Struct RawTensor 

pub struct RawTensor<D: Scalar> { /* private fields */ }

Tensor with scalar type D.

Implementations

impl<D: Scalar> RawTensor<D>

pub fn from_elem<Mapping: M>(elem: Opt<D>) -> Self

Similar to from_vec, but creates a new tensor with uninit elements.

pub fn from_vec<Mapping: M>(data: Vec<Opt<D>>) -> Self

Creates a new tensor from a vector.

Mapping determines the axes of the tensor, and data is a flat vector containing the tensor elements in the order sorted by the axes.

pub fn from_fn<Mapping: M, F>(f: F) -> Self

where F: FnMut(&Vec<Term>, &IxDyn) -> Opt<D>,

Creates a new tensor from a mapping and a function.

pub fn map<D2: Scalar, F>(&self, f: F) -> RawTensor<D2>

where F: FnMut(&Opt<D>) -> Opt<D2>,

Applies a unary function to each element of the tensor.

Examples

let tensor = TensorValue::<i32>::from_vec::<m![A, B]>(...);
let doubled = tensor.map(|&x| x * Opt::Init(2));

pub fn zip_with<D2: Scalar, D3: Scalar, F>( &self, other: &RawTensor<D2>, f: F, ) -> RawTensor<D3>

where F: Fn(Opt<D>, Opt<D2>) -> Opt<D3>,

Applies a binary function element-wise to two tensors with the same shape.

Examples

let a = RawTensor::<i32>::from_vec::<m![A, B]>(...);
let b = RawTensor::<i32>::from_vec::<m![A, B]>(...);
let sum = a.zip_with(&b, |x, y| x + y);

pub fn reduce( &self, retain_axes: &[Term], reduce_fn: impl Fn(Opt<D>, Opt<D>) -> Opt<D>, identity: Opt<D>, ) -> RawTensor<D>

Reduces axes not in retain_axes using a custom binary function and identity.

pub fn reduce_add(&self, retain_axes: &[Term]) -> RawTensor<D>

Performs reduction (sum) over axes. The axes to retain are specified by their terms.

pub fn write_broadcast( &mut self, src: &Self, unicast: FMapping, broadcast: FMapping, src_offset: &Index, dst_offset: &Index, )

Broadcasts and writes data from another tensor based on the given mappings and offsets.

Arguments

• src - Source tensor to read data from.
• unicast - Mapping for 1-to-1 mapping dimensions.
• broadcast - Mapping for broadcasted dimensions.
• src_offset - Offset to apply to the source tensor indices.
• dst_offset - Offset to apply to the destination (self) tensor indices.

Trait Implementations

impl<D: Clone + Scalar> Clone for RawTensor<D>

fn clone(&self) -> RawTensor<D>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<D: Debug + Scalar> Debug for RawTensor<D>

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

Formats the value using the given formatter.

impl<D: PartialEq + Scalar> PartialEq for RawTensor<D>

fn eq(&self, other: &RawTensor<D>) -> 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<D> Eq for RawTensor<D>

where D: Eq + Scalar,

impl<D: Scalar> StructuralPartialEq for RawTensor<D>