BayesFiltSmooth¶

class probnum.filtsmooth.BayesFiltSmooth(dynamics_model, measurement_model, initrv)[source]¶

Bases: abc.ABC

Bayesian filtering and smoothing.

Methods Summary

`filter_step`(start, stop, current_rv, data[, …])	Filter step.
`smooth`(filter_posterior[, _previous_posterior])	Smoothing.

Methods Documentation

abstract filter_step(start, stop, current_rv, data, _linearise_predict_at=None, _linearise_update_at=None, _diffusion=1.0)[source]¶

Filter step.

For Gaussian filters, this means a prediction step followed by an update step.

Return type: Tuple[RandomVariable, Dict]

smooth(filter_posterior, _previous_posterior=None)[source]¶

Smoothing.

Return type: TimeSeriesPosterior