Hi,
it seems as if commit e95defd117b03fe175fbd8d41b17b435457671e5 introduced a bug into priorbounds. If prior_trunc~=0, the first output argument contains the lower bound of the priors, while if prior_trunc==0 it contains the upper bound, i.e. the order is switched. See http://www.dynare.org/phpBB3/viewtopic.php?f=1 http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4067 &t=4067
Best,
Johannes
--------
Johannes Pfeifer
Haußerstr. 29
72076 Tübingen
Tel.: +49-(0)7071-6396184
Mobil: +49-(0)170-6936820
jpfeifer@gmx.de
"Johannes Pfeifer" jpfeifer@gmx.de writes:
it seems as if commit e95defd117b03fe175fbd8d41b17b435457671e5 introduced a bug into priorbounds. If prior_trunc~=0, the first output argument contains the lower bound of the priors, while if prior_trunc==0 it contains the upper bound, i.e. the order is switched. See http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4067
I don't understand how the commit that you point at can trigger such a bug. The commit just adds an error message under Octave, but I don't see how it can possibly switch the order of output arguments. Which test did you perform to arrive at the conclusion that this commit is the culprit?
When looking at the history of prior_bounds in Tortoise Git it showed that the Octave error message was added. But it also showed that the order had been changed. However, I am not sure that the display of diff was correct. It may have been introduced in a different commit.
-------- Johannes Pfeifer Haußerstr. 29 72076 Tübingen Tel.: +49-(0)7071-6396184 Mobil: +49-(0)170-6936820 jpfeifer@gmx.de
-----Ursprüngliche Nachricht----- Von: dev-bounces@dynare.org [mailto:dev-bounces@dynare.org] Im Auftrag von Sébastien Villemot Gesendet: Freitag, 14. September 2012 12:01 An: List for Dynare developers Betreff: Re: [DynareDev] prior_bounds
"Johannes Pfeifer" jpfeifer@gmx.de writes:
it seems as if commit e95defd117b03fe175fbd8d41b17b435457671e5 introduced a bug into priorbounds. If prior_trunc~=0, the first output argument contains the lower bound of the priors, while if prior_trunc==0 it contains the upper bound, i.e. the order is switched. See http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4067
I don't understand how the commit that you point at can trigger such a bug. The commit just adds an error message under Octave, but I don't see how it can possibly switch the order of output arguments. Which test did you perform to arrive at the conclusion that this commit is the culprit?
-- Sébastien Villemot Researcher in Economics http://www.dynare.org/sebastien Phone: +33-1-40-77-84-04
"Johannes Pfeifer" jpfeifer@gmx.de writes:
When looking at the history of prior_bounds in Tortoise Git it showed that the Octave error message was added. But it also showed that the order had been changed. However, I am not sure that the display of diff was correct. It may have been introduced in a different commit.
Looking closely at the diff, I see no change in order. So I think it is a different commit which introduced the bug.
Are you aware of a previous version of Dynare that was not affected by the bug? We could then use git-bisect to quickly find the culprit.
-----Ursprüngliche Nachricht----- Von: dev-bounces@dynare.org [mailto:dev-bounces@dynare.org] Im Auftrag von Sébastien Villemot Gesendet: Freitag, 14. September 2012 12:01 An: List for Dynare developers Betreff: Re: [DynareDev] prior_bounds
"Johannes Pfeifer" jpfeifer@gmx.de writes:
it seems as if commit e95defd117b03fe175fbd8d41b17b435457671e5 introduced a bug into priorbounds. If prior_trunc~=0, the first output argument contains the lower bound of the priors, while if prior_trunc==0 it contains the upper bound, i.e. the order is switched. See http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4067
I don't understand how the commit that you point at can trigger such a bug. The commit just adds an error message under Octave, but I don't see how it can possibly switch the order of output arguments. Which test did you perform to arrive at the conclusion that this commit is the culprit?
-- Sébastien Villemot Researcher in Economics http://www.dynare.org/sebastien Phone: +33-1-40-77-84-04
Dev mailing list Dev@dynare.org https://www.dynare.org/cgi-bin/mailman/listinfo/dev
Sorry for the confusion. I looked closely into prior_bounds and think it is not a bug. I tested several prior specifications and the current order was correct. However, the description in the header was confusing and is corrected. What also confused me was that suddenly the argument prior_trunc and 1- prior_trunc were shifted, but that was due to using the monotously decreasing transformation 1/sqrt(gaminv(prior_trunc, p7(i)/2,2/p6(i)))+p3(i);
Unfortunately, in the process I found a bug in set_prior.m when estimating models with correlations. I pushed the fix to my repository. You might need to import it into the 4.3 branch.
