Struct FixedShapeTensorMetadata 

pub struct FixedShapeTensorMetadata { /* private fields */ }

Metadata for a FixedShapeTensor extension type.

Implementations

impl FixedShapeTensorMetadata

pub fn new(shape: Vec<usize>) -> Self

Creates a new FixedShapeTensorMetadata with the given logical shape.

Use with_dim_names and with_permutation to further configure the metadata.

pub fn with_dim_names(self, names: Vec<String>) -> VortexResult<Self>

Sets the dimension names for this tensor. An empty vec is normalized to None since a 0-dimensional tensor has no dimensions to name.

The number of names must match the number of logical dimensions.

pub fn with_permutation(self, permutation: Vec<usize>) -> VortexResult<Self>

Sets the permutation for this tensor. An empty vec is normalized to None since a 0-dimensional tensor has no dimensions to permute.

The permutation must be a valid permutation of [0, 1, ..., N-1] where N is the number of logical dimensions.

pub fn ndim(&self) -> usize

Returns the number of dimensions (rank) of the tensor.

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

Returns the logical dimensions of the tensor as a slice.

pub fn dim_names(&self) -> Option<&[String]>

Returns the dimension names, if set.

pub fn permutation(&self) -> Option<&[usize]>

Returns the permutation, if set.

pub fn physical_shape(&self) -> impl Iterator<Item = usize> + '_

Returns an iterator over the physical shape of the tensor.

The physical shape describes the row-major memory layout. It is derived from the logical shape by placing each logical dimension’s size at its physical position: physical_shape[permutation[i]] = logical_shape[i].

When no permutation is present, the physical shape is identical to the logical shape.

pub fn strides(&self) -> impl Iterator<Item = usize> + '_

Returns an iterator over the strides for each logical dimension of the tensor.

The stride for a logical dimension is the number of elements to skip in the flat backing array in order to move one step along that logical dimension.

When a permutation is present, the physical memory is laid out in row-major order over the physical dimensions (the logical dimensions reordered by the permutation), so the strides are derived from that physical layout.

Trait Implementations

impl Clone for FixedShapeTensorMetadata

fn clone(&self) -> FixedShapeTensorMetadata

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for FixedShapeTensorMetadata

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

Formats the value using the given formatter.

impl Display for FixedShapeTensorMetadata

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

Formats the value using the given formatter.

impl Hash for FixedShapeTensorMetadata

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for FixedShapeTensorMetadata

fn eq(&self, other: &FixedShapeTensorMetadata) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Eq for FixedShapeTensorMetadata

impl StructuralPartialEq for FixedShapeTensorMetadata