Enum SparseFormat 

pub enum SparseFormat {
    Coo,
    Csr,
    Csc,
    Bsr,
    Dia,
    Dsr,
    Ell,
    Rle,
    Symmetric,
}

Layout format for sparse tensors

Different sparse formats are optimized for different use cases and access patterns. Choose the appropriate format based on your matrix characteristics and operations.

Variants

Coo

Coordinate format (COO) - stores (row, col, value) triplets

Best for: Matrix construction, format conversion, random insertion Memory: 3 * nnz storage (row indices, col indices, values) Operations: Efficient addition, inefficient matrix-vector multiplication

Csr

Compressed Sparse Row (CSR) - row-oriented compressed format

Best for: Matrix-vector multiplication, row slicing, iterating by rows Memory: (nnz + n + 1) storage (values, col_indices, row_ptr) Operations: Fast row access, efficient SpMV, slow column access

Csc

Compressed Sparse Column (CSC) - column-oriented compressed format

Best for: Matrix-vector multiplication (A^T * x), column slicing Memory: (nnz + m + 1) storage (values, row_indices, col_ptr) Operations: Fast column access, efficient transpose operations

Bsr

Block Sparse Row (BSR) - stores dense blocks in sparse locations

Best for: Matrices with dense block structure, finite element methods Memory: Efficient for matrices with natural block structure Operations: BLAS-optimized operations on dense blocks

Dia

Diagonal format (DIA) - stores diagonals efficiently

Best for: Matrices with few non-zero diagonals, finite difference schemes Memory: Very compact for diagonal-dominant matrices Operations: Fast diagonal operations, limited to diagonal patterns

Dsr

Dynamic Sparse Row (DSR) - dynamic insertion/deletion support

Best for: Matrices that change structure frequently during computation Memory: Tree-based storage allows dynamic modifications Operations: Efficient insertion/deletion, slower than static formats

Ell

ELLPACK format (ELL) - fixed-width row storage

Best for: GPU operations, SIMD vectorization, matrices with uniform row density Memory: Can have significant overhead for irregular matrices Operations: SIMD-friendly, efficient on parallel architectures

Rle

Run-Length Encoded format (RLE) - compresses consecutive zeros

Best for: Matrices with long runs of consecutive non-zeros Memory: Excellent compression for specific patterns Operations: Specialized for pattern-specific matrices

Symmetric

Symmetric sparse format (SYM) - stores only lower/upper triangle

Best for: Symmetric matrices from finite element analysis, optimization Memory: Roughly half the storage of equivalent full format Operations: Specialized symmetric operations, automatic symmetry enforcement

Trait Implementations

impl Clone for SparseFormat

fn clone(&self) -> SparseFormat

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SparseFormat

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

Formats the value using the given formatter.

impl From<ScipyFormat> for SparseFormat

fn from(format: ScipyFormat) -> Self

Converts to this type from the input type.

impl From<SparseFormat> for ScipyFormat

fn from(format: SparseFormat) -> Self

Converts to this type from the input type.

impl Hash for SparseFormat

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for SparseFormat

fn eq(&self, other: &SparseFormat) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for SparseFormat

impl Eq for SparseFormat

impl StructuralPartialEq for SparseFormat