OSInstance Class Reference

The in-memory representation of an OSiL instance.. More...

#include <OSInstance.h>

Collaboration diagram for OSInstance:

Public Member Functions
	OSInstance ()
	The OSInstance class constructor.
	~OSInstance ()
	The OSInstance class destructor.
std::string	getInstanceName ()
	Get instance name.
std::string	getInstanceSource ()
	Get instance source.
std::string	getInstanceDescription ()
	Get instance description.
int	getVariableNumber ()
	Get variable number.
std::string *	getVariableNames ()
	Get variable names.
double *	getVariableInitialValues ()
	Get variable initial values.
std::string *	getVariableInitialStringValues ()
	Get variable initial std::string values.
char *	getVariableTypes ()
	Get variable types.
int	getNumberOfIntegerVariables ()
	getNumberOfIntegerVariables
int	getNumberOfBinaryVariables ()
	getNumberOfBinaryVariables
double *	getVariableLowerBounds ()
	Get variable lower bounds.
double *	getVariableUpperBounds ()
	Get variable upper bounds.
int	getObjectiveNumber ()
	Get objective number.
std::string *	getObjectiveNames ()
	Get objective names.
std::string *	getObjectiveMaxOrMins ()
	Get objective maxOrMins.
int *	getObjectiveCoefficientNumbers ()
	Get objective coefficient number.
double *	getObjectiveConstants ()
	Get objective constants.
double *	getObjectiveWeights ()
	Get objective weights.
SparseVector **	getObjectiveCoefficients ()
	Get objective coefficients.
double **	getDenseObjectiveCoefficients ()
	getDenseObjectiveCoefficients.
int	getConstraintNumber ()
	Get constraint number.
std::string *	getConstraintNames ()
	Get constraint names.
double *	getConstraintLowerBounds ()
	Get constraint lower bounds.
double *	getConstraintUpperBounds ()
	Get constraint upper bounds.
char *	getConstraintTypes ()
	Get constraint types.
int	getLinearConstraintCoefficientNumber ()
	Get number of specified (usually nonzero) linear constraint coefficient values.
bool	getLinearConstraintCoefficientMajor ()
	Get whether the constraint coefficients is in column major (true) or row major (false).
SparseMatrix *	getLinearConstraintCoefficientsInColumnMajor ()
	Get linear constraint coefficients in column major.
SparseMatrix *	getLinearConstraintCoefficientsInRowMajor ()
	Get linear constraint coefficients in row major.
int	getNumberOfQuadraticTerms ()
	Get the number of specified (usually nonzero) qTerms in the quadratic coefficients.
QuadraticTerms *	getQuadraticTerms ()
	Get all the quadratic terms in the instance.
int *	getQuadraticRowIndexes ()
	Get the indexes of rows which have a quadratic term.
int	getNumberOfQuadraticRowIndexes ()
	Get the number of rows which have a quadratic term.
int	getNumberOfNonlinearExpressions ()
	Get number of nonlinear expressions.
OSExpressionTree *	getNonlinearExpressionTree (int rowIdx)
	Get the expression tree for a given row index.
OSExpressionTree *	getNonlinearExpressionTreeMod (int rowIdx)
	Get the expression tree for a given row index for the modified expression trees (quadratic terms added).
std::vector< OSnLNode * >	getNonlinearExpressionTreeInPostfix (int rowIdx)
	Get the postfix tokens for a given row index.
std::vector< OSnLNode * >	getNonlinearExpressionTreeModInPostfix (int rowIdx)
	Get the postfix tokens for a given row index for the modified Expression Tree (quadratic terms added).
std::vector< OSnLNode * >	getNonlinearExpressionTreeInPrefix (int rowIdx)
	Get the prefix tokens for a given row index.
std::vector< OSnLNode * >	getNonlinearExpressionTreeModInPrefix (int rowIdx)
	Get the prefix tokens for a given row index for the modified Expression Tree (quadratic terms added).
int	getNumberOfNonlinearObjectives ()
int	getNumberOfNonlinearConstraints ()
std::map< int, OSExpressionTree * >	getAllNonlinearExpressionTrees ()
std::map< int, OSExpressionTree * >	getAllNonlinearExpressionTreesMod ()
int *	getNonlinearExpressionTreeIndexes ()
	Get all the nonlinear expression tree indexes, i.e.
int	getNumberOfNonlinearExpressionTreeIndexes ()
	Get the number of unique Nonlinear exrpession tree indexes.
int *	getNonlinearExpressionTreeModIndexes ()
	Get all the nonlinear expression tree indexes, i.e.
int	getNumberOfNonlinearExpressionTreeModIndexes ()
	Get the number of unique Nonlinear exrpession tree indexes after modifying the expression tree to contain quadratic terms.
bool	setInstanceSource (std::string source)
	set the instance source.
bool	setInstanceDescription (std::string description)
	set the instance description.
bool	setInstanceName (std::string name)
	set the instance name.
bool	setVariableNumber (int number)
	set the variable number.
bool	addVariable (int index, std::string name, double lowerBound, double upperBound, char type, double init, std::string initString)
	add a variable.
bool	setVariables (int number, std::string names, double lowerBounds, double upperBounds, char types, double inits, std::string initsString)
	set all the variable related elements.
bool	setObjectiveNumber (int number)
	set the objective number.
bool	addObjective (int index, std::string name, std::string maxOrMin, double constant, double weight, SparseVector *objectiveCoefficients)
	add an objective.
bool	setObjectives (int number, std::string names, std::string maxOrMins, double constants, double weights, SparseVector **objectitiveCoefficients)
	set all the objectives related elements.
bool	setConstraintNumber (int number)
	set the constraint number.
bool	addConstraint (int index, std::string name, double lowerBound, double upperBound, double constant)
	add a constraint.
bool	setConstraints (int number, std::string names, double lowerBounds, double upperBounds, double constants)
	set all the constraint related elements.
bool	setLinearConstraintCoefficients (int numberOfValues, bool isColumnMajor, double values, int valuesBegin, int valuesEnd, int indexes, int indexesBegin, int indexesEnd, int *starts, int startsBegin, int startsEnd)
	set linear constraint coefficients
bool	setQuadraticTerms (int number, int rowIndexes, int varOneIndexes, int varTwoIndexes, double coefficients, int begin, int end)
	set quadratic terms
bool	setQuadraticTermsInNonlinearExpressions (int number, int rowIndexes, int varOneIndexes, int varTwoIndexes, double coefficients)
	set quadratic terms in nonlinearExpressions
bool	initializeNonLinearStructures ()
	Initialize the data structures for the nonlinear API.
double	calculateFunctionValue (int idx, double *x, bool new_x)
	Calculate the function value for function (constraint or objective) indexed by idx.
double *	calculateAllConstraintFunctionValues (double x, double objLambda, double *conLambda, bool new_x, int highestOrder)
	Calculate all of the constraint function values.
double *	calculateAllConstraintFunctionValues (double *x, bool new_x)
	Calculate all of the constraint function values, we are overloading this function and this version of the method will not use any AD and will evaluate function values from the OS Expression Tree.
double *	calculateAllObjectiveFunctionValues (double x, double objLambda, double *conLambda, bool new_x, int highestOrder)
	Calculate all of the objective function values.
double *	calculateAllObjectiveFunctionValues (double *x, bool new_x)
	Calculate all of the objective function values, we are overloading this function and this version of the method will not use any AD and will evaluate function values from the OS Expression Tree.
SparseJacobianMatrix *	calculateAllConstraintFunctionGradients (double x, double objLambda, double *conLambda, bool new_x, int highestOrder)
	Calculate the gradient of all constraint functions.
SparseVector *	calculateConstraintFunctionGradient (double x, double objLambda, double *conLambda, int idx, bool new_x, int highestOrder)
	Calculate the gradient of the constraint function indexed by idx.
SparseVector *	calculateConstraintFunctionGradient (double *x, int idx, bool new_x)
	Calculate the gradient of the constraint function indexed by idx this function is overloaded.
double **	calculateAllObjectiveFunctionGradients (double x, double objLambda, double *conLambda, bool new_x, int highestOrder)
	Calculate the gradient of all objective functions.
double *	calculateObjectiveFunctionGradient (double x, double objLambda, double *conLambda, int objIdx, bool new_x, int highestOrder)
	Calculate the gradient of the objective function indexed by objIdx.
double *	calculateObjectiveFunctionGradient (double *x, int objIdx, bool new_x)
	Calculate the gradient of the objective function indexed by objIdx this function is overloaded.
SparseHessianMatrix *	calculateLagrangianHessian (double x, double objLambda, double *conLambda, bool new_x, int highestOrder)
	Calculate the Hessian of the Lagrangian Expression Tree This method will build the CppAD expression tree for only the first iteration Use this method on if the value of x does not affect the operations sequence.
SparseHessianMatrix *	calculateHessian (double *x, int idx, bool new_x)
	Calculate the Hessian of a constraint or objective function.
bool	getSparseJacobianFromColumnMajor ()
bool	getSparseJacobianFromRowMajor ()
OSExpressionTree *	getLagrangianExpTree ()
std::map< int, int >	getAllNonlinearVariablesIndexMap ()
SparseHessianMatrix *	getLagrangianHessianSparsityPattern ()
bool	addQTermsToExressionTree ()
SparseJacobianMatrix *	getJacobianSparsityPattern ()
void	duplicateExpressionTreesMap ()
	duplicate the map of expression trees.
bool	createCppADFun (std::vector< double > vdX)
	Create the a CppAD Function object: this is a function where the domain is the set of variables for the problem and the range is the objective function plus constraints.
std::vector< double >	forwardAD (int p, std::vector< double > vdX)
	Perform an AD forward sweep.
std::vector< double >	reverseAD (int p, std::vector< double > vdlambda)
	Perform an AD reverse sweep.
bool	getIterateResults (double x, double objLambda, double *conLambda, bool new_x, int highestOrder)
	end revised AD code
bool	getZeroOrderResults (double x, double objLambda, double *conLambda)
	Calculate function values.
bool	getFirstOrderResults (double x, double objLambda, double *conLambda)
	Calculate first derivatives.
bool	getSecondOrderResults (double x, double objLambda, double *conLambda)
	Calculate second derivatives.
bool	initForAlgDiff ()
	This should be called by nonlinear solvers using callback functions.
bool	initObjGradients ()
	This should be called by initForAlgDiff().
Public Attributes
InstanceHeader *	instanceHeader
	A pointer to an InstanceHeader object.
InstanceData *	instanceData
	A pointer to an InstanceData object.
CppAD::ADFun< double > *	Fad
	F is a CppAD function the range space is the objective + constraints functions, x is the domeain space.
bool	bUseExpTreeForFunEval
	bUseExpTreeForFunEval is set to true if you wish to use the OS Expression Tree for function evaluations instead of AD -- false by default.
Private Member Functions
bool	processVariables ()
	process variables.
bool	processObjectives ()
	process objectives.
bool	processConstraints ()
	process constraints.
bool	processLinearConstraintCoefficients ()
	process linear constraint coefficients.
Private Attributes
std::string	m_sInstanceName
	m_sInstanceName holds the instance name.
std::string	m_sInstanceSource
	m_sInstanceSource holds the instance source.
std::string	m_sInstanceDescription
	m_sInstanceDescription holds the instance description.
bool	m_bProcessVariables
	m_bProcessVariables holds whether the variables are processed.
int	m_iVariableNumber
	m_iVariableNumber holds the variable number.
int	m_iNumberOfIntegerVariables
	m_iNumberOfIntegerVariables holds the number of integer variables.
int	m_iNumberOfBinaryVariables
	m_iNumberOfBinaryVariables holds the number of binary variables.
int	m_iNumberOfQuadraticRowIndexes
