the hessian matrix at the "mode" is not positive definite!

[parantap]

Hi, Anybody can help me how to fix this problem? I tried different guesses for the parameters but it sometimes simply does not work.

POSTERIOR KERNEL OPTIMIZATION PROBLEM!
(minus) the hessian matrix at the "mode" is not positive definite!
=> posterior variance of the estimated parameters are not positive.
You should try to change the initial values of the parameters using
the estimated_params_init block, or use another optimization routine.
Warning: The results below are most likely wrong!
> In dynare_estimation_1 at 711
In dynare_estimation at 84
In quant_banking_stoch7_endo_est at 378
In dynare at 174

[jpfeifer]

Search the forum. See e.g. http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4841. Do you already use mode_compute=6? If nothing else works, use mcmc_jumping_covariance

[parantap]

Many thanks for this lead. Would you kiddly tell me where and how I should insert the mcmc_jumping_covaraince command. I put this in the option but it did not take this. An example command will be very much appreciated.

[parantap]

I have downloaded dynare 4.2.2 and it seems that this problem of negative hessian has gone away. was this a bug in the previous versionof dynare? Does anybody know? version 4.2.2 seems quite stable. It sometimes gives a java warning which I don't understand. Besides this it seems to be a working fine.

[jpfeifer]

You mean 4.4.2? As far as I know, there was no bug in this part. There might have been small changes in the mod-finders. Java warnings are Matlab problems, not Dynare. It usually helps to increase the Java Heap Space in Home -> Preferences -> Matlab -> General -> Java Heap Memory

Regarding your example: in 4.4 you can use

Code: Select all

 
estimation(datafile= ..., mcmc_jumping_covariance=identity_matrix);