Struct TensorView 

pub struct TensorView<'a, T: Scalar> { /* private fields */ }

Borrowed tensor view, lifetime-tied to the source Tensor.

TensorView is the borrowed counterpart to Tensor, following the String/&str pattern. It references the source tensor’s data buffer without copying.

Public vs. internal views

Public API methods (Tensor::tensor_view, etc.) call Tensor::wait before constructing a view, so the returned TensorView always has event = None — data is ready to read.

The crate-internal as_operand_view() skips the wait and propagates the pending event, allowing accelerator operations to chain without CPU synchronization.

Implementations

impl<'a, T: Scalar> TensorView<'a, T>

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

Returns the shape (size of each dimension).

pub fn strides(&self) -> &[isize]

Returns the strides (in units of T).

pub fn ndim(&self) -> usize

Returns the number of dimensions (rank).

pub fn logical_memory_space(&self) -> LogicalMemorySpace

Returns the logical memory space where the source tensor’s data resides.

pub fn preferred_compute_device(&self) -> Option<ComputeDevice>

Returns the preferred compute device override, if set.

pub fn buffer(&self) -> &DataBuffer<T>

Returns a reference to the underlying data buffer.

pub fn offset(&self) -> isize

Returns the element offset into the data buffer.

pub fn permute(&self, _perm: &[usize]) -> Result<TensorView<'a, T>>

Permute (reorder) the dimensions of this view.

Returns a new TensorView with reordered dims and strides (zero-copy).

Errors

Returns an error if perm is not a valid permutation of 0..ndim().

pub fn broadcast(&self, _target_dims: &[usize]) -> Result<TensorView<'a, T>>

Broadcast this view to a larger shape.

Dimensions of size 1 are expanded to the target size (zero-copy via stride 0).

Errors

Returns an error if target_dims is incompatible with the current shape.

pub fn diagonal(&self, _axes: &[(usize, usize)]) -> Result<TensorView<'a, T>>

Extract a diagonal view by merging pairs of axes.

Errors

Returns an error if any axis is out of range or paired dimensions have different sizes.

pub fn select(&self, _dim: usize, _index: usize) -> Result<TensorView<'a, T>>

Select a single index along a dimension, removing that dimension.

Returns a view with ndim() - 1 dimensions. Zero-copy: adjusts offset and removes the selected dimension from dims/strides.

Errors

Returns an error if dim >= ndim() or index >= dims()[dim].

Examples

use tenferro_tensor::{Tensor, MemoryOrder};
use tenferro_device::LogicalMemorySpace;

let a = Tensor::<f64>::zeros(&[3, 4, 10],
    LogicalMemorySpace::MainMemory, MemoryOrder::ColumnMajor);
let tv = a.tensor_view();
// Select batch index 5 → view of shape [3, 4]
let mat = tv.select(2, 5).unwrap();
assert_eq!(mat.dims(), &[3, 4]);

pub fn narrow( &self, _dim: usize, _start: usize, _length: usize, ) -> Result<TensorView<'a, T>>

Narrow (slice) a dimension to a sub-range.

Returns a view with the same number of dimensions, but dims()[dim] reduced to length. Zero-copy: only offset and dim size change.

Errors

Returns an error if dim >= ndim() or start + length > dims()[dim].

Examples

use tenferro_tensor::{Tensor, MemoryOrder};
use tenferro_device::LogicalMemorySpace;

let a = Tensor::<f64>::zeros(&[3, 10],
    LogicalMemorySpace::MainMemory, MemoryOrder::ColumnMajor);
let tv = a.tensor_view();
// Take columns 2..5 → view of shape [3, 3]
let sub = tv.narrow(1, 2, 3).unwrap();
assert_eq!(sub.dims(), &[3, 3]);

pub fn to_tensor(&self, _order: MemoryOrder) -> Tensor<T>

Copy this view into an owned Tensor.

pub fn contiguous(&self, _order: MemoryOrder) -> Tensor<T>

Return a contiguous copy of this view’s data.

pub fn conj(&self) -> Tensor<T>

where T: Conjugate,

Return a tensor with complex-conjugated elements from this view.

For real types, returns a copy unchanged.