<agrum/FMDP/planning/treeOperatorStrategy.h> More...

#include <treeOperatorStrategy.h>

Inheritance diagram for gum::TreeOperatorStrategy< GUM_SCALAR >:

Collaboration diagram for gum::TreeOperatorStrategy< GUM_SCALAR >:

Public Member Functions
Constructor & destructor.
	TreeOperatorStrategy ()
	Default constructor.
	~TreeOperatorStrategy ()
	Default destructor.

MultiDimFunctionGraph< GUM_SCALAR, ExactTerminalNodePolicy > *	getFunctionInstance ()
MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > *	getArgMaxFunctionInstance ()
MultiDimFunctionGraph< ActionSet, SetTerminalNodePolicy > *	getAggregatorInstance ()

Protected Member Functions
INLINE void	deleteFunctionGraph_ (const MultiDimFunctionGraph< GUM_SCALAR > f1, const MultiDimFunctionGraph< GUM_SCALAR > f2, Idx del)
INLINE void	deleteFunctionGraph_ (const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > f1, const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > f2, Idx del)
Graph Function Operations Methods
MultiDimFunctionGraph< GUM_SCALAR > *	regress (const MultiDimFunctionGraph< GUM_SCALAR > Vold, Idx actionId, const FMDP< GUM_SCALAR > fmdp, const gum::VariableSet &elVarSeq)
	Computes Qaction for given actionid.
virtual MultiDimFunctionGraph< GUM_SCALAR > *	maximize (const MultiDimFunctionGraph< GUM_SCALAR > f1, const MultiDimFunctionGraph< GUM_SCALAR > f2, Idx del=3)
virtual MultiDimFunctionGraph< GUM_SCALAR > *	minimize (const MultiDimFunctionGraph< GUM_SCALAR > f1, const MultiDimFunctionGraph< GUM_SCALAR > f2, Idx del=3)
virtual MultiDimFunctionGraph< GUM_SCALAR > *	multiply (const MultiDimFunctionGraph< GUM_SCALAR > f1, const MultiDimFunctionGraph< GUM_SCALAR > f2, Idx del=3)
virtual MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > *	argmaximize (const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > f1, const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > f2, Idx del=3)
	ArgMaximizes between f2 and f1.
virtual MultiDimFunctionGraph< GUM_SCALAR > *	add (const MultiDimFunctionGraph< GUM_SCALAR > f1, const MultiDimFunctionGraph< GUM_SCALAR > f2, Idx del=1)
	Adds reward to given function( whether a f2 or f1).
virtual MultiDimFunctionGraph< GUM_SCALAR > *	subtract (const MultiDimFunctionGraph< GUM_SCALAR > f1, const MultiDimFunctionGraph< GUM_SCALAR > f2, Idx del=0)
	Subtract current f1 from old f1 to see if threshold is reached or not.

Detailed Description

template<typename GUM_SCALAR>
class gum::TreeOperatorStrategy< GUM_SCALAR >

A class to find optimal policy for a given FMDP.

Perform a TreeOperatorStrategy planning on given in parameter factored markov decision process

Definition at line 70 of file treeOperatorStrategy.h.

Constructor & Destructor Documentation

◆ TreeOperatorStrategy()

template<typename GUM_SCALAR>

INLINE gum::TreeOperatorStrategy< GUM_SCALAR >::TreeOperatorStrategy ( )

Default constructor.

Definition at line 85 of file treeOperatorStrategy_tpl.h.

                                                                  {
    GUM_CONSTRUCTOR(TreeOperatorStrategy);
  }

References TreeOperatorStrategy().

Referenced by TreeOperatorStrategy(), and ~TreeOperatorStrategy().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ ~TreeOperatorStrategy()

template<typename GUM_SCALAR>

INLINE gum::TreeOperatorStrategy< GUM_SCALAR >::~TreeOperatorStrategy ( )

Default destructor.

Definition at line 93 of file treeOperatorStrategy_tpl.h.

                                                                   {
    GUM_DESTRUCTOR(TreeOperatorStrategy);
  }

References TreeOperatorStrategy().

Here is the call graph for this function:

Member Function Documentation

◆ add()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR > * gum::TreeOperatorStrategy< GUM_SCALAR >::add	(	const MultiDimFunctionGraph< GUM_SCALAR > *	f1,
		const MultiDimFunctionGraph< GUM_SCALAR > *	f2,
		Idx	del = 1 )

protectedvirtual

Adds reward to given function( whether a f2 or f1).

Warning: given function is deleted, returns the new one

Parameters

reward	: R(s) or R(s,a)
function	: either V(s) or Q(s,a)

Warning: given function is deleted, returns the new one

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 205 of file treeOperatorStrategy_tpl.h.

                                                                                              {
    TreeOperator< GUM_SCALAR, std::plus > opi(f1, f2);
    MultiDimFunctionGraph< GUM_SCALAR >*  ret = opi.compute();
    this->deleteFunctionGraph_(f1, f2, del);
    return ret;
  }

References gum::TreeOperator< GUM_SCALAR, COMBINEOPERATOR, TerminalNodePolicy >::compute(), and gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_().

