Struct Tensor 

pub struct Tensor {
    pub id: TensorId,
    pub name: Option<String>,
    pub data: Buffer,
    pub shape: Shape,
    pub device: Device,
    pub dtype: DType,
    pub requires_grad: bool,
}

The core Tensor struct.

Strictly acts as a metadata wrapper. It points to a physical memory Buffer and maps it using a geometric Shape. It contains NO recursive graph pointers.

Fields

id: TensorId

The globally unique token used by the Tape to track gradients.

name: Option<String>

An optional string for explainability (e.g., “layer_1_weights”).

data: Buffer

The physical memory container (wrapped in Arc for zero-copy sharing).

shape: Shape

The geometric layout and memory strides.

device: Device

The hardware context where the physical memory resides.

dtype: DType

The precision format of the memory buffer.

requires_grad: bool

Flag indicating whether the Autograd engine should track this tensor.

Implementations

impl Tensor

pub fn new( data: Buffer, shape: Shape, device: Device, dtype: DType, requires_grad: bool, ) -> Self

Constructs a new Tensor from raw components. Automatically assigns a new unique TensorId.

pub fn with_name(self, name: &str) -> Self

Builder pattern helper to attach an explainability name to the tensor.

pub fn transpose(&self) -> Self

Performs an O(1) mathematical transpose.

This generates a NEW tensor (with a new ID for the graph) but explicitly clones only the Arc buffer pointer, never the underlying memory.

Trait Implementations

impl Add for &Tensor

type Output = Tensor

The resulting type after applying the + operator.

fn add(self, rhs: Self) -> Self::Output

Performs the + operation.

impl Clone for Tensor

fn clone(&self) -> Tensor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Tensor

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

Formats the value using the given formatter.

impl Mul for &Tensor

type Output = Tensor

The resulting type after applying the * operator.

fn mul(self, rhs: Self) -> Self::Output

Performs the * operation.