This is an adapter for an NLP that converts variable bound constraints to inequality constraints. More...

#include <IpNLPBoundsRemover.hpp>

Inheritance diagram for Ipopt::NLPBoundsRemover:

Public Member Functions
virtual void	GetQuasiNewtonApproximationSpaces (SmartPtr< VectorSpace > &approx_space, SmartPtr< Matrix > &P_approx)
	Method for obtaining the subspace in which the limited-memory Hessian approximation should be done.

SmartPtr< NLP >	nlp ()
	Accessor method to the original NLP.

Constructors / Destructor
	NLPBoundsRemover (NLP &nlp, bool allow_twosided_inequalities=false)
	The constructor is given the NLP of which the bounds are to be replaced by inequality constraints.

virtual	~NLPBoundsRemover ()
	Destructor.

NLP Initialization.
virtual bool	ProcessOptions (const OptionsList &options, const std::string &prefix)
	Overload if you want the chance to process options or parameters that may be specific to the NLP.

virtual bool	GetSpaces (SmartPtr< const VectorSpace > &x_space, SmartPtr< const VectorSpace > &c_space, SmartPtr< const VectorSpace > &d_space, SmartPtr< const VectorSpace > &x_l_space, SmartPtr< const MatrixSpace > &px_l_space, SmartPtr< const VectorSpace > &x_u_space, SmartPtr< const MatrixSpace > &px_u_space, SmartPtr< const VectorSpace > &d_l_space, SmartPtr< const MatrixSpace > &pd_l_space, SmartPtr< const VectorSpace > &d_u_space, SmartPtr< const MatrixSpace > &pd_u_space, SmartPtr< const MatrixSpace > &Jac_c_space, SmartPtr< const MatrixSpace > &Jac_d_space, SmartPtr< const SymMatrixSpace > &Hess_lagrangian_space)
	Method for creating the derived vector / matrix types.

virtual bool	GetBoundsInformation (const Matrix &Px_L, Vector &x_L, const Matrix &Px_U, Vector &x_U, const Matrix &Pd_L, Vector &d_L, const Matrix &Pd_U, Vector &d_U)
	Method for obtaining the bounds information.

virtual bool	GetStartingPoint (SmartPtr< Vector > x, bool need_x, SmartPtr< Vector > y_c, bool need_y_c, SmartPtr< Vector > y_d, bool need_y_d, SmartPtr< Vector > z_L, bool need_z_L, SmartPtr< Vector > z_U, bool need_z_U)
	Method for obtaining the starting point for all the iterates.

virtual bool	GetWarmStartIterate (IteratesVector &warm_start_iterate)
	Method for obtaining an entire iterate as a warmstart point.

NLP evaluation routines.
virtual bool	Eval_f (const Vector &x, Number &f)

virtual bool	Eval_grad_f (const Vector &x, Vector &g_f)

virtual bool	Eval_c (const Vector &x, Vector &c)

virtual bool	Eval_jac_c (const Vector &x, Matrix &jac_c)

virtual bool	Eval_d (const Vector &x, Vector &d)

virtual bool	Eval_jac_d (const Vector &x, Matrix &jac_d)

virtual bool	Eval_h (const Vector &x, Number obj_factor, const Vector &yc, const Vector &yd, SymMatrix &h)

NLP solution routines.
virtual void	FinalizeSolution (SolverReturn status, const Vector &x, const Vector &z_L, const Vector &z_U, const Vector &c, const Vector &d, const Vector &y_c, const Vector &y_d, Number obj_value, const IpoptData ip_data, IpoptCalculatedQuantities ip_cq)
	This method is called at the very end of the optimization.

virtual bool	IntermediateCallBack (AlgorithmMode mode, Index iter, Number obj_value, Number inf_pr, Number inf_du, Number mu, Number d_norm, Number regularization_size, Number alpha_du, Number alpha_pr, Index ls_trials, const IpoptData ip_data, IpoptCalculatedQuantities ip_cq)
	This method is called once per iteration, after the iteration summary output has been printed.


virtual void	GetScalingParameters (const SmartPtr< const VectorSpace > x_space, const SmartPtr< const VectorSpace > c_space, const SmartPtr< const VectorSpace > d_space, Number &obj_scaling, SmartPtr< Vector > &x_scaling, SmartPtr< Vector > &c_scaling, SmartPtr< Vector > &d_scaling) const
	Routines to get the scaling parameters.

