Struct RandomFeatureMap 

pub struct RandomFeatureMap {
    pub weights: Vec<Vec<f64>>,
    pub biases: Vec<f64>,
    pub config: RandomFeatureConfig,
    /* private fields */
}

Random Fourier feature map z: R^d → R^(2D).

Approximates a shift-invariant kernel via:

z(x) = sqrt(2/D) * [cos(ω_1^T x + b_1), ..., cos(ω_D^T x + b_D)]

Fields

weights: Vec<Vec<f64>>

Frequency vectors, shape [n_features][n_dims].

biases: Vec<f64>

Phase offsets, shape [n_features].

config: RandomFeatureConfig

Configuration used to create this map.

Implementations

impl RandomFeatureMap

pub fn new( n_dims: usize, config: RandomFeatureConfig, ) -> Result<Self, InterpolateError>

Create a new random feature map for the given number of input dimensions.

pub fn transform( &self, x: &[Vec<f64>], ) -> Result<Vec<Vec<f64>>, InterpolateError>

Map input points x (shape [n_samples][n_dims]) to features of shape [n_samples][n_features].

Feature formula: z_i(x) = sqrt(2/D) * cos(ω_i^T x + b_i).

pub fn kernel_approx( &self, x1: &[f64], x2: &[f64], ) -> Result<f64, InterpolateError>

Approximate kernel value k(x1, x2) ≈ z(x1)^T z(x2).

Trait Implementations

impl Clone for RandomFeatureMap

fn clone(&self) -> RandomFeatureMap

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RandomFeatureMap

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

Formats the value using the given formatter.