Prof Pfeifer,
When I was doing Bayesian estimation, only mode_compute=6 works. And the result doesn’t converge. So I use the mode_check. However, I can not understand the meaning of the plots. I wonder how the analyse these plots. Thanks for your help. 
The flat likelihood kernel indicates that some parameters are not identified. The parameter eta runs into a bound where the model cannot be solved for some reason. Use options_.debug=1 to get the error code.
Professor Pfeifer：
Thanks for help. I change two parameters‘ prior distribution， and the mode-check plots are better. But from the plots, it seems that parameter h still can not be identified. Then I used options_.debug=1; identification; at the same time. I got these:
==== Identification analysis ====
Testing prior mean
Evaluating simulated moment uncertainty … please wait
Doing 120 replicas of length 300 periods.
Simulated moment uncertainty … done!
All parameters are identified in the model (rank of H).
All parameters are identified by J moments (rank of J)
==== Identification analysis completed ====
However, there isn’t any result for this command: options_.debug=1
what should I do now? The information means there isn’t identification problem, but the estimation result didn’t converge.
Note the y-scale for h. The parameter is identified. The problem now seems to be omegak where you are not at the peak of the posterior.
Professor
Yes, you are right, the estimated result of omegak is nor correct. The prior mean of omegak is 4, but the estimated is close to zero. Is there any method to fix this problem?
You need to understand why the data pushes the parameter towards 0. Maybe there is an economic reason.