Public Member Functions inherited from Ipopt::NLP
	NLP ()
	Default constructor.

virtual	~NLP ()
	Default destructor.

	DECLARE_STD_EXCEPTION (USER_SCALING_NOT_IMPLEMENTED)
	Exceptions.

	DECLARE_STD_EXCEPTION (INVALID_NLP)

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.
	NLPBoundsRemover ()
	Default Constructor.

	NLPBoundsRemover (const NLPBoundsRemover &)
	Copy Constructor.

void	operator= (const NLPBoundsRemover &)
	Default Assignment Operator.

Private Attributes
SmartPtr< NLP >	nlp_
	Pointer to the original NLP.

SmartPtr< const Matrix >	Px_l_orig_
	Pointer to the expansion matrix for the lower x bounds.

SmartPtr< const Matrix >	Px_u_orig_
	Pointer to the expansion matrix for the upper x bounds.

SmartPtr< const VectorSpace >	d_space_orig_
	Pointer to the original d space.

bool	allow_twosided_inequalities_
	Flag indicating whether twosided inequality constraints are allowed.

Detailed Description

This is an adapter for an NLP that converts variable bound constraints to inequality constraints.

This is necessary for the version of Ipopt that uses iterative linear solvers. At this point, none of the original inequality constraints is allowed to have both lower and upper bounds. The NLP visible to Ipopt via this adapter will not have any bounds on variables, but have equivalent inequality constraints.

Definition at line 23 of file IpNLPBoundsRemover.hpp.

Constructor & Destructor Documentation

◆ NLPBoundsRemover() [1/3]

Ipopt::NLPBoundsRemover::NLPBoundsRemover	(	NLP &	nlp,
		bool	allow_twosided_inequalities = `false`
	)

The constructor is given the NLP of which the bounds are to be replaced by inequality constraints.

◆ ~NLPBoundsRemover()

virtual Ipopt::NLPBoundsRemover::~NLPBoundsRemover ( )

inlinevirtual

Destructor.

Definition at line 37 of file IpNLPBoundsRemover.hpp.

◆ NLPBoundsRemover() [2/3]

Ipopt::NLPBoundsRemover::NLPBoundsRemover ( )

private

Default Constructor.

◆ NLPBoundsRemover() [3/3]

Ipopt::NLPBoundsRemover::NLPBoundsRemover ( const NLPBoundsRemover & )

private

Copy Constructor.

Member Function Documentation

◆ ProcessOptions()

virtual bool Ipopt::NLPBoundsRemover::ProcessOptions	(	const OptionsList &	options,
		const std::string &	prefix
	)

inlinevirtual

Overload if you want the chance to process options or parameters that may be specific to the NLP.

Reimplemented from Ipopt::NLP.

Definition at line 46 of file IpNLPBoundsRemover.hpp.

◆ GetSpaces()

virtual bool Ipopt::NLPBoundsRemover::GetSpaces	(	SmartPtr< const VectorSpace > &	x_space,
		SmartPtr< const VectorSpace > &	c_space,
		SmartPtr< const VectorSpace > &	d_space,
		SmartPtr< const VectorSpace > &	x_l_space,
		SmartPtr< const MatrixSpace > &	px_l_space,
		SmartPtr< const VectorSpace > &	x_u_space,
		SmartPtr< const MatrixSpace > &	px_u_space,
		SmartPtr< const VectorSpace > &	d_l_space,
		SmartPtr< const MatrixSpace > &	pd_l_space,
		SmartPtr< const VectorSpace > &	d_u_space,
		SmartPtr< const MatrixSpace > &	pd_u_space,
		SmartPtr< const MatrixSpace > &	Jac_c_space,
		SmartPtr< const MatrixSpace > &	Jac_d_space,
		SmartPtr< const SymMatrixSpace > &	Hess_lagrangian_space
	)

virtual

Method for creating the derived vector / matrix types.

The Hess_lagrangian_space pointer can be NULL if a quasi-Newton options is chosen.

Implements Ipopt::NLP.

◆ GetBoundsInformation()

virtual bool Ipopt::NLPBoundsRemover::GetBoundsInformation	(	const Matrix &	Px_L,
		Vector &	x_L,
		const Matrix &	Px_U,
		Vector &	x_U,
		const Matrix &	Pd_L,
		Vector &	d_L,
		const Matrix &	Pd_U,
		Vector &	d_U
	)

virtual

Method for obtaining the bounds information.

