KalmanConfig

@dataclass(frozen=True)
class KalmanConfig()

KalmanConfig stores the parsed Kalman Filter configuration.

Fields:

Name	Type	Description
y_names	`list[str]`	Names of the included observables.
R	`NDArray \\| None`	Numeric observation-noise covariance matrix (full matrix) built from config parameters.
jitter	`float \\| None`	Jitter term for Cholesky-failure fallback (`None` defers to runtime defaults).
symmetrize	`bool \\| None`	Symmetrization option (`None` defers to runtime defaults).
P0	`P0Config`	`dataclass` storing the mode and values of the initial \(P\) state.
R_symbolic	`sympy.Matrix \\| None`	Symbolic expression of the configured full `R` matrix.
R_param_symbols	`list[sympy.Symbol] \\| None`	Symbols required to build `R_symbolic`.
R_param_names	`list[str] \\| None`	Parameter names (ordered) passed to `R_builder`.
R_builder	`Callable[..., NDArray] \\| None`	Lambdified builder that reconstructs full `R` from `R_param_names`.

Symbolic R Metadata

R_symbolic/R_builder are used by estimation pipelines (e.g. iterative MCMC updates) to rebuild R from the current parameter draw when needed.

@dataclass(frozen=True)
class P0Config()

P0Config stores the required parameters to construct the initial \(P\) state.

P0 Shape

Currently, any P0 produced by P0Config is only populated in the diagonals no matter the configuration. (Zero correlation assumption) A P0 pipeline implementing std and corr fields to build a complete covariance matrix is a planned implementation.

Fields:

Name	Type	Description
mode	`str`	P0 creation mode. `diag` uses given diagonal values, `eye` uses an identity matrix of the appropriate shape.
scale	`float`	Scaling factor of the P0 matrix. (`P0 = P0 * scale`)
diag	`dict[str, float] \\| None`	Variable names and their diagonals (variances, not standard deviation) in the \(P\) matrix.