Page 1 of 1

RBC_Est.mod Errors

PostPosted: Fri Mar 27, 2015 2:03 am
by dale
Hi, think this topic had been dealt many times but face a new error message like below
(already set the mean and variance of psi with gamma.pdf as 1.75, 0.1 respectively)
Do we miss something in last estimation command of RBC_Est.mod?

========================

You did not declare endogenous variables after the estimation/calib_smoother command.

***MESSAGE FROM ROUTINE D9LGIT IN LIBRARY SLATEC.
***FATAL ERROR, PROG ABORTED, TRACEBACK REQUESTED
* NO CONVERGENCE IN 200 TERMS OF CONTINUED FRACTION
* ERROR NUMBER = 3
*
***END OF MESSAGE

***JOB ABORT DUE TO FATAL ERROR.
0 ERROR MESSAGE SUMMARY
LIBRARY SUBROUTINE MESSAGE START NERR LEVEL COUNT
SLATEC D9LGIT NO CONVERGENCE IN 20 3 2 8

error: Due to a bug in Octave, you must choose other values for mean and/or variance of your prior on psi, or use another shape
error: called from:
error: h:\dynare\4.4.3\matlab\draw_prior_density.m at line 58, column 13
error: h:\dynare\4.4.3\matlab\plot_priors.m at line 55, column 38
error: h:\dynare\4.4.3\matlab\dynare_estimation_init.m at line 257, column 13
error: h:\dynare\4.4.3\matlab\dynare_estimation_1.m at line 81, column 63
error: h:\dynare\4.4.3\matlab\dynare_estimation.m at line 89, column 5
error: H:\dynare\work\RBC_Est.m at line 159, column 1

Re: RBC_Est.mod Errors

PostPosted: Fri Mar 27, 2015 2:32 am
by dale
Using Dynare 4.4.3 and Octave 3.8.2 versions now.
It seems that errors has something to do in part with GNU GUI..

If I add graphics_toolkit("gnuplot"); then more graphs come out but a long warning message show up like below


--------------------------------------------------

warning: division by zero
warning: division by zero
warning: division by zero
warning: division by zero
warning: division by zero
warning: division by zero
warning: division by zero
warning: division by zero
Loading 1001 observations from simuldataRBC.m

Restricting the sample to observations 500 to 699. Using in total 200 observations.
Initial value of the log posterior (or likelihood): 668.828
warning: implicit conversion from matrix to real vector
Acceptance ratio [during last 500]: 0.251750 [0.304000] 2% done
warning: implicit conversion from matrix to real vector
Acceptance ratio: 0.362400 100% done

==========================================================
Change in the covariance matrix = 10.
Mode improvement = 6.6106
New value of jscale = 0.0004874
======================================================

Re: RBC_Est.mod Errors

PostPosted: Mon Mar 30, 2015 8:39 am
by dale
Hi, here's another example of incomplete run...
don't know what went wrong...
thanks

Re: RBC_Est.mod Errors

PostPosted: Mon Mar 30, 2015 8:53 am
by jpfeifer
Please provide all files to replicate the issue in a zip or rar-file

Re: RBC_Est.mod Errors

PostPosted: Mon Mar 30, 2015 10:53 am
by dale
Dear Pfeifer,

Attached pls find the files. The RBC_Est.mod is the same as one presented in DYNARE User Guide(2007-208) pp54.

Re: RBC_Est.mod Errors

PostPosted: Thu Apr 02, 2015 9:40 am
by jpfeifer
I was unable to replicate any problems with Matlab. I will have to try it with Octave. Which Dynare version are you using?

Re: RBC_Est.mod Errors

PostPosted: Thu Apr 02, 2015 1:51 pm
by dale
I am using Dynare 4.4.3 and Octave 3.8.2.....
Thanks.

Re: RBC_Est.mod Errors

PostPosted: Thu Apr 02, 2015 7:16 pm
by jpfeifer
It seems there is no reason to worry. Octave seems to have a hard time computing the moments of the prior distribution specified, but finally succeeds. After that, everything runs as expected-

Re: RBC_Est.mod Errors

PostPosted: Fri Apr 03, 2015 5:58 am
by dale
Thanks a lot!
However still not sure if I got all the ouputs and the graphs right and correctly from the estimation. For example, the 4th graph of comparing historical and smoothed values look totally overlapped and same. I think it should be different but something might have gone wrong. I would appreciate if you can send me all outputs and graphs estimated with Matlab or Octave.

Re: RBC_Est.mod Errors

PostPosted: Fri Apr 03, 2015 7:28 am
by jpfeifer
For observed variables, the smoothed and historical values must be the same (unless there is measurement error). The results look OK.