	m_iNumberOfQuadraticRowIndexes holds the number of distinct rows and objectives with quadratic terms.
bool	m_bQuadraticRowIndexesProcessed
	m_bQuadraticRowIndexesProcessed is true if getQuadraticRowIndexes() has been called.
int *	m_miQuadRowIndexes
	m_miQuadRowIndexes is an integer pointer to the distinct rows indexes with a quadratic term.
int	m_iNumberOfNonlinearExpressionTreeIndexes
	m_iNumberOfNonlinearExpressionTreeIndexes holds the number of distinct rows and objectives with nonlinear terms.
bool	m_bNonlinearExpressionTreeIndexesProcessed
	m_bNonlinearExpressionTreeIndexesProcessed is true if getNonlinearExpressionTreeIndexes has been called.
int *	m_miNonlinearExpressionTreeIndexes
	m_miNonlinearExpressionTreeIndexes is an integer pointer to the distinct rows indexes in the nonlinear expression tree map.
int	m_iNumberOfNonlinearExpressionTreeModIndexes
	m_iNumberOfNonlinearExpressionTreeModIndexes holds the number of distinct rows and objectives with nonlinear terms including quadratic terms added to the nonlinear expression trees.
bool	m_bNonlinearExpressionTreeModIndexesProcessed
	m_bNonlinearExpressionTreeModIndexesProcessed is true if getNonlinearExpressionTreeModIndexes has been called.
int *	m_miNonlinearExpressionTreeModIndexes
	m_miNonlinearExpressionTreeModIndexes is an integer pointer to the distinct rows indexes in the modified expression tree map.
std::string *	m_msVariableNames
	m_msVariableNames holds an array of variable names.
double *	m_mdVariableInitialValues
	m_mdVariableInitialValues holds a double array of the initial variable values.
std::string *	m_msVariableInitialStringValues
	m_msVariableInitialStringValues holds a std::string array of the initial variable values.
char *	m_mcVariableTypes
	m_mcVariableTypes holds a char array of variable types (default = 'C').
double *	m_mdVariableLowerBounds
	m_mdVariableLowerBounds holds a double array of variable lower bounds (default = 0.0).
double *	m_mdVariableUpperBounds
	m_mdVariableUpperBounds holds a double array of variable upper bounds (default = INF).
bool	m_bProcessObjectives
	m_bProcessObjectives holds whether the objectives are processed.
int	m_iObjectiveNumber
	m_iObjectiveNumber is the number of objective functions.
int	m_iObjectiveNumberNonlinear
	m_iObjectiveNumber is the number of objective functions with a nonlinear term.
std::string *	m_msObjectiveNames
	m_msObjectiveNames holds an array of objective names.
std::string *	m_msMaxOrMins
	m_msMaxOrMins holds a std::string array of objective maxOrMins ("max" or "min").
int *	m_miNumberOfObjCoef
	m_miNumberOfObjCoef holds an integer array of number of objective coefficients (default = 0.0).
double *	m_mdObjectiveConstants
	m_mdObjectiveConstants holds an array of objective constants (default = 0.0).
double *	m_mdObjectiveWeights
	m_mdObjectiveWeights holds an array of objective weights (default = 1.0).
SparseVector **	m_mObjectiveCoefficients
	m_mObjectiveCoefficients holds an array of objective coefficients, one set of objective coefficients for each objective.
bool	m_bGetDenseObjectives
	m_bGetDenseObjectives holds whether the dense objective functions are processed.
double **	m_mmdDenseObjectiveCoefficients
	m_mmdDenseObjectiveCoefficients holds an array of pointers, each pointer points to a vector of dense objective function coefficients
bool	m_bProcessConstraints
	m_bProcessConstraints holds whether the constraints are processed.
int	m_iConstraintNumber
	m_iConstraintNumber is the number of constraints.
int	m_iConstraintNumberNonlinear
	m_iConstraintNumberNonlinear is the number of constraints that have a nonlinear term.
std::string *	m_msConstraintNames
	m_msConstraintNames holds an array of constraint names.
double *	m_mdConstraintLowerBounds
	m_mdConstraintLowerBounds holds an array of constraint lower bounds (default = -INF).
double *	m_mdConstraintUpperBounds
	m_mdConstraintUpperBounds holds an array of constraint upper bounds (default = INF).
double *	m_mdConstraintConstants
	m_mdConstraintConstants holds an array of constraint constants (default = 0.0).
char *	m_mcConstraintTypes
	m_mcConstraintTypes holds a char array of constraint types (R for range; L for <=; G for >=; E for =; U for unconstrained)
bool	m_bProcessLinearConstraintCoefficients
	m_bProcessLinearConstraintCoefficients holds whether the linear constraint coefficients are processed.
int	m_iLinearConstraintCoefficientNumber
	m_iLinearConstraintCoefficientNumber holds the number of specified (usually nonzero) linear constraint coefficient values.
bool	m_bColumnMajor
	m_bColumnMajor holds whether the linear constraint coefficients are stored in column major.
bool	m_binitForAlgDiff
	m_binitForAlgDiff is true if initForAlgDiff() has been called.
SparseMatrix *	m_linearConstraintCoefficientsInColumnMajor
	m_linearConstraintCoefficientsInColumnMajor holds the standard 3 array data structure for linear constraint coefficients (starts, indexes and values) in column major.
SparseMatrix *	m_linearConstraintCoefficientsInRowMajor
	m_linearConstraintCoefficientsInRowMajor holds the standard 3 array data structure for linear constraint coefficients (starts, indexes and values) in row major.
bool	m_bProcessQuadraticTerms
	m_bProcessQuadraticTerms holds whether the quadratic terms are processed.
int	m_iQuadraticTermNumber
	m_iQuadraticTermNumber holds the number of specified (usually nonzero) qTerms in the quadratic coefficients.
double *	m_mdConstraintFunctionValues
	m_mdConstraintFunctionValues holds a double array of constraint function values -- the size of the array is equal to getConstraintNumber().
double *	m_mdObjectiveFunctionValues
	m_mdObjectiveFunctionValues holds a double array of objective function values -- the size of the array is equal to getObjectiveNumber().
int	m_iJacValueSize
	m_iJacValueSize is the number of nonzero partial derivates in the Jacobian.
int *	m_miJacStart
	m_miJacStart holds a int array of starts for the Jacobian matrix in sparse form (row major).
int *	m_miJacIndex
	m_miJacIndex holds a int array of variable indices for the Jacobian matrix in sparse form (row major).
double *	m_mdJacValue
	m_mdJacValue holds a double array of partial derivatives for the Jacobian matrix in sparse form (row major).
int *	m_miJacNumConTerms
	m_miJacNumConTerms holds a int array of the number of constant terms (gradient does not change) for the Jacobian matrix in sparse form (row major).
SparseJacobianMatrix *	m_sparseJacMatrix
	m_sparseJacMatrix is the Jacobian matrix stored in sparse matrix format
int	m_iHighestTaylorCoeffOrder
	m_iHighestTaylorCoeffOrder is the order of highest calculated Taylor coefficient
QuadraticTerms *	m_quadraticTerms
	m_quadraticTerms the data structure for all the quadratic terms in the instance.
bool	m_bQTermsAdded
	m_bQTermsAdded is true if we add the quadratic terms to the expression tree
unsigned int	m_iNumberOfNonlinearVariables
	m_iNumberOfNonlinearVariables is the number of variables that appear in a nonlinear expression.
bool	m_bProcessNonlinearExpressions
	m_bProcessNonlinearExpressions holds whether the nonlinear expressions are processed.
int	m_iNonlinearExpressionNumber
	m_iNonlinearExpressionNumber holds the number of nonlinear expressions.
int *	m_miNonlinearExpressionIndexes
	m_miNonlinearExpressionIndexes holds an integer array of nonlinear expression indexes, negative indexes correspond to objectives.
bool	m_bProcessExpressionTrees
	m_bProcessExpressionTrees is true if the expression trees have been processed.
bool	m_bProcessExpressionTreesMod
	m_bProcessExpressionTreesMod is true if the modified expression trees have been processed.
std::map< int, OSExpressionTree * >	m_mapExpressionTrees
	m_mapExpressionTrees holds a hash map of expression tree pointers, with the key being the row index and value being the expression tree representing the nonlinear expression of that row.
std::map< int, int >	m_mapCppADFunRangeIndex
OSExpressionTree *	m_LagrangianExpTree
	m_LagrangianExpTree is an OSExpressionTree object that is the expression tree for the Lagrangian function.
bool	m_bLagrangianExpTreeCreated
	m_bLagrangianHessionCreated is true if a Lagrangian function for the Hessian has been created
SparseHessianMatrix *	m_LagrangianSparseHessian
	m_LagrangianSparseHessian is the Hessian Matrix of the Lagrangian function in sparse format
bool	m_bLagrangianSparseHessianCreated
	m_bLagrangianSparseHessianCreated is true if the sparse Hessian Matrix for the Lagrangian was created
std::map< int, int >	m_mapAllNonlinearVariablesIndex
	m_mapAllNonlinearVariablesIndexMap is a map of the variables in the Lagrangian function
int *	m_miNonLinearVarsReverseMap
	m_miNonLinearVarsReverseMap maps the nonlinear variable number back into the original variable space
bool	m_bAllNonlinearVariablesIndex
	m_bAllNonlinearVariablesIndexMap is true if the map of the variables in the Lagrangian function has been constructed
std::map< int, OSExpressionTree * >	m_mapExpressionTreesMod
	m_mapExpressionTreesMod holds a map of expression trees, with the key being the row index and value being the expression tree representing a modification of the nonlinear expression of that row.
bool	m_bCppADFunIsCreated
	m_bCppADFunIsCreated is true if we have created the OSInstanc CppAD Function
bool	m_bCppADTapesBuilt
	is true if a CppAD Expresion Tree has been built for each row and objective with a nonlinear expression.
bool	m_bCppADMustReTape
	is true if a CppAD Expresion Tree has an expression that can change depending on the value of the input, e.g.
bool	m_bDuplicateExpressionTreesMap
	m_bDuplicateExpressionTreeMap is true if m_mapExpressionTrees was duplicated.
bool	m_bNonLinearStructuresInitialized
	m_bNonLinearStructuresInitialized is true if initializeNonLinearStructures( ) has been called.
bool	m_bSparseJacobianCalculated
	m_bSparseJacobianCalculated is true if getJacobianSparsityPattern() has been called.
std::map< int, std::vector < OSnLNode * > >	m_mapExpressionTreesInPostfix
	m_mapExpressionTrees holds a hash map of expression trees in postfix format, with the key being the row index and value being the expression tree representing the nonlinear expression of that row.
int	m_iHighestOrderEvaluated
	m_iHighestOrderEvaluated is the highest order derivative of the current iterate
double **	m_mmdObjGradient
	m_mdObjGradient holds an array of pointers, each pointer points to gradient of each objective function
CppAD::vector< AD< double > >	m_vX
	m_vX is a vector of CppAD indpendent variables.
std::vector< double >	m_vdX
	m_vdX is a vector of primal variables at each iteration
std::vector< double >	m_vdYval
	m_vdYval is a vector of function values
std::vector< bool >	m_vbLagHessNonz
	m_vbLagHessNonz is a boolean vector holding the nonzero pattern of the Lagrangian of the Hessian
std::vector< double >	m_vdYjacval
	m_vdYval is a vector equal to a column or row of the Jacobian
std::vector< double >	m_vdw
	m_vdYval is a vector of derivatives -- output from a reverse sweep
std::vector< double >	m_vdLambda
	m_vdYval is a vector of Lagrange multipliers
std::vector< double >	m_vdDomainUnitVec
	m_vdDomainUnitVec is a unit vector in the domain space
std::vector< double >	m_vdRangeUnitVec
	m_vdRangeUnitVec is a unit vector in the range space
bool	m_bProcessTimeDomain
	m_bProcessTimeDomain holds whether the time domain has been processed.
bool	m_bProcessTimeStages
	m_bProcessTimeStages holds whether the time stages have been processed.
bool	m_bProcessTimeInterval
	m_bProcessTimeInterval holds whether a time interval has been processed.
bool	m_bFiniteTimeStages
	m_bFiniteTimeStages holds whether the time domain has the form of finite (discrete) stages.
int	m_iNumberOfTimeStages
	m_iNumberOfTimeStages holds the number of discrete stages

