May 2015 - Dev - dynare.org

Status of testsuite in master branch
by Dynare Robot 30 May '15

30 May '15

1 0

Status of testsuite in master branch
by Dynare Robot 29 May '15

29 May '15

1 0

Status of testsuite in master branch
by Dynare Robot 27 May '15

27 May '15

1 0

Variable limit
by Johannes Pfeifer 26 May '15

26 May '15

Hi, does anyone know a way around this issue: http://www.dynare.org/phpBB3/viewtopic.php?f=1 <http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=6864&p=19888#p19888> &t=6864&p=19888#p19888 -------- Johannes Pfeifer Friesenwall 104 50672 Köln Mobil: +49-(0)170-6936820 jpfeifer(a)gmx.de

2 1

GSL with Cygwin
by Johannes Pfeifer 26 May '15

26 May '15

Dear all, is there a particular reason, GSL is only contained in the 32 bit file http://www.dynare.org/build/dynare-mingw32-libs.zip but not in http://www.dynare.org/build/dynare-mingw64-libs.zip ? I couldnt find any information on a restriction to 32 bit. Best, Johannes -------- Johannes Pfeifer Friesenwall 104 50672 Köln Mobil: +49-(0)170-6936820 jpfeifer(a)gmx.de

3 8

Status of testsuite in master branch
by Dynare Robot 26 May '15

26 May '15

1 0

Summary of make check
by Johannes Pfeifer 23 May '15

23 May '15

Hi Stéphane, Quick question: when I run make check on Sedna, is there a summary file created that tells me which tests crashed? If yes, where is it stored? Regards, Johannes

2 1

Re: [DynareDev] Dynare build system for Windows
by Michel Juillard 23 May '15

23 May '15

Thanks Sebastien, the issue arose from a remark from Johannes that GSL was not included in dynare-mingw64-libs.zip It then appeared that the directories in the zip file didn't match the organisation that we have for cross-compiling. I think that we would need to spend some time to check, complete and maybe update the entire setup. But if there is no use for it, we are better removing the build on Windows information from the README.md Best Michel Sébastien Villemot writes: > Dear all, > > I haven't compiled Dynare on Windows so I may be wrong, but my impression > is that supporting compilation from Cygwin using MinGW is almost the same > as supporting cross-compilation from Linux using MinGW. So I don't think > you will gain much from that move. But if really nobody ever tests the > compilation from Cygwin, then it's indeed better not to claim that this > platform is supported. > > Best > > Le samedi 23 mai 2015 17:48:10 EEST, Michel Juillard a écrit : >> Dear Johannes, >> >> we are thinking of discontinuing the support for building Dynare on a >> Windows plateform. Building via cross-compiling in Linux has been >> working well for several years now and, as far as I know, you are the >> only person who expressed interest in building Dynare directly in >> Windows. >> >> We would of course continue to assist you if you want to maintain a >> building chain in Windows but we would remove the Windows building >> instruction from the Wiki as well as the two zip files with MINGW >> libraries for this. We would in particular find the solution for the GSL >> library that suits you best. >> >> What do you think? Sebastien may also have an opinion on this topic. >> >> Best >> -- Michel Juillard

1 0

Bug in identification when estimating only standard deviations with ML
by Johannes Pfeifer 23 May '15

23 May '15

Hi Marco, could you please have a look at the attached mod-file rbc_ident. There, only variances of exogenous variables are estimated. In both 4.4.3 and master it triggers various bugs. The first thing is that in identification_analysis.m if ~isempty(estim_params_.var_exo), normaliz1 = estim_params_.var_exo(:,7); % normalize with prior standard deviation else normaliz1=[]; end if ~isempty(estim_params_.param_vals), normaliz1 = [normaliz1; estim_params_.param_vals(:,7)]'; % normalize with prior standard deviation end should be if ~isempty(estim_params_.var_exo), normaliz1 = estim_params_.var_exo(:,7)'; % normalize with prior standard deviation else normaliz1=[]; end if ~isempty(estim_params_.param_vals), normaliz1 = [normaliz1 estim_params_.param_vals(:,7)']; % normalize with prior standard deviation end Otherwise, if isempty(estim_params_.param_vals) then normaliz1 is a column, not a row vector. But even upon fixing this, another error appears. When defining the standard deviations as structural parameters, everything works fine. Best, Johannes

1 0

Status of testsuite in master branch
by Dynare Robot 21 May '15

21 May '15

1 0