MatrixSparse Class Reference

#include <MatrixSparse.hpp>

Inheritance diagram for MatrixSparse:

Detailed Description

Sparse Matrix

Handle a sparse matrix that can be symmetrical, square or not. Storage relies either on Eigen3 Library (see opt_eigen flag) or cs code. Default storage option can be set globally by using setGlobalFlagEigen

Public Member Functions
	MatrixSparse (int nrow=0, int ncol=0, int opt_eigen=-1)
	MatrixSparse (const cs *A)
	MatrixSparse (const MatrixSparse &m)
MatrixSparse &	operator= (const MatrixSparse &m)
virtual	~MatrixSparse ()
bool	isFlagEigen () const
	Cloneable interface. More...
bool	isSparse () const override
	Interface for AMatrix. More...
bool	isDense () const override
void	setValue (int irow, int icol, double value, bool flagCheck=true) override
double	getValue (int row, int col, bool flagCheck=true) const override
void	updValue (int irow, int icol, const EOperator &oper, double value, bool flagCheck=true) override
virtual void	setColumn (int icol, const VectorDouble &tab) override
virtual void	setRow (int irow, const VectorDouble &tab) override
virtual void	setDiagonal (const VectorDouble &tab) override
virtual void	setDiagonalToConstant (double value=1.) override
virtual MatrixSparse *	transpose () const override
virtual void	addScalar (double v) override
virtual void	addScalarDiag (double v) override
virtual void	prodScalar (double v) override
virtual void	fill (double value) override
virtual void	multiplyRow (const VectorDouble &vec) override
virtual void	multiplyColumn (const VectorDouble &vec) override
virtual void	divideRow (const VectorDouble &vec) override
virtual void	divideColumn (const VectorDouble &vec) override
virtual VectorDouble	prodVecMat (const VectorDouble &x, bool transpose=false) const override
virtual VectorDouble	prodMatVec (const VectorDouble &x, bool transpose=false) const override
virtual void	prodMatMatInPlace (const AMatrix *x, const AMatrix *y, bool transposeX=false, bool transposeY=false) override
virtual NF_Triplet	getMatrixToTriplet (int shiftRow=0, int shiftCol=0) const override
virtual String	toString (const AStringFormat *strfmt=nullptr) const override
	Interface to AStringable. More...
virtual void	addMatInPlace (const MatrixSparse &y, double cx=1., double cy=1.)
	The next functions use specific definition of matrix (to avoid dynamic_cast) rather than manipulating AMatrix. They are no more generic of AMatrix. More...
virtual void	prodNormMatMatInPlace (const MatrixSparse &a, const MatrixSparse &m, bool transpose=false)
virtual void	prodNormMatInPlace (const MatrixSparse &a, const VectorDouble &vec=VectorDouble(), bool transpose=false)
const cs *	getCS () const
void	setCS (cs *cs)
void	freeCS ()
cs *	getCSUnprotected () const
virtual void	resetFromValue (int nrows, int ncols, double value) override
	Reset the matrix to new dimensions and fill with a new value. More...
virtual void	resetFromArray (int nrows, int ncols, const double *tab, bool byCol=true) override
	Reset the matrix from an array of double values. More...
virtual void	resetFromVD (int nrows, int ncols, const VectorDouble &tab, bool byCol=true) override
	Reset the matrix from a vector of double values. More...
virtual void	resetFromVVD (const VectorVectorDouble &tab, bool byCol=true) override
	Reset the matrix from an array of double values. More...
void	resetFromTriplet (const NF_Triplet &NF_T)
void	dumpElements (const String &title, int ifrom, int ito) const
void	fillRandom (int seed=432432, double zeroPercent=0.1)
int	computeCholesky ()
int	solveCholesky (const VectorDouble &b, VectorDouble &x)
int	simulateCholesky (const VectorDouble &b, VectorDouble &x)
double	computeCholeskyLogDeterminant ()
void	addValue (int row, int col, double value)
double	L1Norm () const
void	getStats (int *nrows, int *ncols, int *count, double *percent) const
int	scaleByDiag ()
