simulate.vlmc {VLMC}

Simulate a Discrete Time Series from fitted VLMC model

Description

Simulate from fitted VLMC model – basis of the VLMC bootstrap

Usage

## S3 method for class 'vlmc'
simulate(object, nsim = 1, seed = NULL, n,
         n.start = 64 * object$size[["context"]],
         integer.return = FALSE, keep.RSeed = TRUE, ...)

Arguments

object

typically the result of vlmc(..).

nsim, n

non-negative integer, giving the length of the result. Note that n is deprecated and just there for back compatibility.

seed

random seed initializer; see simulate.

n.start

the number of initial values to be discarded (because of initial effects).

integer.return

logical; if TRUE, the result will be an integer vector with values in 0:(k-1); otherwise the resulting vector consists of letters from the alphabet x$alpha.

keep.RSeed

logical indicating if the seed should be stored with the result (as ‘required’ by the generic simulate). Only set this FALSE with good reasons (back compatibility).

...

(potentially further arguments for other simulate methods.

Details

The .Random.seed is used and updated as with other random number generation routines such as rbinom.

Note that if you want to simulate from a given start sequence x0, you'd use predict.vlmc(x, x0, type= "response") — actually not quite yet.

Value

A "simulate.vlmc" object, basically a vector of length nsim. Either integer or character, depending on the integer.return argument, see above. Further, if keep.RSeed was true (as by default), a "seed" attribute with the random seed at the start of the simulation, for reproducibility.

Author(s)

Martin Maechler

See Also

vlmc and predict.vlmc.

Examples

example(vlmc)

simulate(vlmc.dt1, 100)
simulate(vlmc.dt1c01, 100, int = TRUE)
# n.start = 0: 1st few observations will resemble the data
simulate(vlmc.dt1c01, 20, n.start=0, int = TRUE)