Dear Dynare friends,
We are pleased to announce the release of Dynare 4.4.3.
This is a bugfix release.
The Windows packages are already available for download at:
http://www.dynare.org/download/dynare-stable
The Mac and GNU/Linux packages (for Debian and Ubuntu LTS) should
follow soon.
This release is compatible with MATLAB versions 7.3 (R2006b) to 8.4
(R2014a) and with GNU Octave versions 3.6 to 3.8.
Here is a list of the problems identified in version 4.4.2 and that
have been fixed in version 4.4.3:
- When loading a dataset in XLS, XLSX or CSV format, the first
observation was discarded.
- Reading data in an Excel-file with only one variable wasz leading
to a crash.
- When using the k_order_perturbation option (which is implicit at
3rd order) without the use_dll option, crashes or unexpected
behavior could happen if some 2nd or 3rd derivative evaluates to
zero (while not being symbolically zero)
- When using external function, Ramsey policy could crash or return
wrong results.
- For Ramsey policy, the equation numbers associated with the
Lagrange multipliers stored in M_.aux_vars were erroneously one too
low
- When updating deep parameters in the steady state file, the changes
were not fully taken into account (this was only affecting the
Ramsey policy).
- When using external functions and the bytecode option, wrong
results were returned (if second order derivates of the external
functions were needed).
- The confidence level for computations in estimation, conf_sig could
not be changed and was fixed at 0.9. The new option mh_conf_sig is
now used to set this interval
- Conditional forecasts with non-diagonal covariance matrix used an
incorrect decomposition of the covariance matrix. A Cholesky
factorization is used.
- Option geweke_interval was not effective, Dynare always defaulted
to the standard value.
- The mode_file option lacked backward compatibility with older
Dynare versions.
- Loading an mh_mode file with the mode_file option was broken.
- Using identification with var_exo_det leaded to crashes (the
preprocessor now returns an error if they are used simultaneously)
- The identification command did not print results if the initial
parameter set was invalid and then crashed later on if the MC
sample is bigger than 1
- Inconsistencies between static and dynamic models leaded to crashes
instead of error messages (only with block option).
- The use of external functions crashed the preprocessor when the
derivatives of the external function are explicitly called in the
model block. The preprocessor now forbids the use of external
functions derivates in the model block.
- Using the block option when a variable does not appear in the
current period crashed Dynare instead of providing an error
message.
On behalf of the Dynare Team,
Stéphane Adjemian
--
Université du Maine, Gains
Dynare Team
