K-adaptability approximator. More...

#include <Kadaptability.hpp>

Inheritance diagram for Kadaptability:

Collaboration diagram for Kadaptability:

Public Member Functions
	Kadaptability (const map< uint, uint > &numPartitionsMap=map< uint, uint >(), double bigM=100.0, double epsilon=0.0001, string folder=" ")
	Constructor of K-Adaptability approximator. More...

	~Kadaptability ()
	Destructor of K-Adaptability approximator. More...

void	checkWsCompatability (const map< uint, uint > &wsMap)
	Check the compatability of the given model. More...

void	createVariableAndUncMap (dvContainer::const_iterator varsBegin, dvContainer::const_iterator varsEnd, uncContainer::const_iterator uncBegin, uncContainer::const_iterator uncEnd)
	Create the map from the original decision variable and uncertainty to the new variable and uncertainty for each plan and pair of partition and time stage respectively. More...

ROCPPOptModelIF_Ptr	approximate (ROCPPOptModelIF_Ptr pIn)
	Check the compatability of the input model and check whether the model has constraint uncertainty to choose the appropriate method. More...

ROCPPOptModelIF_Ptr	fixSecondStageVariablesToWarmStart (ROCPPOptModelIF_Ptr pIn, ROCPPOptModelIF_Ptr pKadaptModel, const map< string, double > &warmStartResults, const map< uint, uint > &wsMap)
	Return a new model after fixing second stage variables which are solved by the warm start problem to the value provided in the input map. More...

ROCPPOptModelIF_Ptr	fixBinaryVariableValues (ROCPPOptModelIF_Ptr pKadaptModel, const map< string, bool > &varValues) const
	Return a new model after fixing second stage variables to the value provided in the input map. More...

virtual ROCPPOptModelIF_Ptr	addProblemSpecificConstraints (ROCPPOptModelIF_Ptr pKadapt) const

string	getSolutionApproachParameters (string delimiter) const

ROCPPVarIF_Ptr	getDVonPartition (string partition, string OrigDVname) const
	Get the new variable associated with the given original variable on the given contingency plan. More...

ROCPPUnc_Ptr	getUncOnPartition (string OrigUncName, string partition, uint t) const
	Get the new uncertain parameter associated with the given original uncertainty on the given contingency plan at time stage t. More...

void	getWsSolutions (ROCPPOptModelIF_Ptr pIn, const map< string, double > &warmStartResults, const map< uint, uint > &wsMap, map< string, double > &wsSolutions)
	Return the warm start solution of this problem by setting the variables which exist in the warm start problem to their solutions and assigning the variables which don't exist in the warm start problem to the biggest solved kth solutions. More...

void	printParametersToScreen () const

void	printOut (const ROCPPOptModelIF_Ptr pIn, const map< string, double > &resultIn, ROCPPVarIF_Ptr dv, string partition)
	Print out the solution for the given variable for the given plan. More...

void	printOut (const ROCPPOptModelIF_Ptr pIn, const map< string, double > &resultIn, ROCPPVarIF_Ptr dv)
	Print out the solution for the given variable for all plans. More...

void	printOut (const ROCPPOptModelIF_Ptr pIn, const map< string, double > &resultIn, ROCPPUnc_Ptr unc, string partition)
	Print out whether observe or not the given uncertainty for the given plan. More...

void	printOut (const ROCPPOptModelIF_Ptr pIn, const map< string, double > &resultIn, ROCPPUnc_Ptr unc)
	Print out whether we observe or not the given uncertainty for all cases. More...

ROCPPOptModelIF_Ptr	Reformulate (ROCPPOptModelIF_Ptr pIn)

bool	isApplicable (ROCPPOptModelIF_Ptr pIn) const

string	getName () const

Public Member Functions inherited from DecisionRuleIF
	DecisionRuleIF (uint memory=1000)
	Constructor of the DecisionRuleIF class. More...

virtual	~DecisionRuleIF ()
	Destructor of the DecisionRuleIF class. More...

uint	getMemory () const

Public Member Functions inherited from ReformulationStrategyIF
	ReformulationStrategyIF ()

	~ReformulationStrategyIF ()

Protected Member Functions
void	initialize (ROCPPOptModelIF_Ptr pIn)
	Initialize the class ith the given parameters. More...

ROCPPOptModelIF_Ptr	approxObjUnc (ROCPPOptModelIF_Ptr pIn)
	Method for multistage model with only objective uncertainty. More...

ROCPPOptModelIF_Ptr	approxCstrUnc (ROCPPOptModelIF_Ptr pIn)
	Method for two stage model with both objective and constraint uncertainty. More...

Protected Attributes
ROCPPKadaptEncoder_Ptr	m_pPartitionEncoder

map< string, map< string, ROCPPVarIF_Ptr > >	m_DVmap
	Map from original decision variable name to partition string to new decision variable. More...

map< string, map< string, ROCPPVarIF_Ptr > >	m_mapPartitionEnc_mapOrigDVtoDVonPartition
	Map from partition to map from original variable name to new variable on that partition. More...

map< pair< string, uint >, map< string, ROCPPUnc_Ptr > >	m_mapPartitionEncandt_mapOrigUnctoUnconPartition
	Map from pair of partition and time stage to the map from uncertainty name to the uncertainty on that partition at specific time stage. More...