Detailed Description

The in-memory representation of an OSiL instance..

Remarks:

1. Elements become objects of class type (the ComplexType is the class)

2. The attributes, children of the element, and text correspond to members of the class. (Note text does not have a name and becomes .value)

3. Model groups such as choice and sequence and all correspond to arrays

anything specific to XML such as base64, multi, incr do not go into classes
The root OSnLNode of each <nl> element is called ExpressionTree
Root is not called osil it is called osinstance

The OSInstance class is composed of two objects: an InstanceHeader object and and InstanceData object

Definition at line 634 of file OSInstance.h.

Constructor & Destructor Documentation

OSInstance::OSInstance ( )

The OSInstance class constructor.

OSInstance::~OSInstance ( )

The OSInstance class destructor.

Member Function Documentation

bool OSInstance::processVariables ( ) [private]

process variables.

Returns:: true if the variables are processed.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

bool OSInstance::processObjectives ( ) [private]

process objectives.

Returns:: true if the objectives are processed.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

bool OSInstance::processConstraints ( ) [private]

process constraints.

Returns:: true if the constraints are processed.

Exceptions:

Exception

if the elements in constraints are logically inconsistent.

bool OSInstance::processLinearConstraintCoefficients ( ) [private]

process linear constraint coefficients.

Returns:: true if the linear constraint coefficients are processed.

Exceptions:

Exception

if the elements in linear constraint coefficients are logically inconsistent.

std::string OSInstance::getInstanceName ( )

Get instance name.

Returns:: instance name. Null or empty std::string ("") if there is no instance name.

std::string OSInstance::getInstanceSource ( )

Get instance source.

Returns:: instance source. Null or empty std::string ("") if there is no instance source.

std::string OSInstance::getInstanceDescription ( )

Get instance description.

Returns:: instance description. Null or empty std::string ("") if there is no instance description.

int OSInstance::getVariableNumber ( )

Get variable number.

Returns:: variable number.

std::string* OSInstance::getVariableNames ( )

Get variable names.

Returns:: a std::string array of variable names, null if no variable names.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

double* OSInstance::getVariableInitialValues ( )

Get variable initial values.

Returns:: a double array of variable initial values, null if no initial variable values.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

std::string* OSInstance::getVariableInitialStringValues ( )

Get variable initial std::string values.

Returns:: a std::string array of variable initial values, null if no initial variable std::string values.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

char* OSInstance::getVariableTypes ( )

Get variable types.

C for Continuous
B for Binary
I for Integer
S for String

Returns:: a char array of variable types.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

int OSInstance::getNumberOfIntegerVariables ( )

getNumberOfIntegerVariables

Returns:: an integer which is the number of I variables.

int OSInstance::getNumberOfBinaryVariables ( )

getNumberOfBinaryVariables

Returns:: an integer which is the number of B variables.

double* OSInstance::getVariableLowerBounds ( )

Get variable lower bounds.

Returns:: a double array of variable lower bounds.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

double* OSInstance::getVariableUpperBounds ( )

Get variable upper bounds.

Returns:: a double array of variable upper bounds.

