GitLab

Closes issue #315 (https://github.com/DynareTeam/dynare/issues/315)

added tests for mode_compute=4 and mode_compute=6. They increase computing time of the
test suite by about 30 minutes.

Save the names of the estimated parameters in <MODEL_NAME>_mode.mat when mode_compute is equal to six (stochastic optimization).

2) force no analytic derivative with mode_check;

Removed NaNs in M_.Correlation_matrix if they are the consequence of variances (of structural innovations) calibrated to zero.

2) Start adapting identification routines to allow computation of analytic asymptotic Hessian with KF routines

Fixed bug reported by Johannes (http://www.dynare.org/pipermail/dev/2012-May/002015.html), see also trac#257.

estimated with a  particle filter (the non linear  smoother is not yet
implemented).

Fixed bug trac#202. Define the sigma_e_is_diagonal flag when no shocks block is declared in the mod file.

Fixes around latex output: use of longtable for parameter estimates and various fixes for posterior analysis, also for parallel execution.

This includes re-setting the list of output arguments in objective functions
Added test function

oo_.posterior.optimization.mode oo_.posterior.optimization.variance
oo_.posterior.metropolis.mean oo_.posterior.metropolis.variance as
aggregate arrays in addition to previous storage variable by variable

Bug fix. When the estimation command is used, do not issue an error message if options_.order>1. In this
case reset the options_.order equal to one (except if options_.particle.status==1).

Fixed typo. Do not use the non linear filter automatically when order==2, otherwise the non linear filters are used if the estimation command is used after a stoch_simul command with order==2. The non linear filter is used only if options_.particle.status is explicitely set to 1 (before the estimation command). A new estimation option is needed...

added nodisplay and graph_format options and applied to prior plots and graphs in estimstion routine. Tex tables also splitted when more than 50 params are estimated.

nograph: no figure is done
nodisplay: plots are done behind the scenes
graph_format: default only eps figs are saved. it accepts cell vector like {'eps','pdf','fig'}

Revert "added calls to re-initialize persistent variable 'penalty' in dsge_likelihood() and dsge_likelihood_hh() before calling the optimizer"
This reverts commit 108444b7.

The initialization works indeed differently and I will document it in next commit.

dsge_likelihood() and dsge_likelihood_hh() before calling the
optimizer

this function. Suppressed use of globals in CheckPath.

name conflict with cszt

Do not show plots of the smoothed variables, shocks and errors if nograph option is used. Note that with this option the plot are done behind the scene and saved in fig, eps and pdf format (see trac#216).