ROCPPBilinearReform_Ptr	m_pBTR

double	m_epsilon
	Small values used to deal with the infeasible constraint in constraint uncertainty. More...

string	m_folder

map< uint, uint >	m_numPartitionsMap
	Map from time period to value of K for that time-period. More...

Protected Attributes inherited from DecisionRuleIF
uint	m_memory
	Memory of this approximator. More...

Detailed Description

K-adaptability approximator.

Constructor & Destructor Documentation

◆ Kadaptability()

Kadaptability::Kadaptability	(	const map< uint, uint > &	numPartitionsMap = `map<uint, uint>()`,
		double	bigM = `100.0`,
		double	epsilon = `0.0001`,
		string	folder = `" "`
	)

Constructor of K-Adaptability approximator.

Parameters

numPartitionsMap	Map from time period to value of K for that time-period
bigM	Big-M constant value for linearizing the bilinear terms
epsilon	Value of epsilon to build the strictly inequality in the constraint uncertainty case
folder	Name of the folders where to save the log file

◆ ~Kadaptability()

Kadaptability::~Kadaptability ( )

inline

Destructor of K-Adaptability approximator.

Member Function Documentation

◆ addProblemSpecificConstraints()

virtual ROCPPOptModelIF_Ptr Kadaptability::addProblemSpecificConstraints ( ROCPPOptModelIF_Ptr pKadapt ) const

inlinevirtual

◆ approxCstrUnc()

ROCPPOptModelIF_Ptr Kadaptability::approxCstrUnc ( ROCPPOptModelIF_Ptr pIn )

protected

Method for two stage model with both objective and constraint uncertainty.

Warning: Only two stage model is allowed here

◆ approximate()

ROCPPOptModelIF_Ptr Kadaptability::approximate ( ROCPPOptModelIF_Ptr pIn )

virtual

Check the compatability of the input model and check whether the model has constraint uncertainty to choose the appropriate method.

Implements DecisionRuleIF.

◆ approxObjUnc()

ROCPPOptModelIF_Ptr Kadaptability::approxObjUnc ( ROCPPOptModelIF_Ptr pIn )

protected

Method for multistage model with only objective uncertainty.

Warning: No uncertainty in constraint is allowed here

◆ checkWsCompatability()

void Kadaptability::checkWsCompatability ( const map< uint, uint > & wsMap )

Check the compatability of the given model.

Warning: Only binary variables are allowed in the K-adaptability approximator Check the compatability of the given warm start map

Note: the value of K in each time stage must be less than or equal to that in the original approximator; at least one K must be strictly less than the one in the original problem

◆ createVariableAndUncMap()

void Kadaptability::createVariableAndUncMap	(	dvContainer::const_iterator	varsBegin,
		dvContainer::const_iterator	varsEnd,
		uncContainer::const_iterator	uncBegin,
		uncContainer::const_iterator	uncEnd
	)

Create the map from the original decision variable and uncertainty to the new variable and uncertainty for each plan and pair of partition and time stage respectively.

◆ fixBinaryVariableValues()

ROCPPOptModelIF_Ptr Kadaptability::fixBinaryVariableValues	(	ROCPPOptModelIF_Ptr	pKadaptModel,
		const map< string, bool > &	varValues
	)		const

Return a new model after fixing second stage variables to the value provided in the input map.

◆ fixSecondStageVariablesToWarmStart()

ROCPPOptModelIF_Ptr Kadaptability::fixSecondStageVariablesToWarmStart	(	ROCPPOptModelIF_Ptr	pIn,
		ROCPPOptModelIF_Ptr	pKadaptModel,
		const map< string, double > &	warmStartResults,
		const map< uint, uint > &	wsMap
	)

Return a new model after fixing second stage variables which are solved by the warm start problem to the value provided in the input map.

Parameters

pIn	Model before approximating
pKadaptModel	Model after approximating
warmStartResults	Results of the warm start problem
wsMap	Map from time stage to number of k in the warm start problem

◆ getDVonPartition()

ROCPPVarIF_Ptr Kadaptability::getDVonPartition	(	string	partition,
		string	OrigDVname
	)		const

Get the new variable associated with the given original variable on the given contingency plan.

◆ getName()

string Kadaptability::getName ( ) const

inlinevirtual

Implements ReformulationStrategyIF.

◆ getSolutionApproachParameters()

string Kadaptability::getSolutionApproachParameters ( string delimiter ) const

◆ getUncOnPartition()

ROCPPUnc_Ptr Kadaptability::getUncOnPartition	(	string	OrigUncName,
		string	partition,
		uint	t
	)		const

Get the new uncertain parameter associated with the given original uncertainty on the given contingency plan at time stage t.

