Struct DenseRestriction 

pub struct DenseRestriction {
    pub data: Vec<f32>,
    pub rows: usize,
    pub cols: usize,
}

Simple restriction map using a dense matrix.

Fields

data: Vec<f32>

Matrix data in row-major order.

rows: usize

Number of rows (output dimension).

cols: usize

Number of columns (input dimension).

Implementations

impl DenseRestriction

pub fn new(data: Vec<f32>, rows: usize, cols: usize) -> Result<Self, SheafError>

Create a new restriction map.

pub fn identity(dim: usize) -> Self

Create an identity restriction (for same-dimension stalks).

pub fn random_orthogonal(dim: usize) -> Self

Available on crate feature rand only.

Create a random orthogonal restriction map.

Uses QR decomposition of random matrix. Useful for connection Laplacians.

Trait Implementations

impl Clone for DenseRestriction

fn clone(&self) -> DenseRestriction

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for DenseRestriction

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

Formats the value using the given formatter.

impl RestrictionMap for DenseRestriction

type Scalar = f32

Scalar type for the map.

type Vector = Vec<f32>

Vector type for input/output.

fn in_dim(&self) -> usize

Input dimension (stalk dimension at source node).

fn out_dim(&self) -> usize

Output dimension (edge space dimension).

fn apply(&self, x: &Self::Vector) -> Result<Self::Vector, SheafError>

Apply the restriction map to a stalk vector.

fn apply_transpose(&self, x: &Self::Vector) -> Result<Self::Vector, SheafError>

Apply the transpose (adjoint) of the restriction map. Used in Laplacian computation.

fn as_matrix(&self) -> Vec<Vec<Self::Scalar>>

Get the matrix representation (for debugging/serialization).

fn frobenius_norm(&self) -> Self::Scalar

Frobenius norm of the map (for regularization).