tfp.experimental.sequential.ensemble_kalman_filter_predict

Ensemble Kalman Filter Prediction.

tfp.experimental.sequential.ensemble_kalman_filter_predict(
    state, transition_fn, seed=None, inflate_fn=None, name=None
)

The Ensemble Kalman Filter is a Monte Carlo version of the traditional Kalman Filter.

This method is the 'prediction' equation associated with the Ensemble Kalman Filter. This takes in an optional inflate_fn to perform covariance inflation on the ensemble [2].

state Instance of EnsembleKalmanFilterState.
transition_fn callable returning a (joint) distribution over the next latent state, and any information in the extra state. Each component should be an instance of MultivariateNormalLinearOperator.
seed Python int seed for random ops.
inflate_fn Function that takes in the particles and returns a new set of particles. Used for inflating the covariance of points. Note this function should try to preserve the sample mean of the particles, and scale up the sample covariance.
name Python str name for ops created by this method. Default value: None (i.e., 'ensemble_kalman_filter_predict').

next_state EnsembleKalmanFilterState representing particles after applying transition_fn.

References

[1] Geir Evensen. Sequential data assimilation with a nonlinear quasi-geostrophic model using Monte Carlo methods to forecast error statistics. Journal of Geophysical Research, 1994.

[2] Jeffrey L. Anderson and Stephen L. Anderson. A Monte Carlo Implementation of the Nonlinear Filtering Problem to Produce Ensemble Assimilations and Forecasts. Monthly Weather Review, 1999.