Here is the call graph for this function:

◆ argmaximize()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > * gum::TreeOperatorStrategy< GUM_SCALAR >::argmaximize	(	const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > *	f1,
		const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > *	f2,
		Idx	del = 3 )

protectedvirtual

ArgMaximizes between f2 and f1.

Warning: given vFunction and qAction are deleted, returns the new one; given f1 and f2 are deleted, returns the new one

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 185 of file treeOperatorStrategy_tpl.h.

                                                                                                   {
    TreeOperator< ArgMaxSet< GUM_SCALAR, Idx >, ArgumentMaximisesAction, SetTerminalNodePolicy >
                                                                                  argmaxope(f1, f2);
    MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy >* ret
        = argmaxope.compute();
    this->deleteFunctionGraph_(f1, f2, del);
    return ret;
  }

References gum::TreeOperator< GUM_SCALAR, COMBINEOPERATOR, TerminalNodePolicy >::compute(), and gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_().

Here is the call graph for this function:

◆ deleteFunctionGraph_() [1/2]

template<typename GUM_SCALAR>

INLINE void gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_	(	const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > *	f1,
		const MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > *	f2,
		Idx	del )

inlineprotectedinherited

Definition at line 183 of file IOperatorStrategy.h.

                                                                                                 {
      if (del == 1 || del == 3) delete f1;
      if (del >= 2) delete f2;
    }

◆ deleteFunctionGraph_() [2/2]

template<typename GUM_SCALAR>

INLINE void gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_	(	const MultiDimFunctionGraph< GUM_SCALAR > *	f1,
		const MultiDimFunctionGraph< GUM_SCALAR > *	f2,
		Idx	del )

inlineprotectedinherited

Definition at line 176 of file IOperatorStrategy.h.

                                                                                     {
      if (del == 1 || del == 3) delete f1;
      if (del >= 2) delete f2;
    }

Here is the caller graph for this function:

◆ getAggregatorInstance()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< ActionSet, SetTerminalNodePolicy > * gum::TreeOperatorStrategy< GUM_SCALAR >::getAggregatorInstance ( )

inlinevirtual

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 106 of file treeOperatorStrategy.h.

                                                                                       {
      return MultiDimFunctionGraph< ActionSet, SetTerminalNodePolicy >::getTreeInstance();
    }

References gum::MultiDimFunctionGraph< GUM_SCALAR, TerminalNodePolicy >::getTreeInstance().

Here is the call graph for this function:

◆ getArgMaxFunctionInstance()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >, SetTerminalNodePolicy > * gum::TreeOperatorStrategy< GUM_SCALAR >::getArgMaxFunctionInstance ( )

inlinevirtual

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 101 of file treeOperatorStrategy.h.

                                    {
      return MultiDimFunctionGraph< ArgMaxSet< GUM_SCALAR, Idx >,
                                    SetTerminalNodePolicy >::getTreeInstance();
    }

◆ getFunctionInstance()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR, ExactTerminalNodePolicy > * gum::TreeOperatorStrategy< GUM_SCALAR >::getFunctionInstance ( )

inlinevirtual

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 96 of file treeOperatorStrategy.h.

                                                                                        {
      return MultiDimFunctionGraph< GUM_SCALAR >::getTreeInstance();
    }

References gum::MultiDimFunctionGraph< GUM_SCALAR, TerminalNodePolicy >::getTreeInstance().

Here is the call graph for this function:

◆ maximize()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR > * gum::TreeOperatorStrategy< GUM_SCALAR >::maximize	(	const MultiDimFunctionGraph< GUM_SCALAR > *	f1,
		const MultiDimFunctionGraph< GUM_SCALAR > *	f2,
		Idx	del = 3 )

protectedvirtual

Warning: given vFunction and qAction are deleted, returns the new one; given f1 and f2 are deleted, returns the new one

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 143 of file treeOperatorStrategy_tpl.h.

                                                                                                   {
    TreeOperator< GUM_SCALAR, Maximizes > opi(f1, f2);
    MultiDimFunctionGraph< GUM_SCALAR >*  ret = opi.compute();
    this->deleteFunctionGraph_(f1, f2, del);
    return ret;
  }

References gum::TreeOperator< GUM_SCALAR, COMBINEOPERATOR, TerminalNodePolicy >::compute(), and gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_().

Here is the call graph for this function:

◆ minimize()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR > * gum::TreeOperatorStrategy< GUM_SCALAR >::minimize	(	const MultiDimFunctionGraph< GUM_SCALAR > *	f1,
		const MultiDimFunctionGraph< GUM_SCALAR > *	f2,
		Idx	del = 3 )

protectedvirtual

Warning: Minimze the two given functions; given f1 and f2 are deleted, returns the new one

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 157 of file treeOperatorStrategy_tpl.h.

                                                                                                   {
    TreeOperator< GUM_SCALAR, Minimizes > opi(f1, f2);
    MultiDimFunctionGraph< GUM_SCALAR >*  ret = opi.compute();
    this->deleteFunctionGraph_(f1, f2, del);
    return ret;
  }

References gum::TreeOperator< GUM_SCALAR, COMBINEOPERATOR, TerminalNodePolicy >::compute(), and gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_().