Implements Ipopt::NLP.

◆ GetStartingPoint()

virtual bool Ipopt::NLPBoundsRemover::GetStartingPoint	(	SmartPtr< Vector >	x,
		bool	need_x,
		SmartPtr< Vector >	y_c,
		bool	need_y_c,
		SmartPtr< Vector >	y_d,
		bool	need_y_d,
		SmartPtr< Vector >	z_L,
		bool	need_z_L,
		SmartPtr< Vector >	z_U,
		bool	need_z_U
	)

virtual

Method for obtaining the starting point for all the iterates.

Implements Ipopt::NLP.

◆ GetWarmStartIterate()

virtual bool Ipopt::NLPBoundsRemover::GetWarmStartIterate ( IteratesVector & warm_start_iterate )

inlinevirtual

Method for obtaining an entire iterate as a warmstart point.

The incoming IteratesVector has to be filled. This has not yet been implemented for this adapter.

Reimplemented from Ipopt::NLP.

Definition at line 107 of file IpNLPBoundsRemover.hpp.

◆ Eval_f()

virtual bool Ipopt::NLPBoundsRemover::Eval_f	(	const Vector &	x,
		Number &	f
	)

inlinevirtual

Implements Ipopt::NLP.

Definition at line 117 of file IpNLPBoundsRemover.hpp.

◆ Eval_grad_f()

virtual bool Ipopt::NLPBoundsRemover::Eval_grad_f	(	const Vector &	x,
		Vector &	g_f
	)

inlinevirtual

Implements Ipopt::NLP.

Definition at line 125 of file IpNLPBoundsRemover.hpp.

◆ Eval_c()

virtual bool Ipopt::NLPBoundsRemover::Eval_c	(	const Vector &	x,
		Vector &	c
	)

inlinevirtual

Implements Ipopt::NLP.

Definition at line 133 of file IpNLPBoundsRemover.hpp.

◆ Eval_jac_c()

virtual bool Ipopt::NLPBoundsRemover::Eval_jac_c	(	const Vector &	x,
		Matrix &	jac_c
	)

inlinevirtual

Implements Ipopt::NLP.

Definition at line 141 of file IpNLPBoundsRemover.hpp.

◆ Eval_d()

virtual bool Ipopt::NLPBoundsRemover::Eval_d	(	const Vector &	x,
		Vector &	d
	)

virtual

Implements Ipopt::NLP.

◆ Eval_jac_d()

virtual bool Ipopt::NLPBoundsRemover::Eval_jac_d	(	const Vector &	x,
		Matrix &	jac_d
	)

virtual

Implements Ipopt::NLP.

◆ Eval_h()

virtual bool Ipopt::NLPBoundsRemover::Eval_h	(	const Vector &	x,
		Number	obj_factor,
		const Vector &	yc,
		const Vector &	yd,
		SymMatrix &	h
	)

virtual

Implements Ipopt::NLP.

◆ FinalizeSolution()

virtual void Ipopt::NLPBoundsRemover::FinalizeSolution	(	SolverReturn	,
		const Vector &	,
		const Vector &	,
		const Vector &	,
		const Vector &	,
		const Vector &	,
		const Vector &	,
		const Vector &	,
		Number	,
		const IpoptData *	,
		IpoptCalculatedQuantities *
	)

virtual

This method is called at the very end of the optimization.

It provides the final iterate to the user, so that it can be stored as the solution. The status flag indicates the outcome of the optimization, where SolverReturn is defined in IpAlgTypes.hpp.

Reimplemented from Ipopt::NLP.

◆ IntermediateCallBack()

virtual bool Ipopt::NLPBoundsRemover::IntermediateCallBack	(	AlgorithmMode	,
		Index	,
		Number	,
		Number	,
		Number	,
		Number	,
		Number	,
		Number	,
		Number	,
		Number	,
		Index	,
		const IpoptData *	,
		IpoptCalculatedQuantities *
	)

inlinevirtual

This method is called once per iteration, after the iteration summary output has been printed.

It provides the current information to the user to do with it anything she wants. It also allows the user to ask for a premature termination of the optimization by returning false, in which case Ipopt will terminate with a corresponding return status. The basic information provided in the argument list has the quantities values printed in the iteration summary line. If more information is required, a user can obtain it from the IpData and IpCalculatedQuantities objects. However, note that the provided quantities are all for the problem that Ipopt sees, i.e., the quantities might be scaled, fixed variables might be sorted out, etc. The status indicates things like whether the algorithm is in the restoration phase... In the restoration phase, the dual variables are probably not not changing.

