Hi,
I am trying out a little adaption in bayesian estimation, where I am trying to modify “kalman_filter.m” (KF) and dsge_likelihood.m"(DL). I am using dynare 4.5.0.
I have a question on the notation in these files, with a focus on the state equation and the associated T matrix,
Comment for T matrix say:
% transition matrix of the state equation
My model has 36 state variables, as shown by DynareResults.dr.state_var in DL namespace, but my T matrix is 43x43.
There are 7 observables around. I am trying to map certain state variables. From DynareResults.dr.state_var I get 36 indices to map to 43 dimensions of T.
Can it be that, what is called the state equation in KF is actually a vector stacking observables and state variables?
This would be supported by the first 7 columns of T being all zeros.
Do I get the right indices for T when I take DynareResults.dr.state_var and add the number of observables?
Thanx in advance for any help
Best
Sven