Struct LearnedProjection 

pub struct LearnedProjection {
    pub input_dim: usize,
    pub grid_size: usize,
    pub w1: Vec<f32>,
    pub w2: Vec<f32>,
}

Learned toroidal projection parameters.

Implements φ_θ(e) = (σ(W₁e) mod 1, σ(W₂e) mod 1) where σ is sigmoid and e is an embedding vector.

Note: Only available with std feature due to Vec usage.

Fields

input_dim: usize

Input dimension

grid_size: usize

Grid size for output

w1: Vec<f32>

Weight matrix W1 (flattened, grid_size elements)

w2: Vec<f32>

Weight matrix W2 (flattened, grid_size elements)

Implementations

impl LearnedProjection

pub fn new(input_dim: usize, grid_size: usize) -> Self

Create a new projection with random initialization.

pub fn project(&self, embedding: &[f32]) -> (usize, usize)

Project embedding to torus position.

φ_θ(e) = (σ(W₁·e) mod 1, σ(W₂·e) mod 1) * grid_size

Trait Implementations

impl Clone for LearnedProjection

fn clone(&self) -> LearnedProjection

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LearnedProjection

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

Formats the value using the given formatter.