Exceptions:

Exception

if the elements in variables are logically inconsistent.

int OSInstance::getObjectiveNumber ( )

Get objective number.

Returns:: objective number.

std::string* OSInstance::getObjectiveNames ( )

Get objective names.

Returns:: a std::string array of objective names. Null if no objective names.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

std::string* OSInstance::getObjectiveMaxOrMins ( )

Get objective maxOrMins.

One maxOrMin for each objective.

Returns:: a std::string array of objective maxOrMins ("max" or "min"), null if no objectives.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

int* OSInstance::getObjectiveCoefficientNumbers ( )

Get objective coefficient number.

One number for each objective.

Returns:: an integer array of size of which is equal to number of objectives, each element of the array is the number of nonzero coefficients in that objective function, null if no objectives.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

double* OSInstance::getObjectiveConstants ( )

Get objective constants.

One constant for each objective.

Returns:: a double array of objective constants, null if no objectives.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

double* OSInstance::getObjectiveWeights ( )

Get objective weights.

One weight for each objective.

Returns:: a double array of objective weights, null if no objectives.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

SparseVector** OSInstance::getObjectiveCoefficients ( )

Get objective coefficients.

One set of objective coefficients for each objective.

See also:: org.optimizationservices.oscommon.datastructure.SparseVector

Returns:: an array of objective coefficients, null if objective coefficients. Each member of the array is of type ObjectiveCoefficients. The ObjectiveCoefficients class contains two arrays: variableIndexes is an integer array and values is a double array of coefficient values.

Exceptions:

Exception

if the elements in objectives are logically inconsistent.

double** OSInstance::getDenseObjectiveCoefficients ( )

getDenseObjectiveCoefficients.

Returns:: an vector of pointers, each pointer points to a dense vector of ObjectiveCoefficients.

int OSInstance::getConstraintNumber ( )

Get constraint number.

Returns:: constraint number.

std::string* OSInstance::getConstraintNames ( )

Get constraint names.

Returns:: a std::string array of constraint names, null if no constraint names.

Exceptions:

Exception

if the elements in constraints are logically inconsistent.

double* OSInstance::getConstraintLowerBounds ( )

Get constraint lower bounds.

Returns:: a double array of constraint lower bounds, null if no constraints.

Exceptions:

Exception

if the elements in constraints are logically inconsistent.

double* OSInstance::getConstraintUpperBounds ( )

Get constraint upper bounds.

Returns:: a double array of constraint upper bounds, null if constraints.

Exceptions:

Exception

if the elements in constraints are logically inconsistent.

char* OSInstance::getConstraintTypes ( )

Get constraint types.

R for range constraint lb <= constraint <= ub
L for less than constraint -INF <= con <= ub or con <= ub
G for greater than constraint lb <= con <= INF or con >= lb
E for equal to constraint lb <= con <= ub where lb = ub or con = lb/ub
U for unconstrained constraint -INF <= con <= INF

Returns:: a char array of constraint types, null if constraints.

Exceptions:

Exception

if the elements in constraints are logically inconsistent.

int OSInstance::getLinearConstraintCoefficientNumber ( )

Get number of specified (usually nonzero) linear constraint coefficient values.

Returns:: number of specified (usually nonzero) linear constraint coefficient values.

bool OSInstance::getLinearConstraintCoefficientMajor ( )

Get whether the constraint coefficients is in column major (true) or row major (false).

Returns:: whether the constraint coefficients is in column major (true) or row major (false).

Exceptions:

Exception

if the elements in linear constraint coefficients are logically inconsistent.

SparseMatrix* OSInstance::getLinearConstraintCoefficientsInColumnMajor ( )

Get linear constraint coefficients in column major.

Returns:: a sparse matrix reprsentation of linear constraint coefficients in column major, null if no linear constraint coefficients.

Exceptions:

Exception

if the elements in linear constraint coefficients are logically inconsistent.

See also:: org.optimizationservices.oscommon.datastructure.SparseMatrix

SparseMatrix* OSInstance::getLinearConstraintCoefficientsInRowMajor ( )

Get linear constraint coefficients in row major.

Returns:: a sparse matrix reprsentation of linear constraint coefficients in row major, null if no linear constraint coefficients.

Exceptions:

Exception

if the elements in linear constraint coefficients are logically inconsistent.

See also:: org.optimizationservices.oscommon.datastructure.SparseMatrix

int OSInstance::getNumberOfQuadraticTerms ( )

Get the number of specified (usually nonzero) qTerms in the quadratic coefficients.

Returns:: qTerm number.

QuadraticTerms* OSInstance::getQuadraticTerms ( )

Get all the quadratic terms in the instance.

Returns:: the QuadraticTerms data structure for all quadratic terms in the instance, null if no quadratic terms. The QuadraticTerms contains four arrays: rowIndexes, varOneIndexes, varTwoIndexes, coefficients.

Exceptions:

Exception

if the elements in quadratic coefficients are logically inconsistent.

See also:: org.optimizationservices.oscommon.datastructure.QuadraticTerms

int* OSInstance::getQuadraticRowIndexes ( )

Get the indexes of rows which have a quadratic term.

Returns:: an integer pointer to the row indexes of rows with quadratic terms, objectives functions have index < 0 NULL if there are no quadratic terms.

int OSInstance::getNumberOfQuadraticRowIndexes ( )

Get the number of rows which have a quadratic term.

Returns:: an integer which is the number of distinct rows (including obj) with quadratic terms,

int OSInstance::getNumberOfNonlinearExpressions ( )

Get number of nonlinear expressions.

Returns:: the number of nonlinear expressions.

OSExpressionTree* OSInstance::getNonlinearExpressionTree ( int rowIdx )

Get the expression tree for a given row index.

Returns:: an expression tree

OSExpressionTree* OSInstance::getNonlinearExpressionTreeMod ( int rowIdx )

Get the expression tree for a given row index for the modified expression trees (quadratic terms added).

Returns:: an expression tree

std::vector<OSnLNode*> OSInstance::getNonlinearExpressionTreeInPostfix ( int rowIdx )

Get the postfix tokens for a given row index.

Returns:: a vector of pointers to OSnLNodes in postfix, if rowIdx does not index a row with a nonlinear term throw an exception

std::vector<OSnLNode*> OSInstance::getNonlinearExpressionTreeModInPostfix ( int rowIdx )

Get the postfix tokens for a given row index for the modified Expression Tree (quadratic terms added).

Returns:: a vector of pointers to OSnLNodes in postfix, if rowIdx does not index a row with a nonlinear term throw an exception

std::vector<OSnLNode*> OSInstance::getNonlinearExpressionTreeInPrefix ( int rowIdx )

Get the prefix tokens for a given row index.

Returns:: a vector of pointers to OSnLNodes in prefix, if rowIdx does not index a row with a nonlinear term throw an exception

std::vector<OSnLNode*> OSInstance::getNonlinearExpressionTreeModInPrefix ( int rowIdx )

Get the prefix tokens for a given row index for the modified Expression Tree (quadratic terms added).

Returns:: a vector of pointers to OSnLNodes in prefix, if rowIdx does not index a row with a nonlinear term throw an exception

int OSInstance::getNumberOfNonlinearObjectives ( )

Returns:: the number of Objectives with a nonlinear term

int OSInstance::getNumberOfNonlinearConstraints ( )

Returns:: the number of Constraints with a nonlinear term

std::map<int, OSExpressionTree* > OSInstance::getAllNonlinearExpressionTrees ( )

Returns:: a map: the key is the row index and the value is the corresponding expression tree

std::map<int, OSExpressionTree* > OSInstance::getAllNonlinearExpressionTreesMod ( )

Returns:: a map: the key is the row index and the value is the corresponding expression tree

int* OSInstance::getNonlinearExpressionTreeIndexes ( )

Get all the nonlinear expression tree indexes, i.e.

indexes of rows (objectives or constraints) that contain nonlinear expressions.

Returns:: a pointer to an integer array of nonlinear expression tree indexes.

int OSInstance::getNumberOfNonlinearExpressionTreeIndexes ( )

Get the number of unique Nonlinear exrpession tree indexes.

Returns:: the number of unique nonlinear expression tree indexes.

int* OSInstance::getNonlinearExpressionTreeModIndexes ( )

Get all the nonlinear expression tree indexes, i.e.

indexes of rows (objetives or constraints) that contain nonlinear expressions after modifying the expression tree to contain quadratic terms.

Returns:: a pointer to an integer array of nonlinear expression tree indexes (including quadratic terms).

int OSInstance::getNumberOfNonlinearExpressionTreeModIndexes ( )

Get the number of unique Nonlinear exrpession tree indexes after modifying the expression tree to contain quadratic terms.

Returns:: the number of unique nonlinear expression tree indexes (including quadratic terms).

bool OSInstance::setInstanceSource ( std::string source )

set the instance source.

Parameters:

source

holds the instance source.

Returns:: whether the instance source is set successfully.

bool OSInstance::setInstanceDescription ( std::string description )

set the instance description.

Parameters:

description

holds the instance description.

Returns:: whether the instance description is set successfully.

bool OSInstance::setInstanceName ( std::string name )

set the instance name.

