bssm (version 0.1.8)

sim_smoother: Simulation Smoothing

Description

Function sim_smoother performs simulation smoothing i.e. simulates the states from the conditional distribution \(p(\alpha | y, \theta)\).

Usage

sim_smoother(object, nsim, seed, use_antithetic = FALSE, ...)
# S3 method for gssm
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, method = "dk", ...)
# S3 method for bsm
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, method = "dk", ...)
# S3 method for ar1
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, method = "dk", ...)
# S3 method for ngssm
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, ...)
# S3 method for ng_bsm
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, ...)
# S3 method for svm
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, ...)
# S3 method for ng_ar1
sim_smoother(object, nsim = 1,
  seed = sample(.Machine$integer.max, size = 1),
  use_antithetic = FALSE, ...)

Arguments

object

Model object.

nsim

Number of independent samples.

seed

Seed for the random number generator.

use_antithetic

Use an antithetic variable for location. Default is FALSE. Only used if method is "dk".

...

Ignored.

method

If "dk" (default), use simulation smoothing algorithm by Durbin and Koopman (2002). If "psi", use twisted SMC. Only used for Gaussian models of class "gssm", "bsm", and "ar1".

Value

An array containing the generated samples.

Details

For non-Gaussian/non-linear models, the simulation is based on the approximating Gaussian model.

Examples

# NOT RUN {
model <- bsm(rep(NA, 50), sd_level = uniform(1,0,5), sd_y = uniform(1,0,5))
sim <- sim_smoother(model, 12)
ts.plot(sim[, 1, ])
# }