interface: State
    
    Given an array of parameter object classes and an array of their corresponding parameter order groups, returns all possible object instance bindings to the parameters, excluding bindings that are equivalent due to the parameter order grouping.
    
    Specified by:
    
    getPossibleBindingsGivenParamOrderGroups in interface State
    
    Parameters:
    paramClasses - the name of object classes to which the bound object instances must belong
    paramOrderGroups - the parameter order group names.
    
    Returns:
    A list of all possible object instance bindings for the parameters, were a binding is represented by a list of object instance names
  - equals
```
public boolean equals(java.lang.Object obj)
```
    Overrides:
    
    equals in class java.lang.Object
  - setObjectsValue
```
public <T> State setObjectsValue(java.lang.String objectName,
                        java.lang.String attName,
                        T value)
```
    Specified by:
    
    setObjectsValue in interface State

Class TabularBeliefState

Nested Class Summary

Nested classes/interfaces inherited from interface burlap.oomdp.singleagent.pomdp.beliefstate.EnumerableBeliefState

Field Summary

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Field Detail

BELIEFCLASSNAME

BELIEFATTNAME

stateEnumerator

beliefValues

domain

Constructor Detail

TabularBeliefState

TabularBeliefState

TabularBeliefState

Method Detail

getTabularBeliefMDPDomain

belief

belief

sampleStateFromBelief

getUpdatedBeliefState

getStatesAndBeliefsWithNonZeroProbability

numStates

stateForId

setBelief

setBelief

getTransitionProb

setBeliefVector

getBeliefVector

getStateSpace

zeroOutBeliefVector

initializeBeliefsUniformly

initializeAllBeliefValuesTo

copy

addObject

addAllObjects

removeObject

removeObject

removeAllObjects

renameObject

renameObject

getObjectMatchingTo

numTotalObjects

getObject

getAllObjects

getObjectsOfClass

getFirstObjectOfClass

getObjectClassesPresent

getAllObjectsByClass

getCompleteStateDescription

getAllUnsetAttributes

getCompleteStateDescriptionWithUnsetAttributesAsNull

getPossibleBindingsGivenParamOrderGroups

equals

setObjectsValue