Module standard 

Standard (covariance-form) Recursive Least Squares.

Two variants:

• Rls<N, D> — Standard RLS with D-dimensional observation. Uses Cholesky decomposition of the D × D innovation matrix M when D > 1. Model: y = A x where x ∈ ℝⁿ, y ∈ ℝᵈ, A ∈ ℝᵈˣⁿ.

• RlsParallel<N, P> — P independent single-output RLS instances sharing one covariance matrix. The regressor a ∈ ℝⁿ is a vector (scalar observation per output), so the denominator is a scalar—no Cholesky required. Model: yᵀ = a X where X ∈ ℝⁿˣᵖ.

Both include the numerical guards from the indiflight C reference: covariance explosion detection, order decrement limiting, and diagonal clamping.

Structs

Rls

Standard RLS maintaining the covariance matrix P directly.

RlsParallel

Parallel multi-output RLS with shared covariance.