Page 1 of 1

param_names vs parameter_names

PostPosted: Sun Apr 20, 2014 6:51 pm
by excessliquid
Hello,

I'm new to Dynare. I get an error when running a .mod file "structure has no member "parameter_names". When I look at the .m file I see that it is creating "param_names" variables instead. Files such as dynare.estimation_init.m are instead looking for parameter_names. Is there a way to fix this?

running Dynare 4.4.2 and octave 3.6.4
thanks

Re: param_names vs parameter_names

PostPosted: Wed Apr 23, 2014 10:36 pm
by excessliquid
BTW I am simply trying to run a GPM model from Douglas Laxton's site:

http://www.douglaslaxton.org/id4.html
"A Small Multi-Country Global Projection Model with Financial-Real Linkages and Oil Prices"

Re: param_names vs parameter_names

PostPosted: Sat Apr 26, 2014 12:47 pm
by excessliquid
Update:

The error is generated because dynare_estimation_init.m has the line:
Code: Select all
if isequal(mode_file.parameter_names, bayestopt_.name)


But the mode_file (at least Laxton's mode file) only contains a hessian and vector of modes - the structure does not have a parameter_names variable. So if you get rid of the if statement (with something always true) the code works.

Additional question: How do you create a mode_file? I can't find documentation on this. Is it automatically generated if you specify "mode_file=" in the estimation? When I tried to run the .mod with mode_file=soe1bayes_mode and mode_compute=1/2/3/4 it kept saying there was no such mode_file.

Re: param_names vs parameter_names

PostPosted: Tue May 06, 2014 6:40 pm
by jpfeifer
Once you run a mode-compute till the end, a mode-file is created. If load a mode-file and do not set mode_compute=0, Dynare will start mode-finding at the values provided in the mode file and then start a new mode finding, which finally creates a new mode-file. If you did not run a first mode-finding, there will not be any existing mode-file that you can load.