Page 1 of 1
model converges in one computer but not in other?
Posted:
Thu Jan 16, 2014 7:19 pmby Shreekar
Hi:
My model converges in other computer while the same model I got it as attachment via e-mail did not converge in my computer. I did not change anything but it does not converge in my computer at all. It says infinite steady state possible. What could be the problem here? Anybody help.
Re: model converges in one computer but not in other?
Posted:
Thu Jan 16, 2014 7:42 pmby Shreekar
Just add in to what I posted before:
I changed back my dynare version to 4.3.3 and the model converged. I was using the recent version 4.4.0. So, it is due to the version conflict! But the newest version should not have a problem of converging the model constructed in an older version but I got it opposite.
Re: model converges in one computer but not in other?
Posted:
Fri Jan 17, 2014 8:18 amby jpfeifer
Please send me the mod-file.
Re: model converges in one computer but not in other?
Posted:
Mon Mar 03, 2014 12:39 pmby jpfeifer
It seems that the reason is that Dynare 4.3.3 provided a message on convergence, even if there were NaN in the residuals.
Re: model converges in one computer but not in other?
Posted:
Mon Mar 03, 2014 3:54 pmby Shreekar
I have sent you a model in PM. I am having problem on solving that in deterministic setup. The model can be solved in stochastic though. Also it converges in 4.3.3 but does not converge in 4.4.0. I appreciate your response.
Best,
Shreekar
Re: model converges in one computer but not in other?
Posted:
Mon Mar 03, 2014 4:44 pmby jpfeifer
That was my answer to your PM. Dynare 4.3.3 is also not able to solve your model. Take a look at oo_.endo_simul and you will see NaN and Inf in your "solution".
Re: model converges in one computer but not in other?
Posted:
Mon Mar 03, 2014 6:18 pmby Shreekar
That's correct. So, does that mean I should use 4.4.0 to solve that model? Is it the problem with the version or my model setup is wrong. but why does it solve in the stochastic setup but not in deterministic setup?
Re: model converges in one computer but not in other?
Posted:
Tue Mar 04, 2014 3:58 pmby Shreekar
I just sent you PM with the model in stochastic shock. It solves under 4.4.2. I just added an equation with the shock. The same model does not solve under deterministic and says something is wrong with the model. Not sure where is the fault?
Re: model converges in one computer but not in other?
Posted:
Tue Mar 04, 2014 7:24 pmby Shreekar
Also please see endo_simul is empty even when the model converges under stoch_simul. There are IRFs but I do not see anything in endo_simul.