Description
Generate a random (constant or time-varying) object of class
"dlm", along with states and observations from it.Usage
dlmRandom(m, p, nobs = 0, JFF, JV, JGG, JW)
Arguments
m
dimension of the observation vector.
p
dimension of the state vector.
nobs
number of states and observations to simulate from the model.
JFF
should the model have a time-varying FF component?
JV
should the model have a time-varying V component?
JGG
should the model have a time-varying GG component?
JW
should the model have a time-varying W component?
Value
- The function returns a list with the following components.
- modAn object of class
"dlm". - thetaMatrix of simulated state vectors from the model.
- yMatrix of simulated observations from the model.
- If
nobs is zero, only the mod component is returned.
Details
The function generates randomly the system and observation matrices and
the variances of a DLM having the specified state and observation
dimension. The system matrix GG is guaranteed to have
eigenvalues strictly less than one, which implies that a constant DLM is
asymptotically stationary. The default behavior is to generate a
constant DLM. If JFF is TRUE then a model for
nobs observations in which all
the elements of FF are time-varying is generated. Similarly
with JV, JGG, and JW.References
Anderson and Moore, Optimal filtering, Prentice-Hall (1979)