int	addVecInPlace (const VectorDouble &x, VectorDouble &y)
void	setConstant (double value)
VectorDouble	extractDiag (int oper_choice=1) const
void	prodNormDiagVecInPlace (const VectorDouble &vec, int oper=1)
const Eigen::SparseMatrix< double > &	getEigenMatrix () const
void	setEigenMatrix (const Eigen::SparseMatrix< double > &eigenMatrix)
MatrixSparse *	extractSubmatrixByRanks (const VectorInt &rank_rows, const VectorInt &rank_cols)
MatrixSparse *	extractSubmatrixByColor (const VectorInt &colors, int ref_color, bool row_ok, bool col_ok)
VectorInt	colorCoding ()
int	getNonZeros () const
Public Member Functions inherited from AMatrix
	AMatrix (int nrow=0, int ncol=0)
	AMatrix (const AMatrix &m)
AMatrix &	operator= (const AMatrix &m)
virtual	~AMatrix ()
virtual void	reset (int nrows, int ncols)
virtual bool	isSquare (bool printWhyNot=false) const
virtual bool	isValid (int irow, int icol, bool printWhyNot=false) const
virtual bool	isIdentity (bool printWhyNot=false) const
virtual bool	isSymmetric (bool printWhyNot=false, double eps=EPSILON10) const
virtual bool	mustBeSymmetric () const
virtual void	transposeInPlace ()
virtual VectorDouble	getRow (int irow) const
virtual VectorDouble	getColumn (int icol) const
void	addMatInPlace (const AMatrix &y, double cx=1., double cy=1.)
void	prodMatInPlace (const AMatrix *matY, bool transposeY=false)
void	prodNormMatMatInPlace (const AMatrix &a, const AMatrix &m, bool transpose=false)
void	prodNormMatInPlace (const AMatrix &a, const VectorDouble &vec=VectorDouble(), bool transpose=false)
void	resize (int nrows, int ncols)
	Resize the matrix to new dimensions (this method doesn't change the storage type) More...
void	addValue (int irow, int icol, double value)
bool	isSame (const AMatrix &m, double eps=EPSILON4, bool printWhyNot=false)
bool	isSameSize (const AMatrix &m) const
bool	empty () const
double	compare (const AMatrix &mat) const
int	getNRows () const
int	getNCols () const
int	size () const
VectorDouble	getValues (bool byCol=true) const
VectorDouble	getDiagonal (int shift=0) const
bool	isColumnDefined (int icol) const
bool	isRowDefined (int irow) const
int	getNumberColumnDefined () const
int	getNumberRowDefined () const
bool	isNonNegative (bool verbose=false)
void	prodMatVecInPlace (const VectorDouble &x, VectorDouble &y, bool transpose=false) const
void	prodMatVecInPlacePtr (const double *x, double *y, bool transpose=false) const
void	prodVecMatInPlace (const VectorDouble &x, VectorDouble &y, bool transpose=false) const
void	prodVecMatInPlacePtr (const double *x, double *y, bool transpose=false) const
double	quadraticMatrix (const VectorDouble &x, const VectorDouble &y)
int	invert ()
int	solve (const VectorDouble &b, VectorDouble &x) const
void	dumpElements (const String &title, int ifrom, int ito) const
void	setIdentity (double value=1.)
void	fillRandom (int seed=432432, double zeroPercent=0.1)
void	setValues (const VectorDouble &values, bool byCol=true)
double	getMeanByColumn (int icol) const
double	getMinimum () const
double	getMaximum () const
double	getNormInf () const
void	copyReduce (const AMatrix *x, const VectorInt &activeRows, const VectorInt &activeCols)
void	copyElements (const AMatrix &m, double factor=1.)
void	setFlagCheckAddress (bool flagCheckAddress)
void	makePositiveColumn ()
void	linearCombination (double val1, const AMatrix *mat1, double val2=1., const AMatrix *mat2=nullptr)
double	operator() (int row, int col) const
double &	operator() (int row, int col)
Public Member Functions inherited from AStringable
	AStringable ()
	AStringable (const AStringable &r)
AStringable &	operator= (const AStringable &r)
virtual	~AStringable ()
virtual void	display (const AStringFormat *strfmt=nullptr) const final
virtual void	display (int level) const final
Public Member Functions inherited from ICloneable
	ICloneable ()
virtual	~ICloneable ()
virtual ICloneable *	clone () const =0

