This function extracts the approximate Fisher information matrix G
of the log-likelihood and transforms it using the
Jacobian of the parameter map between Markov chain variables and
model parameters.
Usage
simfiGaussianFILL(
ParMapJac = function(x) {
diag(1, length(x))
}
)
Arguments
- ParMapJac
Jacobian of the parMap function
Value
the approximate Fisher information of the Gaussian log-likelihood with respect to
the MCMC variable, useful for SMMALA
Details
The simfi() values are with respect to the raw model parameters,
while this function rephrases them in terms of the Markov chain's
position.