Struct CsMatrix 

pub struct CsMatrix<T> { /* private fields */ }

Compressed Sparse Matrix in either CSR or CSC format.

This is the core sparse matrix structure that can represent either:

• CSR (Compressed Sparse Row) format when rows are compressed
• CSC (Compressed Sparse Column) format when columns are compressed

The matrix stores only non-zero elements and their indices, making it memory-efficient for matrices with few non-zero elements.

Format

The matrix uses three internal arrays:

• secondary_indices: Column indices for CSR or row indices for CSC
• primary_offsets: Offsets for each row (CSR) or column (CSC)
• values: Non-zero values stored in row-major (CSR) or column-major (CSC) order

Examples

use algebra_sparse::{CsMatrix, CsrMatrix};
use nalgebra::DMatrix;

// Create from dense matrix
let dense = DMatrix::from_row_slice(2, 3, &[
    1.0, 0.0, 2.0,
    0.0, 3.0, 0.0,
]);
let sparse = CsrMatrix::from_dense(dense.as_view());

Implementations

impl<T> CsMatrix<T>

where T: Real,

pub fn from_dense( dense_mat: DMatrixView<'_, T>, row_sparse: bool, zero_threshold: T, ) -> Self

Create a CsMatrix from a dense matrix.

Arguments

• dense_mat - The dense matrix to convert
• row_sparse - If true, creates CSR format; if false, creates CSC format
• zero_threshold - Values below this threshold are treated as zero and not stored

Examples

use algebra_sparse::CsMatrix;
use nalgebra::DMatrix;

let dense = DMatrix::from_row_slice(2, 3, &[
    1.0, 0.0, 2.0,
    0.0, 3.0, 0.0,
]);

// Create CSR format
let csr = CsMatrix::from_dense(dense.as_view(), true, 1e-10);

// Create CSC format
let csc = CsMatrix::from_dense(dense.as_view(), false, 1e-10);

pub fn new(secondary_size: usize) -> Self

Creates a new empty CsMatrix with the given secondary axis size.

Arguments

• secondary_size - Number of columns for CSR format or rows for CSC format

Examples

use algebra_sparse::CsMatrix;

// Create an empty CSR matrix with 3 columns
let csr: CsMatrix<f64> = CsMatrix::new(3);

pub fn new_lane_builder(&mut self, zero_threshold: T) -> CsVecBuilder<'_, T>

Creates a new lane (row or column) builder for this matrix.

Returns a CsVecBuilder that can be used to efficiently add non-zero elements to the next lane of the matrix. When the builder is dropped, the lane is finalized.

Arguments

• zero_threshold - Values below this threshold are filtered out

Examples

use algebra_sparse::CsMatrix;

let mut matrix = CsMatrix::new(3);
let mut builder = matrix.new_lane_builder(1e-10);
builder.push(0, 1.0);  // Add element at column 0
builder.push(2, 2.0);  // Add element at column 2

pub fn reset(&mut self, secondary_size: usize)

Resets the matrix to empty state with a new secondary axis size.

This clears all data and allows reuse of the matrix with different dimensions.

Arguments

• secondary_size - New secondary axis size

pub fn clear(&mut self)

Clears all data from the matrix while preserving the secondary axis size.

This removes all lanes but keeps the matrix ready for reuse with the same dimensions.

pub fn num_primary(&self) -> usize

Returns the number of primary axes (rows for CSR, columns for CSC).

pub fn num_secondary(&self) -> usize

Returns the number of secondary axes (columns for CSR, rows for CSC).

pub fn get_lane(&self, lane_index: usize) -> CsVecRef<'_, T>

Gets a lane (row for CSR, column for CSC) as an immutable sparse vector.

Arguments

• lane_index - Index of the lane to retrieve

Returns

A CsVecRef representing the sparse vector for this lane

pub fn get_lane_mut(&mut self, lane_index: usize) -> CsVecMut<'_, T>

Gets a lane (row for CSR, column for CSC) as a mutable sparse vector.

Arguments

• lane_index - Index of the lane to retrieve

Returns

A CsVecMut allowing modification of the lane’s values

pub fn as_view(&self) -> CsMatrixView<'_, T>

Returns an immutable view of this matrix.

The view allows efficient read-only access without allocation.

pub fn dense_rate(&self) -> f32

Returns the density rate of the matrix.

This is the ratio of non-zero elements to total elements. A value of 0.1 means 10% of elements are non-zero.

Trait Implementations

impl<T: Clone> Clone for CsMatrix<T>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<T: Debug> Debug for CsMatrix<T>

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

Formats the value using the given formatter.

impl<T: Default> Default for CsMatrix<T>

fn default() -> CsMatrix<T>

Returns the “default value” for a type.

impl<'a, T> IntoView for &'a CsMatrix<T>

type View = CsMatrixView<'a, T>

The view type produced by this trait.

fn into_view(self) -> Self::View

Converts the type into an immutable view.