Solver for the augmented system with LowRankUpdateSymMatrix Hessian matrices. More...

#include <IpLowRankAugSystemSolver.hpp>

Inheritance diagram for Ipopt::LowRankAugSystemSolver:

Public Member Functions
bool	InitializeImpl (const OptionsList &options, const std::string &prefix)
	overloaded from AlgorithmStrategyObject More...

virtual ESymSolverStatus	Solve (const SymMatrix W, double W_factor, const Vector D_x, double delta_x, const Vector D_s, double delta_s, const Matrix J_c, const Vector D_c, double delta_c, const Matrix J_d, const Vector *D_d, double delta_d, const Vector &rhs_x, const Vector &rhs_s, const Vector &rhs_c, const Vector &rhs_d, Vector &sol_x, Vector &sol_s, Vector &sol_c, Vector &sol_d, bool check_NegEVals, Index numberOfNegEVals)
	Set up the augmented system and solve it for a given right hand side. More...

virtual Index	NumberOfNegEVals () const
	Number of negative eigenvalues detected during last solve. More...

virtual bool	ProvidesInertia () const
	Query whether inertia is computed by linear solver. More...

virtual bool	IncreaseQuality ()
	Request to increase quality of solution for next solve. More...

Constructors/Destructors
	LowRankAugSystemSolver (AugSystemSolver &aug_system_solver)
	Constructor using only a linear solver object. More...

virtual	~LowRankAugSystemSolver ()
	Default destructor. More...

Public Member Functions inherited from Ipopt::AugSystemSolver
virtual ESymSolverStatus	MultiSolve (const SymMatrix W, double W_factor, const Vector D_x, double delta_x, const Vector D_s, double delta_s, const Matrix J_c, const Vector D_c, double delta_c, const Matrix J_d, const Vector *D_d, double delta_d, std::vector< SmartPtr< const Vector > > &rhs_xV, std::vector< SmartPtr< const Vector > > &rhs_sV, std::vector< SmartPtr< const Vector > > &rhs_cV, std::vector< SmartPtr< const Vector > > &rhs_dV, std::vector< SmartPtr< Vector > > &sol_xV, std::vector< SmartPtr< Vector > > &sol_sV, std::vector< SmartPtr< Vector > > &sol_cV, std::vector< SmartPtr< Vector > > &sol_dV, bool check_NegEVals, Index numberOfNegEVals)
	Like Solve, but for multiple right hand sides. More...

	AugSystemSolver ()
	Default constructor. More...

virtual	~AugSystemSolver ()
	Default destructor. More...

Public Member Functions inherited from Ipopt::AlgorithmStrategyObject
bool	Initialize (const Journalist &jnlst, IpoptNLP &ip_nlp, IpoptData &ip_data, IpoptCalculatedQuantities &ip_cq, const OptionsList &options, const std::string &prefix)
	This method is called every time the algorithm starts again - it is used to reset any internal state. More...

bool	ReducedInitialize (const Journalist &jnlst, const OptionsList &options, const std::string &prefix)
	Reduced version of the Initialize method, which does not require special Ipopt information. More...

	AlgorithmStrategyObject ()
	Default Constructor. More...

virtual	~AlgorithmStrategyObject ()
	Default Destructor. More...

Public Member Functions inherited from Ipopt::ReferencedObject
	ReferencedObject ()

virtual	~ReferencedObject ()

Index	ReferenceCount () const

void	AddRef (const Referencer *referencer) const

void	ReleaseRef (const Referencer *referencer) const

Private Member Functions
Default Compiler Generated Methods
(Hidden to avoid implicit creation/calling). These methods are not implemented and we do not want the compiler to implement them for us, so we declare them private and do not define them. This ensures that they will not be implicitly created/called.
	LowRankAugSystemSolver ()
	Default constructor. More...

	LowRankAugSystemSolver (const LowRankAugSystemSolver &)
	Copy Constructor. More...

void	operator= (const LowRankAugSystemSolver &)
	Overloaded Equals Operator. More...

Internal functions
ESymSolverStatus	UpdateFactorization (const SymMatrix W, double W_factor, const Vector D_x, double delta_x, const Vector D_s, double delta_s, const Matrix &J_c, const Vector D_c, double delta_c, const Matrix &J_d, const Vector *D_d, double delta_d, const Vector &proto_rhs_x, const Vector &proto_rhs_s, const Vector &proto_rhs_c, const Vector &proto_rhs_d, bool check_NegEVals, Index numberOfNegEVals)
	Method for updating the factorization, including J1_, J2_, Vtilde1_, Utilde2, Wdiag_, compound_sol_vecspace_. More...