Parameters:

name

holds the instance name.

Returns:: whether the instance name is set successfully.

bool OSInstance::setVariableNumber ( int number )

set the variable number.

Parameters:

number

holds the variable number.

Returns:: whether the variable number is set successfully.

bool OSInstance::addVariable	(	int	index,
		std::string	name,
		double	lowerBound,
		double	upperBound,
		char	type,
		double	init,
		std::string	initString
	)

add a variable.

In order to use the add method, the setVariableNumber must first be called so that the variable number is known ahead of time to assign appropriate memory. If a variable with the given variable index already exists, the old variable will be replaced.

Parameters:

	index	holds the variable index. It is required.
	name	holds the variable name; use null or empty std::string ("") if no variable name.
	lowerBound	holds the variable lower bound; use Double.NEGATIVE_INFINITY if no lower bound.
	upperBound	holds the variable upper bound; use Double.POSITIVE_INFINITY if no upper bound.
	type	holds the variable type character, B for Binary, I for Integer, S for String, C or any other char for Continuous)
	init	holds the double variable initial value; use Double.NaN if no initial value.
	initString	holds the std::string variable initial value; use null or empty std::string ("") if no initial std::string value.

Returns:: whether the variable is added successfully.

bool OSInstance::setVariables	(	int	number,
		std::string *	names,
		double *	lowerBounds,
		double *	upperBounds,
		char *	types,
		double *	inits,
		std::string *	initsString
	)

set all the variable related elements.

All the previous variable-related elements will be deleted.

Parameters:

	number	holds the number of variables. It is required.
	names	holds a std::string array of variable names; use null if no variable names.
	lowerBounds	holds a double array of variable lower bounds; use null if all lower bounds are 0; use Double.NEGATIVE_INFINITY if no lower bound for a specific variable in the array.
	upperBounds	holds a double array of variable upper bounds; use null if no upper bounds; use Double.POSITIVE_INFINITY if no upper bound for a specific variable in the array.
	types	holds a char array of variable types; use null if all variables are continuous; for a specfic variable in the array use B for Binary, I for Integer, S for String, C or any other char for Continuous,)
	inits	holds a double array of varible initial values; use null if no initial values.
	initsString	holds a std::string array of varible initial values; use null if no initial std::string values.

Returns:: whether the variables are set successfully.

bool OSInstance::setObjectiveNumber ( int number )

set the objective number.

Parameters:

number

holds the objective number.

Returns:: whether the objective number is set successfully.

bool OSInstance::addObjective	(	int	index,
		std::string	name,
		std::string	maxOrMin,
		double	constant,
		double	weight,
		SparseVector *	objectiveCoefficients
	)

add an objective.

In order to use the add method, the setObjectiveNumber must first be called so that the objective number is known ahead of time to assign appropriate memory. If a objective with the given objective index already exists, the old objective will be replaced. Objective index will start from -1, -2, -3, ... down, with -1 corresponding to the first objective.

Parameters:

	index	holds the objective index. Remember the first objective index is -1, second -2, ...
	name	holds the objective name; use null or empty std::string ("") if no objective name.
	maxOrMin	holds the objective sense or direction; it can only take two values: "max" or "min".
	constant	holds the objective constant; use 0.0 if no objective constant.
	weight	holds the objective weight; use 1.0 if no objective weight.
	objectiveCoefficients	holds the objective coefficients (null if no objective coefficients) in a sparse representation that holds two arrays: index array and a value array.

Returns:: whether the objective is added successfully.

bool OSInstance::setObjectives	(	int	number,
		std::string *	names,
		std::string *	maxOrMins,
		double *	constants,
		double *	weights,
		SparseVector **	objectitiveCoefficients
	)

set all the objectives related elements.

All the previous objective-related elements will be deleted.

Parameters:

	number	holds the number of objectives. It is required.
	names	holds a std::string array of objective names; use null if no objective names.
	maxOrMins	holds a std::string array of objective objective senses or directions: "max" or "min"; use null if all objectives are "min".
	constants	holds a double array of objective constants; use null if all objective constants are 0.0.
	weights	holds a double array of objective weights; use null if all objective weights are 1.0.
	objectitiveCoefficients	holds an array of objective coefficients, (null if no objective have any coefficeints) For each objective, the coefficients are stored in a sparse representation that holds two arrays: index array and a value array. If for a specific objective, there are no objecitve coefficients, use null for the corresponding array member.

Returns:: whether the objectives are set successfully.

bool OSInstance::setConstraintNumber ( int number )

set the constraint number.

Parameters:

number

holds the constraint number.

Returns:: whether the constraint number is set successfully.

bool OSInstance::addConstraint	(	int	index,
		std::string	name,
		double	lowerBound,
		double	upperBound,
		double	constant
	)

add a constraint.

In order to use the add method, the setConstraintNumber must first be called so that the constraint number is known ahead of time to assign appropriate memory. If a constraint with the given constraint index already exists, the old constraint will be replaced.

Parameters:

	index	holds the constraint index. It is required.
	name	holds the constraint name; use null or empty std::string ("") if no constraint name.
	lowerBound	holds the constraint lower bound; use Double.NEGATIVE_INFINITY if no lower bound.
	upperBound	holds the constraint upper bound; use Double.POSITIVE_INFINITY if no upper bound.

Returns:: whether the constraint is added successfully.

bool OSInstance::setConstraints	(	int	number,
		std::string *	names,
		double *	lowerBounds,
		double *	upperBounds,
		double *	constants
	)

set all the constraint related elements.

All the previous constraint-related elements will be deleted.

Parameters:

	number	holds the number of constraints. It is required.
	names	holds a std::string array of constraint names; use null if no constraint names.
	lowerBounds	holds a double array of constraint lower bounds; use null if no lower bounds; use Double.NEGATIVE_INFINITY if no lower bound for a specific constraint in the array.
	upperBounds	holds a double array of constraint upper bounds; use null if no upper bounds; use Double.POSITIVE_INFINITY if no upper bound for a specific constraint in the array.

Returns:: whether the constraints are set successfully.

bool OSInstance::setLinearConstraintCoefficients	(	int	numberOfValues,
		bool	isColumnMajor,
		double *	values,
		int	valuesBegin,
		int	valuesEnd,
		int *	indexes,
		int	indexesBegin,
		int	indexesEnd,
		int *	starts,
		int	startsBegin,
		int	startsEnd
	)

set linear constraint coefficients

Parameters:

	numberOfValues	holds the number of specified coefficient values (usually nonzero) in the coefficient matrix.
	isColumnMajor	holds whether the coefficient matrix is stored in column major (true) or row major (false).
	values	holds a double array coefficient values in the matrix.
	valuesBegin	holds the begin index of the values array to copy from (usually 0).
	valuesEnd	holds the end index of the values array to copy till (usually values.lenght - 1).
	indexes	holds an integer array column/row indexes for each value in the values array.
	indexesBegin	holds the begin index of the indexes array to copy from (usually 0).
	indexesEnd	holds the end index of the indexes array to copy till (usually indexes.lenght - 1).
	starts	holds an integer array start indexes in the matrix; the first value of starts should always be 0.
	startsBegin	holds the begin index of the starts array to copy from (usually 0).
	startsEnd	holds the end index of the starts array to copy till (usually starts.lenght - 1).

Returns:: whether the linear constraint coefficients are set successfully.

bool OSInstance::setQuadraticTerms	(	int	number,
		int *	rowIndexes,
		int *	varOneIndexes,
		int *	varTwoIndexes,
		double *	coefficients,
		int	begin,
		int	end
	)

set quadratic terms

Parameters:

	number	holds the number of quadratic terms.
	rowIndexes	holds an integer array of row indexes of all the quadratic terms. A negative integer corresponds to an objective row, e.g. -1 for 1st objective and -2 for 2nd.
	varOneIndexes	holds an integer array of the first varialbe indexes of all the quadratic terms.
	varTwoIndexes	holds an integer array of the second varialbe indexes of all the quadratic terms.
	coefficients	holds a double array all the quadratic term coefficients.
	begin	holds the begin index of all the arrays to copy from (usually = 0).
	end	holds the end index of all the arrays to copy till (usually = array length -1).

Returns:: whether the quadratic terms are set successfully.

bool OSInstance::setQuadraticTermsInNonlinearExpressions	(	int	number,
		int *	rowIndexes,
		int *	varOneIndexes,
		int *	varTwoIndexes,
		double *	coefficients
	)

set quadratic terms in nonlinearExpressions

Parameters:

	number	holds the number of quadratic terms.
	rowIndexes	holds an integer array of row indexes of all the quadratic terms. A negative integer corresponds to an objective row, e.g. -1 for 1st objective and -2 for 2nd.
	varOneIndexes	holds an integer array of the first varialbe indexes of all the quadratic terms.
	varTwoIndexes	holds an integer array of the second varialbe indexes of all the quadratic terms.
	coefficients	holds a double array all the quadratic term coefficients.

Returns:: whether the quadratic terms are set successfully.

bool OSInstance::initializeNonLinearStructures ( )

Initialize the data structures for the nonlinear API.

