BCP_lp_relax Class Reference

An object of type BCP_lp_relax holds the description of an lp relaxation. More...

#include <BCP_matrix.hpp>

Inheritance diagram for BCP_lp_relax:

[legend]

Collaboration diagram for BCP_lp_relax:

[legend]

List of all members.

Public Member Functions
Query methods

size_t	colnum () const
	The number of columns.
size_t	rownum () const
	The number of rows.
const BCP_vec< double > &	Objective () const
	A const reference to the vector of objective coefficients.
const BCP_vec< double > &	ColLowerBound () const
	A const reference to the vector of lower bounds on the variables.
const BCP_vec< double > &	ColUpperBound () const
	A const reference to the vector of upper bounds on the variables.
const BCP_vec< double > &	RowLowerBound () const
	A const reference to the vector of lower bounds on the cuts.
const BCP_vec< double > &	RowUpperBound () const
	A const reference to the vector of upper bounds on the cuts.
Methods modifying the whole LP relaxation.

BCP_lp_relax &	operator= (const BCP_lp_relax &mat)
	Copy the content of `x` into the LP relaxation.
void	reserve (const int MaxColNum, const int MaxRowNum, const int MaxNonzeros)
	Reserve space in the LP relaxation for at least `MaxColNum` columns, `MaxRowNum` rows and `MaxNonzeros` nonzero entries.
void	clear ()
	Clear the LP relaxation.
void	copyOf (const CoinPackedMatrix &m, const double OBJ, const double CLB, const double CUB, const double RLB, const double *RUB)
	Set up the LP relaxation by making a copy of the arguments.
void	assign (CoinPackedMatrix &m, double &OBJ, double &CLB, double &CUB, double &RLB, double *&RUB)
	Set up the LP relaxation by taking over the pointers in the arguments.
Methods for expanding/shrinking the LP relaxation.

void	erase_col_set (const BCP_vec< int > &pos)
	Remove the columns whose indices are listed in `pos` from the LP relaxation.
void	erase_row_set (const BCP_vec< int > &pos)
	Remove the rows whose indices are listed in `pos` from the LP relaxation.
Packing/unpacking

void	pack (BCP_buffer &buf) const
	Pack the LP relaxation into the buffer.
void	unpack (BCP_buffer &buf)
	Unpack the LP relaxation from the buffer.
Constructors and destructor

	BCP_lp_relax (const bool colordered=true)
	Create an empty LP relaxation with given ordering.
	BCP_lp_relax (const BCP_lp_relax &mat)
	The copy constructor makes a copy of the argument LP relaxation.
	BCP_lp_relax (BCP_vec< BCP_row * > &rows, BCP_vec< double > &CLB, BCP_vec< double > &CUB, BCP_vec< double > &OBJ)
	Create a row major ordered LP relaxation by assigning the content of the arguments to the LP relaxation.
	BCP_lp_relax (BCP_vec< BCP_row * > &rows, BCP_vec< double > &CLB, BCP_vec< double > &CUB, BCP_vec< double > &OBJ, double extra_gap, double extra_major)
	Same as the previous method except that this method allows extra_gap and extra_major to be specified.
	BCP_lp_relax (BCP_vec< BCP_col * > &cols, BCP_vec< double > &RLB, BCP_vec< double > &RUB)
	Create a column major ordered LP relaxation by assigning the content of the arguments to the LP relaxation.
	BCP_lp_relax (BCP_vec< BCP_col * > &cols, BCP_vec< double > &RLB, BCP_vec< double > &RUB, double extra_gap, double extra_major)
	Same as the previous method except that this method allows extra_gap and extra_major to be specified.
	BCP_lp_relax (const bool colordered, const BCP_vec< int > &VB, const BCP_vec< int > &EI, const BCP_vec< double > &EV, const BCP_vec< double > &OBJ, const BCP_vec< double > &CLB, const BCP_vec< double > &CUB, const BCP_vec< double > &RLB, const BCP_vec< double > &RUB)
	Create an LP relaxation of the given ordering by copying the content of the arguments to the LP relaxation.
	BCP_lp_relax (const bool colordered, const int rownum, const int colnum, const int nznum, int &VB, int &EI, double &EV, double &OBJ, double &CLB, double &CUB, double &RLB, double &RUB)
	Create an LP relaxation of the given ordering by assigning the content of the arguments to the LP relaxation.
	~BCP_lp_relax ()
	The destructor deletes the data members.
Private Member Functions
helper functions for the constructors