Reimplemented from Ipopt::NLP.

Definition at line 184 of file IpNLPBoundsRemover.hpp.

◆ GetScalingParameters()

virtual void Ipopt::NLPBoundsRemover::GetScalingParameters	(	const SmartPtr< const VectorSpace >	x_space,
		const SmartPtr< const VectorSpace >	c_space,
		const SmartPtr< const VectorSpace >	d_space,
		Number &	obj_scaling,
		SmartPtr< Vector > &	x_scaling,
		SmartPtr< Vector > &	c_scaling,
		SmartPtr< Vector > &	d_scaling
	)		const

virtual

Routines to get the scaling parameters.

Reimplemented from Ipopt::NLP.

◆ GetQuasiNewtonApproximationSpaces()

virtual void Ipopt::NLPBoundsRemover::GetQuasiNewtonApproximationSpaces	(	SmartPtr< VectorSpace > &	approx_space,
		SmartPtr< Matrix > &	P_approx
	)

inlinevirtual

Method for obtaining the subspace in which the limited-memory Hessian approximation should be done.

This is only called if the limited-memory Hessian approximation is chosen. Since the Hessian is zero in the space of all variables that appear in the problem functions only linearly, this allows the user to provide a VectorSpace for all nonlinear variables, and an ExpansionMatrix to lift from this VectorSpace to the VectorSpace of the primal variables x. If the returned values are NULL, it is assumed that the Hessian is to be approximated in the space of all x variables. The default instantiation of this method returns NULL, and a user only has to overwrite this method if the approximation is to be done only in a subspace.

Reimplemented from Ipopt::NLP.

Definition at line 218 of file IpNLPBoundsRemover.hpp.

◆ nlp()

SmartPtr< NLP > Ipopt::NLPBoundsRemover::nlp ( )

inline

Accessor method to the original NLP.

Definition at line 227 of file IpNLPBoundsRemover.hpp.

◆ operator=()

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

private

Default Assignment Operator.

Member Data Documentation

◆ nlp_

SmartPtr<NLP> Ipopt::NLPBoundsRemover::nlp_

private

Pointer to the original NLP.

Definition at line 257 of file IpNLPBoundsRemover.hpp.

◆ Px_l_orig_

SmartPtr<const Matrix> Ipopt::NLPBoundsRemover::Px_l_orig_

private

Pointer to the expansion matrix for the lower x bounds.

Definition at line 260 of file IpNLPBoundsRemover.hpp.

◆ Px_u_orig_

SmartPtr<const Matrix> Ipopt::NLPBoundsRemover::Px_u_orig_

private

Pointer to the expansion matrix for the upper x bounds.

Definition at line 263 of file IpNLPBoundsRemover.hpp.

◆ d_space_orig_

SmartPtr<const VectorSpace> Ipopt::NLPBoundsRemover::d_space_orig_

private

Pointer to the original d space.

Definition at line 266 of file IpNLPBoundsRemover.hpp.

◆ allow_twosided_inequalities_

bool Ipopt::NLPBoundsRemover::allow_twosided_inequalities_

private

Flag indicating whether twosided inequality constraints are allowed.

Definition at line 269 of file IpNLPBoundsRemover.hpp.

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

src/Algorithm/IpNLPBoundsRemover.hpp

Public Member Functions

Private Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ NLPBoundsRemover() [1/3]

◆ ~NLPBoundsRemover()

◆ NLPBoundsRemover() [2/3]

◆ NLPBoundsRemover() [3/3]

Member Function Documentation

◆ ProcessOptions()

◆ GetSpaces()

◆ GetBoundsInformation()

◆ GetStartingPoint()

◆ GetWarmStartIterate()

◆ Eval_f()

◆ Eval_grad_f()

◆ Eval_c()

◆ Eval_jac_c()

◆ Eval_d()

◆ Eval_jac_d()

◆ Eval_h()

◆ FinalizeSolution()

◆ IntermediateCallBack()

◆ GetScalingParameters()

◆ GetQuasiNewtonApproximationSpaces()

◆ nlp()

◆ operator=()

Member Data Documentation

◆ nlp_

◆ Px_l_orig_

◆ Px_u_orig_

◆ d_space_orig_

◆ allow_twosided_inequalities_