ESymSolverStatus	SolveMultiVector (const Vector D_x, double delta_x, const Vector D_s, double delta_s, const Matrix &J_c, const Vector D_c, double delta_c, const Matrix &J_d, const Vector D_d, double delta_d, const Vector &proto_rhs_x, const Vector &proto_rhs_s, const Vector &proto_rhs_c, const Vector &proto_rhs_d, const MultiVectorMatrix &V, const SmartPtr< const Matrix > &P_LM, SmartPtr< MultiVectorMatrix > &V_x, SmartPtr< MultiVectorMatrix > &Vtilde1, SmartPtr< MultiVectorMatrix > &Vtilde1_x, bool check_NegEVals, Index numberOfNegEVals)
	Method for solving the augmented system without low-rank update for multiple right hand sides that are provided as MultiVectorMatrix. More...

bool	AugmentedSystemRequiresChange (const SymMatrix W, double W_factor, const Vector D_x, double delta_x, const Vector D_s, double delta_s, const Matrix &J_c, const Vector D_c, double delta_c, const Matrix &J_d, const Vector *D_d, double delta_d)
	Method that compares the tags of the data for the matrix with those from the previous call. More...

Private Attributes
SmartPtr< AugSystemSolver >	aug_system_solver_
	The augmented system solver object that should be used for the factorization of the augmented system without the low-rank update. More...

Index	num_neg_evals_
	Stores the number of negative eigenvalues detected during most recent factorization. More...

Tags and values to track in order to decide whether the
matrix has to be updated compared to the most recent call of the Set method.
TaggedObject::Tag	w_tag_
	Tag for W matrix. More...

double	w_factor_
	Most recent value of W_factor. More...

TaggedObject::Tag	d_x_tag_
	Tag for D_x vector, representing the diagonal matrix D_x. More...

double	delta_x_
	Most recent value of delta_x from Set method. More...

TaggedObject::Tag	d_s_tag_
	Tag for D_s vector, representing the diagonal matrix D_s. More...

double	delta_s_
	Most recent value of delta_s from Set method. More...

TaggedObject::Tag	j_c_tag_
	Tag for J_c matrix. More...

TaggedObject::Tag	d_c_tag_
	Tag for D_c vector, representing the diagonal matrix D_c. More...

double	delta_c_
	Most recent value of delta_c from Set method. More...

TaggedObject::Tag	j_d_tag_
	Tag for J_d matrix. More...

TaggedObject::Tag	d_d_tag_
	Tag for D_d vector, representing the diagonal matrix D_d. More...

double	delta_d_
	Most recent value of delta_d from Set method. More...

Information to be stored in order to resolve for the
same matrix with a different right hand side.
bool	first_call_

SmartPtr< DenseGenMatrix >	J1_

SmartPtr< DenseGenMatrix >	J2_

SmartPtr< MultiVectorMatrix >	Vtilde1_

SmartPtr< MultiVectorMatrix >	Utilde2_

SmartPtr< DiagMatrix >	Wdiag_
	Hessian Matrix passed to the augmented system solver solving the matrix without the low-rank update. More...

SmartPtr< const CompoundVectorSpace >	compound_sol_vecspace_
	Vector space for Compound vectors that capture the entire right hand side and solution vectors . More...

Additional Inherited Members
Protected Member Functions inherited from Ipopt::AlgorithmStrategyObject
const Journalist &	Jnlst () const

IpoptNLP &	IpNLP () const

IpoptData &	IpData () const

IpoptCalculatedQuantities &	IpCq () const

bool	HaveIpData () const

Detailed Description

Solver for the augmented system with LowRankUpdateSymMatrix Hessian matrices.

This version works with the Sherman-Morrison formula and multiple backsolves.

Definition at line 24 of file IpLowRankAugSystemSolver.hpp.

Constructor & Destructor Documentation

Ipopt::LowRankAugSystemSolver::LowRankAugSystemSolver ( AugSystemSolver & aug_system_solver )

Constructor using only a linear solver object.

virtual Ipopt::LowRankAugSystemSolver::~LowRankAugSystemSolver ( )

virtual

Default destructor.

Ipopt::LowRankAugSystemSolver::LowRankAugSystemSolver ( )

private

Default constructor.

Ipopt::LowRankAugSystemSolver::LowRankAugSystemSolver ( const LowRankAugSystemSolver & )

private

Copy Constructor.

