Struct CsfTensor 

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

Compressed Sparse Fiber (CSF) tensor format

Stores a sparse tensor using a hierarchical compressed structure. The modes are ordered from coarsest (level 0) to finest (last level). Values are stored only at the leaf level.

Implementations

impl<T> CsfTensor<T>

where T: Float + SparseElement + Debug + Copy + 'static,

pub fn from_sparse_tensor( tensor: &SparseTensor<T>, mode_order: &[usize], ) -> SparseResult<Self>

where T: Sum,

Create a CSF tensor from a COO-format SparseTensor.

The mode_order specifies which tensor mode corresponds to each CSF level. For example, for a 3D tensor with mode_order = [0, 1, 2], level 0 compresses mode 0, level 1 compresses mode 1, and values are indexed by mode 2.

Arguments

• tensor - The sparse tensor in COO format
• mode_order - Ordering of modes (must be a permutation of 0..ndim)

pub fn shape(&self) -> &[usize]

Get the tensor shape

pub fn ndim(&self) -> usize

Get the number of dimensions

pub fn nnz(&self) -> usize

Get the number of stored non-zero values

pub fn mode_order(&self) -> &[usize]

Get the mode ordering

pub fn fiber_pointers(&self, level: usize) -> Option<&[usize]>

Get the fiber pointers at a given level

pub fn fiber_indices(&self, level: usize) -> Option<&[usize]>

Get the fiber indices at a given level

pub fn values(&self) -> &[T]

Get the stored values

pub fn get(&self, coords: &[usize]) -> T

Look up a value by coordinate

pub fn to_sparse_tensor(&self) -> SparseResult<SparseTensor<T>>

where T: Sum,

Convert back to a COO-format SparseTensor

pub fn contract_vector( &self, mode: usize, vector: &[T], ) -> SparseResult<SparseTensor<T>>

where T: Sum,

Tensor contraction along a specified mode with a vector.

Computes the mode-n product with a vector: result = T x_n v The result is a tensor of one lower dimension.

Arguments

• mode - The tensor mode to contract
• vector - The vector to contract with (length = shape[mode])

pub fn mode_n_product( &self, mode: usize, matrix: &CsrArray<T>, ) -> SparseResult<SparseTensor<T>>

where T: Float + SparseElement + Div<Output = T> + Sum + 'static,

Mode-n product with a matrix: result = T x_n M

The result has the same number of dimensions as the input tensor, but the size of mode n changes from shape[n] to M.nrows.

Arguments

• mode - The tensor mode to multiply along
• matrix - The matrix (nrows x shape[mode])

pub fn memory_usage(&self) -> usize

Get memory usage estimate (bytes)

Trait Implementations

impl<T: Clone> Clone for CsfTensor<T>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<T: Debug> Debug for CsfTensor<T>

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

Formats the value using the given formatter.