-------- Johannes Pfeifer Haußerstr. 29 72076 Tübingen Tel.: +49-(0)7071-6396184 Mobil: +49-(0)170-6936820 jpfeifer@gmx.de
-----Ursprüngliche Nachricht----- Von: dev-bounces@dynare.org [mailto:dev-bounces@dynare.org] Im Auftrag von Sébastien Villemot Gesendet: Freitag, 14. September 2012 15:08 An: List for Dynare developers Betreff: Re: [DynareDev] prior_bounds
"Johannes Pfeifer" jpfeifer@gmx.de writes:
When looking at the history of prior_bounds in Tortoise Git it showed that the Octave error message was added. But it also showed that the order had been changed. However, I am not sure that the display of diff was correct. It may have been introduced in a different commit.
Looking closely at the diff, I see no change in order. So I think it is a different commit which introduced the bug.
Are you aware of a previous version of Dynare that was not affected by the bug? We could then use git-bisect to quickly find the culprit.
-----Ursprüngliche Nachricht----- Von: dev-bounces@dynare.org [mailto:dev-bounces@dynare.org] Im Auftrag von Sébastien Villemot Gesendet: Freitag, 14. September 2012 12:01 An: List for Dynare developers Betreff: Re: [DynareDev] prior_bounds
"Johannes Pfeifer" jpfeifer@gmx.de writes:
it seems as if commit e95defd117b03fe175fbd8d41b17b435457671e5 introduced a bug into priorbounds. If prior_trunc~=0, the first output argument contains the lower bound of the priors, while if prior_trunc==0 it contains the upper bound, i.e. the order is switched. See http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4067
I don't understand how the commit that you point at can trigger such a bug. The commit just adds an error message under Octave, but I don't see how it can possibly switch the order of output arguments. Which test did you perform to arrive at the conclusion that this commit is the culprit?
-- Sébastien Villemot Researcher in Economics http://www.dynare.org/sebastien Phone: +33-1-40-77-84-04
Dev mailing list Dev@dynare.org https://www.dynare.org/cgi-bin/mailman/listinfo/dev
-- Sébastien Villemot Researcher in Economics http://www.dynare.org/sebastien Phone: +33-1-40-77-84-04
"Johannes Pfeifer" jpfeifer@gmx.de writes:
Sorry for the confusion. I looked closely into prior_bounds and think it is not a bug. I tested several prior specifications and the current order was correct. However, the description in the header was confusing and is corrected. What also confused me was that suddenly the argument prior_trunc and 1- prior_trunc were shifted, but that was due to using the monotously decreasing transformation 1/sqrt(gaminv(prior_trunc, p7(i)/2,2/p6(i)))+p3(i);
Unfortunately, in the process I found a bug in set_prior.m when estimating models with correlations. I pushed the fix to my repository. You might need to import it into the 4.3 branch.
Thanks. Pushed in master and backported in 4.3.
-----Ursprüngliche Nachricht----- Von: dev-bounces@dynare.org [mailto:dev-bounces@dynare.org] Im Auftrag von Sébastien Villemot Gesendet: Freitag, 14. September 2012 15:08 An: List for Dynare developers Betreff: Re: [DynareDev] prior_bounds
"Johannes Pfeifer" jpfeifer@gmx.de writes:
When looking at the history of prior_bounds in Tortoise Git it showed that the Octave error message was added. But it also showed that the order had been changed. However, I am not sure that the display of diff was correct. It may have been introduced in a different commit.
Looking closely at the diff, I see no change in order. So I think it is a different commit which introduced the bug.
Are you aware of a previous version of Dynare that was not affected by the bug? We could then use git-bisect to quickly find the culprit.
-----Ursprüngliche Nachricht----- Von: dev-bounces@dynare.org [mailto:dev-bounces@dynare.org] Im Auftrag von Sébastien Villemot Gesendet: Freitag, 14. September 2012 12:01 An: List for Dynare developers Betreff: Re: [DynareDev] prior_bounds
"Johannes Pfeifer" jpfeifer@gmx.de writes:
it seems as if commit e95defd117b03fe175fbd8d41b17b435457671e5 introduced a bug into priorbounds. If prior_trunc~=0, the first output argument contains the lower bound of the priors, while if prior_trunc==0 it contains the upper bound, i.e. the order is switched. See http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=4067
I don't understand how the commit that you point at can trigger such a bug. The commit just adds an error message under Octave, but I don't see how it can possibly switch the order of output arguments. Which test did you perform to arrive at the conclusion that this commit is the culprit?
-- Sébastien Villemot Researcher in Economics http://www.dynare.org/sebastien Phone: +33-1-40-77-84-04
Dev mailing list Dev@dynare.org https://www.dynare.org/cgi-bin/mailman/listinfo/dev
-- Sébastien Villemot Researcher in Economics http://www.dynare.org/sebastien Phone: +33-1-40-77-84-04
Dev mailing list Dev@dynare.org https://www.dynare.org/cgi-bin/mailman/listinfo/dev
-
Johannes Pfeifer -
Sébastien Villemot