Struct SparseOps 

pub struct SparseOps;

Implementations

impl SparseOps

pub fn from_triplets_u32( nrows: usize, ncols: usize, triplets: Vec<(usize, usize, u32)>, ) -> Result<CsrMatrix<u32>>

Create CSR matrix from COO format using nalgebra_sparse native conversion

pub fn from_triplets( nrows: usize, ncols: usize, triplets: Vec<(usize, usize, u8)>, ) -> Result<CsrMatrix<u8>>

Create CSR matrix from triplets with u8 values

pub fn add_matrices( a: &CsrMatrix<u32>, b: &CsrMatrix<u32>, ) -> Result<CsrMatrix<u32>>

Highly optimized matrix addition using nalgebra_sparse’s spadd operation

pub fn filter_columns_u32( matrix: &CsrMatrix<u32>, keep_indices: &[usize], ) -> Result<CsrMatrix<u32>>

Optimized column filtering using sparsity pattern operations

pub fn compute_row_sums(matrix: &CsrMatrix<u32>) -> Vec<u32>

Optimized row sums using native CSR structure access

pub fn compute_col_sums(matrix: &CsrMatrix<u32>) -> Vec<u32>

Optimized column sums using parallel reduction over CSR structure

pub fn element_wise_multiply( a: &CsrMatrix<u32>, b: &CsrMatrix<u8>, ) -> Result<CsrMatrix<u32>>

Element-wise multiplication using optimized two-pointer algorithm for sparse matrices

This implementation uses a sorted two-pointer merge algorithm which is significantly faster than HashMap-based intersection for sparse matrices. CSR format guarantees column indices are already sorted within each row, allowing O(nnz_a + nnz_b) complexity instead of O(nnz_a * log(nnz_b)) with HashMap lookups.

pub fn transpose_u32(matrix: &CsrMatrix<u32>) -> CsrMatrix<u32>

Transpose operation using nalgebra_sparse native transpose

pub fn matrix_vector_multiply( matrix: &CsrMatrix<u32>, vector: &[u32], ) -> Result<Vec<u32>>

Matrix-vector multiplication using native spmv operation

pub fn get_density_stats(matrix: &CsrMatrix<u32>) -> (f64, usize, usize)

Get matrix density statistics