Dynamic size matrix using std::vector for storage

suggest change

Unfortunately as of C++14 there’s no dynamic size matrix class in the C++ standard library. Matrix classes that support dynamic size are however available from a number of 3rd party libraries, including the Boost Matrix library (a sub-library within the Boost library).

If you don’t want a dependency on Boost or some other library, then one poor man’s dynamic size matrix in C++ is just like

vector<vector<int>> m( 3, vector<int>( 7 ) );

… where vector is std::vector. The matrix is here created by copying a row vector n times where n is the number of rows, here 3. It has the advantage of providing the same m[y] indexing notation as for a fixed size raw array matrix, but it’s a bit inefficient because it involves a dynamic allocation for each row, and it’s a bit unsafe because it’s possible to inadvertently resize a row.

A more safe and efficient approach is to use a single vector as storage for the matrix, and map the client code’s (x, y) to a corresponding index in that vector:

#include <algorithm>        // std::copy
#include <cassert>          // assert
#include <initializer_list> // std::initializer_list
#include <vector>           // std::vector
#include <cstddef>          // ptrdiff_t

// A dynamic size matrix using std::vector for storage.

namespace my {
    using Size = ptrdiff_t;
    using std::initializer_list;
    using std::vector;

    template< class Item >
    class Matrix
    {
    private:
        vector<Item>    items_;
        Size            n_cols_;
        
        auto index_for( Size const x, Size const y ) const
            -> Size
        { return y*n_cols_ + x; }

    public:
        auto n_rows() const -> Size { return items_.size()/n_cols_; }
        auto n_cols() const -> Size { return n_cols_; }

        auto item( Size const x, Size const y )
            -> Item&
        { return items_[index_for(x, y)]; }
        
        auto item( Size const x, Size const y ) const
            -> Item const&
        { return items_[index_for(x, y)]; }

        Matrix(): n_cols_( 0 ) {}

        Matrix( Size const n_cols, Size const n_rows )
            : items_( n_cols*n_rows )
            , n_cols_( n_cols )
        {}
        
        Matrix( initializer_list< initializer_list<Item> > const& values )
            : items_()
            , n_cols_( values.size() == 0? 0 : values.begin()->size() )
        {
            for( auto const& row : values )
            {
                assert( Size( row.size() ) == n_cols_ );
                items_.insert( items_.end(), row.begin(), row.end() );
            }
        }
    };
}  // namespace my

//--------------------------------------------- Usage:
using my::Matrix;

auto some_matrix()
    -> Matrix<int>
{
    return
    {
        {  1,  2,  3,  4,  5,  6,  7 },
        {  8,  9, 10, 11, 12, 13, 14 },
        { 15, 16, 17, 18, 19, 20, 21 }
    };
}

#include <iostream>
#include <iomanip>
using namespace std;
auto main() -> int
{
    Matrix<int> const m = some_matrix();
    assert( m.n_cols() == 7 );
    assert( m.n_rows() == 3 );
    for( int y = 0, y_end = m.n_rows(); y < y_end; ++y )
    {
        for( int x = 0, x_end = m.n_cols(); x < x_end; ++x )
        {
            cout << setw( 4 ) << m.item( x, y );        // ← Note: not `m[y][x]`!
        }
        cout << '\n';
    }
}
   1   2   3   4   5   6   7
   8   9  10  11  12  13  14
  15  16  17  18  19  20  21

The above code is not industrial grade: it’s designed to show the basic principles, and serve the needs of students learning C++.

For example, one may define operator() overloads to simplify the indexing notation.

Feedback about page:

Feedback:
Optional: your email if you want me to get back to you:



Table Of Contents