◆ getWsSolutions()

void Kadaptability::getWsSolutions	(	ROCPPOptModelIF_Ptr	pIn,
		const map< string, double > &	warmStartResults,
		const map< uint, uint > &	wsMap,
		map< string, double > &	wsSolutions
	)

Return the warm start solution of this problem by setting the variables which exist in the warm start problem to their solutions and assigning the variables which don't exist in the warm start problem to the biggest solved kth solutions.

Parameters

pIn	Model before approximating
warmStartResults	Results of the warm start problem
wsMap	Map from time stage to number of k in the warm start problem
wsSolutions	Map from variable name to its warm start solution for all variables in the problem to be solved

◆ initialize()

void Kadaptability::initialize ( ROCPPOptModelIF_Ptr pIn )

protected

Initialize the class ith the given parameters.

◆ isApplicable()

bool Kadaptability::isApplicable ( ROCPPOptModelIF_Ptr pIn ) const

virtual

Implements ReformulationStrategyIF.

◆ printOut() [1/4]

void Kadaptability::printOut	(	const ROCPPOptModelIF_Ptr	pIn,
		const map< string, double > &	resultIn,
		ROCPPUnc_Ptr	unc
	)

Print out whether we observe or not the given uncertainty for all cases.

◆ printOut() [2/4]

void Kadaptability::printOut	(	const ROCPPOptModelIF_Ptr	pIn,
		const map< string, double > &	resultIn,
		ROCPPUnc_Ptr	unc,
		string	partition
	)

Print out whether observe or not the given uncertainty for the given plan.

◆ printOut() [3/4]

void Kadaptability::printOut	(	const ROCPPOptModelIF_Ptr	pIn,
		const map< string, double > &	resultIn,
		ROCPPVarIF_Ptr	dv
	)

Print out the solution for the given variable for all plans.

◆ printOut() [4/4]

void Kadaptability::printOut	(	const ROCPPOptModelIF_Ptr	pIn,
		const map< string, double > &	resultIn,
		ROCPPVarIF_Ptr	dv,
		string	partition
	)

Print out the solution for the given variable for the given plan.

◆ printParametersToScreen()

void Kadaptability::printParametersToScreen ( ) const

◆ Reformulate()

ROCPPOptModelIF_Ptr Kadaptability::Reformulate ( ROCPPOptModelIF_Ptr pIn )

inlinevirtual

Implements ReformulationStrategyIF.

Member Data Documentation

◆ m_DVmap

map<string, map<string, ROCPPVarIF_Ptr> > Kadaptability::m_DVmap

protected

Map from original decision variable name to partition string to new decision variable.

◆ m_epsilon

double Kadaptability::m_epsilon

protected

Small values used to deal with the infeasible constraint in constraint uncertainty.

◆ m_folder

string Kadaptability::m_folder

protected

◆ m_mapPartitionEnc_mapOrigDVtoDVonPartition

map< string, map<string, ROCPPVarIF_Ptr> > Kadaptability::m_mapPartitionEnc_mapOrigDVtoDVonPartition

protected

Map from partition to map from original variable name to new variable on that partition.

◆ m_mapPartitionEncandt_mapOrigUnctoUnconPartition

map< pair<string,uint> , map<string, ROCPPUnc_Ptr> > Kadaptability::m_mapPartitionEncandt_mapOrigUnctoUnconPartition

protected

Map from pair of partition and time stage to the map from uncertainty name to the uncertainty on that partition at specific time stage.

◆ m_numPartitionsMap

map<uint,uint> Kadaptability::m_numPartitionsMap

protected

Map from time period to value of K for that time-period.

◆ m_pBTR

ROCPPBilinearReform_Ptr Kadaptability::m_pBTR

protected

◆ m_pPartitionEncoder

ROCPPKadaptEncoder_Ptr Kadaptability::m_pPartitionEncoder

protected

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Constructor & Destructor Documentation

◆ Kadaptability()

◆ ~Kadaptability()

Member Function Documentation

◆ addProblemSpecificConstraints()

◆ approxCstrUnc()

◆ approximate()

◆ approxObjUnc()

◆ checkWsCompatability()

◆ createVariableAndUncMap()

◆ fixBinaryVariableValues()

◆ fixSecondStageVariablesToWarmStart()

◆ getDVonPartition()

◆ getName()

◆ getSolutionApproachParameters()

◆ getUncOnPartition()

◆ getWsSolutions()

◆ initialize()

◆ isApplicable()

◆ printOut() [1/4]

◆ printOut() [2/4]

◆ printOut() [3/4]

◆ printOut() [4/4]

◆ printParametersToScreen()

◆ Reformulate()

Member Data Documentation

◆ m_DVmap

◆ m_epsilon

◆ m_folder

◆ m_mapPartitionEnc_mapOrigDVtoDVonPartition

◆ m_mapPartitionEncandt_mapOrigUnctoUnconPartition

◆ m_numPartitionsMap

◆ m_pBTR

◆ m_pPartitionEncoder