Struct Matrix 

pub struct Matrix<T> { /* private fields */ }

A dense 2D matrix in column-major layout.

Access elements with m[[row, col]] syntax. Data is stored contiguously in column-major order, so flat buffers use row + nrows * col.

Examples

use tensor4all_tensorbackend::Matrix;

let mut m = Matrix::zeros(2, 3);
m[[0, 1]] = 5.0_f64;
assert_eq!(m[[0, 1]], 5.0);
assert_eq!(m[[0, 0]], 0.0);
assert_eq!(m.nrows(), 2);
assert_eq!(m.ncols(), 3);

Implementations

impl<T> Matrix<T>

pub fn try_from_col_major_vec( nrows: usize, ncols: usize, data: Vec<T>, ) -> Result<Self, MatrixShapeError>

Fallibly create a matrix from column-major data after checked shape validation.

Errors

Returns MatrixShapeError::ShapeOverflow when the shape exceeds usize, or MatrixShapeError::DataLengthMismatch when the payload length does not match the shape.

pub fn from_col_major_vec(nrows: usize, ncols: usize, data: Vec<T>) -> Self

Create a matrix from raw column-major data.

Panics

Panics if nrows * ncols overflows or if data.len() != nrows * ncols.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::from_col_major_vec(2, 2, vec![1.0, 3.0, 2.0, 4.0]);
assert_eq!(m[[0, 0]], 1.0);
assert_eq!(m[[0, 1]], 2.0);
assert_eq!(m[[1, 0]], 3.0);
assert_eq!(m[[1, 1]], 4.0);

pub fn as_col_major_slice(&self) -> &[T]

View the underlying column-major data as a contiguous slice.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::from_col_major_vec(2, 2, vec![1, 3, 2, 4]);
assert_eq!(m.as_col_major_slice(), &[1, 3, 2, 4]);

pub fn as_col_major_mut_slice(&mut self) -> &mut [T]

View the underlying column-major data as a mutable contiguous slice.

The slice uses row + nrows * col ordering. This is useful for kernels that validate dimensions once and then operate over contiguous columns.

Examples

use tensor4all_tensorbackend::Matrix;

let mut m = Matrix::from_col_major_vec(2, 2, vec![1, 3, 2, 4]);
m.as_col_major_mut_slice()[1] = 30;
assert_eq!(m[[1, 0]], 30);

pub fn into_col_major_vec(self) -> Vec<T>

Consume the matrix and return its owned column-major buffer.

The returned buffer uses row + nrows * col ordering. Use this when transferring matrix storage to another column-major dense container without cloning.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::from_col_major_vec(2, 2, vec![1.0, 3.0, 2.0, 4.0]);
let data = m.into_col_major_vec();
assert_eq!(data, vec![1.0, 3.0, 2.0, 4.0]);

pub fn to_typed_tensor(&self) -> TypedTensor<T>

where T: TensorScalar,

Borrow this matrix as an owned tenferro [TypedTensor].

This clones the matrix buffer and preserves column-major layout. Use Matrix::into_typed_tensor when the matrix can be consumed.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::from_col_major_vec(2, 2, vec![1.0_f64, 3.0, 2.0, 4.0]);
let tensor = m.to_typed_tensor();
assert_eq!(tensor.shape(), &[2, 2]);
assert_eq!(tensor.as_slice()?, &[1.0, 3.0, 2.0, 4.0]);
assert_eq!(m.as_col_major_slice(), &[1.0, 3.0, 2.0, 4.0]);

pub fn into_typed_tensor(self) -> TypedTensor<T>

where T: TensorScalar,

Consume this matrix as a tenferro [TypedTensor] without cloning.

The tensor shape is [nrows, ncols], and the owned data remains in column-major layout.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::from_col_major_vec(2, 2, vec![1.0_f64, 3.0, 2.0, 4.0]);
let tensor = m.into_typed_tensor();
assert_eq!(tensor.shape(), &[2, 2]);
assert_eq!(tensor.as_slice()?, &[1.0, 3.0, 2.0, 4.0]);

pub fn try_from_typed_tensor( tensor: TypedTensor<T>, ) -> Result<Self, MatrixTensorConversionError>

where T: TensorScalar + Clone,

Consume a rank-2 tenferro [TypedTensor] as a Matrix.

The input tensor must have shape [nrows, ncols] and an owned host buffer. The buffer is reused without cloning and interpreted as column-major matrix storage.

Errors

Returns MatrixTensorConversionError::Rank if the tensor is not rank-2, or MatrixTensorConversionError::HostBuffer if tenferro cannot export the tensor as an owned host buffer.

Examples

use tenferro::TypedTensor;
use tensor4all_tensorbackend::Matrix;

let tensor = TypedTensor::from_vec_col_major(vec![2, 2], vec![1.0_f64, 3.0, 2.0, 4.0])?;
let m = Matrix::try_from_typed_tensor(tensor)?;
assert_eq!(m.nrows(), 2);
assert_eq!(m.ncols(), 2);
assert_eq!(m[[0, 1]], 2.0);

pub fn nrows(&self) -> usize

Number of rows

pub fn ncols(&self) -> usize

Number of columns

impl<T: Clone> Matrix<T>

pub fn from_elem(nrows: usize, ncols: usize, elem: T) -> Self

Create a new matrix filled with a constant value.

Panics

Panics if nrows * ncols overflows.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::from_elem(2, 3, 7.0);
assert_eq!(m[[0, 0]], 7.0);
assert_eq!(m[[1, 2]], 7.0);

impl<T: Clone + Zero> Matrix<T>

pub fn try_zeros(nrows: usize, ncols: usize) -> Result<Self, MatrixShapeError>

Fallibly create a zero-filled matrix after checked shape validation.

Errors

Returns MatrixShapeError::ShapeOverflow when the shape exceeds usize.

pub fn zeros(nrows: usize, ncols: usize) -> Self

Create a zeros matrix

Panics

Panics if nrows * ncols overflows.

Examples

use tensor4all_tensorbackend::Matrix;

let m = Matrix::<f64>::zeros(2, 3);
assert_eq!(m.nrows(), 2);
assert_eq!(m.ncols(), 3);
assert_eq!(m[[0, 0]], 0.0);
assert_eq!(m[[1, 2]], 0.0);

Trait Implementations

impl<T: Clone> Clone for Matrix<T>

fn clone(&self) -> Matrix<T>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<T: Debug> Debug for Matrix<T>

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

Formats the value using the given formatter.

impl<T> Index<[usize; 2]> for Matrix<T>

type Output = T

The returned type after indexing.

fn index(&self, idx: [usize; 2]) -> &Self::Output

Performs the indexing (container[index]) operation.

impl<T> IndexMut<[usize; 2]> for Matrix<T>

fn index_mut(&mut self, idx: [usize; 2]) -> &mut Self::Output

Performs the mutable indexing (container[index]) operation.