Returns:: true if we have initialized the nonlinear data strucutres.

double OSInstance::calculateFunctionValue	(	int	idx,
		double *	x,
		bool	new_x
	)

Calculate the function value for function (constraint or objective) indexed by idx.

Parameters:

	idx	is the index on the constraint (0, 1, 2, 3, ...) or objective function (-1, -2, -3, ...).
	x	is a pointer (double array) to the current variable values
	new_x	is false if any evaluation method was previously called for the current x has been evaluated for the current iterate x use a value of false if not sure

Returns:: the function value as a double.

double* OSInstance::calculateAllConstraintFunctionValues	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		bool	new_x,
		int	highestOrder
	)

Calculate all of the constraint function values.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins
	new_x	is false if any evaluation method was previously called for the current x for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a double array of constraint function values -- the size of the array is equal to getConstraintNumber().

double* OSInstance::calculateAllConstraintFunctionValues	(	double *	x,
		bool	new_x
	)

Calculate all of the constraint function values, we are overloading this function and this version of the method will not use any AD and will evaluate function values from the OS Expression Tree.

Parameters:

	x	is a pointer (double array) to the current variable values
	new_x	is false if any evaluation method was previously called for the current iterate

Returns:: a double array of constraint function values -- the size of the array is equal to getConstraintNumber().

double* OSInstance::calculateAllObjectiveFunctionValues	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		bool	new_x,
		int	highestOrder
	)

Calculate all of the objective function values.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a double array of objective function values -- the size of the array is equal to getObjectiveNumber().

double* OSInstance::calculateAllObjectiveFunctionValues	(	double *	x,
		bool	new_x
	)

Calculate all of the objective function values, we are overloading this function and this version of the method will not use any AD and will evaluate function values from the OS Expression Tree.

Parameters:

	x	is a pointer (double array) to the current variable values
	new_x	is false if any evaluation method was previously called for the current iterate

Returns:: a double array of objective function values -- the size of the array is equal to getConstraintNumber().

SparseJacobianMatrix* OSInstance::calculateAllConstraintFunctionGradients	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		bool	new_x,
		int	highestOrder
	)

Calculate the gradient of all constraint functions.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a pointer a SparseJacobianMatrix.

SparseVector* OSInstance::calculateConstraintFunctionGradient	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		int	idx,
		bool	new_x,
		int	highestOrder
	)

Calculate the gradient of the constraint function indexed by idx.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins idx is the index of the constraint function gradient
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a pointer to a sparse vector of doubles.

SparseVector* OSInstance::calculateConstraintFunctionGradient	(	double *	x,
		int	idx,
		bool	new_x
	)

Calculate the gradient of the constraint function indexed by idx this function is overloaded.

Parameters:

	x	is a pointer (double array) to the current variable values idx is the index of the constraint function gradient
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a pointer to a sparse vector of doubles.

double** OSInstance::calculateAllObjectiveFunctionGradients	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		bool	new_x,
		int	highestOrder
	)

Calculate the gradient of all objective functions.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: an array of pointer to dense objective function gradients.

double* OSInstance::calculateObjectiveFunctionGradient	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		int	objIdx,
		bool	new_x,
		int	highestOrder
	)

Calculate the gradient of the objective function indexed by objIdx.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins objIdx is the index of the objective function being optimized
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a pointer to a dense vector of doubles.

double* OSInstance::calculateObjectiveFunctionGradient	(	double *	x,
		int	objIdx,
		bool	new_x
	)

Calculate the gradient of the objective function indexed by objIdx this function is overloaded.

Parameters:

	x	is a pointer (double array) to the current variable values objIdx is the index of the objective function being optimized
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a pointer to a dense vector of doubles.

SparseHessianMatrix* OSInstance::calculateLagrangianHessian	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		bool	new_x,
		int	highestOrder
	)

Calculate the Hessian of the Lagrangian Expression Tree This method will build the CppAD expression tree for only the first iteration Use this method on if the value of x does not affect the operations sequence.

Parameters:

	x	is a pointer (double array) to the current variable values
	objLambda	is the Lagrange multiplier on the objective function
	conLambda	is pointer (double array) of Lagrange multipliers on the constratins
	new_x	is false if any evaluation method was previously called for the current iterate
	highestOrder	is the highest order of the derivative being calculated

Returns:: a pointer a SparseHessianMatrix. Each array member corresponds to one constraint gradient.

SparseHessianMatrix* OSInstance::calculateHessian	(	double *	x,
		int	idx,
		bool	new_x
	)

Calculate the Hessian of a constraint or objective function.

Parameters:

	x	is a pointer (double array) to the current variable values
	new_x	is false if any evaluation method was previously called for the current iterate idx is the index of the either a constraint or objective function Hessian

Returns:: a pointer a SparseVector. Each array member corresponds to one constraint gradient.

bool OSInstance::getSparseJacobianFromColumnMajor ( )

Returns:: true if successful in generating the constraints gradient.

bool OSInstance::getSparseJacobianFromRowMajor ( )

Returns:: true if successful in generating the constraints gradient.

OSExpressionTree* OSInstance::getLagrangianExpTree ( )

Returns:: a pointer to the ExpressionTree for the Lagrangian function of current instance we only take the Lagrangian of the rows with nonlinear terms

std::map<int, int> OSInstance::getAllNonlinearVariablesIndexMap ( )

Returns:: a pointer to a map of the indices of all of the variables that appear in the Lagrangian function

SparseHessianMatrix* OSInstance::getLagrangianHessianSparsityPattern ( )

Returns:: a pointer to a SparseHessianMatrix with the nonzero structure of the Lagrangian Expression Tree

bool OSInstance::addQTermsToExressionTree ( )

Returns:: true if successful in adding the qTerms to the ExpressionTree.

SparseJacobianMatrix* OSInstance::getJacobianSparsityPattern ( )

Returns:: pointer to a SparseJacobianMatrix.

void OSInstance::duplicateExpressionTreesMap ( )

duplicate the map of expression trees.

bool OSInstance::createCppADFun ( std::vector< double > vdX )

Create the a CppAD Function object: this is a function where the domain is the set of variables for the problem and the range is the objective function plus constraints.

Parameters:

vdX

is a vector of doubles holding the current primal variable values the size of x should equal instanceData->variables->numberOfVariables

Returns:: if successfully created

std::vector<double> OSInstance::forwardAD	(	int	p,
		std::vector< double >	vdX
	)

Perform an AD forward sweep.

Parameters:

	p	is the highest order Taylor coefficient
	vdX	is a vector of doubles of the current primal variable values the size of vdX m_iNumberOfNonlinearVariables

Returns:: a double vector equal to the dimension of the range space the result of the forward p sweep

std::vector<double> OSInstance::reverseAD	(	int	p,
		std::vector< double >	vdlambda
	)

Perform an AD reverse sweep.

Parameters:

	p	is the order of the sweep
	vdlambda	is a vector of doubles of the current dual (lagrange) variable values the size of lambda should equal number of objective functions plus number of constraints

Returns:: a double vector equal to the n*p

bool OSInstance::getIterateResults	(	double *	x,
		double *	objLambda,
		double *	conLambda,
		bool	new_x,
		int	highestOrder
	)

end revised AD code

Get the information for each iteration. Get the functions values, Jacobian and Hessian of the Lagrangian

Parameters:

	x	is a pointer of doubles of primal values for the current iteration
	objLambda	is is a pointer of doubles of the current dual (Lagrange) multipliers on the objective functions
	conLambda	is a pointer of doubles of the current dual (Lagrange) multipliers on the constraints
	new_x	is false if any evaluation method was previously called
	highestOrder	is the highest order derivative to be calculated

Returns:: true if successful

bool OSInstance::getZeroOrderResults	(	double *	x,
		double *	objLambda,
		double *	conLambda
	)

Calculate function values.

Parameters:

	x	is a pointer of doubles of primal values for the current iteration
	objLambda	is is a pointer of doubles of the current dual (Lagrange) multipliers on the objective functions
	conLambda	is a pointer of doubles of the current dual (Lagrange) multipliers on the constraints

Returns:: true if successful

bool OSInstance::getFirstOrderResults	(	double *	x,
		double *	objLambda,
		double *	conLambda
	)

Calculate first derivatives.

Parameters:

	x	is a pointer of doubles of primal values for the current iteration
	objLambda	is is a pointer of doubles of the current dual (Lagrange) multipliers on the objective functions
	conLambda	is a pointer of doubles of the current dual (Lagrange) multipliers on the constraints

Returns:: true if successful

bool OSInstance::getSecondOrderResults	(	double *	x,
		double *	objLambda,
		double *	conLambda
	)

Calculate second derivatives.

Parameters:

	x	is a pointer of doubles of primal values for the current iteration
	objLambda	is is a pointer of doubles of the current dual (Lagrange) multipliers on the objective functions
	conLambda	is a pointer of doubles of the current dual (Lagrange) multipliers on the constraints

Returns:: true if successful

bool OSInstance::initForAlgDiff ( )

This should be called by nonlinear solvers using callback functions.

initForAlgDiff will initialize the correct nonlinear structures in preparation for using the algorithmic differentiation routines.

