Struct CsrMatrix 

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

Compressed Sparse Row (CSR) sparse matrix.

Stores non-zero entries in row-major order using three arrays: indptr (row pointer array of length n_rows + 1), indices (column indices of each non-zero), and data (values of each non-zero).

Type Parameter

T — the scalar element type. No bounds are required for basic structural methods; arithmetic methods impose their own bounds.

Dataset Trait

Implements ferrolearn_core::Dataset when T: Float + Send + Sync + 'static, reporting n_samples() == n_rows(), n_features() == n_cols(), and is_sparse() == true.

Implementations

impl<T> CsrMatrix<T>

where T: Clone,

pub fn new( n_rows: usize, n_cols: usize, indptr: Vec<usize>, indices: Vec<usize>, data: Vec<T>, ) -> Result<Self, FerroError>

Construct a CSR matrix from raw components.

Arguments

• n_rows — number of rows.
• n_cols — number of columns.
• indptr — row pointer array of length n_rows + 1.
• indices — column index of each non-zero entry.
• data — value of each non-zero entry.

Errors

Returns FerroError::InvalidParameter if the data is structurally invalid (wrong lengths, out-of-bound indices, unsorted inner indices).

pub fn n_rows(&self) -> usize

Returns the number of rows.

pub fn n_cols(&self) -> usize

Returns the number of columns.

pub fn nnz(&self) -> usize

Returns the number of stored non-zero entries.

pub fn inner(&self) -> &CsMat<T>

Returns a reference to the underlying sprs::CsMat<T>.

pub fn into_inner(self) -> CsMat<T>

Consume this matrix and return the underlying sprs::CsMat<T>.

pub fn from_coo(coo: &CooMatrix<T>) -> Result<Self, FerroError>

where T: Clone + Add<Output = T> + 'static,

Construct a CsrMatrix from a CooMatrix by converting to CSR.

Duplicate entries at the same position are summed.

Errors

This conversion is always successful for structurally valid inputs.

pub fn from_csc(csc: &CscMatrix<T>) -> Result<Self, FerroError>

where T: Clone + Default + 'static,

Construct a CsrMatrix from a CscMatrix.

Errors

This conversion is always successful.

pub fn to_csc(&self) -> CscMatrix<T>

where T: Clone + Default + 'static,

Convert to CscMatrix.

pub fn to_coo(&self) -> CooMatrix<T>

Convert to CooMatrix.

Each non-zero becomes one triplet entry.

pub fn to_dense(&self) -> Array2<T>

where T: Clone + Zero + 'static,

Convert this sparse matrix to a dense Array2<T>.

pub fn from_dense(dense: &ArrayView2<'_, T>, epsilon: T) -> Self

where T: Copy + Zero + PartialOrd + Signed + 'static,

Construct a CsrMatrix from a dense Array2<T>, dropping entries whose absolute value is less than or equal to epsilon.

Entries v where |v| <= epsilon are treated as structural zeros. For integer types, pass epsilon = 0.

pub fn row_slice( &self, start: usize, end: usize, ) -> Result<CsrMatrix<T>, FerroError>

where T: Clone + Default + 'static,

Return a new CSR matrix containing only the rows in start..end.

Errors

Returns FerroError::InvalidParameter if start > end or end > n_rows().

pub fn scale(&mut self, scalar: T)

where for<'r> T: MulAssign<&'r T>,

Scalar multiplication in-place: multiplies every non-zero by scalar.

Requires T: for<'r> MulAssign<&'r T>, which is satisfied by all primitive numeric types.

pub fn mul_scalar(&self, scalar: T) -> CsrMatrix<T>

where T: Copy + Mul<Output = T> + Zero + 'static,

Scalar multiplication returning a new matrix.

pub fn add(&self, rhs: &CsrMatrix<T>) -> Result<CsrMatrix<T>, FerroError>

where T: Zero + Default + Clone + 'static, for<'r> &'r T: Add<&'r T, Output = T>,

Element-wise addition of two CSR matrices with the same shape.

Errors

Returns FerroError::ShapeMismatch if the matrices have different shapes.

pub fn mul_vec(&self, rhs: &Array1<T>) -> Result<Array1<T>, FerroError>

where T: Clone + Zero + 'static + AddAssign, for<'r> &'r T: Mul<Output = T>,

Sparse matrix-dense vector product: computes self * rhs.

Errors

Returns FerroError::ShapeMismatch if rhs.len() != n_cols().

impl<T> CsrMatrix<T>

where T: Float + Send + Sync + Signed + 'static,

pub fn from_dense_float(dense: &ArrayView2<'_, T>) -> Self

Construct a CsrMatrix from a dense Array2<T>, treating entries with absolute value at or below T::epsilon() as structural zeros.

Trait Implementations

impl<T: Clone> Clone for CsrMatrix<T>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F> Dataset for CsrMatrix<F>

where F: Float + Send + Sync + 'static,

Implements Dataset so that CsrMatrix<F> can be passed to any ferrolearn algorithm that accepts a dataset.

• n_samples() — number of rows (one sample per row).
• n_features() — number of columns (one feature per column).
• is_sparse() — always true.

fn n_samples(&self) -> usize

Returns the number of samples (rows) in the dataset.

fn n_features(&self) -> usize

Returns the number of features (columns) in the dataset.

fn is_sparse(&self) -> bool

Returns true if the dataset uses a sparse representation.

impl<T: Debug> Debug for CsrMatrix<T>

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

Formats the value using the given formatter.