Struct DenseTensor 

pub struct DenseTensor {
    pub shape: Vec<usize>,
    pub data: Vec<f64>,
}

A dense tensor of arbitrary rank stored in row-major (C-order) layout.

The shape is given as a Vecusize` and the flat data length must equal the product of all dimensions.

Fields

shape: Vec<usize>

Shape of the tensor, e.g. [2, 3, 4] for a 2×3×4 tensor.

data: Vec<f64>

Flat row-major storage.

Implementations

impl DenseTensor

pub fn zeros(shape: &[usize]) -> Self

Create a zero tensor with the given shape.

pub fn from_data(shape: &[usize], data: Vec<f64>) -> Self

Create a tensor from flat data (row-major).

Panics

Panics if data.len() != shape.iter().product().

pub fn strides(&self) -> Vec<usize>

Compute the strides for row-major indexing.

pub fn flat_index(&self, idx: &[usize]) -> usize

Flat index from a multi-index.

pub fn get(&self, idx: &[usize]) -> f64

Get element at multi-index.

pub fn set(&mut self, idx: &[usize], val: f64)

Set element at multi-index.

pub fn frobenius_norm(&self) -> f64

Frobenius norm: sqrt of sum of squared elements.

pub fn mode_n_unfold(&self, mode: usize) -> Vec<Vec<f64>>

Mode-n unfolding (matricisation).

Returns a matrix of shape (shape[n], prod(other dims)) stored row-major as a Vec<Vecf64>.

pub fn mode_n_fold(mat: &[Vec<f64>], mode: usize, shape: &[usize]) -> Self

Mode-n folding: reconstruct a tensor from a mode-n unfolded matrix.

mat has shape (shape[mode], prod(other dims)). shape is the target tensor shape.

impl DenseTensor

pub fn clone_tensor(&self) -> DenseTensor

Clone helper (needed since derive(Clone) isn’t on the struct).

pub fn scale(&self, s: f64) -> DenseTensor

Scale all elements by a scalar.

pub fn add_tensor(&self, other: &DenseTensor) -> DenseTensor

Element-wise addition (shapes must match).

pub fn sub_tensor(&self, other: &DenseTensor) -> DenseTensor

Element-wise subtraction (shapes must match).

pub fn max_abs(&self) -> f64

Max absolute element.

pub fn sum(&self) -> f64

Sum all elements.

pub fn rank(&self) -> usize

Rank (number of modes).

pub fn numel(&self) -> usize

Total number of elements.

Trait Implementations

impl Clone for DenseTensor

fn clone(&self) -> DenseTensor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for DenseTensor

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

Formats the value using the given formatter.