Here is the call graph for this function:

◆ multiply()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR > * gum::TreeOperatorStrategy< GUM_SCALAR >::multiply	(	const MultiDimFunctionGraph< GUM_SCALAR > *	f1,
		const MultiDimFunctionGraph< GUM_SCALAR > *	f2,
		Idx	del = 3 )

protectedvirtual

Warning: given f1 and f2 are deleted, returns the new one

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 170 of file treeOperatorStrategy_tpl.h.

                                                                                                   {
    TreeOperator< GUM_SCALAR, std::multiplies > opi(f1, f2);
    MultiDimFunctionGraph< GUM_SCALAR >*        ret = opi.compute();
    this->deleteFunctionGraph_(f1, f2, del);
    return ret;
  }

References gum::TreeOperator< GUM_SCALAR, COMBINEOPERATOR, TerminalNodePolicy >::compute(), and gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_().

Here is the call graph for this function:

◆ regress()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR > * gum::TreeOperatorStrategy< GUM_SCALAR >::regress	(	const MultiDimFunctionGraph< GUM_SCALAR > *	Vold,
		Idx	actionId,
		const FMDP< GUM_SCALAR > *	fmdp,
		const gum::VariableSet &	elVarSeq )

protectedvirtual

Computes Qaction for given actionid.

Performs a multiplication/projection on given qAction.

Returns: qAction : the computed Q(s,a)

Parameters

qAction	: the computed Q(s,a)
pxip	: the transition probabilities for the xip variable
xip	: the variable we eliminate on the projection

Warning: given qAction is deleted, return the new one

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 115 of file treeOperatorStrategy_tpl.h.

                                                                                    {
    // ******************************************************************************
    // Initialisation :
    // Creating a copy of last Vfunction to deduce from the new Qaction
    // And finding the first var to eleminate (the one at the end)
    Bijection< const DiscreteVariable*, const MultiDimFunctionGraph< GUM_SCALAR >* > pxi;
    for (SequenceIteratorSafe< const DiscreteVariable* > varIter
         = Vold->variablesSequence().beginSafe();
         varIter != Vold->variablesSequence().endSafe();
         ++varIter) {
      pxi.insert(*varIter,
                 RECAST(fmdp->transition(actionId, fmdp->mapMainPrime().first(*varIter))));
    }
 
    TreeRegress< GUM_SCALAR, std::multiplies, std::plus > tr(
        Vold,
        pxi);   //, Vold->variablesSequence(), (GUM_SCALAR) 0.0 );
    return tr.compute();
  }

References gum::TreeRegress< GUM_SCALAR, COMBINEOPERATOR, PROJECTOPERATOR, TerminalNodePolicy >::compute(), gum::BijectionImplementation< T1, T2, std::is_scalar< T1 >::value &&std::is_scalar< T2 >::value >::first(), gum::BijectionImplementation< T1, T2, Gen >::insert(), gum::FMDP< GUM_SCALAR >::mapMainPrime(), RECAST, gum::FMDP< GUM_SCALAR >::transition(), and gum::MultiDimImplementation< GUM_SCALAR >::variablesSequence().

Here is the call graph for this function:

◆ subtract()

template<typename GUM_SCALAR>

MultiDimFunctionGraph< GUM_SCALAR > * gum::TreeOperatorStrategy< GUM_SCALAR >::subtract	(	const MultiDimFunctionGraph< GUM_SCALAR > *	f1,
		const MultiDimFunctionGraph< GUM_SCALAR > *	f2,
		Idx	del = 0 )

protectedvirtual

Subtract current f1 from old f1 to see if threshold is reached or not.

Warning: this time, nothing is deleted

Parameters

old	and new VFuntion

Warning: this time, nothing is deleted

Implements gum::IOperatorStrategy< GUM_SCALAR >.

Definition at line 222 of file treeOperatorStrategy_tpl.h.

                                                                                                   {
    TreeOperator< GUM_SCALAR, std::minus > opi(f1, f2);
    MultiDimFunctionGraph< GUM_SCALAR >*   ret = opi.compute();
    this->deleteFunctionGraph_(f1, f2, del);
    return ret;
  }

References gum::TreeOperator< GUM_SCALAR, COMBINEOPERATOR, TerminalNodePolicy >::compute(), and gum::IOperatorStrategy< GUM_SCALAR >::deleteFunctionGraph_().

Here is the call graph for this function:

The documentation for this class was generated from the following files:

agrum/FMDP/planning/treeOperatorStrategy.h
agrum/FMDP/planning/treeOperatorStrategy_tpl.h

Public Member Functions

Protected Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ TreeOperatorStrategy()

◆ ~TreeOperatorStrategy()

Member Function Documentation

◆ add()

◆ argmaximize()

◆ deleteFunctionGraph_() [1/2]

◆ deleteFunctionGraph_() [2/2]

◆ getAggregatorInstance()

◆ getArgMaxFunctionInstance()

◆ getFunctionInstance()

◆ maximize()

◆ minimize()

◆ multiply()

◆ regress()

◆ subtract()