pomp (version 1.6)

Ensemble Kalman filters: Ensemble Kalman filters

Description

The ensemble Kalman filter and ensemble adjustment Kalman filter.

Usage

"enkf"(object, params, Np, h, R, verbose = getOption("verbose"), ...)
"eakf"(object, params, Np, C, R, verbose = getOption("verbose"), ...)
"logLik"(object, ...)
"cond.logLik"(object, ...)
"pred.mean"(object, pars, ...)
"filter.mean"(object, pars, ...)

Arguments

object

An object of class pomp or inheriting class pomp.

params

optional named numeric vector containing the parameters at which the filtering should be performed. By default, params = coef(object).

Np

the number of particles to use.

verbose

logical; if TRUE, progress information is reported.

h

function returning the expected value of the observation given the state.

C

matrix converting state vector into expected value of the observation.

R

matrix; variance of the measurement noise.

pars

Names of variables.

...

additional arguments (currently ignored).

Value

An object of class kalmand.pomp. This class inherits from class pomp.

Methods

logLik

Extracts the estimated log likelihood.

cond.logLik

Extracts the estimated conditional log likelihood $$\ell_t(\theta) = \mathrm{Prob}[y_t \vert y_1, \dots, y_{t-1}],$$ where $y_t$ are the data, at time $t$.

pred.mean

Extract the mean of the approximate prediction distribution. This prediction distribution is that of $$X_t \vert y_1,\dots,y_{t-1},$$ where $X_t$, $y_t$ are the state vector and data, respectively, at time $t$.

filter.mean

Extract the mean of the filtering distribution, which is that of $$X_t \vert y_1,\dots,y_t,$$ where $X_t$, $y_t$ are the state vector and data, respectively, at time $t$.

See Also

pomp, pfilter, and the tutorials on the package website.