Matrix decomposition

[maile]

Dear All
I am trying to figure out how I can decompose the dr_.ghx and dr_.ghu into A, B and C matrices if they are coefficient matrices of the system: y(t)=Ax(t); x(t)=Bx(t-1) + C*u(t). But I can’t understand. Does anyone has an idea?

Thanks
Mai

[MichelJuillard]

For a model with only one lag and in version 4

[quote]
nstatic = oo_.dr.nstatic
npred = oo_.dr.npred
B = oo_.dr.ghx(nstatic+(1:npred),
D = oo_.dr.ghu(nstatic+(1:npred),
A =dr_.ghx*inv(B)

where y is the original vector of all the variables in the model and x(t) is the vector of state (predetermined) variables. Look at the reference manuel for the ordering of the rows of oo_.dr.ghx and oo_.dr.ghu

This conversion won’t work if there exist linear combinations among state variables.

In version 3, oo_.dr is dr_

Best

Michel[/quote]