#include <Model.hpp>

Collaboration diagram for ipo::Model::Parameters:

Public Member Functions
	Parameters ()
	Constructor: sets default values. More...

double	getMu () const
	Get value of mu. More...

double	getEpsilon () const
	Get value of epsilon. More...

std::ostream *	getOutputStream () const
	Get output stream. More...

void	setMu (double const mu)
	Set value of mu. More...

void	setEpsilon (double const epsilon)
	Set value of epsilon. More...

void	setOutputStream (std::ostream *outputStream)
	Set value of output stream. More...

Private Attributes
double	mu
	A value used for scaling: the default is 10. More...

double	epsilon
	A limit used for convergence. More...

std::ostream *	outputStream
	A stream (default is nullptr) for sending output to. More...

Detailed Description

Definition at line 328 of file Model.hpp.

Constructor & Destructor Documentation

Model::Parameters::Parameters ( )

Constructor: sets default values.

Definition at line 313 of file Model.cc.

Member Function Documentation

double ipo::Model::Parameters::getEpsilon ( ) const

inline

Get value of epsilon.

Returns: value of epsilon

Definition at line 343 of file Model.hpp.

References epsilon.

Referenced by ipo::Model::stoppingCriterion().

double ipo::Model::Parameters::getMu ( ) const

inline

Get value of mu.

Returns: value of mu

Definition at line 338 of file Model.hpp.

References mu.

Referenced by ipo::Model::optimise().

std::ostream* ipo::Model::Parameters::getOutputStream ( ) const

inline

Get output stream.

Returns: output stream

Definition at line 348 of file Model.hpp.

References outputStream.

Referenced by ipo::Model::optimise().

void Model::Parameters::setEpsilon ( double const epsilon )

Set value of epsilon.

Parameters

epsilon The new value

Definition at line 323 of file Model.cc.

void Model::Parameters::setMu ( double const mu )

Set value of mu.

Parameters

mu	The new value

Definition at line 320 of file Model.cc.

void Model::Parameters::setOutputStream ( std::ostream * outputStream )

Set value of output stream.

Parameters

outputStream The new value

Definition at line 327 of file Model.cc.

Member Data Documentation

double ipo::Model::Parameters::epsilon

private

A limit used for convergence.

The default is square root of machine epsilon. The optimisation improves an estimate of the objective function until it is within epsilon of the optimum.

Note that convergence also depends on (1) the problem being well formed, and (2) how well the NewtonGradient and LineSearch algorithms used internally perform.

Definition at line 386 of file Model.hpp.

Referenced by getEpsilon().

double ipo::Model::Parameters::mu

private

A value used for scaling: the default is 10.

The value must exceed 1. The interior point optimisation uses a logarithmic barrier in which the natural logarithm of the negative of each constraint function is scaled by a value $1/t$ . The barrier function approximates an indicator function, which is the limit as $t\to\infty$ of the barrier function. The optimisation algorithm periodically increases $t$ and mu is use to scale $t$ from a smaller to a larger value: $t\to\mu t$ .