Returns:: true if successful

bool OSInstance::initObjGradients ( )

This should be called by initForAlgDiff().

initObjGradients will initialize the objective function gradients to be equal to the coefficients given in the <coef> section of the OSiL instance

Returns:: true if successful

Member Data Documentation

InstanceHeader* OSInstance::instanceHeader

A pointer to an InstanceHeader object.

Definition at line 644 of file OSInstance.h.

InstanceData* OSInstance::instanceData

A pointer to an InstanceData object.

Definition at line 647 of file OSInstance.h.

std::string OSInstance::m_sInstanceName [private]

m_sInstanceName holds the instance name.

Definition at line 653 of file OSInstance.h.

std::string OSInstance::m_sInstanceSource [private]

m_sInstanceSource holds the instance source.

Definition at line 657 of file OSInstance.h.

std::string OSInstance::m_sInstanceDescription [private]

m_sInstanceDescription holds the instance description.

Definition at line 661 of file OSInstance.h.

bool OSInstance::m_bProcessVariables [private]

m_bProcessVariables holds whether the variables are processed.

Definition at line 666 of file OSInstance.h.

int OSInstance::m_iVariableNumber [private]

m_iVariableNumber holds the variable number.

Definition at line 671 of file OSInstance.h.

int OSInstance::m_iNumberOfIntegerVariables [private]

m_iNumberOfIntegerVariables holds the number of integer variables.

Definition at line 676 of file OSInstance.h.

int OSInstance::m_iNumberOfBinaryVariables [private]

m_iNumberOfBinaryVariables holds the number of binary variables.

Definition at line 681 of file OSInstance.h.

int OSInstance::m_iNumberOfQuadraticRowIndexes [private]

m_iNumberOfQuadraticRowIndexes holds the number of distinct rows and objectives with quadratic terms.

Definition at line 686 of file OSInstance.h.

bool OSInstance::m_bQuadraticRowIndexesProcessed [private]

m_bQuadraticRowIndexesProcessed is true if getQuadraticRowIndexes() has been called.

Definition at line 691 of file OSInstance.h.

int* OSInstance::m_miQuadRowIndexes [private]

m_miQuadRowIndexes is an integer pointer to the distinct rows indexes with a quadratic term.

Definition at line 696 of file OSInstance.h.

int OSInstance::m_iNumberOfNonlinearExpressionTreeIndexes [private]

m_iNumberOfNonlinearExpressionTreeIndexes holds the number of distinct rows and objectives with nonlinear terms.

Definition at line 701 of file OSInstance.h.

bool OSInstance::m_bNonlinearExpressionTreeIndexesProcessed [private]

m_bNonlinearExpressionTreeIndexesProcessed is true if getNonlinearExpressionTreeIndexes has been called.

Definition at line 706 of file OSInstance.h.

int* OSInstance::m_miNonlinearExpressionTreeIndexes [private]

m_miNonlinearExpressionTreeIndexes is an integer pointer to the distinct rows indexes in the nonlinear expression tree map.

Definition at line 712 of file OSInstance.h.

int OSInstance::m_iNumberOfNonlinearExpressionTreeModIndexes [private]

m_iNumberOfNonlinearExpressionTreeModIndexes holds the number of distinct rows and objectives with nonlinear terms including quadratic terms added to the nonlinear expression trees.

Definition at line 718 of file OSInstance.h.

bool OSInstance::m_bNonlinearExpressionTreeModIndexesProcessed [private]

m_bNonlinearExpressionTreeModIndexesProcessed is true if getNonlinearExpressionTreeModIndexes has been called.

Definition at line 723 of file OSInstance.h.

int* OSInstance::m_miNonlinearExpressionTreeModIndexes [private]

m_miNonlinearExpressionTreeModIndexes is an integer pointer to the distinct rows indexes in the modified expression tree map.

Definition at line 729 of file OSInstance.h.

std::string* OSInstance::m_msVariableNames [private]

m_msVariableNames holds an array of variable names.

Definition at line 734 of file OSInstance.h.

double* OSInstance::m_mdVariableInitialValues [private]

m_mdVariableInitialValues holds a double array of the initial variable values.

Definition at line 739 of file OSInstance.h.

std::string* OSInstance::m_msVariableInitialStringValues [private]

m_msVariableInitialStringValues holds a std::string array of the initial variable values.

Definition at line 744 of file OSInstance.h.

char* OSInstance::m_mcVariableTypes [private]

m_mcVariableTypes holds a char array of variable types (default = 'C').

(C for Continuous; B for Binary; I for Integer; S for String)

Definition at line 750 of file OSInstance.h.

double* OSInstance::m_mdVariableLowerBounds [private]

m_mdVariableLowerBounds holds a double array of variable lower bounds (default = 0.0).

Definition at line 755 of file OSInstance.h.

double* OSInstance::m_mdVariableUpperBounds [private]

m_mdVariableUpperBounds holds a double array of variable upper bounds (default = INF).

Definition at line 760 of file OSInstance.h.

bool OSInstance::m_bProcessObjectives [private]

m_bProcessObjectives holds whether the objectives are processed.

Definition at line 765 of file OSInstance.h.

int OSInstance::m_iObjectiveNumber [private]

m_iObjectiveNumber is the number of objective functions.

Definition at line 770 of file OSInstance.h.

int OSInstance::m_iObjectiveNumberNonlinear [private]

m_iObjectiveNumber is the number of objective functions with a nonlinear term.

Definition at line 775 of file OSInstance.h.

std::string* OSInstance::m_msObjectiveNames [private]

m_msObjectiveNames holds an array of objective names.

Definition at line 780 of file OSInstance.h.

std::string* OSInstance::m_msMaxOrMins [private]

m_msMaxOrMins holds a std::string array of objective maxOrMins ("max" or "min").

Definition at line 785 of file OSInstance.h.

int* OSInstance::m_miNumberOfObjCoef [private]

m_miNumberOfObjCoef holds an integer array of number of objective coefficients (default = 0.0).

Definition at line 790 of file OSInstance.h.

double* OSInstance::m_mdObjectiveConstants [private]

m_mdObjectiveConstants holds an array of objective constants (default = 0.0).

Definition at line 795 of file OSInstance.h.

double* OSInstance::m_mdObjectiveWeights [private]

m_mdObjectiveWeights holds an array of objective weights (default = 1.0).

Definition at line 800 of file OSInstance.h.

SparseVector** OSInstance::m_mObjectiveCoefficients [private]

m_mObjectiveCoefficients holds an array of objective coefficients, one set of objective coefficients for each objective.

Definition at line 806 of file OSInstance.h.

bool OSInstance::m_bGetDenseObjectives [private]

m_bGetDenseObjectives holds whether the dense objective functions are processed.

Definition at line 811 of file OSInstance.h.

double** OSInstance::m_mmdDenseObjectiveCoefficients [private]

m_mmdDenseObjectiveCoefficients holds an array of pointers, each pointer points to a vector of dense objective function coefficients

Definition at line 817 of file OSInstance.h.

bool OSInstance::m_bProcessConstraints [private]

m_bProcessConstraints holds whether the constraints are processed.

Definition at line 822 of file OSInstance.h.

int OSInstance::m_iConstraintNumber [private]

m_iConstraintNumber is the number of constraints.

Definition at line 827 of file OSInstance.h.

int OSInstance::m_iConstraintNumberNonlinear [private]

m_iConstraintNumberNonlinear is the number of constraints that have a nonlinear term.

Definition at line 832 of file OSInstance.h.

std::string* OSInstance::m_msConstraintNames [private]

m_msConstraintNames holds an array of constraint names.

Definition at line 837 of file OSInstance.h.

double* OSInstance::m_mdConstraintLowerBounds [private]

m_mdConstraintLowerBounds holds an array of constraint lower bounds (default = -INF).

Definition at line 842 of file OSInstance.h.

double* OSInstance::m_mdConstraintUpperBounds [private]

m_mdConstraintUpperBounds holds an array of constraint upper bounds (default = INF).

Definition at line 847 of file OSInstance.h.

double* OSInstance::m_mdConstraintConstants [private]

m_mdConstraintConstants holds an array of constraint constants (default = 0.0).

Definition at line 853 of file OSInstance.h.

char* OSInstance::m_mcConstraintTypes [private]

m_mcConstraintTypes holds a char array of constraint types (R for range; L for <=; G for >=; E for =; U for unconstrained)

Definition at line 859 of file OSInstance.h.

bool OSInstance::m_bProcessLinearConstraintCoefficients [private]

m_bProcessLinearConstraintCoefficients holds whether the linear constraint coefficients are processed.

Definition at line 864 of file OSInstance.h.

int OSInstance::m_iLinearConstraintCoefficientNumber [private]

m_iLinearConstraintCoefficientNumber holds the number of specified (usually nonzero) linear constraint coefficient values.

Definition at line 870 of file OSInstance.h.

bool OSInstance::m_bColumnMajor [private]

m_bColumnMajor holds whether the linear constraint coefficients are stored in column major.

Definition at line 875 of file OSInstance.h.

bool OSInstance::m_binitForAlgDiff [private]

m_binitForAlgDiff is true if initForAlgDiff() has been called.

