sparse matrix class implementation

[mariaczi]

Hi,
I code class to storage sparse matrix row compressed and i have a
problem with implements method to setVal and addVal.
I will replace later this methods overloaded operator().
Please, can You help me?
Thanks in advance.

I write this, but it's not so good :(
#include <iostream>
#include <vector>

using namespace std;

template <class T>
class SparseMatrixCR {
public:
SparseMatrixCR() {}
SparseMatrixCR(size_t m, size_t n) : rows_(m),
cols_(n) {
row_ptr.resize(m+1);
}

// read
T operator()(size_t row, size_t col) const;
// write
T& operator()(size_t row, size_t col);

int setSize(size_t _size);
void setVal(size_t row, size_t col, T val);
void addVal(size_t row, size_t col, T val);

// -- friend func --
friend ostream& operator<< (ostream& os, const
SparseMatrixCR<T>& rha) {
// os << "[" << rha.GetRows() << "][" <<
rha.GetCols() << "]" << endl;

for(size_t i(0); i < rha.GetRows(); ++i) {
for(size_t j(0); j < rha.GetCols();
++j) {
os.setf(ios::scientific,
ios::floatfield);
os.width(14);
os << rha(i, j) << " ";
}
os << endl;
}
return os;
}
private:
vector< T > values;
vector< size_t > row_ptr;
vector< size_t > colidx;

size_t rows_;
size_t cols_;

}; // end of: class SparseMatrixCR

template <class T>
void SparseMatrixCR<T>::setVal(size_t row, size_t col, T val) {

if(val != 0 && values.size() == 0) {
colidx.push_back(col);
values.push_back(val);
for(size_t i(row+1); i < row_ptr.size(); ++i) {
row_ptr[i] += 1;
}
}
else {

bool exist(false);

for(size_t idx(row_ptr[row]); idx < row_ptr[row+1];
++idx) {
if(colidx[idx] == col) {
values[idx] = val;
exist = true;
break;
}

}
if(exist == false) {
if(val != 0) {
colidx.insert( colidx.begin() +
row_ptr[row], col );
values.insert( values.begin() +
row_ptr[row], val );
for(size_t i(row+1); i <
row_ptr.size(); ++i) {
row_ptr[i] += 1;
}
}
else {
colidx.erase( colidx.begin() +
row_ptr[row] );
values.erase( values.begin() +
row_ptr[row] );
for(size_t i(row+1); i <
row_ptr.size(); ++i) {
row_ptr[i] -= 1;
}
}
}
}

} // end of: setVal()


template <class T>
void SparseMatrixCR<T>::addVal(size_t row, size_t col, T val) {

if(values.size() == 0) {
colidx.push_back(col);
values.push_back(val);
for(size_t i(row+1); i < row_ptr.size(); ++i) {
row_ptr[i] += 1;
}
}
else {

bool exist(false);

for(size_t idx(row_ptr[row]); idx < row_ptr[row+1];
++idx) {
if(colidx[idx] == col) {
values[idx] += val;
exist = true;
break;
}
if(colidx[idx] > col) {
poz = idx;
}
}
if(exist == false) {
if(val != 0) {
colidx.insert( colidx.begin() +
row_ptr[row], col );
values.insert( values.begin() +
row_ptr[row], val );
for(size_t i(row+1); i <
row_ptr.size(); ++i) {
row_ptr[i] += 1;
}
}
}
}
} // end of: addVal()

[Daniel T.]

mariaczi wrote:

Hi,
I code class to storage sparse matrix row compressed and i have a
problem with implements method to setVal and addVal.
I will replace later this methods overloaded operator().
Please, can You help me?

Here's a simple sparse matrix class that you can use and is already
pretty efficient:

template < typename T >
class SparseMatrix {
std::map< std::pair< size_t, size_t >, T > rep;
public:
T operator()( size_t x, size_t y ) const {
return rep[ std::make_pair( x, y ) ];
}
T& operator()( size_t x, size_t y ) {
return rep[ std::make_pair( x, y ) ];
}
};

Add operations to taste. For example you might want to add a
member-function that tells wether a particular node exists.

[Dervish]

> Here's a simple sparse matrix class that you can use and is already

pretty efficient:

Here is even more simple sparse matrix (not it is zero initialized)

template <class T> class Matrix : public map<size_t, map<size_t, T> >
{};

int main ()
{
Matrix<double> mat;
mat[4238749][6234234] = 123.456;
cout << mat[4238749][6234234] << endl;
cout << mat[10000][2323436] << endl;
return 0;
}

[mariaczi]

On 8 Jun 2006 04:06:59 -0700, "Daniel T." <da******@earthlink.net>
wrote:

Here's a simple sparse matrix class that you can use and is already
pretty efficient: [cut]Add operations to taste. For example you might want to add a
member-function that tells wether a particular node exists.

it's ok, but i need my implementation to comparision with sparse
matrix implemented that You write

[rwf_20]

Daniel T. wrote:

template < typename T >
class SparseMatrix {
std::map< std::pair< size_t, size_t >, T > rep;
public:
T operator()( size_t x, size_t y ) const {
return rep[ std::make_pair( x, y ) ];
}
T& operator()( size_t x, size_t y ) {
return rep[ std::make_pair( x, y ) ];
}
};

Add operations to taste. For example you might want to add a
member-function that tells wether a particular node exists.

IIRC, operator[] on a std::map will allocate a new entry if one does
not already exist, no?

T operator()(size_t x, size_t y) const {
std::map<std::pair<size_t, size_t>, T>::const_iterator mapItr =
rep.find(std::make_pair<size_t, size_t>(x, y));
if (mapItr != rep.end()) return mapItr->second;
else return 0;
}

This is of course overkill if you can control calls to operator() such
that it will never be called on a nonexistent location. But, since I
believe [] reduces to a map lookup anyway, the performace overhead is
minimal. The advantage is that routines written for standard matrices
can work with the above with filling your map with a bunch of zeroes.

Ryan

[Daniel T.]

mariaczi wrote:

On 8 Jun 2006 04:06:59 -0700, "Daniel T." <da******@earthlink.net>
wrote:

Here's a simple sparse matrix class that you can use and is already
pretty efficient:

[cut]

Add operations to taste. For example you might want to add a
member-function that tells wether a particular node exists.

it's ok, but i need my implementation to comparision with sparse
matrix implemented that You write

If I understand you... Just write an op== and it will.

[Noah Roberts]

rwf_20 wrote:

IIRC, operator[] on a std::map will allocate a new entry if one does
not already exist, no?

That is true.

[mariaczi]

On 8 Jun 2006 12:14:33 -0700, "Daniel T." <da******@earthlink.net>
wrote:

If I understand you... Just write an op== and it will.

Hmm... i use my sparse matrix implementation to solve linear
equations, and i need two assorted implementation of sparse matrix to
compare solving time with this and this and test usage memory for all

sorry for my english :(
regards