Ipopt::Matrix Class Reference

Matrix Base Class. More...

#include <IpMatrix.hpp>

Inheritance diagram for Ipopt::Matrix:

[legend]

Collaboration diagram for Ipopt::Matrix:

[legend]

List of all members.

Public Member Functions
bool	HasValidNumbers () const
	Method for determining if all stored numbers are valid (i.e., no Inf or Nan).
SmartPtr< const MatrixSpace >	OwnerSpace () const
	Return the owner MatrixSpace.
Constructor/Destructor

	Matrix (const MatrixSpace *owner_space)
	Constructor.
virtual	~Matrix ()
	Destructor.
Operations of the Matrix on a Vector

void	MultVector (Number alpha, const Vector &x, Number beta, Vector &y) const
	Matrix-vector multiply.
void	TransMultVector (Number alpha, const Vector &x, Number beta, Vector &y) const
	Matrix(transpose) vector multiply.
Methods for specialized operations. A prototype
implementation is provided, but for efficient implementation those should be specially implemented.
void	AddMSinvZ (Number alpha, const Vector &S, const Vector &Z, Vector &X) const
	X = X + alpha*(Matrix S^{-1} Z).
void	SinvBlrmZMTdBr (Number alpha, const Vector &S, const Vector &R, const Vector &Z, const Vector &D, Vector &X) const
	X = S^{-1} (r + alphaZM^Td).


Index	NRows () const
	Number of rows.
Index	NCols () const
	Number of columns.
virtual void	Print (SmartPtr< const Journalist > jnlst, EJournalLevel level, EJournalCategory category, const std::string &name, Index indent=0, const std::string &prefix="") const
	Print detailed information about the matrix.
virtual void	Print (const Journalist &jnlst, EJournalLevel level, EJournalCategory category, const std::string &name, Index indent=0, const std::string &prefix="") const
	Number of rows.
Protected Member Functions
implementation methods (derived classes MUST
overload these pure virtual protected methods.
virtual void	MultVectorImpl (Number alpha, const Vector &x, Number beta, Vector &y) const =0
	Matrix-vector multiply.
virtual void	TransMultVectorImpl (Number alpha, const Vector &x, Number beta, Vector &y) const =0
	Matrix(transpose) vector multiply.
virtual void	AddMSinvZImpl (Number alpha, const Vector &S, const Vector &Z, Vector &X) const
	X = X + alpha*(Matrix S^{-1} Z).
virtual void	SinvBlrmZMTdBrImpl (Number alpha, const Vector &S, const Vector &R, const Vector &Z, const Vector &D, Vector &X) const
	X = S^{-1} (r + alphaZM^Td).
virtual bool	HasValidNumbersImpl () const
	Method for determining if all stored numbers are valid (i.e., no Inf or Nan).
virtual void	PrintImpl (const Journalist &jnlst, EJournalLevel level, EJournalCategory category, const std::string &name, Index indent, const std::string &prefix) const =0
	Print detailed information about the matrix.
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.
	Matrix ()
	default constructor
	Matrix (const Matrix &)
	Copy constructor.
Matrix &	operator= (const Matrix &)
	Overloaded Equals Operator.
Private Attributes
const SmartPtr< const MatrixSpace >	owner_space_
CachedResults data members

TaggedObject::Tag	valid_cache_tag_
bool	cached_valid_

Detailed Description

Matrix Base Class.

This is the base class for all derived matrix types. All Matrices, such as Jacobian and Hessian matrices, as well as possibly the iteration matrices needed for the step computation, are of this type.

Deriving from Matrix: Overload the protected XXX_Impl method.

Definition at line 27 of file IpMatrix.hpp.

Constructor & Destructor Documentation

Ipopt::Matrix::Matrix ( const MatrixSpace * owner_space ) [inline]

Constructor.

It has to be given a pointer to the corresponding MatrixSpace.

Definition at line 35 of file IpMatrix.hpp.

virtual Ipopt::Matrix::~Matrix ( ) [inline, virtual]

Destructor.

Definition at line 43 of file IpMatrix.hpp.

Ipopt::Matrix::Matrix ( ) [private]

default constructor

Ipopt::Matrix::Matrix ( const Matrix & ) [private]

Copy constructor.

Member Function Documentation

void Ipopt::Matrix::MultVector	(	Number	alpha,
		const Vector &	x,
		Number	beta,
		Vector &	y
	)			const `[inline]`

Matrix-vector multiply.

Computes y = alpha * Matrix * x + beta * y. Do not overload. Overload MultVectorImpl instead.

Definition at line 52 of file IpMatrix.hpp.

void Ipopt::Matrix::TransMultVector	(	Number	alpha,
		const Vector &	x,
		Number	beta,
		Vector &	y
	)			const `[inline]`

Matrix(transpose) vector multiply.

Computes y = alpha * Matrix^T * x + beta * y. Do not overload. Overload TransMultVectorImpl instead.

Definition at line 62 of file IpMatrix.hpp.

void Ipopt::Matrix::AddMSinvZ	(	Number	alpha,
		const Vector &	S,
		const Vector &	Z,
		Vector &	X
	)			const

X = X + alpha*(Matrix S^{-1} Z).

