amplObj Class Reference

Class for an AMPL-objective-function. More...

#include <ampl.h>

Inheritance diagram for amplObj:

[legend]Collaboration diagram for amplObj:


Public Member Functions
	amplObj (int dim_=0, Pointer< SparsityInfo > sparsity_=NULL, Pointer< ostream > out_func_p_=out_out_p, Pointer< ostream > out_func_log_p_=out_log_p)
	(Standard-)Constructor with optional the dimension.
double	eval (const UserVector< double > &x) const
	Evaluates the objective for a UserVector<double>.
void	grad (dvector &g, const dvector &x) const
	Computes the gradient of the objective for a dvector.
void	grad (UserVector< double > &g, const UserVector< double > &x) const
	Computes the (sub)gradient for a UserVector<double>.
void	HessMult (dvector &y, const dvector &x, const dvector &z) const
	Computes the product of a Hessian and a dvector.
void	HessMult (UserVector< double > &y, const UserVector< double > &x, const UserVector< double > &z) const
	Evaluates the product of the Hessian and a UserVector<double>.
void	set_curvature (CurvatureType ct)
CurvatureType	get_curvature () const
void	print (ostream &out) const
	Prints some information about this function.
Private Attributes
Func::CurvatureType	curv_type

Detailed Description

Class for an AMPL-objective-function.

This class is used to compute a nonlinear objective of a problem, which was described by an ampl-file.

Definition at line 66 of file ampl.h.

Constructor & Destructor Documentation

amplObj::amplObj	(	int	dim_ = `0`,
		Pointer< SparsityInfo >	sparsity_ = `NULL`,
		Pointer< ostream >	out_func_p_ = `out_out_p`,
		Pointer< ostream >	out_func_log_p_ = `out_log_p`
	)			`[inline]`

(Standard-)Constructor with optional the dimension.

Parameters:

	dim_	The dimension of the function.
	out_func_p_	An ostream to print function-related output to, default is out_out_p.
	out_func_log_p_	An ostream to print function-related logging-output to, default is out_func_p.

Definition at line 75 of file ampl.h.

References Func::sparsity.

Member Function Documentation

double amplObj::eval ( const UserVector< double > & x ) const [inline, virtual]

Evaluates the objective for a UserVector<double>.

Calls objval(...) from the AMPL/Solver Interface Library (ASL).

Parameters:

The UserVector<double> to evaluate.

Returns:: The value of the objective.

Implements Func.

Definition at line 85 of file ampl.h.

References asl, UserVector< Type >::dim(), and out_log.

Referenced by HessMult().

void amplObj::grad	(	dvector &	g,
		const dvector &	x
	)			const `[inline]`

Computes the gradient of the objective for a dvector.

Calls objgrd(...) from the AMPL/Solver Interface Library (ASL).

Parameters:

	g	The dvector to store the result in.
	x	The dvector to compute the gradient for.

Definition at line 103 of file ampl.h.

References asl, DenseVector< Type >::dim(), and out_log.

void amplObj::grad	(	UserVector< double > &	g,
		const UserVector< double > &	x
	)			const `[inline, virtual]`

Computes the (sub)gradient for a UserVector<double>.

Abstract.

Parameters:

	g	The UserVector<double> to store the result in.
	x	The UserVector<double> to compute the gradient for.

See also:

grad(const dvector&)

valgrad(double&, UserVector<double>&, const UserVector<double>&)

add_grad(UserVector<double>&, UserVector<double>&)

add_valgrad(double&, UserVector<double>&, UserVector<double>&)

Implements Func.

Definition at line 116 of file ampl.h.

References asl, UserVector< Type >::dim(), out_log, UserVector< Type >::set(), and UserVector< Type >::size().

void amplObj::HessMult	(	dvector &	y,
		const dvector &	x,
		const dvector &	z
	)			const `[inline]`

Computes the product of a Hessian and a dvector.

Parameters:

	y	The dvector to store the result in.
	x	The dvector to compute the Hessian for.
	z	The dvector to multiply with the Hessian.

Definition at line 139 of file ampl.h.

References eval().

void amplObj::HessMult	(	UserVector< double > &	y,
		const UserVector< double > &	x,
		const UserVector< double > &	z
	)			const `[inline, virtual]`

Evaluates the product of the Hessian and a UserVector<double>.

Abstract.

Parameters:

	y	The UserVector<double> to store the result in: $y = \nabla^2 f(x) \cdot z$ .
	x	The UserVector<double> to evaluate the Hessian for.
	z	The UserVector<double> to multiply with the Hessian.