Member Function Documentation

bool Ipopt::LowRankAugSystemSolver::InitializeImpl	(	const OptionsList &	options,
		const std::string &	prefix
	)

virtual

overloaded from AlgorithmStrategyObject

Implements Ipopt::AugSystemSolver.

virtual ESymSolverStatus Ipopt::LowRankAugSystemSolver::Solve	(	const SymMatrix *	W,
		double	W_factor,
		const Vector *	D_x,
		double	delta_x,
		const Vector *	D_s,
		double	delta_s,
		const Matrix *	J_c,
		const Vector *	D_c,
		double	delta_c,
		const Matrix *	J_d,
		const Vector *	D_d,
		double	delta_d,
		const Vector &	rhs_x,
		const Vector &	rhs_s,
		const Vector &	rhs_c,
		const Vector &	rhs_d,
		Vector &	sol_x,
		Vector &	sol_s,
		Vector &	sol_c,
		Vector &	sol_d,
		bool	check_NegEVals,
		Index	numberOfNegEVals
	)

virtual

Set up the augmented system and solve it for a given right hand side.

Reimplemented from Ipopt::AugSystemSolver.

virtual Index Ipopt::LowRankAugSystemSolver::NumberOfNegEVals ( ) const

virtual

Number of negative eigenvalues detected during last solve.

Returns the number of negative eigenvalues of the most recent factorized matrix. This must not be called if the linear solver does not compute this quantities (see ProvidesInertia).

Implements Ipopt::AugSystemSolver.

virtual bool Ipopt::LowRankAugSystemSolver::ProvidesInertia ( ) const

virtual

Query whether inertia is computed by linear solver.

Returns true, if linear solver provides inertia.

Implements Ipopt::AugSystemSolver.

virtual bool Ipopt::LowRankAugSystemSolver::IncreaseQuality ( )

virtual

Request to increase quality of solution for next solve.

Ask underlying linear solver to increase quality of solution for the next solve (e.g. increase pivot tolerance). Returns false, if this is not possible (e.g. maximal pivot tolerance already used.)

Implements Ipopt::AugSystemSolver.

void Ipopt::LowRankAugSystemSolver::operator= ( const LowRankAugSystemSolver & )

private

Overloaded Equals Operator.

ESymSolverStatus Ipopt::LowRankAugSystemSolver::UpdateFactorization	(	const SymMatrix *	W,
		double	W_factor,
		const Vector *	D_x,
		double	delta_x,
		const Vector *	D_s,
		double	delta_s,
		const Matrix &	J_c,
		const Vector *	D_c,
		double	delta_c,
		const Matrix &	J_d,
		const Vector *	D_d,
		double	delta_d,
		const Vector &	proto_rhs_x,
		const Vector &	proto_rhs_s,
		const Vector &	proto_rhs_c,
		const Vector &	proto_rhs_d,
		bool	check_NegEVals,
		Index	numberOfNegEVals
	)

private

Method for updating the factorization, including J1_, J2_, Vtilde1_, Utilde2, Wdiag_, compound_sol_vecspace_.

ESymSolverStatus Ipopt::LowRankAugSystemSolver::SolveMultiVector	(	const Vector *	D_x,
		double	delta_x,
		const Vector *	D_s,
		double	delta_s,
		const Matrix &	J_c,
		const Vector *	D_c,
		double	delta_c,
		const Matrix &	J_d,
		const Vector *	D_d,
		double	delta_d,
		const Vector &	proto_rhs_x,
		const Vector &	proto_rhs_s,
		const Vector &	proto_rhs_c,
		const Vector &	proto_rhs_d,
		const MultiVectorMatrix &	V,
		const SmartPtr< const Matrix > &	P_LM,
		SmartPtr< MultiVectorMatrix > &	V_x,
		SmartPtr< MultiVectorMatrix > &	Vtilde1,
		SmartPtr< MultiVectorMatrix > &	Vtilde1_x,
		bool	check_NegEVals,
		Index	numberOfNegEVals
	)

private

Method for solving the augmented system without low-rank update for multiple right hand sides that are provided as MultiVectorMatrix.

The result is returned as a MultiVectorMatrix in Vtilde1. V_x and Vtilde1_x are V and Vtilde1 in the x-space.

bool Ipopt::LowRankAugSystemSolver::AugmentedSystemRequiresChange	(	const SymMatrix *	W,
		double	W_factor,
		const Vector *	D_x,
		double	delta_x,
		const Vector *	D_s,
		double	delta_s,
		const Matrix &	J_c,
		const Vector *	D_c,
		double	delta_c,
		const Matrix &	J_d,
		const Vector *	D_d,
		double	delta_d
	)

private

Method that compares the tags of the data for the matrix with those from the previous call.