Static Public Member Functions
static MatrixSparse *	createFromTriplet (const NF_Triplet &NF_T, int nrow=0, int ncol=0, int opt_eigen=-1)
static MatrixSparse *	addMatMat (const MatrixSparse *x, const MatrixSparse *y, double cx=1., double cy=1.)
static MatrixSparse *	diagVec (const VectorDouble &vec, int opt_eigen=-1)
static MatrixSparse *	diagConstant (int number, double value=1., int opt_eigen=-1)
static MatrixSparse *	diagMat (MatrixSparse *A, int oper_choice, int opt_eigen=-1)
static MatrixSparse *	glue (const MatrixSparse *A1, const MatrixSparse *A2, bool flagShiftRow, bool flagShiftCol)

Public Attributes
	DECLARE_TOTL
	Has a specific implementation in the Target language. More...

Constructor & Destructor Documentation

MatrixSparse() [1/3]

MatrixSparse::MatrixSparse	(	int	nrow = 0,
		int	ncol = 0,
		int	opt_eigen = -1
	)

MatrixSparse() [2/3]

MatrixSparse::MatrixSparse

(

const cs *

A

)

MatrixSparse() [3/3]

MatrixSparse::MatrixSparse

(

const MatrixSparse &

m

)

~MatrixSparse()

MatrixSparse::~MatrixSparse ( )

virtual

Member Function Documentation

addMatInPlace()

void MatrixSparse::addMatInPlace ( const MatrixSparse &  y, double  cx = 1., double  cy = 1.  )

virtual

The next functions use specific definition of matrix (to avoid dynamic_cast) rather than manipulating AMatrix. They are no more generic of AMatrix.

Add a matrix (multiplied by a constant)

Updates the current Matrix as a linear combination of matrices as follows: this <- cx * this + cy * y

Parameters

cx	Coefficient applied to the current Matrix
cy	Coefficient applied to the Matrix 'y'
y	Second Matrix in the Linear combination

addMatMat()

MatrixSparse * MatrixSparse::addMatMat ( const MatrixSparse *  x, const MatrixSparse *  y, double  cx = 1., double  cy = 1.  )

static

addScalar()

void MatrixSparse::addScalar ( double  v )

overridevirtual

Add a value to each matrix component

Parameters

v	Add a scalar value to all terms of the current matrix

Reimplemented from AMatrix.

addScalarDiag()

void MatrixSparse::addScalarDiag ( double  v )

overridevirtual

Add value to matrix diagonal

Parameters

v	Add constant value to the diagonal of the current Matrix

Reimplemented from AMatrix.

addValue()

void MatrixSparse::addValue	(	int	row,
		int	col,
		double	value
	)

addVecInPlace()

int MatrixSparse::addVecInPlace	(	const VectorDouble &	x,
		VectorDouble &	y
	)

colorCoding()

VectorInt MatrixSparse::colorCoding

(

)

computeCholesky()

int MatrixSparse::computeCholesky

(

)

computeCholeskyLogDeterminant()

double MatrixSparse::computeCholeskyLogDeterminant

(

)

createFromTriplet()

MatrixSparse * MatrixSparse::createFromTriplet ( const NF_Triplet &  NF_T, int  nrow = 0, int  ncol = 0, int  opt_eigen = -1  )

static

diagConstant()

MatrixSparse * MatrixSparse::diagConstant ( int  number, double  value = 1., int  opt_eigen = -1  )

static

diagMat()

MatrixSparse * MatrixSparse::diagMat ( MatrixSparse *  A, int  oper_choice, int  opt_eigen = -1  )

static

Construct a sparse matrix with the diagonal of 'A', where each element is transformed

Parameters

A	Input sparse matrix
oper_choice	Operation on the diagonal term (see Utilities::operate_XXX)
opt_eigen	Option for choosing Eigen Library or not

Returns

diagVec()

MatrixSparse * MatrixSparse::diagVec ( const VectorDouble &  vec, int  opt_eigen = -1  )

static

divideColumn()

void MatrixSparse::divideColumn ( const VectorDouble &  vec )

overridevirtual

Divide the matrix column-wise

Divide a Matrix column-wise

Reimplemented from AMatrix.

divideRow()

void MatrixSparse::divideRow ( const VectorDouble &  vec )

overridevirtual

Divide the matrix row-wise

Divide a Matrix row-wise

Reimplemented from AMatrix.

dumpElements()

void MatrixSparse::dumpElements	(	const String &	title,
		int	ifrom,
		int	ito
	)		const

Dump a specific range of samples from the internal storage

extractDiag()

VectorDouble MatrixSparse::extractDiag

(

int

oper_choice = 1

)

const

extractSubmatrixByColor()

MatrixSparse * MatrixSparse::extractSubmatrixByColor	(	const VectorInt &	colors,
		int	ref_color,
		bool	row_ok,
		bool	col_ok
	)

extractSubmatrixByRanks()

MatrixSparse * MatrixSparse::extractSubmatrixByRanks	(	const VectorInt &	rank_rows,
		const VectorInt &	rank_cols
	)

fill()

void MatrixSparse::fill ( double  value )

overridevirtual

Set all the values of the matrix at once

Change any nonzero term to 'value'

Parameters

value

Constant value used for filling 'this'

Reimplemented from AMatrix.