Should be implemented efficiently for the ExansionMatrix

void Ipopt::Matrix::SinvBlrmZMTdBr	(	Number	alpha,
		const Vector &	S,
		const Vector &	R,
		const Vector &	Z,
		const Vector &	D,
		Vector &	X
	)			const

X = S^{-1} (r + alpha*Z*M^Td).

Should be implemented efficiently for the ExansionMatrix

bool Ipopt::Matrix::HasValidNumbers ( ) const

Method for determining if all stored numbers are valid (i.e., no Inf or Nan).

Index Matrix::NRows ( ) const [inline]

Number of rows.

Definition at line 276 of file IpMatrix.hpp.

Index Matrix::NCols ( ) const [inline]

Number of columns.

Definition at line 282 of file IpMatrix.hpp.

virtual void Ipopt::Matrix::Print	(	SmartPtr< const Journalist >	jnlst,
		EJournalLevel	level,
		EJournalCategory	category,
		const std::string &	name,
		Index	indent = `0`,
		const std::string &	prefix = `""`
	)			const `[virtual]`

Print detailed information about the matrix.

Do not overload. Overload PrintImpl instead.

virtual void Ipopt::Matrix::Print	(	const Journalist &	jnlst,
		EJournalLevel	level,
		EJournalCategory	category,
		const std::string &	name,
		Index	indent = `0`,
		const std::string &	prefix = `""`
	)			const `[virtual]`

Number of rows.

SmartPtr< const MatrixSpace > Matrix::OwnerSpace ( ) const [inline]

Return the owner MatrixSpace.

Definition at line 288 of file IpMatrix.hpp.

virtual void Ipopt::Matrix::MultVectorImpl	(	Number	alpha,
		const Vector &	x,
		Number	beta,
		Vector &	y
	)			const `[protected, pure virtual]`

Matrix-vector multiply.

Computes y = alpha * Matrix * x + beta * y

Implemented in Ipopt::CompoundMatrix, Ipopt::CompoundSymMatrix, Ipopt::DenseGenMatrix, Ipopt::DenseSymMatrix, Ipopt::DiagMatrix, Ipopt::ExpansionMatrix, Ipopt::IdentityMatrix, Ipopt::LowRankUpdateSymMatrix, Ipopt::MultiVectorMatrix, Ipopt::ScaledMatrix, Ipopt::SumMatrix, Ipopt::SumSymMatrix, Ipopt::SymScaledMatrix, Ipopt::ZeroMatrix, Ipopt::GenTMatrix, and Ipopt::SymTMatrix.

virtual void Ipopt::Matrix::TransMultVectorImpl	(	Number	alpha,
		const Vector &	x,
		Number	beta,
		Vector &	y
	)			const `[protected, pure virtual]`

Matrix(transpose) vector multiply.

Computes y = alpha * Matrix^T * x + beta * y

Implemented in Ipopt::CompoundMatrix, Ipopt::DenseGenMatrix, Ipopt::ExpansionMatrix, Ipopt::MultiVectorMatrix, Ipopt::ScaledMatrix, Ipopt::SumMatrix, Ipopt::SymMatrix, Ipopt::ZeroMatrix, and Ipopt::GenTMatrix.

virtual void Ipopt::Matrix::AddMSinvZImpl	(	Number	alpha,
		const Vector &	S,
		const Vector &	Z,
		Vector &	X
	)			const `[protected, virtual]`

X = X + alpha*(Matrix S^{-1} Z).

Prototype for this specialize method is provided, but for efficient implementation it should be overloaded for the expansion matrix.

Reimplemented in Ipopt::CompoundMatrix, Ipopt::ExpansionMatrix, Ipopt::IdentityMatrix, and Ipopt::ScaledMatrix.

virtual void Ipopt::Matrix::SinvBlrmZMTdBrImpl	(	Number	alpha,
		const Vector &	S,
		const Vector &	R,
		const Vector &	Z,
		const Vector &	D,
		Vector &	X
	)			const `[protected, virtual]`

X = S^{-1} (r + alpha*Z*M^Td).

Should be implemented efficiently for the ExpansionMatrix.

Reimplemented in Ipopt::CompoundMatrix, Ipopt::ExpansionMatrix, and Ipopt::ScaledMatrix.

virtual bool Ipopt::Matrix::HasValidNumbersImpl ( ) const [inline, protected, virtual]

Method for determining if all stored numbers are valid (i.e., no Inf or Nan).

A default implementation always returning true is provided, but if possible it should be implemented.

Reimplemented in Ipopt::CompoundMatrix, Ipopt::CompoundSymMatrix, Ipopt::DenseGenMatrix, Ipopt::DenseSymMatrix, Ipopt::DiagMatrix, Ipopt::IdentityMatrix, Ipopt::LowRankUpdateSymMatrix, Ipopt::MultiVectorMatrix, Ipopt::ScaledMatrix, Ipopt::SumMatrix, Ipopt::SumSymMatrix, Ipopt::SymScaledMatrix, Ipopt::GenTMatrix, and Ipopt::SymTMatrix.

Definition at line 154 of file IpMatrix.hpp.

virtual void Ipopt::Matrix::PrintImpl	(	const Journalist &	jnlst,
		EJournalLevel	level,
		EJournalCategory	category,
		const std::string &	name,
		Index	indent,
		const std::string &	prefix
	)			const `[protected, pure virtual]`