Module sparse 

Sparse Tensor Support

Provides sparse tensor representations for memory-efficient storage and computation when tensors have many zero elements.

Formats

• COO (Coordinate): Best for construction and random access
• CSR (Compressed Sparse Row): Best for row-wise operations and matrix-vector products

Example

use axonml_tensor::sparse::{SparseTensor, SparseFormat};

// Create from COO format
let indices = vec![(0, 1), (1, 0), (2, 2)];
let values = vec![1.0, 2.0, 3.0];
let sparse = SparseTensor::from_coo(&indices, &values, &[3, 3]);

// Convert to dense
let dense = sparse.to_dense();

@version 0.1.0

Structs

SparseCOO

Sparse tensor in COO (Coordinate) format.

SparseCSR

Sparse tensor in CSR (Compressed Sparse Row) format.

Enums

SparseFormat

Sparse tensor storage format.

SparseTensor

Unified sparse tensor interface supporting multiple formats.