fillRandom()

void MatrixSparse::fillRandom	(	int	seed = 432432,
		double	zeroPercent = 0.1
	)

Set all the values of the Matrix with random values

freeCS()

void MatrixSparse::freeCS

(

)

getCS()

const cs * MatrixSparse::getCS

(

)

const

Returns a pointer to the Sparse storage

getCSUnprotected()

cs * MatrixSparse::getCSUnprotected

(

)

const

Temporary function to get the CS contents of Sparse Matrix

getEigenMatrix()

const Eigen::SparseMatrix<double>& MatrixSparse::getEigenMatrix ( ) const

inline

getMatrixToTriplet()

NF_Triplet MatrixSparse::getMatrixToTriplet ( int  shiftRow = 0, int  shiftCol = 0  ) const

overridevirtual

Extract the contents of the matrix

From a matrix of any type, creates the triplet (specific format for creating efficiently a Sparse matrix) It only takes the only non-zero elements of the matrix

Reimplemented from AMatrix.

getNonZeros()

int MatrixSparse::getNonZeros ( ) const

inline

getStats()

void MatrixSparse::getStats	(	int *	nrows,
		int *	ncols,
		int *	count,
		double *	percent
	)		const

getValue()

double MatrixSparse::getValue ( int  row, int  col, bool  flagCheck = true  ) const

overridevirtual

Get the value from a matrix cell

Implements AMatrix.

glue()

MatrixSparse * MatrixSparse::glue ( const MatrixSparse *  A1, const MatrixSparse *  A2, bool  flagShiftRow, bool  flagShiftCol  )

static

isDense()

bool MatrixSparse::isDense ( ) const

inlineoverridevirtual

Returns if the matrix belongs to the MatrixSparse class (avoids dynamic_cast)

Implements AMatrix.

isFlagEigen()

bool MatrixSparse::isFlagEigen ( ) const

inline

Cloneable interface.

isSparse()

bool MatrixSparse::isSparse ( ) const

inlineoverridevirtual

Interface for AMatrix.

Returns if the current matrix is Sparse

Implements AMatrix.

L1Norm()

double MatrixSparse::L1Norm

(

)

const

multiplyColumn()

void MatrixSparse::multiplyColumn ( const VectorDouble &  vec )

overridevirtual

Multiply the matrix column-wise

Multiply a Matrix column-wise

Reimplemented from AMatrix.

multiplyRow()

void MatrixSparse::multiplyRow ( const VectorDouble &  vec )

overridevirtual

Multiply the matrix row-wise

Multiply a Matrix row-wise

Reimplemented from AMatrix.

operator=()

MatrixSparse & MatrixSparse::operator=

(

const MatrixSparse &

m

)

prodMatMatInPlace()

void MatrixSparse::prodMatMatInPlace ( const AMatrix *  x, const AMatrix *  y, bool  transposeX = false, bool  transposeY = false  )

overridevirtual

Multiply a matrix by another and stored in 'this'

Store the product of 'x' by 'y' in this

Parameters

x	First Matrix
y	Second matrix
transposeX	True if First matrix is transposed
transposeY	True if Second matrix is transposed

Reimplemented from AMatrix.

prodMatVec()

VectorDouble MatrixSparse::prodMatVec ( const VectorDouble &  x, bool  transpose = false  ) const

overridevirtual

Perform y = 'this' %*% x

Reimplemented from AMatrix.

prodNormDiagVecInPlace()

void MatrixSparse::prodNormDiagVecInPlace	(	const VectorDouble &	vec,
		int	oper_choice = 1
	)

Perform: 'this' = diag('vec') %*% 'A' %*% diag('vec')

Parameters

vec	Input Vector
oper_choice	Type of transformation

prodNormMatInPlace()

void MatrixSparse::prodNormMatInPlace ( const MatrixSparse &  a, const VectorDouble &  vec = VectorDouble(), bool  transpose = false  )

virtual

Product 't(A)' %*% ['vec'] %*% 'A' or 'A' %*% ['vec'] %*% 't(A)' stored in 'this'

prodNormMatMatInPlace()

void MatrixSparse::prodNormMatMatInPlace ( const MatrixSparse &  a, const MatrixSparse &  m, bool  transpose = false  )

virtual

Product 't(A)' %*% 'M' %*% 'A' or 'A' %*% 'M' %*% 't(A)' stored in 'this'

prodScalar()

void MatrixSparse::prodScalar ( double  v )

overridevirtual

Multiply each matrix component by a value

Parameters

v	Multiply all the terms of the matrix by the scalar 'v'

Reimplemented from AMatrix.

prodVecMat()

VectorDouble MatrixSparse::prodVecMat ( const VectorDouble &  x, bool  transpose = false  ) const

overridevirtual

Perform y = x %*% 'this'

Reimplemented from AMatrix.

resetFromArray()

void MatrixSparse::resetFromArray ( int  nrows, int  ncols, const double *  tab, bool  byCol = true  )

overridevirtual

Reset the matrix from an array of double values.

