Are there sparse matrix algorithms in scirust?

[zkbpkp]

I need a library to work with large-scale sparse matrices. Does scirust support sparse algorithms?

[zkbpkp]

I've explored the code and found out that there's no sparse matrices support. So I want to suggest a refactoring: Matrix becomes a trait, and different matrices representations (like dense matrices, CSR and CSC) implement this trait.

pub trait Matrix<T: MagmaBase> {
    /// Returns the capacity of the matrix i.e. the number of elements it can
    /// hold
    fn capacity(&self) -> usize;

    /// Reshapes the matrix
    fn reshape(&mut self, rows: usize, cols: usize) -> bool;

    /// Returns an iterator over a specific row of the matrix
    fn row_iter(&self, r: isize) -> RowIterator<T>;

    /// Returns an iterator over a specific column of the matrix
    fn col_iter(&self, c: isize) -> ColIterator<T>;

    /// Returns an iterator over all cells of the matrix
    fn cell_iter(&self) -> CellIterator<T>;

    /// Provide the main diagonal elements
    fn diagonal_iter(&self) -> DiagIterator<T>;

    // ...more methods...
}

impl<T: MagmaBase> Matrix<T> for DenseMatrix<T> {
    // ...
}

impl<T: MagmaBase> Matrix<T> for SparseMatrix<T> {
    // ...
}

// ...more impl's...

With these changes it will be possible to operate with matrices of different representation, e.g. add "regular" matrix with CSR matrix.
If you are interested in such changes, I can work on it (actually, I've already started), but maybe I'll need some help.

P.S. Sorry if my English isn't very good.

[shailesh1729]

Yes, at the moment there is no sparse matrix support. I will be able to merge the changes if and when you are ready. We can chat if you need any explanation on any part.