void	BCP_createColumnOrderedMatrix (BCP_vec< BCP_row * > &rows, BCP_vec< double > &CLB, BCP_vec< double > &CUB, BCP_vec< double > &OBJ)
void	BCP_createRowOrderedMatrix (BCP_vec< BCP_col * > &cols, BCP_vec< double > &RLB, BCP_vec< double > &RUB)
Private Attributes
Data members

BCP_vec< double >	_Objective
	The objective coefficients of the variables.
BCP_vec< double >	_ColLowerBound
	The lower bounds on the variables.
BCP_vec< double >	_ColUpperBound
	The upper bounds on the variables.
BCP_vec< double >	_RowLowerBound
	The lower bounds on the cuts.
BCP_vec< double >	_RowUpperBound
	The upper bounds on the cuts.

Detailed Description

An object of type BCP_lp_relax holds the description of an lp relaxation.

The matrix, lower/upper bounds on the variables and cuts and objective coefficients for the variables.

Definition at line 267 of file BCP_matrix.hpp.

Constructor & Destructor Documentation

BCP_lp_relax::BCP_lp_relax ( const bool colordered = true ) [inline]

Create an empty LP relaxation with given ordering.

Definition at line 377 of file BCP_matrix.hpp.

BCP_lp_relax::BCP_lp_relax ( const BCP_lp_relax & mat )

The copy constructor makes a copy of the argument LP relaxation.

BCP_lp_relax::BCP_lp_relax	(	BCP_vec< BCP_row * > &	rows,
		BCP_vec< double > &	CLB,
		BCP_vec< double > &	CUB,
		BCP_vec< double > &	OBJ
	)

Create a row major ordered LP relaxation by assigning the content of the arguments to the LP relaxation.

When the constructor returns the content of 'rows' doesn't change while that of CLB, CUB and OBJ will be whatever the default constructor for those arguments create.

BCP_lp_relax::BCP_lp_relax	(	BCP_vec< BCP_row * > &	rows,
		BCP_vec< double > &	CLB,
		BCP_vec< double > &	CUB,
		BCP_vec< double > &	OBJ,
		double	extra_gap,
		double	extra_major
	)

Same as the previous method except that this method allows extra_gap and extra_major to be specified.

For the description of those see the documentation of CoinPackedMatrix. (extra_gap will be used for adding columns, while extra major for adding rows)

BCP_lp_relax::BCP_lp_relax	(	BCP_vec< BCP_col * > &	cols,
		BCP_vec< double > &	RLB,
		BCP_vec< double > &	RUB
	)

Create a column major ordered LP relaxation by assigning the content of the arguments to the LP relaxation.

When the constructor returns the content of 'cols' doesn't change while that of RLB and RUB will be whatever the default constructor for those arguments create.

BCP_lp_relax::BCP_lp_relax	(	BCP_vec< BCP_col * > &	cols,
		BCP_vec< double > &	RLB,
		BCP_vec< double > &	RUB,
		double	extra_gap,
		double	extra_major
	)

Same as the previous method except that this method allows extra_gap and extra_major to be specified.

For the description of those see the documentation of CoinPackedMatrix. (extra_gap will be used for adding rows, while extra major for adding columns)

BCP_lp_relax::BCP_lp_relax	(	const bool	colordered,
		const BCP_vec< int > &	VB,
		const BCP_vec< int > &	EI,
		const BCP_vec< double > &	EV,
		const BCP_vec< double > &	OBJ,
		const BCP_vec< double > &	CLB,
		const BCP_vec< double > &	CUB,
		const BCP_vec< double > &	RLB,
		const BCP_vec< double > &	RUB
	)

Create an LP relaxation of the given ordering by copying the content of the arguments to the LP relaxation.

BCP_lp_relax::BCP_lp_relax	(	const bool	colordered,
		const int	rownum,
		const int	colnum,
		const int	nznum,
		int *&	VB,
		int *&	EI,
		double *&	EV,
		double *&	OBJ,
		double *&	CLB,
		double *&	CUB,
		double *&	RLB,
		double *&	RUB
	)

Create an LP relaxation of the given ordering by assigning the content of the arguments to the LP relaxation.

When the constructor returns the content of the arguments will be NULL pointers.

BCP_lp_relax::~BCP_lp_relax ( ) [inline]

The destructor deletes the data members.

Definition at line 435 of file BCP_matrix.hpp.

Member Function Documentation

size_t BCP_lp_relax::colnum ( ) const [inline]

The number of columns.

Definition at line 288 of file BCP_matrix.hpp.

size_t BCP_lp_relax::rownum ( ) const [inline]

The number of rows.

Definition at line 290 of file BCP_matrix.hpp.

const BCP_vec<double>& BCP_lp_relax::Objective ( ) const [inline]

A const reference to the vector of objective coefficients.

Definition at line 292 of file BCP_matrix.hpp.

const BCP_vec<double>& BCP_lp_relax::ColLowerBound ( ) const [inline]

A const reference to the vector of lower bounds on the variables.

Definition at line 294 of file BCP_matrix.hpp.

const BCP_vec<double>& BCP_lp_relax::ColUpperBound ( ) const [inline]

A const reference to the vector of upper bounds on the variables.

Definition at line 296 of file BCP_matrix.hpp.

const BCP_vec<double>& BCP_lp_relax::RowLowerBound ( ) const [inline]

A const reference to the vector of lower bounds on the cuts.

Definition at line 298 of file BCP_matrix.hpp.

const BCP_vec<double>& BCP_lp_relax::RowUpperBound ( ) const [inline]

A const reference to the vector of upper bounds on the cuts.

Definition at line 300 of file BCP_matrix.hpp.

BCP_lp_relax& BCP_lp_relax::operator= ( const BCP_lp_relax & mat )

Copy the content of x into the LP relaxation.

Reimplemented from CoinPackedMatrix.

void BCP_lp_relax::reserve	(	const int	MaxColNum,
		const int	MaxRowNum,
		const int	MaxNonzeros
	)

Reserve space in the LP relaxation for at least MaxColNum columns, MaxRowNum rows and MaxNonzeros nonzero entries.

This is useful when columns/rows are added to the LP relaxation in small chunks and to avoid a series of reallocation we reserve sufficient space up front.

void BCP_lp_relax::clear ( )

Clear the LP relaxation.

Reimplemented from CoinPackedMatrix.

void BCP_lp_relax::copyOf	(	const CoinPackedMatrix &	m,
		const double *	OBJ,
		const double *	CLB,
		const double *	CUB,
		const double *	RLB,
		const double *	RUB
	)

Set up the LP relaxation by making a copy of the arguments.

void BCP_lp_relax::assign	(	CoinPackedMatrix &	m,
		double *&	OBJ,
		double *&	CLB,
		double *&	CUB,
		double *&	RLB,
		double *&	RUB
	)

Set up the LP relaxation by taking over the pointers in the arguments.

void BCP_lp_relax::erase_col_set ( const BCP_vec< int > & pos )

Remove the columns whose indices are listed in pos from the LP relaxation.

void BCP_lp_relax::erase_row_set ( const BCP_vec< int > & pos )

Remove the rows whose indices are listed in pos from the LP relaxation.

void BCP_lp_relax::pack ( BCP_buffer & buf ) const

Pack the LP relaxation into the buffer.

void BCP_lp_relax::unpack ( BCP_buffer & buf )

Unpack the LP relaxation from the buffer.

void BCP_lp_relax::BCP_createColumnOrderedMatrix	(	BCP_vec< BCP_row * > &	rows,
		BCP_vec< double > &	CLB,
		BCP_vec< double > &	CUB,
		BCP_vec< double > &	OBJ
	)			`[private]`