Definition at line 880 of file OSInstance.h.

SparseMatrix* OSInstance::m_linearConstraintCoefficientsInColumnMajor [private]

m_linearConstraintCoefficientsInColumnMajor holds the standard 3 array data structure for linear constraint coefficients (starts, indexes and values) in column major.

Definition at line 887 of file OSInstance.h.

SparseMatrix* OSInstance::m_linearConstraintCoefficientsInRowMajor [private]

m_linearConstraintCoefficientsInRowMajor holds the standard 3 array data structure for linear constraint coefficients (starts, indexes and values) in row major.

Definition at line 893 of file OSInstance.h.

bool OSInstance::m_bProcessQuadraticTerms [private]

m_bProcessQuadraticTerms holds whether the quadratic terms are processed.

Definition at line 899 of file OSInstance.h.

int OSInstance::m_iQuadraticTermNumber [private]

m_iQuadraticTermNumber holds the number of specified (usually nonzero) qTerms in the quadratic coefficients.

Definition at line 905 of file OSInstance.h.

double* OSInstance::m_mdConstraintFunctionValues [private]

m_mdConstraintFunctionValues holds a double array of constraint function values -- the size of the array is equal to getConstraintNumber().

Definition at line 910 of file OSInstance.h.

double* OSInstance::m_mdObjectiveFunctionValues [private]

m_mdObjectiveFunctionValues holds a double array of objective function values -- the size of the array is equal to getObjectiveNumber().

Definition at line 915 of file OSInstance.h.

int OSInstance::m_iJacValueSize [private]

m_iJacValueSize is the number of nonzero partial derivates in the Jacobian.

Definition at line 920 of file OSInstance.h.

int* OSInstance::m_miJacStart [private]

m_miJacStart holds a int array of starts for the Jacobian matrix in sparse form (row major).

Definition at line 925 of file OSInstance.h.

int* OSInstance::m_miJacIndex [private]

m_miJacIndex holds a int array of variable indices for the Jacobian matrix in sparse form (row major).

Definition at line 930 of file OSInstance.h.

double* OSInstance::m_mdJacValue [private]

m_mdJacValue holds a double array of partial derivatives for the Jacobian matrix in sparse form (row major).

Definition at line 935 of file OSInstance.h.

int* OSInstance::m_miJacNumConTerms [private]

m_miJacNumConTerms holds a int array of the number of constant terms (gradient does not change) for the Jacobian matrix in sparse form (row major).

Definition at line 942 of file OSInstance.h.

SparseJacobianMatrix* OSInstance::m_sparseJacMatrix [private]

m_sparseJacMatrix is the Jacobian matrix stored in sparse matrix format

Definition at line 947 of file OSInstance.h.

int OSInstance::m_iHighestTaylorCoeffOrder [private]

m_iHighestTaylorCoeffOrder is the order of highest calculated Taylor coefficient

Definition at line 953 of file OSInstance.h.

QuadraticTerms* OSInstance::m_quadraticTerms [private]

m_quadraticTerms the data structure for all the quadratic terms in the instance.

` (rowIdx, varOneIdx, varTwoIdx, coef)

Definition at line 959 of file OSInstance.h.

bool OSInstance::m_bQTermsAdded [private]

m_bQTermsAdded is true if we add the quadratic terms to the expression tree

Definition at line 963 of file OSInstance.h.

unsigned int OSInstance::m_iNumberOfNonlinearVariables [private]

m_iNumberOfNonlinearVariables is the number of variables that appear in a nonlinear expression.

Definition at line 969 of file OSInstance.h.

bool OSInstance::m_bProcessNonlinearExpressions [private]

m_bProcessNonlinearExpressions holds whether the nonlinear expressions are processed.

Definition at line 974 of file OSInstance.h.

int OSInstance::m_iNonlinearExpressionNumber [private]

m_iNonlinearExpressionNumber holds the number of nonlinear expressions.

Definition at line 979 of file OSInstance.h.

int* OSInstance::m_miNonlinearExpressionIndexes [private]

m_miNonlinearExpressionIndexes holds an integer array of nonlinear expression indexes, negative indexes correspond to objectives.

Definition at line 985 of file OSInstance.h.

bool OSInstance::m_bProcessExpressionTrees [private]

m_bProcessExpressionTrees is true if the expression trees have been processed.

Definition at line 990 of file OSInstance.h.

bool OSInstance::m_bProcessExpressionTreesMod [private]

m_bProcessExpressionTreesMod is true if the modified expression trees have been processed.

Definition at line 995 of file OSInstance.h.

std::map<int, OSExpressionTree*> OSInstance::m_mapExpressionTrees [private]

m_mapExpressionTrees holds a hash map of expression tree pointers, with the key being the row index and value being the expression tree representing the nonlinear expression of that row.

Definition at line 1001 of file OSInstance.h.

std::map<int, int> OSInstance::m_mapCppADFunRangeIndex [private]

Definition at line 1005 of file OSInstance.h.

OSExpressionTree* OSInstance::m_LagrangianExpTree [private]

m_LagrangianExpTree is an OSExpressionTree object that is the expression tree for the Lagrangian function.

Definition at line 1011 of file OSInstance.h.

bool OSInstance::m_bLagrangianExpTreeCreated [private]

m_bLagrangianHessionCreated is true if a Lagrangian function for the Hessian has been created

Definition at line 1016 of file OSInstance.h.

SparseHessianMatrix* OSInstance::m_LagrangianSparseHessian [private]

m_LagrangianSparseHessian is the Hessian Matrix of the Lagrangian function in sparse format

Definition at line 1021 of file OSInstance.h.

bool OSInstance::m_bLagrangianSparseHessianCreated [private]

m_bLagrangianSparseHessianCreated is true if the sparse Hessian Matrix for the Lagrangian was created

Definition at line 1027 of file OSInstance.h.

std::map<int, int> OSInstance::m_mapAllNonlinearVariablesIndex [private]

m_mapAllNonlinearVariablesIndexMap is a map of the variables in the Lagrangian function

Definition at line 1032 of file OSInstance.h.

int* OSInstance::m_miNonLinearVarsReverseMap [private]

m_miNonLinearVarsReverseMap maps the nonlinear variable number back into the original variable space

Definition at line 1037 of file OSInstance.h.

bool OSInstance::m_bAllNonlinearVariablesIndex [private]

m_bAllNonlinearVariablesIndexMap is true if the map of the variables in the Lagrangian function has been constructed

Definition at line 1043 of file OSInstance.h.

std::map<int, OSExpressionTree*> OSInstance::m_mapExpressionTreesMod [private]

m_mapExpressionTreesMod holds a map of expression trees, with the key being the row index and value being the expression tree representing a modification of the nonlinear expression of that row.

We incorporate the linear and quadratic term for a variable into the corresponding expression tree before gradient and Hessian calculations

Definition at line 1051 of file OSInstance.h.

bool OSInstance::m_bCppADFunIsCreated [private]

m_bCppADFunIsCreated is true if we have created the OSInstanc CppAD Function

Definition at line 1057 of file OSInstance.h.

bool OSInstance::m_bCppADTapesBuilt [private]

is true if a CppAD Expresion Tree has been built for each row and objective with a nonlinear expression.

Definition at line 1063 of file OSInstance.h.

bool OSInstance::m_bCppADMustReTape [private]

is true if a CppAD Expresion Tree has an expression that can change depending on the value of the input, e.g.

an if statement -- false by default

Definition at line 1069 of file OSInstance.h.

bool OSInstance::m_bDuplicateExpressionTreesMap [private]

m_bDuplicateExpressionTreeMap is true if m_mapExpressionTrees was duplicated.

Definition at line 1074 of file OSInstance.h.

bool OSInstance::m_bNonLinearStructuresInitialized [private]

m_bNonLinearStructuresInitialized is true if initializeNonLinearStructures( ) has been called.

Definition at line 1079 of file OSInstance.h.

bool OSInstance::m_bSparseJacobianCalculated [private]

m_bSparseJacobianCalculated is true if getJacobianSparsityPattern() has been called.

Definition at line 1084 of file OSInstance.h.

std::map<int, std::vector<OSnLNode*> > OSInstance::m_mapExpressionTreesInPostfix [private]

m_mapExpressionTrees holds a hash map of expression trees in postfix format, with the key being the row index and value being the expression tree representing the nonlinear expression of that row.

Definition at line 1090 of file OSInstance.h.

int OSInstance::m_iHighestOrderEvaluated [private]

m_iHighestOrderEvaluated is the highest order derivative of the current iterate

Definition at line 1097 of file OSInstance.h.

double** OSInstance::m_mmdObjGradient [private]

m_mdObjGradient holds an array of pointers, each pointer points to gradient of each objective function

Definition at line 1103 of file OSInstance.h.

CppAD::vector< AD<double> > OSInstance::m_vX [private]

m_vX is a vector of CppAD indpendent variables.

Definition at line 1110 of file OSInstance.h.

std::vector<double> OSInstance::m_vdX [private]

m_vdX is a vector of primal variables at each iteration