Struct KalmanFilter 

pub struct KalmanFilter { /* private fields */ }

A discrete linear Kalman filter for tracking state estimates through noisy measurements.

The state evolves as: x(k) = F * x(k-1) + B * u(k) + w(k), w ~ N(0, Q) Measurements are: z(k) = H * x(k) + v(k), v ~ N(0, R)

Implementations

impl KalmanFilter

pub fn new(state_dim: usize, measure_dim: usize) -> Self

Creates a new Kalman filter with the given state and measurement dimensions.

All matrices are initialized to sensible defaults:

• F = identity
• H = identity (top-left sub-block) or zero-padded
• Q, R = small diagonal noise
• x = zero vector
• P = identity (high initial uncertainty)

pub fn set_transition(&mut self, f: Vec<Vec<f64>>) -> Result<()>

Sets the state transition matrix F.

pub fn set_measurement(&mut self, h: Vec<Vec<f64>>) -> Result<()>

Sets the measurement matrix H.

pub fn set_process_noise(&mut self, q: Vec<Vec<f64>>) -> Result<()>

Sets the process noise covariance Q.

pub fn set_measurement_noise(&mut self, r: Vec<Vec<f64>>) -> Result<()>

Sets the measurement noise covariance R.

pub fn set_initial_state(&mut self, x: Vec<f64>) -> Result<()>

Sets the initial state vector.

pub fn predict(&mut self)

Prediction step: x̂ = F * x̂, P = F * P * F^T + Q.

pub fn update(&mut self, z: &[f64]) -> Result<()>

Update step incorporating a measurement z.

Computes:

• y = z - H * x̂ (innovation)
• S = H * P * H^T + R (innovation covariance)
• K = P * H^T * S^{-1} (Kalman gain)
• x̂ = x̂ + K * y (updated state)
• P = (I - K * H) * P (updated covariance)

pub fn state(&self) -> &[f64]

Returns the current state estimate.

pub fn covariance_flat(&self) -> Vec<f64>

Returns the current error covariance matrix (flattened row-major).

pub fn covariance(&self) -> &[Vec<f64>]

Returns a reference to the covariance matrix.