View source on GitHub
|
Ensemble Kalman Filter Update.
tfp.experimental.sequential.ensemble_kalman_filter_update(
state, observation, observation_fn, damping=1.0, seed=None, name=None
)
The Ensemble Kalman Filter is a Monte Carlo version of the traditional Kalman Filter.
This method is the 'update' equation associated with the Ensemble Kalman Filter. In expectation, the ensemble covariance will match that of the true posterior (under a Linear Gaussian State Space Model).
Args | |
|---|---|
state
|
Instance of EnsembleKalmanFilterState.
|
observation
|
Tensor representing the observation for this timestep.
|
observation_fn
|
callable returning an instance of
tfd.MultivariateNormalLinearOperator along with an extra information
to be returned in the EnsembleKalmanFilterState.
|
damping
|
Floating-point Tensor representing how much to damp the
update by. Used to mitigate filter divergence. Default value: 1.
|
seed
|
Python int seed for random ops.
|
name
|
Python str name for ops created by this method.
Default value: None (i.e., 'ensemble_kalman_filter_update').
|
Returns | |
|---|---|
next_state
|
EnsembleKalmanFilterState representing particles at next
timestep, after applying Kalman update equations.
|