Ipopt::CGPenaltyCq Class Reference

Class for all Chen-Goldfarb penalty method specific calculated quantities. More...

#include <IpCGPenaltyCq.hpp>

Inheritance diagram for Ipopt::CGPenaltyCq:

[legend]

Collaboration diagram for Ipopt::CGPenaltyCq:

[legend]

List of all members.

Public Member Functions
bool	Initialize (const Journalist &jnlst, const OptionsList &options, const std::string &prefix)
	This method must be called to initialize the global algorithmic parameters.
Constructors/Destructors

	CGPenaltyCq (IpoptNLP ip_nlp, IpoptData ip_data, IpoptCalculatedQuantities *ip_cg)
	Constructor.
virtual	~CGPenaltyCq ()
	Default destructor.
Methods for the Chen-Goldfarb line search

Number	curr_jac_cd_norm (Index nrm_type)
	Compute \|\|delta_c, delta_d\|\|_infty.
Number	curr_scaled_y_Amax ()
	Compute gradient scaling based y->Amax.
Number	curr_added_y_nrm2 ()
	Compute the 2-norm of y plus delta y.
Number	curr_penalty_function ()
	Method for the penalty function at current point.
Number	trial_penalty_function ()
	Method for the penalty function at trial point.
Number	curr_direct_deriv_penalty_function ()
	Method for the directional derivative of the penalty function at current point with current step in delta.
Number	curr_fast_direct_deriv_penalty_function ()
	Method for the directional derivative of the penalty function at current point with current "fast" step in delta_cgpen.
Number	dT_times_barH_times_d ()
	Quality of d^T Aug(H) d.
Number	curr_cg_pert_fact ()
	Method for the current value for the perturbation factor for the Chen-Goldfarb method.
Number	compute_curr_cg_penalty (const Number)
	Method for choose line search penalty parameter.
Number	compute_curr_cg_penalty_scale ()
	Method for choose penalty parameters for scaling the KKT system.
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.
	CGPenaltyCq ()
	Default Constructor.
	CGPenaltyCq (const CGPenaltyCq &)
	Copy Constructor.
void	operator= (const CGPenaltyCq &)
	Overloaded Equals Operator.
Private Attributes
bool	initialize_called_
	flag indicating if Initialize method has been called (for debugging)
Pointers for easy access to data and NLP information. To
avoid circular references of Smart Pointers, we use a regular pointer here.
IpoptNLP *	ip_nlp_
IpoptData *	ip_data_
IpoptCalculatedQuantities *	ip_cq_
Caches for the Chen-Goldfarb line search

CachedResults< Number >	curr_fast_direct_deriv_penalty_function_cache_
	Cache for the penalty function at current point.
CachedResults< Number >	curr_jac_cd_norm_cache_
	Cache for the penalty function at current point.
CachedResults< Number >	curr_scaled_y_Amax_cache_
	Cache for the penalty function at current point.
CachedResults< Number >	curr_added_y_nrm2_cache_
	Cache for the penalty function at current point.
CachedResults< Number >	curr_penalty_function_cache_
	Cache for the penalty function at current point.
CachedResults< Number >	trial_penalty_function_cache_
	Cache for the penalty function at trial point.
CachedResults< Number >	curr_direct_deriv_penalty_function_cache_
	Cache for the directional derivative of the penalty function at current point with step in delta.
CachedResults< Number >	curr_cg_pert_fact_cache_
	Cache for Chen-Goldfarb perturbation factor.

Number	reference_infeasibility_
	Parameters for penalty method.
static void	RegisterOptions (const SmartPtr< RegisteredOptions > &roptions)
	Methods for IpoptType.

Detailed Description

Class for all Chen-Goldfarb penalty method specific calculated quantities.

Definition at line 21 of file IpCGPenaltyCq.hpp.

Constructor & Destructor Documentation

Ipopt::CGPenaltyCq::CGPenaltyCq	(	IpoptNLP *	ip_nlp,
		IpoptData *	ip_data,
		IpoptCalculatedQuantities *	ip_cg
	)

Constructor.

virtual Ipopt::CGPenaltyCq::~CGPenaltyCq ( ) [virtual]

Default destructor.

Ipopt::CGPenaltyCq::CGPenaltyCq ( ) [private]

Default Constructor.

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

Copy Constructor.

Member Function Documentation

bool Ipopt::CGPenaltyCq::Initialize	(	const Journalist &	jnlst,
		const OptionsList &	options,
		const std::string &	prefix
	)

This method must be called to initialize the global algorithmic parameters.

The parameters are taken from the OptionsList object.

Number Ipopt::CGPenaltyCq::curr_jac_cd_norm ( Index nrm_type )

Compute ||delta_c, delta_d||_infty.

Number Ipopt::CGPenaltyCq::curr_scaled_y_Amax ( )

Compute gradient scaling based y->Amax.

Number Ipopt::CGPenaltyCq::curr_added_y_nrm2 ( )

Compute the 2-norm of y plus delta y.

Number Ipopt::CGPenaltyCq::curr_penalty_function ( )

Method for the penalty function at current point.

Number Ipopt::CGPenaltyCq::trial_penalty_function ( )

Method for the penalty function at trial point.

Number Ipopt::CGPenaltyCq::curr_direct_deriv_penalty_function ( )

Method for the directional derivative of the penalty function at current point with current step in delta.

Number Ipopt::CGPenaltyCq::curr_fast_direct_deriv_penalty_function ( )

Method for the directional derivative of the penalty function at current point with current "fast" step in delta_cgpen.

Number Ipopt::CGPenaltyCq::dT_times_barH_times_d ( )

Quality of d^T Aug(H) d.

Number Ipopt::CGPenaltyCq::curr_cg_pert_fact ( )

Method for the current value for the perturbation factor for the Chen-Goldfarb method.

The factor is computed as 2-norm of the constraints devided by the current penbalty parameter

Number Ipopt::CGPenaltyCq::compute_curr_cg_penalty ( const Number )

Method for choose line search penalty parameter.

Number Ipopt::CGPenaltyCq::compute_curr_cg_penalty_scale ( )

Method for choose penalty parameters for scaling the KKT system.

static void Ipopt::CGPenaltyCq::RegisterOptions ( const SmartPtr< RegisteredOptions > & roptions ) [static]

Methods for IpoptType.

void Ipopt::CGPenaltyCq::operator= ( const CGPenaltyCq & ) [private]

Overloaded Equals Operator.

Member Data Documentation

IpoptNLP* Ipopt::CGPenaltyCq::ip_nlp_ [private]

Definition at line 108 of file IpCGPenaltyCq.hpp.

IpoptData* Ipopt::CGPenaltyCq::ip_data_ [private]

Definition at line 109 of file IpCGPenaltyCq.hpp.

IpoptCalculatedQuantities* Ipopt::CGPenaltyCq::ip_cq_ [private]

Definition at line 110 of file IpCGPenaltyCq.hpp.

CachedResults<Number> Ipopt::CGPenaltyCq::curr_fast_direct_deriv_penalty_function_cache_ [private]