ergo
SparseMatrix Class Reference

Sparse matrix structure optimized for XC data access pattern. More...

#include <sparse_matrix.h>

Classes

class  Exception

Public Member Functions

 SparseMatrix (const SparsePattern &pattern_)
 Constructs a square matrix and preallocate according to the specified pattern.
 SparseMatrix (const SparsePattern &pattern_, const symmMatrix &m, const int *aoMap, std::vector< int > const &permutationHML)
 ~SparseMatrix ()
void print (const char *title) const
void addSymmetrizedTo (symmMatrix &sMat, const int *aoMap, std::vector< int > const &permutationHML) const
 Assigns itself to a given hierarchic matrix.
void add (int row, int col, ergo_real val)
 Adds given value to an element in given row and column.
ergo_real at (int row, int col) const

Private Member Functions

void createOffsets (const SparsePattern &pattern)
 Fills in offsets and his based on pattern.

Private Attributes

const SparsePattern & pattern
ergo_real ** columns
int ** offsets
 for accelerated at() and add() methods.
int ** his
 for accelerated at() and add() methods.
int * cnt
 for accelerated at() and add() methods.
int n

Detailed Description

Sparse matrix structure optimized for XC data access pattern.

Constructor & Destructor Documentation

◆ SparseMatrix() [1/2]

SparseMatrix::SparseMatrix ( const SparsePattern & pattern_)
explicit

Constructs a square matrix and preallocate according to the specified pattern.

References columns, createOffsets(), n, and pattern.

◆ SparseMatrix() [2/2]

SparseMatrix::SparseMatrix ( const SparsePattern & pattern_,
const symmMatrix & m,
const int * aoMap,
std::vector< int > const & permutationHML )

References SparsePattern::Column::begin(), columns, createOffsets(), SparsePattern::Column::end(), n, and pattern.

◆ ~SparseMatrix()

SparseMatrix::~SparseMatrix ( )
inline

References cnt, columns, his, n, and offsets.

Member Function Documentation

◆ add()

void SparseMatrix::add ( int row,
int col,
ergo_real val )
inline

Adds given value to an element in given row and column.

Checking against intervals.end() is terribly expensive!!! Luckily, we do not have to do it.

References cnt, columns, his, and offsets.

◆ addSymmetrizedTo()

void SparseMatrix::addSymmetrizedTo ( symmMatrix & sMat,
const int * aoMap,
std::vector< int > const & permutationHML ) const

Assigns itself to a given hierarchic matrix.

References SparsePattern::Column::begin(), columns, SparsePattern::Column::end(), n, and pattern.

Referenced by XCEvaluatorRestricted::getXC().

◆ at()

ergo_real SparseMatrix::at ( int row,
int col ) const
inline

References cnt, columns, his, and offsets.

◆ createOffsets()

void SparseMatrix::createOffsets ( const SparsePattern & pattern)
private

Fills in offsets and his based on pattern.

References cnt, his, n, offsets, and SparsePattern::size().

Referenced by SparseMatrix(), and SparseMatrix().

◆ print()

void SparseMatrix::print ( const char * title) const

References SparsePattern::Column::begin(), columns, SparsePattern::Column::end(), n, and pattern.

Member Data Documentation

◆ cnt

int* SparseMatrix::cnt
private

for accelerated at() and add() methods.

Referenced by add(), at(), createOffsets(), and ~SparseMatrix().

◆ columns

ergo_real** SparseMatrix::columns
private

Referenced by add(), addSymmetrizedTo(), at(), print(), SparseMatrix(), SparseMatrix(), and ~SparseMatrix().

◆ his

int** SparseMatrix::his
private

for accelerated at() and add() methods.

Referenced by add(), at(), createOffsets(), and ~SparseMatrix().

◆ n

int SparseMatrix::n
private

Referenced by addSymmetrizedTo(), createOffsets(), print(), SparseMatrix(), SparseMatrix(), and ~SparseMatrix().

◆ offsets

int** SparseMatrix::offsets
private

for accelerated at() and add() methods.

Referenced by add(), at(), createOffsets(), and ~SparseMatrix().

◆ pattern

const SparsePattern& SparseMatrix::pattern
private

Referenced by addSymmetrizedTo(), print(), SparseMatrix(), and SparseMatrix().

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