Struct BSCMatrix 

pub struct BSCMatrix<T> {
    pub nrows: usize,
    pub ncols: usize,
    pub block_size: (usize, usize),
    pub block_rows: usize,
    pub block_cols: usize,
    pub data: Vec<T>,
    pub indices: Vec<usize>,
    pub indptr: Vec<usize>,
}

Block Sparse Column matrix with flat block storage.

Each non-zero block is stored contiguously in data (row-major within the block). block_size = (r, c) means each block has r rows and c columns.

Fields

nrows: usize

Total number of matrix rows.

ncols: usize

Total number of matrix columns.

block_size: (usize, usize)

Block dimensions (rows per block, cols per block).

block_rows: usize

Number of block-rows.

block_cols: usize

Number of block-columns.

data: Vec<T>

Flat block storage: length = nnz_blocks * r * c.

indices: Vec<usize>

Block-row indices: length = nnz_blocks.

indptr: Vec<usize>

Column pointer array: length = block_cols + 1.

Implementations

impl<T> BSCMatrix<T>

where T: Clone + Copy + Zero + One + SparseElement + Debug + PartialEq,

pub fn new( data: Vec<T>, indices: Vec<usize>, indptr: Vec<usize>, shape: (usize, usize), block_size: (usize, usize), ) -> SparseResult<Self>

Create a BSCMatrix from pre-built flat data.

Arguments

• data: flat block data, length must equal indices.len() * r * c.
• indices: block-row indices.
• indptr: column pointer, length block_cols + 1.
• shape: (nrows, ncols) of the full matrix.
• block_size: (r, c) block dimensions.

pub fn zeros( shape: (usize, usize), block_size: (usize, usize), ) -> SparseResult<Self>

Create an empty BSCMatrix (all-zero) of the given shape and block size.

pub fn from_bsr(bsr: &BSRMatrix<T>) -> SparseResult<Self>

where T: Add<Output = T> + Mul<Output = T>,

Convert a BSRMatrix to BSCMatrix (essentially a block-level transpose and re-index).

The resulting BSCMatrix represents the same matrix but stored column-major.

pub fn from_dense( dense: &[T], nrows: usize, ncols: usize, block_size: (usize, usize), ) -> SparseResult<Self>

Build a BSCMatrix from a row-major dense slice.

pub fn to_dense(&self) -> Vec<T>

Convert the BSCMatrix to a row-major dense vector (nrows × ncols).

pub fn to_bsr(&self) -> SparseResult<BSRMatrix<T>>

where T: Add<Output = T> + Mul<Output = T>,

Convert to a BSRMatrix (re-index as row-based).

pub fn spmv(&self, x: &[T]) -> SparseResult<Vec<T>>

where T: Add<Output = T> + Mul<Output = T>,

Sparse matrix-vector product: y = A * x.

Iterates over block-columns and scatters partial sums back to y.

pub fn transpose_to_bsr(&self) -> SparseResult<BSRMatrix<T>>

where T: Add<Output = T> + Mul<Output = T>,

Compute the transpose as a BSRMatrix with swapped block_size.

pub fn nnz_blocks(&self) -> usize

Return the number of non-zero blocks.

pub fn nnz(&self) -> usize

Return the total number of stored scalar values.

pub fn shape(&self) -> (usize, usize)

Return (nrows, ncols).

pub fn get(&self, row: usize, col: usize) -> T

Get the scalar value at position (row, col).

Trait Implementations

impl<T: Clone> Clone for BSCMatrix<T>

fn clone(&self) -> BSCMatrix<T>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<T: Debug> Debug for BSCMatrix<T>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.