type

states.KalmanLinear

class KalmanLinear:

Linear-Gaussian Kalman predict step.

Reads ``cs.{field}`` (mean, shape ``(N, D)``) and ``cs.{field}_cov`` (covariance, shape ``(N, D, D)``) and writes both back as .. math:: x' = F x, \qquad P' = F P F^T + Q \cdot dt.

Parameters

NameTypeDescription
fieldstrField name for the mean. Covariance lives in ``f"{field}_cov"``.
Ftorch.Tensor``(D, D)`` state-transition matrix.
Htorch.Tensor``(M, D)`` measurement matrix. Stored here for the paired :class:`KalmanUpdate` factory; not used by the predict step itself.
Qtorch.Tensor``(D, D)`` process-noise covariance (per unit time when ``dt_scale_q=True``).
Rtorch.Tensor``(M, M)`` measurement-noise covariance. Stored for the paired update; not used by the predict step itself.
dt_scale_q = TrueboolMultiply ``Q`` by ``ctx.delta`` before adding to the predicted covariance. Default ``True``.

Raises

  • ValueError — If ``F``, ``H``, ``Q``, and ``R`` do not all share the same dtype (would force silent per-call casting).

Members

method

property

constant

Source: unitrack/states/kalman/base.py:14