Steady state without initial values

[jd_prada]

Hi

I just have a question. When I try to find the steady state without giving explicit initial values, how does the fsolve routine of Matlab works, since it need the seed to solve the system. My question is, if I do not give initial values, from where dynare begins to find the steady state?

Another question: I have a large model, When I find the steady state with the command "steady;", algorithm 0, without initial values, it converges to the steady state that I know is right. But when i add some initial values that I know are right, none of the algorithms converge. Why this happens?

[MichelJuillard]

1) variables that aren't initialized in INITVAL are set to zero
2) All the various algorithms used by STEADY are variant of the Newton method. When one of the variables is set to its current value instead of zero, the residuals of the first iterations are different and the path of Newton improvments are different. In your case, it doesn't converge.

Best

Michel