Returns true, if there was a change and the factorization has to be updated.

Member Data Documentation

SmartPtr<AugSystemSolver> Ipopt::LowRankAugSystemSolver::aug_system_solver_

private

The augmented system solver object that should be used for the factorization of the augmented system without the low-rank update.

Definition at line 110 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::w_tag_

private

Tag for W matrix.

If W has been given to Set as NULL, then this tag is set to 0

Definition at line 120 of file IpLowRankAugSystemSolver.hpp.

double Ipopt::LowRankAugSystemSolver::w_factor_

private

Most recent value of W_factor.

Definition at line 122 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::d_x_tag_

private

Tag for D_x vector, representing the diagonal matrix D_x.

If D_x has been given to Set as NULL, then this tag is set to 0

Definition at line 126 of file IpLowRankAugSystemSolver.hpp.

double Ipopt::LowRankAugSystemSolver::delta_x_

private

Most recent value of delta_x from Set method.

Definition at line 128 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::d_s_tag_

private

Tag for D_s vector, representing the diagonal matrix D_s.

If D_s has been given to Set as NULL, then this tag is set to 0

Definition at line 132 of file IpLowRankAugSystemSolver.hpp.

double Ipopt::LowRankAugSystemSolver::delta_s_

private

Most recent value of delta_s from Set method.

Definition at line 134 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::j_c_tag_

private

Tag for J_c matrix.

If J_c has been given to Set as NULL, then this tag is set to 0

Definition at line 138 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::d_c_tag_

private

Tag for D_c vector, representing the diagonal matrix D_c.

If D_c has been given to Set as NULL, then this tag is set to 0

Definition at line 142 of file IpLowRankAugSystemSolver.hpp.

double Ipopt::LowRankAugSystemSolver::delta_c_

private

Most recent value of delta_c from Set method.

Definition at line 144 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::j_d_tag_

private

Tag for J_d matrix.

If J_d has been given to Set as NULL, then this tag is set to 0

Definition at line 148 of file IpLowRankAugSystemSolver.hpp.

TaggedObject::Tag Ipopt::LowRankAugSystemSolver::d_d_tag_

private

Tag for D_d vector, representing the diagonal matrix D_d.

If D_d has been given to Set as NULL, then this tag is set to 0

Definition at line 152 of file IpLowRankAugSystemSolver.hpp.

double Ipopt::LowRankAugSystemSolver::delta_d_

private

Most recent value of delta_d from Set method.

Definition at line 154 of file IpLowRankAugSystemSolver.hpp.

bool Ipopt::LowRankAugSystemSolver::first_call_

private

Definition at line 160 of file IpLowRankAugSystemSolver.hpp.

SmartPtr<DenseGenMatrix> Ipopt::LowRankAugSystemSolver::J1_

private

Definition at line 161 of file IpLowRankAugSystemSolver.hpp.

SmartPtr<DenseGenMatrix> Ipopt::LowRankAugSystemSolver::J2_

private

Definition at line 162 of file IpLowRankAugSystemSolver.hpp.

SmartPtr<MultiVectorMatrix> Ipopt::LowRankAugSystemSolver::Vtilde1_

private

Definition at line 163 of file IpLowRankAugSystemSolver.hpp.

SmartPtr<MultiVectorMatrix> Ipopt::LowRankAugSystemSolver::Utilde2_

private

Definition at line 164 of file IpLowRankAugSystemSolver.hpp.

SmartPtr<DiagMatrix> Ipopt::LowRankAugSystemSolver::Wdiag_

private

Hessian Matrix passed to the augmented system solver solving the matrix without the low-rank update.

Definition at line 167 of file IpLowRankAugSystemSolver.hpp.

SmartPtr<const CompoundVectorSpace> Ipopt::LowRankAugSystemSolver::compound_sol_vecspace_

private

Vector space for Compound vectors that capture the entire right hand side and solution vectors .

Definition at line 170 of file IpLowRankAugSystemSolver.hpp.

Index Ipopt::LowRankAugSystemSolver::num_neg_evals_

private

Stores the number of negative eigenvalues detected during most recent factorization.

This is what is returned by NumberOfNegEVals() of this class. It usually is the number of negative eigenvalues retured from the aug_system_solver solve, but if a Cholesky factorization could not be performed, the returned value is one more than this what the aug_system_solver returned.

Definition at line 180 of file IpLowRankAugSystemSolver.hpp.

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

src/Algorithm/IpLowRankAugSystemSolver.hpp

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation