abstract base class for linear operators More...

#include <flexLinearOperator.h>

Inheritance diagram for flexLinearOperator< T >:

Public Member Functions
	flexLinearOperator (int aNumRows, int aNumCols, linOp aType, bool aIsMinus)
	initializes the linear operator More...

int	getNumCols () const
	returns number of columns of the linear operator More...

int	getNumRows () const
	returns number of rows of the linear operator More...

void	setNumCols (int aNumCols)
	sets the number of columns of the linear operator More...

void	setNumRows (int aNumRows)
	sets the number of rows of the linear operator More...

void	setMinus (bool aIsMinus)
	constrols if operator should be negated or not More...

virtual flexLinearOperator< T > *	copy ()=0
	copies the linear operator More...

virtual void	times (bool transposed, const Tdata &input, Tdata &output)=0
	applies linear operator on vector More...

virtual void	timesPlus (bool transposed, const Tdata &input, Tdata &output)=0
	applies linear operator on vector and adds its result to y More...

virtual void	timesMinus (bool transposed, const Tdata &input, Tdata &output)=0
	applies linear operator on vector and substracts its result from y More...

virtual std::vector< T >	getAbsRowSum (bool transposed)=0
	returns a vector of sum of absolute values per row used for preconditioning More...

virtual thrust::device_vector< T >	getAbsRowSumCUDA (bool transposed)=0
	same function as getAbsRowSum() but implemented in CUDA More...

virtual T	getMaxRowSumAbs (bool transposed)=0
	returns the maximum sum of absolute values per row used for preconditioning More...

Public Attributes
linOp	type
	type of linear operator More...

bool	isMinus
	determines if operator is negated

Detailed Description

template<typename T>
class flexLinearOperator< T >

abstract base class for linear operators

flexLinearOperator combines the interface for all usable operators

Constructor & Destructor Documentation

◆ flexLinearOperator()

template<typename T>

flexLinearOperator< T >::flexLinearOperator	(	int	aNumRows,
		int	aNumCols,
		linOp	aType,
		bool	aIsMinus
	)

inline

initializes the linear operator

Parameters

aNumRows	number of rows of liner operator
aNumCols	number of columns of liner operator
aType	type of linear operator sa type

See also: type

Parameters

aIsMinus determines if operator is negated

See also: isMinus

Member Function Documentation

◆ copy()

template<typename T>

virtual flexLinearOperator<T>* flexLinearOperator< T >::copy ( )

pure virtual

copies the linear operator

Returns: copy of linear operator

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexDiagonalOperator< T >, flexConcatOperator< T >, flexSuperpixelOperator< T >, flexMatrix< T >, flexMatrixLogical< T >, flexFullMatrix< T >, flexIdentityOperator< T >, and flexZeroOperator< T >.

◆ getAbsRowSum()

template<typename T>

virtual std::vector<T> flexLinearOperator< T >::getAbsRowSum ( bool transposed )

pure virtual

returns a vector of sum of absolute values per row used for preconditioning

Parameters

transposed is true if operator should be (temporarily) transposed before usage

Returns: vector of sum of absolute values per row

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexConcatOperator< T >, flexMatrix< T >, flexMatrixLogical< T >, flexDiagonalOperator< T >, flexIdentityOperator< T >, flexFullMatrix< T >, flexSuperpixelOperator< T >, and flexZeroOperator< T >.

◆ getAbsRowSumCUDA()

template<typename T>

virtual thrust::device_vector<T> flexLinearOperator< T >::getAbsRowSumCUDA ( bool transposed )

pure virtual

same function as getAbsRowSum() but implemented in CUDA

Parameters

transposed is true if operator should be (temporarily) transposed before usage

Returns: vector of sum of absolute values per row

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexConcatOperator< T >, flexMatrix< T >, flexMatrixLogical< T >, flexFullMatrix< T >, flexDiagonalOperator< T >, flexIdentityOperator< T >, flexSuperpixelOperator< T >, and flexZeroOperator< T >.

◆ getMaxRowSumAbs()

template<typename T>

virtual T flexLinearOperator< T >::getMaxRowSumAbs ( bool transposed )

pure virtual

returns the maximum sum of absolute values per row used for preconditioning

Parameters

transposed is true if operator should be (temporarily) transposed before usage

Returns: maximum sum of absolute values per row

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexConcatOperator< T >, flexMatrix< T >, flexDiagonalOperator< T >, flexMatrixLogical< T >, flexIdentityOperator< T >, flexFullMatrix< T >, flexSuperpixelOperator< T >, and flexZeroOperator< T >.

◆ getNumCols()

template<typename T>

int flexLinearOperator< T >::getNumCols ( ) const

inline

returns number of columns of the linear operator

Returns: number of columns

◆ getNumRows()

template<typename T>

int flexLinearOperator< T >::getNumRows ( ) const

inline

returns number of rows of the linear operator

Returns: number of rows

◆ setMinus()

template<typename T>

void flexLinearOperator< T >::setMinus ( bool aIsMinus )

inline

constrols if operator should be negated or not

Parameters

aIsMinus true if operator should be negated, otherwise false.

◆ setNumCols()

template<typename T>

void flexLinearOperator< T >::setNumCols ( int aNumCols )

inline

sets the number of columns of the linear operator

Parameters

aNumCols number of columns

◆ setNumRows()

template<typename T>

void flexLinearOperator< T >::setNumRows ( int aNumRows )

inline

sets the number of rows of the linear operator

Parameters

aNumRows number of rows

◆ times()

template<typename T>

virtual void flexLinearOperator< T >::times	(	bool	transposed,
		const Tdata &	input,
		Tdata &	output
	)

pure virtual

applies linear operator on vector

equals $ y = Ax $

Parameters

transposed	is true if operator should be (temporarily) transposed before usage
input	data to be processed
output	output data

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexDiagonalOperator< T >, flexConcatOperator< T >, flexMatrix< T >, flexMatrixLogical< T >, flexSuperpixelOperator< T >, flexFullMatrix< T >, flexIdentityOperator< T >, and flexZeroOperator< T >.

◆ timesMinus()

template<typename T>

virtual void flexLinearOperator< T >::timesMinus	(	bool	transposed,
		const Tdata &	input,
		Tdata &	output
	)

pure virtual

applies linear operator on vector and substracts its result from y

equals $ y = y - Ax $

Parameters

transposed	is true if operator should be (temporarily) transposed before usage
input	data to be processed
output	output data

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexConcatOperator< T >, flexDiagonalOperator< T >, flexIdentityOperator< T >, flexMatrix< T >, flexMatrixLogical< T >, flexSuperpixelOperator< T >, flexFullMatrix< T >, and flexZeroOperator< T >.

◆ timesPlus()

template<typename T>

virtual void flexLinearOperator< T >::timesPlus	(	bool	transposed,
		const Tdata &	input,
		Tdata &	output
	)

pure virtual

applies linear operator on vector and adds its result to y

equals $ y = y + Ax $

Parameters

transposed	is true if operator should be (temporarily) transposed before usage
input	data to be processed
output	output data

Implemented in flexGradientOperator< T >, flexMatrixGPU< T >, flexDiagonalOperator< T >, flexIdentityOperator< T >, flexConcatOperator< T >, flexMatrix< T >, flexMatrixLogical< T >, flexSuperpixelOperator< T >, flexFullMatrix< T >, and flexZeroOperator< T >.

Member Data Documentation

◆ type

template<typename T>

linOp flexLinearOperator< T >::type

type of linear operator

See also: linOp

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

flexLinearOperator.h

Public Member Functions

Public Attributes

Detailed Description

template<typename T> class flexLinearOperator< T >

Constructor & Destructor Documentation

◆ flexLinearOperator()

Member Function Documentation

◆ copy()

◆ getAbsRowSum()

◆ getAbsRowSumCUDA()

◆ getMaxRowSumAbs()

◆ getNumCols()

◆ getNumRows()

◆ setMinus()

◆ setNumCols()

◆ setNumRows()

◆ times()

◆ timesMinus()

◆ timesPlus()

Member Data Documentation

◆ type

template<typename T>
class flexLinearOperator< T >