simulate
Simulate a TSmodel
Description
Simulate a model to produce artificial data.
Usage
simulate(model, input=NULL, input0=NULL,
sampleT=100, Start = NULL, freq = NULL,
noise=NULL, noise.model=NULL,sd=1, SIGMA=NULL, seed=NULL, fortran=T)
Required Arguments
- model
-
An object of class TSmodel or TSestModel.
Optional Arguments
- input
- Data for the exogenous variable if specified in the model.
- sampleT
- The length of the sample to simulate.
- Start
- Start date for resulting data.
- freq
- freq for resulting data.
- y0, input0
-
Lagged values prior to t=1 for y and u, in reverse order so y0[1,] and
input0[1,]correspond to t=0. These arguments are not implemented for state
space models. If not specified initial values are set to zero.
- noise
-
Noise can be supplied. Otherwise it will be generated.
If supplied it should
be a list as described below under returned value.
- SIGMA
-
The covariance of the noise process. If this is specified then sd is ignored.
A vector or scalar is treated as a diagonal matrix. For an object of class
TSestModel, if neither SIGMA nor sd are specified, then SIGMA is set to
the estimated covariance (model$estimates$cov).
- sd
-
The standard deviation of the noise. This can be a vector.
- noise.model
-
A TSmodel to be used for generating noise (not yet supported by SS methods).
- seed
- The seed can be specified to regenerate a simulation.
- fortran
- Specifies the fortran version of the code should be used (instead of the S code version).
Value
The value returned is an object of class TSdata which can be supplied as an
argument to estimation routines. (See TSdata). In addition to the usual
elements (see the description of a TSdata object)
there are some additional elements:
$model- the generating model,
$seed - the initial seed,
$version - the version of S used (random number generators may vary)
$SIGMA as specified
$sd as specified
$noise - a list of $e, $w and $w0 - the noise processes. $w0 is w for t=0 in
the state space model and prior lags in ARMA models. For VAR models B has
no lags so w0 has no effect.
$state - the state variable for state space models.
Details
A state space or ARMA model as described in TSmodel is simulated with pseudo
random noise (The default noise is a normally distributed processes.
An object of class TSdata is
returned. This can be used as input to estimation algorithms. If Start and freq are specified, or if input or noise$w (in that order)
have time series properties, these are given to the output.
If $w0 is supplied as a matrix (rather than a vector) for a state space model
simulation (as it is for ARMA simulations) then it is set to a
vector of zeros. This provides compatability with VAR models (ARMA
models with no lags in B). In general ARMA and SS simulations will
not produce exactly the same results because it is impossible to
determine necessary transformation of initial conditions and w0.
The seed will be set first if it is specified.
See Also
Examples
return to Table of Contents