Parameters

nrows	New number of rows
ncols	New number of columns
tab	The array of values
byCol	True if values are column-major in the array

Reimplemented from AMatrix.

resetFromTriplet()

void MatrixSparse::resetFromTriplet

(

const NF_Triplet &

NF_T

)

resetFromValue()

void MatrixSparse::resetFromValue ( int  nrows, int  ncols, double  value  )

overridevirtual

Reset the matrix to new dimensions and fill with a new value.

Parameters

nrows	New number of rows
ncols	New number of columns
value	The new value used to fill the matrix

Reimplemented from AMatrix.

resetFromVD()

void MatrixSparse::resetFromVD ( int  nrows, int  ncols, const VectorDouble &  tab, bool  byCol = true  )

overridevirtual

Reset the matrix from a vector of double values.

Parameters

nrows	New number of rows
ncols	New number of columns
tab	The vector of values
byCol	True if values are column-major in the vector

Reimplemented from AMatrix.

resetFromVVD()

void MatrixSparse::resetFromVVD ( const VectorVectorDouble &  tab, bool  byCol = true  )

overridevirtual

Reset the matrix from an array of double values.

Parameters

tab	The array of values
byCol	True if values are column-major in the array

Reimplemented from AMatrix.

scaleByDiag()

int MatrixSparse::scaleByDiag

(

)

setColumn()

void MatrixSparse::setColumn ( int  icol, const VectorDouble &  tab  )

overridevirtual

Set the contents of a Column

Fill a column of an already existing Sparse matrix, using 'tab' as entry The input 'tab' corresponds to the whole column contents

Parameters

icol	Column rank
tab	Vector containing the information (Dimension: nrows)

Warning

: This method only copies the values at the non-zero existing entries

Reimplemented from AMatrix.

setConstant()

void MatrixSparse::setConstant

(

double

value

)

setCS()

void MatrixSparse::setCS

(

cs *

cs

)

setDiagonal()

void MatrixSparse::setDiagonal ( const VectorDouble &  tab )

overridevirtual

Set the contents of the (main) Diagonal

Reimplemented from AMatrix.

setDiagonalToConstant()

void MatrixSparse::setDiagonalToConstant ( double  value = 1. )

overridevirtual

Set the contents of the (main) Diagonal to a constant value

Reimplemented from AMatrix.

setEigenMatrix()

void MatrixSparse::setEigenMatrix ( const Eigen::SparseMatrix< double > &  eigenMatrix )

inline

setRow()

void MatrixSparse::setRow ( int  irow, const VectorDouble &  tab  )

overridevirtual

Set the contents of a Row

Fill a row of an already existing Sparse matrix, using 'tab' as entry The input 'tab' corresponds to the whole row contents

Parameters

irow	Row rank
tab	Vector containing the information (Dimension: ncols)

Warning

: This method only copies the values at the non-zero existing entries

Reimplemented from AMatrix.

setValue()

void MatrixSparse::setValue ( int  irow, int  icol, double  value, bool  flagCheck = true  )

overridevirtual

Set the value for a matrix cell

Implements AMatrix.

simulateCholesky()

int MatrixSparse::simulateCholesky	(	const VectorDouble &	b,
		VectorDouble &	x
	)

solveCholesky()

int MatrixSparse::solveCholesky	(	const VectorDouble &	b,
		VectorDouble &	x
	)

toString()

String MatrixSparse::toString ( const AStringFormat *  strfmt = nullptr ) const

overridevirtual

Interface to AStringable.

Reimplemented from AMatrix.

Reimplemented in ProjMatrix.

transpose()

MatrixSparse * MatrixSparse::transpose ( ) const

overridevirtual

Transpose the matrix and return it as a copy

Reimplemented from AMatrix.

updValue()

void MatrixSparse::updValue ( int  irow, int  icol, const EOperator &  oper, double  value, bool  flagCheck = true  )

overridevirtual

Modifies the contents of a matrix cell

Implements AMatrix.

Member Data Documentation

DECLARE_TOTL

MatrixSparse::DECLARE_TOTL

Has a specific implementation in the Target language.

---

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

• include/Matrix/MatrixSparse.hpp
• src/Matrix/MatrixSparse.cpp