Struct Rls 

pub struct Rls<const N: usize, const D: usize> { /* private fields */ }

Standard RLS maintaining the covariance matrix P directly.

Supports multi-dimensional observations (D > 1) via Cholesky decomposition of the D × D innovation matrix M = λI + A P Aᵀ.

Model

y = A x    where  y ∈ ℝᵈ,  x ∈ ℝⁿ,  A ∈ ℝᵈˣⁿ

The regressor matrix is passed as Aᵀ ∈ ℝⁿˣᵈ (column-major, matching the indiflight C convention).

Type parameters

• N — number of parameters (regressor dimension)
• D — observation dimension (number of outputs per sample)

Example

use flight_solver::rls::{Rls, CovarianceGuards};
use nalgebra::SMatrix;

// 3 parameters, 2-dimensional observations
let mut rls = Rls::<3, 2>::new(1e2, 0.995, CovarianceGuards::default());

let a_t = SMatrix::<f32, 3, 2>::new(
    0.034, 0.227,
    0.135, -0.436,
    -0.169, -0.059,
);
let y = nalgebra::SVector::<f32, 2>::new(0.338, -0.165);

let stats = rls.update(&a_t, &y);

Implementations

impl<const N: usize, const D: usize> Rls<N, D>

where Const<N>: DimName, Const<D>: DimName,

pub fn new(gamma: f32, lambda: f32, guards: CovarianceGuards) -> Self

Create a new standard RLS with initial covariance P = γ I.

Arguments

• gamma — initial covariance diagonal (higher = more uncertain = faster learning)
• lambda — forgetting factor in (0, 1] (lower = faster forgetting of old data)
• guards — numerical guard configuration

pub fn update( &mut self, a_t: &OMatrix<f32, Const<N>, Const<D>>, y: &OVector<f32, Const<D>>, ) -> UpdateStats

Process one observation and update parameter estimate.

Arguments

• a_t — regressor matrix transposed, Aᵀ ∈ ℝⁿˣᵈ (column-major). Each column is one regressor vector.
• y — observation vector y ∈ ℝᵈ

Algorithm

1. Covariance explosion check (temporarily increase λ if P[i,i] > cov_max)
2. Form M = λI + A P Aᵀ and Cholesky-solve for K = P Aᵀ M⁻¹
3. Order decrement limiting on trace(K A P)
4. Covariance update: P ← (KAPmult · P − P Aᵀ Kᵀ) / λ
5. Parameter update: x ← x + K (y − A x)

pub fn params(&self) -> &OVector<f32, Const<N>>

Current parameter estimate x ∈ ℝⁿ.

pub fn params_mut(&mut self) -> &mut OVector<f32, Const<N>>

Mutable access to the parameter estimate.

pub fn covariance(&self) -> &OMatrix<f32, Const<N>, Const<N>>

Current covariance matrix P ∈ ℝⁿˣⁿ.

pub fn lambda(&self) -> f32

Forgetting factor λ.

pub fn set_lambda(&mut self, lambda: f32)

Set the forgetting factor λ ∈ (0, 1].

pub fn samples(&self) -> u32

Number of samples processed so far.

Trait Implementations

impl<const N: usize, const D: usize> Clone for Rls<N, D>

fn clone(&self) -> Rls<N, D>

Returns a duplicate of the value.

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

Performs copy-assignment from source.